Cambridge Encyclopedia of Language Sciences PDF [PDF]

Zitiervorschau

The Cambridge Encyclopedia of the Language Sciences

Have you lost track of developments in generative linguistics, fi nding yourself unsure about the distinctive features of Minimalism? Would you like to know more about recent advances in the genetics of language, or about right hemisphere linguistic operation? Has your interest in narrative drawn you to question the relation between stories and grammars? The Cambridge Encyclopedia of the Language Sciences addresses these issues, along with hundreds of others. It includes basic entries for those unfamiliar with a given topic and more specific entries for those seeking more specialized knowledge. It incorporates both well-established fi ndings and cutting-edge research as well as classical approaches and new theoretical innovations. The volume is aimed at readers who have an interest in some aspect of language science but wish to learn more about the broad range of ideas, fi ndings, practices, and prospects that constitute this rapidly expanding field, a field arguably at the center of current research on the human mind and human society. Patrick Colm Hogan is a professor in the Department of English and the Program in Cognitive Science at the University of Connecticut. He is the author of ten books, including Cognitive Science, Literature, and the Arts: A Guide for Humanists and The Mind and Its Stories: Narrative Universals and Human Emotion (Cambridge University Press, 2003).

Advance Praise for The Cambridge Encyclopedia of the Language Sciences

“For both range and depth of exposition and commentary on the diverse disciplinary angles that exist on the nature of language, there is no single volume to match this fi ne work of reference.” – Akeel Bilgrami, Columbia University “The Cambridge Encyclopedia of the Language Sciences is a very welcome addition to the field of language sciences. Its comprehensiveness is praiseworthy, as is the quality of its entries and discussions.” – Seymour Chatman, University of California, Berkeley “Th is ambitious and comprehensive work, and the very high quality of the editors and contributors, ensure that it will be a valuable contribution to the understanding of language and its uses, for both professionals and a more general audience.” – Noam Chomsky, Massachusetts Institute of Technology

THE CAMBRIDGE ENCYCLOPEDIA OF

THE LANGUAGE SCIENCES Edited by PATRICK COLM HOGAN University of Connecticut

C AM BRIDG E U N I VE RSI T Y PRE SS Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo, Delhi, Dubai, Tokyo, Mexico City Cambridge University Press 32 Avenue of the Americas, New York, NY 10013-2473, USA www.cambridge.org Information on this title: www.cambridge.org/9780521866897 © Cambridge University Press 2011 This publication is in copyright. Subject to statutory exception and to the provisions of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press. First published 2011 Printed in the United States of America A catalog record for this publication is available from the British Library. Library of Congress Cataloging in Publication data The Cambridge encyclopedia of the language sciences / edited by Patrick Colm Hogan. p. cm. Includes bibliographical references and index. ISBN 978-0-521-86689-7 (hardback) 1. Linguistics – Encyclopedias. I. Hogan, Patrick Colm. II. Title. P 29.C28 2009 410ʹ.3–dc22 2008041978 ISBN 978-0-521-86689-7 Hardback Cambridge University Press has no responsibility for the persistence or accuracy of URLs for external or third-party Internet Web sites referred to in this publication and does not guarantee that any content on such Web sites is, or will remain, accurate or appropriate.

GENERAL EDITOR Patrick Colm Hogan University of Connecticut, Storrs

ADVISORY EDITORIAL BOARD Florian Coulmas German Institute of Japanese Studies and Duisberg-Essen University

Barbara Lust Cornell University

William Croft University of New Mexico

Lee Osterhout University of Washington

Lyle Jenkins Biolinguistics Institute

James Pustejovsky Brandeis University

CONSULTING EDITORIAL BOARD Mark Baker Rutgers University

Howard Lasnik University of Maryland

Deborah Cameron University of Oxford

Loraine Obler City University of New York

Nigel Fabb University of Strathclyde

William O’Grady University of Hawaii

Carol Ann Fowler Haskins Laboratories and University of Connecticut

Susan Pintzuk University of York

Ronald Geluykens University of Oldenburg

Eleanor Rosch University of California, Berkeley

Margaret Harris Oxford Brookes University

Jay Rueckl University of Connecticut

Zoltán Kövecses Eötvös Loránd University

Mark Turner Case Western Reserve University

To the memory of B. N. Pandit (1916–2007) – philosopher, Sanskritist, father-in-law Puruṣa-artha-śūnyānāṃ guṇānāṃ pratiprasavaḥ kaivalyaṃ sva-rūpa-pratiṣṭhā vā citi-śaktir-iti – Patañjali

CONTENTS

List of Entries A Note on Cross-References and the Alphabetization of the Entries Preface: On the Very Idea of Language Sciences Acknowledgments

1

Language Structure in Its Human Context: New Directions for the Language Sciences in the Twenty-First Century

page xi xv xvii xxiii

1

William Croft

2

The Psychology of Linguistic Form

12

Lee Osterhout, Richard A. Wright, and Mark D. Allen

3

The Structure of Meaning

23

James Pustejovsky

4

Social Practices of Speech and Writing

35

Florian Coulmas

5

Explaining Language: Neuroscience, Genetics, and Evolution

46

Lyle Jenkins

6

Acquisition of Language

56

Barbara Lust

7

Elaborating Speech and Writing: Verbal Art

65

Patrick Colm Hogan

ENTRIES

77

List of Contributors

941

Index

953

ix

ENTRIES

A Abduction Absolute and Statistical Universals Accessibility Hierarchy Acoustic Phonetics Adaptation Ad Hoc Categories Adjacency Pair Age Groups Aging and Language Agreement Agreement Maximization Alliteration Ambiguity Amygdala Analogy Analogy: Synchronic and Diachronic Analyticity Anaphora Animacy Animal Communication and Human Language Aphasia Areal Distinctness and Literature Art, Languages of Articulatory Phonetics Artificial Languages Aspect Auditory Processing Autism and Language Autonomy of Syntax B Babbling Basal Ganglia Basic Level Concepts Bilingual Education Bilingualism, Neurobiology of Bilingualism and Multilingualism Binding

Biolinguistics Birdsong and Human Language Blended Space Blindness and Language Bounding Brain and Language Broca’s Area C Cartesian Linguistics Case Categorial Grammar Categorization Causative Constructions C-Command Cerebellum Charity, Principle of Children’s Grammatical Errors Chirographic Culture Clitics and Cliticization Codeswitching Cognitive Architecture Cognitive Grammar Cognitive Linguistics and Language Learning Cognitive Linguistics, Language Science, and Metatheory Cognitive Poetics Coherence, Discourse Coherence, Logical Colonialism and Language Color Classification Communication Communication, Prelinguistic Communicative Action Communicative Intention Comparative Method Competence Competence and Performance, Literary Compositionality

Computational Linguistics Concepts Conceptual Blending Conceptual Development and Change Conceptual Metaphor Conduit Metaphor Connectionism and Grammar Connectionism, Language Science, and Meaning Connectionist Models, Language Structure, and Representation Consciousness and Language Consistency, Truth, and Paradox Constituent Structure Constraints in Language Acquisition Construction Grammars Contact, Language Context and Co-Text Control Structures Conversational Implicature Conversational Repair Conversation Analysis Cooperative Principle Core and Periphery Corpus Callosum Corpus Linguistics Creativity in Language Use Creoles Critical Discourse Analysis Critical Periods Culture and Language Cycle, The

D Deconstruction Defi nite Descriptions Deixis Descriptive, Observational, and Explanatory Adequacy Dhvani and Rasa

xi

List of Entries Dialect Dialogism and Heteroglossia Diff usion Digital Media Diglossia Discourse Analysis (Foucaultian) Discourse Analysis (Linguistic) Discrete Infi nity Disorders of Reading and Writing Division of Linguistic Labor Dyslexia

E Ellipsis Embodiment Emergentism Emergent Structure Emotion and Language Emotion, Speech, and Writing Emotion Words Emplotment Encoding Énoncé/Statement (Foucault) Essentialism and Meaning Ethics and Language Ethnolinguistic Identity Event Structure and Grammar Evidentiality Evolutionary Psychology Exemplar Exemplar Theory Extinction of Languages

F Family Resemblance Feature Analysis Felicity Conditions Field (Bourdieu) Film and Language Filters Focus Foregrounding Forensic Linguistics Formal Semantics Forms of Life Frame Semantics Framing Effects Frontal Lobe Functional Linguistics

G Games and Language Gender and Language Gender Marking Generative Grammar Generative Poetics Generative Semantics Generic- and Specific-Level Metaphors

xii

Genes and Language Gesture Government and Binding Grammaticality Grammaticality Judgments Grammaticalization Grooming, Gossip, and Language

H Habitus, Linguistic Head-Driven Phrase Structure Grammar Hippocampus Historical Linguistics Historical Reconstruction Holophrastic Stage, The Homologies and Transformation Sets

I Icon, Index, and Symbol Ideal Speech Situation Identity, Language and Ideology and Language Idioms Idle Talk and Authenticity Ijtihād (Interpretive Effort) I-Language and E-Language Illocutionary Force and Sentence Types Image Schema Implicational Universals Indeterminacy of Translation Indexicals Inequality, Linguistic and Communicative Infantile Responses to Language Information Structure in Discourse Information Theory Innateness and Innatism Integrational Linguistics Intension and Extension Intentionality Internal Reconstruction Interpretation and Explanation Interpretive Community Intertextuality Intonation Irony

L Language, Natural and Symbolic Language Acquisition Device (LAD) Language Change, Universals of Language Families Language-Game Language-Learning Environment Language of Thought Language Policy Laws of Language Learnability

Left Hemisphere Language Processing Legal Interpretation Lexical Acquisition Lexical-Functional Grammar Lexical Learning Hypothesis Lexical Processing, Neurobiology of Lexical Relations Lexical Semantics Lexicography Linguistic Relativism Literacy Literariness Literary Character and Character Types Literary Universals Literature, Empirical Study of Logic and Language Logical Form Logical Positivism

M Mapping Markedness Market, Linguistic Marxism and Language Meaning and Belief Meaning and Stipulation Meaning Externalism and Internalism Media of Communication Memes and Language Memory and Language Mental Models and Language Mental Space Merge Metalanguage Metaphor Metaphor, Acquisition of Metaphor, Information Transfer in Metaphor, Neural Substrates of Metaphor, Universals of Meter Methodological Solipsism Methodology Metonymy Minimalism Mirror Systems, Imitation, and Language Modality Modern World-System, Language and the Modularity Montague Grammar Mood Morpheme Morphological Change Morphological Typology Morphology Morphology, Acquisition of Morphology, Evolution and

List of Entries Morphology, Neurobiology of Morphology, Universals of Motif Movement Music, Language and

N Narrative, Grammar and Narrative, Neurobiology of Narrative, Scientific Approaches to Narrative Universals Narratives of Personal Experience Narratology Nationalism and Language Natural Kind Terms Necessary and Sufficient Conditions Negation and Negative Polarity Network Theory Neurochemistry and Language Neuroimaging Number

O Occipital Lobe Optimality Theory Oral Composition Oral Culture Ordinary Language Philosophy Origins of Language Overregularizations

P Parable Paralanguage Parameters Parietal Lobe Parsing, Human Parsing, Machine Passing Theories Performance Performative and Constative Perisylvian Cortex Perlocution Person Philology and Hermeneutics Phoneme Phonetics Phonetics and Phonology, Neurobiology of Phonological Awareness Phonology Phonology, Acquisition of Phonology, Evolution of Phonology, Universals of Phrase Structure Pidgins Pitch Poetic Form, Universals of

Poetic Language, Neurobiology of Poetic Metaphor Poetics Point of View Politeness Politics of Language Possible Worlds Semantics Possible Worlds Semantics and Fiction Pragmatic Competence, Acquisition of Pragmatics Pragmatics, Evolution and Pragmatics, Neuroscience of Pragmatics, Universals in Pragmatism and Language Predicate and Argument Preference Rules Prestige Presupposition Primate Vocal Communication Priming, Semantic Principles and Parameters Theory Principles and Parameters Theory and Language Acquisition Print Culture Private Language Projectibility of Predicates Projection (Blending Theory) Projection Principle Proposition Propositional Attitudes Prototypes Proverbs Psychoanalysis and Language Psycholinguistics Psychonarratology Psychophysics of Speech

Q Qualia Roles Quantification Quantitative Linguistics

R Radical Interpretation Reading Realization Structure Rectification of Names (Zheng Ming) Recursion, Iteration, and Metarepresentation Reference and Extension Reference Tracking Register Regularization Relevance Theory Religion and Language Representations Rhetoric and Persuasion Rhyme and Assonance

Rhythm Right Hemisphere Language Processing Role and Reference Grammar Rule-Following

S Schema Scripts Second Language Acquisition Self-Concept Self-Organizing Systems Semantic Change Semantic Fields Semantic Memory Semantic Primitives (Primes) Semantics Semantics, Acquisition of Semantics, Evolution and Semantics, Neurobiology of Semantics, Universals of Semantics-Phonology Interface Semantics-Pragmatics Interaction Semiotics Sense and Reference Sentence Sentence Meaning Sexuality and Language Signed Languages, Neurobiology of Sign Language, Acquisition of Sign Languages Sinewave Synthesis Situation Semantics Socially Distributed Cognition Sociolinguistics Source and Target Specific Language Impairment Speech-Acts Speech Anatomy, Evolution of Speech-Language Pathology Speech Perception Speech Perception in Infants Speech Production Spelling Spreading Activation Standardization Standard Theory and Extended Standard Theory Stereotypes Story and Discourse Story Grammar Story Schemas, Scripts, and Prototypes Stress Structuralism Stylistics Stylometrics Subjacency Principle Suggestion Structure Syllable

xiii

List of Entries Synchrony and Diachrony Syntactic Change Syntax Syntax, Acquisition of Syntax, Evolution of Syntax, Neurobiology of Syntax, Universals of Syntax-Phonology Interface

T Teaching Language Teaching Reading Teaching Writing Temporal Lobe Tense Text Text Linguistics Thalamus Thematic Roles Theory of Mind and Language Acquisition Tone Topicalization Topic and Comment Traces

xiv

Transformational Grammar Translation Truth Truth Conditional Semantics Two-Word Stage Typology

Verbal Reasoning Verbal Reasoning, Development of Verifiability Criterion Verse Line Voice Voice Interaction Design

U

W

Underlying Structure and Surface Structure Universal Grammar Universal Pragmatics Universals, Nongenetic Usage-Based Theory Use and Mention

Wernicke’s Area Word Classes (Parts of Speech) Word Meaning Word Order Word Recognition, Auditory Word Recognition, Visual Words Working Memory and Language Processing Writing, Origin and History of Writing and Reading, Acquisition of Writing and Reading, Neurobiology of Writing Systems

V Vagueness Verbal Art, Evolution and Verbal Art, Neuropsychology of Verbal Display Verbal Humor Verbal Humor, Development of Verbal Humor, Neurobiology of

X X-Bar Theory

A NOTE ON CROSS-REFERENCES AND THE ALPHABETIZATION OF THE ENTRIES

Cross-references are signaled by small capitals (boldface when implicit). They are designed to indicate the general relevance of the cross-referenced entry and do not necessarily imply that the entries support one another. Note that the phrasing of the cross-references does not always match the entry headings precisely. In order to minimize the disruption of reading, entries often use shortened forms of the entry headings for cross-references. For example, “this process involves parietal structures” points to the entry “Parietal Lobe.” In some cases, a cross-reference may refer to a set of entries. For example, “architectures of this sort are found in connectionism” alerts the reader to the presence of entries on connectionism generally, rather than to a single entry. Finally, a cross-reference may present a heading in a different word order. For example, the target entry for “here we see another universal of phonology ” would be listed as “Phonology, Universals of.” In general, entries with multiword headings are alphabetized under their main language term. Thus, the entry for “Universals of Phonology” is listed as “Phonology, Universals of.” The main exceptions to this involve the words language and linguistic or linguistics, where another term in the heading seemed more informative or distinctive in the context of language sciences (e.g., “Linguistic Market” is listed as “Market, Linguistic”).

xv

PREFACE: ON THE VERY IDEA OF LANGUAGE SCIENCES

A title referring to language sciences tacitly raises at least three questions. First, what is a science? Second, what is language? Finally, what is a language science? I cannot propose answers to these questions in a short preface. Moreover, it would not be appropriate to give answers here. The questions form a sort of background to the essays and entries in the following pages, essays and entries that often differ in their (explicit or implicit) answers. However, a preface of this sort can – and should – indicate the general ideas about science and language that governed the development of The Cambridge Encyclopedia of the Language Sciences.

WHAT IS SCIENCE? Philosophers of science have often been concerned to defi ne a demarcation criterion, separating science from nonscience. I have not found any single criterion, or any combination of criteria, compelling in the sense that I have not found any argument that, to my satisfaction, successfully provides necessary and sufficient conditions for what constitutes a science. In many ways, one’s acceptance of a demarcation criterion is guided by what one already considers to be a science. More exactly, one’s formulation of a demarcation criterion tends to be a function of what one takes to be a paradigmatic science or, in some cases, an exemplary case of scientific practice. The advocates of strict demarcation criteria meet their mirror opposites in writers who assert the social construction of science, writers who maintain that the difference between science and nonscience is simply the difference between distinct positions within institutions, distinct relations to power. Suppose we say that one discipline or theory is a science and another is not. Th is is just to say that the former is treated as science, while the latter is not. The former is given authority in academic departments, in relevant institutions (e.g., banks, in the case of economics), and so on. Again, this is not the place for a treatise on the philosophy of science. Here it is enough to note that I believe both sides are partially correct and partially incorrect. First, as already noted, I do not believe that there is a strict, defi nitive demarcation criterion for science. However, I do believe that there is

a complex of principles that roughly defi ne scientific method. These principles do not apply in the same way to chemical interactions and group relations – and that is one reason why narrow demarcation criteria fail. However, they are the same general principles across different domains. Very simply, scientific method involves inter alia the following practices: 1) the systematic study of empirically ascertainable patterns in a given area of research; 2) the formulation of general principles that govern those patterns; 3) the attempt to uncover cases where these principles do not govern observed patterns; 4) the attempt to eliminate gaps, vagueness, ambiguity, and the like from one’s principles and from the sequences of principles and data that produce particular explanations; and 5) the attempt to increase the simplicity of one’s principles and particular explanations. Discourses are scientific to the extent that they routinely involve these and related practices. Note that none of this requires, for example, strict falsification or detailed prediction. For example, social phenomena are most often too complex to allow for significant prediction, in part because one cannot gather all the relevant data beforehand. Th is does not mean that they are closed to systematic explanations after the fact, as more data become available. Of course, following such methodological guidelines is not all there is to the actual practice of science. There are always multiple options for formulating general principles that fit the current data. The evaluation of simplicity is never entirely straightforward. Theories almost invariably encounter anomalous data in some areas and fail to examine other areas. Moreover, in many cases, the very status of the data is unclear. Despite all this, we hierarchize theories. We teach some and do not teach others. Agencies fund some and do not fund others. The very nature of the enterprise indicates that even in ideal circumstances, this cannot be purely meritocratic. Moreover, real circumstances are far from ideal. Thus, in the real world, adherence to methodological principles may be very limited (see, for example, Faust 1984, Mahoney 1977, and Peters and Ceci 1982). Th is is where social constructionism enters. It seems undeniable that relations of institutional power, the political economy of professions, and ideologies of nation or gender guide what is institutionalized, valued, funded, and so forth.

xvii

Preface In putting together a volume on science, then, I have tried to incorporate the insights of both the more positive views of science and the more social constructionist views. Put in a way that may seem paradoxical, I have tried to include all approaches that fit the loose criteria for science just mentioned. I believe that these loose criteria apply not only to paradigmatic sciences themselves but also to many social critiques of science that stress social construction. I have therefore included a wide range of what are sometimes called the human sciences. Indeed, the volume could be understood as encompassing the language-relevant part of the human sciences – which leads to our second question.

Group Dynamics Society

Mind

Individual Interactions

Mental Representations Intentions

WHAT IS LANGUAGE? Like “science,” one’s defi nition of “language” depends to a great extent on just what the word calls to mind. One’s view of language is likely to vary if one has in mind syntax or semantics, hearers or speakers, dialogues or diaries, brain damage or propaganda, storytelling or acoustic phonetics. A fi rst impulse may be to see one view of language as correct and the others as false. And, of course, some views are false. However, I believe that our understanding of language can and, indeed, should sustain a great deal of pluralism. In many ways, my own paradigm for human sciences is cognitive science. Cognitive science brings together work from a remarkable array of disciplines – literally, from Anthropology to Zoology. Moreover, it sustains a range of cognitive architectures, as well as a range of theories within those architectures. Thus, it is almost by its very nature pluralistic. Of course, some writers wish to constrain this pluralism, insisting that one architecture is right and the others are wrong. Certainly, one can argue that particular architectures are wrong. However, perhaps the most noteworthy aspect of cognitive science is that it sustains a series of types of cognitive architecture. In Cognitive Science, Literature, and the Arts (2003), I argued that these types capture patterns at different levels of analysis. Thus, all are scientifically valuable. More exactly, we may distinguish three levels of cognitive analysis: bodies, minds, and groups or societies. These levels stand in a hierarchical relation such that bodies are more explanatorily basic than minds, and minds are more explanatorily basic than groups or societies. Lower levels provide causally necessary principles for higher levels. Minds do not operate without brains. People without minds do not interact in groups. In other words, lower levels explain higher levels. However, higher-level patterns provide interpretive principles for understanding lower levels (see interpretation and explanation). We explain the (mental) feeling of fear by reference to the (bodily) operation of the amygdala. But, at the same time, we understand amygdala activity as fear because we interpret that activity in terms of the mental level. In the analysis of cognition, the most basic, bodily cognitive architecture is provided by neurobiology. However, due to the intricate particularity of neurobiology, we often draw on more abstract associative models at this level. These models serve to make the isolation and explanation of patterns less computationally complex and individually variable. The

xviii

Body

Associative Networks Brains

Figure 1. Levels of cognitive analysis. Between the levels, black arrows represent the direction of explanation, while hollow arrows represent the direction of interpretation. Within the levels, the superior items are more computationally tractable or algorithmically specifiable models of the inferior items, either singly (in the case of brains and intentions) or collectively (in the case of individual interactions). Tractability may be produced by simplification (as in the case of bodily architectures), by systematic objectification (as in the case of mental architectures), or by statistical abstraction (as in the case of social analysis).

most important architectures of the latter sort are found in connectionism. As a wide range of writers have stressed, the distinctive feature of mind – our second level of analysis – is intentionality. However, intentionality, as subjective and experiential, is often not well suited for scientific study. Many theorists have therefore sought to systematize and objectify our understanding of mind. Most cognitive treatments of the mental level have their roots in folk psychology, a minimal, commonsense objectification of intention in terms of beliefs and aims. But these cognitive treatments draw on empirical research and principles of scientific method to develop models of the human mind that are sometimes very far removed from folk psychology. Specifically, they most often replace belief by mental representations and algorithmically specifiable operations on those representations. We may therefore refer to these models as representational. Representationalism serves to make intention more tractable through a mentalistic architecture that is precisely articulated in its structures, processes, and contents. Finally, our treatment of societies may be loosely divided into the more intentional or mental pole of individual interaction and the less subjective, more broadly statistical pole of group dynamics. (See Figure 1.) These divisions apply to language no less than they apply to other areas of human science. We draw on our representational account of syntax to understand certain brain processes in the perisylvian cortex . Conversely, we explain

Preface the impairment of (mental) syntactic capacities by reference to (bodily) lesions in that area. For our purposes, the crucial part of this analysis is its implication that language includes all three levels and that the sciences of language should therefore encompass brains, associative networks, intentions, mental representations, individual interactions, and group dynamics. Th is takes us to our third question.

WHAT IS A SCIENCE OF LANGUAGE? The preceding sections converge on a broad, pluralistic – but not indiscriminate – account of what constitutes a language science. Specifically, a language science is the application of general principles of scientific method to language phenomena at any level. At the level of brain, we have neurolinguistics (see brain and language). At the level of associative networks, we have connectionism. Intentionalism leads us to certain forms of ordinary language philosophy. Representational architectures are particularly well developed, including Noam Chomsky’s various theories (see, for example, minimalism), cognitive linguistics, and other approaches. Personal interaction and group dynamics are taken up in pragmatics, linguistic discourse analysis, and sociolinguistics. Just as language encompasses patterns at all these levels, language science necessarily includes systematic accounts of language at all these levels. Again, the levels of language are interrelated without being reducible. Similarly, the various sciences are interrelated – systematically interrelated through “upward” explanation and “downward” understanding or interpretation – without being reducible. The preceding points should serve to clarify something that is obvious, but rather vague, in ordinary speech: Language science is not the same as language. Language science is a systematic treatment of language that seeks to provide both explanation and understanding. Thus, an encyclopedia of the language sciences does not present the same material as an encyclopedia of language. It presents the current state of theoretical explanation and understanding (along with some historical background that is important for contextualizing current theories). It does not present the current state of knowledge about particular features of particular languages – except insofar as these features enter into research programs that aim toward broader explanatory accounts or principles of more general understanding. Thus, the phonology of Urdu, the morphology of Quechua, the metrical principles of English verse lines, and the story and discourse structure of Chinese narratives enter into the following essays and entries only insofar as they enter into larger theoretical concerns. Of course, to say this is only to mark out a general area for an encyclopedia of the language sciences. It does not determine precisely what essays and/or entries should make up such a work. Th is leads to our fi nal concern.

THE STRUCTURE OF THE VOLUME The preceding view of language science guided the formulation of topics for the entries and the organization of the introductory

essays. However, it was not the only factor. In language sciences, and indeed in human sciences generally, we need to add two further considerations. The preceding analysis implicitly treats language patterns as if they are comparable to any patterns isolated in the natural sciences. However, there are two differences between patterns in language and, say, the patterns isolated by physicists. First, language patterns are mutable. They are mutable in three ways – at the level of groups, at the level of individual minds or brains, and at the level of common genetic inheritance. Insofar as language patterns change at the level of groups, this mutability is comprehended by group dynamics and related processes (most obviously in historical linguistics). But mental and neurobiological theories do not necessarily treat the other two sorts of mutability, for such theories tend to focus on steady states of language. We therefore account for changes in the individual mind or brain by reference to development or acquisition (see phonology, acquisition of; syntax, acquisition of ; and so on). We account for changes in common genetic properties through the evolution of language (see phonology, evolution of; syntax, evolution of; and so on). The second difference between patterns in language and patterns isolated by physicists is related to the fi rst. Just as we may be insufficient in language, we may be more than sufficient. In other words, there is a difference between ordinary usage and skilled usage. Rocks do not fall well or badly. They simply fall, and they do so at the same rate. People, however, speak well or badly, effectively or ineffectively, in a manner that is clichéd or unusually creative (see creativity in language use). The point is most obvious in verbal art – which leads us to “the most sweet and pleasing sciences of poetry,” as Cervantes put it (1950, 426). In keeping with the preceding analysis, then, the main topics in language science are treated initially in a series of seven overview essays. The fi rst essay provides a general introduction to the study of language. Its purpose is to orient readers toward the field as a whole. The second and third essays turn to the mental level of language since this is the most widely analyzed. Due to the amount of work in this area, and due to the diversity of approaches, the treatment of this level is divided into two chapters. The fi rst addresses “formal” aspects of language – syntax, phonology, and so forth. The second takes up meaning. The fourth and fi fth chapters address the other two levels of language – society (at the “top”) and the brain (at the “bottom”). The latter also addresses the topics of genetics and evolution, integrating these with the treatment of the brain. The sixth chapter takes up language acquisition. Thus, it turns from the evolution of the general language capacities of the human brain to the development of the particular language competence of individual human minds. Finally, the seventh chapter considers the nonordinary use of language in verbal art. The subsequent entries specify, elaborate, augment, and revise the ideas of these essays. Here, of course, the editor of a volume on language sciences faces the problem of just what entries should be included. In other words, if language sciences encompass the language-related part of neuroscience, social science, and so forth, just what is that language-related part? What does it include, and what does it exclude? One might defi ne

xix

Preface this part very narrowly as including only phenomena that are necessarily bound up with oral speech, sign languages, or writing. More theoretically, one might defi ne this part as comprising neurobiological, mental, or social phenomena that occur only in connection with distinctive properties of speech, signing, or writing. Certainly, an encyclopedia treating language will focus on phenomena that are inseparable from speech, sign languages, and/or writing and on such distinctive aspects of natural language as syntax. However, here, too, I believe it would be a mistake to confi ne language sciences within a narrowly defi ned domain. Therefore, I have adopted a looser criterion. The volume centrally addresses distinctive properties of natural language. However, it takes up a wider range of phenomena that are closely connected with the architectural or, even more importantly, the functional features of speech, sign languages, and writing. There are several cognitive operations for which speech, signing, and writing appear to have direct functional consequences. One is referential – the specification, compilation, and interrelation of intentional objects (see the entries on reference). Here I have in mind phenomena ranging from the division of the color spectrum to the elaboration of causal relations. A second area is mnemonic – the facilitation and partial organization of memory (see, for example, encoding). A third is inferential – the derivation of logical implications. A fourth is imaginative – the expansion and partial structuring of simulation. One could think of the fi rst and second functions as bearing on direct, experiential knowledge of present or past objects and events. The third and fourth functions bear, rather, on indirect knowledge of actual or possible objects and events. Two other functions are connected with action rather than with knowledge. The fi rst is motivational – the extension or elaboration of the possibilities for emotional response (see emotion and language). A fi nal area is interpersonal – the communication of referential intents, memories, inferences, simulations, and motivations. In determining what should be included in the volume, I have taken these functions into account, along with architectural considerations. Thus I see issues of interpretation and emplotment (one of the key ways in which we organize causal relations) as no less important than phonology or syntactic structure. Of course, we have more fi rmly established and systematic knowledge in some areas than in others. Thus some entries will necessarily be more tentative, and make reference to a broader variety of opinion or a more limited research base. But that is not a reason to leave such entries aside. Again, the purpose of an encyclopedia of language science is not to present a compilation of well-established particular facts, but rather to present our current state of knowledge with respect to explanation and understanding. In keeping with this, when generating the entries (e.g., “Phonology,” “Syntax,” “Neurobiology of Phonology,” “Neurobiology of Syntax,” “Acquisition of Phonology,” and so on), I have tried to be as systematic as possible. Thus the volume includes some topics that have been under-researched and under-theorized. For example, if neurobiology does in fact provide a causal substrate for higher levels, then there should

xx

be important things to say, not only about the neurobiology of syntax, but also about the neurobiology of pragmatics and the neuropsychology of verbal art. The fi rst has certainly been more fully researched than the second or third. But that is only more reason to stress the importance of the second and third, to bring together what research has been done, and to point to areas where this research could be productively extended. While it is possible to be systematic with research areas, one cannot be systematic with theories. Theories are more idiosyncratic. They differ from one another along many axes and cannot be generated as a set of logical possibilities. I have sought to represent theories that have achieved some level of acceptance in scientific communities. Given limitations of space, decisions on this score have often been difficult – particularly because social constructionist and related analyses show that acceptance in scientific communities is by no means a straightforward function of objective scientific value. Th is leads us once again to the issue of the validity of theories. It should come as no surprise that my view of the issue in effect combines a pluralistic realism with a roughly Lakatosian advocacy of research programs and a Feyerabend-like practical anarchism (Feyerabend 1975; Lakatos 1970). Specifically, I take it that some theories are true and others are not. However, I do not believe that only one theory is true. Different theories may organize the world in different ways. There is no correct way of organizing the world (though some ways will be more useful than others for particular purposes). On the other hand, once the world is organized in a certain way, then certain accounts of the world are correct and certain accounts are incorrect. To take a simple example, we may divide the color spectrum in different ways (see color classification). No division is correct or incorrect. But once we have a division, there are facts about the color of particular objects. (Th is view is related to Donald Davidson’s (1984) argument that truth is not relative to a conceptual scheme, though it is, of course, relative to the meaning of one’s words. It also may have some similarity to Hilary Putnam’s (1981) “internal realism,” depending on how that is interpreted.) Moreover, even for one organization of the world, we can never defi nitively say that a given theory is or is not true. Note that this means we cannot even strictly falsify a theory. We can refer to the ongoing success of a research program – and that is important. Yet I do not share Imre Lakatos’s (1970) optimism about research programs. To some extent, research programs appear to succeed insofar as they have powerful institutional support, often for not very good intellectual reasons. Here, then, I agree with Paul Feyerabend (1975) that orthodoxy in theorization is wrong. It is wrong not only in explicitly or implicitly identifying institutional support with validity. Thus, it is wrong not only for social constructionist reasons. It is wrong also for, so to speak, positivist reasons. It is wrong in diminishing the likelihood of intellectual progress, the likelihood of increasing the validity of our theories, which is to say the scope of explanation and understanding produced by these theories. Whether or not this very brief sketch points toward a good philosophy of science, it does, I believe, point toward a good philosophy for an encyclopedia of science – perhaps

Preface particularly language science. I have tried to follow this philosophy throughout the volume. Specifically, I have sought to present a range of theoretical ideas (as well as more theoryindependent topics), placing them together in such a way as to encourage a mutual sharpening of ideas and insights. To borrow M. M. Bakhtin’s terms (1981), I have not set out to provide a monological source of authoritative discourse. Rather, I have sought to present a heteroglot volume with which readers may interact dialogically (see dialogism and heteroglossia) – hopefully, to produce more intellectually adequate theories later. Toward this end, I have encouraged authors to be open about their own judgments and attitudes. There is a common view that a piece of writing is biased if the speaker frankly advocates one point of view. But, in fact, the opposite is the case. A piece of writing is biased if a speaker acts as though he or she is simply reporting undisputed facts, when in fact he or she is articulating a partisan argument. Being open, dialogical, and multivocal does not mean being bland. Indeed, insight is more likely to be produced through the tension among ideas and hypotheses that are clearly delineated in their differences. Th is is no less true in the language sciences than elsewhere. Indeed, that is one reason why this volume treats language sciences, not the science of language. – Patrick Colm Hogan

WORKS CITED AND SUGGESTIONS FOR FURTHER READING Bakhtin, M. M. 1981. The Dialogic Imagination: Four Essays. Ed. Michael Holquist. Trans. Caryl Emerson and Michael Holquist. Austin: University of Texas Press. Cervantes Saavedra, Miguel de. 1950. The Adventures of Don Quixote. Trans. J. M. Cohen . New York : Penguin. Davidson, Donald. 1984. “On the very idea of a conceptual scheme.” In Inquiries into Truth and Interpretation, 183–98. Oxford: Oxford University Press. Faust, David. 1984. The Limits of Scientific Reasoning. Minneapolis: University of Minnesota Press. Feyerabend, Paul. 1975. Against Method: Outline of an Anarchistic Theory of Knowledge. London: Verso. Hogan, Patrick Colm. 2003. Cognitive Science, Literature, and the Arts: A Guide for Humanists. New York : Routledge. Lakatos, Imre. 1970. “Falsification and the methodology of scientific research programmes.” In Criticism and the Growth of Knowledge, ed. Imre Lakatos and Alan Musgrave, 91–195. Cambridge: Cambridge University Press. Mahoney, Michael. 1977. “Publication prejudices: An experimental study of confi rmatory bias in the peer review system.” Cognitive Therapy and Research 1: 161–75. Peters, Douglas, and Stephen Ceci. 1982. “Peer-review practices of psychological journals: The fate of published articles, submitted again.” Behavioral and Brain Sciences 5.2: 187–95. Putnam, Hilary. 1981. Reason, Truth, and History. Cambridge: Cambridge University Press.

xxi

ACKNOWLEDGMENTS

First of all, I must thank Phil Laughlin, who (inspired by the exemplary MIT Encyclopedia of the Cognitive Sciences) suggested the project initially and invited me to make a more formal proposal. Without Phil’s initial idea and subsequent encouragement, this volume would not now exist. Eric Schwartz took over from Phil, then Simina Calin took over from Eric; both were supportive and helpful, as were the several editorial assistants, most recently April Potenciano, Christie Polchowski, and Jeanie Lee. Regina Paleski and Mark Fox ably shepherded this complex project through the production process; Phyllis Berk worked with devotion on copy editing the manuscript; and Robert Swanson took on the tough job of indexing. The members of the editorial board kindly provided comments on the list of entries and suggested possible authors. They also served as second readers for most of the entries. I am indebted to them all. It is difficult and unrewarding work, but extremely valuable. Some entries were evaluated by specialists not on the editorial board. I am deeply grateful to the following scholars who agreed to read and comment on entries: J. Abutalebi, E. Ahlsén, A. Aikhenvald, S. Anderson, A. Atkin, S. Barker, J. Beall, D. Beaver, H. Bejoint, H. Ben-Yami, A. Berger, D. Bickerton, A. Bilgrami, S. Blackmore, J. Blommaert, C. Bowern, E. Brenowitz, J. Bybee, J. Carroll, T. Deacon, M. DeGraff, J.-L. Dessalles, A. Edgar, C. Elgin, R. Ferrer i Cancho, J. Field, H. Filip, D. Finkelstein, J. Forrester, R. Gibbs, R. Gibson, R. Giora, R. Gleave, K. Gluer-Pagin, M. Goral, M. Hashimoto, J. Heath, D. Herman, R. Hilpinen, J. Hintikka, K. Hoff man, K. Holyoak, P. Hutchinson, J. Hyun, P. Indefrey, M. Israel, K. Johnson, M. Johnson, J. Kane, P. Kay, A. Kibort, S. Kiran, C. Kitzinger, W. Labov, B. Lafford, C. Larkosh, A. Libert, P. Livingston, K. Ludwig, M. Lynch, J. Magnuson, G. Marcus, R. May, J. McGilvray, A. Mehler, S. Mills, D. Moyal-Sharrock, K. Oatley, B. O’Connor, L. Pandit, B. Partee, J. Pennebaker, P. Portner, C. Potts, J. Robinson, S. Rosen, S. Ross, J. Saul, R. Schleifer, M. Shibatani, R. Skousen, S. Small, W. Snyder, M. Solms, F. Staal, P. Stockwell, L. Talmy, H. Truckenbrodt, J. P. Van Bendeghem, W. van Peer, S. Wheeler, and L. Zhang. Thanks to M. Cutter for help with the illustrations. For some time, a graduate assistant, Karen Renner, took care of many secretarial duties. Th is work was supported by the English Department at the University of Connecticut, with some added funding from the University of Connecticut Research Foundation. The support of the English Department was due to the kindness and commitment of our department head, Bob Tilton – without his help, this project would not have been possible. Work for the entry on “Ad Hoc Categories” was supported by National Science Foundation Grant BCS-0212134 and DARPA Contract FA8650–05-C-7256 to Lawrence W. Barsalou. The entry on “Dyslexia” was prepared with support from a British Academy Research Readership to Margaret J. Snowling. Preparation of the manuscript for “Speech Production” was supported by NIDCD A-93 and NIDCD grant DC-03782, both to Haskins Laboratories. Research for Paisley Livingston’s entries benefited from fi nancial support from the Research and Postgraduate Studies Committee of Lingnan University, Hong Kong.

xxiii

1 LANGUAGE STRUCTURE IN ITS HUMAN CONTEXT: NEW DIRECTIONS FOR THE LANGUAGE SCIENCES IN THE TWENTY-FIRST CENTURY William Croft

The science of language in the twenty-fi rst century is likely to expand its scope compared to that of the twentieth century. The twentieth-century science of language focused its attention on the analysis of language structure: the sound system of a language (phonology) and its grammatical system (morphology and syntax). The analysis of linguistic structure, or form, is central to the science of language. After the middle of the twentieth century, however, greater attention was placed on the relationship between language form and its psychological and social context. The analysis of linguistic structure will remain central to the science of language. However, understanding language in context will undoubtedly be a crucial feature of language science in the twenty-fi rst century. Th is essay focuses on the basic principles that have emerged in research on language in its social and cognitive context, the ways that this context constrains language structure and use, and the new directions in research implied by the integration of language structure and context. Th is essay is necessarily selective in the topics covered, and the selection represents a particular way to integrate language form and its context. It also brings together theories that have originated in philosophy, psychology, and sociology, as well as different branches of linguistics. Such effort is necessary in order to treat language as a unitary phenomenon, and also to relate central questions of linguistic analysis to other scientific domains. Language structure cannot be fully understood without situating it with respect to current theories of joint action, social cognition, conceptualization of experience, memory and learning, cultural transmission and evolution, shared knowledge and practice in communities, and demographic processes in human history.

WHY TALK? THE PRAGMATICS OF LANGUAGE Why do we talk? Why does language exist? It is only by answering these questions that we can understand how language fits in its context. The answer is that language plays an essential role in social interaction, fundamentally at the level of joint action between two or more individuals (Clark 1996; Tomasello, 2008). What makes a joint action joint is that it is more than just

the sum of individual actions performed by separate persons; in particular, each individual involved must take into consideration the other individual’s beliefs, intentions, and actions in a way that can be described as cooperative. A shared cooperative activity between two individuals can be defi ned in terms of a set of attitudes held by the cooperating individuals and as a way of carrying out the individual action (Bratman 1992). The attitudes are as follows: (a) Each individual participant intends to perform the joint action. That is, each participant’s intention is not directed simply toward his/her individual action but toward the joint action that is carried out by both participants together. (b) Each participant intends to perform the joint action in accordance with and because of each one’s meshing subplans. That is, each participant’s individual actions are intended to mesh with the other participant’s actions in order to successfully achieve the joint action. (c) Neither participant is coercing the other. (d) Each participant has a commitment to mutual support. That is, each one will help the other to carry out the subplans; each participant is thus responsible for more than just execution of his/her own subplan. (e) All of (a)–(d) are common ground, or shared knowledge between the individuals. The concept of common ground plays a central role in understanding the function of language in social interaction; it is discussed more fully toward the end of this essay. Finally, in addition to these mental attitudes on the part of the participants, there must be mutual responsiveness in action. That is, the participants will coordinate their individual actions as they are executed in order to ensure that they mesh with each other and, hence, that the joint action will be successfully carried out (to the best of their abilities). Coordination is essential in carrying out joint actions successfully, and this is where language plays a central role in joint actions. The social cognitive abilities necessary for shared cooperative activity appear to be unique to humans, providing what Michael Tomasello (2008) calls the social cognitive infrastructure necessary for the evolution of the capacity for modern human language. Other species than humans have a capacity for imitative learning of complex vocalizations (see animal communication and human language). Th is has not been sufficient to lead to the evolution of human-like language among these species. Nonhuman primates have the ability to plan actions and to recognize regularities in behavior of other creatures, enough to manipulate their behavior. These abilities are preconditions for executing complex actions such as joint actions, but they are not sufficient for doing so. Research on primate behavior in natural and experimental settings suggest that some primates even have the ability to recognize conspecifics as beings with intentional states like themselves in some circumstances (Tomasello, 2008; this ability develops in humans only at around nine months of age). Nevertheless, it has not been demonstrated that nonhuman primates have the ability to engage in shared cooperative activity

1

The Cambridge Encyclopedia of the Language Sciences as already defi ned. Tomasello (ibid.) suggests that in particular, helpfulness, Michael Bratman’s condition (d), may be critical to the evolution of the ability to carry out joint actions. The fi nal condition for joint action is that the individual actions must be coordinated in accordance with the shared attitudes of the participants. Any joint action poses coordination problems between the participants (Lewis 1969). Any means that is used to solve the coordination problem on a particular occasion is a coordination device. There are various coordination devices that human beings use to solve the coordination problems of joint actions, of which the simplest is joint attention to jointly salient properties of the environment (Tomasello 1999, 2003). But by far the most effective coordination device is for the participants to communicate with each other: By communicating their mental states, the participants greatly facilitate the execution of any joint action. communication is itself a joint action, however. The speaker and hearer must converge on a recognition of the speaker’s intention by the hearer (see communicative intention; see also cooperative principle). Th is is H. Paul Grice’s defi nition of meaning ([1948] 1989), or Herbert Clark’s informative act (Clark 1992; see the next section). And this joint action poses coordination problems of its own. The essential problem for the joint action of communication is that the participants cannot read each other’s minds. Language is the primary coordination device used to solve the coordination problem of communication, which is in turn used to solve the coordination problem for joint actions in general. Indeed, that is the ultimate purpose of language: to solve the coordination problem for joint actions, ranging from the mundane to the monumental (Clark 1999). Th is fact is essential for understanding the structure of discourse and the linguistic expressions used in it, as Clark (1992, 1996) has shown for many aspects of conversational interaction, and it also accounts for many fundamental properties of linguistic structure.

LANGUAGE, COMMUNICATION, AND CONVENTION A language can be provisionally described as a conventional system for communication (this defi nition is modified later in this section). David Lewis (1969) and Clark (1996, Chapter 5) defi ne convention as follows: (i) A regularity in behavior (ii) that is partly arbitrary (that is, we could have equally chosen an alternative regularity of behavior), (iii) that is common ground in the community, (iv) as a coordination device (v) for a recurrent coordination problem.

butterfly) or a grammatical construction (such as the ModifierHead construction for English noun phrases) emerges as a convention when it becomes a regularly used means for solving the recurrent coordination problem of referring to a specific experience that is to be communicated. Linguistic convention actually operates at two levels: the grammatical level of words and constructions, at which the speaker’s intentions are formulated; and the phonological level of the articulation and perception of the sounds that make up the grammatical units. Th is is the phenomenon described as duality of patterning in language (Hockett 1960). One could imagine in principle that linguistic convention possessed only one level: perceivable sounds (or gestures or written images, depending on the medium), corresponding to part (i) in the defi nition of convention, that directly conveyed the speaker’s intention (the recurrent coordination problem) as a whole, corresponding to part (v) in the defi nition of convention. These exist in interjections with specific functions such as Hello and Thanks. However, most linguistic expressions are complex, consisting of discrete, meaningful units. Complex linguistic expressions evolved for two reasons: First, the number of different speaker intentions to be communicated grew to be indefinitely large; and second, a speaker’s intended message came to be broken down into recurrent conceptual parts that could be recombined to produce the indefi nite variety of messages. Again, one could imagine that each conventional linguistic unit consisted of a unique sound (gesture, image). But languages have distinct meaningful units that are made up of different sequences of the same sounds: bat, sat, Sam, same, tame, time, etc. Th is system has evolved for the same two reasons: the increasing number of meaningful units (even the recurring ones) necessary to convey the indefi nitely large number of speaker intentions, and an ability to break down a sound signal (or gesture, or image) into parts that can be recombined as a sequence of sounds (or gestures or images). Thus, the duality of patterning characteristic of human language has evolved to accommodate the huge number of speaker intentions that people want to convey, and to exploit the facts that intentions can be broken down into recombinable conceptual units and that the medium of expression can be broken down into recombinable units as well. Language is therefore a joint action that operates simultaneously at four levels (Clark 1996). The higher-numbered levels are dependent on the lower-numbered levels; the individual actions of the interlocutors are given in italics: (4) proposing and taking up a joint project (joint action); (3) signaling and recognizing the communicative intention; (2) formulating and identifying the proposition; (1) producing and attending to the utterance.

In other words, conventions can emerge when members of the community have shared knowledge that a certain repeated behavior can act among them as a coordination device for a recurrent coordination problem. Th is defi nition of convention is general: It applies to conventions such as shaking hands (or kissing on the cheek) for greeting, or driving on the right (left) side of the road. The defi nition also applies straightforwardly to language: A string of sounds (i.e., a word or morpheme, such as

2

The highest level corresponds to the illocutionary force in speech-act theory (Austin 1962); the next level to Gricean meaning, or the informative act (Clark 1992); the next level to the propositional act (Searle 1969); and the lowest level to the utterance act (Austin 1962; Searle 1969). Each level enables the level(s) above it, and succeeds only if the level(s) below has been successfully achieved (e.g., one cannot recognize the

Language Structure in Its Human Context communicative intention if one did not pay attention to the utterance produced).

THE INCOMPLETENESS OF CONVENTION The model of language as joint action describes the social cognitive system that must have evolved in the human species for modern human language to have emerged. It describes what appears to be a stable system that led to the emergence of highly complex cooperative activity among humans, namely, what is called society or culture. But it is not a complete picture of the nature of language in social interaction. Linguistic convention can function as a coordination device for communication because there are recurrent coordination problems in communication: People have repeatedly wished to convey similar intentions formulated in similar concepts. Convention, linguistic or otherwise, is a regularity of behavior that emerges in a community or society. But convention must emerge from previous successful communication events where a convention did not previously exist. In other words, there must be a precedent: You and I use a coordination device because we used it before (or observed it used before), and it worked. Following a precedent is a coordination device, but it is not (yet) convention; it is based not on regular behavior that is mutually known in the community but only on previous successful uses that we are aware of (Lewis 1969). Following a precedent cannot be the ultimate root of convention either. It always requires a successfully coordinated communicative act as a precedent. The ultimate coordination device is joint salience: Each participant can assume that in a particular situation, certain features are salient to both participants (Lewis 1969). Joint salience is possible because humans have the social cognitive capacity for joint attention to their environment (Tomasello 2003). Joint attention forms a basis for common ground, as discussed later in this article. Linguistic convention, however, is not perfect; it does not trump or replace the nonconventional coordination devices of precedent and joint salience in the act of communication. Th is is partly because of the kind of conventions found in language, and partly because of the nature of convention itself. Linguistic conventions are incomplete because of the phenomena of indexicality and ambiguity (Clark 1996). A linguistic convention such as hat or find represents a type, but on a particular occasion of use, we often intend to convey a particular token of the category. Thus, I found the hat communicates a particular taking event involving a specific hat. In order to identify which fi nding event and which hat, the interlocutors must rely on joint salience in the context, facilitated in part by the past tense of find and the article the combined with hat, to coordinate successfully on the right fi nding event and the right hat. Linguistic shifters, such as the pronoun I, more explicitly require joint salience, namely, who is the speaker in the context. Proper names denote tokens, but even a proper name such as William Croft may be (and is) used for more than one individual, for example, the contemporary linguist and the English Baroque musical composer. Most words are also highly ambiguous; that is, the same regularity of behavior is used as a coordination device to solve

different recurrent coordination problems. For example, patient is ambiguous between the linguistic semantic role (The patient in sentence 25 is Roland ) and a role in the domain of medicine (The patient in room 25 is Roland ). Linguistic convention alone cannot tell which meaning is intended by the speaker. Only joint salience, provided in the example sentences by the meanings of the other words and the broader context of conversation, will successfully solve the coordination problem of what is meant by patient. Indexicality and ambiguity are so pervasive in language that no utterance can be successfully conveyed without recourse to nonconventional coordination devices. But convention itself is also incomplete. Th is is because every situation being communicated is unique and can be construed as the recurrence of different coordination problems. The simplest example of this phenomenon is that different words can be used to describe the current situation, each representing a different construal of the current situation in comparison to prior situations. For example, one can refer to an individual as the prime minister, Tony Blair, the Labour Party leader, my friend, that guy, he, etc.; each expression construes reference to the current person as the recurrence of a different coordination problem. The need to use nonconventional coordination devices as well as linguistic convention in communication is not generally a problem for successful joint actions by cooperative human beings. However, in some contexts, successful coordination is quite difficult. For example, scholarly discourse on abstract theoretical concepts often leads to alternative construals of what is intended by particular scholars. What do we take Plato to have meant? Th is changes over time and across persons. Alternative construals, not always accurately described as “misunderstandings,” occur in more everyday circumstances as well, as readers can verify for themselves. In addition, human beings are not always cooperative. The complexity of language as joint action here leaves open many possible means of language abuse. For example, lying abuses linguistic convention in its role of helping coordinate a shared cooperative activity, namely, coming to a shared belief. Other types of language abuse exploit nonconventional coordination devices. For example, in one lawsuit, the courts ordered a government agency to destroy certain documents, intending the term to denote their information content; the agency destroyed the documents, that is, the physical objects, after making copies of them (Bolinger 1980). Here, the ambiguity of documents requires recourse to joint salience, but the agency abused this nonconventional coordination device (the lawsuit was about privacy of information). Finally, the fact that a current situation can be construed as an instance of different recurrent coordination problems leads to alternative framings of the situation, such as referring to an entity as a fetus or an unborn baby. These alternative framings bias the conceptualization of the current situation in ways that invite certain inferences and courses of action, rather than others.

THE LINGUISTIC SYSTEM IN CONTEXT In the preceding sections, language is described as a conventional system for communication, and the role of convention

3

The Cambridge Encyclopedia of the Language Sciences in language and of language in communication was discussed. In this section, the linguistic system is described in broad outline. Linguistic structure has been intensively studied over the past century ever since Ferdinand de Saussure inaugurated the modern analysis of linguistic structure, Structuralism (Saussure [1916] 1966). Th is section focuses on those aspects of linguistic structure that are generally agreed upon and shows the extent to which they emerge from the principles that have been presented in the preceding section. The most fundamental structuralist principle is the centrality of the linguistic sign or symbol, that is, the notion that language pairs form and meaning, and that particular linguistic forms convey particular meanings. Th is principle fits directly with the defi nition of convention. The regularity in behavior in part (i) of the defi nition of convention is the expression of a linguistic form by a speaker; the recurrent coordination problem in part (v) of the defi nition is the communication of a meaning between the interlocutors. Also central to the structural analysis of language is the arbitrariness of the linguistic sign. That is, arbitrariness exists in the particular form and meaning that are paired. Th is conforms with part (ii) of the defi nition of convention, namely, that the convention is partly arbitrary. Arbitrariness is usually defi ned in structuralist analysis as the principle that one cannot entirely predict the form used from the meaning that is intended. From a communicative point of view, arbitrariness means that another choice could have served approximately equally well. For example, the choice of producing the string of sounds butterfly for a particular meaning could have been replaced with the choice of producing the string of sounds Schmetterling – a choice made by members of the German speech community. Two different choices are communicatively equivalent in that neither choice is preferred for the meaning intended – and that is usually because the choice of one expression over the other is arbitrary in the structuralist sense. Another principle that can be traced back to Saussure is the distinction between the paradigmatic and syntagmatic contrast of linguistic units. In a complex (multiword or multimorpheme) grammatical construction, such as The cat sat on the mat, each word enters into two different types of contrast. For example, the fi rst word the contrasts with the word cat in that the ’s role in the construction (determiner) contrasts with cat ’s role (head noun). Th is is a syntagmatic contrast. But the also contrasts with another possible fi ller of the same role in the construction, such as a in A cat sat on the mat ; and cat contrasts with hamster, parakeet, etc. in the same way. These are paradigmatic contrasts. More recent grammatical theories represent paradigmatic contrast in terms of a set of elements belonging to a grammatical category. Thus, the and a belong to the category determiner, and cat, hamster, parakeet, etc. belong to the category noun. Syntagmatic contrasts are represented by contrasting roles in the syntactic structure or constructions used in the utterance. For example, the determiner category is functioning as a modifier of the noun category in a noun phrase construction. Finally, the syntagmatic–paradigmatic distinction also applies to phonology (sound structure): Paradigmatic contrast is represented

4

by phonological categories, and syntagmatic contrasts by the phonological structure of words and larger prosodic units. The syntagmatic–paradigmatic distinction is the most basic way to describe the fact that the linguistic system allows a (re-) combination of meaningful units in different ways. The adaptive motivation for the emergence of such a communication system was described previously: The number of intentions to be communicated is so great that a set of simple (atomic) symbols will not suffice, but experience is such that it can be broken down into recurrent parts for which conventional linguistic expressions can develop. The same motivations gave rise to the syntagmatic–paradigmatic distinction in phonology as well. Paradigmatic principles of structure in grammar and phonology are represented in terms of linguistic categories, phonological and grammatical. These abstract linguistic categories can be mapped onto the substantive categories of the actual phonetic realization (for phonology) and of utterance meaning (for grammar). Linguistic typology (Comrie 1989; Croft 2003), which takes a cross-linguistic perspective on grammatical analysis, has demonstrated that the ways in which phonological categories are mapped onto phonetic space, and grammatical or lexical categories are mapped onto conceptual space, are not unlimited. For example, phonetic similarities and conceptual similarities constrain the structure of phonological and grammatical categories, respectively. Syntagmatic principles of structure are represented in various ways, but all such representations reflect another basic principle, the hierarchical organization of the structure of utterances. Sentences are organized in a hierarchical structure, representing groupings of words at different levels. So The cat sat on the mat is not just a string of roles that contrast syntagmatically, as in [Determiner Noun Copula Preposition Determiner Noun]. Instead, it is a set of nested groupings of words: [[Determiner Noun] [Copula] [Preposition [Determiner Noun]]]. The nested groupings are frequently represented in a variety of ways, such as the syntactic trees of phrase (constituent) structure analysis. They can also be represented as dependency diagrams (for example, the determiner is related to the noun as its modifier, which in turn is related to the copula as its subject), and representations combining constituency and dependency also exist. The structure of a construction often appears to be motivated, though not entirely predicted, by the structure of the meaning that it is intended to convey. For example, the syntactic groupings in [[The cat] is [on [the mat]]] are motivated semantically; the in the cat modifies cat semantically as well as syntactically (indicating that the cat’s identity is known to both speaker and hearer). The (partial) motivation of syntactic structure by its meaning is captured by general principles in different theories. These principles can be described as variants of the broader principle of diagrammatic iconicity (Peirce 1932): roughly, that the abstract structure of the linguistic expression parallels the abstract structure of the meaning intended, to a great extent. It is difficult to evaluate the structure of meaning independently of the structure of linguistic form. However, different speech communities settle on a similar range of constructions to express the same complex meaning – the regularities discovered in linguistic typology (see, for example, the studies

Language Structure in Its Human Context published in Typological Studies in Language and the Oxford Studies in Typology and Linguistic Theory). Th is fact suggests that there are regularities in the meaning to be conveyed that are then reflected in the grammatical constructions used to express them.

GRAMMAR AND THE VERBALIZATION OF EXPERIENCE The preceding sections have described the general context of language use and the basic principles of language structure. The grammars of particular languages conform to the basic principles of language structure in the preceding section. But the grammars of particular languages, while diverse in many ways, are similar to a much greater degree than would be predicted from the general principles in the preceding section, or even the context of language use described in the earlier sections. For example, all languages have structures like clauses in which some concept (prototypically an action concept, usually labeled a verb) is predicated on one or more concepts that are referred to (prototypically an object or person, usually labeled a noun). The noun-like expressions are in turn organized into phrases with modifiers. Clauses are related to each other by varying degrees of grammatical integration. Certain semantic categories are repeatedly expressed across languages as grammatical inflections or “function words” (e.g., articles, prepositions, auxiliaries) that combine with the major lexical categories of words in sentences. These universal patterns in grammar are attributable to the way that experience is verbalized by human beings. The fundamental problem of verbalization is that each experience that a speaker wishes to verbalize is a unique whole. But a linguistic utterance is unlike an experience: An utterance is broken down into parts, and these parts are not unique; they have been used before in other utterances. (Th is latter point is the fact of convention; a particular linguistic form is used regularly and repeatedly for a recurrent coordination problem.) The process by which the unique whole of experience is turned into a linguistic utterance made up of reusable parts has been described by Wallace Chafe (1977). The fi rst step is that the speaker subchunks the experience into smaller parts, each also a unique Gestalt similar in this way to the original experience. The subchunking process may be iterated (in later work, Chafe emphasizes how consciousness shifts from one chunk to another in the experience to be verbalized). A subchunk of the experience is then propositionalized; this is the second step. Propositionalizing involves breaking up an experience by extracting certain entities that are (at least prototypically) persistent, existing across subchunks. These entities are the referents that function as arguments of the predicate; the predicate is what is left of the subchunk after the arguments have been separated. Propositionalizing therefore breaks down the experience into parts – arguments and the predicate – that are not of the same type as the original experience (i.e., not a Gestalt). Once the whole has been broken down into these parts, the parts must be categorized, that is, assigned a category that relates the parts of the current experience to similar parts of prior experiences. Categorizing is the third step in the verbalization process. These categories are what are expressed by

content words, such as nouns and verbs. In this way, the speaker has transformed the unique whole of the original experience into parts that can be expressed by language. Th is is not the end of the verbalization process. Content words denote only general categories of parts of the experience to be verbalized. In order to communicate the original experience, the speaker must tie down the categories to the unique instances of objects, events, and so forth in the experience, and the speaker must assemble the parts into a structure representing the original whole that the speaker intends to verbalize. That is to say, corresponding to the categorizing step in verbalizing the parts of the experience, there is a particularizing step that indicates the unique parts; and corresponding to the steps of propositionalizing and subchunking are integrative steps of structuring and cohering, respectively (Croft 2007). These latter three steps give rise to grammar in the sense of grammatical constructions, inflections, and particles, and the semantic commonalities among grammatical categories across languages. The particularizing step takes a category (a type) and selects an instance (token) or set of tokens, and also identifies it by situating it in space and time. For object concepts, selecting can be accomplished via the inflectional category of number, and via the grammatical categories of number and quantification (three books, an ounce of gold ). For action concepts, selecting is done via grammatical aspect, which helps to individuate events in time (ate vs. was eating), and via agreement with subject and/or object, since events are also individuated by the participants in them (I read the paper and She read the magazine describe different reading events). Objects and events can be situated in space via deictic expressions and other sorts of locative expressions (this book, the book on the table). Events and some types of objects can be situated in time via tense and temporal expressions (I ate two hours ago; ex-mayor). Events and objects can also be situated relative to the mental states of the interlocutors: The article in the book indicates that the particular object is known to both speaker and hearer, and the modal auxiliary in She should come indicates that the event exists not in the real world but in the attitude of obligation in the mind of the speaker. The structuring step takes participants and the predicated event in a clause and puts them together, reassembling the predicate and the argument(s) into the subchunk from which they were derived by propositionalizing. Grammatically this is a complex area. It includes the expression of grammatical relations in what is called the argument structure of a predicate, so that She put the clock on the mantle indicates which referent is the agent (the subject), which the thing moved (the object), and which the destination of the motion (the prepositional phrase). But it also includes alternative formulations of the same event, such as The clock was put on the mantle (the passive voice construction) and It was the mantle where she put the clock (a cleft construction). The alternative constructions function to present the information in the proposition in different ways to the hearer, depending on the way the discourse is unfolding; they are referred to as information structure or discourse function. Finally, the cohering step takes the clauses (subchunks) and reassembles them into a whole that evokes the original whole

5

The Cambridge Encyclopedia of the Language Sciences experience for the hearer. Th is step can be accomplished by various clause-linking devices, including subordination of various kinds, coordination, and other clause-linking constructions found in the world’s languages. Coherence of clauses in discourse is also brought about by discourse particles and reference tracking, that is, grammatical devices, such as pronouns or ellipsis, which show that an event is related to another event via a shared participant (Harry filled out the form and _ mailed it to the customs office). The three steps of particularizing, structuring, and cohering result in a grammatical structure that evokes a reconstituted version of the original unique whole. These six steps in verbalization are not necessarily processed sequentially or independently. The steps in the verbalization process are dependent on the grammatical resources available in the language, which constrain the possibilities available to the speaker. For example, when a speaker takes a subchunk and extracts participants from it, there must be a construction available in the language to relate the participants to the predicate, as with put in the earlier example. Thus, subchunking must be coordinated with propositionalizing and structuring. Also, the steps may not be overtly expressed by grammatical inflections or particles. For example, The book fell does not overtly express the singular number of book, or that the event is situated in the real world rather than a nonreal mental space of the speaker. Finally, the reconstituted experience evoked by the linguistic utterance is not the same as the unique whole with which the speaker began. The cognitive processes humans use in verbalization do not simply carry out one or more of the six steps described. They also conceptualize the experience in different ways, depending on the speaker’s choices. These choices range from the subtle difference between describing something as leaves or foliage, or the more dramatic framing differences between fetus and unborn baby referred to previously. There are a wide range of conceptualization processes or construal operations that have been identified in language (see, e.g., Langacker 1987; Talmy 2000). The construal operations can be accounted for by processes familiar from cognitive psychology: attention, comparison, perspective, and Gestalt (Croft and Cruse 2004, Chapter 4). These psychological processes are part of the meaning of all linguistic units: words, inflections, and constructions. As a consequence, every utterance presents a complex conceptualization of the original experience that the speaker intends to verbalize for the hearer. The conventionalized conceptualizations embodied in the grammatical resources of a language represent cultural traditions of ways to verbalize experience in the speech community.

VARIATION AND THE USAGE - BASED MODEL One of the results of recent research on language structure and language use is the focus on the ubiquity of variation in language use, that is, in the verbalization of experience and its phonetic realization. The ubiquity of variation in language use has led to new models of the representation of linguistic knowledge in the mind that incorporate variation as an essential characteristic of language. These models are more developed in phonetics and phonology. The phonological model is described

6

fi rst and then recent proposals to apply it to grammar (syntax and semantics) are examined. One of the major results of instrumental phonetics is the discovery that phonetic variation in speech is ubiquitous. Variation in the realization of phonemes is found not just across speakers but also in the speech of a single speaker. There are at least two reasons why such variation in the speech signal would exist. Presumably, the level of neuromuscular control over articulatory gestures needed for identical (invariant) productions of a phoneme is beyond a speaker’s ability. At least as important, the variation in the speech signal does not prevent successful communication (or not enough of the time to lead to the evolution of even fi ner neuromuscular control abilities in humans). There is evidence, moreover, that the mental representation of phonological categories includes the representation of individual tokens of sounds and the words that contain them. Speakers retain knowledge of fi ne-grained phonetic detail (Bybee 2001; Pierrehumbert 2003). Also, there are many frequency effects on phonological patterns (Bybee 2001). For example, higher-frequency forms tend to have more reduced phonetic realizations of phonemes than lower-frequency forms. Finally, human beings are extremely good pattern detectors from infancy on into adulthood. Infants are able to develop sensitivity to subtle statistical patterns of the phonetic signals they are exposed to. Th is type of learning, which occurs without actively attending to the stimulus or an intention to learn is called implicit learning (Vihman and Gathercole, unpublished manuscript). It contrasts with explicit learning, which takes place under attention from the learner – particularly joint attention between an infant learning language and an adult – and is involved in the formation of categories and symbolic processing. There is neuroscientific evidence that implicit learning is associated with the neocortex and explicit learning with the hippocampus (ibid.). A number of researchers have proposed a usage-based or exemplar model of phonological representation to account for these patterns (Bybee 2001; Pierrehumbert 2003). In this model, phonological categories are not represented by specific phonetic values for the phoneme in the language, but by a cluster of remembered tokens that form a density distribution over a space of phonetic parameters. The phonetic space represents the phonetic similarities of tokens of the phonological category. Th is model includes properties of implicit learning (the cluster of individual tokens) and explicit learning (the labeling of the density distribution as representing tokens of, say, /e/ and not /i/). Consolidation of token memories also takes place – individual tokens decay in memory, highly similar tokens are merged, and the distribution of tokens can be restructured – but new tokens are constantly being incorporated into the representation and influencing it. Marilyn Vihman and S. Kunnari (2006) propose three types of learning for an exemplar model. First, there is an initial implicit learning of statistical regularities of the sensory input. Second, explicit learning of linguistic categories, such as the words that are templates containing the sound segments, takes place. Finally, a second layer of implicit learning of statistical

Language Structure in Its Human Context regularities gives rise to probability distributions for each linguistic phonological and lexical category. The result of this last layer of learning is the exemplar or usage-based model described by Janet Pierrehumbert and Joan Bybee. The application of the usage-based/exemplar model to grammar is more complex. Most research in this area has compared the range of uses of a particular word or grammatical construction. However, this does not represent the process of language production (that is, verbalization), analogous to the phonetic variation found in the production of phonemes. Studies of parallel verbalizations of particular scenes demonstrate that variation in the verbalization of the same scene by speakers in similar circumstances is ubiquitous, much like the phonetic realization of phonological categories (Croft 2010). There is also substantial evidence for frequency effects in grammar. For example, English has a grammatical category of auxiliary verb that has distinctive syntax in negation (I ca n’t sing vs. I didn’t sing), questions (Can he sing? vs. Did he sing?). These syntactic patterns are actually a relic of an earlier stage of English when word order was freer; it has survived in the auxiliaries of modern English because of their higher token frequency (Bybee and Thompson 1997), as well as their semantic coherence. Frequency plays a central role in the historical process of Grammaticalization (Hopper and Traugott 2003), in which certain constructions develop a “grammatical” function (more precisely, they are recruited to serve the particularizing, structuring, and cohering steps of the verbalization process). Part of the grammaticalization process is that the construction increases in frequency; it therefore undergoes grammatical and phonological changes, such as fi xation of word order, loss of syntactic flexibility, and phonetic reduction (Bybee 2003). A well-known example is the recruitment of the go + Infi nitive construction for the future tense: She is going (to Sears) to buy a food processor becomes future She’s going to buy a food processor, with no possibility of inserting a phrase between go and the infi nitive, and is fi nally reduced to She’s gonna buy a food processor. Finally, early syntactic acquisition is driven by implicit learning of patterns in the linguistic input (Tomasello 2003). The process of syntactic acquisition is very gradual and inductive, involving an interplay between detection of statistical regularities and the formation of categories that permit productive extension of grammatical constructions. Children occasionally produce overregularization errors, and these are also sensitive to frequency (more frequent forms are more likely to be produced correctly, and less frequent forms are more likely to be subject to regularization). A usage-based model of grammatical form and meaning is gradually emerging from this research. An exemplar model of grammatical knowledge would treat linguistic meanings as possessing a frequency distribution of tokens of remembered constructions used for that meaning. Those constructions would be organized in a multidimensional syntactic space organized by structural similarity (e.g., Croft 2001, Chapter 8) and whose dimensions are organized by the function played by the construction in the verbalization process. The meanings of constructions are themselves organized in a conceptual space whose structure can be inferred empirically via cross-

linguistic comparison of the meanings expressed by grammatical categories and constructions. The typological approach to grammar has constructed conceptual spaces for a number of semantic domains using techniques such as the semantic map model (see Haspelmath 2003 for a survey of recent studies) and multidimensional scaling (Croft and Poole 2008). To sum up, the usage-based/exemplar model can be applied to both phonological patterns in words and grammatical structures in constructions. A speaker’s knowledge of language is the result of the interplay between two learning processes. One learning process is the tallying of statistical regularities of tokens of words and constructions with a particular phonetic realization, performing a particular communicative act in a specific social interaction. The other is the organization of these tokens into categories and the formation of generalizations that allow the reuse or replication of these grammatical structures to solve future coordination problems in communication.

VARIATION AND CHANGE: AN EVOLUTIONARY APPROACH The view of language described in the preceding sections roots both language structure and a speaker’s linguistic knowledge in the individual acts of linguistic behavior that a speaker has engaged in and will engage in. It is a dynamic view of language in that linguistic behavior is essentially historical: a temporal series of utterances, each one linked to prior utterances as repeated behavior to solve recurrent coordination problems in social interaction. Each member of a speech community has a history of his or her participation in linguistic events, either as speaker or hearer. Th is history is remembered in the exemplarbased representation of that member’s linguistic knowledge, but also consolidated and organized in such a way that each unique experience is broken down and categorized in ways that allow for reuse of words and constructions in future communication events. Each time a speaker produces an utterance, he or she replicates tokens of linguistic structures – sounds, words, and constructions – based on the remembering of prior tokens of linguistic structures, following the principles of convention and verbalization described earlier. However, the replication process is never perfect: Variation is generated all of the time, as described in the preceding section. The variation generated in the process of language use can be called fi rst-order variation. Variation in replication is the starting point for language change. Language change is an instance of change by replication (rather than inherent change); change by replication is the domain of an evolutionary model of change (Hull 1988; Croft 2000). Change by replication is a two-step process. The fi rst step is the generation of variation in replication. Th is requires a replicator and a process of replication by which copies are produced that preserve much of the structure of the original. In biological evolution, the canonical replicator is the gene, and the process of replication takes place in meiosis (which in sexual organisms occurs in sexual reproduction). Copies of the gene are produced, preserving much of the structure of the original gene. Variation is generated by random mutation processes and by recombination in sexual reproduction.

7

The Cambridge Encyclopedia of the Language Sciences In language, replication occurs in language use. The replicators are tokens of linguistic structures in utterances (called linguemes in Croft 2000). These tokens are instances of linguistic behavior. The process of language change is therefore an example of cultural transmission, governed by principles of evolutionary change. The replication process in language change is governed by the principle of convention. As we have seen in the preceding section, variation is generated in the process of verbalization, including the recombination of linguistic forms. Th is represents innovation in language change. Firstorder variation is the source of language change. Experiments in phonological perception and production indicate that “sound change is drawn from a pool of synchronic variation” (the title of Ohala 1989). Indeterminacy in the interpretation of a complex acoustic signal can lead to reanalysis of the phonological categories in that signal. Likewise, it appears that grammatical change is also drawn from a pool of synchronic variation, namely, variation in verbalization. There is an indeterminacy in the understanding of the meaning of a word or construction because we cannot read each other’s minds, our knowledge of linguistic conventions differs because we have been exposed to different exemplars, and every situation is unique and can be construed in different ways. Th is indeterminacy gives rise to variation in verbalization (Croft 2010), and can lead to the reanalysis of the mapping of function into grammatical form (Croft 2000). The second step of the evolutionary process is the selection of variants. Selection requires an entity other than the replicator, namely, the interactor. The interactor interacts with its environment in such a way that this interaction causes replication to be differential (Hull 1988). In biological evolution, the canonical interactor is the organism. The organism interacts with its environment. In natural selection, some organisms survive to reproduce and therefore replicate their genes while others do not; this process causes differential replication. In language, selection occurs in language use as well. The interactor is the speaker. The speaker has variant linguistic forms available and chooses one over others based on his or her environment. In language, the most important environmental interaction is the social relationship between speaker and hearer and the social context of the speech event. Th is is, of course, the realm of sociolinguistics (see, e.g., Labov 2001, and the following section). Selection goes under the name of propagation in language change. Selection (propagation) is a function of the social value that variants acquire in language use. First-order variation does not have a social value. Socially conditioned variation is second-order variation. Once a variant is propagated in a speech community, it can lead to third-order variation, that is, variation in linguistic conventions across dialects and languages. Linguistic diversity is the result of language change. The evolutionary model requires a revision to the defi nition of language offered near the beginning of this essay. In the evolutionary model, a language is a population of utterances, the result of the employment of linguistic conventions in a speech community. The linguistic system is the result of the ways in which speakers have consolidated the uses of language in which they have participated into their knowledge of the

8

conventions of the speech community. Each speaker’s systematic knowledge of his or her language is different, because of differences in the range of language use to which each speaker is exposed.

SPEECH COMMUNITIES AND COMMON GROUND Language in this revised sense is the product of a speech community: the utterances produced by communicative interactions among speakers. A speech community is defi ned by its social interactions involving language: Members of the speech community communicate with one another, and the community is defi ned by communicative isolation from other communities. Communicative isolation is relative, of course, and in fact the structure of human speech communities is far more complex than the structure of biological populations. Two related phenomena serve to defi ne communities: common ground and shared practice. Common ground plays an essential role in defi ning joint action and convention, both central to understanding the nature of language. Common ground consists of knowledge, beliefs, and attitudes presumed by two or more individuals to be shared between them. Common ground can be divided into two types: personal common ground and communal common ground (Clark 1996, Chapter 4). Personal common ground is shared directly in face-to-face interaction by the persons. Personal common ground has two bases. The fi rst is the perceptual basis: We share knowledge of what is in our shared perceptual field. The perceptual basis is provided by virtue of joint attention and salience, as mentioned earlier. A shared basis for common ground has the following properties: The shared basis provides information to the persons involved that the shared basis holds; the shared basis indicates to each person that it provides information to every person that the shared basis holds; and the shared basis indicates the proposition in the common ground (Clark 1996, 94). A basis for common ground varies in how well it is justified; hence, we may not always be certain of what is common ground or not. The second basis for personal common ground is a discourse basis. When I report on situations I have experienced to you in conversation, and vice versa, these become part of our personal common ground. Although we did not experience them perceptually together, we did experience the reporting of them linguistically together. The discourse basis thus involves joint attention (to the linguistic signal), as well as the common ground of a shared language. The discourse basis and the perceptual basis both require direct interaction by the interlocutors. They correspond to social networks, which are instrumental in language maintenance and change (Milroy 1987). The other type of common ground is communal common ground. Communal common ground is shared by virtue of common community membership. A person can establish common ground with a stranger if they both belong to a common community (e.g., Americans, linguists, etc.). Some communities are quite specialized while other communities are very broad and even all-encompassing, such as the community of human beings in this world, which gives rise to the possibility of communication in the fi rst place.

Language Structure in Its Human Context Clark argues that the basis of communal common ground is shared expertise. Étienne Wenger, on the other hand, defi nes communities of practice in terms of shared practice: Individuals engage in joint actions together, and this gives them common ground and creates a community (Wenger 1998). Wenger’s defi nition of a community of practice, therefore, requires faceto-face interaction, like personal common ground. However, shared practice can be passed on as new members enter the community and share practice with remaining current members. Th is is cultural transmission and can lead to individuals being members of the same community through a history of shared practice, even if they do not interact directly with every other member of the community. Since communities are defi ned by shared practice, and human beings engage in a great variety of joint actions with different groups of people, the community structure of human society is very complex. Every society is made up of multiple communities. Each person in the society is a member of multiple communities, depending on the range of shared activities he or she engages in. The different communities have only partially overlapping memberships. As a consequence, the structure of a language is equally complex. A linguistic structure – a pronunciation, a word, a construction – is associated with a particular community, or set of communities, in a society. A pronunciation is recognized as an accent characteristic of a particular community. Words will have different meanings in different communities (e.g., subject is a grammatical relation for linguists but a person in an experiment for psychologists). The same concept will have different forms in different communities (e.g., Zinfandel for the general layperson, Zin to a wine aficionado). Thus, a linguistic convention is not just a symbol – a pairing of form and meaning – but includes a third part, the community in which it is conventional. Th is is part (iii) of the defi nition of convention given in an earlier section. Finally, each individual has a linguistic repertoire that reflects his or her knowledge and exposure to the communities in which he or she acts. The choice of a linguistic form on the part of a speaker is an act of identification with the community that uses it. Th is is the chief mechanism for selection (propagation) in language change: Ihe propagation of variants reflects the dynamics of social change. More recent work in sociolinguistics has argued that linguistic acts of social identity are not always passive: Individuals institute linguistic conventions to construct an identity as well as to adopt one (Eckert 2000).

LANGUAGE DIVERSITY AND ITS ENDANGERMENT Variation in language can lead to language change if it is propagated through a speech community. Social processes over human history have led to the enormous linguistic diversity we fi nd today – a diversity that newer social processes also threaten. The basic social process giving rise to linguistic diversity is the expansion and separation of populations into distinct societies. As groups of people divide for whatever reason, they become communicatively isolated, and the common language that they once spoke changes in different directions,

leading to distinct dialects and eventually to mutually unintelligible languages. Th is ubiquitous demographic process is reflected in the family trees of languages that have been constructed by linguists working on genetic classification. These family trees allow for the possibility of reconstructing not just protolanguages but also the underlying social processes that are traced in them. Even sociolinguistic situations that obscure family trees leave linguistic evidence of other social processes. Extensive borrowing indicates a period of intensive social contact. Difficulty in separating branches of a linguistic family tree indicates an expansion through a new area but continued low-level contact between the former dialects. These can be seen in the dialect continua found in much of Europe, where the Romance, Germanic, and Slavic peoples expanded over a mostly continuous terrain (Chambers and Trudgill 1998). Shared typological (structural) traits may be due to intimate contact between languages with continued language maintenance, or to a major language shift by a social group, resulting in a large proportion of non-native speakers at one point in a language’s history. The spread of human beings across the globe led to the creation of a huge number of distinct societies whose languages diverged. The number of distinct languages that have survived until the beginning of the twenty-fi rst century is about 6,000. Most linguists generally accept the hypothesis that modern human language evolved just once in human history, probably no later than 70,000 to 100,000 years ago. So in principle, all modern human languages may have originated in a single common ancestor. Tracing back the actual lineages of contemporary languages deep into human prehistory appears to be extremely difficult, if not impossible. Nevertheless, there is no doubt that contemporary linguistic diversity is extremely ancient in human history. What we can discover about linguistic history by the comparison of existing languages can potentially shed important light on human history and prehistory. There are linguistic descriptions of a small proportion of existing human languages, though descriptive work has increased and the overall quality of descriptions has improved dramatically, thanks to advances in linguistic science throughout the twentieth century. It would be safe to say that the diversity of linguistic structure, and how that structure is manifested in phonetic reality on the one hand and in the expression of meaning on the other, is truly remarkable and often unexpected. Many proposed universals of language have had to be revised or even abandoned as a consequence, although systematic analysis of existing linguistic descriptions by typologists have revealed many other language universals that appear to be valid. Linguistic diversity has revealed alternative ways of conceptualizing experience in other societies, as well as alternative methods of learning and alternative means for communication for the accomplishment of joint actions. But the single most important fact about the diversity of human language is that it is severely endangered. Of the 6,000 different languages extant today, 5,000 are spoken by fewer than 100,000 people. The median number of speakers for a language is only 6,000 (Crystal 2000). Many languages are no longer spoken by children in the community, and therefore will go extinct in another generation. The loss for the science

9

The Cambridge Encyclopedia of the Language Sciences of language, and more generally for our understanding of human history, human thought, and human social behavior, is immense. But the loss is at least as great for the speakers themselves. Language use is a mode of social identity, not just in terms of identifying with a speech community but as the vehicle of cultural transmission. The loss of languages, like other linguistic phenomena, is a reflection of social processes. The most common social processes leading to language loss are disruption, dislocation, or destruction of the society (language loss rarely occurs via genocide of its speakers). The enormous consequences of language loss has led to a shift in linguistic fieldwork from mere language description and documentation to language revitalization in collaboration with members of the speech community. But reversing language shift ultimately requires a change in the social status of the speech community in the local and global socioeconomic system.

SUMMARY The scientific study of language in its pragmatic, cognitive, and social context beginning in the latter half of the twentieth century is converging on a new perspective on language in the twenty-fi rst century. Linguistic conventions coordinate communication, which in turn coordinates joint actions. The fragility of social interaction by individuals leads to creativity, variation, and dynamism in the verbalization and vocalization of language. Individual linguistic knowledge (the linguistic system) reflects the remembered history of language use and mediates processes of language change. The continually changing structure of society, defi ned by common ground emerging from shared practices (joint actions), guides the evolution of linguistic conventions throughout its history. Human history in turn has spawned tremendous linguistic diversity, which reflects the diversity of human social and cognitive capacity. But the unchecked operation of contemporary social forces is leading to the destruction of speech communities and the mass extinction of human languages today. WORKS CITED AND SUGGESTIONS FOR FURTHER READING Austin, J. L. 1962. How to Do Things with Words. Cambridge: Harvard University Press. Bolinger, Dwight. 1980. Language, the Loaded Weapon. London: Longmans. Bratman, Michael. 1992. “Shared cooperative activity.” Philosophical Review 101: 327–41. Bybee, Joan L. 2001. Phonology and Language Use. Cambridge Cambridge University Press. ———. 2003. “Mechanisms of change in grammaticalization: The role of frequency.” In Handbook of Historical Linguistics, ed. Brian Joseph and Richard Janda, 602–23. Oxford: Blackwell. Bybee, Joan L ., and Sandra A. Thompson. 1997. “Th ree frequency effects in syntax.” In Proceedings of the 23rd Annual Meeting of the Berkeley Linguistics Society, ed. Matthew L. Juge and Jeri O. Moxley, 378–88. Berkeley : Berkeley Linguistics Society. Chafe, Wallace. 1977. “The recall and verbalization of past experience.” In Current Issues in Linguistic Theory, ed. Peter Cole, 215–46. Bloomington: Indiana University Press. Chambers, J. K., and Peter Trudgill. 1998. Dialectology. 2d ed. Cambridge: Cambridge University Press.

10

Clark , Herbert H. 1992. Arenas of Language Use. Chicago and Stanford: University of Chicago Press and the Center for the Study of Language and Information. ———. 1996. Using Language. Cambridge: Cambridge University Press. Clark , Herbert H.. 1999. “On the origins of conversation.” Verbum 21: 147–61. Comrie, Bernard. 1989. Language Universals and Linguistic Typology. 2d ed. Chicago: University of Chicago Press. Croft, William. 2000. Explaining language change: An evolutionary approach. Harlow, Essex : Longman. ———. 2001. Radical Construction Grammar: Syntactic Theory in Typological Perspective. Oxford: Oxford University Press. ———. 2003. Typology and Universals. 2d ed. Cambridge: Cambridge University Press. ———. 2007. “The origins of grammar in the verbalization of experience.” Cognitive Linguistics 18: 339–82. ———. 2010. “The origins of grammaticalization in the verbalization of experience.” Linguistics 48: 1–48. Croft, William, and D. Alan Cruse. 2004. Cognitive Linguistics. Cambridge: Cambridge University Press. Croft, William, and Keith T. Poole. 2008. “Inferring universals from grammatical variation: Multidimensional scaling for typological analysis.” Theoretical Linguistics 34: 1–37. Crystal, David. 2000. Language Death. Cambridge: Cambridge University Press. Eckert, Penelope. 2000. Linguistic Variation as Social Practice: The Linguistic Construction of Identity in Belten High . Oxford: Blackwell. Grice, H. Paul. [1948] 1989. “Meaning.” In Studies in the Way of Words, 213–23. Cambridge: Harvard University Press. Haspelmath, Martin. 2003. “The geometry of grammatical meaning: Semantic maps and cross-linguistic comparison.” In The New Psychology of Language. Vol. 2. Ed. Michael Tomasello, 211–42. Mahwah, NJ: Lawrence Erlbaum Associates. Hockett, Charles F. 1960. “The origin of speech.” Scientific American 203: 88–96. Hopper, Paul, and Elizabeth Traugott. 2003. Grammaticalization. 2d ed. Cambridge: Cambridge University Press. Hull, David L. 1988. Science as a Process: An Evolutionary Account of the Social and Conceptual Development of Science. Chicago: University of Chicago Press. Labov, William. 2001. Principles of Linguistic Change. Vol. 2. Social Factors. Oxford: Blackwell. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar. Vol. 1. Theoretical Prerequisites. Stanford: Stanford University Press. Lewis, David. 1969. Convention. Cambridge, MA : MIT Press. Milroy, Lesley. 1987. Language and Social Networks. 2d ed. Oxford: Basil Blackwell. Ohala, John. 1989. “Sound change is drawn from a pool of synchronic variation.” In Language Change: Contributions to the Study of its Causes, ed. Leiv Egil Breivik and Ernst Håkon Jahr, 173–98. Berlin: Mouton de Gruyter. Peirce, Charles Sanders. 1932. “Ground, object and interpretant.” In Collected Papers of Charles Sanders Peirce. Vol. 2: Elements of Logic, ed. Charles Hartshorne and Paul Weiss, 134–55. Cambridge: Harvard University Press. Pierrehumbert, Janet B. 2003. “Probabilistic phonology: discrimination and robustness.” In Probabilistic Linguistics, ed. Rens Bod, Jennifer Hay, and Stefanie Jannedy, 177–228. Cambridge, MA: MIT Press. Saussure, Ferdinand de. [1916] 1966. Cours de linguistique générale. Ed. Ch. Bally and A. Sechehaye. (Course in General Linguistics. Trans. Wade Baskin. New York : McGraw-Hill.) Searle, John R. 1969. Speech Acts: An Essay in the Philosophy of Language. Cambridge: Cambridge University Press.

Language Structure in Its Human Context Talmy, Leonard. 2000. Toward a Cognitive Semantics. Vol. 1. Concept Structuring Systems Cambridge, MA : MIT Press. Tomasello, Michael. 1999. The Cultural Origins of Human Cognition. Cambridge: Harvard University Press. ———. 2003. Constructing a Language: A Usage-Based Theory of Language Acquisition. Cambridge: Harvard University Press. ———. 2008. The Origins of Human Communication. Cambridge, MA : MIT Press.

Vihman, Marilyn M., and V. M. Gathercole. Language Development. Unpublished manuscript. Vihman, Marilyn M., and S. Kunnari. 2006. “The sources of phonological knowledge: A crosslinguistic perspective.” Recherches Linguistiques de Vincennes 35: 133–64. Wenger, Étienne. 1998. Communities of Practice: Learning, Meaning and Identity. Cambridge: Cambridge University Press.

11

2

with a speaker’s (or hearer’s) semantic and contextual knowledge. Here, we review some of what we have learned about the psychology of linguistic form, as it pertains to sounds, words, and sentences.

SOUNDS

THE PSYCHOLOGY OF LINGUISTIC FORM Lee Osterhout, Richard A. Wright, and Mark D. Allen

Humans can generate and comprehend a stunning variety of conceptual messages, ranging from sophisticated types of mental representations, such as ideas, intentions, and propositions, to more primal messages that satisfy demands of the immediate environment, such as salutations and warnings. In order for these messages to be transmitted and received, however, they must be put into a physical form, such as a sound wave or a visual marking. As noted by the Swiss linguist Ferdinand de Saussure ([2002] 2006), the relationship between mental concepts and physical manifestations of language is almost always arbitrary. The words cat, sat, and mat are quite similar in terms of how they sound but are very dissimilar in meaning; one would expect otherwise if the relationship between sound and meaning was principled instead of arbitrary. Although the relationship between linguistic form and meaning is arbitrary, it is also highly systematic. For example, changing a phoneme in a word predictably also changes its meaning (as in the cat, sat, and mat example). Human language is perhaps unique in the complexity of its linguistic forms (and, by implication, the system underlying these forms). Human language is compositional; that is, every sentence is made up of smaller linguistic units that have been combined in highly constrained ways. A standard view (Chomsky 1965; Pinker 1999) is that units and rules of combination exist at the levels of sound (phonemes and phonology), words (morphemes and morphology), and sentences (words and phrases and syntax). Collectively, these rules comprise a grammar that defi nes the permissible linguistic forms in the language. These forms are systematically related to, but distinct from, linguistic meaning (semantics). Linguistic theories, however, are based on linguistic description and observation and therefore have an uncertain relation to the psychological underpinnings of human language. Researchers interested in describing the psychologically relevant aspects of linguistic form require their own methods and evidence. Furthermore, psychological theories must describe not only the relevant linguistic forms but also the processes that assemble these forms (during language production) and disassemble them (during language comprehension). Such theories should also explain how these forms are associated

12

Sound units. Since the advent of speech research, one of the most intensively pursued topics in speech science has been the search for the fundamental sound units of language. Many researchers have found evidence for phonological units that are abstract (i.e., generalizations across any number of heard utterances, rather than memories of specific utterances) and componential (constituent elements that operate as part of a combinatorial system). However, there is other evidence for less abstract phonological forms that may be stored as whole words. As a result, two competing hypotheses about phonological units have emerged: an abstract componential one versus a holistic one. The more widespread view is the componential one. It posits abstract units that typically relate either to abstract versions of the articulatory gestures used to produce the speech sounds (Liberman and Mattingly 1985, Browman and Goldstein 1990), or to ones derived from descriptive units of phonological theory. Such descriptive units include the feature (see feature analysis), an abstract subphonemic unit of contrast; the phoneme, an abstract unit of lexical contrast that is either a consonant or a vowel; the phone or allophone, a surface variant of the phoneme; the syllable, a timing unit that is made up of a vowel and one or more of its flanking consonants; the prosodic word , the rhythmic structure that relates to patterns of emphasized syllables; or various structures that related to tone, the lexically contrastive use of the voice’s pitch, and intonation, the pitch-based tune that relates to the meaning of a sentence (for reviews, see Frazier 1995; Studdert-Kennedy 1980). In the holistic view, the word is the basic unit, whereas other smaller units are considered to be epiphenomenal (e.g., Goldinger, Pisoni, and Logan, 1991). Instance-specific memory traces of particular spoken words are often referred to as episodes. Proponents of this view point out that while abstract units are convenient for description and relate transparently to segment-based writing systems, such as those based on the alphabet, there is evidence that listeners draw on a variety of highly detailed and instance-specific aspects of a word’s pronunciation in making lexical decisions (for reviews, see Goldinger and Azuma 2003; Nygaard and Pisoni 1995). Some researchers have proposed hybrid models in which there are two layers of representation: the episodic layer, in which highly detailed memory traces are stored, and an abstract layer organized into features or phones (Scharenborg et al. 2005). The proponents of hybrid models try to capture the instance-specific effects in perception that inspire episodic approaches, as well as the highly abstracted lexical contrast effects. PROCESSES. speech production refers to the process by which the sounds of language are produced. The process necessarily involves both a planning stage, in which the words and other linguistic units that make up an utterance are assembled

The Psychology of Linguistic Form in some fashion, and an implementation stage, in which the various parts of the vocal tract, for example the articulators, execute a motor plan to generate the acoustic signal. See Carol A. Fowler (1995) for a detailed review of the stages involved in speech production. It is worth noting here that even if abstract phonological units such as features are involved in planning an utterance, at some point the linguistic string must be implemented as a motor plan and a set of highly coordinated movements. Th is has motivated gestural representations that include movement plans, rather than static featural ones (Browman and Goldstein 1990; Fowler 1986, 1996; Saltzman and Munhall 1989; Stetson 1951). speech perception is the process by which human listeners identify and interpret the sounds of language. It, too, necessarily involves at least two stages: 1) the conversion of the acoustic signal into an electrochemical response at the auditory periphery and 2) the extraction of meaning from the neurophysiological response at the cortical levels. Brian C. J. Moore (1989) presents a thorough review of the physiological processes and some of the issues involved in speech perception. A fundamental point of interest here is perceptual constancy in the face of a massively variable signal. Restated as a question, how is it that a human listener is able to perceive speech sounds and understand the meaning of an utterance, given the massive variability created by physiological idiosyncrasies and contextual variation? The various answers to this question involve positing some sort of perceptual units, be they individual segments, subsegmental features, coordinated speech gestures, or higher-level units like syllables, morphemes, or words. It is worth noting here that the transmission of linguistic information does not necessarily rely exclusively on the auditory channel; the visible articulators, the lips and to a lesser degree the tongue and jaw, also transmit information. A listener presented with both auditory and visual stimuli will integrate the two signals in the perceptual process (e.g., Massaro 1987). When the information in the visual signal is unambiguous (as when the lips are the main articulators), the visual signal may even dominate the acoustic one (e.g., McGurk and Macdonald 1976). Moreover, writing systems convey linguistic information, albeit in a low-dimensional fashion. Most strikingly, sign languages are fully as powerful as speech-based communication systems and are restricted to the visual domain. Despite the differences between signed and spoken languages in terms of the articulators and their perceptual modalities, they draw on the same sorts of linguistic constituents, at least so far as the higher-level units are concerned: syllable, morpheme, word, sentence, and prosodic phrase (e.g., Brentari 1998). Some linguists have also proposed the decomposition of signed languages into smaller units using manual analogs of phonological features, despite the obvious differences in the articulators and the transmission media (for a review see Emmory 2002). The parallel of signed and spoken language structure despite the differences in transmission modalities is often interpreted as evidence for abstract phonological units at the level of the mental lexicon (Meier, Cormier, and Quinto-Pozos 2002). THE HISTORY OF THE DEBATE: EARLY PHONOLOGICAL UNITS. The current debate about how to characterize speech sounds has

its roots in research that dates back over a century. Prior to the advent of instrumental and experimental methods in the late nineteenth century, it was commonly accepted that the basic units of speech were discrete segments that were alphabetic in nature and serially ordered. While it was recognized that speech sounds varied systematically depending on the phonetic context, the variants themselves were thought to be static allophones of an abstract and lexically contrastive sound unit, that is, a phoneme. Translated into modern terminology, phonological planning involved two stages: 1) determining the contextually determined set of discrete surface variants, given a particular lexical string, and 2) concatenating the resulting allophones. The physiological implementation of the concatenated string was thought to result in a series of articulatory steady states, or postures. The only continuous aspects of sound production were believed to be brief transitional periods created by articulatory transitions from one state to the next. The transitional movements were thought to be wholly predictable and determined by the physiology of a particular speaker’s vocal tract. Translated again into modern terminology, perception (when considered) was thought to be simply the process of translating the allophones back into their underlying phonemes for lexical access. The earliest example of the phonemeallophone relationship is attributed to Pāṇini, around 500 b.c.e. whose sophisticated system of phonological rules and relationships influenced structuralist linguists of the early twentieth century, as well as generative linguists of the late twentieth century (for a review, see Anderson 1985; Kiparsky 1979). The predominant view at the end of the nineteenth century was typified by Alexander M. Bell’s (1867) descriptive work on English pronunciation. In it, he presented a set of alphabet-inspired symbols whose shapes and orientations were intended to encode both the articulatory steady states and their resulting steady-state sounds. A fundamental assumption in the endeavor was that all sounds of human language could be encoded as a sequence of universal articulatory posture complexes whose subcomponents were shared by related sounds. For example, all labial consonants (p, b, m, f, v, w, etc.) shared a letter shape and orientation, while all voiced sounds (b, d, g, v, z, etc.) shared an additional mark to distinguish them from their voiceless counterparts (p, t, k, f, s, etc.). Bell’s formalization of a set of universal and invariant articulatory constituents, aligned as an alphabetic string, influenced other universal transcription systems such as Henry Sweet’s (1881) Romic alphabet, which laid the foundation for the development of the International Phonetic Alphabet (Passy 1888). It also foreshadowed the use of articulatory features, such as those proposed by Noam Chomsky and Morris Halle (1968) in modern phonology, in that each speech sound, and therefore each symbol, was made up of a set of universal articulatory components. A second way in which Bell’s work presaged modern research was the connection between perception and production. Implicit in his system of writing was the belief that perception of speech sounds was the process of extracting the articulations that produced them. Later perceptual models would incorporate this relationship in one way or another (Chistovich 1960; Dudley 1940; Fowler 1986, 1996; Joos 1948; Ladefoged and McKinney 1963; Liberman and Mattingly 1985; Stetson 1951).

13

The Cambridge Encyclopedia of the Language Sciences THE HISTORY OF THE DEBATE: EARLY EXPERIMENTAL RESEARCH. Prior to the introduction of experimental methods into phonetics, the dominant methodologies were introspection about one’s own articulations and careful but subjective observations of others’ speech, and the measurement units were letter-based symbols. Thus, the observer and the observed were inextricably linked while the resolution of the measurement device was coarse. Th is view was challenged when a handful of phoneticians and psychologists adopted the scientific method and took advantage of newly available instrumentation, such as the kymograph, in the late 1800s. They discovered that there were no segmental boundaries in the speech stream and that the pronunciation of a particular sound varied dramatically from one instance to the next (for a review of early experimental phonetics, see Kühnert and Nolan 1999 and Minifie 1999). In the face of the new instrumental evidence, some scholars, like Eduard Sievers (1876), P.-J. Rousselot (1897), and Edward Wheeler Scripture (1902), proposed that the speech stream, and the articulations that produced it, were continuous, overlapping, and highly variable, rather than being discrete, invariant, and linear. For them, the fundamental sound units were the syllable or even the word or morpheme. Rousselot’s research (1897–1901) revealed several articulatory patterns that were confi rmed by later work (e.g., Stetson 1951). For example, he observed that when sounds that are transcribed as sequential are generated by independent articulators (such as the lips and tongue tip), they are initiated and produced simultaneously. He also observed that one articulatory gesture may significantly precede the syllable it is contrastive in, thereby presenting an early challenge to the notion of sequential ordering in speech. Laboratory researchers like Raymond H. Stetson (1905, 1951) proposed that spoken language was a series of motor complexes organized around the syllable. Stetson also fi rst proposed that perception was the process of perceiving the articulatory movements that generate the speech signal. However, outside of the experimental phonetics laboratory, most speech researchers, particularly such phonologists as Leonard Bloomfield (1933), continued to use phonological units that remained abstract, invariant, sequential, and letter-like. Th ree events that occurred in the late 1940s and early 1950s changed this view dramatically. The fi rst event was the application to speech research of modern acoustic tools like the spectrogram (Potter 1945), sophisticated models of vocal tract acoustics (e.g., House and Fairbanks 1953), reliable articulatory instrumentation, such as high-speed X-ray cineflourography (e.g., Delattre and Freeman 1968), and electromyographic studies of muscle activation (Draper, Ladefoged, and Whitteridge 1959). The second was the advent of modern perception research in which researchers discovered complex relationships between speech perception and the acoustic patterns present in the signal (Delattre, Liberman, and Cooper 1955). The third was the development of distinctive feature theory in which phonemes were treated as feature matrices that captured the relationships between sounds (Jakobson 1939; Jakobson, Fant, and Halle 1952). When researchers began to apply modern acoustic and articulatory tools to the study of speech production, they

14

rediscovered and improved on the earlier observation that the speech signal and the articulations that create it are continuous, dynamic, and overlapping. Stetson (1951) can be seen as responsible for introducing kinematics into research on speech production. His research introduced the notions of coproduction, in which articulatory gestures were initiated simultaneously, and gestural masking, in which the closure of one articulatory gesture hides another, giving rise to the auditory percept of deletion. Stetson’s work provided the foundation for current language models that incorporate articulatory gestures and their movements as the fundamental phonological units (e.g., Browman and Goldstein 1990; Byrd and Saltzman 2003; Saltzman and Munhall 1989). In the perceptual and acoustic domains, the identification of perceptual cues to consonants and vowels raised a series of questions that remain at the heart of the debate to this day. The coextensive and covarying movements that produce the speech signal result in acoustic information that exhibits a high degree of overlap and covariance with information about adjacent units (e.g., Delattre, Liberman, and Cooper 1955). Any single perceptual cue to a particular speech sound can also be a cue to another speech sound. For example, the onset of a vowel immediately following a consonant provides the listener with cues that identify both the consonant and vowel (Liberman et al. 1954). At the same time, multiple cues may identify a single speech sound. For example, the duration of a fricative (e.g., “s”), the fricative’s noise intensity, and the duration of the preceding vowel all give information about whether the fricative is voiced (e.g., “z”) or voiceless (e.g., “s”) (Soli 1982). Finally, the cues to one phone may precede or follow cues to adjacent phones. The many-to-one, the one-to-many, and the nonlinear relationships between acoustic cues and their speech sounds pose a serious problem for perceptual models in which features or phones are thought to bear a linear relationship to each other. More recently, researchers studying perceptual learning have discovered that listeners encode speaker-specific details and even utterance-specific details when they are learning new speech sounds (Goldinger and Azuma 2003). The latest set of fi ndings poses a problem for models in which linguistic sounds are stored as abstract units. In distinctive feature theory, each phoneme is made up of a matrix of binary features that encodes both the distinctions and the similarities between one class of sounds and the next in a particular language (Jakobson, Fant, and Halle 1952; Chomsky and Halle, 1968). The features are thought to be drawn from a language universal set, and thus allow linguists to observe similarities across languages in the patterning of sounds. Moreover, segmenting the speech signal into units that are hierarchically organized permits a duality of patterning of sound and meaning that is thought to give language its communicative power. That is, smaller units such as phonemes may be combined according to language-specific phonotactic (sound combination) constraints into morphemes and words, and words may be organized according to grammatical constraints into sentences. Th is means that with a small set of canonical sound units, together with recursion, the talker may produce and the hearer may decode and parse a virtually unbounded number of utterances in the language.

The Psychology of Linguistic Form WORDS In this section, we focus on those representations of form that encode meaning and other abstract linguistic content at the most minimally analyzable units of analysis – namely, words and morphemes. As such, we give a brief overview of the study of lexical morphology, investigations in morphological processing, and theories about the structure of the mental lexicon. LEXICAL FORM. What is the nature of a representation at the level of lexical form? We limit our discussion here largely to phonological codes, but recognize that a great many of the theoretical and processing issues we raise apply to orthographic codes as well. It is virtually impossible for the brain to store exact representations for all possible physical manifestations of linguistic tokens that one might encounter or produce. Instead, representations of lexical form are better thought of as somewhat abstract structured groupings of phonemes (or graphemes) that are stored as designated units in long-term memory, either as whole words or as individual morpheme constituents, and associated with any other sources of conceptual or linguistic content encoded in the lexical entries that these form representations map onto. As structured sequences of phonological segments, then, these hypothesized representational units of lexical form must be able to account for essentially all the same meaning-to-form mapping problems and demands that individual phonological segments themselves encounter during on-line performance, due to idiosyncratic variation among speakers and communicative environments. More specifically, representations of morphemes and words at the level of form must be abstract enough to accommodate significant variation in the actual physical energy profi les produced by the motor systems of individual speakers/writers under various environmental conditions. Likewise, in terms of language production, units of lexical form must be abstract enough to accommodate random variation in the transient shape and status of the mouth of the language producer. FORM AND MEANING: INDEPENDENT LEVELS OF LEXICAL REPRESENTATION. The previous description of words and morphemes to some degree rests on the assumption that lexical form is represented independently from other forms of cognitive and linguistic information, such as meaning and lexical syntax (e.g., lexical category, nominal class, gender, verbal subcategory, etc.). Many theories of the lexicon have crucially relied on the assumption of separable levels of representation within the lexicon. In some sense, as explained by Allport and Funnell (1981), this assumption follows naturally from the arbitrariness of mapping between meaning and form, and would thus appear to be a relatively noncontroversial assumption. The skeptical scientist, however, is not inclined to simply accept assumptions of this sort at face value without considering alternative possibilities. Imagine, for example, that the various types of lexical information stored in a lexical entry are represented within a single data structure of highly interconnected, independent, distributed features. Th is sort of arrangement is easy to imagine within the architecture of a connectionist model (McClelland and Rumelhart 1986).

Using the lexical entry cat as an example, imagine a connectionist system in which all the semantic features associated with “cat,” such as [whiskers], [domestic pet], and so on (which are also shared with all other conceptual lexical entities bearing those features, such as , , etc.), are directly associated with the phonological units that comprise its word form /k/, /ae/, /t/ (which are likewise shared with all other word forms containing these phonemes) by means of individual association links that directly tie individual semantic features with individual phonological units (Rueckl et al. 1997). One important consequence of this hypothetical arrangement is that individual word forms do not exist as free-standing representations. Instead, the entire lexical entry is represented as a vector of weighted links connecting individual phonemes to individual lexical semantic and syntactic features. It logically follows from this model, then, that if all or most of the semantic features of the word “cat,” for example, were destroyed or otherwise made unavailable to the processor, then the set of phonological forms /k/ /ae/ /t/, having nothing to link to, would have no means for mental representation, and would therefore not be available to the language processor. We will present experimental evidence against this model that, instead, favors models in which a full phonological word (e.g., /kaet/) is represented in a localist fashion and is accessible to the language processor, even when access to its semantic features is partially or entirely disrupted. Several of the most prominent theories of morphology and lexical structure within formal linguistics make explicit claims about modularity of meaning and form. Ray Jackendoff (1997), for example, presents a theory that has a tripartite structure, in which words have separate identities at three levels of representation – form, syntax, and meaning – and that these three levels are sufficient to encode the full array of linguistic information encoded by each word. His model provides further details in which it is proposed that our ability to store, retrieve, and use words correctly, as well as our ability to correctly compose morphemes into complex words, derives from a memorized inventory of mapping functions that picks out the unique representations or feature sets for a word at each level and associates these elements with one another in a given linguistic structure. While most psycholinguistic models of language processing have not typically addressed the mapping operations assumed by Jackendoff, they do overlap significantly in terms of addressing the psychological reality of his hypothetical tripartite structure in the mental lexicon. Although most experimental treatments of the multilevel nature of the lexicon have been developed within models of language production, as will be seen, there is an equally compelling body of evidence for multilevel processing from studies of language comprehension as well. The most influential lexical processing models over the last two decades make a distinction between at least two levels: the lemma level, where meaning and syntax are stored, and the lexeme level, where phonological and orthographic descriptions are represented. These terms and the functions associated with them were introduced in the context of a computational production model by Gerard Kempen and Pieter

15

The Cambridge Encyclopedia of the Language Sciences Huijbers (1983) and receive further refi nement with respect to human psycholinguistic performance in the foundational lexical production models of Merrill F. Garrett (1975) and Willem Levelt (1989). Much compelling evidence for a basic lemma/ lexeme distinction has come from analyses of naturally occurring speech errors generated by neurologically unimpaired subjects, including tip-of-the-tongue phenomena (Meyer and Bock 1992), as well as from systematic analyses of performance errors observed in patients with acquired brain lesions. A more common experimental approach, however, is the picture–word interference naming paradigm, in which it has been shown that lemma- and lexeme-level information can be selectively disrupted during the course of speech production (Schriefers, Meyer, and Levelt 1990). In terms of lexical comprehension models, perhaps the most straightforward sources of evidence for a meaning/form distinction have come from analyses of the performance of brain-damaged patients. A particularly compelling case for the independence of meaning and form might be demonstrated if an individual with acquired language pathology were to show an intact ability to access word forms in his/ her lexicon, yet remain unable to access meaning from those form representations. Th is is precisely the pattern observed in patients designated as suffering from word meaning deafness. These patients show a highly selective pattern of marked deficit in comprehending word meanings, but with perfect or nearperfect access to word forms. A good example is patient “WBN” as described in Mark D. Allen (2005), who showed an entirely intact ability to access spoken word-form representations. In an auditory lexical decision task, WBN scored 175/182 (96%) correct, which shows that he could correctly distinguish real words from nonwords (e.g., flag vs. fl ig), presumably relying on preserved knowledge of stored lexemes to do so. However, on tasks that required WBN to access meaning from spoken words, such as picture-to-word matching tasks, he performed with only 40%–60% accuracy (at chance in many cases). LEXICAL STRUCTURE: COMPLEX WORDS. A particularly important issue in lexical representation and processing concerns the cognitive structure of complex words, that is, words composed of more than one morpheme. One of the biggest debates surrounding this issue stems from the fact that in virtually all languages with complex word structures, lexical information is encoded in consistent, rule-like structures, as well as in idiosyncratic, irregular structures. Th is issue can be put more concretely in terms of the role of morphological decomposition in single-word comprehension theories within psycholinguistics. Consider the written word wanted, for example. A question for lexical recognition theories is whether the semantic/syntactic properties of this word [WANT, Verb, +Past, … ] are extracted and computed in a combinatorial fashion each time wanted is encountered – by accessing the content associated with the stem want- [WANT, Verb] and combining it with the content extracted from the affi x -ed [+Past] – or whether instead, a single whole-word form wanted is stored at the lexeme level and associated directly with all of its semantic/syntactic content. To understand the plausibility that a lexical system could in principle store whole-word representations such as wanted,

16

one must recognize that in many other cases, such as those involving irregularly inflected words, such as taught, the system cannot store a stem and affi x at the level of form, as there are no clear morpheme boundaries to distinguish these constituents, but must instead obligatorily store it as a whole word at the lexeme level. Many prominent theories have favored the latter, nondecompositional hypothesis for all words, including irregular words like taught, as well as regular compositional words like wanted (Bybee 1988). Other influential processing models propose that complex words are represented as whole-word units at the lexeme level, but that paradigms of inflectionally related words (want, wants, wanted ) map onto a common representation at the lemma level (Fowler, Napps, and Feldman 1985). In addition to this, another class of models, which has received perhaps the strongest empirical support, posits full morphological decomposition at the lexeme level whenever possible (Allen and Badecker 1999). According to these fully decompositional models, a complex word like wanted is represented and accessed in terms of its decomposed constituents want- and -ed at the level of form, such that the very same stem want- is used during the recognition of want, wants, and wanted. According to these models, then, the recognition routines that are exploited by morphological decomposition at the level of form resemble those in theoretical approaches to sentence processing, in which meaning is derived compositionally by accessing independent units of representation of form and combining the content that these forms access into larger linguistic units, according to algorithms of composition specified by the grammar. While there is compelling empirical support for decompositional models of morphological processing, researchers are becoming increasingly aware of important factors that might limit decomposition. These factors are regularity, formal and semantic transparency, and productivity. Regularity refers to the reliability of a particular word-formation process. For example, the plural noun kids expresses noun plurality in a regular, reliable way, while the plural noun children does not. Formal transparency refers to the degree to which the morpheme constituents of a complex structure are obvious from its surface form. For example, morpheme boundaries are fairly obvious in the transparently inflected word wanted, compared to those of the opaquely (and irregularly) inflected word taught. Although an irregular form like taught is formally opaque, as defi ned here, it is nonetheless semantically transparent, because its meaning is a straightforward combination of the semantics of the verb teach and the feature [+Past]. In contrast, an example of a complex word that is formally transparent yet semantically opaque is the compound word dumbbell, which is composed of two recognizable morphemes, but the content associated with these two surface morphemes do not combine semantically to form the meaning of the whole word. Productivity describes the extent to which a word-formation process can be used to form new words freely. For example, the suffi x -ness is easily used to derive novel nouns from adjectives (e.g., nerdiness, awesomeness, catchiness), while the ability to

The Psychology of Linguistic Form form novel nouns using the analogous suffi x -ity is awkward at best (?nerdity) if not impossible. Another phenomenon associated with these lexical properties is that they tend to cluster together in classes of morphologically complex word types across a given language, such that there will often exist a set of highly familiar, frequently used forms that are irregular, formally opaque, and nonproductive, and also a large body of forms that are morphologically regular, formally transparent, and productive. Given the large variety of complex word types found in human languages with respect to these dimensions of combinability, as well as the idiosyncratic nature of the tendency for these dimensions to cluster together from language to language, it would appear that empirical evidence for morphological decomposition must be established on a “case-by-case” basis for each word-formation type within each language. Th is indeed appears to be the direction that most researchers have taken.

SENTENCES On the surface, a sentence is a linear sequence of words. But in order to extract the intended meaning, the listener or reader must combine the words in just the right way. That much is obvious. What is not obvious is how we do that in real time, as we read or listen to a sentence. Of particular relevance to this essay are the following questions: Is there a representational level of syntactic form that is distinct from the meaning of a sentence? And if so, exactly how do we extract the implicit structure in a spoken or written sentence as we process it? One can ask similar questions about the process of sentence production: When planning a sentence, is there a planning stage that encodes a specifically syntactic form? And if so, how do these representations relate to the sound and meaning of the intended utterance? For purely practical reasons, there is far more research on extracting the syntactic form during sentence comprehension (a process known as parsing ; see parsing , human) than on planning the syntactic form of to-be-spoken sentences. Nonetheless, research in both areas has led to substantive advances in our understanding of the psychology of sentence form. SYNTAX AND SEMANTICS. A fundamental claim of a generative grammar is that syntax and semantics are clearly distinct. A fundamental claim of a cognitive grammar is that syntax and semantics are so entwined that they cannot be easily separated. Th is debate among linguists is mirrored by a similar debate among researchers who study language processing. A standard assumption underlying much psycholinguistic work is that a relatively direct mapping exists between the levels of knowledge posited within generative linguistic theories and the cognitive and neural processes underlying comprehension (Bock and Kroch 1989). Distinct language-specific processes are thought to interpret a sentence at each level of analysis, and distinct representations are thought to result from these computations. But other theorists, most notably those working in the connectionist framework, deny that this mapping exists (Elman et al. 1996). Instead, the meaning of the sentence is

claimed to be derived directly, without an intervening level of syntactic structure. The initial evidence of separable syntactic and semantic processing streams came from studies of brain-damaged patients suffering from aphasia , in particular the syndromes known as Broca’s and Wernicke’s aphasia. Broca’s aphasics typically produce slow, labored speech; their speech is generally coherent in meaning but very disordered in terms of sentence structure. Many syntactically important words are omitted (e.g., the, is), as are the inflectional morphemes involved in morphosyntax (e.g., -ing, -ed, -s). Wernicke’s aphasics, by contrast, typically produce fluent, grammatical sentences that tend to be incoherent. Initially, these disorders were assumed to reflect deficits in sensorimotor function; Broca’s aphasia was claimed to result from a motoric deficit, whereas Wernicke’s aphasia was claimed to reflect a sensory deficit. The standard assumptions about aphasia changed in the 1970s, when theorists began to stress the ungrammatical aspects of Broca’s aphasics’ speech; the term agrammatism became synonymous with Broca’s aphasia. Particularly important in motivating this shift was evidence that some Broca’s aphasics have a language-comprehension problem that mirrors their speech-production problems. Specifically, some Broca’s aphasics have trouble understanding syntactically complex sentences (e.g., John was finally kissed by Louise) in which the intended meaning is crucially dependent on syntactic cues – in this case, the grammatical words was and by (Caramazza and Zurif 1976). Th is evidence seemed to rule out a purely motor explanation for the disorder; instead, Broca’s aphasia was viewed as fundamentally a problem constructing syntactic representations, both for production and comprehension. By contrast, Wernicke’s aphasia was assumed to reflect a problem in accessing the meanings of words. These claims about the nature of the aphasic disorders are still quite influential. Closer consideration, however, raises many questions. “Pure” functional deficits affecting a single linguistically defi ned function are rare; most patients have a mixture of problems, some of which seem linguistic but others of which seem to involve motor or sensory processing (Alexander 2006). Many of the Broca’s patients who produce agrammatic speech are relatively good at making explicit grammaticality judgments (Linebarger, Schwartz, and Saff ran 1983), suggesting that their knowledge of syntax is largely intact. Similarly, it is not uncommon for Broca’s aphasics to speak agrammatically but to have relatively normal comprehension, bringing into question the claim that Broca’s aphasia reflects damage to an abstract “syntax” area used in production and comprehension (Miceli et al. 1983). Taken together, then, the available evidence from the aphasia literature does not provide compelling evidence for distinct syntactic and semantic processing streams. Another source of evidence comes from neuroimaging studies of neurologically normal subjects. One useful method involves the recording of event-related brain potentials (ERPs) from a person’s scalp as he or she reads or listens to sentences. ERPs reflect the summed, simultaneously occurring postsynaptic activity in groups of cortical pyramidal neurons. A particularly fruitful approach has involved the presentation of sentences containing linguistic anomalies. If syntactic and semantic aspects of sentence comprehension are segregated

17

The Cambridge Encyclopedia of the Language Sciences into distinct streams of processing, then syntactic and semantic anomalies might affect the comprehension system in distinct ways. A large body of evidence suggests that syntactic and semantic anomalies do in fact elicit qualitatively distinct ERP effects, and that these effects are characterized by distinct and consistent temporal properties. Semantic anomalies (e.g., The cat will bake the food … ) elicit a negative wave that peaks at about 400 milliseconds after the anomalous word appears (the N400 effect) (Kutas and Hillyard 1980). By contrast, syntactic anomalies (e.g., The cat will eating the food … ) elicit a large positive wave that onsets at about 500 milliseconds after presentation of the anomalous word and persists for at least half a second (the P600 effect [Osterhout and Holcomb 1992]). In some studies, syntactic anomalies have also elicited a negativity over anterior regions of the scalp, with onsets ranging from 100 to 300 milliseconds. These results generalize well across types of anomaly, languages, and various methodological factors. The robustness of the effects seems to indicate that the human brain does in fact honor the distinction between the form and the meaning of a sentence. SENTENCE COMPREHENSION. Assuming that sentence processing involves distinct syntactic and semantic processing streams, the question arises as to how these streams interact during comprehension. A great deal of evidence indicates that sentence processing is incremental, that is, that each successive word in a sentence is integrated into the preceding sentence material almost immediately. Such a strategy, however, introduces a tremendous amount of ambiguity – that is, uncertainty about the intended syntactic and semantic role of a particular word or phrase. Consider, for example, the sentence fragment The cat scratched.… There are actually two ways to parse this fragment. One could parse it as a simple active sentence, in which the cat is playing the syntactic role of subject of the verb scratched and the semantic role of the entity doing the scratching (as in The cat scratched the ratty old sofa). Or one could parse it as a more complex relative clause structure, in which the verb scratched is the start of a second, embedded clause, and the cat is the entity being scratched, rather than the one doing the scratching (as in The cat scratched by the raccoon was taken to the pet hospital ). The ambiguity is resolved once the disambiguating information (the ratty sofa or by the raccoon) is encountered downstream, but that provides little help for a parser that assigns roles to words as soon as they are encountered. How does an incremental sentence-processing system handle such ambiguities? An early answer to this question was provided by the garden-path (or modular) parsing models developed in the1980s. The primary claim was that the initial parse of the sentence is controlled entirely by the syntactic cues in the sentence (Ferreira and Clifton 1986). As words arrive in the linguistic input, they are rapidly organized into a structural analysis by a process that is not influenced by semantic knowledge. The output of this syntactic process then guides semantic interpretation. Th is model can be contrasted with interactive models, in which a wide variety of information (e.g., semantics and conceptual/world knowledge) influences the earliest stages of sentence parsing. Initial results of numerous studies

18

(mostly involving the measurement of subjects’ eye movements as they read sentences) indicated that readers tended to read straight through such syntactically simple sentences as The cat scratched the ratty old sofa but experienced longer eye fi xations and more eye regressions when they encountered by the raccoon in the more complex sentences. When confronted with syntactic uncertainty, readers seemed to immediately choose the simplest syntactic representation available (Frazier 1987). When this analysis turned out to be an erroneous choice (that is, when the disambiguating material in the sentence required a more complex structure), longer eye fi xations and more regressions occurred as the readers attempted to “reanalyze” the sentence. A stronger test of the garden-path model, however, requires the examination of situations in which the semantic cues in the sentence are clearly consistent with a syntactically complex parsing alternative. A truly modular, syntax-driven parser would be unaffected by the semantic cues in the sentence. Consider, for example, the sentence fragment The sofa scratched.… Sofas are soft and inanimate and therefore unlikely to scratch anything. Consequently, the semantic cues in the fragment favor the more complex relative clause analysis, in which the sofa is the entity being scratched (as in The sofa scratched by the cat was given to Goodwill ). Initial results seemed to suggest that the semantic cues had no effect on the initial parse of the sentence; readers seemed to build the syntactically simplest analysis possible, even when it was inconsistent with the available semantic information. Such evidence led to the hypothesis that the language processor is comprised of a number of autonomously functioning components, each of which corresponds to a level of linguistic analysis (Ferreira and Clifton 1986). The syntactic component was presumed to function independently of the other components. The modular syntax-fi rst model has been increasingly challenged, most notably by advocates of constraint-satisfaction models (Trueswell and Tanenhaus 1994). These models propose that all sources of relevant information (including statistical, semantic, and real-world information) simultaneously and rapidly influence the actions of the parser. Hence, the implausibility of a sofa scratching something is predicted to cause the parser to initially attempt the syntactically more complex relative-clause analysis. Consistent with this claim, numerous studies have subsequently demonstrated compelling influences of semantics and world knowledge on the parser’s response to syntactic ambiguity (ibid.). There is, however, a fundamental assumption underlying most of the syntactic ambiguity research (regardless of theoretical perspective): that syntax always controls combinatory processing when the syntactic cues are unambiguous. Recently, this assumption has also been challenged. The challenge centers on the nature of thematic roles, which help to defi ne the types of arguments licensed by a particular verb (McRae, Ferretti, and Amyote 1997; Trueswell and Tanenhaus 1994). Exactly what is meant by thematic role varies widely, especially with respect to the amount of semantic and conceptual content it is assumed to hold (McRae, Ferretti, and Amyote 1997). For most syntax-fi rst proponents, a thematic role is limited to a few syntactically relevant selectional restrictions, such as animacy

The Psychology of Linguistic Form (Chomsky 1965); thematic roles are treated as (largely meaningless) slots to be fi lled by syntactically appropriate fi llers. A second view is that there is a limited number of thematic roles (agent, theme, benefactor, and so on), and that a verb selects a subset of these (Fillmore 1968). Although this approach attributes a richer semantics to thematic roles, the required generalizations across large classes of verbs obscure many subtleties in the meaning and usage of these verbs. Both of these conceptions of thematic roles exclude knowledge that people possess concerning who tends to do what to whom in particular situations. Ken McRae and others have proposed a third view of thematic roles that dramatically expands their semantic scope: Thematic roles are claimed to be rich, verb-specific concepts that reflect a person’s collective experience with particular actions and objects (McRae, Ferretti, and Amyote 1997). These rich representations are claimed to be stored as a set of features that defi ne gradients of typicality (“situation schemas”), and to comprise a large part of each verb’s meaning. One implication is that this rich knowledge will become immediately available once a verb’s meaning has been retrieved from memory. As a consequence, the plausibility of a particular word combination need not be evaluated by means of a potentially complex inferential process, but rather can be evaluated immediately in the context of the verb’s meaning. One might therefore predict that semantic and conceptual knowledge of events will have profound and immediate effects on the way in which words are combined during sentence processing. McRae and others have provided evidence consistent with these claims, including semantic influences on syntactic ambiguity resolution. The most compelling argument against the absolute primacy of syntax, however, would be evidence that semantic and conceptual knowledge can “take control” of sentence processing even when opposed by contradicting and unambiguous syntactic cues. Recent work by Ferranda Ferreira (2003) suggests that this might happen on some occasions. She reported that when plausible sentences (e.g., The mouse ate the cheese) were passivized to form implausible sentences (e.g., The mouse was eaten by the cheese), participants tended to name the wrong entity as “do-er” or “acted-on,” as if coercing the sentences to be plausible. However, the processing implications of these results are uncertain, due to the use of postsentence ruminative responses, which do not indicate whether semantic influences reflect the listeners’ initial responses to the input or some later aspect of processing. Researchers have also begun to explore the influence of semantic and conceptual knowledge on the on-line processing of syntactically unambiguous sentences. An illustrative example is a recent ERP study by Albert Kim and Lee Osterhout (2005). The stimuli in this study were anomalous sentences that began with a active structure, for example, The mysterious crime was solving.… The syntactic cues in the sentence require that the noun crime be the Agent of the verb solving. If syntax drives sentence processing, then the verb solving would be perceived to be semantically anomalous, as crime is a poor Agent for the verb solve, and therefore should elicit an N400 effect. However, although crime is a poor Agent, it is an excellent Theme (as in solved the crime). The Theme role can be accommodated simply

by changing the inflectional morpheme at the end of the verb to a passive form (“The mysterious crime was solved … ”). Therefore, if meaning drives sentence processing in this situation, then the verb solving would be perceived to be in the wrong syntactic form (-ing instead of –ed ), and should therefore elicit a P600 effect. Kim and Osterhout observed that verbs like solving elicited a P600 effect, showing that a strong “semantic attraction” between a predicate and an argument can determine how words are combined, even when the semantic attraction contradicts unambiguous syntactic cues. Conversely, in anomalous sentences with an identical structure but with no semantic attraction between the subject noun and the verb (e.g., The envelope was devouring … ”), the critical verb elicited an N400 effect rather than a P600 effect. These results demonstrate that semantics, rather than syntax, can “drive” word combinations during sentence comprehension. SENTENCE PRODUCTION. Generating a sentence requires the rapid construction of novel combinations of linguistic units, involves multiple levels of analysis, and is constrained by a variety of rules (about word order, the formation of complex words, word pronunciation, etc). Errors are a natural consequence of these complexities (Dell 1995). Because they tend to be highly systematic, speech errors have provided much of the data upon which current models of sentence production are based. For example, word exchanges tend to obey a syntactic category rule, in that the exchanged words are from the same syntactic category (for example, two nouns have been exchanged in the utterance Stop hitting your brick against a head wall). The systematicity of speech errors suggests that regularities described in theories of linguistic form also play a role in the speech-planning process. The dominant model of sentence production is based on speech error data (Dell 1995; Garrett 1975; Levelt 1989). According to this model, the process of preparing to speak a sentence involves three stages of planning: conceptualization, formulation, and articulation, in that order. During the conceptualization stage, the speaker decides what thought to express and how to order the relevant concepts sequentially. The formulation stage begins with the selection of a syntactic frame to encode the thought; the frame contains slots that act as place holders for concepts and, eventually, specific words. The phonological string is translated into a string of phonological features, which then drive the motor plan manifested in articulation. Th is model, therefore, posits the existence of representations of syntactic structure that are distinct from the representations of meaning and sound. Other evidence in support of this view comes from the phenomenon of syntactic priming : Having heard or produced a particular syntactic structure, a person is more likely to produce sentences using the same syntactic structure (Bock 1986). Syntactic priming occurs independently of sentence meaning, suggesting that the syntactic frames are independent forms of representation that are quite distinct from meaning.

CONCLUSIONS Collectively, the evidence reviewed in this essay indicates that psychologically relevant representations of linguistic form

19

The Cambridge Encyclopedia of the Language Sciences exist at all levels of language, from sounds to sentences. At each level, units of linguistic form are combined in systematic ways to form larger units of representation. For the most part, these representations seem to be abstract; that is, they are distinct from the motor movements, sensory experiences, and episodic memories associated with particular utterances. However, it is also clear that more holistic (that is, nondecompositional) representations of linguistic form, some of which are rooted in specific episodic memories, also play a role in language processing. It also seems to be true that linguistic forms (e.g., the morphological structure of a word or the syntactic structure of a sentence) are dissociable from the meanings they convey. At the same time, semantic and conceptual knowledge can strongly influence the processing of linguistic forms, as exemplified by semantic transparency effects on word decomposition and thematic role effects on sentence parsing. These conclusions represent substantive progress in our understanding of linguistic form and the role it plays in language processing. Nonetheless, answers to some of the most basic questions remain contentiously debated, such as the precise nature of the “rules” of combination, the relative roles of compositional and holistic representations, and the pervasiveness of interactions between meaning and form. WORKS CITED AND SUGGESTIONS FOR FURTHER READING Alexander, M. P. 2006 . “Aphasia I: Clinical and anatomical issues.” In Patient-Based Approaches to Cognitive Neuroscience (2d ed.), ed. M. J. Farah and T. E. Feinberg , 165–82. Cambridge, MA : MIT Press . Allen, Mark D. 2005. “The preservation of verb subcategory knowledge in a spoken language comprehension deficit.” Brain and Language 95: 255 –64. Allen, Mark , and William Badecker. 1999. “Stem homograph inhibition and stem allomorphy: Representing and processing inflected forms in a multilevel lexical system.” Journal of Memory and Language 41: 105 –23. Allport, D. A., and E. Funnell. 1981. “Components of the mental lexicon.” Philosophical Transactions of the Royal Society of London B 295: 397–410. Anderson, Stephen R . 1985. Phonology in the Twentieth Century: Theories of rules and Theories of Representations. Chicago: The University of Chicago Press. Bell, Alexander M. 1867. Visible Speech: The Science of Universal Alphabetics. London: Simpkin, Marshal. Bloomfield, Leonard. 1933. Language. New York : H. Holt & Co. Bock , J. Katherine. 1986. “Syntactic persistence in language production.” Cognitive Psychology 18: 355 –87. Bock, J. K., and Anthony S. Kroch. 1989. “The isolability of syntactic processing.” In Linguistic Structure in Language Processing, ed.. G. N. Carlson and M. K. Tanenhaus, 157–96. Boston: Kluwer Academic. Brentari, Dianne. 1998. A Prosody Model of Sign Language Phonology. Cambridge, MA : MIT Press. Browman, Catherine P., and Louis Goldstein. 1990. “Gestural specification using dynamically-defi ned articulatory structures.” Journal of Phonetics 18: 299 –320. Bybee, Joan . 1988. “Morphology as lexical organization.” In Th eoretical Morphology: Approaches in Modern Linguistics, ed. M. Hammond and M. Noonan, 119–41. San Diego, CA : Academic Press.

20

Byrd, Dani, and Elliot Saltzman. 2003. “The elastic phrase: Modeling the dynamics of boundary-adjacent lengthening.” Journal of Phonetics 31: 149 –80. Caplan, David. 1995. “Issues arising in contemporary studies of disorders of syntactic processing in sentence comprehension in agrammatic patients.” Brain and Language 50: 325 –38. Caramazza, Alfonzo, and Edgar Zurif. 1976. “Dissociations of algorithmic and heuristic processes in language comprehension: Evidence from aphasia.” Brain and Language 3: 572–82. Chistovich, Ludmilla A 1960. “Classification of rapidly repeated speech sounds.” Akustichneskii Zhurnal 6: 392–98. Chomsky, Noam. 1957. Syntactic Structures. The Hague: Mouton. ———. 1965. Aspects of the Theory of Syntax . Cambridge, MA : MIT Press. Chomsky, Noam, and Morris Halle. 1968. The Sound Pattern of English. New York : Harper and Row. Delattre, Pierre, and Donald Freeman. 1968. “A dialect study of American R’s by x-ray motion picture.” Linguistic s 44: 29 –68. Delattre, Pierre C., Avin M. Liberman, and Franklin S. Cooper. 1955. “Acoustic loci and transitional cues for consonants.” Journal of the Acoustical Society of America 27: 769 –73. Dell, Gary S. 1995. “Speaking and misspeaking.” In An Invitation to Cognitive Science: Language. Cambridge, MA : MIT Press. Draper, M., P. Ladefoged, and D. Whitteridge. 1959. “Respiratory muscles in speech .” Journal of Speech and Hearing Research 2: 16 –27. Dudley, Homer. 1940. “The carrier nature of speech.” Bell System Technical Journal 14: 495 –515. Elman, Jeff rey L. 1990. “Representation and structure in connectionist models.” In Cognitive Models of Speech Processing, ed. G. T. M. Altmann, 227–60. Cambridge, MA : MIT Press. Elman, Jeff rey L., Elizabeth A. Bates, A. Karmiloff-Smith, D. Parisi, and K. Plunkett. 1996. “Rethinking innateness.” Cambridge, MA: MIT Press. Emmory, Karen. 2002. Language, Cognition, and the Brain: Insights from Sign Language Research. Mahwah, NJ: Lawrence Erlbaum Associates. Ferreira, Fernanda. 2003. “The misinterpretation of noncanonical sentences.” Cognitive Psychology 47: 164 –203. Ferreira, Fernanda, and Charles Clifton, Jr. 1986. “The independence of syntactic processing.” Journal of Memory and Language 25: 348 –68. Fillmore, Charles. 1968. “The case for case.” In Universals of Linguistic Theory, ed. E. Bach, 1–80. New York : Holt, Rinehart, & Winston. Fowler, Carol A. 1986. “An event approach to the study of speech perception from a direct-realist perspective.” Journal of Phonetics 14: 3–28. ———. 1995. “Speech production.” In Speech, Language, and Communication, ed. J. L. Miller and P. D. Eimas, 29–61. New York : Academic Press. ———. 1996. “Listeners do hear sounds not tongues.” Journal of the Acoustical Society of America 99: 1730 –41. Fowler, Carol, Susan Napps, and Laurie Feldman. 1985. “Relations among regular and irregular morphologically related words in the lexicon as revealed by repetition priming.” Memory and Cognition 13: 241–55. Franklin, S., J. Lambon Ralph, J. Morris, and P. Bailey. 1996. “A distinctive case of word meaning deafness?” Cognitive Neuropsychology 13: 1139 –62. Frazier, L. 1987. “Sentence processing: A tutorial review.” In Attention and Performance XII: The Psychology of Reading, ed. M. Coltheart, 3–30. Hillsdale, NJ: Lawrence Erlbaum Associates. ———. 1995. “Representation in psycholinguistics.” In Speech, Language, and Communication, ed. J. L. Miller and P. D. Eimas, 1–27. New York : Academic Press.

The Psychology of Linguistic Form Garrett , Merrill F. 1975. “The analysis of sentence production.” In Th e Psychology of Learning and Motivation , ed. G. Bower, 133–77. New York : Academic Press . Goldinger, Stephen D., and Tamiko Azuma . 2003. “Puzzle-solving science: The quixotic quest for units of speech perception .” Journal of Phonetics 31: 305 –20. Goldinger, Stephen D., David B. Pisoni, and John S. Logan . 1991. “On the nature of talker variability effects on recall of spoken word lists .” Journal of Experimental Psychology: Learning, Memory and Cognition 17: 152 –62. Hall, D. A., and M. J. Riddoch . 1997. “ Word meaning deafness: Spelling words that are not understood .” Cognitive Neuropsychology 14: 1131–64. Hillis , Argye E. 2000. “The organization of the lexical system.” In What Defi cits Reveal about the Human Mind/Brain: Handbook of Cognitive Neuropsychology, ed. B. Rapp, 185–210. New York : Psychology Press . House, Arthur S., and Grant Fairbanks. 1953. “ The influence of consonant enviroment upon the secondary acoustical characteristics of vowels .” Journal of the Acoustical Society of America 25: 105 –13. Jackendoff , Ray. 1997. Th e Architecture of the Language Faculty. Cambridge, MA : MIT Press . Jakobson, Roman . 1939. “Observations sur le classment phonologique des consonnes .” Proceedings of the 3rd International Conference of Phonetic Sciences, 34 –41. Ghent. Jakobson, Roman, Gunnar Fant , and Morris Halle. 1952 . Preliminaries to Speech Analysis . Cambridge, MA : MIT Press . Joos , Martin . 1948 . “Acoustic Phonetics .” Language Monograph 23, Supplement to Language 24: 1–36. Kempen, Gerard, and Pieter Huijbers . 1983. “ The lexicalization process in sentence production and naming: Indirect election of words .” Cognition 14: 185 –209. Kim , Albert, and Lee Osterhout . 2005. “ The independence of combinatory semantic processing: Evidence from event-related potentials .” Journal of Memory and Language 52: 205 –25. Kiparsky, Paul. 1979. Panini as a Variationist . Cambridge, MA : MIT Press . Kühnert , Barbara , and Francis Nolan . 1999. “The origin of coarticulation.” In Coarticulation: Th eory, Data and Techniques , ed. B. Rapp, 7–30. Cambridge: Cambridge University Press . Kutas , Marta, and Steven A. Hillyard . 1980. “Reading senseless sentences: Brain potentials reflect semantic anomaly.” Science 207: 203 –5. Ladefoged, P., and N. McKinney. 1963. “Loudness, sound pressure, and subglottal pressure in speech .” Journal of the Acoustical Society of America 35: 454 –60. Lambon, Ralph M., K. Sage, and A. Ellis . 1996 . “ Word meaning blindness: A new form of acquired dyslexia .” Cognitive Neuropsychology 13: 617–39. Levelt , Willem . 1989. Speaking: From intention to articulation . Cambridge, MA : MIT Press . Liberman, Alvin M., Pierre C. Delattre, Franklin S. Cooper, and Lou J. Gerstman . 1954 . “ The role of consonant vowel transitions in the perception of the stop and nasal consonants .” Journal of Experimental Psychology 52: 127–37. Liberman, Alvin M., and Ignatius G. Mattingly. 1985. “ Th e motor theory of speech perception revised .” Cognition 21: 1–36. Linebarger, Marcia, Myrna Schwartz , and Eleanor Saff ran . 1983. “Sensitivity to grammatical structure in so-called agrammatic aphasics .” Cognition 13: 361–93. Massaro, Dominic W. 1987. Speech Perception by Ear and Eye: A Paradigm for Psychological Inquiry. Hillsdale, NJ: Lawrence Erlbaum Associates .

McClelland, Jamesv, and David Rumelhart. 1986. Parallel Distributed Processing: Explorations in the Microstructure of Cognition . Vol. 1. Cambridge, MA : MIT Press. McGurk , Harry, and John Macdonald. 1976. “Hearing lips and seeing voices.” Nature 264: 746 –8. McRae, Ken, Todd R. Ferretti, and Liane Amyote. 1997. “Thematic roles as verb-specific concepts.” Language and Cognitive Processes 12.2: 137–76. Meier, Richard P., Kearsy Cormier, and David Quinto-Pozos. 2002. Modularity in Signed and Spoken Languages. Cambridge: Cambridge University Press. Meyer, Antje, and Kathryn Bock . 1992. “Tip-of-the-tongue phenomenon: Blocking or partial activation?” Memory and Cognition 20: 715 –26. Miceli, G., L. Mazzuchi, L. Menn, and H. Goodglass. 1983. “Contrasting cases of Italian agrammatic aphasia without comprehension disorder.” Brain and Language 19: 65 –97. Miller, George A., and Patricia E. Nicely. 1955. “An analysis of perceptual confusions among some English consonants.” Journal of the Acoustical Society of America 27: 329 –35. Minifie, Fred D. 1999. “The history of physiological phonetics in the United States.” In A Guide to the History of the Phonetic Sciences in the United States: Issued on the Occasion of the 14th International Congress of Phonetic Sciences, San Francisco, 1–7 August 1999, ed. J. Ohala, A. Bronstein, M. Busà, L. Grazio, J. Lewis, and W. Weigel. Berkeley : University of California Press.. Moore, Brian C. J. 1989. An Introduction to the Psychology of Hearing. 3d ed. London: Academic Press. Nygaard, Lynn C., and David B. Pisoni. 1995. “Speech Perception: New directions in research and theory.” In Speech, Language, and Communication, ed. J. Miller and P. Eimas, 63–96. New York : Academic Press. Osterhout, Lee, and Philip J. Holcomb. 1992. “Event-related brain potentials elicited by syntactic anomaly.” Journal of Memory and Language 31: 785 –806. Passy, Paul. 1888. “Our revised alphabet.” The Phonetic Teacher, 57–60. Pinker, Steven. 1999. Words and Rules: The Ingredients of Language. New York : Basic Books. Potter, Ralph K. 1945. “ Visible patterns of sound.” Science 102: 463–70. Rousselot, P.-J. 1897–1901. Principes de Phonétique Experimentale. Paris: H. Welter. Rueckl, Jay, Michelle Mikolinski, Michal Raveh, Caroline Miner, and F. Mars. 1997. “Morphological priming, fragment completion, and connectionist networks.” Journal of Memory and Language 36: 382–405. Saltzman, Elliot L., and Kevin G. Munhall. 1989. “A dynamical approach to gestural patterning in speech production.” Ecological Psychology 1: 333–82. Saussure, Ferdinand de. [2002] 2006. Écrits de linguistique générale, ed. Simon Bouquet and Rudolf Engler. Paris: Gallimard. English translation: Writings in General Linguistics, Oxford: Oxford University Press. Scharenborg , O., D. Norris, L. ten Bosch, and J. M. McQueen. 2005. “How should a speech recognizer work? ” Cognitive Science 29: 867–918. Schriefers, Herbert, Antje Meyer and Willem Levelt. 1990. “Exploring the time course of lexical access in language production: Pictureword interference studies.” Journal of Memory and Language 29: 86 –102. Scripture, Edward Wheeler. 1902. The Elements of Experimental Phonetics. New York : Charles Scribner’s Sons. Sievers, Eduard. 1876. Grundzuge der Lautphysiologie zur Einfuhrung in das Studium der Lautlehere der Indogermanischen Sprachen. Leipzig: Breitkopf and Hartel.

21

The Cambridge Encyclopedia of the Language Sciences Soli, Sig D. 1982. “Structure and duration of vowels together specify fricative voicing.” Journal of the Acoustical Society of America 72: 366 –78. Stetson, Raymond H. 1905. “A motor theory of rhythm and discrete succession II.” Psychological Review 12: 293–350. ———. 1951. Motor Phonetics: A Study of Speech Movement in Action. 2d ed. Amsterdam: North-Holland. Studdert-Kennedy, Michael. 1980. “Speech perception.” Language and Speech 23: 45 –65.

22

Sweet, Henry. 1881. “Sound notation.” Transactions of the Philological Society: 177–235. Trueswell, John C., and Michael K. Tanenhaus. 1994. “Toward a lexicalist framework of constraint-based syntactic ambiguity resolution.” In Perspectives on Sentence Processing , ed. C. Clifton, L. Frazier, and K. Rayner, 155–80. Hillsdale, NJ: Lawrence Erlbaum Associates.

(2)

a. The girl laughed and sang. b. The girl laughed.

Th is is an example of structural entailment, because it is the structure itself that allows the inference (i.e., “if someone does both A and B, then someone does A”). Th is particular rule is essentially the classical inference rule of conjunction elimination from propositional logic; that is,

3 THE STRUCTURE OF MEANING

(3)

James Pustejovsky

A∧B A

While this relies on a largely syntactic notion of entailment, semantics should also explain how (4b) is a legitimate inference from (4a). (4)

1 INTRODUCTION Semantics is the systematic study of meaning in language. As a discipline, it is directed toward the determination of how humans reason with language, and more specifically, discovering the patterns of inference we employ through linguistic expressions.The study of semantics has diverse traditions, and the current literature is quite heterogeneous and divided on approaches to some of the basic issues facing the field (cf. semantics). While most things in the world have meaning to us, they do not carry meaning in the same way as linguistic expressions do. For example, they do not have the properties of being true or false, or ambiguous or contradictory. (See Davis and Gillon [2004] for discussion and development of this argument.) For this and other reasons, this overview essay addresses the question of how linguistic expressions carry meaning and what they denote in the world. Where syntax determines the constituent structure of a sentence along with the assignment of grammatical and thematic relations, it is the role of semantics to compute the deeper meaning of the resulting expression. For example, the two sentences in (1) differ in their syntactic structures (through their voice), but they mean essentially the same thing; that is, their propositional content is identical. (1)

a. The child ate a cookie. b. A cookie was eaten by the child.

Early on, such observations led philosophers and linguists to distinguish meaning from the pure structural form of a sentence (Saussure [1916] 1983; Russell 1905). Semantic theories in linguistics assume that some sort of logical form is computed from the constituent structure associated with a sentence, and it is this meaning representation that allows us to make categorical and truth-conditional judgments, such as the equivalence in meaning of the two sentences in (1). Another role played by semantics is in the computation of inferences from our utterances, such as entailments, implicatures, and presuppositions. For example, consider the various notions of entailment. From the logical form (LF) of the sentence in (2a), semantics enables us to infer (2b) as a legitimate inference.

a. The drought killed the crops. b. The crops died.

Such lexical entailments involve an inference that is tied directly to the meaning of a word, namely, the verb kill; that is, “when something is killed, it dies.” Hence, the role of lexical information in the construction of logical forms and the inferences we can compute from our utterances is an important area of linguistics, and one we return to in Section 3.5 below. There is an important distinction in semantics among propositions, sentences, and utterances. We can think of an utterance as a speech-act, situated in time and space, that is, which happens at a particular time and location. A sentence, on the other hand, is a expression that is inherently linguistic, and can be expressed on multiple occasions by multiple utterances. The notion of a proposition is more complex and contentious, but it is that object that is traditionally taken as being true or false, expressed by the sentence when uttered in a specific context.

1.1 Historical Remarks The study of meaning has occupied philosophers for centuries, beginning at least with Plato’s theory of forms and Aristotle’s theory of meaning. Locke, Hume, and Reid all pay particular attention to the meanings of words in composition, but not until the late nineteenth century do we see a systematic approach to the study of logical syntax emerge, with the work of Bertrand Russell and Gottlob Frege. Russell and Frege were not interested in language as a linguistic phenomenon, but rather as a medium through which judgments can be formed and expressed. Frege’s focus lay in formulating the rules that create meaningful expressions in a compositional manner, while also introducing an important distinction between an expression’s sense and its reference (cf. sense and reference, reference and extension). Russell’s work on the way in which linguistic expressions denote introduced the problem of definite descriptions and referential failure, and what later came to be recognized as the problem of presupposition (cf. pragmatics). Ferdinand de Saussure ([1916] 1983), working within an emerging structuralist tradition, developed relational techniques for linguistic analysis, which were elaborated into a framework of componential analysis for language meaning. The idea behind componential analysis is the reduction of a word’s meaning into its ultimate contrastive elements. These

23

The Cambridge Encyclopedia of the Language Sciences contrastive elements are structured in a matrix, allowing for dimensional analysis and generalizations to be made about lexical sets occupying the cells in the matrix. Th is technique developed into a general framework for linguistic description called distinctive FEATURE ANALYSIS. Th is is essentially the inspiration for J. Katz and J. Fodor’s 1963 theory of lexical semantics within transformational grammar. On this theory, usually referred to as markerese, a lexical entry in the language consists of grammatical and semantic markers, and a special feature called a semantic distinguisher. In the subsequent discussion by U. Weinreich (1972) and many others, this model was demonstrated to be far too impoverished to characterize the compositional mechanisms inherent in language. In the late 1960s and early 1970s, alternative models of word meaning emerged (Fillmore 1968 [frame semantics]; Lakoff [1965] 1970 [generative semantics]; Gruber 1976; Jackendoff 1972), which respected the relational structure of sentence meaning while encoding the named semantic functions in lexical entries. In D. R. Dowty (1979), a model theoretic interpretation of the decompositional techniques of G. Lakoff, J. D. McCawley, and J. R. Ross was developed. In the later twentieth century, montague grammar (Montague 1973, 1974) was perhaps the most significant development in the formal analysis of linguistic semantics, as it brought together a systematic, logically grounded theory of compositionality, with a model theoretic interpretation. Subsequent work enriched this approach with insights from D. Davidson (1967), H. P. Grice (1969), Saul Kripke ([1972] 1980), David Lewis (1976), and other philosophers of language (cf. Partee 1976; Davidson and Harman 1972). Recently, the role of lexical-syntactic mapping has become more evident, particularly with the growing concern over projection from lexical semantic form, the problem of verbal alternations and polyvalency, and the phenomenon of polysemy. The work of R. Jackendoff (1983, 1997) on conceptual semantics has come to the fore, as the field of lexical semantics has developed into a more systematic and formal area of study (Pustejovsky and Boguraev 1993; Copestake and Briscoe 1995, 15–67). Finally, one of the most significant developments in the study of meaning has been the “dynamic turn” in how sentences are interpreted in discourse. Inspired by the work of Irene Heim (1982) and H. Kamp (1981), the formal analysis of discourse has become an active and growing area of research, as seen in the works of Jeroen Groenendijk and Martin Stokhof (1991), Kamp and U. eyle (1993), and Nicholas Asher and Alex Lascarides (2003). In the remainder of this essay, we examine the basic principle of how meanings are constructed. First, we introduce the notion of compositionality in language. Since words are the building blocks of larger meanings, we explore various approaches to lexical semantics. Then, we focus on how units of meaning are put together compositionally to create propositions. Finally, we examine the meaning of expressions above the level of the sentence, within a discourse.

1.2 Compositionality Because semantics focuses on how linguistic expressions come to have meaning, one of the most crucial concepts in the field

24

is the notion of compositionality (cf. compositionality). As speakers of language, we understand a sentence by understanding its parts and how they are put together. The principle of compositionality characterizes how smaller units of meaning are put together to form larger, more meaningful expressions in language. The most famous formulation of this notion comes from Frege, paraphrased as follows: The meaning of an expression is a function of the meanings of its parts and the way they are syntactically combined. (Partee 1984)

Th is view has been extremely influential in semantics research over the past 40 years. If one assumes a compositional approach to the study of meaning, then two things immediately follow: 1) One must specify the specific meaning of the basic elements of the language, and 2) one must formulate the rules of combination for how these elements go together to make more complex expressions. The fi rst aspect includes determining what words and morphemes mean, that is, lexical semantics, which we address in the next section. The second aspect entails defi ning a calculus for how these elements compose to form larger expressions, that is, argument selection and modification. Needless to say, in both of these areas, there is much divergence of opinion, but semanticists generally agree on the basic assumptions inherent in compositionality.

2 LEXICAL MEANING Semantic interpretation requires access to knowledge about words. The lexicon of a grammar must provide a systematic and efficient way of encoding the information associated with words in a language. lexical semantics is the study of what words mean and how these meanings are structured. The lexicon is not merely a collection of words with their semantic forms, but rather a set of structured objects that participate in larger operations and compositions, both enabling syntactic environments and acting as signatures to semantic entailments and implicatures in the context of larger discourse. There are four basic questions in modeling the semantic content and structure of the lexicon: 1) What semantic information goes into a lexical entry? 2) How do lexical entries relate semantically to one another? 3) How is this information exploited compositionally by the grammar? 4) How is this information available to semantic interpretation generally? The lexicon and lexical semantics have traditionally been viewed as the most passive modules of language, acting in the service of the more dynamic components of the grammar. This view has its origins in the generative tradition (Chomsky [1955] 1975) and has been an integral part of the notion of the lexicon ever since. While the Aspects-model of selectional features (Chomsky 1965) restricted the relation of selection to that between lexical items, work by McCawley (1968) and Jackendoff (1972) showed that selectional restrictions must be available to computations at the level of derived semantic representation rather than at deep structure. Subsequent work by Joan Bresnan (1982), Gerald Gazdar et al. (1985), and C. Pollard and I. Sag (1994) extend the range of phenomena that can be handled by the projection and exploitation of lexically derived information in the grammar.

The Structure of Meaning and alternations, among other relations (cf. Pustejovsky and Boguraev 1993).

Natural Entity

Physical

Abstract

2.2 Argument Structure Mass

Individuated

inanimate

animate

rock

human

Mental

Experiential

Figure 1.

Recently, with the convergence of several areas in linguistics (lexical semantics, computational lexicons, type theories), several models for the determination of selection have emerged that put even more compostional power in the lexicon, making explicit reference to the paradigmatic systems that allow for grammatical constructions to be partially determined by selection. Examples of this approach are generative lexicon theory (Pustejovsky 1995; Bouillon and Busa 2001), and construction grammar (Goldberg, 1995; Jackendoff 1997, 2002). These developments have helped to characterize the approaches to lexical design in terms of a hierarchy of semantic expressiveness. There are at least three such classes of lexical description, defi ned as follows: sense enumerative lexicons, where lexical items have a single type and meaning, and ambiguity is treated by multiple listings of words; polymorphic lexicons, where lexical items are active objects, contributing to the determination of meaning in context, under well-defi ned constraints; and unrestricted sense lexicons, where the meanings of lexical items are determined mostly by context and conventional use. It seems clear that the most promising direction seems to be a careful and formal elucidation of the polymorphic lexicons, and this will form the basis of our subsequent discussion. Lexical items can be systematically grouped according to their syntactic and semantic behavior in the language. For this reason, there have been two major traditions of word clustering, corresponding to this distinction. Broadly speaking, for those concerned mainly with grammatical behavior, the most salient aspect of a lexical item is its argument structure; for those focusing on a word’s entailment properties, the most important aspect is its semantic class. In this section, we examine these two approaches and see how their concerns can be integrated into a common lexical representation.

Once the basic semantic types for the lexical items in the language have been specified, their subcategorization and selectional information must be encoded in some form. The argument structure for a word can be seen as the simplest specification of its semantics, indicating the number and type of parameters associated with the lexical item as a predicate. For example, the verb die can be represented as a predicate taking one argument, and kill as taking two arguments, while the verb give takes three arguments. (5)

What originally began as the simple listing of the parameters or arguments associated with a predicate has developed into a sophisticated view of the way arguments are mapped onto syntactic expressions. E. Williams’s (1981) distinction between external (the underlined arguments for “kill” and “give”) and internal arguments and J. Grimshaw’s proposal for a hierarchically structured representation (cf. Grimshaw 1990) provide us with the basic syntax for one aspect of a word’s meaning. Similar remarks hold for the argument list structure in HPSG (head-driven phrase structure grammar) and LFG (lexical-functional grammar). One influential way of encoding selectional behavior has been the theory of thematic relations (cf. thematic roles; Gruber 1976; Jackendoff 1972). Thematic relations are now generally defi ned as partial semantic functions of the event being denoted by the verb or noun, and behave according to a predefi ned calculus of role relations (e.g., Carlson 1984; Dowty 1991; Chierchia 1989). For example, semantic roles, such as agent, theme, and goal, can be used to partially determine the meaning of a predicate when they are associated with the grammatical arguments to a verb. (6)

One of the most common ways to organize lexical knowledge is by means of type or feature inheritance mechanisms (Evans and Gazdar 1990; Carpenter 1992; Copestake and Briscoe 1992; Pollard and Sag 1994). Furthermore, T. Briscoe, V. de Paiva, and A. Copestake (1993) describe a rich system of types for allowing default mechanisms into lexical type descriptions. Similarly, type structures, such as that shown in Figure 1, can express the inheritance of syntactic and semantic features, as well as the relationship between syntactic classes

a. put< AGENT,THEME,LOCATION> b. borrow

Thematic roles can be ordered relative to each other in terms of an implicational hierarchy. For example, there is considerable use of a universal subject hierarchy such as shown in the following (cf. Fillmore 1968; Comrie 1981). (7)

2.1 Semantic Classes

a. die(x) b. kill(x,y) c. give(x,y,z)

AGENT > RECIPIENT/BENEFACTIVE > THEME/PATIENT > INSTRUMENT > LOCATION

Many linguists have questioned the general explanatory coverage of thematic roles, however, and have chosen alternative methods for capturing the generalizations they promised. Dowty (1991) suggests that “theta-role” generalizations are best captured by entailments associated with the predicate itself. A theta-role can then be seen as the set of predicate entailments that are properties of a particular argument to the verb. Characteristic entailments might be thought of as prototype roles, or proto-roles; this allows for degrees or shades of

25

The Cambridge Encyclopedia of the Language Sciences meaning associated with the arguments to a predicate. Others have opted for a more semantically neutral set of labels to assign to the parameters of a relation, whether it is realized as a verb, noun, or adjective. For example, the theory of argument structure as developed by Williams (1981), Grimshaw (1990), and others can be seen as a move toward a more minimalist description of semantic differentiation in the verb’s list of parameters. The interaction of a structured argument list and a rich system of types, such as that presented previously, provides a mechanism for semantic selection through inheritance. Consider, for instance the sentence pairs in (8). (8)

a. The man / the rock fell. b. The man / *the rock died.

Now consider how the selectional distinction for a feature such as animacy is modeled so as to explain the selectional constraints of predicates. For the purpose of illustration, the arguments of a verb will be identified as being typed from the system shown previously. (9)

a. λx :physical[fall(x)] b. λx :animate[die(x)]

In the sentences in (8), it is clear how rocks can’t die and men can, but it is still not obvious how this judgment is computed, given what we would assume are the types associated with the nouns rock and man, respectively. What accomplishes this computation is a rule of subtyping, Θ, that allows the type associated with the noun man (i.e., human) to also be accepted as the type animate, which is what the predicate die requires of its argument as stated in (9b) (cf. Gunter 1992; Carpenter 1992): (10) Θ [human ⊆ animate]: human → animate

The rule applies since the concept human is subtyped under animate in the type hierarchy. Parallel considerations rule out the noun rock as a legitimate argument to die since it is not subtyped under animate. Hence, one of the concerns given for the way that syntactic processes can systematically keep track of which selectional features are entailed and which are not is partially addressed by such lattice traversal rules as the one presented here.

2.3 Decomposition The second approach to the aforementioned lexical specification is to defi ne constraints internally to the predicate itself. Traditionally, this has been known as lexical decomposition. Since the 1960s, lexical semanticists have attempted to formally model the semantic relations between such lexical items as the adjective dead and the verbs die and kill (cf. Lakoff [1965] 1970; McCawley 1968) in the sentences that follow. (11) a. John killed Bill. b. Bill died. c. Bill is dead.

Assuming that the underlying form for a verb like kill directly encodes the stative predicate in (11c) and the relation of causation, generative semanticists posited representations such as (12).

26

12 CAUSE x, BECOME NOT ALIVE y

Here, the predicate CAUSE is represented as a relation between an individual causer x and an expression involving a change of state in the argument y. R. Carter ([1976] 1988) proposes a representation quite similar, shown here for the causative verb darken: 13 x CAUSE  y BE.DARK CHANGE

Although there is an intuition that the cause relation involves a causer and an event, neither Lakoff nor Carter makes this commitment explicitly. In fact, it has taken several decades for Davidson’s (1967) observations regarding the role of events in the determination of verb meaning to fi nd their way convincingly into the major linguistic frameworks. Recently, a new synthesis has emerged that attempts to model verb meanings as complex predicative structures with rich event structures (cf. Parsons 1990; Pustejovsky 1991b; Tenny 1992; Krifka 1992). Th is research has developed the idea that the meaning of a verb can be analyzed into a structured representation of the event that the verb designates, and has furthermore contributed to the realization that verbs may have complex, internal event structures. Recent work has converged on the view that complex events are structured into an inner and an outer event, where the outer event is associated with causation and agency and the inner event is associated with telicity (completion) and change of state (cf. Tenny and Pustejovsky 2000; Levin and Rappaport Hovav 2005). Jackendoff (1990) develops an extensive system of what he calls Conceptual Representations, which parallel the syntactic representations of sentences of natural language. These employ a set of canonical predicates, including CAUSE, GO, TO, and ON, and canonical elements, including Thing, Path, and Event. These approaches represent verb meaning by decomposing the predicate into more basic predicates. Th is work owes obvious debt to the innovative work within generative semantics, as illustrated by McCawley’s (1968) analysis of the verb kill. Recent versions of lexical representations inspired by generative semantics can be seen in the Lexical Relational Structures of K. Hale and S. J. Keyser (1993), where syntactic tree structures are employed to capture the same elements of causation and change of state as in the representations of Carter, Levin and T. Rapoport, Jackendoff, and Dowty. The work of Levin and Rappaport, building on Jackendoff ’s Lexical Conceptual Structures, has been influential in further articulating the internal structure of verb meanings (see Levin and Rappaport 1995). J. Pustejovsky (1991b) extends the decompositional approach presented in Dowty (1979) by explicitly reifying the events and subevents in the predicative expressions. Unlike Dowty’s treatment of lexical semantics, where the decompositional calculus builds on propositional or predicative units (as discussed earlier), a “syntax of event structure” makes explicit reference to quantified events as part of the word meaning. Pustejovsky further introduces a tree structure to represent the temporal ordering and dominance constraints on an event and its subevents. For example, a predicate such as build is associated with a complex event such as that shown in the following (cf. also Moens and Steedman 1988).

The Structure of Meaning (14) [transition [e1:PROCESS ] [e2:STATE ] ]

The process consists of the building activity itself, while the State represents the result of there being the object built. Grimshaw (1990) adopts this theory in her work on argument structure, where complex events such as break are given a similar representation. In such structures, the process consists of what x does to cause the breaking, and the state is the resultant state of the broken item. The process corresponds to the outer causing event as discussed earlier, and the state corresponds in part to the inner change of state event. Both Pustejovsky and Grimshaw differ from earlier authors in assuming a specific level of representation for event structure, distinct from the representation of other lexical properties. Furthermore, they follow J. Higginbotham (1989) in adopting an explicit reference to the event place in the verbal semantics. Recently, Levin and Rappaport (2001, 2005) have adopted a large component of the event structure model for their analysis of verb meaning composition.

2.4 Noun Meaning Thus far, we have focused on the lexical semantics of verb entries. All of the major categories, however, are encoded with syntactic and semantic feature structures that determine their constructional behavior and subsequent meaning at logical form. In Generative Lexicon Theory (Pustejovsky, 1995), it is assumed that word meaning is structured on the basis of four generative factors (qualia roles) that capture how humans understand objects and relations in the world and provide the minimal explanation for the linguistic behavior of lexical items (these are inspired in large part by Moravcsik’s (1975, 1990) interpretation of Aristotelian aitia). These are: the formal role: the basic category that distinguishes the object within a larger domain; constitutive role: the relation between an object and its constituent parts; the telic role: its purpose and function; and the agentive role: factors involved in the object’s origin or “coming into being.” Qualia structure is at the core of the generative properties of the lexicon, since it provides a general strategy for creating new types. For example, consider the properties of nouns such as rock and chair. These nouns can be distinguished on the basis of semantic criteria that classify them in terms of general categories, such as natural kind or artifact object. Although very useful, this is not sufficient to discriminate semantic types in a way that also accounts for their grammatical behavior. A crucial distinction between rock and chair concerns the properties that differentiate natural kinds from artifacts : Functionality plays a crucial role in the process of individuation of artifacts, but not of natural kinds. Th is is reflected in grammatical behavior, whereby “a good chair” or “enjoy the chair” are wellformed expressions reflecting the specific purpose for which an artifact is designed, but “good rock” or “enjoy a rock” are semantically ill-formed since for rock the functionality (i.e., telic) is undefi ned. Exceptions exist when new concepts are referred to, such as when the object is construed relative to a specific activity, for example, as in “The climber enjoyed that rock”; rock itself takes on a new meaning, by virtue of having telicity associated with it, and this is accomplished by integration with the semantics of the subject noun phrase (NP).

Although chair and rock are both physical objects, they differ in their mode of coming into being (i.e., agentive): Chairs are man-made; rocks develop in nature. Similarly, a concept such as food or cookie has a physical manifestation or denotation, but also a functional grounding pertaining to the relation of “eating.” These apparently contradictory aspects of a category are orthogonally represented by the qualia structure for that concept, which provides a coherent structuring for different dimensions of meaning.

2.5 The Problem of Polysemy Given the compactness of a lexicon relative to the number of objects and relations in the world, and the concepts we have for them, lexical ambiguity is inevitable. Add to this the cultural, historical, and linguistic blending that contributes to the meanings of our lexical items, and ambiguity can appear arbitrary as well. Hence, homonymy – where one lexical form has many meanings – is to be expected in a language. Examples of homonyms are illustrated in the following sentences: (15) a. Mary walked along the bank of the river. b. She works for the largest bank in the city. (16) a. The judge asked the defendant to approach the bar. b. The defendant was in the pub at the bar.

Weinreich (1964) calls such lexical distinctions contrastive ambiguity, where it is clear that the senses associated with the lexical item are unrelated. For this reason, it is generally assumed that homonyms are represented as separate lexical entries within the organization of the lexicon. Th is accords with a view of lexical organization that has been termed a sense enumeration lexicon (cf. Pustejovsky 1995). Words with multiple senses are simply listed separately in the lexicon, but this does not seem to compromise or complicate the compositional process of how words combine in the interpretation of a sentence. Th is model becomes difficult to maintain, however, when we consider the phenomenon known as polysemy. Polysemy is the relationship that exists between different senses of a word that are related in some logical manner, rather than arbitrarily, as in the previous examples. It is illustrated in the following sentences (cf. Apresjan 1973; Pustejovsky 1991a, 1998). (17) a. Mary carried the book home. b. Mary doesn’t agree with the book. (18) a. b. (19) a. b.

Mary has her lunch in her backpack. Lunch was longer today than it was yesterday. The fl ight lasted three hours. The fl ight landed on time in Los Angeles.

Notice that in each of these pairs, the same nominal form is assuming different semantic interpretations relative to its selective context. For example, in (17a), the noun book refers to a physical object, while in (17b), it refers to the informational content. In (18a), lunch refers to the physical manifestation of the food, while in (18b), it refers to the eating event. Finally, in (19a), flight refers to the fl ying event, while in (19b), it refers to the plane. Th is phenomenon of regular (or logical) polysemy is one of the most challenging in semantics

27

The Cambridge Encyclopedia of the Language Sciences and has stimulated much research recently (Bouillon 1997; Bouillon and Busa 2001; Cooper 2006). The determination of what such lexical items denote will of course have consequences for one’s theory of compositionality, as we will see in a later section.

3 BUILDING SENTENCE MEANINGS 3.1 Function Application The principle of compositionality follows the view that syntax is an initial guide to the interpretation process. Hence, there would appear to be a strong relationship between the meaning of a phrase and where it appears in a sentence, as is apparent from grammatical function in the following sentences. (20) a. The woman loves the child. b. The child loves the woman.

However, this is not always a reliable association, as seen in languages that have freer word order restrictions, such as German. (21) a. Die Frau liebt das Kind. The woman loves the child. b. Das Kind liebt die Frau. The child loves the woman.

In German, both word orders are ambiguous, since information about the grammatical case and gender of the two NPs is neutralized. Although there is often a correlation between the grammatical relation associated with a phrase and the meaning assigned to it, this is not always a reliable association. Subjects are not always “doers” and objects are not always “undergoers” in a sentence. For example, notice how in both (22a) and (22b), the NP the watch is playing the same role; that is, it is “undergoing a change,” even though it is the subject in one sentence and the object in the other. (22) a. The boy broke the watch. b. The watch broke.

To handle such verbal alternations compositionally requires either positing separate lexical entries for each syntactic construction associated with a given verb, or expressing a deeper relation between different verb forms. For most semantic theories, the basic mechanism of compositionality is assumed to be function application of some sort. A rule of application, apply, acts as the glue to assign (or discharge) the argument role or position to the appropriate candidate phrase in the syntax. Thus, for a simple transitive sentence such as (23a), two applications derive the propositional interpretation of the sentence in (23d). (23) a. b. c. d.

John loves Mary. love(Arg1,Arg2) APPLY love(Arg1,Arg2) to Mary = love(Arg1,Mary) APPLY love(Arg1,Mary) to John = love(John,Mary)

One model used to defi ne the calculus of compositional combinations is the λ-calculus (Barendregt 1984). Using the

28

Figure 2. language of types, we can express the rule of APPLY as a property associated with predicates (or functions), and application as a relationship between expressions of specific types in the language. (24) Function Application: If α is of type a, and β is of type a → b, then β(α) is of type b.

Viewed as typed expressions, the separate linguistic units in (23a) combine as function application, as illustrated in Figure 2. As one can see, the λ-calculus is an expressive mechanism for modeling the relation between verbs and their arguments interpreted as function application. One important extension to the type language used here provides a compositional analysis of the semantics of propositional attitude verbs, such as believe and think (Montague 1973). The sentential complements of such verbs, as is well known, create opaque contexts for substitutions under identity. For example, if Lois is unaware of Superman’s true identity, then the belief statement in (25b) is false, even though (25a) is true. (25) a. Lois believes Superman rescued the people. b. Lois believes Clark Kent rescued the people.

On this view, verbs such as believe introduce an intensional context for the propositional argument, instead of an extensional one. In such a context, substitution under identity is not permitted without possibly affecting the truth value (truth conditional semantics). Th is is an important contribution to the theory of meaning, in that a property of opacity is associated with specific types within a compositional framework. One potential challenge to a theory of function application is the problem of ambiguity in language. Syntactic ambiguities arise because of the ways in which phrases are bracketed in a sentence, while lexical ambiguity arises when a word has multiple interpretations in a given context. For example, in the following sentence, the verb treat can mean one of two things: (26) The doctor treated the patient well.

Either 1) the patient is undergoing medical care, or 2) the doctor was kind to the patient. More often than not, however, the context of a sentence will eliminate such ambiguities, as shown in (27). (27) a. The doctor treated the patient with antibiotics. (Sense 1) b. The doctor treated the patient with care. (Sense 2)

In this case, the interpretation is constructed from the appropriate meaning of the verb and how it combines with its arguments.

The Structure of Meaning 3.2 Quantifiers and Scope Another type of ambiguity, one that is not associated with the constituent structure of the sentence or lexical senses in any obvious way, involves quantified noun phrases (e.g., every cookie, some cake, and most pies). It is interesting that when a sentence has more than one of these phrases, one often sees more than one interpretation possible because of the ways the quantified NPs relate to each other. Th is is not the case in the following sentence, however, where there is only one interpretation as to what happened with the cookie. (28) Some student ate a cookie.

Now consider the sentences in (29), where there is a combination of a some-NP and an every-NP. (29) a. Every student saw a movie. b. Every cookie was eaten by a student.

(32) a. Every woman sang a song. b. ∀x∃y[woman(x) → [song ( y) & sang (x, y)]] c. ∃y∀x[[song ( y) & woman(x)] → sang (x, y)]]

An alternative treatment for handling such cases is to posit a rule of quantifier raising, where the scope ambiguity is reduced to a difference in syntactic structures associated with each interpretation (May 1985).

3.3 Semantic Modification In constructing the meaning of expressions, a semantic theory must also account for how the attribution of properties to an entity is computed, what is known as the problem of modification. The simplest type of modification one can imagine is intersective attribution. Notice that in the phrases in (33), the object denoted correctly has both properties expressed in the NP:

The sentence in (29a) can mean one of two things: 1) that there was one movie, for example, Star Wars, that every student saw; or 2) that everyone saw a movie, but it didn’t have to be the same one. Similarly, for (29b), there could be one student who ate all the cookies, or each cookie that was eaten by a different student. Th is kind of quantifier scope ambiguity has to be resolved in order to determine what kind of inferences one can make from a sentence. Syntax and semantics must interact to resolve this kind of ambiguity, and it is the theory of sentence meaning that defi nes this interaction (cf. quantification). One of the roles of semantic theory is to correctly derive the entailment relations associated with a sentence’s logical form, since this has an obvious impact on the valid reasoning patterns in the language. How these interpretations are computed has been an area of intense research, and one of the most influential approaches has been the theory of generalized quantifiers (cf. Barwise and Cooper 1981). On this approach, the denotation of an NP is treated as a set of sets of individuals, and a sentence structure such as [NP VP] is true if and only if the denotation of the VP is a member of the family of sets denoted by the NP. That is, the sentence in (30) is true if and only if singing (the denotation of the VP) is a member of the set of properties denoted by every woman.

(33) a. black coffee λ x[black (x) & coffee (x)] b. Italian singer λ x[Italian(x) & singer (x)] c. metal cup λ x[metal(x) & cup (x)]

(30) Every woman sang.

In each of these sentences, good is a manner modifier whose interpretation is dependent on the noun it modifies; in (35a), it means “to teach well”; in (35b), it means a “tasty meal”; and in (35c), it means “something keeping you dry.” Similar remarks hold for the adjective dangerous.

On this view, quantifiers such as most, every, some, and so on are actually second-order relations between predicates, and it is partly this property that allows for the compositional interpretation of quantifier scope variation seen previously. The intended interpretation of (30) is (31b), where the subject NP every woman is interpreted as a function, taking the VP as its argument. (31) a. ∀λ P x[woman(x) → P (x)](sang) b. ∀x[woman(x) → sang (x)]

When combined with another quantified expression, as in (32a), the relational interpretation of the generalized quantifiers is crucial for being able to determine both scope interpretations shown in (32).

There are two general solutions to computing the meaning of such expressions: a) Let adjectives be functions over common noun denotations, or b) let adjectives be normal predicates, and have a semantic rule associated with the syntax of modification. Computing the proper inferences for relative clauses will involve a similar strategy, since they are a sort of intersective modification. That is, for the relative clause in (34), the desired logical form will include an intersection of the head noun and the relation predicated in the subordinated clause. (34) a. writer who John knows b. λx[writer (x) & know ( j, x)]

Unfortunately, however, most instances of adjectival modification do not work so straightforwardly, as illustrated in (35). Adjectives such as good, dangerous, and fast modify polysemously in the following sentences. (35) a. John is a a good teacher. b. A good meal is what we need now. c. Mary took a good umbrella with her into the rain.

(36) a. Th is is a dangerous road at night. b. She used a dangerous knife for the turkey.

That is, the road is dangerous in (36a) when “one drives on it,” and the knife is dangerous in (36b) when “one cuts with it.” Finally, the adjective fast in the following sentences acts as though it is an adverb, modifying an activity implicit in the noun, that is, programming in (37a) and driving in (37b). (37) a. Mary is the fastest programmer we have on staff. b. The turnpike is a faster road than Main Street.

29

The Cambridge Encyclopedia of the Language Sciences To account for such cases, it is necessary to enrich the mode of composition beyond simple property intersection, to accommodate the context dependency of the interpretation. Analyses taking this approach include Borschev and Partee (2001), Bouillon (1997), and Pustejovsky (1995).

3.4 Arguments versus Adjuncts In our discussion thus far of how predicates select arguments to create compositionally complex expressions, we have assumed that the matrix predicate (the main verb of the sentence) acts as the only function over other phrases. In fact, what an argument of the verb is and what an adjunct is are questions just as much of meaning as of syntax. In this section, we examine the semantic issues involved. In this overview, we have adopted the position that language reflects the workings of our deeper conceptual systems in some direct and nonidiosyncratic manner. Lexical choice as well as specific grammatical phenomena can be constrained by underlying conceptual bias. Well-known examples of this transparency include count/mass noun distinctions in the lexicon, and case marking and valence distinctions in the syntax. For example, concepts entailing unindividuated stuff or material will systematically be semantically typed as mass nouns in the grammar, whereas naturally individuating (countable) substances will assume the status of count nouns, with their respective grammatical consequences, as illustrated in (38). (Some mass terms are not shared by all languages, such as the concept of “paper” or “furniture.”) (38) a. {not much/all/lots of } gold/water/dirt/sand b. {every/two/several} chairs/girls/beaches

Similarly, as presented in previous sections, the classification of verbs appears to reflect their underlying relational structure in fairly obvious ways. (39) a. Mary arrived. b. John greeted Mary. c. Mary gave a book to John.

That is, the argument structure of each verb encodes the semantics of the underlying concept, which in turn is reflected in the projection to the specific syntactic constructions, that is, as intransitive, transitive, and ditransitive constructions, respectively. For unary, binary, and ternary predicates, there is a visible or transparent projection to syntax from the underlying conceptual structure, as well as a predictable compositional derivation as function application. So, the question arises as to what we do with nonselected arguments and adjuncts within the sentence. It is well known, for example, that arguments not selected by the predicate appear in certain contexts (cf. Jackendoff 1992; Levin and Rappaport Hovav 2005). (40) a. The man laughed himself sick. b. The girl danced her way to fame. c. Mary nailed the window shut.

30

Each of the italicized phrases is an argument of something, but is it selected by the matrix predicate? Jackendoff has proposed a solution that relies on the notion of construction, as introduced by A. E. Goldberg (1995) (cf. construction grammars). Another problem in compositionality emerges from the interpretation of adjuncts. The question posed by the examples in (41) is this: Which NPs are arguments semantically and which are merely adjuncts? (41) a. b. c. d.

Mary ate the soup. Mary ate the soup with a spoon. Mary ate the soup with a spoon in the kitchen. Mary ate the soup with a spoon in the kitchen at 3:00 p.m.

For Davidson (1967), there is no semantic distinction between arguments and adjuncts in the logical form. Under his proposal, a two-place predicate such as eat contains an additional argument, the event variable, e, which allows each event participant a specific role in the interpretation (cf. Parsons 1990; event structure and grammar). (42) λyλxλe[eat(e, x, y)]

Then, any additional adjunct information (such as locations, instruments, etc.) is added by conjunction to the meaning of the main predicate, in a fashion similar to the interpretation of intersective modification over a noun. In this manner, Davidson is able to capture the appropriate entailments between propositions involving action and event expressions through conventional mechanisms of logical entailment. For example, to capture the entailments between (41b–d) and (41a) in the following, each more specifically described event entails the one above it by virtue of conjunction elimination (already encountered) on the expression. (43) a. ∃e[eat(e, m, the-soup)] b. ∃e[eat(e, m, the-soup) & with(e, a spoon)] c. ∃e[eat(e, m, the-soup) & with(e, a spoon) & in(e, the kitchen)] d. ∃e[eat(e, m, the-soup) & with(e, a spoon) & in(e, the kitchen) & at (e, 3:00 p.m.)]

Th is approach has the advantage that no special inference mechanisms are needed to derive the entailment relations between the core propositional content in (43a) and forms modified through adjunction. Th is solution, however, does not extend to cases of verbs with argument alternations that result in different meanings. For example, how do we determine what the core arguments are for a verb like sweep? (44) a. b. c. d. e. f.

John swept. John swept the floor. John swept the dirt. John swept the dirt off the sidewalk. John swept the floor clean. John swept the dirt into a pile.

The semantics of such a verb should determine what its arguments are, and how the different possible syntactic realizations relate to each other semantically. These cases pose an interest-

The Structure of Meaning ing challenge for the theory of compositionality (cf. Jackendoff 2002).

3.5 Presupposition In computing the meaning of a sentence, we have focused on that semantic content that is asserted by the proposition. Th is is in contrast to what is presupposed. A presupposition is that propositional meaning that must be true for the sentence containing it to have a proper semantic value (Stalnaker 1970; Karttunen 1974; Potts 2005). (Stalnaker makes the distinction between what a speaker says and what a speaker presupposes.) Such knowledge can be associated with a word, a grammatical feature, or a syntactic construction (so-called presupposition triggers). For example, in (45) and (46), the complement proposition to each verb is assumed to be true, regardless of the polarity assigned to the matrix predicate. (45) a. Mary realized that she was lost. b. Mary didn’t realize that she was lost. (46) a. John knows that Mary is sick. b. John doesn’t know that Mary is sick.

There are similar presuppositions associated with aspectual predicates, such as stop and finish, as seen in (47). (47) a. Fred stopped smoking. b. John fi nished painting his house.

In these constructions, the complement proposition is assumed to have been true before the assertion of the sentence. Such conventional presuppositions are also triggered by interrogative contexts, such as seen in (48). (48) a. Why did you go the store? b. When did you see Mary?

As with all presuppositions, however, they are defeasible, as the answer to (48b) in (49) illustrates. (49) But I didn’t see Mary.

Conversational presuppositions, on the other hand, are implicated propositions by virtue of a context and discourse situation. The response in (50b) conversationally implicates that I am not hungry (Recanati 2002); conversational implicature). (50) a. Are you hungry? b. I’ve had a very large breakfast.

The meaning of such implicatures is not part of the asserted content of the proposition, but computed within a conversational context in a discourse. We will return to this topic in a later section.

3.6 Noncompositionality While semantic theory seems to conform to the principles of compositionality in most cases, there are many constructions that do not fit into the conventional function application paradigm. A phrase is noncompositional if its meaning cannot

be predicted from the meaning of its parts. We have already encountered modification constructions that do not conform to simple intersective interpretations, for example, good teacher. There are two other constructions that pose a problem for the principle of compostionality in semantics: (51) a. Idioms: hear it through the grapevine, kick the bucket ; b. Coercions: begin the book , enjoy a coffee.

The meaning of an idiom such as leave well enough alone is in no transparent way composed of the meanings of its parts. Although there are many interesting syntactic properties and constraints on the use of idiomatic expressions in languages, from a semantic point of view its meaning is clearly associated with the entire phrase. Hence, the logical form for (52), (52) Every person kicked the bucket.

will make reference to quantification over “persons,” but not over “buckets” (cf. [53]). (53) ∃x[ person(x) & kick.the.bucket (x)]

We confront another kind of noncompositionality in semantics when predicates seem to appear with arguments of the “wrong type.” For example, in (54a), a countable individual entity is being “coerced” into the food associated with that animal, namely, bits of chicken, while in (54b), the mass terms water and beer are being packaged into unit measures (Pelletier, 1975). In (55), the aspectual verbs normally select for an event, but here are coercing entities into event denotations. Similarly, in (56), both object NPs are being coerced into propositional interpretations. (Cf. Pustejovsky 1995 and Jackendoff 2002 for discussions of coercion phenomena and their treatment.) (54) a. There’s chicken in the salad. b. We’ll have a water and two beers. (55) a. Roser fi nished her thesis. b. Mary began the novel. (56) a. Mary believes John’s story. b. Mary believes John.

These examples illustrate that semantics must accommodate specific type-shifting and coercing operations in the language in order to remain compositional. In order to explain just such cases, Pustejovsky (2007) presents a general theory of composition that distinguishes between four distinct modes of argument selection: a) function application, b) accommodation, c) coercion by introduction, and d) coercion by exploitation.

4 DISCOURSE STRUCTURE Thus far we have been concentrating on the meaning of single sentences. But no sentence is really ever uttered outside of a context. Language is used as a means of communication and is as much a way of acting as a means of representing (Austin 1975; Searle 1969). In this section, we briefly survey the major areas of research in discourse semantics. We begin by examining the semantic models that have emerged to account for “dynamic phenomena” in discourse, such as intersentential

31

The Cambridge Encyclopedia of the Language Sciences anaphora. We then look at how discourse relations can be used to model larger units of meaning. From our previous discussion, we have assumed the sentence as the unit for semantic interpretation, including the level for the interpretation of quantifier scope and anaphoric binding, as in (57). (57) a. Every actress said she was happy. b. Every actress came in and said hello.

Notice that the anaphoric link between the quantifier and the pronoun in (57a) is acceptable, while such a binding is not possible within a larger discourse setting, as in (58) and (59). (58) a. Every actress came in. b. *She said she was happy. (59) a. Every actress came in. b. *She said hello.

So, in a larger unit of semantic analysis, a bound variable interpretation of the pronoun does not seem permitted. Now notice that indefi nites do in fact allow binding across the level of the sentence. (60) a. An actress came in. b. She said hello.

The desired interpretation, however, is one that the semantic model we have sketched out is unable to provide. (61) a. ∃x[actress (x) & come.in(x)] b. [& say.hello (x)]

What this example points out is that the view of meaning we have been working with so far is too static to account for phenomena that are inherently dynamic in nature (Chierchia 1995; Groenendijk and Stokhof 1991; Karttunen 1976). In this example, the indefi nite NP “an actress” is being used as a discourse referent, and is available for subsequent reference as the story unfolds in the discourse. Following Kamp and Reyle’s (1993) view, an indefi nite NP introduces a “novel discourse referent,” while a pronoun or definite description says something about an existing discourse referent. Using the two notions of novelty and familiarity, we can explain why she in (60b) is able to bind to the indefi nite; namely, she looks for an accessible discourse referent, the indefi nite. The reason that (58) and (59) are not good discourses is due to the universally quantified NP “every actress,” which is inaccessible as an antecedent to the pronoun. One influential formalization of this approach is Dynamic Predicate Logic (Groenendijk and Stokhof, 1991), which combines conventional interpretations of indefi nites as existentials with the insight from incremental interpretations, mentioned previously. On this view, the interpretation of a sentence is a function of an ordered pair of assignments, rather than a static single assignment. The “output condition” for a sentence with an indefi nite NP, such as (60a), specifies that a subsequent sentence with a pronoun can share that variable assignment: “The meaning of a sentence lies in the way it changes the representation of the information of the interpreter” (ibid.). That is, when a quantified expression is used in discourse, something new

32

is added to the listener’s interpretation state so that the listener can use the quantifier to help understand future utterances. In this way, the meaning of a sentence is interpreted dynamically. The dynamics of discourse, of course, involve more than the binding of anaphors to antecedents across adjacent sentences. Every utterance is made in the context of a common ground of shared knowledge (presuppositions), with a communicative intent, and in a particular time and place (cf. discourse analysis, communicative intention). Just as sentences have internal structure, with both syntactic and semantic dependencies, discourse can also be viewed as a sequence of structured segments, with named dependencies between them. For example, the sentences in (62) form a discourse structured by a relation of narration, implying temporal sequence (Dowty, 1986). (62) a. John entered the room. b. He sat down.

In (63), on the other hand, the two sentences are related by the dependency of explanation, where (63b) temporally precedes and explains (63a). (63) a. Max fell. b. John pushed him.

Theories of discourse relations, such as rhetorical structure theory (Mann and Thompson 1986), segmented discourse representation theory (SDRT) (Asher and Lascarides 3), and that of Hobbs (1985) attempt to model the rhetorical functions of the utterances in the discourse (hence, they are more expressive of discourse structure and speaker intent than discourse representation theory [DRT], which does not model such parameters). For the simple discourses above, SDRT, for example extends the approach from dynamic semantics with rhetorical relations and their semantic values, while providing a more complex process of discourse updates. Rhetorical relations, as used in SDRT, carry specific types of illocutionary force (cf. Austin 1975; Searle 1969, 9), namely, explanation, elaboration, giving backgrounds, and describing results.

5 CONCLUSION In this essay, I have attempted to outline the basic components for a theory of linguistic meaning. Many areas of semantics were not touched on in this overview, such as issues relating to the philosophy of language and mind and the psychological consequences of various semantic positions. Many of the accompanying entries herein, however, address these issues directly. WORKS CITED AND SUGGESTIONS FOR FURTHER READING Apresjan, J. D. 1973. “Synonymy and synonyms.” In Trends in Soviet Theoretical Linguistics, ed. F. Kiefer, 173–99. Dordrecht, the Netherlands: Reidel. Asher, Nicholas, and Alex Lascarides. 2003. Logics of Conversation. Cambridge: Cambridge University Press. Austin, J. L. 1975. How to Do Things with Words. Cambridge: Harvard University Press.

The Structure of Meaning Barendregt, Henk. 1984. The Lambda Calculus, Its Syntax and Semantics. Amsterdam: North-Holland. Barwise, Jon, and Robin Cooper. 1981. “Generalized quantifiers and natural language.” Linguistics and Philosophy 4.1: 159 –219. Borschev, Vladimir, and Barbara H. Partee. 2001. “Genitive modifiers, sorts, and metonymy.” Nordic Journal of Linguistics 24.2: 140 –60. Bouillon, P. 1997. Polymorphie et semantique lexicale: Le cas des adjectifs. Lille: Presses Universitaires du Spetentrion. Bouillon, P., and F. Busa, eds. 2001. The Language of Word Meaning. Cambridge: Cambridge University Press. Bresnan, J., ed. 1982. The Mental Representation of Grammatical Relations. Cambridge, MA : MIT Press. Briscoe, T., V. de Paiva, and A. Copestake, eds. 1993. Inheritance, Defaults, and the Lexicon. Cambridge: Cambridge University Press. Busa, F. 1996. “Compositionality and the semantics of nominals.” Ph.D. diss., Brandeis University. Carlson, G. 1984. “Thematic roles and their role in semantic interpretation.” Linguistics 22: 259 –79. Carpenter, B. 1992. The Logic of Typed Feature Structures. Cambridge: Cambridge University Press. Carter, R. [1976] 1988. On Linking: Papers by Richard Carter. Ed. B. Levin and C. Tenny. MIT Lexicon Project Working Papers 25, Center for Cognitive Science. Cambridge, MA : MIT Press. Chierchia, G. 1989. “Structured meanings, thematic roles, and control.” In Properties, Types, and Meaning, ed. G. Chierchia, B. Partee, and R. Turner, II: 131–66. Dordrecht, the Netherlands: Kluwer Academic Publishers. ———. 1995. The Dynamics of Meaning. Chicago: University of Chicago Press. Chomsky, N. [1955] 1975. The Logical Structure of Linguistic Theory. Chicago: University of Chicago Press. ———. 1965. Aspects of the Theory of Syntax . Cambridge, MA : MIT Press. Comrie, Bernard. 1981. Language Universals and Linguistic Typology: Syntax and Morphology. Chicago: University of Chicago Press. Cooper, Robin . 2006 . “A record type theoretic account of copredication and dynamic generalized quantification.” In Kvantifikator for en Dag, Essays Dedicated to Dag Westerstahl on His Sixtieth Birthday. Avilable online at: http://www.phil.gu.se/posters/ festskrift3/. Copestake, A. 1992. The Representation of Lexical Semantic Information . Cognitive Research Paper CSRP 280, School of Cognitive and Computing Science, University of Sussex , Brighton, England. ———. 1993. “Defaults in Lexical Representation.” In Inheritance, Defaults, and the Lexicon, ed. Ted Briscoe, Valeria de Paiva, and Ann Copestake, 223–45. Cambridge: Cambridge University Press. Copestake, A., and E. Briscoe. 1992. “Lexical operations in a unification-based framework.” In Lexical Semantics and Knowledge Representation, ed. J. Pustejovsky and S. Bergler, 101–19. New York : Springer Verlag. Copestake, A., and T. Briscoe. 1995. “Semi-productive polysemy and sense extension.” Journal of Semantics 12: 15 –67. Davidson, D. 1967. “The logical form of action sentences.” In The Logic of Decision and Action, ed. N. Rescher, 81–95. Pittsburgh: Pittsburgh University Press. Davidson, D., and G. Harman, eds. 1972. Semantics of Natural Language. Dordrecht, the Netherlands: Reidel. Davis, Steven, and Brendan Gillon. 2004. Semantics: A Reader. Oxford: Oxford University Press. Dolling , J. 1992. “Flexible Interpretationen durch Sortenverschiebung.” In Fügungspotenzen, ed. Ilse Zimmermann and Anatoli Strigen, 23–62. Berlin: Akademie Verlag.

Dowty, D. R . 1979. Word Meaning and Montague Grammar. Dordrecht, the Netherlands: D. Reidel. ———. 1986. “The effects of aspectual class on the temporal structure of discourse: Semantics or pragmatics. Linguistics and Philosophy 9.1: 37–61. ———. 1991. “Thematic proto-roles and argument selection.” Language 67: 547–619. Evans, R., and G. Gazdar. 1990. “The DATR papers: February 1990.” Cognitive Science Research Paper CSRP 139, School of Cognitive and Computing Science, University of Sussex, Brighton, England. Fillmore, C. 1968. “The case for case.” In Universals in Linguistic Theory, ed. E. Bach and R. Harms, 1–88. New York : Holt, Rinehart, and Winston. Gazdar, G., E. Klein, G. Pullum, and I. Sag. 1985. Generalized Phrase Structure Grammar. Cambridge: Harvard University Press. Goldberg , A. E. 1995. Constructions: A Construction Grammar Approach to Argument Structure. Chicago: University of Chicago Press. Grice, H. P. 1969. “Utterer’s meaning and intentions.” Philosophical Review 78: 147–77. ———.1989. Studies in the Way of Words. Cambridge: Harvard University Press. Grimshaw, J. 1979. “Complement selection and the lexicon” Linguistic Inquiry 10: 279 –326. ———. 1990. Argument Structure. Cambridge, MA : MIT Press. Groenendijk , Jeroen, and Martin Stokhof. 1991. “Dynamic predicate logic.” Linguistics and Philosophy 14.1: 39–100. Gruber, J. S. 1976. Lexical Structures in Syntax and Semantics. Amsterdam: North-Holland. Gunter, C. 1992. Semantics of Programming Languages. Cambridge, MA : MIT Press. Hale, K. and S. J. Keyser. 1993. “On argument structure and the lexical expression of syntactic relations.” In The View from Building 20: Essays in Honor of Sylvain Bromberger, ed. K. Hale and S. J. Keyser, 53–109. Cambridge, MA : MIT Press. Heim, Irene. 1982. The semantics of definite and indefinite noun phrases. Ph.D. thesis, University of Massachussets, Amherst. Higginbotham, J. 1989. “Elucidations of meaning.” Linguistics and Philosophy 12: 465 –517. Hobbs, Jerry R. 1985. “On the coherence and structure of discourse.” Report No. CSLI-85–37, Center for the Study of Language and Information, Stanford University. Jackendoff, R. 1972. Semantic Interpretation in Generative Grammar. Cambridge, MA : MIT Press. ———. 1983. Semantics and Cognition. Cambridge, MA : MIT Press. ———. 1990. Semantic Structures. Cambridge, MA: MIT Press. ———. 1992. “Babe Ruth homered his way into the hearts of America.” In Syntax and the Lexicon, ed. T. Stowell and E. Wehrli, 155–78. San Diego: Academic Press. ———. 1997. The Architecture of the Language Faculty. Cambridge, MA: MIT Press. ———. 2002. Foundations of Language. Oxford: Oxford University Press. Kamp, H. 1981. “A theory of truth and semantic representation.” In Formal Methods in the Study of Language, ed. J. A. G. Groenendijk , T. M. V. Janssen, and M. B. J. Stokhof. Mathematical Centre Tracts 135, 277–322. Amsterdam: Mathematical Centre. Kamp, H., and U. Reyle. 1993. From Discourse to Logic. Dordrecht, the Netherlands: Kluwer Academic Publishers. Karttunen, L. 1974. “Presupposition and linguistic context.” Theoretical Linguistics 1: 181–93. ———. 1976. “Discourse referents.” In Syntax and Semantics, ed. James D. McCawley. Vol. 7: Notes from the Linguistic Underground , 363–85. New York : Academic Press.

33

The Cambridge Encyclopedia of the Language Sciences Katz, J., and J. Fodor. 1963. “The structure of a semantic theory.” Language 39: 170 –210. Krifka, M. 1992. “Thematic relations as links between nominal reference and temporal constitution.” In Lexical Matters, CSLI Lecture Notes, ed. I. Sag and A. Szabolcsi, 29–53. Chicago: University of Chicago Press. Kripke, Saul. [1972] 1980. “Naming and necessity.” In Semantics of Natural Language, ed. D. Davidson and G. Harman, 253–355. Dordrecht and Boston: Reidel. Lakoff, G. [1965] 1970. Irregularity in Syntax . New York : Holt, Rinehart, and Winston. Levin, B., and M. Rappaport Hovav. 1995. Unaccusativity: At the SyntaxSemantics Interface. Cambridge, MA : MIT Press. ———. (2005. Argument Realization. Cambridge: Cambridge University Press. Lewis, David. 1976. “General semantics.” In Montague Grammar, ed. Barbara H. Partee, 1–50. New York : Academic Press. Lyons, John. 1968. Introduction to Theoretical Linguistics. Cambridge: Cambridge University Press. Mann, William C., and Sandra A. Thompson. 1986. Rhetorical Structure Theory: Description and Construction of Text Structures. ISI/RS-86–174, 1–15. Nijmegen, the Netherlands: Information Sciences Institute. May, Robert. 1985. Logical Form: Its Structure and Derivation. Cambridge, MA : MIT Press. McCawley, J. D. 1968. “The role of semantics in a grammar.” In Universals in Linguistic Theory, ed. E. Bach and R. T. Harms, 124–69. New York : Holt, Rinehart, and Winston. Moens, M., and M. Steedman. 1988. “Temporal ontology and temporal reference.” Computational Linguistics 14: 15 –28. Montague, Richard. 1973. “The proper treatment of quantification in ordinary English.” In Approaches to Natural Language, ed. Jaakko Hintikka, Julius Matthew, Emil Moravcisk, and Patrick Suppes, 221–42. Dordrecht, the Netherlands: D. Reidel. Repr. in Montague 1974, 247–70. ———. 1974. Formal Philosophy: Selected Papers of Richard Montague. New Haven, CT: Yale University Press. Moravcsik , J. M. 1975. “Aitia as generative factor in Aristotle’s philosophy.” Dialogue 14: 622–36. ———. 1990. Thought and Language. London: Routledge. Nunberg , G. 1979. “The non-uniqueness of semantic solutions: Polysemy.” Linguistics and Philosophy 3: 143–84. Parsons, T. 1990. Events in the Semantics of English. Cambridge, MA: MIT Press. Partee, Barbara. 1984. “Compositionality.” In Varieties of Formal Semantics, ed. Fred Landman and Frank Veltman, 281–311. Dordrecht, the Netherlands: Foris.

34

Partee, Barbara , ed. 1976 . Montague Grammar. New York : Academic Press . Partee, B., and M. Rooth. 1983. “Generalized conjunction and type ambiguity.” In Meaning, Use, and Interpretation of Language, ed. Rainer Bäuerle, Christoph Schwarze, and Arnim von Stechow, 361–83. Berlin: Walter de Gruyter. Pelletier, F. J. 1975. “Non-singular reference: Some Preliminaries.” In Mass Terms: Some Philosophical Problems, ed. F. J. Pelletier, 1–14. Dordrecht, the Netherlands: Reidel. Pollard, C., and I. Sag. 1994. Head-Driven Phrase Structure Grammar. Chicago: University of Chicago Press and Stanford CSLI. Potts, Christopher. 2005. The Logic of Conventional Implicatures. Oxford: Oxford University Press. Pustejovsky, J. 1991a. “The generative lexicon.” Computational Linguistics 17: 409 –41. ———. 1991b. “The syntax of event structure.” Cognition 41: 47–81. ———. 1995. The Generative Lexicon. Cambridge, MA: MIT Press. ———. 1998. “The semantics of lexical underspecification.” Folia Linguistica 32: 323–47. ———. 2007. The Multiplicity of Meaning. Cambridge, MA : MIT Press. Pustejovsky, J., and B. Boguraev. 1993. “Lexical knowledge representation and natural language processing.” Artificial Intelligence 63: 193–223. Recanati, Francois. 2002. “Unarticulated constituents.” Linguistics and Philosophy 25: 299 –345. Russell, Bertrand, 1905. “On Denoting.” Mind 14: 479 –93. Saussure, Ferdinand de. [1916] 1983. Course in General Linguistics. Trans. R. Harris. London: Duckworth. Searle, John. 1969. Speech Acts: An Essay in the Philosophy of Language. Cambridge: Cambridge University Press. ———. 1979. Expression and Meaning. Cambridge: Cambridge University Press. Stalnaker, Robert. 1970. “Pragmatics.” Synthese 22.1/2: 272–78. Strawson, P. F. 1971. Logico-Linguistic Papers. London: Methuen. Tenny, C. 1992. “The Aspectual Interface Hypothesis.” In Lexical Matters, CSLI Lecture Notes, ed. I. Sag and A. Szabolcsi, 1–27. Chicago: University of Chicago Press. Tenny, C., and J. Pustejovsky. 2000. Events as Grammatical Objects. Stanford, CA: CSLI Publications. Chicago: University of Chicago Press. Weinreich, U. 1964. “Webster’s Th ird: A Critique of its Semantics.” International Journal of American Linguistics 30: 405 –9. ———. 1972. Explorations in Semantic Theory. The Hague: Mouton. Williams, E. 1981. “Argument structure and morphology.” Linguistic Review 1: 81–114.

diglossia, culture and language, digital media, literacy). This essay is focused on the issue of the media of com-

4 SOCIAL PRACTICES OF SPEECH AND WRITING Florian Coulmas

INTRODUCTION Language is constitutive for human society. As a social fact it cannot be thought of in the abstract, for the medium of communication is what allows it to serve social functions. The nature of the social relationship that exists by virtue of language partially depends on the externalization of language, that is, on how it is transmitted from one actor to another as speech, writing, sign, or Braille. The anatomy of speech organs (cf. Liberman and Blumstein 1988) provides the biological foundation of human society in the most general sense, which is why oral speech is considered fundamental for socialization both in the phylogenetic and ontogenetic sense. But unless we study human society like that of other primates from the point of view of physical anthropology, other forms of language externalization must also be taken into account as communication potential from the beginning. There are two reasons for this. One is that the invention of writing (sign language, Braille) cannot be undone. The other, which follows therefrom, is that writing has brought about basic changes in the nature of human communication. It brought in its wake a literate mindset that cannot be reversed. Research about language in literate societies is carried out by researchers who, growing up, were socialized into a literate world organized by and large on the basis of literate principles. It is not fortuitous, therefore, that social practices of speech and writing are dealt with here under one heading. The scientific enterprise in general, linguistics in particular, is a social practice involving speech and writing. Even the investigation of unwritten languages happens against the background of literate society and by means of the tools developed for what Goody (1977, 151) felicitously called the “technology of the intellect.” For the language sciences, it is important to keep in mind that it is not just the technicians who use a tool to do what they need to do and want to do, but that the tool restricts what can be done. This holds true for the hardware, that is, the writing implements, as well as for the software, the code or the writing systems. The social aspects of speech and writing encompass a wide range of topics many of which are dealt with in other entries of this encyclopedia (sociolinguistics, discourse analysis,

munication and the social conditions and consequences of their evolution. The reason is that the social practices of speech and writing both depend on the available technology and lead to technological and social innovation. As has been argued by Marshall McLuhan (1964), Elizabeth L. Eisenstein (1979), Jan Assmann (1991), David R. Olson (1994), and Nicholas Negroponte (1995) among others, civilizations are characterized by the media they predominantly use and which shape the way they exchange, store, and administer information, thus exercising a profound influence on social practice. The nexus between speech and writing is variable and more obvious in some cases than in others. For instance, when the lyrics of a song are read on the monitor and sung in a karaoke bar, speech and writing are joined together in one activity. On the other hand, the songs of bards handed down by word of mouth from one generation to another are part of the culture of spoken, as opposed to written, language (Ong 1982, Olson 1991; oral culture, oral composition). However, the very idea of orality is predicated on literacy and would not have become an object of study without it. Just as there is no silence without sound, illiteracy exists but in a literate society. On the face of it, many kinds of verbal behavior, such as speech contests, bidding at an auction, and election campaign addresses, do not involve writing. The institutional frameworks in which they take place in modern society, school, trade and government, though, rely to a very large extent on written texts. To analyze social practices of speech and writing, then, it is necessary to consider technological aspects of writing and institutional aspects of literacy.

TECHNOLOGICAL ASPECTS OF WRITING Many social practices and ligatures of contemporary society would be impossible without writing. This does not imply that the externalization of language by technological means is the only force that shaped modern society. The assumption of an unmediated cause-and-effect relationship between writing and social organization, of a watershed between primitive oral life and complex literate civilization, is a simplification that fails to do justice to the complexity of the interaction. It is surely tempting to argue that what all great early civilizations had in common was writing and that it was hence writing that caused complex societies to come into existence. However, if we look at the uses of writing in early civilizations, many differences are apparent. For example, economic administration was preeminent in Mesopotamia (Nissen, Damerow, and Englund 1990), whereas cult stood out in Egypt (Assmann 1991). In both cases, it is untenable to argue that accounting and the cult of the dead, respectively, were an outflow of the invention of writing. Yet the opposite proposition, claiming that the demands of bookkeepers and priests led to the creation of writing, is no less simplistic. Similarly, the invention of the printing press and movable type has often been seen as a technological breakthrough with vast social consequences (Febvre and Martin [1958] 1999; Eisenstein 1979). In our day, the digital turn (Fischer 2006), described as the third media revolution after chirographic culture (Schmandt-Besserat 1992) and print culture

35

The Cambridge Encyclopedia of the Language Sciences (Olson 1994), is regarded as a driving force of globalization (Kressel 2007). Both of these propositions are defensible, but not in a unidirectional, monocausal sense. Equally true are the opposite propositions, that socioeconomic developments led to the emergence of a larger reading public, thus paving the way for a simpler and more efficient reproduction technology than the copying of manuscripts, and that modern industrial society with mass participation generated pressure for the development of a technology of mass dissemination of information. The invention of writing facilitated complex social organization, and the printing press was conducive to the spread of education. However, writing has been a common possession of humanity for more than 5,000 years and the printing press for half a millennium, if we disregard the use of cast-metal movable type in Korea in the early thirteenth century. Yet we are living in a world with hundreds of millions of adult illiterates, even, or rather particularly, where writing first emerged, that is, in Mesopotamia, in Egypt, in China, and in Mesoamerica. According to unesco (2006), there were 781 million adult illiterates worldwide and 100 million school-age children not attending school in 2005. In spite of the uneven distribution of illiterates in the world, these figures suffice to discredit the notion that a new technology of communication of and by itself brings about social change. Economic development, social structure, ethnic and linguistic composition, fecundity, ideology, and tradition are intervening variables that determine how a society makes use of and adjusts to a new technology. It is necessary, therefore, to reckon with the contemporaneity of different levels of development, different technologies, and different literacies. Assuming a dialectic relationship of mutual influence between writing and social change is a more promising approach for understanding the transition from oral to literate society. New technologies both respond to practical needs and create new practices. Any technology is an artifact, but to conclude that its creators rule over it is a fallacy, for the applications of technological innovations are often recognized not in advance but after the fact, when they have been used for some time. Like the genie let out of the bottle, they may have unplanned and sometimes unwelcome consequences. The material and functional properties of writing technologies determine their potential uses, which, however, are not necessarily evident at the outset. The locus of writing is the city. Even a superficial look at present-day urban environments reveals that city dwellers are surrounded by written messages wherever they go. Of late, this has given rise to a new branch of scholarship known as “linguistic landscape research” (Landry and Bourhis 1997; Backhaus 2007), as it were, a social epigraphy for posterity. The variety of writing surfaces on which the literate culture of modern cityscapes manifests itself is striking. It testifies to the traces of history in the present and to the contemporaneity of different stages of development, for it includes some of the oldest materials used for writing side by side with the most recent devices. This contemporaneity is one of the foremost characteristics of writing. For writing arrests change and enables accumulation of information. Some genuine monuments from antiquity speak to us today, such as the Egyptian obelisk of Ramses II of the 19th Dynasty, 1304–1237 b.c.e., re-erected on the Place de la Concorde in the center of Paris. Around the corner, the passerby

36

can read the latest stock quotes off a scrolling light-emitting diode (LED) display. Brand-new buildings are adorned with the old technique of cutting messages in stone. Stelae with commemorative inscriptions, gravestones, and buildings bearing the names of their owners or occupants are still being put up, much as in ancient times. There are hardly any material objects to which writing cannot be affixed. Since the earliest times of literate culture, few have been discarded and many added. The choice continues to expand. Hard surfaces made for endurance are stone, marble, metal, ceramics, wood, and, today, plastics. Inscriptions are incised, engraved, etched, carved, and chiseled into them as they were in the past, and malleable surfaces such as moist clay and molten metal are impressed or molded into shape. In addition to monumental inscriptions, writing is found on various other surfaces, such as whitewashed walls, street signs, posters, billboards, handbills, notice boards, memorial plaques, cloth, clothing, commercials carried around by “sandwichmen” and mounted on trucks, advertising pillars, buses and other vehicles covered with commercials, shop windows, and digital display panels. These and some other surfaces, such as palm leaves, papyrus, parchment, and wax tablets that have gone out of fashion, are variously suitable for realizing the functional potential of writing. Two fundamental functions of writing are memory support and communication. They are not mutually exclusive, but different surfaces lend themselves better to one than to the other. Hard surfaces answer the requirement of durability. They are inscribed only once, but with a lasting trace that can be recovered after years, decades, even millennia. Baked clay tablets, the hallmark of cuneiform civilization, and mural inscriptions on Egyptian monuments embody this type. Memory is in time turned into history, the recording of the past and the collection of knowledge, which are the cardinal functional characteristics of this technology. Inscriptions on hard surfaces are, of course, also communicative but stationary. Clay tablets can be transported in limited numbers only, and monumental inscriptions have to be visited to be read. In order to allow written signs to travel and thus to realize a potential that fundamentally distinguishes writing from speech, freeing the message from the copresence of sender and receiver, lighter materials are needed. In antiquity, three main writing surfaces met this requirement: papyrus, parchment, and paper. For millennia, Egypt was practically the only producer of papyrus because the reed of which it is made grows in abundance along the banks of the Nile. The papyrus scroll hieroglyph is attested in the most ancient known Egyptian inscriptions, and the oldest papyrus fragments covered with writing date from the third millennium b.c.e. Papyrus came to be commonly used for documentary and literary purposes throughout Greece, Asia Minor, and the Roman Empire. As of the fourth century c.e., parchment (processed animal hide), a more durable writing material than the brittle papyrus, began to be more widely used in Europe, where the scroll was gradually edged out by the book in codex form (Roberts and Skeat 1983). The word paper is derived from papyrus, but paper making is quite different from papyrus making. It was invented by the Chinese some 1,900 years ago (Twitchett 1983). The earliest Chinese documents on paper date

Social Practices of Speech and Writing from the second century c.e. In the wake of the Islamic expansion to Central Asia, the Arabs acquired the paper-making technology in the eighth century c.e., which they in turn introduced to Europe in the eleventh century. Relatively cheap, flexible, and convenient to carry, paper replaced parchment as the principal writing surface in Europe and in other parts of the world. Since its invention in China, paper, which Pierre-Marc De Biasi (1999) called “the greatest invention of all time,” gave a boost to the production of written text wherever it was introduced. In China, it was used for block printing as of the seventh century. In the tenth century, the entire Buddhist scripture was printed using 130,000 printing blocks (Taylor and Taylor 1995, 156). Paper was the first writing material that spread around the world. In the West, Johannes Gutenberg’s invention of printing with movable type would hardly have had the same impact without it. Of the 180 copies of the Bible he printed, 150 were on paper and only 30 on parchment, one indication of the importance of paper for the dissemination of written text. Its position in this regard is undiminished. The paperless office is far from a reality even in the most advanced countries; rather, many ancillary devices that presuppose writing on paper form the mainstay of the thriving office machines industry: printers, scanners, copiers, and fax machines. Although nowadays paper holds only a tiny fraction of all new information released, it is still the unchallenged surface for the formal publication of information. World paper consumption for information storage and distribution is at an all-time high. Notwithstanding the shift of many periodicals and scholarly journals to online publication most continue to be printed on paper for archival purposes, for paper has a much longer duration of life than can be guaranteed for any digital storage medium. This brings to light a more general trade-off of information processing. Weight and storage capacity are inversely related. A tablet measuring about 10 × 10 cm is the ideal size for writing on wet clay. It holds about 300 characters. Depending on the thickness of the tablet, this yields an information/weight ratio of .2 kg to 1 kg per 1,000 characters. A text of 300,000 characters would weigh between 200 kg and 1000 kg. In modern terms, that would be a short book of fewer than 190 pages, assuming an information density of 1,600 characters per page. Give it a solid cover and it comes to a total of 250 g. With respect to the information/ weight ratio, paper thus outperforms clay by a factor of 4,000. Such a rough-and-ready calculation may suffice to illustrate the point. Papyrus was similarly superior to clay with regard to storage capacity and transportability; however, many more clay tablets than papyrus documents have come down to us through the millennia. How many papyrus rolls were lost in the legendary blaze of the library of Alexandria is not known, but when in 2004 a fire broke out in the Anna Amalia Library in Weimar, 50,000 books were destroyed, many of them unique or rare. Another 65,000 volumes were severely damaged by fire and water. Baked clay tablets would have withstood the flames and the water used to put them out. This line of thought can be extended into the digital age by another calculation. Computer technology has exponentially increased information storage density. The 50,000 burnt books of the Anna Amalia Library took up some 1,660 meters of shelf space. Assuming an average of 300 pages per book, their

digitalized contents would require some 750 gigabyte (GB) storage space, which easily fits on an external hard disk the size of a small book. As compared to print, digital information storage thus reduces the necessary physical space by a factor of 50,000. Again, this is a coarse measure only. There is considerable variation in the “bytes per book page,” both in actual fact and in model calculations, but the correlation between information amount and storage space of print media and digital media transpires from it. In sum, as clay was followed by paper and paper by digital storage media, information density per square centimeter increased exponentially, while the weight and size of written records decreased. It became, accordingly, ever easier to store and to transmit written text with many consequences for reading and writing behavior, for reproducing text, and for the role texts play in everyday life. What the history of writing shows is that new technologies do not always replace old ones. Rather, the new supplements the old and often transforms its use. For instance, parchment was marginalized by paper but for centuries never completely driven out; print never replaced handwriting and has not become obsolete by texts typed on a cell phone keypad. Advances in writing technology have greatly expanded the repertoire of tools that humanity has acquired for handling information in the form of written language. The fact that old technologies continue to be used side by side with new ones testifies not just to inertia and path dependency but also to the different properties of the materials used. For centuries after the introduction of paper, it was considered too feeble a material to hold contracts and other important documents, which were preferably executed on parchment. Similarly, although it is technically possible to keep birth registers as electronic files only, birth certificates on paper continue to be issued. One of the reasons is that digital information storage is subject to physical decay and technical obsolence not less but much more than predecessor media. Writing has made record keeping for posterity and accumulation of knowledge possible. However, with the introduction of every new writing material, the storage problem that it seemed to solve became more acute. An archive of records on paper takes up much less space than one for clay tablets, but it is beset by dangers that pose no threat to baked clay: dust, humidity, insects, fire, and water. Theft, too, is a greater threat to libraries than to clay tablet archives and a greater threat to computers than to libraries. Keeping books in a usable physical state requires more work than keeping clay tablets. The same kind of relationship holds between libraries and digital data archives. Much more can be stored, but preservation for future use becomes ever more difficult as time intervals of technical innovation shrink. Only a few specialists are able to handle data stored with software that ceased to be produced 20 years ago, whereas a book would last for centuries. The problem of preserving and organizing the everswelling flood of information remains unsolved, and at the same time many traditional libraries and collections of documents fall into decay. Technology has hugely expanded human memory, but it has not yet eliminated the risk that many parts of the heritage committed to writing will disappear forever. To guard against collective memory loss, the United Nations Educational, Scientific and Social Orgnization (UNESCO) has

37

The Cambridge Encyclopedia of the Language Sciences launched the Memory of the World Programme to assist in the preservation of archive holdings and library collections all over the world. For the time being, this is the endpoint of a development begun with the advent of literacy in ancient civilizations: the institutionalization of writing and the bureaucratization of society. The more serviceable writing became to human society, the more it penetrated social relations and the more attention it came to require on the part of society.

INSTITUTIONAL ASPECTS OF LITERACY From its inception, writing has been an instrument of power. In ancient civilizations of restricted literacy, its mastery was jealously guarded by the elite. It was indispensable for the workings of the temple-centered economies of ancient Near Eastern city states (Nissen 1988), symbolized the rule of the pharaohs in Egypt (Posener 1956), became the bedrock of China’s Confucian bureaucratic state (Lewis 1999), and was a sine qua non of Athenian democracy (Thomas 1992). Certainly, literacy levels varied widely as did the uses of literacy, but the general tendency of the extension of the human mind by means of writing to engender and necessitate institutions is unmistakable. The most important institutions produced by literate culture have to do with government, cult, schooling, and economic organization.

Government Writing was used early on to extend the reach of authority and to mediate the relationship between ruler and ruled. Monuments that embody power, such as the Rosetta Stone inscribed with a decree to commemorate “the reception of rulership” by Pharaoh Ptolemy V on March 27, 197 b.c.e. (Parkinson 1999, 29), as well as stelae appealing with regulations to the literate public, were erected throughout the Ancient Near East. Their inscriptions were drafted by scribes who created the first bureaucratic states. The Egyptian vizier was responsible for the collection of taxes, the maintenance of archives, and the appointment of officials. The skills of the scribe afforded privilege in Egyptian society. As one of them put it, “the scribe is released from manual tasks; it is he who commands” (Goody and Watt 1968, 37). A thousand years later, Mencius (372–289 b.c.e.) made the same point in China: “Some labour with their hearts and minds; some labour with their strength. Those who labour with their hearts and minds govern others. Those who labour with their strength are governed by others” (“Book of Mencius,” quoted from Lloyd and Sivin 2002, 16). These observations bring to light the connection between literacy and social hierarchy that persists to this day. Wherever societies obtained the art of writing, the literati were close to the powers that be, but the Chinese institutionalized literacy like no other culture. In Confucius’s day, literacy was already the preeminent mark of the gentleman, and as of the second century b.c.e., the study of the Confucian classics gradually became institutionalized as the key to attaining public office and influence. The civil service examination system was employed with few adjustments from the Han Dynasty (206 b.c.e.–220 c.e.) until the closing stages of the Qing Dynasty at the beginning of the twentiethth century. It was based entirely on the study of texts. To prepare for the highest degree, students spent up to 20

38

years memorizing the Confucian classics and commentaries. They were then able to recite, understand, and interpret every clause of the five canonical works – “Book of Changes,” “Book of Documents,” “Book of Poetry,” “Records of Rites,” and “Spring and Autumn Annals” – said to have been redacted by Confucius himself, as well as a collection of commentaries by subsequent scholars. That such an education was an adequate preparation for bureaucrats charged with administering the country was rarely called into question. It was firmly rooted in the past, and the classics were thought to hold a solution to any problem that might arise. The authority of writing and the conservatism of literature were never more effective. The strength of the system lay in the fact that it encouraged respect for learning and provided the emperor with a bureaucracy educated in one standard curriculum. Its weakness was its emphasis on commentary that stifled inquisitiveness and deviation from the trodden path. The civil service exam system institutionalized the transmission of texts down a lineage and was, thus, inherently averse to change. In its early days, it helped to loosen the hereditary aristocracy’s grip on power by rewarding merit rather than birth for recruiting bureaucrats. In actual fact, however, learning remained largely a prerogative of aristocratic families out of reach for most commoners. Women were not permitted to sit for the exams. In the end, the civil service examinations served as a system to perpetuate the power of the thin elite of literati bureaucrats. Controlling literacy has always been the other side of its relation to authority. The powerful have lived in fear of the power of the pen and have had little interest in promoting the art of writing among the masses. Illiterates are powerless, unable to challenge the letter of the law or to invoke the laws on the books to their own advantage. That which is written down acquires authority in its own right as a reference source independent of the ruler. While helping to project power far beyond earshot, it gains a measure of objectivity, thereby reducing the arbitrariness of rule. But only the literate can hold the ruler accountable to his own decrees. The institutionalization of writing to this end occurred in fifth-century Greece where government was established in the polis through written laws. These were not Godgiven but man-made laws, aiding the development of a division between cosmic order and human society (Stratton 1980), so characteristic of the Greek Weltanschauung and so different from the Chinese. From the objectification of language in writing follows another function with important implications for the exercising and curbing of power. Writing detaches the author from his message, which makes it easier and less risky to express the unspeakable. Two examples suffice to illustrate. In Athens, ostracism was institutionalized as a procedure to protect democratic rule. In the event that a charismatic politician (a demagogue) became too influential or was otherwise disruptive, the demos (people) were entitled to exile him from the polis. To this end, every citizen was given a potsherd (ostrakon) on which to write the name of the man to be sent into exile. The degree of literacy in Athens that can be inferred from this practice is a question that has given rise to much research and speculation (Burns 1981; Harvey 1966; Havelock 1982; W. Harris 1989; Thomas 1992). It is unlikely that it will be possible ever to quote even approximately

Social Practices of Speech and Writing correct figures, but what we know of the literate culture of the Greek city-states is that, unlike in China, there was no scribal class. Minimal competence such as was necessary to scratch a name on a potsherd was relatively widespread. Both written law as the basis of social order and ostracism as a check on power exemplify the institutionalization of writing as a precondition for political participation and as a seed of the public sphere. It enabled the people to express discontent with a leader without raising their hand individually or speaking out. Anonymity was a protection from reprisals. The other example is from Babylon, as reported in the Old Testament. Writing means empowerment and, therefore, has to be controlled. The proverbial writing on the wall, mene-tekel-parsin, an Aramaic phrase that, according to Daniel 5:25, mysteriously appeared on the wall of King Belshazzar’s palace, cautioning him that his days were numbered and his empire was doomed, exemplifies the potential of criticism. A message by an unknown author exposing abuse of power to the public is a direct challenge to authority. While it would be problematic to voice disapproval in the presence of others, writing affords the originator the protection of anonymity. The dialectics of technological innovation come to bear here. While the mighty use writing to establish, exercise, and consolidate their power, it also lends itself to embarrassing and undermining them. This was clearly understood in antiquity. Power holders always tried to curtail the power of the written word. Confucius’s Analects were burned in 200 b.c.e. on order of Emperor Pinyin Qin Shi Huang Di. Plato was convinced that the state had to control the content of what pupils read. Tacitus refers to banned books (Anales 4,34), and his own books were later banned in return. Censorship is ubiquitous throughout history (Jones 2001). Many of the finest literary works were at one time or another put on a list of forbidden books, such as the papal Index Auctorum et Librorum Prohibitorum, first published in 1559. It took another 500 years for censorship to be universally censured as an illegitimate means of exercising power. Article 19 of the Universal Declaration of Human Rights adopted by the United Nations in 1948 states that everyone has the right to freedom of expression and information, regardless of frontiers. The struggle against censorship worldwide is far from over. With every new writing technology it is rekindled, as the debate about controlling the Internet illustrates. The power elites try to control the new media, which they perceive as a threat, although, ironically, they are often involved in its development. The battlefield shifts with technological advance. What were scriptoria and publishing houses in the past are servers and the flow of data through cyberspace today. In the long run, attempts on the part of governments to defend their lead and keep new information technology out of reach of their adversaries have failed because it is the very nature of these technologies that they can be utilized to uphold and to counter the claim to power.

Cult In nonliterate societies religious, order and social order are merged without clear division. Writing tends to introduce a fracture between spheres, although the differentiation of the

sacred and the profane may take a long time to complete. The Ten Commandments, written by God and given to Moses, were a code of conduct regulating the relations between God and the people, as well as the people among themselves. The spheres of spiritual and worldly power were only beginning to be separated in antiquity, a process to which writing and the ensuing interpretation of texts contributed a great deal. Of Moses’ legendary Ten Commandments stone tablets no archaeological evidence remains, but the Aśoka edicts are a tangible example of the closeness of cult and social order. Engraved in rocks and stone pillars that have been discovered in dozens of sites throughout northern India, Nepal, Pakistan, and Afghanistan, these edicts served to inform the subjects of King Aśoka’s (304–232 b.c.e.) reforms and make them lead a moral life according to the truth of Buddhism. Moral precepts hewn in stone on public display once again raise the question of the degree of literacy. It must have been high enough to disseminate the message throughout the vast realm. In the rich spiritual world of the Indian subcontinent, much emphasis was always given to the importance of oral tradition. Competing with other religious movements, Buddhism established itself by rigidly regulating monastic life and assembling a canon of scriptures. At the time of Aśoka, Buddhism was an institutionalized religion that spread from northern India to other parts of the subcontinent, first and subsequently to South and Southeast Asia, as well as Central Asia and China. The history of Buddhism and its emergence as a major world religion is a history of translation. Translation means writing, turning one text into another and all it involves: exegesis, doctrine, scholasticism, and schism. As Ivan Illich and Barry Sanders (1988: 52) have argued, there is no translation in orality, but only the collaborative endeavor to understand one’s partner in discourse. Writing eternalizes spiritual enlightenment (e.g., the word of God), which must be preserved with the greatest care and does not allow alteration at will. Of all religions, Buddhism has produced by far the largest corpus of sacred texts. Significantly, different canons resulted from translation, the Pali canon, the Tibetan canon, and the Chinese canon. With translation came schism, and with that the delimitation of religious sects and districts. In the evolution of other book religions, translation had similar consequences. These other religions share with Buddhism the vital importance they attach to scriptures. The holy book is what distinguishes world religions from other cults. Their legitimacy is derived from the revelation embodied in the scriptures, which are considered the true source of enlightenment. The major world religions vary in the role they assign sacred texts, in how they make use of them for purposes of propaganda and the regulation of life, but the reverence accorded to writing is a feature they share, as is the question of translation. Translation is a problem for two reasons that can be labeled authenticity and correspondence. First, the idea of authenticity of the word of God does not agree with its transposition into another form. Some book religions are very strict in this regard. According to believers, God himself chose the Arabic language for his final testament, the Quran, and created every one of the letters of the Arabic alphabet. Consequently, the only true version of the holy book of Islam is the Quran in Classical Arabic.

39

The Cambridge Encyclopedia of the Language Sciences The Christian tradition until the Reformation movement knew similar limitations, recognizing only the three sacred tongues of Hebrew, Greek, and Latin as legitimate languages of the Bible. In many cases, other languages were considered unfit for the expression of divine revelations, if only because they lacked the fixity that comes with being reduced to writing. Indeed, translations of religious scriptures, when they were eventually produced, were for many languages their first literary texts, serving as an anchor in the fluidity of oral discourse and as a starting point of literacy and textual transmission. The other reason for the problematic nature of translation is that through it, a stable correspondence is to be established between two codes that though obviously different in lexical, grammatical, and phonetic makeup, must express the same contents. In order to translate, the meaning of the text has to be established unequivocally, the assumption being that this is possible. The text is thus elevated to the source of meaning. Authority that formerly accrued to the sage, the reciter, and the soothsayer was relocated in text. This transition “from utterance to text” (Olson 1977) implies that in order to understand a message, at issue no longer is what the speaker means but what the text contains. This is what translation is all about. Language itself in objectified form thus becomes a force in its own right with farreaching implications in the domains of knowledge production and social regulation. Preservation of the word of God in text has provided an objectified reference plane incorporating the “true” meaning waiting to be extracted from it. Olson’s notion of autonomous text that results from the transition from utterance to text as he conceptualizes it has been criticized because it ignores the reader’s involvement in constructing meaning (Nystrand 1986) and because it underestimates the oral elements in literate culture (Wiley 1996). The soundness of these criticisms in detail, however, does not invalidate the general idea that writing gives language a measure of stability that it does not have in speech, and brings with it a shift from the intentional to the conventional aspects of linguistic meaning, a shift from “I mean” to “the word means.” The high prestige that the written word acquired through its association with and instrumentalization by organized religion has greatly contributed to the coming into existence of autonomous text. The reverence for holy books had various consequences for language attitudes and practices, two of which can be mentioned here: code choice and script choice (Coulmas 2005). Writing introduces functional domain differentiation into a community’s linguistic ecology. That the language of cult differs from that of everyday pursuits has always been the rule rather than the exception, but with writing the distinction becomes more pronounced. The important position of religious texts in combination with restricted literacy encouraged the perpetuation of the split between spoken and written language. While the codification of the latter was aided by the desire to transmit the sacred texts inviolately to later generations, the former was subject to perpetual change. The result was a situation of coexisting codes, called diglossia in modern scholarship (Ferguson 1959; Krishnamurti 1986; Schiffman 1996). Although every case of diglossia is different, the defining characteristic is a domainspecific usage of varieties that coincides by and large with the spoken/written language divide. These varieties can be different

40

languages or varieties of the same stock. In multilingual environments like medieval Europe, where Latin was the only written language, or in present-day India, cultivated languages are often reserved for writing and formal communication, while vernacular varieties are used in informal settings. A similar division is found between linguistic varieties of the same stock, where one is defined by reference to a corpus of classical texts, such as Classical Arabic, whereas the others fluctuate without artificial restriction. Writing introduces an element of art and artificiality into the history of language. Every language is the collective product of its speakers, but a written language is more clearly an artefact than a vernacular, and the script that it uses more clearly yet. Historically, the diffusion of scripts coincided in large measure with that of religions, a connection that is still visible today. Chinese characters arrived in Japan together with Buddhism. The spread of Roman traces the expansion of both the Roman Empire and the Catholic Church, while Orthodox Christianity uses Cyrillic. Armenian Christians have their own alphabet designed in the fifth century by St. Mesrob. Estrangela is the script of the Syrian Church. The Hebrew square script is the script of Judaism, the Arabic alphabet that of Islam. Many other examples could be added; clerks were churchmen (Coulmas 1996, 435 f.). The historical interconnectedness of writing and religion is one of the reasons that scripts tend to function as symbols of identity, but ethnic, national, and political identity are also readily expressed by means of a distinct script or even slightly different orthographic conventions. As Peter Unseth (2005) has pointed out, there are clear sociolinguistic parallels between choosing scripts and languages. Because of the visibility and the artificial nature of writing, however, the choice of scripts is generally a more deliberate departure from tradition in that it involves conscious planning.

Schooling Language is a natural faculty, writing an artifact. That is the reason why children acquire language, but not writing, without guidance. The difficult art of writing requires skills that must be taught, memorized, and laboriously practiced. The place to do it is school. For writing to be useful to the community, conventions have to be established, individual variation curtailed, norms set. Collective instruction following a curriculum is a more efficient way to achieve this than is private tutoring. Already in antiquity, school became, and still is, the institution that most explicitly exercises authority over the written language by controlling its transmission from one generation to the next. With schooling came the regimentation and the decontextualization of language. Because in writing the continuous flow of speech has to be broken down into discrete units, analytic reflection about the units of language was fostered. As writing, language became an object of investigation and normalization. Both grammar and lexicon are products of writing. This is not to deny the grammaticality of oral speech or that oral people have a mental lexicon. It just means that the notions of grammar and lexicon as we know them are entirely dependent upon writing. At school, units of writing had to be practiced with the stylus, the brush, or the pen, mechanically through repetition without any

Social Practices of Speech and Writing communicative intent. These units could be given a phonetic interpretation; they could be pronounced and thus acquired as words, an existence as units of language. In time, the image became the model. Since the correct form of the written sign was a matter to which the scribe and his pupils had to pay attention, standards of correctness first developed with reference to writing and written language. Only much later, and as an effect of schooling, did these notions come to be applied to speech. The twin questions of what the units of writing were and how they were to be conjoined led to the more general and profound question “What is a language?” Right up to the present, answers to this question exhibit what Roy Harris (1980) has called “a scriptist bias.” Only trained linguists readily recognize unwritten vernaculars as languages, and even they have to admit that while it is easy to count written languages, the question of how many speech forms on this planet qualify as distinct languages is impossible to answer without laying down analytic criteria that are no less arbitrary than decisions as to what dialects, varieties, idioms, and speech forms should be reduced to writing. Languages, as well as the units into which they are analyzed, are a product of writing, for only in writing can the flow of speech be arrested and broken down into independent stable components with a presumed inherent, rather than vague and contextually determined meaning. Among the first results of school mastering the language in the Ancient Near East were word lists, the paradigm case of decontextualized language. These lists were the foundation of lexicography (Civil 1995), the science of words. In China, lexicography began with lists of characters, and characters are still the basic units of dictionaries. Dictionaries provide entries for lexical items. A lexical item is a word contained in a dictionary. More refined and less circular definitions of the orthographic word, as distinct from the phonological word and the abstract lexeme have been proposed in great number, but it remains difficult if not impossible to define word without reference to writing. The word stands at the beginning of grammatical scholarship, which was, as the word grammatical itself indicates, exclusively concerned with written language. Grammatike, combining the Greek words grammata (“letters”) and techne (“art”), was the art of knowing letters. These beginnings of the systematic study of language left a lasting imprint which, as Per Linell (2005) has convincingly shown, still informs modern linguistics. The word, the sentence, and even the phoneme are analytic concepts derived from the discrete segments of writing, not vice versa. The conceptualization of writing as a representation of speech is therefore highly problematic (R. Harris 1980, 2000). To sum up this section, the institutionalization of writing in school resulted in a changed attitude to language. It became an object of study and regulation. Both of these concepts were not in the first instance developed for, and applied to, speech. Under conditions of restricted literacy and diglossia, a wide divide between spoken and written language was taken for granted. Speech and writing were two modes of communication involving varieties or languages that were both grammatically and stylistically quite separate from each other. It was only when literacy became accessible to wider sections of the population that the relationship between speech and writing became an

issue. In medieval Europe, the ad litteras reform during the reign of Charlemagne aimed at unifying spoken and written language, as the widening gap between both was perceived as a problem. It was eventually reduced, not so much by enforcing a uniform standard for pronouncing Latin than by dethroning it as the only written language and transforming “lingua illiteratae” (BlancheBenveniste 1994), that is, Romance, Germanic, and Slavonic vernaculars, into written languages in their own right. Literacy in these emerging “national” languages was bolstered by the Reformation movement that wrested the interpretation monopoly of Christian scriptures from the Catholic clergy. “Write as you speak,” a maxim that can be traced back to antiquity, became an increasingly important principle for teaching writing (Müller 1990). Although it unrealistically denies the speech/ writing distinction, generations of teachers have repeated it to their pupils. It never meant that their writing should be as elliptical, situation-bound, and variable as their speech. The implication is that if you cannot write as you speak, something must be wrong with your speech. Universal education on the basis of this maxim resulted in a conceptual reduction of the distance between speech and writing, with some notable consequences. Mass literacy through schooling led to the disappearance of diglossia from most European speech communities, although the dichotomy of speech and writing continued to be expressed in stylistic differences. In other parts of the world, where universal education was realized later or is still only a desideratum, the split between spoken and written language remained. In today’s world, the 1953 UNESCO declaration recommending mother tongue literacy notwithstanding, literacy education in the language of the nation-state often means learning to read and write in a second language. The extent to which progress in the promotion of literacy depends on the language of instruction is still a matter of controversy, as is whether the writing system is a significant variable. To some extent, this can be explained by the fact that definitions of literacy are shifting with changing socioeconomic needs and technical innovations, and because the range of what are considered varieties of a given language is variable (as is evident, for example, in the context of decreolization and discussions about elevating certain varieties, such as black English in the United States, to language status). There is, however, wide agreement that the crucial variable is the effectiveness of the educational system. Mastering the written language is a difficult task, which is best executed by the institution that at the same time administers the written language: school. Since the time of the French Revolution, schools have been charged with establishing the national language and, by way of spreading the national language ideology, a link between language and nationalism. As a result, the demands of multilingual education are often at variance with the state-sponsored educational system. Because of the nationalization of languages in the modern nation-state and their privileged position in the school system, however, the language of literacy training became a political issue. Minority speech communities in many industrialized countries aspired to the prestige for their language that comes with a written standard and started to lobby for the inclusion of their language in the school curriculum. Fueled by the

41

The Cambridge Encyclopedia of the Language Sciences growing awareness of the importance of minority protection, such movements have met with a measure of success, leading to a highly complex situation of multiple and multilingual literacies in modern societies, which of late has attracted scholarly attention (Martin-Jones and Jones 2000; Daswani 2001; Cook and Bassetti 2005). The prevailing view sees the establishment of a single national language with a unified written standard as facilitating universal literacy. International comparisons of literacy rates are notoriously difficult (Guérin-Pace and Blum 1999), but there is little doubt that Europe, where national language academies first implemented the national language ideology, led the way. Today, however, the monolingual model of literacy is called into question by developments that, on the one hand, favor English as a supplementary universal written language in many non-Englishspeaking countries, and on the other, allow minority languages to make inroads into the domains of writing. The question of whether the diversification of literacy will help achieve the goal of eradicating illiteracy or whether it will compromise the alleged economic advantage of having one written standard language continues to be discussed by academics and politicians, while the complementary developments of globalization of markets and (re)localization of cultures unfold.

Economic Organization The critical importance of writing for economic processes in early civilizations is best documented for Mesopotamia. It is widely agreed now that in Sumer, number concepts and numerical representation stood at the beginning of writing that evolved into cuneiform (Schmandt-Besserat 1992). The overwhelming majority of archaeological finds from Ur, Uruk, Nineveh, and Babylon are records of economic transactions kept in clay tablet archives by accountants of the palace administration (Nissen, Damerow, and Englund 1990). The Sumerians and their Akkadian successors were the first to develop bookkeeping into a sophisticated technique of balancing income and expenditure. Hammurabi, King of Babylon (1728–1686 b.c.e.), created the first commercial code, 282 articles written on a large stone monument, which was erected in a public place for all to observe. Every business transaction was to be in writing and signed, usually with a seal, by the contracting parties. At the time, the public-sector economy was far too highly developed and too complex to function without writing. Tribute quota lists had to be kept, rations for laborers involved in public works calculated, inventories recorded. Deeds were issued in duplicate and receipts stored for future reference. Large-scale trading, often involving credit and “futures,” had to be regulated and overseen by the bureaucracy, consisting of a huge scribal class charged with creating and servicing these documents. Ancient Mesopotamia is the paradigm case of the close interconnectedness of economy and the institutionalization of writing, but if economic behavior is understood in the wide sense of human adaptations to the needs and aspirations of society at a given moment in history, this interconnectedness can be identified in every literate society. Complex socioeconomic systems of managed production, distribution and trade, taxation, and credit did not evolve in cultures that had no writing. Yet the nature of

42

the relationship between writing and economy is not a simple one. For one thing, the degree of literacy necessary for economic growth is a matter of controversy and depends on how economic growth is measured. Illiteracy is considered a strong indicator of economic underdevelopment, correlating as it does with infant mortality, low life expectancy, and poverty; but the question of whether high literacy drives economic growth or is a function thereof remains unresolved. Functional illiteracy rates of up to 20 percent in Western countries, notably in the United States, suggest that at least in the developed world, literacy rates are more indicative of economic disparity than of overall national economic development. Similarly, in developing countries, illiterates are largely rural and without access to public services (Varghese 2000). The distribution of wealth and the division of labor in today’s national economies are such that they allow for, or perhaps even sustain, substantial residues of illiteracy. Both in the developed and the developing world, people who partake of the practice of writing live side by side with others who continue to conduct their life in the oral mode. It is fair to say that the evolution of writing in antiquity both happened in response to and propelled economic development. Yet although writing technology has been available for economic pursuits for more than five millennia, fully literate societies remain an ideal. This shows that the relationship between economy and institutionalized writing is subject to interference by other variables, notably those discussed in the previous sections, that is, government, religion, culture, and education. In antiquity, these spheres and the economy were not separate. It was a long process that led to their differentiation in modern times. As the chief instrument of knowledge management and growth, writing was indispensable for rationalization and what Max Weber (1922) called the “disenchantment of the world,” which, as the religious associations and emotional values attached to various written languages testify, is still incomplete. The interaction of writing with the economy has been studied from various points of view. Writing can accordingly be understood as - a tool, - social choice, - a common good, - human capital, - transaction cost. Writing is a tool that enables individuals to expand their communication range and communities to increase social integration and differentiation. It is useful and valuable. Since writing systems, as discussed previously, are artifacts, this raises the question of how this tool evolves so that it is functional. Students of writing, notably I. J. Gelb (1963), have invoked the principle of least effort, which predicts that in time, writing systems become simpler and easier to use. The relative simplicity of the Greek alphabet, as compared, for example, with Egyptian hieroglyphs and early cuneiform, lends support to this hypothesis. However, intuitive though it is, there are problems. Obvious counterexamples are found outside the ancient Near Eastern civilizations mainly studied by Gelb. In

Social Practices of Speech and Writing its long history, the Chinese writing system and its adaptation in Japan increased rather than decreased in complexity. On the other hand, it took 500 years for the much simpler Korean Han’gul to be accepted as an alternative to Chinese. If simplicity and adaptability were the forces that drive the spread of writing systems, Han’gul should have supplanted Chinese characters, not just in Korea but in China and elsewhere long ago. Looking at the evolution of individual systems, certain letter forms in the Tibetan script were simplified to the extent of becoming indistinguishable, rendering the system almost dysfunctional. One has to conclude that if the principle of least effort is at work in the evolution of the technology of the intellect, then it is often compromised by other principles, such as path dependency, that is, the fact that change in society depends on established ways, identity affirmation (see ethnolinguistic identity ), and cultural exclusivism. Considering writing and written language from the point of view of social choice leads to a similar conclusion. A written language can be understood as an institution with certain properties that is shaped by the agents involved in its operation. However, a single criterion of optimization of reasonable target functions cannot explain the diversity of systems that evolved, the multiplicity of uses, or the actual shape of individual systems, such as, for example, English spelling or the Japanese usage of Chinese characters (Coulmas 1994). Clearly, established writing conventions are the result of public choice. No individual can introduce a better standard, even though its superior qualities may be indubitable, for conformity with the existing norm is crucial for the functionality of the institution. Penalties on norm violations are therefore high, and proposals for changing established practice invariably meet with strong resistance. Change is not impossible, but it comes at a cost and rarely uncompromisingly follows functional optimization. A written language, if used by a community, has properties of a common good. Like a public transport system, it is beneficial for most members of that community and therefore deserving of the attention of the state. This is the rationale underlying the legal protection enjoyed by national languages, explicitly or implicitly, in modern nation-states. A written language not used by everyone is not a common good and treated accordingly. Not serving anyone’s interest, dead languages are of interest to the historian at best. For the same reason, it has always been difficult to elevate a vernacular to the status of written language; not providing access to information and not being used by many members of the community in the beginning, it does not count as a common good. Its claim to written language status is supported not by its instrumental value but only by the symbolic value for its community. To reconcile the idea of a dominant language as a common good with the recognition of minority rights is therefore problematic. With Pierre Bourdieu (1982), it can be conceptualized as a struggle over legitimacy, competence, and access (market, linguistic; habitus, linguistic). Only if social harmony is made part of the equation will accommodation be possible. A written language, the cognitive abilities its mastery implies, and the information to which it provides access are a resource. Partaking in it adds to an individual’s human capital and hence to his or her marketability. In the information economy of the knowledge society, this view on written language is more

pertinent than ever (Levine 1994) and can explain processes such as the accelerated spread of English. The globalization of markets and the information technology revolution offer ever more people the opportunity to enhance their human capital and, at the same time, compel them to do so. However, the commodification of written language (Heller 2003) and the new forms and uses it takes on in the new media have consequences, some of which become apparent only as technology spreads. Conceptualizing written language as transaction cost brings market forces into view. Reducing transaction costs is considered a quasi-natural precondition of economic growth, which partly explains the spread of trade languages. Once in place and controlled by relevant agents, their use is less costly than translation. In today’s world, this principle seems strongly to favor the further spread of English. However, the effects of technological innovation are, as always, hard to foresee and even to assess when it unfolds in front of our eyes. When the Internet ceased to be a communication domain reserved to the U.S. military, partly due to the available software at the time, it seemed to be poised to become an English-only medium. But it turned out that as the technology caught on, the share of English in cyberspace communication rapidly declined. The new information technology made it much easier for speech communities big and small around the world to communicate in writing, a possibility eagerly exploited wherever the hardware became available. For some communities this meant using their language in writing for the first time. In others it led to the suspension of norms, blurring, in many ways yet to be explored, the traditional distinctions between spoken and written language. David Crystal (2001) suggests that “Netspeak” or computermediated language is a new medium, neither spoken language nor written language. The implications of Internet communication for literacy in the electronic age are only beginning to be explored (Richards 2000). New multilayered literacies are evolving, responding as they do to the complementary and sometimes conflicting demands of economic rationality, social reproduction through education, ideology, and the technical properties of the medium. These developments open up a huge new area of research into how, since the invention of writing, the range of linguistic communication options has been constantly expanding.

CONCLUSION Writing is a technology that interacts with social practices in complex ways, exercising a profound influence on the way we think, communicate, and conceptualize language. Since it is an artifact, it can be adjusted deliberately to the evolving needs of society, but it also follows its own inherent rules that derive from the properties of the medium. This essay has analyzed the tension between the properties of the medium and the designs of its users from two points of view, technological advance and institutionalization. Harnessed by institutions, the technology of writing is made serviceable to the intellectual advance of society and modified in the process, sometimes gradually and sometimes in revolutionary steps. Three consequences of writing remain constant: segmentation, linearization, and accumulation. The linear and discrete-segmental structure that all writing systems both derive from and superimpose on language

43

The Cambridge Encyclopedia of the Language Sciences forever informs the perception of language in literate society. And by making language visible and permanent, it enables and compels its users to accumulate information far beyond the capacity of human memory, engendering ever new challenges for storage and organization. WORKS CITED AND SUGGESTIONS FOR FURTHER READING Assmann, Jan. 1991. Stein und Zeit: Mensch und Gesellschaft im alten Ägypten. Munich: Wilhelm Fink Verlag. Backhaus, Peter. 2007. Linguistic Landscapes: A Comparative Study of Urban Multilingualism in Tokyo. Clevedon, UK: Multilingual Matters. Blanche-Benveniste, Claire. 1994. “The construct of oral and written language.” In Functional Literacy: Theoretical Issues and Educational Applications, ed. Ludo Verhoeven, 61–74. Amsterdam and Philadelphia: John Benjamins. Bourdieu, Pierre. 1982. Ce que parler veut dire. Paris: Fayard. Burns, Alfred. 1981. “Athenian literacy in the fifth century B.C.” Journal of the History of Ideas 42.3: 371–87. Civil, Miguel. 1995. “Ancient Mesopotamian lexicography.” In Civilizations of the Ancient Near East, ed. Jack M. Sasson, 2305–14. New York: Charles Scribner’s Sons. Cook, Vivian, and Benedetta Bassetti, eds. 2005. Second Language Writing Systems. Clevedon, UK: Multilingual Matters. Coulmas, Florian. 1994. “Writing systems and literacy: The alphabetic myth revisited.” In Functional Literacy: Theoretical Issues and Educational Implications, ed. Ludo Verhoeven, 305–20. Amsterdam and Philadelphia: John Benjamins. ———. 1996. The Blackwell Encyclopaedia of Writing Systems. Oxford: Blackwell. ———. 2003. Writing Systems: An Introduction to Their Linguistic Analysis. Cambridge: Cambridge University Press. ———. 2005. Sociolinguistics: The Study of Speakers’ Choices. Cambridge: Cambridge University Press. Crystal, David. 2001. Language and the Internet. Cambridge: Cambridge University Press. Daswani, C. J., ed. 2001. Language Education in Multilingual India. New Delhi: UNESCO. De Biasi, Pierre-Marc. 1999. Le papier: Une aventure au quotidien. Paris: Gallimard. Eisenstein, Elizabeth L. 1979. The Printing Press as an Agent of Change. Cambridge: Cambridge University Press. Febvre, Lucien, and Henri-Jean Martin. [1958] 1999. L’apparition du livre. Paris: Albin Michel (Bibliothèque Evolution Humanité). Ferguson, Charles A. 1959. “Diglossia.” Word 15: 325–40. Fischer, Hervé. 2006. Digital Shock: Confronting the New Reality. Montreal: McGill-Queen’s University Press. Gelb, I. J. 1963. A Study of Writing. Chicago and London: University of Chicago Press. Goody, Jack. 1977. The Domestication of the Savage Mind. Cambridge: Cambridge University Press. Goody, Jack, and Ian Watt. 1968. “The consequences of literacy.” In Literacy in Traditional Societies, ed. Jack Goody, 27–68. Cambridge: Cambridge University Press. Guérin-Pace, F., and A. Blum. 1999. “L’illusion comparative: Les logiques d’élaboration et d’utilisation d’une enquête internationale sur l’illettrisme.” Population 54: 271–302. Harris, Roy. 1980. The Language Makers. Ithaca, NY: Cornell University Press. ———. 2000. Rethinking Writing. London: Athlon Press. Harris, William V. 1989. Ancient Literacy. Cambridge: Harvard University Press.

44

Harvey, F. D. 1966. “Literacy in Athenian democracy.” Revue des Études Grecques 79: 585–635. Havelock, Eric A. 1982. The Literate Revolution in Greece and Its Cultural Consequences. Princeton, NJ: Princeton University Press. Heller, Monica. 2003. “Globalization, the new economy, and the commodification of language and identity.” Journal of Sociolinguistics 7.4: 473–92. Illich, Ivan, and Barry Sanders. 1988. The Alphabetization of the Popular Mind. San Francisco: North Point. Jones, Derek, ed. 2001. Censorship: A World Encyclopedia. Vols.1–4. London: Fitzroy Dearborn. Kapitzke, Cushla. 1995. Literacy and Religion: The Textual Politics and Practice of Seventh-Day Adventism. Amsterdam and Philadelphia: John Benjamins. Kressel, Henry, with Thomas V. Lento. 2007. Competing for the Future – How Digital Innovations Are Changing the World. Cambridge: Cambridge University Press. Krishnamurti, Bh., ed. 1986. South Asian Languages: Structure, Convergence and Diglossia. Delhi: Motilal Banarsidass. Landry, Rodrigue, and Richard Y. Bourhis. 1997. “Linguistic landscape and ethnolinguistic vitality.” Journal of Language and Social Psychology 16.1: 23–49. Levine, Kenneth. 1994. “Functional literacy in a changing world.” In Functional Literacy: Theoretical Issues and Educational Applications, ed. Ludo Verhoeven, 113–31. Amsterdam and Philadelphia: John Benjamins. Lewis, Mark Edward. 1999. Writing and Authority in Early China. SUNY Series in Chinese Philosophy and Culture. Albany: State University of New York Press. Liberman, Philip, and Sheila Blumstein. 1988. Speech Physiology, Speech Perception, and Acoustic Phonetics. New York: Cambridge University Press. Linell, Per. 2005. The Written Language Bias in Linguistics: Its Nature, Origin and Transformation. London: Routledge. Lloyd, Geoffrey and Nathan Sivin. 2002. The Way and the Word: Science and Medicine in Early China and Greece. New Haven and London: Yale University Press. Martin-Jones, Marilyn, and Kathryn Jones. 2000. Multilingual Literacies: Reading and Writing Different Worlds. Amsterdam and Philadelphia: John Benjamins. McLuhan, Marshall. 1964. Understanding Media: The Extension of Man. New York: McGraw-Hill. Müller, Karin. 1990. “Schreibe, wie du sprichst!”: Eine Maxime im Spannungsfeld von Mündlichkeit und Schriftlichkeit. Frankfurt am Main and Bern: Peter Lang. Negroponte, Nicholas. 1995. Being Digital. New York: Alfred A. Knopf. A cyberspace extension is available online at: http://archives.obs-us. com/obs/english/books/nn/bdintro.htm. Nissen, Hans J. 1988. The Early History of the Ancient Near East. Chicago and London: The University of Chicago Press. Nissen, Hans J., Peter Damerow, and Robert K. Englund. 1990. Frühe Schrift und Techniken der Wirtschaftsverwaltung im alten Vorderen Orient: Informationsspeicherung und –verarbeitung vor 5000 Jahren. N.p.: Verlag Franzbecker. Nystrand, Martin. 1986. The Structure of Written Communication. Orlando, FL: Academic Press. Olson, David R. 1977. “From utterance to text: The bias of language in speech and writing.” Harvard Educational Review 47.3: 257–86. ———. 1991. Literacy and Orality. Cambridge: Cambridge University Press. ———. 1994. The World on Paper. Cambridge: Cambridge University Press.

Social Practices of Speech and Writing Ong, Walter J. 1982. Orality and Literacy: The Technologizing of the Word. London: Methuen. Parkinson, Richard. 1999. Cracking Codes: The Rosetta Stone and Decipherment. London: British Museum Press. Posener, Georges. 1956. Littérature et politique dans l’Égypte de la XIIe dynastie. Paris: Honoré Champion (Bibliothèque l’École des hautes études. 3007.) Richards, Cameron. 2000. “Hypermedia, Internet communication, and the challenge of redefining literacy in the electronic age.” Language Learning and Technology 4.2: 59–77. Roberts, Colin H., and T. C. Skeat. 1983. The Birth of the Codex. Oxford: Oxford University Press. Sassoon, Rosemary. 1995. The Acquisition of a Second Writing System. Oxford: Intellect. Schiffman, Harold. 1996. Linguistic Culture and Language Policy. London and New York: Routledge. Schmandt-Besserat, Denise. 1992. Before Writing. Austin: The University of Texas Press. Stratton, Jon. 1980. “Writing and the concept of law in Ancient Greece.” Visible Language 14.2: 99–121.

Taylor, Insup, and M. Martin Taylor. 1995. Writing and Literacy in Chinese, Korean and Japanese. Amsterdam and Philadelphia: John Benjamins. Thomas, Rosalind. 1992. Literacy and Ortality in Ancient Greece. Cambridge: Cambridge University Press. Twitchett, Denis C. 1983. Printing and Publishing in Medieval China. London: The Wynkyn de Worde Society. UNESCO Institute for Statistics. 2006. Available online at: http://portal. unesco.org/education/en/ev.php-URL_ID=40338&URL_DO=DO_ TOPIC&URL_SECTION=201.html. Unseth, Peter. 2005. “Sociolinguistic parallels between choosing scripts and languages.” Written Language and Literacy 8.1: 19–42. Varghese, N. V. 2000. “Costing of total literacy campaigns in India.” In Adult Education in India, ed. C. J. Daswani and S. Y. Shah, 227–50. New Delhi: UNESCO. Weber, Max. 1922. Wirtschaft und Gesellschaft: Grundriss der verstehenden Soziologie. Tübingen: J. C. B. Mohr. Wiley, T. G. 1996 . Literacy and Language Diversity in the United States. Washington, DC: Center of Applied Linguistics and Delta Systems.

45

5 EXPLAINING LANGUAGE: NEUROSCIENCE, GENETICS, AND EVOLUTION Lyle Jenkins

Before undertaking a discussion of the explanation of language, we should point out that we are using this word in a special sense. As Noam Chomsky has noted (see Jenkins 2000), while we cannot ask serious questions about general notions like “vision” or “language,” we can ask them of specific systems like insect vision or human language. In what follows, our focus is on the biological foundations of language. As a result, certain areas commonly referred to as languages are excluded from consideration, for example, some invented logical systems, computer languages such as Java, encrypted languages, the language of DNA, and so on. These are all important and interesting areas of investigation. In fact, significant insights into human language may be gained from the study of some of these fields. For example, it has been argued that particular systems of logical form may shed light on the semantics of human language. Both biological factors and nonbiological factors interact in such areas as pragmatics and sociolinguistics. In addition, the study of formal languages (e.g., the “Chomsky hierarchy”) has also led to some important contributions. However, these areas cannot be completely accounted for by a consideration of the biology of human language. It is important to keep in mind that an account of cognitive systems like human language (as well as systems of animal communication) often exhibit a significant degree of modularity. In this view, biological factors interact with other factors to provide a unified explanation. Sometimes the term i-language is used to distinguish the biological object in the mind-brain that biolinguists study from other uses of word language. The same is true of other areas of study that are even more closely related to human language, such as the role of language in poems and novels, or the influence of economic factors and conquests and invasions on language change. historical linguistics may involve factors both within and outside of the scope of human language biology. Again, although analysis of the biology of human language may shed some light on these areas (e.g., the study of phonetics and phonology may be useful in the analysis of poetry), in general it will not provide an exhaustive account of these areas. Similarly, there has been great interest in language as a system of communication. For the reasons discussed here, there is not much to say about arbitrary

46

systems of communication (semaphores, bee language, etc.), but the study of the biology of language might shed some light on the case of human language communication. In this essay, I consider some diverging ideas about the role of communication in driving language evolution. The study of the biology of language (see biolinguistics) is traditionally divided into the investigation of three questions: 1) What is knowledge of language? 2) How does language develop in the child? 3) And how does language evolve in the species? (See Chomsky 1980, 2006; Jenkins 2000, 2004.) Note that the study of the neuroscience of language cross-cuts with all three questions. That is, we can ask: 1) What are the neurological underpinnings of the faculty of language? 2) How does the language faculty develop in the nervous system of the individual? 3) How did the language faculty evolve in the species? (See Chomsky and Lasnik 1995.) These three questions are sometimes referred to as the what and how questions of biolinguistics. There is another question, the why question, which is more difficult to answer. This is the question of why the principles of language are what they are (Chomsky 2004). Investigations into these why questions has in recent years been termed the minimalist program (or minimalism), but interest in and discussion of these questions go back to the earliest days of biolinguistics (Chomsky 1995; Boeckx 2006). Properties of the attained language derive from three factors (Chomsky 2005b): 1) genetic endowment for language, 2) experience, and 3) principles not specific to the faculty of language. Principles in 3) might even be non-domain-specific or nonorganism-specific principles. Examples of such principles are principles of efficient computation. Note that similar questions can be posed about any biological system – viruses, protein folding, bacterial cell division, sunflowers, bumblebees, falling cats, nervous systems, and so on. Furthermore, similar kinds of questions arise in every linguistics research program (Chomsky 2005b). To make the discussion manageable, in what follows I draw examples and discussion from minimalism. However, the issues and problems carry over to all other research programs concerned with providing explanations for properties of human language and accounting for them in terms of neurobiology, genetics, and evolution. For example, any theory of the language faculty will generate infinitely many expressions that provide instructions to the sensorimotor and semantic interfaces. All such generative systems will have an operation that combines structures (e.g., in minimalism, merge), such as the phrase the boy with the phrase saw the cat, and so on, formed from the lexical items of the language. Applying this operation over and over (unbounded Merge), we get a discrete infinity of expressions, part of property 1), our genetic endowment, in this particular case of the genetic component of the language faculty. Many well-known accounts of the evolution of language propose communication as the primary selective force involved in the origin of language (see Works Cited). Here, for purposes of comparison, we present a noncommunicative account of the origins of language, suggested by Chomsky and based on work in minimalism. However, we stress again that this kind of account is also compatible with approaches based on other linguistic research programs. We then discuss another viewpoint

Explaining Language based on the idea that language evolved from gestures, an idea that is represented by a number of approaches to the evolution of language. After that, we present an example of the comparative approach to evolution in biolinguistics and discuss language and the neurosciences, focusing on left–right asymmetries of the language areas to illustrate this research. Finally, we discuss the genetics of language, using the studies of the FOXP2 gene system to show how studies of language phenotype, neurobiology, and molecular biology, as well as comparative and evolutionary studies with other animal systems, are being carried out. Work on the principles of efficient computation governing the application of the operation of Merge seem to suggest an asymmetry, namely, that the computational principles are optimized for the semantic interface, not the sensorimotor interface. This is because conditions of computational efficiency and ease of communication conflict, as is familiar from the theory of language parsing. This in turn has led Chomsky to suggest that externalization of language, and hence communication, was a secondary adaptation of language. This would mean that language arose primarily as an internal language of thought. Supporting this idea is the existence of sign languages, which develop in a different modality but in other respects are very similar to spoken language (Kegl 2004; Pettito 2005). These design features of language have led Chomsky to propose the following scenario for the evolution of language. Several decades ago Chomsky suggested, on the basis of results from nonhuman primate studies, that the higher apes might well have a conceptual system with a system of object reference and notions such as agent, goal, instrument, and so on. What they lack, however, is the central design feature of human language, namely, the “capacity to deal with discrete infinities through recursive rules” (Chomsky 2004, 47). He proposed that when you link that capacity to the conceptual system of the other primates, “you get human language, which provides the capacity for thought, planning, evaluation and so on, over an unbounded range, and then you have a totally new organism” (Chomsky 2004, 48). Thus, let us assume that some individual in the lineage to modern humans underwent a genetic change such that some neural circuit(s) were reorganized to support the capacity for recursion. This in turn provided the capacity for “thought” and so on over an unbounded range. This in itself provided that individual and any offspring with a selective advantage that then spread through the group. Thus, “the earliest stage of language would have been just that: a language of thought, used internally” (Chomsky 2005a, 6). At some later stage, there was an advantage to externalization, so that the capacity would be “linked as a secondary process to the sensorimotor system for externalization and interaction, including communication” (7). The evolutionary scenario just outlined is derived from design principles suggested from work on human languages. Many other evolutionary scenarios have been proposed that assume that communication or other social factors played a more primary role. These include accounts involving manual and facial gestures (Corballis 2002), protolanguage (Bickerton 1996), grooming (Dunbar 1998), and action and motor control (Rizzolatti and Arbib 1998). However, the two kinds of accounts are not incompatible and may represent different aspects or stages in the evolution of language. We cannot review all of these research

directions here (but see the discussion of population dynamics in a later secion; Nowak and Komarova 2001). A number of these accounts attempt to interconnect the evolution of manual gestures, sign language, spoken language, and motor control in various ways. Some of this work is based on the discovery of a system of “mirror neurons” (mirror systems, imitation, and language) (Gallese, Fadiga, et al. 1996). This and later work demonstrated the existence of neurons in area F5 of the premotor cortex of the monkey, which are activated when the monkey executes an action, for example, grasping an object, and also when the monkey observes and recognizes the action carried out by another monkey, or even the experimenter. In addition, a subset of “audiovisual” mirror neurons were discovered that are activated when the sound of an action is perceived, for example, the tearing of paper (Kohler et al. 2002). In addition to hand mirror neurons, “communicative” mouth mirror neurons were discovered that were activated both for ingestive actions and for mouth actions with communicative content, such as lip smacking in the monkey (Ferrari et al. 2003). Since it had been suggested on the basis of cytoarchitectonic studies that there was a homology between area F5 of the monkey and area 44 (in broca’s area) of the human brain (Petrides and Pandya 1994), researchers looked for and found mirror neuron systems in humans, using fMRI (Iacoboni et al. 1999) and event-related magnetoencephalography (MEG) (Nishitani and Hari 2000) in place of single neuron studies. Mirror neurons discharge whether the action is executed, observed, or heard. Moreover, they even discharge in the human system when subjects are exposed to syntactic structures that describe goal-directed actions (Tettamanti et al. 2005). In an fMRI study, 17 Italian speakers were asked to listen to sentences describing actions performed with the mouth (“I bite an apple”), with the hand (“I grasp a knife”), and with the leg (“I kick the ball”). In addition, they were presented a control sentence with abstract content (“I appreciate sincerity”). In the case of the actionrelated words, the left-hemispheric fronto-parietal–temporal network containing the pars opercularis of the inferior frontal gyrus (Broca’s area) was activated. Other areas were differentially activated, depending on the body part. They conclude that the experiment showed that the role of Broca’s area was in “the access to abstract action representations, rather than in syntactic processing per se” (277). On the basis of these outlined findings, it has been suggested that speech may have evolved from gesture rather than from vocal communication by utilizing the mirror neuron system (Rizzolatti and Arbib 1998; Gentilucci and Corballis 2006; Fogassi and Ferrari 2007). Leonardo Fogassi and Pier Francesco Ferrari note that the motor theory of speech perception fits well with an account in terms of mirror neurons in that “the objects of speech perception are the intended phonetic gestures of the speaker, represented in the brain as invariant motor commands” (Liberman and Mattingly 1985, 2). However, Fogassi and Ferrari note that, even if the mirror neuron system is involved in speech, the currently available evidence does not appear to support a “dedicated” mirror-neuron system for language in humans. Additional accounts of the evolution of language may also be found in Maynard Smith and Szathmary (1998), Lieberman (2006), and Christiansen and Kirby (2003). See also adaptation;

47

The Cambridge Encyclopedia of the Language Sciences grooming, gossip, and language; mirror systems, imitation, and language; evolutionary psychology; morphology, evolution and; origins of language; phonology, evolution of; pragmatics, evolution and; primate vocalizations; semantics, evolution and; speech anatomy, evolution of; syntax, evolution of; verbal art, evolution and.

THE COMPARATIVE APPROACH IN BIOLINGUISTICS Throughout the modern era of biolinguistics, a question that has been much debated is to what degree the faculty of language is uniquely human. Marc D. Hauser and colleages (2002) have stressed the importance of the comparative approach in the study of this question. Early on, Chomsky, in making a case for the genetic determination of language, used arguments from animal behavior (ethology) to note the similarities in learning between birdsong and animals, in particular, rapidity of learning, underdetermination of data, and so on (Chomsky 1959). Hauser, Chomsky, and their colleagues have emphasized a number of methodological points concerning the comparative approach, using a distinction between the faculty of language in the broad sense (FLB) and the faculty of language in the narrow sense (FLN). The basic idea is that before concluding that some property of language is uniquely human, one should study a wide variety of species with a wide variety of methods. And before concluding that some property of language is unique to language, one should consider the possibility that the property is present in some other (cognitive) domain, for example, music or mathematics. They tentatively conclude that recursion may be a property of language that is unique to language and, if so, belongs to the faculty of language in the narrow sense. An example of the application of the comparative method is the investigation of the computational abilities of nonhuman primates by W. Tecumseh Fitch and Hauser (2004), who tested the ability of cotton-top tamarins, a New World primate species, as well as human controls, to process different kinds of grammars. Using a familiarization/discrimination paradigm, they found that the tamarins were able to spontaneously process finite-state grammars, which generate strings of syllables of the form (AB)n, such as ABAB, ABABAB. However, they were unable to process context-free grammars, which generate strings of syllables of the form AnBn, such as AABB, AAABBB. It is known from formal language theory (the Chomsky hierarchy) that context-free grammars are more powerful than finite-state grammars. Moreover, the humans tested were able to rapidly learn either grammar. The authors conclude that the acquisition of “hierarchical processing ability,” that is, the ability to learn context-free grammars, “may have represented a critical juncture in the evolution of the human language faculty” (380). In a later study, Timothy Q. Gentner and colleagues (2006) showed that European starlings, in contrast to tamarins, can recognize acoustic patterns generated by context-free grammars. Using an operant conditioning paradigm, they found that 9 of 11 starlings were able to learn both finite-state grammar and context-free grammar sequences accurately. The (AB)n and AnBn sequences in this case were made up of acoustic units (“motifs”) from the song of the starlings. In this case, A corresponded to

48

a “rattle” and B to a “warble.” As for possible reasons why the starlings succeeded with the context-free task while the tamarins failed, and for alternative explanations of the learning results, see additional discussion in Fitch and Hauser (2004), M. Liberman (2004), Gentner et al. (2006), and Hauser et al. (2007).

LANGUAGE AND THE NEUROSCIENCES Turning to the neurosciences (see questions 1–3), with the advent of more tools – both experimental, such as imaging and the methods of comparative genomics, and theoretical, such as computational theories of linguistics – we can look forward to more informative models of language and the brain. David Poeppel and Gregory Hickok (2004) note three problems with classical models of language (involving such concepts as Broca’s area and wernicke’s area): 1) They are inadequate for aphasic symptomatology, 2) they are based on an impoverished linguistic model, and 3) there are anatomical problems. As for aphasic symptomatology, the classical models do not explain certain subtypes of aphasia, like anomic aphasia. Also, clusters of aphasic symptoms are highly variable and dissociable, indicating that there is a more complex architecture underlying the syndromes. As for the second problem, there has been a tendency of some implementations of classical models to incorporate a monolithic picture of linguistic components, for example, production versus comprehension, or semantics versus syntax, without regard for finer computational subdivisions. And finally, certain anatomical problems came into view. For example, Broca’s aphasia and Wernicke’s aphasia did not always correspond to damage in the areas that bore their names. In addition, these classical areas were not always anatomically or functionally homogenous. In addition, areas outside these regions were found to be implicated in language processing, including, for example, the anterior superior temporal lobe, the middle temporal gyrus (MTG), the temporo-parietal junction, the basal ganglia, and many right hemisphere homologs (Poeppel and Hickok 2004, 5). As noted, in the last few decades many new approaches and tools have been developed in neurolinguistics and neurology, genetics and molecular neurobiology (examples follow), and these have helped to overcome the kinds of issues pointed out by Poeppel and Hickok. For a review of some of the attempts to develop a “new functional anatomy of language,” see the essays published along with Poeppel and Hickock (2004) in Cognition 92 (2004). For new approaches to the study of Broca’s area, see Grodzinsky and Amunts (2006).

LEFT–RIGHT ASYMMETRIES OF THE LANGUAGE AREAS Ever since Paul Pierre Broca’s and Carl Wernicke’s seminal discoveries of areas involved in language processing, questions about asymmetries in the brain has been a lively area of research. As early as 1892, Daniel J. Cunningham reports that he “found the left sylvian fissure longer in the chimpanzee and macaque.” Cunningham, in turn, cites earlier work by Oskar Eberstaller on the Sylvian fissure in humans, who had concluded that it was longer in the left hemisphere than in the right (on average). He

Explaining Language postulated that this region held the key to what he called the “sensible Sprachcentrum” (“sensible/cognizant language center”). Claudio Cantalupo and William D. Hopkins (2001) report finding an anatomical asymmetry in Broca’s area in three great ape species. They obtained magnetic resonance images (MRI) (neuroimaging) from 20 chimpanzees (P. troglodytes), 5 bonobos (P. paniscus), and 2 gorillas (G. gorilla). In humans, Brodmann’s area 44 corresponds to part of Broca’s area within the inferior frontal gyrus (IFG). This area is larger in the left hemisphere than in the right. Furthermore, it is known that Broca’s area is vital for speech production (with the qualifications discussed earlier). Although the great apes were known to have a homolog of area 44 on the basis of cytoarchitectonic and electrical stimulation studies, no left–right anatomical asymmetry had been shown. Cantalupo and Hopkins found a pattern of morphological asymmetry similar to that found in the homologous area in humans. This would place the origin of the asymmetry for the anatomical substrate for speech production to at least five million years ago. Since the great apes exhibit only primitive vocalizations, these authors speculate that this area might have subserved a gestural system (see earlier discussion). They note the presence in monkeys of mirror neurons in area 44 that subserve the imitation of hand grasping and manipulation (Rizzolatti and Arbib 1998). They also observe that captive great apes have a greater righthand bias when gesturing is accompanied by vocalization. Hence in the great apes, the asymmetry may have subserved the “production of gestures accompanied by vocalizations,” whereas, for humans, this ability was “selected for the development of speech systems,” accompanied by the expansion of Brodmann’s area 45 (which, along with Brodmann’s area 44, makes up Broca’s area) (Cantalupo and Hopkins 2001, 505). However, additional studies of Brodmann’s area 44 in African great apes (P. troglodytes and G. gorilla) call into question whether the techniques used in Cantalupo and Hopkins’ study were sufficient to demonstrate the left–right asymmetry. Chet C. Sherwood and colleagues (2003) found considerable variation in the distribution of the inferior frontal sulci among great ape brains. They also constructed cytoarchitectural maps of Brodmann’s area 44, examining myeloarchitecture and immunohistochemical staining patterns. When they studied the IFG of great ape brains, they found a poor correspondence between the borders observed in the cytoarchitectural maps and the borders in the surface anatomy (e.g., sulcal landmarks). There were similar findings for human brains in an earlier study (Amunts et al. 1999). Sherwood and colleagues conclude that in the study by Cantalupo and Hopkins, it is “unlikely that the sulci used to define the pars opercularis coincided with the borders of cytoarchitectural area 44” (2003, 284). In general then, macrostructure is a poor predictor of microstructure. Sherwood and colleagues also point out that even if humanlike asymmetries of the inferior frontal gyrus and of the planum temporale are confirmed, these gross asymmetries will not suffice to explain “the unique neural wiring that supports human language” (284). To that end, comparative studies of microstructure in humans and great apes are needed. For example, a computerized imaging program was used to examine minicolumns

in a region of the planum temporale in human, chimpanzee, and rhesus monkey brains. It was found that only human brains exhibited asymmetries in minicolumn morphology, in particular, wider columns and more neuropil space (Buxhoeveden et al. 2001). It is possible that circuits could be reorganized within a language region without a significant volumetric change so that a novel function in language could evolve. Sherwood and colleagues conclude: “Therefore, it is likely that Brodmann’s area 44 homolog in great apes, while similar in basic structure to that in humans, differs in subtle aspects of connectivity and lacks homologous function” (284). Allen Braun (2003) notes that MRI could still turn out to be useful for the study of microstructure at higher field strengths, with the addition of MR contrast agents, and with the use of diffusion-weighted MR methods. He also notes that the pars orbitalis has often been arbitrarily excluded from the definition of Broca’s area, and might be important in the search for antecedents of language in nonhuman primates. In particular, some studies suggest that the pars orbitalis is selectively activated by semantic processes (as opposed to phonological or syntactic processes) (Bookheimer 2002). It is known that nonhuman primates have structures homologous to the perisylvian areas involved in human language, that is, support both expressive and receptive language (Galaburda and Pandya 1983; Deacon 1989). Ricardo Gil-da-Costa and colleagues (2006) presented species-specific vocalizations in rhesus macaques and found that the vocalizations produced distinct patterns of brain activity in areas homologous to the perisylvian language areas in humans using H215O positron emission tomography (PET). Two classes of auditory stimuli were presented to the monkeys. One was species-specific macaque vocalizations (“coos” and “screams”). As a control, nonbiological sounds were presented that matched the species-specific vocalizations in frequency, rate, scale, and duration. They found, for example, a greater response to species-specific calls than to nonbiological sounds in the perisylvian system with homologs in humans, for example, to the area Tpt of the temporal planum and to the anterior perisylvian cortex, roughly corresponding to the areas studied by Wernicke and Broca in humans. However, they did not find any clear lateralization effects in the macaque brain comparable to the anatomical and functional asymmetries documented in humans and anatomical asymmetries in apes (Gannon et al. 1998). Gil-da-Costa and colleagues (2006) note that the perisylvian regions are not performing “linguistic computations” in the macaque, but could be performing a “prelinguistic” function in associating “the sound and meaning of species-specific vocalizations.” Furthermore, this would position the perisylvian system to be “recruited for use during the evolution of language.” More specifically, it may have been “exapted during the emergence of more complex neural mechanisms that couple sound and meaning in human language” (2006, 1070). Although I have focused here on the perisylvian system, it should be emphasized that areas outside this system have also been demonstrated to be involved in language. K. A. Shapiro and colleagues (2006) provide another example of the application of imaging studies to investigate how linguistic

49

The Cambridge Encyclopedia of the Language Sciences categories like nouns, verbs, and adjectives are organized in the brain. An event-related functional MRI imaging study has found specific brain sites that are activated by either nouns or verbs, but not both. In a series of experiments, subjects were asked to produce nouns and verbs in short phrases as real words (the ducks, he plays), as well as pseudowords (the wugs, he zibs), both with regular inflections and irregular inflections (geese, wrote), including both concrete and abstract words. Specific brain areas were selectively activated for either verb production (left prefrontal cortex and left superior parietal lobule) or for noun production (left anterior fusiform gyrus) across the entire battery of tests. Moreover, the areas were nonoverlapping, leading the authors to conclude that these regions “are involved in representing core conceptual properties of nouns and verbs” (2006, 1644). In recent years, it has become possible to study asymmetries on a molecular level as well (Sun and Walsh 2006). As discussed earlier, there are functional, anatomical, and cytoarchitectonic differences between the left and right cerebral hemispheres in humans. To determine what the molecular basis for these asymmetries might be, Tao Sun and colleagues (2005) compared left and right embryonic cerebral hemispheres for left–right differences in gene expression, using serial analysis of gene expression (SAGE). They discovered 27 genes whose transcriptions were differentially expressed on the left and right sides. In particular, the gene LMO4, which asymmetrically expressed the Lim Domain Only 4 transcription factor, is more highly expressed in the perisylvian regions of the right hemisphere than in the left at 12 weeks and 14 weeks. Further studies are needed to determine how LMO4 expression is regulated by factors still earlier in development. Mouse cortex was also examined, and it was found that although Lmo4 expression was moderately asymmetrical in every individual mouse brain, the expression was not consistently lateralized to either the left or the right. This may be related to the fact that asymmetries like paw preference are seen in individual mice but are not biased in the population as a whole, as hand preference is in humans. The results of this study are also consistent with the observation that the genes involved in visceral asymmetries (e.g., of the heart) are not measurably implicated in cerebral asymmetries. It had been noted earlier that situs inversus mutations in humans do not appear to interfere with the left-hemisphere localization of language and handedness (Kennedy et al. 1999). In these earlier studies, it had been found that the pattern of language lateralization in patients with situs inversus was identical to that found in 95 percent of right-handed individuals with normal situs. It was concluded that the pathway affecting language dominance and handedness was most likely distinct from that affecting the asymmetry of the visceral organs.

GENETICS AND SPEECH DISORDERS Biolinguists would like to understand the wiring of networks underlying language function at the level of genes. We have seen that one way to study this question is to use such differential gene expression methods as SAGE. Another key way of investigating the genetics of language is by studying language disorders (see

50

dyslexia, specific language impairment, and autism and language). By understanding how genetic changes can cause the operation of language to break down, we get an idea of the genes that are important for language acquisition, processing, and, ultimately, language evolution. An autosomal-dominant and monogenic disorder of speech and language was found in the KE family with a 3-generation pedigree. A monogenic disorder involves a mutation in a single gene, and here the individuals have one copy of the mutant gene and one normal gene on two autosomal (non-sex) chromosomes. The disorder was mapped to the region 7q31 on chromosome 7, and it was shown that the gene (called FOXP2) had a mutation in a forkhead-domain of the protein it encoded in the affected family (Lai et al. 2001). The individuals were diagnosed as having developmental verbal dyspraxia. The phenotype was found to be quite complex, affecting orofacial sequencing, articulation, grammar, and cognition and is still incompletely understood and under investigation (see also genes and language; Jenkins 2000). The FOXP2 gene was found to code for a transcription factor, that is, a protein that regulates gene expression by turning a gene on or off or otherwise modulating its activity. It is natural to ask what other genes FOXP2 may regulate and how it regulates these genes (turning them on or off, for example), as well as to determine whether any of these genes “downstream” are involved in speech and language in a more direct way. To find the gene targets of FOXP2 in the brain and to determine the effects on those genes, methods are being developed to identify these neural targets both in vitro and in vivo (Geschwind and Miller 2001). The laboratory of D. H. Geschwind has developed a genomic screening approach that combines 1) chromatin immunoprecipitation and 2) microarray analysis (ChIP-Chip). In chromatin immunoprecipitation, an antibody that recognizes the protein of interest (e.g., foxp) is used to fish out a proteinDNA complex. The DNA is then hybridized to arrays with DNA from thousands of human genes. This allows the identification of binding sites for transcription factors (in this case, FOXP2). The goal is to discover potential gene candidates involved in the development of neural circuits supporting speech and language. The homologue to the human FOXP2 gene has been discovered in a number of different species, including mice (Foxp2) and songbirds, such as the zebra finch (FoxP2). Whatever one’s views on the relationship between human language and other animal communication systems, it is important to study the evolutionary origin of genes that affect language, such as FOXP2, for one can learn about the neural pathways constructed by these genes, which might not otherwise be possible in experiments with humans. It has been found that the zebra finch and human protein sequence is 100 percent identical within the DNA-binding domain, suggesting a possible shared function (White et al. 2006; Haesler et al. 2004). In addition, Constance Scharff and Sebastian Haesler (2005) report that the FoxP2 pattern of expression in the brain of birds that learn songs by imitation resembles that found in rodents and humans. In particular, FoxP2 is expressed in the same cell types, for example, striatal medium spiny neurons. Moreover, FoxP2 is expressed both in the embryo and in the adult. To find out whether FoxP2 is required for song behavior in

Explaining Language the zebra finch, the Scharff laboratory is using RNA to downregulate FoxP2 in Area A, a striatal region important for song learning (Scharff and Nottebohm 1991). It is known that young male zebra finches express more FoxP2 bilaterally in Area X when learning to sing (Haesler et al. 2004). They will then be able to determine whether the bird is still able to sing normal song as well as copy the song of an adult male tutor. Weiguo Shu and colleagues (2005) found that disrupting the Foxp2 gene in mice resulted in impairing their ultrasonic vocalization. In addition to their more familiar sonic vocalizations, mice also make ultrasonic sounds, for example, when they are separated from their mothers. In order to study the effect of disrupting the Foxp2 gene on vocalization, these researchers constructed two versions of knockout mice. One version had two copies of the defective Foxp2 gene (the homozygous mice) and the other version had one copy of the defective Foxp2 gene, as well as one gene that functioned normally (the heterozygous mice). The homozygous mice (“double knockout”) suffered severe motor impairment, premature death, and an absence of ultrasonic vocalizations that are normally produced when they are separated from their mother. On the other hand, the heterozygous mice, with a single working copy of the gene, exhibited modest developmental delay and produced fewer ultrasonic vocalizations than normal. In addition, it was found that the Purkinje cells in the cerebellum, responsible for fine motor control, were abnormal. It is concluded that the findings “support a role for Foxp2 in cerebellar development and in a developmental process that subsumes social communication functions” (Shu et al. 2001, 9643). Timothy E. Holy and Zhongsheng Guo (2005) studied the ultrasonic vocalizations that male mice emit, when they encounter female mice or their pheromones. They discovered that the vocalizations, which have frequencies ranging from 30–100 kHz, have some of the characteristics of song, for example, birdsong. In particular, they were able to classify different syllable types and found a temporal sequencing structure in the vocalizations. In addition, individual males, though genetically identical, produced songs with characteristic syllabic and temporal structure. These traits reliably distinguish them from other males. Holy notes that “these discoveries increase the attractiveness of mice as model systems for study of vocalizations” (White et al. 2006, 10378). We have focused on the FOXP2 gene here because there is no other gene affecting speech and language about which so much information is available that bears on questions of neurology and evolution of language. However, the search is underway for other additional genes. For example, genetics researchers have also discovered a DNA duplication in a nine-year-old boy with expressive language delay (Patient 1) (Fisher 2005; Somerville et al. 2005). Although his comprehension of language was at the level of a seven-yearold child, his expression of language was comparable to that of only a two-and-a-half-year-old. The region of DNA duplication in Patient 1 was found to be on chromosome 7, and interestingly, was found to be identical to the region that is deleted in Williams-Beuren syndrome (WBS). Patients with WBS have relatively good expressive language but are impaired in the area of spatial construction. Lucy Osborne,

one of the researchers on the study, noted that, in contrast, Patient 1 had normal spatial ability but could form next to no complete words. When asked what animal has long ears and eats carrots, he could only pronounce the r in the word rabbit but was able to draw the letter on the blackboard and add features such as whiskers (McMaster 2005). The duplicated region on chromosome 7 contains around 27 genes, but it is not yet known which of the duplicate gene copies are involved in the expressive language delay, although certain genes have been ruled out. A gene (SRPX2) responsible for rolandic seizures that are associated with oral and speech dyspraxia and mental retardation has been identified (Roll et al. 2006). It is located on Xq22. The SPRX2 protein is expressed and secreted from neurons in the human adult brain, including the rolandic area. This study characterizes two different mutations. The first mutation was found in a patient with oro-facial dyspraxia and severe speech delay. The second mutation was found in a male patient with rolandic seizures and bilateral perisylvian polymicrogyria. The authors also note that Sprx2 is not expressed during murine embryogenesis, suggesting that SPRX2 might play a specific role in human brain development, particularly of the rolandic and sylvian areas.

RECOVERING ANCIENT DNA In addition to classical studies of fossils, there is currently renewed interest in work on “ancient DNA.” The Neanderthals (Homo neanderthalensis) were an extinct group of hominids that are most closely related to modern humans (Homo sapiens). Up to now, information about Neanderthals has been limited to archaeological data and a few hominid remains. Comparing the genetic sequences of Neanderthals and currently living humans would allow one to pinpoint genetic changes that have occurred during the last few hundred thousand years. In particular, one would be able to examine and compare differences in such genes as FOXP2 in living humans, Neanderthals, and nonhuman primates. Partial DNA sequences of the Neanderthal have now been published by two groups led by Svante Pääbo and Edward Rubin (Green et al. 2006; Noonan et al. 2006). Pääbo’s group identified a 38,000-year-old Neanderthal fossil from the Vindija cave (the Neanderthals became extinct around 30,000 years ago). The fossil was free enough from contamination to permit DNA to be extracted and subjected to large-scale parallel 454 sequencing, a newer and faster system for sequencing DNA. Pääbo’s group was able to sequence and analyze about one million base pairs of Neanderthal DNA. Note that the genomes of modern humans and Neanderthals each have about three billion base pairs (3 gigabases). Among the conclusions they reached on the basis of the comparison to human DNA was that modern human and Neanderthal DNA sequences diverged on average about 500,000 years ago. They also expected to produce a draft of the Neanderthal genome within two years. (For their preliminary results, see Pennisi 2009.) Rubin’s group obtained around 65,000 base pairs of Neanderthal sequence. They used a combination of the Sanger method of DNA sequencing and the (faster) pyrosequencing

51

The Cambridge Encyclopedia of the Language Sciences method. Although the Sanger method yields less amounts of sequence than pyrosequencing, the error rate is lower. From the sequence data they estimated that Neanderthals and humans shared a most recent common ancestor around 706,000 years ago. There has been interest in whether Neanderthals might have contributed to the European gene pool. For example, one study suggests that humans may have acquired the microcephalin gene, which regulates brain size during development, by interbreeding with another species, possibly the Neanderthals. However, the data here do not support this possibility, although more sequence data will be needed to answer the admixture question definitively (Evans et al. 2006). However, they did establish the validity of their sequencing approach, which allows for “the rapid recovery of Neanderthal sequences of interest from multiple independent specimens, without the need for whole-genome resequencing” (Noonan et al. 2006, 1118).

GENETICS AND EVOLUTION The work on comparative animal studies and comparative neuroanatomy discussed earlier are being increasingly informed by the rapidly emerging field of comparative genomics. Work in the field of evolutionary development (evo-devo) has provided substantial support for the idea that gene regulation is key to understanding evolution. Among the methods now available to us to study the evolution of anatomy (including neural circuits) and behavior at the genome level are comparative genomics, gene-expression profiling, and population genetics analysis. We have already seen an example of the profiling of gene expression in the left and right cerebral hemispheres using serial analysis of gene expression. Such methods have also been applied to FOXP2. As we noted earlier, the FOXP2 gene was discovered to code for a transcription factor and is therefore involved in regulating the expression of other genes. In general, transcription factors are highly conserved. In this case, the FOXP2 protein is different from the chimpanzee and gorilla sequence at two amino acids, and from the orangutan sequence at three amino acids. The human and mouse protein sequence differ at three amino acids. The question arises whether these amino acid replacements are of functional significance, that is, whether they played a crucial role in the evolution of language. A population genetic study of the FOXP2 locus concluded that it had been subjected to a “selective sweep” during the past 200,000 years, correlating closely with estimated age of Homo sapiens. However, Sean B. Carroll (2005) notes that one cannot immediately conclude from the fact that the FOXP2 gene has been a target of selection during human evolution that it is the amino acid replacements just discussed that were the functionally important targets. Since the FOXP2 gene is 267 kilobases in size, we should find more than 2,000 differences in DNA sequence between chimpanzees and humans (assuming an average base pair divergence of 1.2%). This means that there are many more possibilities for functionally significant mutations in noncoding regulatory areas than in “coding regions” (the FOXP2 protein is made from coding regions, while the “noncoding regions” contain the regulatory information). It is, of course,

52

difficult to discover the significance of nucleotide changes in noncoding regions, since one cannot determine their functional significance by visual inspection. Nonetheless, until this information is available, there is no reason to favor the idea that the two changes in the FOXP2 protein are functional. Carroll notes that while it may be tempting to reach for the “low-hanging fruit of coding sequence changes, the task of unravelling the regulatory puzzle is yet to come” (1165). In fact, some data favor the regulatory sequence hypothesis. Carroll notes that in evaluating whether FOXP2 is involved in the evolution of the neural circuits of language, one must ask several questions. The first is, “is the gene product used in multiple tissues?” (1164). It is known that FOXP2 appears to act as a repressor in lung tissue (Shu et al. 2001). Moreover, studies in mice have revealed that in addition to Foxp2, two other transcription factors, Foxp1 and Foxp4 are expressed in pulmonary and gut tissues (Lu et al. 2002). Foxp4 is also expressed in neural tissues during embryonic development. It is not at all surprising to find that a gene that may be involved in language development is also active in nonlanguage areas. The reason for this is that transcription factors like FOXP2 often act in a combinatorial fashion in conjunction with other transcription factors in different ways in different tissues. As another example, suppose that a cell has three regulatory proteins and that with each cell division a new regulatory protein becomes expressed in one of the daughter cells, but not the other. Hence, we only need three regulatory proteins acting in combination to produce eight different cell types. Combinatorial control is thought to be widespread as a means of eukaryotic gene regulation. For those familiar with linguistic models of language acquisition, it may help to think of parameter settings, whereby three binary parameters can specify eight different language types with respect to some structural feature. The idea, then, is that FOXP2 can work in conjunction with certain factors in the lung to repress gene activity in the epithelium, whereas it might work together with other factors in the brain to regulate genes there directly (or indirectly) involved in speech and language. The second question to ask is “are mutations in the coding sequence known or likely to be pleiotropic [i.e., causing multiple effects]”? (1164) It is well known that patients with the FOXP2 mutation have multiple defects in speech, orofacial praxis, language, and cognition. The third question to ask is “does the locus contain multiple cis-regulatory elements?” (ibid.) (Cis-elements are regulatory elements located on the same nucleic acid strand as the gene they regulate.) Again, since FOXP2 is expressed in multiple areas in the brain and in other organs, this is a clear indication that it does. Carroll concludes on this basis that “regulatory sequence evolution is the more likely mode of evolution than coding sequence evolution” (ibid.). Finally, Carroll notes that some of the data from experiments with birdsong learners and nonlearners also support the idea of regulatory evolution. When FoxP2 mRNA and protein levels during development were studied, a significant increase of FoxP2 expression was found in Area X, a center required for vocal learning. The increase occurred at a time when vocal learning in zebra finches was underway. Moreover, FoxP2 expression

Explaining Language levels in adult canaries varied with the season and correlated with changes in birdsong. These facts suggest regulatory control in development and in the adult brain.

EVOLUTION AND DYNAMICAL SYSTEMS Another tool used in the study of language evolution is the theory of dynamical systems (see also self-organizing systems) (Nowak 2006). However, applications of dynamical systems to language include studies not only of evolution of language but of language change and language acquisition as well (Niyogi 2004, 2006). Martin A. Nowak and Natalia Komarova (2001, 292) assume for the case of language acquisition that Universal Grammar (UG) contains two parts: 1) a rule system that generates a “search space” of candidate grammars and 2) a “[learning] mechanism to evaluate input sentences and to choose one of the candidate grammars that are contained in his [the learner’s] search space.” One of the main questions to be determined is “what is the maximum size of the search space such that a specific learning mechanism will converge (after a number of input sentences, with a certain probability) to the target grammar.” The question for language evolution then is “what makes a population of speakers converge to a coherent grammatical system.” A homogenous population (all individuals have the same UG) is assumed. Nowak and Komarova derive a set of equations, which they call the “language dynamics equations” (293), which give the population dynamics of grammar evolution. The equations represent the average payoff (for mutual understanding) for all those individuals who use a particular grammar and contribute to biological fitness (the number of offspring they leave), and include a quantity to “measure the accuracy of grammar acquisition” of the offspring from their parents. Another variable denotes “the relative abundance of individuals” who use a particular grammar. Still another variable denotes the “average fitness or ‘grammatical coherence’ of the population,” the “measure for successful communication in a population” (293). Nowak and Komarova use the language dynamics equations to study the conditions under which UG will result in “grammatical coherence.” A number of factors can be varied in order to run computer simulations: population size, assumptions about UG’s search space, and assumptions about the learning mechanism (e.g., “memoryless” or “batch” learning, etc.). A similar kind of dynamical systems analysis has been proposed by Partha Niyogi for language change (2004, 58), for what he calls the “emerging field of population linguistics.” Such concepts as symmetry and stability (stable and unstable equilibria) are used in the study of the language dynamics equations. Niyogi uses symmetric and asymmetric nonlinear dynamical models to study lexical and syntactic change. Nowak and Komarova note that dynamical systems analysis is compatible with a wide range of different kinds of linguistic analysis and learning theories. There are a number of other approaches to the study of language evolution with dynamical systems and simulation, some of which may be found in the suggestions for further reading (Christiansen and Kirby 2003; Cangelosi and Parisi 2001; Lyon et al. 2006). More than 25 years ago, Chomsky observed that “the study of the biological basis for human language capacities may prove to be one of the most exciting frontiers of science in coming years”

(1980, 216). Not only has that proven to be the case, but with the explosion of knowledge in many areas, including (comparative) linguistics, comparative neuroanatomy, evolutionary development, comparative genomics, to take just a few examples, biolinguistics promises to be a fascinating field for decades to come. WORKS CITED AND SUGGESTIONS FOR FURTHER READING Amunts, K., A. Schleicher, et al. 1999. “Broca’s region revisited: Cytoarchitecture and intersubject variability.” Journal of Comparative Neurology 412: 319–41. Bickerton, Derek. 1996. Language and Human Behavior. Seattle: University of Washington Press. Boeckx, Cedric. 2006. Linguistic Minimalism. Oxford: Oxford University Press. Bookheimer, Susan. 2002. “Functional MRI of language: New approaches to understanding the cortical organization of semantic processing.” Annual Review of Neuroscience 25 (March): 151–88. Braun, Allen. 2003. “New findings on cortical anatomy and implications for investigating the evolution of language.” The Anatomical Record Part A: Discoveries in Molecular, Cellular, and Evolutionary Biology 271A (March): 273–75. Buxhoeveden, D. P., A. E. Switala, et al. 2001. “Lateralization of minicolumns in human planum temporale is absent in nonhuman primate cortex.” Brain, Behavior and Evolution 57 (June): 349–58. Cangelosi, Angelo, and Domenico Parisi, eds. 2001. Simulating the Evolution of Languge. New York: Springer. Cantalupo, Claudio, and William D. Hopkins. 2001. “Asymmetric Broca’s area in great apes.” Nature 414 (November): 505. Carroll, Sean B. 2005. “Evolution at two levels: On genes and form.” PLoS Biology 3 (July): 1159–66. Available online at: www.plosbiology.org. Chomsky, Noam. 1959. “A review of B. F. Skinner’s Verbal Behavior.” Language 35.1: 26–58. ———. 1980. “On the biological basis of language capacities.” In Rules and Representations, 185–216. New York: Columbia University Press. ———. 1995. The Minimalist Program. Cambridge, MA: MIT Press. ———. 2004. The Generative Enterprise Revisited: Discussions with Riny Huybregts, Henk Van Riemsdijk, Naoki Fukui and Mihoko Zushi. Berlin: Mouton de Gruyter. ———. 2005a. “Some simple evo-devo theses: How true might they be for language?” Paper presented at the Morris Symposium, SUNY at Stony Brook. ———. 2005b. “Three factors in language design.” Linguistic Inquiry 36.1: 1–22. ———. 2006. Language and Mind. Cambridge: Cambridge University Press. Chomsky, N., and H. Lasnik. 1995. “The theory of principles and parameters.” In The Minimalist Program, N. Chomsky, 13–127. Cambridge, MA: MIT Press. Christiansen, Morten H., and Simon Kirby, eds. 2003. Language Evolution. New York: Oxford University Press. Corballis, Michael C. 2002. From Hand to Mouth: The Origins of Language. Princeton, NJ: Princeton University Press. Cunningham, Daniel J. 1892. Contribution to the Surface Anatomy of the Cerebral Hemispheres [Cunningham Memoirs]. Dublin: Royal Irish Academy of Science. Deacon, T. W. 1989. “The neural circuitry underlying primate calls and human language.” Human Evolution 4 (October): 367–401. Dunbar, Robin. 1998. Grooming, Gossip and the Evolution of Language. Cambridge: Harvard University Press. Evans, Patrick D., Nitzan Mekel-Bobrov, et al. 2006. “Evidence that the adaptive allele of the brain size gene microcephalin introgressed

53

The Cambridge Encyclopedia of the Language Sciences into Homo sapiens from an archaic Homo lineage.” PNAS 103 (November): 18178–83. Ferrari, Pier Francesco, Vittorio Gallese, et al. 2003. “Mirror neurons responding to the observation of ingestive and communicative mouth actions in the monkey ventral premotor cortex.” European Journal of Neuroscience 17: 1703–14. Fisher, S. E. 2005. “On genes, speech, and language.” New England Journal of Medicine 353.16: 1655–7. Fitch, W. Tecumseh, and Marc D. Hauser. 2004. “Computational constraints on syntactic processing in a nonhuman primate.” Science 303 (January): 377–80. Fogassi, Leonardo, and Pier Francesco Ferrari. 2007. “Mirror neurons and the evolution of embodied language.” Current Directions in Psychological Science 16:136–41. Galaburda, Albert M., and Deepak N. Pandya. 1983. “The intrinsic architectonic and connectional organization of the superior temporal region of the rhesus monkey.” Journal of Comparative Neurology 221 (December): 169–84. Gallese, Vittorio, Luciano Fadiga, et al. 1996. “Action recognition in the premotor cortex.” Brain 119: 593–609. Gannon, Patrick J., Ralph L. Holloway, et al. 1998. “Asymmetry of chimpanzee planum temporale: Humanlike pattern of Wernicke’s brain language area homolog.” Science 279 (January): 220–2. Gentilucci, Maurizio, and Michael C. Corballis. 2006. “From manual gesture to speech: A gradual transition.” Neuroscience and Biobehavioral Reviews 30: 949–60. Gentner, Timothy Q., Kimberly M. Fenn, et al. 2006. “Recursive syntactic pattern learning by songbirds.” Nature 440 (April): 1204–7. Geschwind, D. H., and B. L. Miller. 2001. “Molecular approaches to cerebral laterality: Development and neurodegeneration.” American Journal of Medical Genetics 101: 370–81. Gil-da-Costa, Ricardo, Alex Martin, et al. 2006. “Species-specific calls activate homologs of Broca’s and Wernicke’s areas in the macaque.” Nature Neuroscience 9 (July): 1064–70. Green, Richard E., Johannes Krause, et al. 2006. “Analysis of one million base pairs of Neanderthal DNA.” Nature 444 (November): 330–6. Grodzinsky, Yosef, and Katrin Amunts, eds. 2006. Broca’s Region. Oxford: Oxford University Press. Haesler, Sebastian, Kazuhiro Wada, et al. 2004. “FoxP2 expression in avian vocal learners and non-learners.” Journal of Neuroscience 24 (March): 3164–75. Hauser, Marc D., David Barner, et al. 2007. “Evolutionary linguistics: A new look at an old landscape.” Language Learning and Development 3.2: 101–32. Hauser, Marc D., Noam Chomsky, and W. Tecumseh Fitch. 2002. “The faculty of language: What is it, who has it, and how did it evolve?” Science 298 (November): 1569–79. Holy, Timothy E., and Zhongsheng Guo. 2005. “Ultrasonic songs of male mice.” PLoS Biology 3 (December): 2177–86. Available online at: www. plosbiology.org. Iacoboni, Marco, Roger Woods, et al. 1999. “Cortical mechanisms of human imitation.” Science 286: 2526–8. Jenkins, Lyle. 2000. Biolinguistics: Exploring the Biology of Language. Cambridge: Cambridge University Press. Jenkins, Lyle, ed. 2004. Variation and Universals in Biolinguistics. Amsterdam: Elsevier. Kegl, Judy. 2004. “Language emergence in a language-ready brain: Acquisition.” In Variation and Universals in Biolinguistics, ed. L. Jenkins, 195–236. Amsterdam: Elsevier. Kennedy, D. N., K. M. O’Craven, et al. 1999. “Structural and functional brain asymmetries in human situs inversus totalis.” Neurology 53 (October): 1260–5.

54

Kohler, Evelyne, Christian Keysers, et al. 2002. “ Hearing sounds, understanding actions: Action representation in mirror neurons.” Science 297: 846–8. Lai, Cecilia S. L., Simon E. Fisher, et al. 2001. “A forkhead-domain gene is mutated in a severe speech and language disorder.” Nature 413 (October): 519–23. Liberman, Alvin M., and Ignatius G. Mattingly. 1985. “The motor theory of speech perception revised.” Cognition 21: 1–36. Liberman, Mark. 2004. “Hi lo hi lo, It’s off to formal language theory we go.” Language Log (January 17). Available online at: http://itre.cis. upenn.edu/~myl/languagelog/archives/000355.html. Lieberman, Philip. 2006. Toward an Evolutionary Biology of Language. Cambridge: Belknap Press of Harvard University Press. Lu, Min Min, Shanru Li, et al. 2002. “Foxp4: A novel member of the Foxp subfamily of winged-helix genes co-expressed with Foxp1 and Foxp2 in pulmonary and gut tissues.” Mechanisms of Development 119, Supplement 1 (December): S197–S202. Lyon, Caroline, Chrystopher L. Nehaniv, et al., eds. 2006. Emergence of Communication and Language. New York: Springer. Maynard Smith, John, and Erös Szathmáry. 1998. The Major Transitions in Evolution. New York: Oxford University Press. McMaster, Geoff. 2005. “Researchers discover cause of speech defect.” University of Alberta ExpressNews (October 25). Available online at: http://www.expressnews.ualberta.ca. Nishitani, N., and R. Hari. 2000. “Temporal dynamics of cortical representation for action.” Proceedings of the National Academy of Sciences USA 97: 913–18. Niyogi, Partha. 2004. “Phase transitions in language evolution.” In Variation and Universals in Biolinguistics, ed. L. Jenkins, 57–74. Amsterdam: Elsevier. ———. 2006. The Computational Nature of Language Learning and Evolution. Cambridge, MA: MIT Press. Noonan, James P., Graham Coop, et al. 2006. “Sequencing and analysis of Neanderthal genomic DNA.” Science 314 (November): 1113–18. Nowak, Martin A. 2006. Evolutionary Dynamics: Exploring the Equations of Life. Cambridge: Harvard University Press. Nowak, Martin A., and Natalia L. Komarova. 2001. “Towards an evolutionary theory of language.” Trends in Cognitive Sciences 5 (July): 288–95. Pennisi, Elizabeth. 2009. “Neanderthal Genomics: Tales of a Prehistoric Human Genome.” Science 323. 5916: 866–71. Pettito, Laura-Ann. 2005. “How the brain begets language.” In Cambridge Companion to Chomsky, ed. J. McGilvray, 84–101. Cambridge: Cambridge University Press. Petrides, M., and D. N. Pandya. 1994. “Comparative architectonic analysis of the human and the macaque frontal cortex.” In Handbook of Neuropsychology, ed. F. Boller and J. Grafman, 17–58. Amsterdam: Elsevier. Poeppel, David, and Gregory Hickok. 2004. “Towards a new functional anatomy of language.” Cognition 92 (May/June): 1–12. Rizzolatti, Giacomo, and Michael A. Arbib. 1998. “Language within our grasp.” Trends in Neurosciences 21 (May): 188–94. Roll, Patrice, Gabrielle Rudolf, et al. 2006. “Srpx2 mutations in disorders of language cortex and cognition.” Human Molecular Genetics 15 (April): 1195–1207. Scharff, Constance, and Sebastian Haesler. 2005. “An evolutionary perspective on FoxP2: Strictly for the birds?” Current Opinion in Neurobiology 15 (December): 694–703. Scharff, C., and F. Nottebohm. 1991. “A comparative study of the behavioral deficits following lesions of various parts of the zebra finch song system: Implications for vocal learning.” Journal of Neuroscience 11 (September): 2896–913.

Explaining Language Shapiro, K. A., L. R. Moo, et al. 2006. “Cortical signatures of noun and verb production.” Proceedings of the National Academy of Sciences USA 103 (January): 1644–49. Sherwood, Chet C., Douglas C. Broadfield, et al. 2003. “Variability of Broca’s area homologue in African great apes: Implications for language evolution.” The Anatomical Record Part A: Discoveries in Molecular, Cellular, and Evolutionary Biology 271A (March): 276–85. Shu, Weiguo, Julie Y. Cho, et al. 2005. “Altered ultrasonic vocalization in mice with a disruption in the Foxp2 gene.” Proceedings of the National Academy of Sciences USA 102 (July 5): 9643–8. Shu, W., H. Yang, et al. 2001. “Characterization of a new subfamily of winged-helix/forkhead (fox) genes that are expressed in the lung and act as transcriptional repressors.” Journal of Biological Chemistry 276.29: 27488–97.

Somerville, M. J., C. B. Mervis, et al. 2005. “Severe expressive-language delay related to duplication of the Williams-Beuren locus.” New England Journal of Medicine 353.16: 1694–701. Sun, Tao, Christina Patoine, et al. 2005. “Early asymmetry of gene transcription in embryonic human left and right cerebral cortex.” Science 308 (June ): 1794–8. Sun, Tao, and Christopher A. Walsh. 2006. “Molecular approaches to brain asymmetry and handedness.” Nature Reviews Neuroscience 7 (August): 655–62. Tettamanti, Marco, Giovanni Buccino, et al. 2005. “Listening to actionrelated sentences activates fronto-parietal motor circuits.” Journal of Cognitive Neuroscience 17: 273–81. White, Stephanie A., Simon E. Fisher, et al. 2006. “Singing mice, songbirds, and more: Models for Foxp2 function and dysfunction in human speech and language.” Journal of Neuroscience 26 (October): 10376–9.

55

6 ACQUISITION OF LANGUAGE Barbara Lust

Yet normally, by about three years of age, a child will have attained the fundamental knowledge of an infinite combinatorial multileveled system as well as essential constraints on this infinite system, no matter what the language, no matter what the country or culture the child is born into, no matter what the limitless contextual, environmental variations across children, cultures, and countries. This mystery, and the paradox of constrained infinity, drives the field of language acquisition. How is this fundamental knowledge acquired? How is it represented and effected in the mind and brain?

THE FIELD OF LANGUAGE ACQUISITION

INTRODUCTION How does the child, within the first few years of life, come to understand and to produce “Hop on Pop” or “Cat in the Hat” (Dr. Seuss)? (1) We all play ball Up on a wall Fall off the Wall (Dr. Seuss [1963] 199l, 10–13) (2) We looked! And we saw him! The Cat in the Hat! And he said to us … (Dr. Seuss [1957] 1985, 6)

How does the child come to know the precise sound variations that distinguish “hop” and “pop” or “wall” and “ball,” the multiple meanings of words like “play” or “in,” the reference of pronouns like “we” or “him,” the meaning of quantifiers (see quantification) like “all,” the functional categories determining noun phrases like “the,” the idioms like “play ball,” the systematic order and structure necessary for even simple sentences, the infinite recursion of sentences allowed by the coordinate connective “and,” the infinite possibilities for propositions based on the manipulation of even a few sound units and word units, and the infinite possibilities for meaning and truth, which Dr. Seuss continually demonstrated? Even more stunning, how does the child both attain the infinite set of possibilities and, at the same time, know the infinite set of what is impossible, given this infinitely creative combinatorial system? Just as there is no limit to what is possible, there is no limit to what is impossible in language, for example, (3), or an infinite set of other impossible word and structure combinations, for example, (4). This provides an essential paradox for language learners: How do they both acquire the infinite possibilities and, at the same time, the constraints on the infinite set of impossible expressions? (3) *Ylay *lfla (4) *Play all we ball *in cat the hat

56

In many ways, the study of language acquisition stands at the center of the language sciences, subsumes all of its areas, and thus, perhaps, supersedes all in its complexity. It entails the study of linguistic theory in order that the systems and complexities of the end state of language knowledge be understood, so that the status of their acquisition can be evaluated. This involves all areas of linguistics, in addition to phonetics: phonology, morphology, syntax, semantics, and pragmatics (cf. Chierchia 1999). It entails language typology because acquisition of any specific language is possible, and because this acquisition reflects any possible typological variation. Barbara Grimes (2000) proposes that between 6,000 and 7,000 languages now exist in the end state. The study of language acquisition also entails the study of language typology because language acquisition at any state of development represents a form of manifestation of language knowledge, itself subject to typological variation. It entails language change, because the course of acquisition of language over time reveals changing manifestations of language; language change must be related to language acquisition, in ways still not understood (although see Baron 1977, deGraff 1999, and Crain, Gloro, and Thornton 2006 for initial attempts to relate these areas; see also age groups.). It involves psycholinguistics because any particular manifestation of language during language acquisition is revealed through the cognitive and psychological infrastructure of the mind and brain during “hearing” and “speaking” in real time, potentially within a visual modality as in sign languages, thus always involving language processing integrated with language knowledge. It involves neurolinguistics, not only because the knowledge and use of language must in some way relate to its biological and neural instantiation in the brain, but also because explanations of the gradual course of language acquisition over time must critically evaluate the role of biological change. (See brain and language and biolinguistics.) At the same time, the study of language acquisition also stands at the center of Cognitive Science. It cannot be divorced from other closely related disciplines within Cognitive Science. The relation between language and thought is never dissociated in natural language, thus crucially calling for an understanding of cognitive psychology and philosophy; and the formal computation involved in grammatical systems is not fruitfully divorced from computer science. Any use of language involves possibilities for vagueness and ambiguities in interpretation, given the particular pragmatic context of any language use. For example, how would one fully understand the meaning of an ostensibly simple sentence like

Acquisition of Language “She’s going to leave” without understanding the particular context in which this sentence is used, and there is an infinite set of these contexts possible. The challenge of developing an infinitely productive but constrained language system embedded in the infinite possibilities for pragmatic determination of interpretation remains a central challenge distinguishing machine language and the language of the human species. All areas of cognitive science must be invoked to study this area. Finally, especially with regard to language acquisition in the child, developmental psychology must be consulted critically. The child is a biologically and cognitively changing organism. Indeed, an understanding of commonalities and/or differences between language acquisition in child and adult requires expertise in the field of developmental psychology. In addition, the fact that the human species, either child or adult, is capable not only of single language acquisition but also of multilanguage acquisition, either simultaneously or sequentially, exponentially increases the complexity of the area (see bilingualism and multilingualism). The fact that these languages may vary in their parametric values (see parameters) and be either oral or visual further complicates the area.

THEORETICAL FOUNDATIONS Not surprisingly, then, the field of research involving the study of language acquisition is characterized by all the complexity and variation in theoretical positions that characterize the field of linguistics and the language sciences in general. The area of language acquisition research reflects varying approaches to the study of grammar, that is, variations regarding viewpoints on what constitutes the end state of language knowledge that must be acquired, for example, generative or functionalist approaches. At the same time, it is characterized by variation in approach to the study of language acquisition, in particular, ranging from various forms of logical to empirical analyses, and including disputes regarding methodological foundations and varying attempts at explanatory theories (ranging from rationalist to empiricist types). (Lust 2006, Chapter 4, provides a review.) The field is led by a strong theory of what is necessary logically for a strong explanatory theory of language acquisition in the form of Noam Chomsky’s early proposal for a language acquisition device (LAD). This model not only involved explication of what needed to be explained but also spawned decades of research either for or against various proposals bearing on components of this model. Yet, today, no comprehensive theory of language acquisition exists, that is, no theory that would fully account for all logically necessary aspects of a LAD. At the same time, decades of research have now produced an explosion of new discoveries regarding the nature of language acquisition. These empirical discoveries, combined with theoretical advances in linguistic theory and with the development of the interdisciplinary field of cognitive science, bring us today to the verge of such a comprehensive theory, one with firm scientific foundations.

STRUCTURE OF THIS OVERVIEW In this overview, I first briefly characterize and distill focal tensions in the field of language acquisition and briefly survey

varying positions in the field on these focal issues, then exemplify a range of research results, displaying crucial new discoveries in the field. I’ll conclude by formulating leading questions for the future. For those interested in pursuing these topics further, I close with selected references for future inquiry in the field.

Focal Tensions Classically, approaches to the study of language acquisition have been categorized as nativist or empiricist (see innateness and innatism). These approaches, which correspond generally to rationalist or empiricist approaches to epistemology (see Lust 2006 for a review), have been typically associated with claims that essential faculties responsible for language acquisition in the human species involve either innate capabilities for language or not. However, now the debates have become more restricted, allowing more precise scientific inquiry. No current proposals, to our knowledge, suggest that nothing at all is innate. (See Elman et al, 1996 for example). No one proposes that every aspect of language knowledge is innate. For example, in the now-classic Piaget-Chomsky Debate (see Piatelli-Palmarini 1980), Chomsky did not deny that experience was necessary for a comprehensive theory of language acquisition. On the other hand, Jean Piaget proposed an essentially biological model of cognitive development and an antiempiricist theory, coherent with Chomsky’s position in this way. Rather, the current issue that is most central to the field of language acquisition now is whether what is innate regarding language acquisition involves components specific to linguistic knowledge, for example, specific linguistic principles and parameters, or whether more general cognitive knowledge or learning principles – themselves potentially innate – can account for this knowledge. This issue corresponds to the question of whether organization of the mental representation of language knowledge is modular or not. Proponents of the view that specifically linguistic factors are critical to an explanation of human language acquisition generally work within a linguistic theory of a modular language faculty, that is, a theory of universal grammar (UG), which is proposed to characterize the initial state of the human organism, that is, the state of the human organism before experience. This current descendant of Chomsky’s LAD provides specific hypotheses (working within a generative grammar framework) regarding the identity of linguistic principles and parameters that may be innately, or biologically, determined in the human species (e.g., Anderson and Lightfoot 2002). Proponents of the alternative view generally assume a functionalist theory and a model of cultural learning, that is, a model where culture in some ways provides the structure of language, potentially without specific linguistic printiples, with only general learning principles (e.g., Van Valin 1991; Tomasello 2003; Tomasello, Kruger, and Ratner 1993).

Positions in the Field Debates in the field of language acquisition continue today around issues of innateness and modularity (e.g., Pinker, 1994; Tomasello 1995; Edelman 2008). Research in the field is often polarized according to one paradigm or the other. Several

57

The Cambridge Encyclopedia of the Language Sciences specific issues focalize these debates. Two of the most pressing currently concern the nature of development, and the nature of the child’s use of language input. Critically, all approaches must confront the empirical fact that children’s language production and comprehension develop over time. Language acquisition is not instantaneous. The question is whether children’s language knowledge also develops over time, and if it does, does it change qualitatively, and if it does change qualitatively, what causes the change. This question of development is parallel to questions regarding cognitive development in general, where “stage theories” are debated against “continuity” theories. Positions in the field today vary in several ways regarding their understanding of the nature of development in language acquisition. Even within researchers working within a rationalist paradigm that pursues universal grammar as a specifically linguistic theory of a human language faculty, there is disagreement. The theory of UG does not a priori include an obvious developmental component. Positions in the field can be summarized as in i–iv. (i) Some propose that essential language knowledge is fundamentally innate, and do not centrally address the issue of what change in language knowledge may occur during language development (e.g., Crain 1991). In this view, apparent cases of language knowledge change in a young child’s production or comprehension are often mainly attributed to methodological failures, for example, involving the researchers’ choice of specific tasks for testing children’s knowledge. The view of language development is characterized by the following: “On the account we envision, children’s linguistic experience drives children through an innately specified space of grammars, until they hit upon one that is sufficiently like those of adult speakers around them, with the result that further data no longer prompts further language change” (Crain and Pietroski 2002, 182). (Compare LAD and a recent proposal by Yang 2006.) On this view, it would appear that grammars are predetermined and available for choice. Presumably, specific lexicons, as well as other peripheral aspects of language knowledge, would be exempt from this claim of predetermined knowledge. The range and nature of the “innately specified space of grammars” would have to be explicated, as this framework develops. (ii) Some propose that UG itself develops over time in a manner that is biologically driven, that is, through “maturation.” Major aspects of language development, for example late acquisition of the passive construction (such as “The ball was thrown by Mary” in English), are attributed to biologically determined changes in UG (e.g., Wexler 1999; Radford 1990). This maturation is proposed to be the major explanation of language development. Specific language grammars (SLGs) are not, for the most part, treated as essential independent components of the language acquisition challenge, but often as simply “triggered” by UG parameters. Any learning that would be involved for the child in language development would involve merely “peripheral” aspects of language, with the nature of the periphery yet to be specified. (iii) Some propose that UG is continuous, but UG is interpreted as involving just those principles and parameters that constrain and guide the language acquisition process. SLG develops over time through experience with data from a specific language and through the child’s active construction of it. (See examples

58

of a proposal for “grammatical mapping” in Lust 1999, 2006 and Santelmann et al. 2002.) The task of language acquisition goes beyond the “periphery” of grammatical knowledge and beyond mere triggering of parameters; it lies in constructive synthesis of a specific language grammatical system, which is constrained by UG but not fully determined by it. (iv) Some propose that UG is irrelevant to language acquisition; the process can be described by an alternative mechanism, critically involving some form of “usage-based” learning (e.g., Elman et al. 1996; Tomasello 2003, 2005): “In general, the only fully adequate accounts of language acquisition are those that give a prominent role to children’s comprehension of communicative function in everything from words to grammatical morphemes to complex syntactic constructions” (Tomasello 2005, 183). Within the proposals of i–iii which all work within a generative UG framework, (i) and (iii) propose a strong continuity hypothesis of language development, although they differ crucially in what is claimed to be continuous. In (iii), it is only the essential set of principles and parameters of UG constituting the initial state that is proposed to be biologically programmed and to remain constant over development, while in (i), it is comprehensive grammatical knowledge, with no distinction made between UG and SLG. While a “maturational” approach such as in (ii) would appear to maintain the premise of biological programming of language knowledge, and thus be consistent with a theory of UG, it raises several critical theoretical and empirical issues that are still unresolved. For example, theoretically, the question arises: What explains the change from one UG state to another, if this determination is not programmed in UG itself? Empirically, in each area where a maturational account has been proposed, further advanced research has often revealed grammatical competence that was thought to be missing in early development, for example, early knowledge of functional categories. (See Lust 1999 and Wexler 1999 for debate.) All proposals must now confront the fundamental developmental issues: what actually changes during language development and why. Those within the generative paradigm must sharpen their vision of what in fact constitutes UG, and those outside of it must sharpen their view of how infinitely creative but infinitely constrained grammatical knowledge can be attained on the basis of “communicative function” alone. All proposals must be accountable to the wide and cumulative array of empirical data now available. (See, for example, “Discoveries” in the next section.) Researchers are also opposed in their view of how the child uses input data. (See Valian 1999 for a review.) All approaches must confront this fundamental area, whether the UG framework is involved (e.g., in explaining how the environment “triggers” parameter setting) or whether an empiricist framework is involved (where the mechanism of imitation of individual words and/or utterances may be viewed as an essential mechanism of data processing and grammar development for the child, e.g., Tomasello 1995, 2002, 2005). Indeed, the very nature of imitation of language stimuli is under investigation (e.g., Lust, Flynn, and Foley 1996). In each case, the mechanism proposed must reliably account for the child’s mapping from these input data to the specific knowledge of the adult state, thus solving what has been called the “logical problem of language acquisition” and it must

Acquisition of Language also be empirically verified, that is, be a veridical account of how the child actually does use surrounding data. Those approaching the logical problem of language acquisition are formally diagnosing the properties of syntactic and semantic knowledge that must be acquired, and assessing varied formal “learnability theories” that may possibly account for the child’s mapping from actual input data to this knowledge. (e.g., Lightfoot 1989). (See also projection principle.)

Discoveries Fortunately, not only have the last decades seen continuous theoretical developments in areas of linguistics regarding the nature of language knowledge, and continuous sharpening of the debate on the nature of language development, but the field of language acquisition has also produced a wealth of new empirical discoveries, all of which promise to inform the crucial questions and debates in the field, and to eventuate in a more comprehensive theory of language acquisition than has yet been available. I’ll exemplify only some of these here, all of which are foci of intense current research. (See Lust 2006 for a more comprehensive review.) THE FIRST 12 MONTHS. Informed by developed methodologies for investigating infant perception of language, we now know that neonates show early sensitivities to language variations and categorize these variations (e.g., Ramus et al. 2000; Mehler et al. 1996, 1988). For example, given varying sound stimuli from Japanese and varying sound stimuli from English speech, newborn infants distinguish variation from Japanese to English significantly more than variation within either English or Japanese, including speaker variation. In this sense, the infant is seen to be categorizing variation across languages. (See infantile responses to language.) More specifically, we also know that formal aspects of language (phonetic, phonological, and syntactic) begin to develop at birth, even before language comprehension, and provide the foundations for the appearance of overt language production and comprehension in the child at about 12 months of age. Now we also know something of how that development proceeds even during the first 12 months of life. In the area of acquisition of phonology, we have discovered the very fine, precise, and extensive phonetic sensitivities of the newborn, sensitivities that appear to be exactly the right ones to underlie all potential cross-linguistic phonological contrasts, for example, contrasts in voicing or place and manner features that characterize speech sounds. Once again, these sensitivities are categorical (as in categorical perception; see speech perception in infants). We know that by about 6 months, these sensitivities reveal appropriate cross-linguistic selection and modulation, and by 12 months, this process is nicely attuned to the child’s first productive words of their specific language (e.g., Werker 1994; see Jusczyk 1997 and de Boysson-Bardies 1999 for reviews). For example, while the 12-month-old infant acquiring Hindi will have maintained sensitivity to Hindi contrasts in aspiration, the infant acquiring English will show diminished response to such distinctions, which are not linguistically contrastive in English. Although labials appear in first words across

languages, the proportion of labials in both late babbling and first words will reflect the input of labials in the adult language being acquired (de Boysson-Bardies, Vihman, and Vihman 1991). In fact, children are picking out words as formal elements from the speech stream even before they understand them. For example, 8-month-olds who were exposed to stories read to them for 10 days (30 minutes of prerecorded speech, including three short stories for children) during a two-week period, two weeks later distinguished lists of words that had appeared in these stories, for example, (5a), from those which had not, for example, (5b) (Jusczyk and Hohne 1997). (5) a. sneeze elephant python peccaries b. aches apricot sloth burp

Given the age of the children and the nature of the words tested, it is clear that children are engaged both in word segmentation and in long-term storage of these formal elements, even without their semantic content. Moreover, like the acquisition of phonology, children’s sensitivities to the language-specific structure of words begins to show refinement after the sixth month. American 9-month-olds, though not 6-month-olds, listened longer to English words than to Dutch words, while Dutch 9-month-old infants showed the reverse preference (Jusczyk et al. 1993). Simultaneously, and in parallel, in the area of syntax, infants have also been found to be laying the formal foundations for language knowledge even within the first 12 months. They are carving out the formal elements that will form the basis for the syntax of the language they are acquiring. Precise sensitivities to linear order, as well as to constituent structure, have been discovered in these first few months of life. For example, infants as young as 4 months of age have been found to distinguish natural well-formed clause structure, like (6a) from non-natural ones, like (6b), in stories read to them (where / represents clause breaks through pauses, and the stimuli are matched in semantic and lexical content) (Hirsh-Pasek et al. 1987). (6) a. Cinderella lived in a great big house/ but it was sort of dark/ because she had this mean, mean, mean stepmother … b. …in a great big house, but it was/ sort of dark because she had/ this mean …

Experimental research has begun to reveal how infants accomplish this discrimination in the speech stream, suggesting that the mapping of prosodic phrasing to linguistic units may play an important role. Again, sensitivities become more language specific after the sixth month. Although 6-month-olds did not distinguish natural (e.g., 7a) and non-natural phrasal (verb phrase) structures (e.g., 7b), that is, phrasal constituents smaller than the clause, 9-month-olds did:

59

The Cambridge Encyclopedia of the Language Sciences (7) a. The little boy at the piano/ is having a birthday party … b. The little boy at the piano is having/ a birthday party

In all these cases, both phonological and syntactic development does not reduce either to simple loss of initial sensitivities or to simple accrual or addition of new ones, but a gradual integration of a specific language grammatical system. More recently, research has begun to reveal even more precisely how these early sensitivities are related to later language development, thus foreshadowing a truly comprehensive theory of language acquisition (Newman et al. 2006; Kuhl et al. 2005). BEYOND FIRST WORDS: LINGUISTIC STRUCTURE. Continuous with advances in our understanding of early infant development, we are now also seeing a potential revolution in our understanding of the early stages of overt language acquisition, that is, those periods within the first three years of life, where the child is beginning to overtly manifest language knowledge in terms of language production and comprehension. Child language in these early periods has traditionally been referred to as holophrastic or, a little later, telegraphic (see two-word stage) in nature. Even in these early periods, numerous studies are now revealing children’s very early sensitivity to functional categories (FCs) in language, that is, to grammatical elements that function formally to a large degree, often with little or no semantic content. These functional categories play a critical role in defining constituent structure in language, and in defining the locus of syntactic operations. Thus, the evidence that infants and toddlers are accessing these FCs in their early language knowledge begins to provide crucial data on the foundations for linguistic systems in the child. (Such FC are reflected in different ways across languages. In English they are reflected in determiners such as “the,” auxiliary verbs like “do,” complementizers like “that,” or inflection of verbs, for example.) Early research had revealed that young children perceive and consult functional categories, such as determiners, even when they are not overtly producing them regularly in their natural speech. For example, L. Gerken, B. Landau, and R. Remez (1990) showed that 2-year-olds recognize the distinction between grammatical and ungrammatical function words, contrasting these, for example (8a) and (8b), in their elicited imitation of these sentences. Gerken and Bonnie McIntosh (1993) showed that 2-year-olds used this knowledge in a picture identification task, discriminating between (9a) and (9b), where grammatical forms facilitated semantic reference. (8) a. Pete pushes the dog b. Pete pusho na dog (9) a. Find the bird for me b. Find was bird for me

More recently, a wide range of experimental researchers working with expanded infant testing methods have replicated these results, and also revealed similar functional category distinctions even in younger children. For example Yarden Kedar, Marianella Casasola, and Barbara Lust (2006) showed that infants as young as 18 months also distinguish sentences like (9a) and (9b) in a preferential looking task, and again, object reference is facilitated by the grammatical form. Precursors to these functional

60

category sensitivities appear to be available even within the first 12 months (e.g., Shady 1996; Demuth 1994). Contrary to the widespread view that the contentful lexicon (involving nouns and verbs) is the privileged basis for acquisition of syntax in early language acquisition, these results are beginning to suggest that, in fact, functional categories are fundamental (see lexical learning hypothesis). PRINCIPLES AND PARAMETERS. Principles and parameters that are hypothesized to provide the linguistic content of UG and of the human language faculty provide leading hypotheses for language acquisition research. (See principles and parameters theory and language acquisition.) Investigators continue to search not only for theoretical motivation for such principles and parameters but also for empirical evidence of the role of UG-based principles and parameters in early language acquisition. A wide range of empirical research has now accrued in this paradigm, paralleling theoretical developments (e.g., Snyder 2007; Crain and Lillo-Martin 1999; Guasti 2002; Roeper 2007; Lust 2006, among others). This research reveals a wide array of evidence regarding very fundamental linguistic principles, including the central and perhaps most fundamental UG principle of structure dependence: The rules operate on expressions that are assigned a certain structure in terms of a hierarchy of phrases of various types. (Chomsky 1988, 45)

Evidence for this fundamental linguistic principle has been adduced in studies of children’s acquisition of several areas, including various types of question formation (e.g., Crain and McKee 1985; Crain and Nakayama 1987; deVilliers, Roeper, and Vainikka 1990), empty category and pronoun interpretation (e.g., Cohen Sherman and Lust 1993; Nuñez del Prado, Foley, and Lust 1993; Lust and Clifford 1986), and quantifier scope (e.g., Chien 1994). Results from young children at early stages of development across languages have shown that they very early distinguish coordinate and subordinate structures and that they differentiate syntactic processes in these different structures accordingly. For example, in English, they very early distinguish sentences like (10a) and (10b) in both comprehension and production (Cohen Sherman and Lust 1993): (10) a. [The turtle tickles the skunk] and [0 bumps the car] b. Tom [promises/tells Billy [0 to eat the ice cream cone]]

Chinese children differentiate coordinate and embedded structures and differentiate subjects and topics (see topic and comment) in Chinese accordingly (Chien and Lust 1985). Across English, Japanese, and Sinhala, children differentiate possibilities for anaphora according to the embedded or adjoined structure in which proforms appear (Lust and Clifford 1986; Oshima and Lust 1997; Gair et al. 1998; Eisele and Lust 1996). In general, very early linguistic knowledge – including knowledge of language involving diminished content, where direct overt phonetic information is not available – is attested in studies of children’s acquisition of sentences with ellipses. For example, in sentences such as (11), young children have been found to reveal not only competence for empty category interpretation

Acquisition of Language

(in the “does too” clause, where the second clause does not state what Bert did), but also competence for construction of the multiple interpretations allowed in this ambiguous structure (as in 11a–d), and constraint against the ungrammatical possibilities (as in 11 e–i). (See Foley et al. 2003 for an example.) In other words, they evidence early control of and constraints on empty category interpretation and on other forms of ellipsis. Studies of Chinese acquisition show similar results (Guo et al. 1996). Without structure dependence and grammatical computation over abstract structure underlying such sentences (e.g., reconstructing the verb phrase [VP] in the second clause), children would not be expected to evidence this competence (see Foley et al. 2003). All interpretations are pragmatically possible. (11) Oscar bites his banana and Bert does too. Possible Interpretations: a. O bites O’s banana and B bites B’s banana. b. O bites O’s banana and B bites O’s banana. c. O bites B’s banana and B bites B’s banana. d. O bites E’s banana and B bites E’s banana. Impossible Interpretations *e. O bites O’s banana and B bites E’s banana. *f. O bites B’s banana and B bites O’s banana. *g. O bites B’s banana and B bites E’s banana. *h. O bites E’s banana and B bites O’s banana. *i. O bites E’s banana and B bites B’s banana.

As is the case in this study of VP ellipsis acquisition, empirical research results in the Principles and Parameters framework mutually inform theoretical development; they contribute to and help to resolve theoretical debates on the representation of this area of linguistic knowledge. Evidence for early parameter setting with regard to linear order in natural language is now provided by a wide range of cross-linguistic studies showing very early language-specific sensitivity to the clausal head direction and directionality of RECURSION in the language being acquired (e.g., Lust, in preparation; Lust and Chien 1984). Very early differentiation of the pro drop (i.e., argument omission wherein subjects may not be overtly expressed) possibilities of a language have been attested across languages (e.g., Italian and English, Spanish and English) (Valian 1991; Austin et al. 1997, after Hyams 1986). In fact, children have been found to critically consult the complementizer phrase, and the subordinate structure domain in order to make this differentiation (Nuñez del Prado, Foley, and Lust 1993). FROM DATA TO GRAMMAR. Our understanding of how the infant, from birth, consults the surrounding speech stream is now also expanding quickly. Research has revealed very fine sensitivities to particular aspects of the speech stream (e.g., Saffran, Aslin, and Newport 1996), and research has begun to isolate the precise role of particular cues in this process, for example, STRESS, phonotactic constraints, and statistical distributions (e.g, Johnson and Jusczyk 2001). Various forms of bootstrapping may be available to the child (phonological or prosodic or syntactic bootstrapping, for example). Here, bootstrapping generally refers to that process by which the child, in the initial state, might initiate new

learning by “hanging on” to some aspect of the input that it can access. Research is beginning to provide evidence on how and when these forms of bootstrapping may work in the young child (e.g, work of Gleitman 1990 and others on “syntactic bootstrapping” and its role in early lexical learning and the the collection of papers in Morgan and Demuth 1996). Precise hypotheses are now being formed regarding the mechanisms by which certain parameters may be set very early by the infant, even before the first word, by consulting prosodic and other aspects of the speech stream (Mazuka 1996). CROSS-SPECIES COMPARATIVE METHOD. Advances in crossspecies comparisons now provide an additional dimension to the study of language acquisition, allowing refinement of our specification of what is particularly human and of what is particularly linguistic about human acquisition of language (e.g., Hauser, Chomsky, and Fitch 2002; Call and Tomasello 2007; Ramus et al. 2000). For example, comparative studies with cotton-top tamarin monkeys revealed certain capabilities in this species to discriminate language stimuli (e.g., Dutch and Japanese) that were comparable to human infants (Ramus et al. 2000). This implied that a general auditory process was involved in the discrimination. In contrast, other processes discovered in early acquisition of phonology have been found not to generalize. (See, for example, the work of Kuhl et al. 2005.) RESILIENCE. Finally, the tremendous resilience of the language acquisition feat in the face of varying input, including dearth of input, has been revealed through important work on young deaf children’s spontaneously created home sign (Goldin-Meadow 2003). The role of community in sign language creation is also revealed through in-depth studies of the creation of Nicaraguan Sign Language in young children (Senghas 1995; Kegl, Senghas, and Coppola 1999).

Toward the Future FIRST AND SECOND LANGUAGE IN CHILD AND ADULT. Current and future studies that seek to “triangulate the language faculty” by precise comparative studies between child first language acquisition and adult second language and between child monolingual first language acquisition and multilanguage acquisition promise to be able to dissociate factors related to biological maturation, universal grammar, and specific language grammar in ways not achievable by studies of first language acquisition alone (e.g., Flynn and Martohardjono 1994; Yang and Lust 2005). PRAGMATIC AND COGNITIVE DEVELOPMENT AND GRAMMATICAL DEVELOPMENT. Studies of the integration of children’s developing pragmatic knowledge with their grammatical knowledge have only begun; yet this integration characterizes every aspect of a child’s use of language (cf. Clark 2003). Such studies may critically inform our understanding of children’s early syntax development (e.g., Blume 2002). Similarly, studies involving interactions between general cognition and specific aspects of linguistic knowledge (e.g., Gentner and Goldin-Meadow 2003) will be critical for resolving issues of modularity in language acquisition. (See constraints in language acquisition.)

61

The Cambridge Encyclopedia of the Language Sciences A NEUROSCIENCE PERSPECTIVE AND BEYOND. Current advances in brain imaging methods (such as fMRI, and neurophysiological measures, such as EEG [electroencephalography] and MEG [magnetoencephalography]) and in genome mapping (Greally 2007) promise new discoveries regarding fundamental issues still open in the field of language acquisition: How are language knowledge and language acquisition represented in the brain? (See neuroimaging and genes and language). More precisely, what is the content of the initial state and how is it biologically represented? How is development in language knowledge either determined by biological changes or a cause of them? There have been many neuroscientific results regarding language in the adult state (e.g., Friederici 2002; Binder and Price 2001; see syntax, neurobiology of; phonetics and phonology, neurobiology of; and semantics, neurobiology of, for example), including multilingual language (e.g. Paradis 2000; see bilingualism, neurobiology of). Research on brain development has also begun to yield fundamental descriptive results (e.g., Casey, Galvan and Hare 2005). Nevertheless, “it appears that the issue of the language-brain relationship during early development … is a terra incognita.… The issue of how brain development and cognitive development in the area of language development co-occur in early development and over a lifetime will be one of the key issues in the coming decades of the third millennium” (Friederici 2000, 66). Advances in the description of brain development (e.g., Almi et al. 2007) have not yet been related precisely to advances in language acquisition. Not only will advances in this area require technology applicable to a young and changing brain but also advances in the interdisciplinary field of cognitive science. They will depend on the development of strong linguistic theory, but also on the development of a theory that maps the elements of linguistic theory to the activation of language in the brain. At present, there is still no theory of how “an NP relates to a neuron” (Marshall 1980) or how an abstract linguistic principle like structure dependence or a particular linguistic constraint could be represented at a neural level. Most imaging results to date provide evidence on the processing of language, rather than on the knowledge of, or representations of, language per se. METHODOLOGY. Advances in this area will also require advancement on issues of methodology: “Simply put, we need to understand, precisely, how our behavioral methodologies reflect on, and interact with, the language processes we are attempting to study and define” (Swinney 2000, 241). Advancement in this area is a prerequisite to advancement in our understanding of language processing, language knowledge, and their interaction in language acquisition, as well as in the interpretation of brain imaging results. Many persistent sources of disagreement in the field of language acquisition depend on resolution of issues related to methodology (e.g., Crain and Wexler 1999; Lust et al. 1999), including those surrounding new infant-based methods related to preferential looking. LINGUISTIC THEORY AND LANGUAGE ACQUISITION. Finally, the study of language acquisition will be advanced when linguistic studies of the knowledge that must underlie the adult state of language knowledge, no matter what language, are brought into

62

line more fully with studies of the language acquisition process. (See Baker 2005 for an example of an argument for representation of this integration.) Many current disputes regarding the fundamentals of language acquisition cannot be resolved until disputes regarding the nature of the adult state of language knowledge are further resolved, and until the field of Linguistics and Psychology are more fully integrated in the field of Cognitive Science.

WORKS CITED AND SUGGESTIONS FOR FURTHER READING Aitchison, Jean. 2007. The Articulate Mammal. 5th ed. London: Routledge. Almi, C. R., M. J. Rivkin, R. C. McKinstry, and Brain Development Cooperative Group. 2007. “The NIH MRI study of normal brain development (Objective-2): Newborns, infants, toddlers, and preschoolers.” NeuroImage 35: 308–25. Anderson, Stephen R., and David W. Lightfoot. 2002. The Language Organ. Cambridge: Cambridge University Press. Austin, Jennifer, Maria Blume, David Parkinson, Zelmira Nuñez del Prado, Rayna Proman, and Barbara Lust. 1997. “The status of prodrop in the initial state: Results from new analyses of Spanish.” In Contemporary Perspectives on the Acquisition of Spanish, Vol 1: Developing Grammars, ed. Anna Perez-Leroux and W. Glass, 35–51. Boston: Cascadilla Press. Baker, Mark C. 2005. “Mapping the terrain of language learning.” Language Learning and Development 1.1: 93–129. Baron, Naomi. 1977. Language Acquisition and Historical Change. Amsterdam: North Holland. Binder, Jeffrey R., and Cathy J. Price. 2001. “Functional neuroimaging of language.” In Handbook of Functional Neuroimaging of Cognition, ed. Robert Cabeza and Alan Kingstone, 187–251. Cambridge, MA: MIT Press. Blake, Joanna. 2000. Routes to Child Language: Evolutionary and Developmental Precursors. Cambridge: Cambridge University Press. Blume, Maria. 2002. “Discourse-morphosyntax in Spanish non-finite verbs: A comparison between adult and child grammars. Unpublished Ph.D. diss., Cornell University. Boysson-Bardies, Bénédicte de. 1999. How Language Comes to Children: From Birth to Two Years. Cambridge, MA: MIT Press. Boysson-Bardies, Bénédicte de, B. Vihman, and M. M. Vihman. 1991. “Adaptation to language: Evidence from babbling and first words in four languages.” Language 67.2: 297–319. Call, Josep, and Michael Tomasello. 2007. The Gestural Communication of Apes and Monkeys. Mahwah, NJ: Lawrence Erlbaum. Casey, B. J., Adriana Galvan, and Todd Hare. 2005. “Changes in cerebral functional organization during cognitive development.” Current Opinion in Neurobiology 15: 239–44. Chien, Yu-Chin. 1994. “Structural determinants of quantifier scope: An experimental study of Chinese first language acquisition.” In Syntactic Theory and First Language Acquisition: Cross-Linguistic Perspectives. Vol 2: Binding, Dependencies, and Learnability. Ed. B. Lust, G. Hermon, and J. Kornfilt, 391–416. Mahwah, NJ: Lawrence Erlbaum. Chien, Yu-Chin, and Barbara Lust. 1985. “The concepts of topic and subject in first language acquisition of Mandarin Chinese.” Child Development 56: 1359–75. Chierchia, Gennaro. 1999. “Linguistics and language.” In The MIT Encyclopedia of the Cognitive Sciences, ed. R. Wilson and F. Keil, xci– cix. Cambridge, MA: MIT Press. Chomsky, Noam. 1988. Language and Problems of Knowledge. Cambridge, MA: MIT Press. Clark, Eve V. 2003. First Language Acquisition. Cambridge: Cambridge University Press.

Acquisition of Language Cohen Sherman, Janet, and Barbara Lust. 1993. “Children are in control.” Cognition 46: 1–51. Crain, Stephen. 1991. “Language acquisition in the absence of experience.” Behavioral and Brain Sciences. 14.4: 597–650. Crain, Stephen., Takuya Gloro, and Rosalind Thornton. 2006. “Language acquisition is language change.” Journal of Psycholinguistic Research 35: 31–49. Crain, Stephen, and Diane Lillo-Martin. 1999. An Introduction to Linguistic Theory and Language Acquisition. Malden, MA: Basil Blackwell. Crain, Stephen, and Cecile McKee. 1985. “Acquisition of structural restrictions on anaphors.” In Proceedings of the Sixteenth Annual Meeting of the North Eastern Linguistics Society, ed. S. Berman, J. Chloe, and J. McDonough, 94–110. Montreal: McGill University. Crain, Stephen, and Mineharu Nakayama. 1987. “Structure dependence in grammar formation.” Language 63: 522–43. Crain, Stephen, and Paul Pietroski. 2002. Why language acquisition is a snap. Linguistic Review 19: 163–84. Crain, Stephen, and Kenneth Wexler. 1999. “Methodology in the study of language acquisition: A modular approach.” In Handbook of First Language Acquisition, ed. William C. Ritchie and Tej K. Bhatia, 387– 426. San Diego, CA: Academic Press. deGraff, Michele. 1999. “Creolization, language change and language acquisition: A prolegomenon.” In Language Creation and Language Change, ed. M. deGraff, 1–46. Cambidge, MA: MIT Press. Demuth, Katherine. 1994. “On the underspecification of functional categories in early grammars.” In Syntactic Theory and First Language Acquisition: Cross-Linguistic Perspectives. Vol. 1. Ed. B. Lust, M. Suner, and J. Whitman, 119–34. Hillsdale, NJ: Lawrence Erlbaum. deVilliers, Jill, Tom Roeper, and Ann Vainikka. 1990. “The acquisition of long-distance rules. In Language Processing and Language Acquisition, ed. Lynn Frazier and Jill deVilliers, 257–98. Dordrecht, the Netherlands: Kluwer. Edelman, Shimon. 2008. Computing the Mind: How the Mind Really Works. Oxford: Oxford University Press. Eisele, Julie, and Barbara Lust. 1996. “Knowledge about pronouns: A developmental study using a truth-value judgment task.” Child Development 67: 3086–100. Elman, Jeffrey L., Elizabeth A. Bates, Mark H. Johnson, Annette Karmiloff-Smith, Domenico Parisi, and Kim Plunkett. 1966. Rethinking Innateness. Cambridge, MA: MIT Press. Flynn, Suzanne, and Gita Martohardjono. 1994. “ Mapping from the initial state to the final stage: The separation of universal principles and language specific principles.” In Syntactic Theory and First Language Acquisition: Cross Linguistic Perspectives, ed. Barbara Lust, Magui Suner, and John Whitman, 319–36. Hillsdale, NJ: Lawrence Erlbaum. Foley, Claire, Zelmira Nuñez del Prado, Isabella Barbier, and Barbara Lust. 2003. “Knowledge of variable binding in VP-ellipsis: Language acquisition research and theory converge.” Syntax 6.1: 52–83. Friederici, A. D. 2000. “The developmental cognitive neuroscience of language: A new research domain.” Brain and Language 71: 65–8. ———. 2002. “Towards a neural basis of auditory sentence processing.” Trends in Cognitive Science 6.2: 78–84. Gair, James, Barbara Lust, Lewala Sumangala, and Milan Rodrigo. 1998. “Acquisition of null subjects and control in some Sinhala adverbial clauses.” In Studies in South Asian Linguistics: Sinhala and Other South Asian Languages, 271–85. Oxford: Oxford University Press. Gentner, Dedre, and Susan Goldin-Meadow, eds. 2003. Language in Mind. Cambridge, MA: MIT Press. Gerken, L., B. Landau, and R. Remez. 1990. “Function morphemes in young children’s speech perception and production.” Developmental Psychology 26: 204–16. Gerken, Louann, and Bonnie McIntosh. 1993. “Interplay of function morphemes and prosody in early language.” Developmental Psychology 29.3: 448–57.

Gleitman, Lila. 1990. “The structural sources of verb meanings.” Language Acquisition 1.1: 3–55. Goldin-Meadow, Susan. 2003. The Resilience of Language: What Gesture Creation in Deaf Children Can Tell Us About How All Children Learn Language. New York: Psychology Press. ———. 2005. “What language creation in the manual modality tells us about the foundations of language.” Linguistic Review 22: 199–226. Greally, John M. 2007. “Encyclopaedia of humble DNA.” Nature 447: 782–3. Grimes, Barbara, ed. 2000. Ethonologue: Languages of the World. Dallas: SIL International. Guasti, Maria Teresa. 2002. Language Acquisition: The Growth of Grammar. Cambridge, MA: MIT Press. Guo, F. F., C. Foley, Y.-C. Chien, B. Lust, and C.-P. Chiang. 1996. “Operator–variable binding in the initial state: A cross-linguistic study of VP ellipsis structures in Chinese and English.” In Cahiers de Linguistique Asie Orientale 25.1: 3–34. Hauser, Marc, Noam Chomsky, and Tecumseh Fitch. 2002. “The faculty of language: What is it, who has it, and how did it evolve.” Science 298: 1569–79. Hirsh-Pasek, Kathy, Diane Kemler-Nelson, Peter Jusczyk, K. WrightCassidy, B. Druss, and L. Kennedy. 1987. “Clauses are perceptual units for young children”. Cognition 26: 269–86. Hirsh-Pasek, Kathy, and Roberta Michnick Golinkoff. 1996. The Origins of Grammar. Cambridge, MA: MIT Press. Hyams, Nina. 1986. Language Acquisition and the Theory of Parameters. Dorcrecht and Boston: Reidel. Johnson, Elizabeth, and Peter Jusczyk. 2001. “Word segmentation by 8-month-olds: When speech cues count more than statistics.” Journal of Memory and Language 44: 548–67. Jusczyk, Peter. 1997. The Discovery of Spoken Language. Cambridge, MA: MIT Press. Jusczyk, W., A. D. Friederici, J. M. I. Wessels, V. Y. Svenkerud, and A. M. Jusczyk. 1993. “Infants’ sensitivity to the sound patterns of native language words.” Journal of Memory and Language 32: 402–20. Jusczyk, Peter W, and Elizabeth A. Hohne. 1997. “Infants’ memory for spoken words.” Science 277: 1984–6. Kedar, Yarden, Marianella Casasola, and Barbara Lust. 2006. “Getting there faster: 18- and 24-month-old infants’ use of function words to determine reference.” Child Development 77.2: 325–38. Kegl, Judith, Ann Senghas, and M. Coppola. 1999. “Creation through contact: Sign language emergence and sign language change in Nicaragua. In Language Creation and Language Change: Creolization, Diachrony and Development, ed. M. DeGraff, 179–237. Cambridge, MA: MIT Press. Kuhl, Patricia, Barbara Conboy, Denise Padden, Tobey Nelson, and Jessica Pruitt. 2005. “Early speech perception and later language development: Implications for the critical period.” Language Learning and Development 1.3/4: 237–64. Lightfoot, David. 1989. “The child’s trigger experience: Degree 0 learnability.” Behavioral and Brain Sciences 12: 321–34. Lust, Barbara. 1999. “Universal grammar: The strong continuity hypothesis in first language acquisition.” In Handbook of First Language Acquisition, ed. William C. Ritchie and Tej K. Bhatia, 111–55. San Diego, CA: Academic Press. ———. 2006. Child Language: Acquisition and Growth. Cambridge: Cambridge University Press. ———. Universal Grammar and the Initial State: Cross-Linguistic Studies of Directionality. In preparation. Lust, Barbara, and Yu-chin Chien. 1984. The structure of coordination in first language acquisition of Mandarin Chinese. Cognition 17: 49–83. Lust, Barbara, and Teresa Clifford. 1986. “The 3-D study: Effects of depth, distance and directionality on children’s acquisition of anaphora: Comparison of prepositional phrase and subordinate clause

63

The Cambridge Encyclopedia of the Language Sciences embedding.” In Studies of the Acquisition of Anaphora: Defining the Constraints, ed. B. Lust, 203–44. Dordrecht, the Netherlands: Reidel Press. Lust, Barbara, Suzanne Flynn, and Claire Foley. 1996. “What children know about what they say: Elicited imitation as a research method.” In Methods for Assessing Children’s Syntax, ed. Dana McDaniel, Cecile McKee, and Helen Cairns. Cambridge, MA: MIT Press. Lust, Barbara, Suzanne Flynn, Claire Foley, and Yu-Chin Chien. 1999. “How do we know what children know? Problems and advances in establishing scientific methods for the study of language acquisition and linguistic theory.” In Handbook of First Language Acquisition, ed. William C. Ritchie and Tej K. Bhatia, 427–56. San Diego, CA: Academic Press. Marshall, John C. 1980. “On the biology of language acquisition.” In Biological Studies of Mental Processes, ed. D. Caplan, 301–20. Cambridge, MA: MIT Press. Mazuka, Reiko. 1996.” Can a grammatical parameter be set before the first word? Prosodic contributions to early setting of a grammatical parameter.” In Signal to Syntax: Bootstrapping from Speech to Grammar in Early Acquisition, ed. James Morgan and Katherine Demuth, 313–30. Hillsdale, NJ: Lawrence Erlbaum. Mehler, Jacques, Emmanuel Dupoux, Thierry Nazzi, and G. DehaeneLambertz. 1996. “Coping with linguistic diversity: The infant’s viewpoint.” In Signal to Syntax: Bootstrapping from Speech to Grammar in Early Acquisition, ed. James L. Morgan and Katherine Demuth, 101–16. Mahwah, NJ: Lawrence Erlbaum. Mehler, Jaques, Peter Jusczyk, Ghislaine Lambertz, Nilofar Halsted, Josiane Bertoncini, and Claudine Amiel-Tison. 1988. “A precursor of language acquisition in young infants.” Cognition 29: 144–78. Morgan, James L., and Katherine Demuth, eds. 1996. Signal to Syntax: Bootstrapping from Speech to Grammar in Early Acquisition. Mahwah, NJ: Lawrence Erlbaum. Newman, Rochelle, Nan Bernstein Ratner, Ann Marie Jusczyk, Peter W. Jusczyk, and Kathy Ayala Dow. 2006. “Infants’ early ability to segment the conversational speech signal predicts later language development: A retrospective analysis.” Developmental Psychology 42.4: 643–55. Nuñez del Prado, Zelmira, Claire Foley, and Barbara Lust. 1993. “The significance of CP to the pro-drop parameter: An experimental study of Spanish-English comparison.” In Proceedings of the Twenty-Fifth Child Language Research Forum, ed. Eve Clark, 146–57. Stanford, CA: CSLI. Oshima, Shin, and Barbara Lust. 1997. “Remarks on anaphora in Japanese adverbial clauses.” In Papers on Language Acquisition: Cornell University Working Papers in Linguistics, ed. Shamita Somashekar, Kyoko Yamakoshi, Maria Blume, and Claire Foley. Ithaca, NY: Cornell University Press. Paradis, Michel. 2000. “The neurolinguistics of bilingualism in the next decades.” Brain and Language 71: 178–80. Piatelli-Palmarini, Massimo, ed. 1980. Language and Learning: The Debate Between Jean Piaget and Noam Chomsky. Cambridge: Harvard University Press. Pinker, Steven. 1994. The Language Instinct. New York: W. W. Morrow. Radford, Andrew. 1990. Syntactic Theory and the Acquisition of English Syntax. Cambridge: Cambridge University Press. Ramus, Frank, Marc D. Hauser, Cory Miller, Dylan Morris, and Jaques Mehler. 2000. “Language discrimination by human newborns and by cotton-top tamarin monkeys.” Science 288.5464: 349–51.

64

Roeper, Tom. 2007. The Prism of Grammar: How Child Language Illuminates Humanism. Cambridge, MA: MIT Press. Saffran, Jenny R., Richard N. Aslin, and Elissa L. Newport. 1996. “Statistical learning by 8-month-old infants.” Science 274: 1926–8. Santelmann, Lynn, Stephanie Berk, Shamita Somashekar, Jennifer Austin, and Barbara Lust. 2002. “Continuity and development in the acquisition of inversion in yes/no questions: Dissociating movement and inflection.” Journal of Child Language 29: 813–42. Senghas, A. 1995. Children’s Contribution to the Birth of Nicaraguan Sign Language. Cambridge, MA: MIT Press. Seuss, Dr. [1957] 1985. Cat in the Hat. New York: Random House. ———. [1963] 1991. Hop on Pop. New York: Random House. Shady, M. E. 1996. “Infants’ sensitivity to function morphemes.” Ph.D. diss., State University of New York at Buffalo. Snyder, William. 2007. Child Language. Oxford: Oxford University Press. Swinney, David. 2000. “Understanding the behavioral-methodology/ language-processing interface.” Brain and Language. 71: 241–4. Tomasello, Michael. 1995. “Language is not an instinct.” Cognitive Development 10: 131–56. ———. 2002. The New Psychology of Language. Cognitive and Functional Approaches to Language Structure. Mahwah, NJ: Lawrence Erlbaum. ———. 2003. Constructing a Language: A Usage-Based Theory of Language Acquisition. Cambridge: Harvard University Press. ———. 2005. “Beyond formalities: The case of language acquisition.” Linguistic Review 22: 183–98. Tomasello, Michael, Ann Cale Kruger, and Hilary Horn Ratner. 1993. “Cultural learning.” Behavioral and Brain Sciences 16: 495–552. Valian, Virginia. 1991. “Syntactic subjects in the early speech of American and Italian Children.” Cognition 40: 21–49. ———. 1999. “Input and language acquisition.” In Handbook of Child Language Acquisition, ed. William Ritchie and Tej Bhatia, 497–530. San Diego, CA: Academic Press. Van Valin, Robert D. 1991. “Functionalist theory and language acquisition.” First Language 11: 7–40. Werker, Janet F. 1994. “Cross-language speech perception: Developmental change does not involve loss.” In The Development of Speech Perception: The Transition from Speech Sounds to Spoken Words, ed. J. Goodman and H. Nusbaum, 93–120. Cambridge, MA: MIT Press. Werker, Janet F., and Richard C. Tees. 1984. “Cross-language speech perception: Evidence for perceptual reorganization during the first year of life.” Infants’ Behavior and Development 7: 49–63. Wexler, Kenneth. 1999. “Maturation and growth of grammar.” In Handbook of First Language Acquisition, ed. T. Bhatia and Wm. Ritchie, 55–110. San Diego, CA: Academic Press. Yang, Charles D. 2006. The Infinite Gift. New York: Scribner. Yang, S., and B. Lust. 2005. “Testing effects of bilingualism on executive attention: Comparison of cognitive performance on two nonverbal tests.” BUCLD 29 Proceedings Online Supplement. Somerville, MA: Cascadilla. Available online at http://www.bu.edu/linguistics/ APPLIED/BUCLD/supp29.html. ———. 2007. “Cross-linguistic differences in cognitive effects due to bilingualism: Experimental study of lexicon and executive attention in 2 typologically distinct language groups.” BUCLD 31 Proceedings. Somerville, MA: Cascadilla.

7 ELABORATING SPEECH AND WRITING: VERBAL ART Patrick Colm Hogan

The past half century has seen considerable interaction between the sciences of language and the study of literature. But this “interaction” has been largely unidirectional, with influence flowing from language science to literature. This may be seen most clearly in the massive impact of Ferdinand de Saussure on literary study since the 1960s. However, generative grammar, cognitive linguistics, connectionism, and other approaches have also had effects on poetics and literary theory. In the following pages, I wish to consider the general issues of what distinguishes verbal art as an object of study in language science. However, before I turn to this, it is important to get a sense of how the analysis of literature and the analysis of language have been interconnected since the current phase of language study began 50 years ago.

THEORIES OF LANGUAGE AND THEORIES OF LITERATURE When the influence of language science on literary theory is considered, it is helpful to begin with a division between literary theorists who have drawn on broad methodological principles and literary theorists who have taken up particular linguistic theories. For example, my own work on literary universals (Hogan 2003) does not rely on any particular account of language universals. However, I do follow the general methodological principles for isolating genetic and areal distinctness (though see areal distinctness and literature), distinguishing different varieties of statistically unexpected cross-cultural patterns, and so on. This type of influence, however, is the exception rather than the rule. Other writers have drawn on particular theories of language, using them literally or analogically to explore literature. Taking up the structure set out in the Preface, we may distinguish neurobiological, mentalistic, and social theories, as well as theories that bear on acquisition and evolution. Mentalistic theories have been the most prominent. As noted in the Preface, within mentalistic theories, we may distinguish intentionalist accounts of language (referring to human subjectivity and intention) from representationalist accounts (treating algorithmic operations on mental symbols). Intentionalist

theories of language have been developed most prominently within ordinary language philosophy. The ideas of Ludwig Wittgenstein and the principles of speech-act theory have been taken up by literary theorists, such as Mary Louise Pratt and Stanley Fish. Some philosophers working in other areas of the analytic philosophy of language, such as Donald Davidson, have also had considerable influence (see Dasenbrock 1993). While intentionalist theories of language have certainly been influential in literary study, their use has been limited and, so to speak, instrumental. They tend to be taken up for particular interpretive aims. This is often the case with literary borrowings from linguistics and the philosophy of language. However, the uses of representationalist theories have been different. In these cases, the literary and linguistic theories have been much more thoroughly integrated. This is the result of two factors. First, the research areas of the linguistic and literary theories overlap. Second, there has been collaboration between linguists and literary theorists in treating these areas. More exactly, there are two important representationalist schools that have had significant impact in literary study. One is cognitive linguistics. Some of the most important work in cognitive linguistics has treated metaphor. While cognitive linguists were initially interested in ordinary uses of metaphor, they quickly extended their analyses to poetic metaphor. This was facilitated by the collaboration of a linguist, George Lakoff, and a literary theorist, Mark Turner. A similar point may be made about Chomskyan or generative representationalism. One part of Chomskyan theory treats phonology. Patterns in sound and stress are of obvious importance in verse. Thus, certain aspects of poetic form may be included within the framework of generative linguistic theories. Work in this area has been facilitated by collaborations between linguists and literary critics as well (see Fabb and Halle 2006). (Theorists have used Chomskyan generative principles as a model for other aspects of literary theory also; see generative poetics.) Brain-related theorization on language and literature is less developed, in part because it is highly technical and in part because neuroscientific theorization about language itself is much more recent. Turning again to the divisions in the Preface, we may distinguish between connectionist approaches and neurobiological approaches proper. There has been some work on verbal art and parallel distributed processing. For example, some writers have used connectionist models to discuss creativity (see Martindale 1995) and there has been some work on connectionism and metaphor (e.g., Chandler 1991). Though limited, the work done in this field is generally well integrated into connectionist theories (i.e., it is not solely instrumental). Recently, there has been considerable interest in neurobiology and art. This work has addressed many nonlinguistic aspects of brain function. However, some has focused specifically on language. Much of this addresses hemispheric specialization, exploring distinctive features of verbal art (see Kane 2004; poetic language, neurobiology of). Given the general orientation of literary critics, it is unsurprising that social aspects of speech and language have figured more prominently in literary study. At the level of dialogue, literary theorists have drawn on, for example, Paul Grice’s account

65

The Cambridge Encyclopedia of the Language Sciences of conversational implicature, as well as various ideas of Mikhail Bakhtin (see dialogism and heteroglossia). In terms of larger groups, literary theory has been highly influenced by historian Michel Foucault’s ideas about discourse and social power (see discourse analysis [foucaultian]). They have also drawn on the sociological ideas of Pierre Bourdieu (see field; market, linguistic; and habitus, linguistic) and others. Broader social movements, such as Marxism and feminism (see marxism and language, gender and language, and sexuality and language) and their associated theories, have also contributed importantly to literary discussion, though not necessarily in a way that bears particularly on language science. The literary use of most social theories has tended to be instrumental. However, more narrowly sociolinguistic analyses of literature have been integrated into research programs in sociolinguistics. This is largely because, here too, the areas of research overlap, and language scientists have been involved in research along with literary interpreters. For example, William Labov’s studies of oral narrative and the researchers of writers in corpus linguistics have contributed to the advancement of both sociolinguistics and literary study. The same general point holds for the study of acquisition. There has been valuable work done on, for example, the acquisition of metaphor and the development of verbal humor. Finally, evolutionary theory has inspired many literary theorists in recent years (see verbal art, evolution and). Its advocates propose sweeping evolutionary explanations for a wide range of literary phenomena. It is not clear that this program has gone beyond the stage of conjecture. In any case, it is general and often not tied specifically to language. As I have suggested, much work in the preceding areas is very significant. However, much of it begins, like the classical European epic, in medias res. It does not set out a clear field of study for a language science of verbal art. Rather, the most successful work tends to focus on those areas of verbal art that fall within the purview of nonliterary research programs. Put differently, it tends to treat verbal art as a case of something else (e.g., cognitive metaphor or phonology). In the remainder of this essay, then, I do not explore these particular approaches and connections in more detail. Such a review is, in any case, redundant, as this material is covered in the following entries. Instead, I consider what is distinctive about verbal art and why, as a result, it is an important area of study for language science.

THE PARTIAL AUTONOMY OF VERBAL ART: INDIRECT ADDRESS, SIDE PARTICIPATION, AND PLAY Perhaps the most obvious differentiating characteristics of verbal art are that it is normative and rare. While all cultures have verbal art (see Kiparsky 1987, 195–6), few people in any culture produce works of verbal art (though, of course, they do produce many constituents of such works – novel metaphors, allusions, wit, and so forth). On the other hand, these differences are not definitive in themselves. Rather, they seem to result from other factors. Consider two samples of speech actions: 1) the following excerpt from recorded speech – “Go away, I’m cooking.… Excuse

66

me please, I’m trying to cook. I haven’t got enough potatoes” (Biber, Conrad, and Reppen 2006, 69); and 2) the following excerpt from Shakespeare’s Sonnet 97 – “How like a winter hath my absence been/From thee, the pleasure of the fleeting year!” The first piece of speech calls to mind a particular, active context. The second is removed from any such context. Perhaps, then, works of verbal art are more separable from their immediate context. As a first approximation, we might say that verbal art has other verbal art as its primary, distinctive context, and the current material context has largely inhibitory force. In other words, our understanding and evaluation of a work of verbal art are distinguished from other sorts of understanding and evaluation, first of all, by their relation to other works of verbal art – our sense of storytelling techniques, our awareness of the larger story cycle in which a particular narrative occurs, our expectations about characters, and so forth. This does not mean that verbal art is entirely insensitive to immediate context. Our response to stories may be inflected, primarily in an inhibitory way, by current material circumstances. Consider jokes, a form of verbal art. One standard type of joke is ethnic. The standard formats of such jokes (e.g., “How many x’s does it take to screw in a light bulb?”), the general function of jokes, and so forth provide the broad, distinctive context for interpretation and response. The most obvious function of the immediate, material context comes when a member of the relevant ethnic community is present – and that function is usually inhibitory. Removal from immediate context cannot be quite the crucial property, however. Consider, for example, the present essay. Physically, I am alone as I am writing. Anyone who reads this essay will be removed from the material context in which I am writing, and that material context will be irrelevant to the reader’s response and understanding. But that does not make this essay verbal art. Perhaps, then, the most important difference between the aforementioned speech actions is not context per se but something closely related to context. The joke suggests that this has something to do with the audience. Perhaps the difference is a matter of the way the speaker addresses his or her audience. To consider this, we might return to those cases. It is clear that the person who says, “Go away, I’m cooking,” is talking to his or her audience. I, too, am addressing my readership in writing this essay – even if my idea of that readership is rather amorphous. But the sonnet, as a socially circulated poem, is not addressing its readership. Even if Shakespeare initially drafted the poem as a private message to a particular person, he made the decision that its readership would not be its addressee when he made it a public poem. More exactly, works of verbal art tend to be marked by indirect address, rather than direct address. When considered from the perspective of the reader rather than the author, indirect address is roughly the same as “side participation,” as discussed by Richard Gerrig and Deborah Prentice (1996). Gerrig and Prentice distinguish several possible roles in a conversation. Obvious roles include speaker, addressee, and overhearer. The authors add a fourth role – side participant. Suppose I am with my wife at the grocery store. She sees von Humboldt, a colleague of hers. The colleague says, “Oh, that meeting with de Saussure, our new provost, is on the twelfth.” When she says “the new provost” she is doing so for my benefit. My wife knows perfectly well

Elaborating Speech and Writing who de Saussure is. The conversation does not really concern me. If I had been standing a few feet away, I would have merely been an overhearer and von Humboldt would have said only, “Oh, that meeting with de Saussure is on the twelfth.” But since I was closer, I became a side participant and von Humboldt had to take my knowledge and interest into account when speaking. The difficulty with the account of Gerrig and Prentice is that it is, for the most part, analogical. It suggests an area of research and theorization. However, it does not develop this in algorithmic specificity (i.e., spelling out how it will work, step by step) in relation to the structures, processes, and contents of human cognition. To explore the idea further, we might consider a simple model of speech (oral or written). This model begins from the premise that language is a form of action. Thus, it has the usual components of action – goals, motivations or emotions, anticipated outcomes, and so forth. Here, I wish to isolate two stages in the production of speech. One is the basic generative phase in which the initial utterance is formulated. The second is an adjustment phase, which follows our realization of just what we are saying. Intuitively, we might expect that awareness of speech would precede generation. But it does not. Indeed, a moment’s reflection suggests why. In order to realize that I am about to say “Hello,” I must in some sense have already generated the “Hello,” even if my lips have not yet moved. More importantly, empirical research indicates that human action generally involves just this sort of duality. For example, as Henrik Walter points out, our brains initiate or project actions approximately .5 to .7 seconds before the actions are performed. We are able to modify or inhibit the action .3 to .5 seconds after it is projected (see Walter 2001, 248–50). In keeping with this temporal sequence, adjustment may precede, interrupt, or follow the execution of an action. Suppose someone asks me if I need a ride. I begin to say, “No, thanks. Duns is picking me up.” I realize that the person does not know who Duns is. I may adjust the sentence before speaking – “My cousin Duns is picking me up.” Or I may succeed in making the change only after beginning the sentence – “Duns – uh, he’s my cousin – he’s picking me up.” When the adjustment takes place before speaking, we might refer to it as “implicit.” When it occurs in the course of speaking or after speaking, we may refer to it as “explicit.” Finally, it is important to note that actions have two broad sources. One is extrinsic, or externally derived; the other is intrinsic, or internally derived (see MacNeilage 1998, 225–6 on the neural substrates for this division). Now we are in a position to clarify the nature of indirect address – or, more generally, indirect action. Indirect action, as I am using the phrase, involves a relative decrease in the extrinsic aspects of action. Thus, the sources of both generation and adjustment are more intrinsic than is commonly the case. Moreover, when they occur, extrinsic adjustments tend to be implicit rather than explicit. To get a better sense of how indirect action operates, it is useful to look at a paradigmatic case of such action – play. Indeed, play and side participant interaction have a great deal in common. When Jane and Jill play school, each of them must keep in mind that the other person has a real identity outside the role she is playing. Moreover, each of them must continually adjust her speech and behavior to take that into account. For example,

suppose Jane is in third grade and Jill is in second grade. Jane begins to play a third grade teacher. She starts by referring to what the class did last week. However, she has to keep in mind that Jill does not know what the third grade class did last week. Thus, she may have to say “Now, class, you remember that last week we began by discussing government and binding theory.” As this suggests, play is a purer case of indirection than ordinary side participation. In the case of side participation, our adjustments are more likely to be explicit. For example, my wife’s colleague is likely to turn to me when explaining that de Saussure is the new provost. In play, explicit adjustments occur when the pretense of play is disrupted. Suppose Jill accidentally addresses Jane as “Jane,” then explicitly adjusts that to “I mean, Frau Doktor Wittgenstein.” Jane is likely to get annoyed with this breach in the play, perhaps responding, “Jill, you should never call your teacher by her first name. Now you have to sit in the corner and wear the Aversive Stimulus Operant Conditioning Cap.” Following this usage, we may say that verbal art is a form of indirect speech action in the same sense as one finds in play. Indeed, verbal art is, in certain respects, a form of play. Note, however, that in verbal art, in play, and elsewhere, there is not an absolute separation between direct and indirect action. Again, indirect action reduces extrinsic factors. It does not eliminate them. The difference is one of degree. Indeed, the difference between extrinsic and intrinsic is not fine grained enough to clarify the distinctiveness of verbal art. This becomes clear as soon as we notice that almost all writing is predominantly intrinsic in that it is not generated or adjusted primarily by reference to external circumstances. Moreover, the opportunities for extensive revision in writing allow for virtually all adjustments to be implicit in the final text. How, then, do we distinguish ordinary writing from writing that is verbal art? Here, too, the distinction is a matter of degree. Writing involves different gradations of removal from direct address. Take, for example, a letter. Suppose von Humboldt is not speaking to my wife about a faculty meeting, but is instead writing her a note. In an obvious way, von Humboldt’s address to my wife is indirect and intrinsic. After all, my wife is not present. However, von Humboldt’s address to my wife is direct in another way, for it is oriented precisely toward her. It remains guided by her as von Humboldt imagines her. In order to understand how this works, and how it bears on verbal art, we need to have a clearer understanding of imagination and action. When I begin to act, I tacitly project – or imagine – possible outcomes for my action. For example, when I see a car coming toward me, I run to the curb. This is bound up with a tacit imagination of where I should be in order to fulfill my purpose of not being run over. More exactly, we may understand imagination, like speech, as involving two (continuously interacting) processes. One generates possible scenarios. The other makes adjustments. I project running to the curb, but then notice a manhole in my path. This leads me to adjust my imagination of the precise trajectory. The nature of the generation and the nature of the adjustment will change, depending on the guiding purposes of the action. In some cases of speech action, the purpose involves a real addressee. In some cases, it does not. In a face-to-face dialogue, a speaker will continually generate, adjust, and regenerate what he or she is saying in part due to the addressee’s actual

67

The Cambridge Encyclopedia of the Language Sciences response. In writing a letter, a writer will be guided by his or her tacit imagination of the addressee. Though that imagination is not adjusted by reference to actual responses, it nonetheless serves as a guide for the generation of the speech. Crucially, this imagined addressee maintains his or her external, independent existence for the speaker. As such, that addressee is, so to speak, a pseudoextrinsic guide to the generation of the speech. Here, then, we may adjust our account of indirect address in verbal art. Verbal art minimizes both extrinsic and pseudoextrinsic elements in the production and adjustment of speech and in the imaginations that are connected with speech. Moreover, it minimizes explicit markers of adjustments for “side participants.” We began with the idea that verbal art is relatively independent of direct, material context. That context involves authors and readers. In this connection, we have been considering the relation between the author and readers as one of indirect address. The other crucial element of material context is, of course, reference. Indeed, verbal art, as commonly understood, has an even more obvious relation to reference than to address, for verbal art is paradigmatically conceived of as fiction. As such, it has an unusual degree of referential autonomy. In other words, it tends to be characterized by indirect reference as well as indirect address. This, too, is illustrated most clearly in play. Suppose Jane and Jill are playing telephone. Jane picks up a banana, puts it up to the side of her face, then holds it out to Jill and says, “It’s for you.” In saying “it,” she is not referring to a banana. She is referring to a telephone. In sum, verbal art is not just partially autonomous with respect to immediate circumstances. It is largely independent of the extrinsic and pseudoextrinsic aspects of the generation and adjustment of speech action and associated imagination. This is true with respect to both address and reference.

ART AND ACTION: THE PURPOSES OF VERBAL ART In discussing action-related imagination, we noted that the precise nature of imagination varies according to the purposes of the action. Generally speaking, the goals of physical action concern the alteration of some situation. To a great extent, speech actions have the same function. One crucial difference between speech actions and bodily actions is that speech actions have their effects only through minds. However much I plead with the packet of noodles, they won’t turn themselves into pad thai. But, if I sweetly ask someone to make the noodles into pad thai, perhaps that person will do so. Speech actions, then, tend to aim at altering the world by altering people’s actions. In order to alter people’s actions, they aim at altering two things – first, the way those people understand the world and, second, the way they feel about it, including the way they feel about the speaker himself or herself. More exactly, we may distinguish two psychological purposes of speech actions. These are informational and emotional. Informational purposes may be further divided into pragmatic and regulative. Pragmatic information concerns factual, directional, or other information that facilitates our pursuit of goals. Regulative information concerns broad ethical, prudential, or related principles, which tend to serve an adjusting function. For example, feeling hungry, I form the goal of eating something.

68

This combines with pragmatic information about the presence of cookies in the cookie jar to initiate the action of getting a cookie out of the cookie jar. However, before I reach for the cookie jar, prudential information stops me as I recall the deleterious effects that cookies are likely to have on my lithe and shapely form. For imaginations or actions to result from pragmatic information, there must be some motivation present as well. In other words, I must have some emotion. Emotion is what leads us to act, whether the action is speech or larger bodily movement. It is also what leads us to refrain from acting. We may divide emotions into two sorts, depending on their function in action sequences. The first might be called the initiating emotion. An initiating emotion is any emotion that one feels in current circumstances and that impels one to act. Action here includes the initiation of movement and the initiation of imagination. We imagine various outcomes of our actions. For example, feeling hungry, I swiftly and unreflectively imagine eating a cookie. But my imagination does not stop there. I may imagine my wife seeing me with the crescent of cookie in my hand and the telltale crumbs sticking to my lips, then chastising me, reminding me of the doctor’s warnings, and explaining once again that the cookies are for my nieces and nephew. Or I may suddenly see an image of myself with wobbly love-handles. In each case, I experience what might be called a hypothetical emotion. A hypothetical emotion is a feeling that I experience in the course of imagining the possible trajectories of my action. While initiating emotions give rise to (or generate) the action sequence initially, hypothetical emotions qualify (or adjust) that action sequence. Hypothetical emotions may intensify the initiating emotion; they may inhibit it; they may respecify the precise goals of the action sequence (e.g., from eating cookies to eating carrots), or they may affect the precise means that I adopt (e.g., checking that my wife is not around). From the preceding example, it may seem that hypothetical emotions are all egocentric. However, hypothetical emotions may also be empathic. For example, I may forego my plan to eat cookies because I imagine my tiny nephew’s disappointed face when he reaches into the empty jar. Empathic hypothetical emotions may have the same qualifying effects on initiating emotions and actions as do egocentric hypothetical emotions. Hypothetical emotions are critical for all types of action, including verbal action. Consider again the case where von Humboldt explains that de Saussure is the new provost. In doing this, von Humboldt tacitly imagines the conversation from my perspective and, sensing that I may not follow – and, more importantly, that I might feel left out – she provides the information. If I wish to alter someone’s emotions through speech, I will appeal primarily to initiating emotions. Thus, the alteration of initiating emotions is usually a central purpose of speech action. However, certain sorts of hypothetical emotions are important as well. For example, when requesting a favor, I may foreground how grateful I will be. One purpose is to encourage my addressee to imagine my gratitude and experience the related hypothetical emotion (roughly, feeling appreciated). In sum, ordinary speech actions involve an informational aim and an emotional aim, usually coordinated to produce some change in the world, currently or at some time in the future. The informational aim involves both pragmatic and regulatory components, though in ordinary speech, the pragmatic component

Elaborating Speech and Writing is probably dominant. The emotional aim involves both initiating and hypothetical emotions. In ordinary speech, the initiating emotion is probably dominant. In being removed from direct address and direct reference, works of verbal art commonly do not have a goal of directly altering particular material conditions. Nonetheless, verbal art is animated by the same two psychological goals that animate other forms of speech action. Verbal art, too, has an informational component and an emotional component. Indeed, cross-culturally, aestheticians and critics tend to see verbal art as successful insofar as it affects us emotionally and insofar as it develops significant ideas or themes. The emphasis in verbal art, however, tends to be different from that of other speech actions in both cases. Specifically, verbal art commonly does not stress pragmatic aspects of information. Certainly, there are, for example, political works that set out to give the audience pragmatic information. However, this is not the general tendency of verbal art. Rather, verbal art tends to develop regulative concerns. These regulative concerns make their appearance in literary themes. When we interpret a work and seek to understand its “point,” we are commonly looking for a theme, which is to say, some sort of regulative information. Moreover, in terms of emotion, verbal art tends to inspire not initiating emotions but hypothetical emotions – particularly empathic hypothetical emotions.

THE MAXIMIZATION OF RELEVANCE Up to this point, I have been speaking of the content of speech action generally and verbal art in particular. But the form of verbal art is widely seen as crucial, perhaps its definitive feature. The content/form division is somewhat too crude to form the basis for a sustained analysis of verbal art. However, it does point to an important aspect of verbal art, and an important differentiating tendency – the partial autonomy and patterning of distinct linguistic levels in verbal art. All linguistic levels – phonology, morphology, syntax, and so forth – are, of course, relevant to speech actions of all types. However, most components are, in most cases, only instrumentally relevant. Morphology is relevant for communicating whether I want one cookie or two cookies. But it has no separate function. Put differently, the specification of phonology, morphology, and syntax is a sort of by-product of my pragmatic aims. I want to get two cookies. In English, I signal that I want two cookies rather than one by using the word “two,” putting that word before “cookie,” adding “s” to “cookie,” and so forth. I do not set out to do anything with phonology, morphology, or syntax. One might even argue that I do not set out to do anything special with semantics or with discourse principles. I just run my language processors to achieve the active goal. Features of phonology and so forth are relevant to my action only insofar as my speech processor makes them relevant. They have, in this way, “minimal” relevance. A number of literary theorists, prominently the Russian Formalists, have stressed that verbal art “foregrounds” its language (see foregrounding). In cognitive terms, we might say that verbal art tends to enhance linguistic patterns to the point where they are likely to be encoded by readers (i.e., perceived

and integrated with other information; see encoding) and may even draw attentional focus. This occurs most obviously in the phonetic/phonological aspect of language. Poets pattern stress beyond what occurs spontaneously. They organize vowel sounds and initial consonants to produce assonance and alliteration. The point is less clear in morphology, though corpus linguistic studies have pointed to differential tendencies (see Biber, Conrad, and Reppen 2006, 58–65) Syntax is a complex and thus particularly interesting case. Verbal art does undoubtedly tend to pattern syntactic usage in encodable ways. But the poetic patterns in these cases tend to be intensifications of the ways in which ordinary speech actions pattern syntax. For example, in ordinary speech, we have some tendency to use parallel syntactic structures. In poetry, we are more likely to use these. A more obtrusive foregrounding of syntax occurs when we violate expectations of patterning in verbal art. Such violation is the other primary way in which verbal art treats linguistic levels more autonomously. In the case of syntax, there are several ways in which this may occur. One obvious way is through disjoining syntactic units and line units, such that line breaks do not coincide with syntactic breaks (see Tsur 2006, 146–52). Another is by rejecting standard principles of word order. Verbal art also manifests distinctive tendencies in semantics. These are found most obviously in lexical preferences and metaphor. In the case of lexical preferences, verbal art may draw on rarer or more unexpected words. Perhaps more importantly, it may pattern the suggestions and associations of terms. In ordinary speech, we tend to concern ourselves with the associative resonance of a term only in extreme cases (e.g., when its connotations may be offensive to a particular addressee). In verbal art, however, the writer is much more likely to organize his or her lexical choices so that the suggestions of the terms are consistent (e.g., in emotional valence; see dhvani and rasa). As to metaphor, our use of tropes in ordinary speech is surprisingly constrained. There are aspects of everyday metaphor that are creative. However, on the whole, we follow well-worn paths. Lakoff and Turner have argued that a great deal of literary metaphor draws on the same broad structures as ordinary speech. However, literary metaphors extend, elaborate, and combine these structures in surprising ways (see poetic metaphor). Finally, we find parallel tendencies in discourse practices. Consider the principles of conversation articulated by Grice (see cooperative principle). These principles form a set of practical conditions for any sort of interaction between speakers. For example, it is a fundamental principle of conversation that one should not say things that are irrelevant to the conversation. Grice points out, however, that one may flout these principles, violating them in gross and obvious ways. Flouting a principle of conversation gives rise to interpretation. Jones and Smith are discussing who should be hired as the new assistant professor in the Hermeneutics of Suspicion. Jones says, “There’s a new applicant today – Heidegger. What do you think of him?” Smith replies, “Nice penmanship.” Since penmanship is clearly irrelevant to the topic, Smith may be said to be flouting the principle of relevance. Jones is likely to interpret Smith’s comment as indicating a dim view of Heidegger’s qualifications. Literary works often flout conversational principles.

69

The Cambridge Encyclopedia of the Language Sciences All these points indicate significant differences between verbal art and other sorts of speech action. Again, these differences do not create a hard-and-fast division between verbal art and all other types of speech. There is a continuum here, with many parameters and degrees of variation. Nonetheless, there is a clear differential tendency. The differences we have just been considering are a matter of the various linguistic levels of a literary text bearing autonomously or separately on our understanding and experience of the work. This is first of all and most obviously a matter of creating patterns. However, given the violations of syntactic rules and the flouting of conversational principles, it seems clear that verbal art does not simply create extra patterns on top of the usual, “instrumental” patterns produced by ordinary language processes. It also violates patterns of ordinary language processes. Most importantly, in both cases, the result renders the linguistic level in some way directly (rather than just instrumentally) relevant to our experience of the work. Thus, it maximizes the relevance of language features. But in what way are these features relevant? As with any sort of speech action, relevance is, first of all, relevance to the aims of the action. Again, the primary aims of verbal art are thematic and emotional. Thus, the maximization of relevance is the thematic or emotional use of (ordinarily unexpected) noninstrumental patterns – or violations of (ordinarily expected) instrumental patterns – from different linguistic levels. In touching on these issues, literary critics have tended to emphasize thematic relevance. Some writers have seen patterns and violations of patterns in phonology, morphology, and syntax as consequential for interpretation. It is probably true that such formal features are thematically interpretable in some cases. However, it seems doubtful that such features are generally relevant to interpretation. In contrast, extended patterns or violations in semantics and pragmatics are almost always interpretively relevant. In these aspects, the main difference between verbal art and other forms of speech action is where the interpretive process ends. We will consider this issue later. If anything, the maximization of relevance applies more fully to the communication of emotion than to the communication of themes. There are two types of affective response that bear importantly on verbal art, and thus on the maximization of relevance. We might refer to the first as pre-emotional and the second as specific emotional. Pre-emotional effects are effects of interest. Interest is pre-emotional in two senses. First, it is often an initial stage in an emotion episode. Second, it is a component of all specific emotions after those emotions have been activated. Specific emotions are simply our ordinary feelings – sorrow, joy, disgust, and so on. More exactly, interest is the activation of our attention system. That activation occurs whenever we experience something new or unexpected (see Frijda 1986, 272–3, 318, 325, 386). Such activation prepares us for events that have emotional significance. It directs our attention to aspects of the environment or our own bodies that may be emotion triggers. Once a specific emotion is activated, that emotion system reactivates our attention system, focusing it on properties relevant to that emotion in particular. For example, suppose I am out in the woods and hear something move. That arouses my attention. I carefully listen and look. If I see a friend at a distance, I feel some joy. That joy keeps my

70

attention on the friend and simultaneously directs my attention to ways in which I can reach him. If I see some sort of animal, I feel fear. That fear keeps my attention on the animal and simultaneously directs my attention to ways that I can escape. The arousal of interest is obviously crucial to literary experience. It is also important to other speech actions. However, in nonliterary cases, a great deal of the interest comes from the direct relation between the people involved (thus, the speaker and addressee), as well as the practical situation referenced in the speech action. In other words, a great deal comes from direct address and direct reference, both of which are greatly inhibited in verbal art. Nonetheless, verbal art has other means of producing interest. We have just been discussing two such means – the multiplication of non-normal patterns and violations of normal or expected patterns at various linguistic levels. Both are precisely the sorts of deviation from normalcy that produce interest. Specific emotional effects are fostered most obviously by semantics and pragmatics. For example, a great deal of our emotional response to verbal art seems to be bound up with the patterning of associative networks that spread out from lexical items (see Oatley 2002 and suggestion structure). The extension and elaboration of metaphorical structures are clearly consequential in this regard, particularly as the concreteness of metaphors often enhances associations with concrete experiential memories, including emotional memories. The specific emotional impact of phonological, morphological, and syntactic features is less obvious, but no less real, at least in some cases. For example, the interplay between nonlinguistic organization (e.g., in line divisions) and linguistic organization (e.g., in sentence divisions) may serve to convey a sense of a speaker’s voice and, along with this, an emotional tone. Thus, once more, we see both continuity and difference between verbal art and other sorts of speech action. Most speech actions involve a minimal, incidental patterning of linguistic levels. In contrast, the general tendency of literary art is toward the maximization of the relevance of different linguistic levels. This relevance is a function of the main purposes of the text, thematic and emotional. Again, this need not be relevance that increases informational content. Indeed, I suspect that it most often is not. It need not even contribute to the specific emotions of the text. It may be a matter of enhancing interest or of qualifying the specific emotions fostered by the text. In each case, though, there is some partially autonomous variation in the organization of the linguistic level in question, through the addition of unexpected patterning, through the violation of expected patterning, or both.

ON INTERPRETATION AND THE USES OF TEXTS In the preceding section, I referred briefly to the point at which interpretation ends. When verbal art is considered in relation to interpretation, the first thing to remark is that verbal art is notorious for its hermeneutic promiscuity (see philology and hermeneutics). It is widely seen as almost infinitely interpretable. At one level, this is surprising. Novels, for example, are developed in great detail and with considerable elaboration of the characters’ attitudes and actions. It might seem that this would constrain interpretation relative to the much vaguer and more elliptical speech actions of ordinary life. But that is not generally

Elaborating Speech and Writing believed to be the case. This suggests that the degree of interpretability of a text is not a function of its elaboration. Rather, interpretability appears to be a function of a text’s distance from direct address and direct reference to practical conditions, particularly as these are connected with pragmatic information. Put differently, the “limits” of interpretation are not so much a matter of the words themselves. They are, first of all, a matter of action. Here as elsewhere, action is animated by the agent’s goals, emotions, expectations, and so forth. In ordinary life, then, we usually understand validity in interpretation as a function of the speaker’s intention (see intentionality). More technically, our prototype of interpretive validity almost certainly includes the intention of the speaker or author as a critical norm. (The point is related to the argument of Steven Knapp and Walter Benn Michaels that we invariably interpret for speaker’s intention, though it is not identical, for there are cases where we do not adhere to this prototype and thus do not interpret for the speaker’s intention.) However, intention is an abstract criterion. We do not have access to speakers’ intentions. So even interpreting for intention, we need an operational criterion for validity as well. That is where our own action enters. Action commonly guides our sense of when we have gotten an interpretation right. At the dinner table, someone says, “Could you pass the that?” I am not sure what he means by “that.” Unable to read his mind, I engage in an action – either a bodily action (passing the beans) or a speech action (asking if he meant the beans). When I pass the beans, he knows that I have understood his intention. When he accepts the beans, I infer that I understood his intention. The interpretation of verbal art is as removed from such practical action as possible. Thus, our ordinary operational criterion is rendered ineffective. The only obvious practical behaviors relating to literary interpretation are professional – for example, the acceptance or rejection of articles in academic journals. (The point is codified in Fish’s contention that validity in literary interpretation is defined by the norms of interpretive communities.) A number of further problems for intentional inference arise in connection with this removal of literary interpretation from practical action. First, many texts are read or performed, and thus interpreted, far from their authors and even after their authors are dead. If we take a strict intentionalist view of validity, then the author is the only one who has the authority to determine that a given action does indeed satisfy an operational criterion. Suppose Jones leaves instructions for his funeral. The funeral director follows them as well as she can. But Jones is not around to confirm that she got things right. Obviously, there are things that she might do to ascertain Jones’s intention. For example, she might talk to Jones’s friends and relatives or she might try to learn something about Jones’s religion and ethnic background. These are the sorts of concerns that lead writers such as Hans-Georg Gadamer to stress tradition as a crucial guide to interpretation. A second problem is more distinctively connected with verbal art per se. Both informational and affective patterns are more complex in verbal art than in most other speech actions. For example, a literary work communicates thematically relevant information by maximizing the relevance of virtually every semantic and discursive detail in the text. If I am meeting

someone for the first time, I may describe my outfit so that he or she can recognize me. It has pragmatic consequences. But if an author describes a character’s outfit, that description may bear on our understanding of the character’s emotions, class status, or religious beliefs. Those features may in turn bear on our broader understanding of human relations, class conflict, or religious practice as portrayed in the work. In short, ordinarily incidental details may have thematic (thus regulative) consequences. Moreover, literary narratives tend to develop subtle and variable affinities and ambivalences (see Hogan 2003, 122–51). In some cases, the development of these affinities actually runs contrary to the author’s self-conscious sense of his or her own aims. The most famous case may be Milton’s portrayal of Satan. Satan has often been seen as the most engaging figure in Paradise Lost, but this was certainly not Milton’s self-conscious intention. This is part of a larger point that intention is not a single, unified operation. There are different sorts of intention with different objects, constraints, and processes. Perhaps the most important form of intention for verbal art is what we might call aesthetical intent (see Hogan 1996, 163–93). The aesthetical intent of an author is to produce a work that has the right sort of experiential effect. This “right” effect is not something that the author is likely to be able to articulate separately from the work itself. It is simply what he or she experiences when he or she feels that the work is now complete. In composing the work, the author generates and adjusts the text, testing the outcome against his or her own response. The author’s sense that the work is complete need not mean that the work conforms to the author’s self-conscious attitudes and commitments. One way of explaining aesthetical intent is that the author adopts an aesthetical attitude toward the work, or a “dhvani attitude,” as Anand Amaladass put it. This is a matter of broadening one’s attention to the work, expanding one’s encoding of the work, to include its multiple resonances and emotive patterns. In short, it involves approaching the text as a work in which different linguistic levels are maximally relevant. Correlatively, it involves approaching the text as a work that is removed from constraints of direct reference or address, constraints that ordinarily orient our judgment of informational relevance and our construal of emotional bearing. When approaching a work as verbal art, readers and audience members adopt this attitude as well to varying degrees. The mention of readers’ approaches to texts brings us to a final complication. We have been considering ways in which works produced as verbal art tend to be different from other speech actions. But the nature of a given text is not determined solely by authorial intent. Despite our prototypical concern for authorial intent, we are free to approach works of verbal art in pragmatic ways and to approach other works with an aesthetical attitude. As writers in cultural studies have stressed, practices of literary analysis may be applied to a wide range of texts that were not initially intended to be literary. In short, literariness may be defined by interpretation or reception no less than it may be defined by production; it may be defined by readers no less than by authors. In this way, the usual characteristics of verbal art may be extended to other texts – or withdrawn from texts (as when a novel is studied for its author’s psychology).

71

The Cambridge Encyclopedia of the Language Sciences On the other hand, the expansion of hermeneutic liberality results from the existence of verbal art. Interpretive autonomy arises in the first place through the removal of speech action from direct address and direct reference, along with the attenuation of pragmatic information and initiating emotion. In this way, interpretive practices that bridge literary and nonliterary speech actions are themselves a distinctive product of literary speech action.

SHAKESPEAREAN INDIRECTION As the preceding discussions have been rather abstract, it is valuable to end with a more developed example. Consider Shakespeare’s Sonnet 97: How like a winter hath my absence been From thee, the pleasure of the fleeting year! What freezings have I felt, what dark days seen! What old December’s bareness every where! And yet this time removed was summer’s time, The teeming autumn big with rich increase, Bearing the wanton burthen of the prime, Like widowed wombs after their lords’ decease: Yet this abundant issue seem’d to me But hope of orphans, and unfathered fruit, For summer and his pleasures wait on thee, And thou away, the very birds are mute; Or if they sing, ’tis with so dull a cheer That leaves look pale, dreading the winter’s near.

The poem presents a straightforward instance of indirect address. There is, of course, a narratee in the poem, an explicit “thee.” But the poem only plays at addressing this “thee.” The point is clear, for example, when the reader is brought in as a side participant with the otherwise superfluous information, “this time removed was summer’s time.” If there were a defining, material context and if there were direct address in this speech action, the speaker would not need to explain that the separation had occurred over the summer. The beloved would surely remember this. The point is reinforced more subtly by the shifts in spatial orientation in the poem (what some writers see as particular types of deixis [see Stockwell 2002, 41–57]). These may be understood as a matter of indirect reference. If this speech action were grounded in a particular situation, then there would almost certainly be some fixed reference point, a specific home that would define which of the lovers was stationary and which had departed. But the poem is contradictory on this score. In the opening lines, the speaker refers to “my absence …/From thee” and his “time removed.” But toward the end of the poem, he reverses the spatial orientation and related direction of movement. Now, the beloved is “away” and summer will “wait” for her return. The introduction of summer in this context suggests something else. The poet is not only like little Jane, covertly explaining what the third graders did in their last class. He may also be like Jane in referring to a telephone by way of a banana. Put differently, the indirectness of both address and reference make the reference of “summer” itself uncertain. Without a context, readers are free to associate summer tacitly with anything that can,

72

in play, be summer (e.g., a time of professional success for the speaker). In short, “summer” becomes a metaphor. The mention of metaphors points us toward the maximization of relevance. However, before going on to this, it is important to remark on some other contextual features of the poem, contextual features that are themselves bound up with metaphor. As I indicated earlier, the entire tradition of poetry forms an implicit context for any given poem (a point stressed by T. S. Eliot). One obvious way in which Shakespeare suggests a context of verbal art is in his use of seasonal imagery. Winter is a standard literary image of romantic separation. This is, of course, related to the conceptual metaphor, discussed by Lakoff and others: LIFE IS A YEAR. As Lakoff and Turner explain, there are several processes poets use to create novel instances of these schemas. Shakespeare is clearly creating such a novel instance when he maps the source metaphor of winter onto the target, summer (see source and target). Moreover, in making summer into (metaphorical) winter, he is intensifying the effect of the metaphor by contrast. But he does not say summer only. He expands the target time to summer and autumn. In a way, this is peculiar. The contrast would have been more obviously enhanced by making the target time period spring and summer. Why does Shakespeare choose autumn? There are several reasons. Two are closely interrelated. First, he wishes to intensify the tacit emplotment of the speaker’s isolation. That speaker is now anticipating the very worst, for if summer was like winter, how devastating will winter be? Second, he wishes to hold out hope for something beyond that period of terrible loneliness – spring. That hope is possible only through an indirect reference to the broader, literary context in which spring is the season when the lovers are finally reunited. The indirection of the poem makes it unlikely that it will have any pragmatic information as its aim. Pragmatic information most often involves reference to particular situations, or to forms of general knowledge. Of course, there is not a strict division between general pragmatic knowledge and, say, prudential regulatory information. But to the degree that the poem communicates thematic points, those points do incline toward prudential regulatory information. For example, suppose “summer” is metaphorical (i.e., “winter” is metaphorical for “summer,” which is itself metaphorical for, say, professional success). Then the poem suggests that the enjoyment of external conditions (e.g., professional success) is lost when it cannot be shared – a point with clear regulatory consequences. The indirection of the poem also indicates that it is unlikely to be aimed at the elicitation of initiating emotions. This can be seen if we contrast it with a similar letter, sent by the poet to his beloved. Such a letter could be aimed at convincing the beloved to return home. The poem, in contrast, does not have any such initiating emotional aim. Its emotional aim is, rather, confined to provoking empathic hypothetical emotions. The maximization of relevance contributes to the poem’s achievement of its primary aims. As usual, this maximization is most obvious at the phonological level. Consider only the first two lines. The poem is in iambic pentameter. However, when spoken naturally, these lines do not follow the meter. Indeed, there is considerable tension between spontaneous stress

Elaborating Speech and Writing patterns and the meter. One natural way of speaking the lines would be as follows: Hów líke ă wíntĕr hăth mў ábsĕnce bĕen Frŏm thée, thĕ pléasŭre ŏf thĕ fléetĭng yéar!

There are several things one might remark on here. First, in natural speech, the lines are in a somewhat irregular tetrameter. However, this irregularity is not unpatterned. There is a striking rhythmic motif that occurs in the middle of both lines. The sequences “líke ă wíntĕr hăth mў ábsĕnce” and “thée, thĕ pléasŭre ŏf thĕ fléetĭng” have the same stress pattern. This is not monotonous because the lines also manifest three variations. First, a caesura appears in only one of the sequences. Second, the first word of the line changes from stressed to unstressed. Finally, the last word changes from unstressed to stressed. Thus we find novelty, therefore the triggering of at least mild interest, at two levels. In addition, there are interpretive and specific emotional consequences. The disjunction of syntactic and verse breaks may help to give the poem a sense of voice. Specifically, it suggests to me a speaker who pauses before saying what is most painful about his absence. Being away from home means that one is absent from many things. But, here, there is only one crucial attachment. The line break imitates the emotion that so often interrupts one’s speech when such occasions are real and directly addressed. There may also be effects of tempo in these lines. “Hów líke ă wíntĕr” is heavy with accents and thus slower than “thĕ fléetĭng yéar.” This provides an instance of sound echoing sense. It also suggests an interpretive point – time flies in the sense that life passes quickly, leaving us little time together; but, being apart, we experience each moment of that fleeting time as a slow drag. We would have many further points to interpret if we were to consider lexical choice, metaphorical patterning, the flouting of conversational principles, and so forth. Each of these directs us toward the endless interpretability of verbal art. As in other works, there is no clear operational criterion that would tell us that we have reached the end of our interpretation or that our interpretation is correct. Of course, as Fish indicates, there are professional constraints on our judgments in these areas. A psychoanalytic critic might discuss how the poet seems to shift between the position of a separated lover and that of an orphan, suggesting an oedipal relation to the beloved. A writer in queer theory might stress that the poet puts his beloved in the position of the (deceased) father, not the mother, thus suggesting a male beloved. This brings us back to interpretation and reception. Again, we are always free to take indirect address and put it into a more directly referential context. For example, we may seek to read through the sonnets to Shakespeare’s own life and sexual feelings. Indeed, as Walter Ong pointed out many years ago, we have a strong tendency to do just that, placing the decontextualized voice of the poem back in a human body at a particular place in a particular time. That sense of concrete human embodiment is itself no doubt a crucial part not only of literary response but also of all human communication. The preceding point suggests once again the distinctiveness of verbal art and its continuity with other speech actions.

Verbal art both fosters the proliferation of interpretations and sharpens our sense of the human embodiment that limits those interpretations. We find the same tension between sameness and difference in all the characteristics we have considered. Although I have been stressing difference, the sameness is no less consequential. For example, in the context of an encyclopedia of language sciences, it is important that in studying verbal art, we are likely to isolate properties and relations in language that we might otherwise have passed over – properties and relations of address, reference, informational structure and orientation, and type and force of emotional consequence or function. In short, verbal art is a critical element of human life. As such, it is a critical object of study in its own right. It is also a crucial part of human speech action. As such it is a crucial, if sometimes neglected, part of the language sciences as well.

WORKS CITED AND SUGGESTIONS FOR FURTHER READING Amaladass, Anand. 1984. Philosophical Implications of Dhvani: Experience of Symbol Language in Indian Aesthetics. Vienna: De Nobili Research Library. Biber, Douglas, Susan Conrad, and Randi Reppen. 2006. Corpus Linguistics: Investigating Language Structure and Use. Cambridge: Cambridge University Press. Chandler, Steven. 1991. “Metaphor comprehension: A connectionist approach to implications for the mental lexicon.” Metaphor and Symbolic Activity 6.4: 227–58. Dasenbrock, Reed Way, ed. 1993. Literary Theory After Davidson. University Park: Pennsylvania State University Press. Eliot, T. S. 2001. “Tradition and the individual talent.” In The Norton Anthology of Theory and Criticism, ed. Vincent B. Leitch, 1092–8. New York: W. W. Norton. Fabb, Nigel, and Morris Halle. 2006. “Metrical complexity in Christina Rossetti’s verse.” College Literature 33.2: 91–114. Fish, Stanley. 1980. Is There a Text in This Class? The Authority of Interpretive Communities. Cambridge: Harvard University Press. Frijda, Nico. 1986. The Emotions. Cambridge: Cambridge University Press. Gadamer, Hans-Georg. 1989. Truth and Method. 2d ed. Trans. Joel Weinsheimer and Donald Marshall. New York: Crossroad. Gerrig, Richard, and Deborah Prentice. 1996. “Notes on audience response.” In Post-Theory: Reconstructing Film Studies, ed. David Bordwell and Noël Carroll, 388–403. Madison: University of Wisconsin Press. Grice, Paul. 1989. Studies in the Way of Words. Cambridge: Harvard University Press. Hogan, Patrick Colm. 1996. On Interpretation: Meaning and Inference in Law, Psychoanalysis, and Literature. Athens: University of Georgia Press. ———. 2003. The Mind and Its Stories: Narrative Universals and Human Emotion. Cambridge: Cambridge University Press. Jakobson, Roman. 1987. Language in Literature, ed. Krystyna Pomorska and Stephen Rudy. Cambridge, MA: Belknap Press. Kane, Julie. 2004. “Poetry as right-hemispheric language.” Journal of Consciousness Studies 11.5/6: 21–59. Kiparsky, Paul. 1987. “On theory and interpretation.” In The Linguistics of Writing: Arguments Between Language and Literature, ed. Nigel Fabb, Derek Attridge, Alan Durant, and Colin MacCabe, 185–98. New York: Methuen.

73

The Cambridge Encyclopedia of the Language Sciences Knapp, Steven, and Walter Benn Michaels. 1985. “Against theory.” In Against Theory: Literary Studies and the New Pragmatism, ed. W. J. T. Mitchell, 11–30. Chicago: University of Chicago Press. Labov, William. 1972. Language in the Inner City: Studies in the Black English Vernacular. Philadelphia: University of Pennsylvania Press. Lakoff, George, and Mark Turner. 1989. More Than Cool Reason: A Field Guide to Poetic Metaphor. Chicago: University of Chicago Press. MacNeilage, Peter. 1998. “Evolution of the mechanism of language output: Comparative neurobiology of vocal and manual communication.” In Approaches to the Evolution of Language: Social and Cognitive Bases, ed. James Hurford, Michael Studdert-Kennedy, and Chris Knight, 222–41. Cambridge: Cambridge University Press. Martindale, Colin. 1995. “Creativity and connectionism.” In The Creative Cognition Approach, ed. Steven Smith, Thomas Ward, and Ronald Finke, 249–68. Cambridge, MA: MIT Press. Oatley, Keith. 2002. “Emotions and the story worlds of fiction.” In Narrative Impact: Social and Cognitive Foundations, ed. Melanie Green, Jeffrey Strange, and Timothy Brock, 39–69. Mahwah, NJ: Erlbaum.

74

Ong, Walter J., S.J. “The jinnee in the well wrought urn.” In The Barbarian Within and Other Fugitive Essays and Studies, 15–25. New York: Macmillan. Pratt, Mary Louise. 1977. Toward a Speech Act Theory of Literary Discourse. Bloomington: Indiana University Press. Shakespeare, William. 2006. The Sonnets, ed. G. Blakemore Evans. Cambridge: Cambridge University Press. Stockwell, Peter. 2002. Cognitive Poetics: An Introduction. London: Routledge. Tsur, Reuven. 2006. Kubla Khan: Poetic Structure, Hypnotic Quality and Cognitive Style – A Study in Mental, Vocal and Critical Performance. Amsterdam: John Benjamins. Walter, Henrik. 2001. Neurophilosophy of Free Will: From Libertarian Illusions to a Concept of Natural Autonomy. Trans. Cynthia Klohr. Cambridge, MA: MIT Press.

THE CAMBRIDGE ENCYCLOPEDIA OF

THE LANGUAGE SCIENCES

Abduction

Absolute and Statistical Universals and system of language that following generations infer often differ from the system earlier generations are using. This often results in semantic change, syntactic change, and sound change.

A ABDUCTION

– Albert Atkin

Abduction is a form of reasoning first explicated by the nineteenth-century philosopher C. S. Peirce. The central concept he wishes to introduce is that of generating new hypotheses to explain observed phenomena partly by guesswork or speculation. In his early work, Peirce tried to explain abductive reasoning, as distinct from deductive and inductive reasoning, by reference to syllogistic form. For instance, the following schema is an example of deductive reasoning: All the beans in the bag are white These beans came from this bag Therefore, these beans are white

WORKS CITED AND SUGGESTIONS FOR FURTHER READING Anderson, Henning. 1973. “Abductive and deductive change.” Language 49: 765–93. Burks, Arthur. 1946. “Peirce’s theory of abduction.” Philosophy of Science 13: 301–6. McMahon, April. 1994. Understanding Language Change. Cambridge: Cambridge University Press. Peirce, C. S. 1935. The Collected Papers of Charles S. Peirce. Vol. 5. Cambridge: Harvard University Press.

ABSOLUTE AND STATISTICAL UNIVERSALS

This is distinct from inductive reasoning which, Peirce argues, follows this pattern: These beans came from this bag These beans are white Therefore, all the beans in this bag are white And both these forms are distinct from abductive reasoning which, Peirce argues, follows this pattern: These beans are white All the beans in this bag are white Therefore, the beans came from this bag In later work, however, Peirce felt that trying to fit abductive reasoning into such a strict syllogistic form was restrictive, and instead he opted for the following schema to explain abduction: The surprising fact C is observed But if A were true, C would be a matter of course Hence, there is a reason to suspect that A is true. (Peirce 1935, 189)

For example, suppose I observe that my car will not start. One good explanation for this would be that it is out of fuel. Consequently, it seems that we have a good reason to think that my car’s refusal to start is due to its being out of fuel. Of course, we may very quickly discover that my car has plenty of fuel, and a different hypothesis must be adopted, but Peirce always intended that abductive reasoning was fallible and conjectural, awaiting confirmation from other testing. Peirce’s account of abduction has been widely adopted in the philosophy of science, but it has also been of some interest to linguists. One particularly prominent use of abduction has been in historical linguistics for explaining language change (see, for instance, Anderson 1973). The systematic features of a language that govern the use of one generation are opaque to the following generation as they acquire that language – the only access is through language output. It appears, then, that following generations must use abductive inferences to access the rules of language before applying those rules to new cases. And, of course, since abduction is fallible, the rules

Language universals are statements that are true of all languages; for example, “all languages have stop consonants.” But beneath this simple definition lurks deep ambiguity, and this triggers misunderstanding in both interdisciplinary discourse and within linguistics itself. A core dimension of the ambiguity is captured by the opposition “absolute versus statistical universal,” although the literature uses these terms in varied ways. Many textbooks draw the boundary between absolute and statistical according to whether a sample of languages contains exceptions to a universal. But the notion of an exception-free sample is not very revealing, even if the sample contained all known languages: There is always a chance that an as yet undescribed language, or an unknown language from the past or future, will provide an exception. It is impossible, in principle, to survey all languages of our species. If we nevertheless want to make claims about all languages, only two routes are open: a priori deduction of necessarily true statements or statistical extrapolation from empirical samples to the entire set. Absolute universals can then be defined as those that are necessarily true, statistical universals as those that are extrapolated from samples.

Absolute Universals For statements to be necessarily true, they must follow from a priori assumptions. The assumptions that linguists make are diverse and heavily debated. An example is the assumption that words consist of morphemes, that is, minimal form-meaning pairs. If one accepts this, then it is necessarily true that all languages have morphemes, and there cannot be exceptions. Why? Suppose someone claims to have discovered a language without morphemes. One can of course simply analyze the language without mentioning morphemes, but obviously that cannot challenge the universal just because one can always defend it by reanalyzing the language with morphemes. The only true challenge would be to show that analyzing some data in terms of morphemes leads to structures that are in conflict with other assumptions, for example, that form-meaning pairs combine exclusively by linear concatenation. The conflict can be illustrated by languages with morphologies like the English plural geese, where the meanings plural and goose do not correspond to linear

77

Absolute and Statistical Universals

(iii) add additional assumptions that reconcile the conflict.

This view of absolute universals is highly controversial: Many linguists limit absolute universals to what is descriptively necessary in every language; many psychologists propose that children apply different and much more general principles in acquiring a language than those found in linguists’ metalanguages; and to date, no absolute universal has been confirmed by genetic research.

On any of these options, the universal remains exceptionless: On solution (i), no language has morphemes; on solutions (ii) and (iii), all languages have morphemes. As a result, absolute universals can never be falsified by individual data. Their validity can only be evaluated by exploring whether they are consistent with other absolute universals that are claimed simultaneously. Absolute universals can also be thought of as those aspects of one’s descriptive metalanguage – often called a theoretical framework – that are necessarily referred to in the analysis of every language, that is, that constitute the descriptive a priori. Depending on one’s a priori, this includes, apart from the morpheme, such notions as distinctive feature, constituent (see constituent structure), argument, predicate (see predicate and argument), reference, agent, speaker, and so on. In some metalanguages, the a priori also includes more specific assumptions, for example, that constituents can only be described by uniform branching (all to the left, or all to the right), or only by binary branching, and so on. The status of absolute universals is controversial. For many linguists, especially in typology and historical linguistics, absolute universals are simply the descriptive a priori, with no additional claim on biological or psychological reality. The choice between equally consistent universals/metalanguages – for example, among options (i), (ii), and (iii) in the previous example – is guided by their success in describing structures and in defining variables that capture distributional patterns, an evaluation procedure comparable to the way in which technical instruments for analyzing objects are evaluated in the natural sciences. In the morphology problem, typologists would most likely chose option (ii) because it allows for defining a variable of stem-internal versus affixal plural realization that has an interesting distribution (suggesting, for example, that within-stem realization is favored by a few families in Africa and the Near East). In generative grammar, by contrast, absolute universals are thought of not only as descriptively a priori but also as biologically given in what is called universal grammar: they are claimed to be innate (see innateness and innatism) and to be identical to the generalizations that a child makes when learning language. Thus, if the morpheme is accepted as a universal, that is, a priori term of our metalanguage, it will also be claimed to be part of what makes languages learnable (see learnability) and to be part of our genetic endowment. An immediate consequence of such an approach is that something can be claimed as universal even if it is not in fact necessary in the analysis of every language. For example, even if some language (e.g., the Rotokas language of Bougainville) lacks evidence for nasal sounds, one could still include a distinctive feature [± nasal] in Universal Grammar. Rotokas speakers are then said to have the feature as part of their genetic endowment even if they don’t use it.

What is not an absolute universal is a variable (or character, or parameter): some languages have a certain structure or they don’t have it, or to different degrees. It is interesting to note that most variables show some skewing in their distribution; some values of a variable are favored only in certain geographical areas (relative pronouns in Europe) or only in certain families (stem-internal inflection in Afroasiatic). But some values are globally favored (e.g., nasals) or, what is more typical, globally favored under certain structural conditions (e.g., postnominal relative clauses among languages with objects following the verb). These global preferences are called unconditional (unrestricted) and conditional (restricted) statistical universals, respectively. (An alternative term for conditional statistical universals is implicational universals, but this invites confusion because their probablistic nature differentiates them from logical implications; cf. Cysouw 2005) Statistical universals are mostly motivated by theories of how languages develop, how they are used, how they are learned, and how they are processed. One such theory, for example, proposes that processing preferences in the brain lead to a universal increase in the odds for postnominal structures among verb-object languages (Hawkins 2004). Statistical universals take the same forms as statistical hypotheses in any other science – for example, they can be formulated in terms of regression models. They can be tested with the same range of statistical methods as in any other science, and, again as in other sciences, the appropriate choice of models, population assumptions, and testing methods is an issue of ongoing research (e.g. Cysouw 2005; Janssen, Bickel, and Zúñiga 2006; Maslova 2008). A central concern when testing statistical universals is to ascertain true globality, that is, independence of area and family. Areas can be controlled for by standard factorial analysis, but it is an unsettled question just what the relevant areal relations are; for example, should one control for the influence of Europe or the entire Eurasia or both? A quick solution is to assume a standard set of five or six macroareas in the world and accept as universal if a distribution is independent of these areas (Dryer 1989). But the rationale for such a set is problematic, and this has led to a steep surge of interest in research on areas and their historical background (e.g., Nichols 1992; Haspelmath et al. 2005). Controlling for family relations poses another problem. Under standard statistical procedures, one would draw random samples of equal size within each family and then model families as levels of a factor. However, over a third of all known families are isolates, containing only one member each. And picking one member at random in larger families is impossible if at the same time one wants to control for areas (e.g., admitting an Indo-European language from both Europe and South Asia).

strings of morphemes. Confronted with such data, there are three options: (i) give up the notion of morpheme; (ii) give up the assumption of linear concatenation;

78

Statistical Universals

Absolute and Statistical Universals

Accessibility Hierarchy

In response to this problem, typologists seek to ensure representativity of a sample not by random selection within families but by exhaustive sampling of known families, stratified by area. In order to then control for unequal family sizes, one usually admits only as many data points per family as there are different values on the variables of interest (Dryer 1989; Bickel 2008). Samples that are not based on random sampling do not support parametric inference by statistical tests. An alternative to this is randomization methods (Janssen, Bickel, and Zúñiga 2006): The null hypothesis in these methods is that an observed preference can be predicted from the totals of the sample (e.g., that an observed 90% postnominal relatives in VO [verb-object] languages could be predicted if 90% of the entire sample had postnominal relatives) – not that the sample stems from a population without the observed preference. Extrapolation to the total population (the entire of set of human languages) can then only be based on plausibility arguments: If a preference significantly deviates from what is expected from the totals of the observed sample, it is likely that the preference holds in all languages. A key issue in such argumentation is whether the tested variables are sufficiently unstable over time so that a present sample can be assumed to not reflect accidental population skewings from early times in prehistory (Maslova 2000). In response to this, typologists now also seek to test universals by sampling language changes instead of language states – a move that is sometimes called the dynamization of typology (Greenberg 1995; Croft 2003). While the number of proposed statistical universals is impressive – the Universals Archive at Konstanz has collected more than 2,000 (Plank and Filimonova 2000) – very few of them have been rigorously tested for independence of area, family, and time. – Balthasar Bickel WORKS CITED AND SUGGESTIONS FOR FURTHER READING Bickel, B. 2008. “A refined sampling procedure for genealogical control.” Sprachtypologie und Universalienforschung 61: 22–33. Croft, W. 2003. Typology and universals. 2d ed. Cambridge: Cambridge University Press. Cysouw, M. 2005. “Quantitative methods in typology.” In Quantitative Linguistics: An International Handbook, ed. G. Altmann, R. Köhler, and R. Piotrowski, 554–78. Berlin: Mouton de Gruyter. Cysouw, M, ed. 2008. Special issue on analyzing The World Atlas of Language Structures. Sprachtypologie und Universalienforschung 61. Dryer, M. S. 1989. “Large linguistic areas and language sampling.” Studies in Language 13: 257–92. Greenberg, J. H. 1995. “The diachronic typological approach to language.” In Approaches to Language Typology, ed. M. Shibatani and T. Bynon, 143–66. Oxford: Clarendon. Haspelmath, M., M. S. Dryer, D. Gil, and B. Comrie, eds. 2005. The World Atlas of Language Structures. Oxford: Oxford University Press. Hauser, M. D., N. Chomsky, and W. T. Fitch. 2002. “The faculty of language: What it is, who has it, and how did it evolve?” Science 298: 1569– 79. This paper and the response by S. Pinker and R. Jackendoff 2005 launched an ongoing debate on the nature and extent of absolute universals in generative grammar. Hawkins, J. A. 2004. Efficiency and Complexity in Grammars. Oxford: Oxford University Press. Janssen, D., B. Bickel, and F. Zúñiga. 2006. “Randomization tests in language typology.” Linguistic Typology 10: 419–40.

Maslova, E.. 2000. “A dynamic approach to the verification of distributional universals.” Linguistic Typology 4: 307–33. ———. 2008. “Meta-typological distributions.” Sprachtypologie und Universalienforschung 61: 199–207. Newmeyer, F. J. 2005. Possible and Probable Languages: A Generative Perspective on Linguistic Typology. New York: Oxford University Press. Nichols, J. 1992. Language Diversity in Space and Time. Chicago: The University of Chicago Press. Pinker, S., and R. Jackendoff. 2005. “The faculty of language: What’s special about it?” Cognition 95: 201–36. Plank, F., and E. Filimonova. 2000. “The Universals Archive: A brief introduction to prospective users.” Sprachtypologie und Universalienforschung 53: 109–23. Plank, F., ed. 2007. Linguistic Typology 11.1 (Special issue treating the state of typology.)

ACCESSIBILITY HIERARCHY Edward L. Keenan and Bernard Comrie (1972, 1977) introduce the accessibility hierarchy (AH) as a basis for several crosslinguistic generalizations regarding the formation of relative clauses (RCs). AH: SUBJ > DO > IO > OBL > GEN > OCOMP

The terms of the AH are main clause subject, direct object, indirect object, object of pre- or postposition, genitive (possessor), and object of comparison. Keenan and Comrie cross-classified RCs along two parameters: 1) the head noun precedes or follows the restrictive clause (RestCl), and 2) the case of the position relativized, NPrel, is pronominally marked or not. In (1), from German, the RestCl, underlined, follows the head in (1a) and precedes it in (1b). In (2a,b), from Hebrew and Russian, NPrel is pronominally marked but not in English. (1) a. der Mann, der in seinem Büro arbeitet the man, who in his study is+working the man who is working in his study b. der in seinem Büro arbeitende Mann the in his study working man the man who is working in his study (2) a. ha-isha she Dan natan la et ha-sefer the-woman that Dan gave to+her acc the-book the woman that Dan gave the book to b. devuška, kotoruju Petr ljubit girl, who(acc) Peter loves the girl who Peter loves

A given choice of values for the two parameters defines an RC-forming strategy. A strategy that applies to SUBJ is called primary. German has two primary strategies, a postnominal, +case one, (1a), and a prenominal, –case one, (1b). Keenan and Comrie support three hierarchy generalizations: (3) a. All languages have a primary strategy b. A given RC-forming strategy must apply to a continuous segment of the AH c. A primary strategy may cease to apply at any position on the AH

For example, many West Austronesian languages, such as Malagasy (Madagascar), only have primary strategies. So we can only relativize the agent in (4a).

79

Accessibility Hierarchy (4) a. Manolotra (m+aN+tolotra) vary ho an’ny vahiny amin’ny lovia vaovao ny tovovavy offers (pres+act+offer) rice for’the guests on’the dishes new the young+woman The young woman offers rice to the guests on the new dishes b. ny tovovavy (izay) manolotra vary ho an’ny vahiny amin’ny lovia vaovao the woman (that) offers rice to’the guests on’the dishes new the young woman who offers rice to the guests on the new dishes c. *ny vary (izay) manolotra ho an’ny vahiny amin’ny lovia vaovao ny tovovavy the rice (that) offers to’the guests on’the dishes new the young+woman the rice that the young woman offers to the guests on the new dishes

The first four words in (4c) claim that the rice is doing the offering – a nonsense. Malagasy does not, however, have an expressivity gap here since it has a rich voice system allowing any major NP in a clause as subject. The form of offer that takes theme subjects is atolotra, recipient subjects tolorana, and oblique subjects anolorana. (5a,b) illustrate Theme and Instrument RCs. (5) a. ny vary (izay) atolo-dRasoa ho an’ny vahiny amin’ny lovia vaovao the rice (that) offered-by+Rasoa for’the guests on’the new dishes the rice that the young+woman offers to the guests on the new dishes b. ny lovia vaovao (izay) anoloran-dRasoa vary ho an’ny vahiny the dishes new (that) offered-by+Rasoa rice for ‘the guests the new dishes on which the young woman offered rice to the guests

Bantu languages, such as Luganda, (6), illustrate the DO cutoff. Only subjects and objects are directly relativizable. Obliques can be promoted to object using applicative affixed verbs. So the instrumental in (6a) is only relativizable from (6c). (6) a. John yatta enkoko n’ (= na) ekiso John killed chicken with knife b. *ekiso John kye-yatta enkoko (na) knife John rel-killed chicken (with) c. John yattisa (yatt+is+a) ekiso enkoko John kill+with knife chicken John killed+with a knife the chicken d. ekiso John kye-yattisa enkoko knife John rel-kill+with chicken the knife John killed the chicken with

Independent support for the AH: Keenan (1975) shows that stylistically simple texts used RCs formed high on the AH proportionately more than texts independently judged stylistically hard. Second, Comrie’s work (1976) supports the conclusion that the positioning of demoted subjects in morphological causatives tends to assume the highest function on the AH not already filled. Thus in the French J’ai fait rire les enfants “I made-laugh the children,” the children surfaces as a DO as laugh lacks a DO. But in

80

Acoustic Phonetics causativizing a transitive verb, its agent argument may surface as an IO (J’ai fait manger les épinards aux enfants “I made-eat the spinach to the children). Lastly, S. Hawkins and Keenan (1987) show psycholinguistically that recall of RCs formed on high positions on the AH was better than recall of ones formed on low positions. One interesting modification to the hierarchy generalizations concerns syntactic ergativity. Keenan and Comrie noted that Dyirbal (Dixon 1972) relativizes absolutives – intransitive subjects and transitive objects, but not transitive subjects. A verbal affix (antipassive) derives intransitive verbs from transitive ones with the agent as subject, hence relativizable. Mayan languages such as Jacaltec (Craig 1977, 196) are similar. This is an elegant solution to the requirement that agents be relativizable, analogous to Bantu applicatives or Austronesian voices affixes. (7) a. x – ɩ – s – watx’e naj hun-ti’ asp 3abs 3erg make cl:man one-this He made this b. naj x – ɩ – watx’e – n hun-ti cl:man asp 3abs make ap one-this the man (who) made this

– Edward L. Keenan WORKS CITED AND SUGGESTIONS FOR FURTHER READING Comrie, Bernard. 1976. “The syntax of causative constructions: Crosslanguage similarities and divergences.” In The Grammar of Causative Constructions, Syntax and Semantics 6, ed. Masayoshi Shibatani, 261– 312. Amsterdam: Academic Press. Craig, Colette. 1977. Jacaltec. Austin: University of Texas Press. Dixon, Robert M. W. 1972. The Dyirbal Language of North Queensland. Oxford: Cambridge University Press. Hawkins, Sarah, and Edward L. Keenan. 1987. “The psychological validity of the accessibility hierarchy.” In Universal Grammar: 15 Essays, ed. E. L. Keenan, 60–89. London: Croom Helm. Keenan, Edward L. 1975. “Variation in Universal Grammar.” In Analyzing Variation in English, ed. R. Fasold and R. Shuy, 136–48. Washington, DC: Georgetown University Press. Keenan, Edward L., and Bernard Comrie. 1972. “Noun phrase accessibility and Universal Grammar.” Paper presented at the Annual Meetings of the Linguistic Society of America, Atlanta. ———. 1977. “Noun phrase accessibility and Universal Grammar.” Linguistic Inquiry 8.1: 63–99.

ACOUSTIC PHONETICS Like the rest of linguistics, acoustic phonetics combines description with theory. Descriptions are images of acoustic properties and quantitative measures taken from these images, and theory accounts for the way in which a sound’s articulation determines its acoustics.

Description The three most commonly used images of speech are the waveform, spectrum, and spectrogram (Figure 1). The waveform displays differences in sound pressure level (in pascals) over time (Figure 1a, d), the spectrum differences in sound pressure level

Acoustic Phonetics

Figure 1. Waveforms, spectra, and spectrograms of 30 ms intervals of the vowel [i] (a–c), and the fricative [s] (d–f).

(in deciBels) over frequency (Figure 1b, e), and the spectrogram differences in frequency over time (Figure 1c, f,); darkness indicates the sound pressure level at particular frequencies and moments in the spectrogram. The images in Figures 1a–c differ from those in Figure 1d–f in every conceivable respect: Sound pressure level varies more or less regularly and repetitively, every 0.0073 second, in the vowel [i] (as in heed), while in the fricative [s] (as in see), it instead varies nearly randomly. The vowel is thus nearly periodic, while the fricative is decidedly aperiodic. This difference gives the vowel a clear pitch, while making the fricative instead sound noisy. A single cycle’s duration in a periodic sound is its period (T); the distance it travels in space is its wavelength (λ). As measures of a single cycle’s extent, both period and wavelength are reciprocally related to frequency (F), the number of cycles per second, or Hz: (1)

F (cycles/sec) =

1 T (sec/cycle)

(2)

F (cycles/sec) =

c (cm/sec) λ (cm/cycle)

Note that the numerator in (2) is not 1 but instead c, the speed of sound. The spectrum and spectrogram of [i] (Figures 1b, c) show peaks and horizontal bands, respectively, known as formants, at roughly 300, 2,200, and 2,800 Hz. The corresponding images of [s] (Figures 1e, f) show a broad energy band spanning 4,000–7,000 Hz. Whether a sound is periodic and where in its spectrum energy is concentrated are nearly sufficient to distinguish all speech sounds from one another acoustically, and these two properties also reflect the two components of the theoretical model for transforming articulations into acoustics. All voiced sounds are periodic, as are trills. Sonorants (vowels, glides, liquids, nasals) are usually periodic, while obstruents (fricatives, stops, affricates) are aperiodic. Voiced obstruents are both periodic and aperiodic. Differences in vowel quality and consonantal place of articulation are both realized acoustically as differences in where energy is concentrated in their spectra. The remaining property is duration, which besides conveying short:long contrasts also contributes to the conveying of tense:lax contrasts between vowels and the voicing and manner contrasts between consonants.

Theory Speech sounds are the product of the application of a filter that determines the frequencies in which energy is concentrated

81

Acoustic Phonetics First resonance

Glottis = closed

Minimum

0

released or continuously through a fricative’s narrow constriction. In strident sounds, the jet breaks up against a baffle just downstream, increasing turbulence and noise intensity considerably.

Lips = open

Maximum

2.5

5

7.5

10

12.5

15

17.5

(a)

Distance from glottis (cm)

Second resonance

0

2.5

5

7.5

10

12.5

15

17.5

(b)

Distance from glottis (cm)

Third resonance

0

2.5

5

7.5

10

Distance from glottis (cm)

12.5

15

17.5

(c)

Figure 2. The oral cavity as a tube closed at one end and open at the other: (a–c) standing waves corresponding to the first three resonances, each with a velocity minimum at the closed end and a velocity maximum at the open end.

RESONANCE. Both periodic and aperiodic sound sources introduce acoustic energy into the oral cavity across a broad enough range of frequencies to excite any resonance of the air inside the oral cavity. If the articulators are in their rest positions and vocal folds are in the voicing position, this cavity’s shape approximates a uniform tube, closed at the glottis and open at the lips (Figure 2). A resonance is produced by the propagation of acoustic energy away from the source and its reflection back and forth off the two ends of the tube, which establishes a standing wave. In a standing wave resonance, the locations of zero and maximum pressure variation are fixed.To understand how air resonates, it is easier to consider the change in pressure level in the standing wave, rather than pressure level itself, that is, the extent to which the air molecules are being displaced longitudinally, or equivalently the velocity of the pressure change, rather than the extent of their instantaneous compression or rarefaction. Air is most freely displaced longitudinally at the open end, the lips, and least freely at the closed end, the glottis. As a result, the standing waves that fit best inside the oral cavity are those whose wavelengths, and thus frequencies, are such that they have a velocity maximum (antinode) at the lips, and a velocity minimum (node) at the glottis. Because the air resonates more robustly at these frequencies than at others, the oral cavity’s resonant response filters the sound source, passing energy in the source at some frequencies and stopping it at others. Figures 2a–c show the three lowest-frequency standing waves that fit these boundary conditions. How are their frequencies determined? Figures 2a–c show that one-quarter of the first resonance’s wavelength spans the distance from the glottis to the lips (the oral cavity’s length, Loc), threequarters of the second’s, and five-quarters of the third’s. More generally: (3)

in the sound’s spectrum to a periodic and/or aperiodic sound source. SOUND SOURCES. Sound sources are produced by using valves that control air flow through the vocal tract to transform the energy in that flow into sound.In periodic sound sources, the flow of air causes a valve to open and close rapidly and regularly, which in turn causes air pressure to rise and fall just downstream. The repeated opening and closing of the glottis, known as vocal fold vibration, is the most common periodic sound source; others are uvular, alveolar, and bilabial trills. Aperiodic sound sources are produced by keeping a valve completely or nearly closed, in stops and fricatives, respectively. Either way, oral air flow is obstructed enough that oral air pressure rises behind the obstruction. This pressure rise speeds up flow enough to turn it into a turbulent and thus noisy jet, in either a brief burst when a stop closure is

82

Loc =

2n − 1 λn 4

where n is the resonance number. Solving for wavelength and substituting into (2) yields:

(4)

Fn =

c 4 Loc 2n − 1

Substituting 35,000 cm/sec for c and 17.5 cm for Loc (the average adult male’s oral cavity length) yields 500, 1,500, and 2,500 Hz as the first three resonances’ frequencies, values close to schwa’s. Because the variable Loc is in the denominator, resonance frequencies are lower in adults’ and men’s longer oral cavities than in children’s or women’s shorter ones, and likewise when the lips are protruded in a rounded vowel, such

Acoustic Phonetics

Figure 3. Spectra of the vowels (a) [i] as in heed, (b) [u] as in who’d, and (c) [a] as in hod. The individual peaks are the harmonics of the fundamental frequency (F0) of voice sound source, and the formants are the ranges of amplified harmonics. Peaks corresponding to F1–F3 are labeled at the top of each panel.

as the [u] in who’d, rather than spread in an unrounded one, such as the [i] in heed. These observations yield the length rule: Resonance frequencies vary inversely with resonating cavity length. The first three formants of [i] (Figures 1b, c) differ decidedly in frequency from those of schwa, because raising the tongue body toward the front of the palate decreases the oral cavity’s cross-sectional area there and increases it in the pharynx, while spreading the lips, which shortens the oral cavity. Although the length rule predicts how shortening changes formant frequencies, an additional rule is needed to predict how decreasing and increasing cross-sectional area affects formant frequencies. The predictions of two heuristics that are widely used for this purpose are tested here against the observed formant frequency differences between the three vowels [i, u, a], and between the three places of articulation of the stops in [bæ, dæ, gæ]. F1 is lower in the high vowels [i, u] (Figures 3a, b) than in schwa, but higher in the low vowel [a] (Figure 3c). F2

Figure 4. Spectrograms of the first 150 ms of the words (a) bad, (b) dad, and (c) gad. The onsets of F1–F3 are labeled.

is higher in front unrounded [i] than schwa but lower in low back unrounded [a] and especially high back rounded [u]. F1 varies inversely with tongue height, and F2 varies directly with tongue backness and lip rounding. The formant frequencies of all other vowel qualities lie between the extremes observed in these vowels, just as all other vowels’ lingual and labial articulations lie between those of these vowels. F1 starts low and rises following [b, d, g] (Figure 4), both F2 and F3 start low following [b] (Figure 4a), both formants start higher following [d] (Figure 4b), and they diverge from very similar frequencies following [g] (Figure 4c). Although consonants are articulated at other places of articulation, these three places are distinguished in nearly all languages, and many distinguish only these three. The first heuristic treats the constriction as dividing the oral cavity into separate resonating cavities (Figure 5), and it applies the length rule independently to each of them. The first three formants are the three lowest of the six resonances produced by the two cavities. This heuristic may be called the cavity association heuristic because each formant can be associated with the cavity from which it came. There are two complications. First, the cavity behind the constriction is effectively closed at both ends, and

83

Acoustic Phonetics constriction

back

8 cm2

1 cm2

front

3 cm 6 cm

8.5 cm

Figure 5. The configuration of the oral cavity with a constriction partway along its length.

so its resonances must have velocity minima at both ends. Their frequencies are predicted by:

(5)

Fn =

c 2 n Lrc

where Lrc is the length of the resonating cavity. The second complication is that the acoustic interaction of the constriction with the cavity behind it produces a Helmholtz resonance. Its frequency (Fh) is: (6) Fh =

c 2π

Ac Ab Lb Lc

Ac is the constriction’s cross-sectional area, Lc is its length, and Ab and Lb are the cross-sectional area and length of the cavity behind the constriction. If a 3 cm–long constriction with a crosssectional area of 1 cm2 is moved incrementally from 3 cm above the glottis to 0.5 cm back of the lips along a 17.5 cm oral cavity, the back and front cavities produce the resonance frequencies displayed in Figure 6, along with the Helmholtz resonance. The arrows projected down from the intersections between back and front cavity resonances show where F2 and F3 change association from the front to the back cavity. The constriction centers in [a, u, i] are roughly one-quarter (4.5 cm), two-thirds (11.5 cm), and three-quarters (13 cm) of the distance from the glottis to the lips. The constriction of [g] is close to [u]’s, while [d]’s is about seven-eighths the distance from the glottis (15.5 cm), and [b]’s is of course at the lips (17.5 cm). The Helmholtz resonance is lowest for all constriction locations and thus constitutes F1. It also lowers progressively as the constriction is moved forward because the cavity behind the constriction lengthens. The cavity-association heuristic’s successful predictions include the following : 1) The low or pharyngeal vowel [a] has a higher F1 than the high or velar and palatal vowels [u, i]; 2) F1 is low following [g, d, b]; 3) [a]’s F2 (the front cavity’s first resonance) is low, 4) F2 and F3 (the front and back cavities’ first resonances) start at very similar frequencies following [g], because a velar constriction is close to where the front and back cavities’ first resonances cross at 11 cm from the glottis; 5) F2 and F3 start low following [b] (the back cavity’s first and second resonances). It incorrectly predicts: 6) The F2 of [i] (the back cavity’s first resonance) is low, indeed lower than [u]’s; and 7) F2 and F3 (the back cavity’s first and second resonances) are low following [d]. For [u], the calculations leave out the acoustic

84

Figure 6. The first three resonance frequencies of the back cavity (filled symbols) and front cavity (empty symbols) and the Helmholtz resonance (crosses) produced by incremental movement of the constriction in Figure 5.

effects of lip rounding, which closes the front cavity at both ends and introduces another Helmholtz resonance. None of the resonances produced by this front cavity are lower than the back cavity resonances, but the additional Helmholtz resonance is low enough to constitute the F2 observed in [u] (657 Hz if the labial constriction has a cross-sectional area of 1 cm2 and a length of 2 cm, and the front cavity is 4.5 cm long). In the second, perturbation, heuristic, a constriction’s proximity to a resonance’s velocity minimum or maximum determines how it perturbs that resonance’s frequency away from its schwa-like value: A constriction near a velocity minimum raises the formant’s frequency, while one near a maximum lowers it instead (expansions have the opposite effects). Figures 4a–c show that minima occur at even quarters of a resonance’s wavelength and maxima at odd quarters, and that their locations are at fixed proportions of the length of the oral cavity. Because constriction locations are also a certain proportion of the distance from the glottis to the lips, whether they coincide with a minimum or maximum can be calculated by multiplying both sides of (4) by the proportion of the oral cavity’s length that corresponds to the constriction’s location and rounding the result on the right-hand side to the nearest quarter (Table I). The perturbation heuristic successfully predicts the effects of the bilabial, palatal, velar, and pharyngeal constrictions on all three formants of [b, i, g, a], and likewise the effects of the alveolar and velar constrictions on F1 and F3 in [d, u], but it fails to predict F2 raising after [d], and F2 lowering in [u]. The latter can

Adaptation Table 1. Calculating a constriction’s proximity to a resonance’s velocity minimum or maximum from the constriction’s proportional distance from the glottis to the lips. Place of constriction

Segment

Proportion of oral cavity length

Proximity to minimum or maximum Calculation

Odd/even

Lower/higher

Labial

b

1

1*λ1/4=1/4 1*3λ2/4=3/4 1*5λ3/4=1/4

Odd Odd Odd

F1 lower F2 lower F3 lower

Alveolar

d

7/8

7/8*λ1/4=7/32≈1/4 7/8*3λ2/4=21/32≈3/4 7/8*5λ3/4=35/32≈4/4

Odd Odd Even

F1 lower F2 lower F3 higher

Palatal

i

3/4

3/4*λ1/4=3/16≈1/4 3/4*3λ2/4=9/16≈2/4 3/4*5λ3/4=15/16≈4/4

Odd Even Even

F1 lower F2 higher F3 higher

Velar

g, u

2/3

2/3*λ1/4=2/12≈1/4 2/3*3λ2/4=6/12≈2/4 2/3*5λ3/4=10/12≈3/4

Odd Even Odd

F1 lower F2 higher F3 lower

Pharyngeal

a

1/4

1/4*λ1/4=1/16≈0/4 1/4*3λ2/4=3/16≈1/4 1/4*5λ3/4=5/16≈1/4

Even Odd Odd

F1 higher F2 lower F3 lower

again be predicted once the acoustic effects of lip rounding are added, as the simultaneous labial constriction, together with the protrusion of the lips, lowers F2 along with all other formants.

Summary Speech sounds’ articulations produce sound sources by transforming aerodynamic energy into acoustic form, and those sound sources in turn cause air inside the oral cavity to resonate, at frequencies determined by the length of the resonating cavities and where they are constricted. – John Kingston WORKS CITED AND SUGGESTIONS FOR FURTHER READING Fant, C., and M. Gunnar. 1960. Acoustic Theory of Speech Production. The Hague: Mouton. Jakobson, Roman, C. Fant, M. Gunnar, and Morris Halle. 1952. Preliminaries to Speech Analysis, Cambridge, MA: MIT Press. Ladefoged, Peter, and Ian Maddieson. 1996. Sounds of the World’s Languages. Oxford: Blackwell Publishers. Stevens, Kenneth N. 1998. Acoustic Phonetics. Cambridge, MA: MIT Press.

ADAPTATION An adaptation is a characteristic in an organism that evolved because it helped the organism or its relatives to survive and reproduce. Examples include the vertebrate eye, claws, mammary glands, the immune system, and the brain structures that underlie the human capacity for language. More completely, an adaptation is 1) a reliably developing set of characteristics 2) whose genetic basis became established and organized in the species (or population) over evolutionary time because 3) the adaptation interacted with recurring features of the body or environment 4) in a way that, across generations, typically caused this genetic basis to increase its gene frequency.

If a characteristic lacks any of these features, it is not an adaptation. An adaptation is not, therefore, simply anything in an individual with a “good” or “functional” outcome, or that has useful effects by intuitive standards. Rice cultivation, useful as it is, is not a biological adaptation because it lacks a specific genetic basis. Similarly, the English language is not an adaptation, however useful it might be. In contrast, if a mutation occurred that modified a neural structure so that the vocal chords could more reliably produce distinct phonemes, and this gene spread throughout the species because its bearers prospered due to the advantages resulting from a lifetime of more efficient communication, then the modified neural structure would qualify as an adaptation. Researchers judge whether something is an adaptation by assessing how likely or unlikely it is that its functional organization was produced by random mutation and spread by genetic drift. For example, the eye has hundreds of elements that are arranged with great precision to produce useful visual inputs. It is astronomically unlikely that they would have arrived at such high levels of mutual coordination and organization for that function unless the process of natural selection had differentially retained them and spread them throughout the species. Consequently, the eye and the visual system are widely considered to be obvious examples of adaptations. For the same reason, evolutionary scientists consider it overwhelmingly likely that many neurocognitive mechanisms underlying language are adaptations for communication (a proposition that Noam Chomsky has disputed; see Lyle Jenkins’s essay, “Explaining Language,” in this volume). Language competence reliably develops, is believed to have a species-typical genetic basis, and exhibits immensely complex internal coordination that is functionally organized to produce efficient communication, which vastly enhances the achievement of instrumental goals, plausibly including those linked to fitness.

85

Ad hoc Categories Within the evolutionary sciences, the concept of adaptation plays an indispensable role not only in explaining and understanding how the properties of organisms came to be what they are, but also in predicting and discovering previously unknown characteristics in the brains and bodies of species.Evolutionary psychologists, for example, analyze the adaptive problems our ancestors were subjected to, predict the properties of previously unknown cognitive mechanisms that are expected to have evolved to solve these adaptive problems, and then conduct experimental studies to test for the existence of psychological adaptations with the predicted design (see evolutionary psychology). An understanding that organisms embody sets of adaptations rather than just being accidental agglomerations of random properties allows organisms to be properly studied as functional systems. If language is accepted as being the product of adaptations, then there is a scientific justification for studying the underlying components as part of a functional system. The concept of adaptation became more contentious when human behavior and the human psychological architecture began to be studied from an adaptationist perspective. Critics have argued that not every characteristic is an adaptation – an error adaptationists also criticize. More substantively, critics have argued that it is impossible to know what the past was like well enough to recognize whether something is an adaptation. Adaptationists counter that we know many thousands of things about the past with precision and certainty, such as the threedimensional nature of space and the properties of chemicals, the existence of predators, genetic relatives, eyes, infants, food and fertile matings, and the acoustical properties of the atmosphere, and that these can be used to gain an engineer’s insight into why organisms (including humans) are designed as they are. – Julian Lim, John Tooby, and Leda Cosmides WORKS CITED AND SUGGESTIONS FOR FURTHER READING Gould, S. J., and R. C. Lewontin. 1979. “The spandrels of San Marco and the Panglossian paradigm: A critique of the adaptationist programme.” Proceedings of The Royal Society of London, Series B 205.1161: 581–98. Pinker, Steven. 1994. The Language Instinct. New York: Morrow. ———. 2003. “Language as an adaptation to the cognitive niche.” In Language Evolution, ed. M. Christiansen and S. Kirby, 16–37. New York: Oxford University Press. Tooby, John, and I. DeVore. 1987. “The reconstruction of hominid behavioral evolution through strategic modeling.” In The Evolution of Primate Behavior: Primate Models. ed. W. G. Kinsey, 183–237. New York: SUNY Press. Williams, George C. 1966. Adaptation and Natural Selection: A Critique of Some Current Evolutionary Thought. Princeton, NJ: Princeton University Press.

AD HOC CATEGORIES An ad hoc category is a novel category constructed spontaneously to achieve a goal relevant in the current situation (e.g., constructing tourist activities to perform in Beijing while planning a vacation). These categories are novel because they typically have not been entertained previously. They are constructed spontaneously because they do not reside as knowledge structures in long-term memory waiting to be retrieved. They help achieve a

86

relevant goal by organizing the current situation in a way that supports effective goal pursuit. Ad hoc categories contrast with thousands of well-established categories associated with familiar words (e.g., cat, eat, happy). Extensive knowledge about these latter categories resides in memory and may often become active even when irrelevant to current goals. When ad hoc categories are used frequently, however, they, too, become highly familiar and well established in memory. The first time that someone packs a suitcase, the category things to pack in a suitcase is ad hoc. Following many trips, however, it becomes entrenched in memory. Ad hoc categories constitute a subset of role categories, where roles provide arguments for verbs, relations, and schemata. Some role categories are so familiar that they become lexicalized (e.g., seller, buyer, merchandise, and payment name the agent, recipient, theme, and instrument roles of buy). When the conceptualization of a role is novel, however, an ad hoc category results (e.g., potential sellers of gypsy jazz guitars). Pursuing goals requires the constant specification and instantiation of roles necessary for achieving them. When a well-established category for a role doesn’t exist, an ad hoc category is constructed to represent it. Both conceptual and linguistic mechanisms appear central to forming ad hoc categories. Conceptually, people combine existing concepts for objects, events, settings, mental states, properties, and so on to form novel conceptual structures. Linguistically, people combine words in novel ways to index these concepts. Sometimes, novel concepts result from perceiving something novel and then describing it (e.g., seeing a traditional opera set in a modern context and describing this newly encountered genre as “modernized operas”). On other occasions, people combine words for conceptual elements before ever encountering an actual category instance (e.g., describing mezzo sopranos who have power, tone, and flexibility before experiencing one). The conceptual and linguistic mechanisms that formulate ad hoc categories are highly productive, given that components of these categories can be replaced systematically with alternative values from semantic fields (e.g., tourist activities to perform in X, where X could be Rome, Florence, Venice, etc.). Syntactic structures are also central for integrating the conceptual/linguistic components in these categories (e.g., the syntax and accompanying closed class words in tourist activities to perform in Rome). Lawrence Barsalou (1983) introduced the construct of ad hoc categories in experiments showing that these categories are not well established in memory and do not become apparent without context. Once constructed, however, they function as coherent categories, exhibiting internal structures as indexed by typicality gradients. Barsalou (1985) showed that these gradients are organized around ideal values that support goal achievement and also around frequency of instantiation. He also showed (1987) that these internal structures are generally as stable and robust as those in familiar taxonomic categories. Barsalou (1991) offered a theoretical framework for ad hoc categories (see also Barsalou 2003). Within this framework, ad hoc categories provide an interface between roles in knowledge structures (e.g., schemata) and the environment. When a role must be instantiated in order to pursue a goal but knowledge of possible instantiations does not exist, people construct

Ad hoc Categories

Adjacency Pair

an ad hoc category of possible instantiations (e.g., when going camping for the first time, constructing and instantiating activities to perform on a camping trip). The particular instantiations selected reflect their fit with a) ideals that optimize goal achievement and b) constraints from the instantiations of other roles in the knowledge structure (e.g., activities to perform on a camping trip should, ideally, be enjoyable and safe and should depend on constraints such as the vacation location and time of year). Once established, the instantiations of an ad hoc category are encoded into memory and become increasingly well established through frequent use (e.g., establishing touring back roads and socializing around the campground as instances of activities to perform on a camping trip). Barsalou (1999) describes how this framework can be realized within a perceptual symbol system. Specifically, categories (including ad hoc categories) are sets of simulated instances that can instantiate the same space-time region of a larger mental simulation (where a simulation is the reenactment of modality-specific states, as in mental imagery). Ad hoc categories have been studied in a variety of empirical contexts. S. Glucksberg and B. Keysar (1990) proposed that ad hoc categories underlie metaphor (e.g., the metaphor jobs are jails conceptualizes the category of confining jobs). C. J. Cech, E. J. Shoben, and M. Love (1990) found that ad hoc categories are constructed spontaneously during the magnitude comparison task (e.g., forming the ad hoc category of small furniture, such that its largest instances anchor the upper end of the size dimension). F. Vallee-Torangeau, S. H. Anthony, and N. G. Austin (1998) found that people situate taxonomic categories in background settings to form ad hoc categories (e.g., situating fruit to produce fruit in the produce section of a grocery store). E. G. Chrysikou (2006) found that people rapidly organize objects into ad hoc categories that support problem solving (e.g., objects that serve as platforms). Research has also addressed ad hoc categories that become well established in memory, what Barsalou (1985, 1991) termed “goal-derived categories” (also called script categories, slot filler categories, and thematic categories). J. Luciarello and K. Nelson (1985) found that children acquire goal-derived categories associated with scripts before they acquire taxonomic categories (e.g., places to eat). B. H. Ross and G. L. Murphy (1999) examined how taxonomic and goal-derived concepts simultaneously organize foods (e.g., apples as belonging simultaneously to fruit and snack foods). D. L. Medin and colleagues (2006) found that goal-derived categories play central roles in cultural expertise (e.g., tree experts form categories relevant to their work, such as junk trees). Although ad hoc and goal-derived categories are ubiquitous in everyday cognition, they have been the subject of relatively little research. Much further study is needed to understand their structure and role in cognition. Important issues include the following: How do productive conceptual and linguistic mechanisms produce ad hoc categories? How do these categories support goal pursuit during situated action? How do these categories become established in memory through frequent use? How does the acquisition of these categories contribute to expertise in a domain? – Lawrence W. Barsalou

WORKS CITED AND SUGGESTIONS FOR FURTHER READING Barsalou, L. W. 1983. “Ad hoc categories.” Memory & Cognition 11: 211–27. ———. 1985. “Ideals, central tendency, and frequency of instantiation as determinants of graded structure in categories.” Journal of Experimental Psychology: Learning, Memory, and Cognition 11: 629–54. ———. 1987. “The instability of graded structure: Implications for the nature of concepts.” In Concepts and Conceptual Development: Ecological and Intellectual Factors in Categorization, ed. U. Neiser, 101–40. Cambridge: Cambridge University Press. ———. 1991. “Deriving categories to achieve goals.” In The Psychology of Learning and Motivation: Advances in Research and Theory. Vol. 27. Ed. G. Bower, 1–64. San Diego, CA: Academic Press. ———. 1999. “Perceptual symbol systems.” Behavioral and Brain Sciences 22: 577–660. ———. 2003. “Situated simulation in the human conceptual system.” Language and Cognitive Processes 18: 513–62. Cech, C. J., E. J. Shoben, and M. Love. 1990. “Multiple congruity effects in judgments of magnitude.” Journal of Experimental Psychology: Learning, Memory, and Cognition 16: 1142–52. Chrysikou, E. G. 2006. “When shoes become hammers: Goal-derived categorization training enhances problem-solving performance.” Journal of Experimental Psychology: Human Learning and Performance 32: 935–42. Glucksberg, S., and B. Keysar. 1990. “Understanding metaphorical comparisons: Beyond similarity.” Psychological Review 97: 3–18. Lucariello, J., and K. Nelson. 1985. “Slot-filler categories as memory organizers for young children.” Developmental Psychology 21: 272–82. Medin, D. L., N. Ross, S. Atran, D. Cox, J. Coley, J. Proffitt, and S. Blok. 2006. “Folkbiology of freshwater fish.” Cognition 99: 237–73. Ross, B. H., and G. L. Murphy. 1999. “Food for thought: Cross-classification and category organization in a complex real-world domain.” Cognitive Psychology 38: 495–553. Vallée-Tourangeau, F., S. H. Anthony, and N. G. Austin. 1998. “Strategies for generating multiple instances of common and ad hoc categories.” Memory 6: 555–92.

ADJACENCY PAIR conversation analysis, an inductive approach to the microanalysis of conversational data pioneered by Harvey Sacks (1992), attempts to describe the sequential organization of pieces of talk by examining the mechanics of the turn-taking system. Adjacency pairs reflect one of the basic rules for turn-taking (Sacks, Schegloff, and Jefferson 1974), in which a speaker allocates the conversational floor to another participant by uttering the first part of a paired sequence, prompting the latter to provide the second part. Examples are question-answer, greetinggreeting as in (1), and complaint-excuse: (1) A: Hi there B: Oh hi

The constitutive turns in adjacency pairs have the following structural characteristics: (i) They are produced by two different speakers. (ii) They are, as the term suggests, adjacent. This is not a strict requirement, as the two parts can be separated by a so-called insertion sequence, as in (2):

87

Age Groups (2) A: B: A: B:

What’s the time now? (Question 1) Don’t you have a watch? (Question 2) No. (Answer 2) I think it’s around three. (Answer 1)

Static versus Dynamic Theories

(iii) They are organized as a first and a second part, that is, they are nonreversible. This is the case, incidentally, even in ostensibly identical first and second parts, such as the greeting-greeting pair in (1), where reversing the order results in an aberrant sequence. (iv) They are ordered, so that a particular first part requires a relevant second part (e.g., greetings do not follow questions). The fact that the second part is conditionally relevant on the first part does not mean that only one option is available; in fact, certain first parts typically allow for a range of possible second parts. If two (or more) options are possible, one will be the more socially acceptable, preferred response, the other(s) being dispreferred; this phenomenon is known as preference organization, as in: (3) A: Have a piece of cake (first part) B1: Great thanks I will (preferred second) B2: Ehm actually I’ve just eaten but thanks anyway (dispreferred second)

As illustrated in (3), dispreferred second parts tend to be structurally different from preferred seconds (B2 being indirect, including an explanatory account, and containing hesitation markers, unlike B1). Another, related phenomenon that merits mention here is presequencing: Certain adjacency pairs can be introduced or foreshadowed by a preceding exchange, as in: (4) A1: B1: A2: B2:

Do you sell fresh semiskimmed milk? We sure do I’ll have two bottles then please OK

Real-Time Change

This whole exchange forms one unit, in the sense that the occurrence of the question-answer pair A1-B1 is only interpretable given the subsequent request-compliance adjacency pair A2-B2. Phenomena such as this are indicative of a level of sequential organization in conversation beyond two-turn sequencing (see Schegloff 2007). – Ronald Geluykens WORKS CITED AND SUGGESTIONS FOR FURTHER READING Sacks, Harvey. 1992. Lectures on Conversation. Oxford: Blackwell. Sacks, Harvey, Emanuel A. Schegloff, and Gail Jefferson. 1974. “A simplest systematics for the organization of turn-taking in conversation.” Language 50: 696–735. Schegloff, Emanuel A. 2007. The Language of Turn and Sequence. Cambridge: Cambridge University Press.

AGE GROUPS Age is one of the primary independent variables in sociolinguistics, along with social class, sex, ethnicity, and region. Age is the primary social correlate for determining that language is changing and for estimating the rate of change and its progress.

88

Mundane observations make it abundantly clear that language change is not punctual but gradual, and not categorical but variable. Traditional views of language change imposed methodological restrictions to avoid viewing change while it was progressing. Linguists had little confidence in their ability to discern change precisely and accurately amid “the apparent lawlessness of social phenomena,” as Edward Sapir incisively put it (1929, 213). However, in the 1960s, linguists began studying language in its social context. By that time, economics, anthropology, sociology, and other social sciences were well established, and linguistic studies belatedly admitted sociolinguistics, its social science adjunct (Chambers 2002b). Viewing language changes as they progressed entailed the admission of coexisting linguistic entities as data. Linguists were required to study the social distribution of, for example, both sneaked and snuck as variants of the past tense of sneak (as in the example to be discussed). Dealing coherently with variables necessitated determining the distribution of variants with certain social factors, including the age of the speakers. Uriel Weinreich, William Labov, and Marvin I. Herzog, in the document that became the manifesto for viewing language change in its social context, said: “A model of language which accommodates the facts of variable usage and its social and stylistic determinants not only leads to more adequate descriptions of linguistic competence, but also naturally yields a theory of language change that bypasses the fruitless paradoxes with which historical linguistics has been struggling for half a century” (1968, 99). Primary among those paradoxes, of course, were concepts of change as punctual, categorical, and static. By admitting social variables, it became possible to view change as gradual, variable, and dynamic, consistent with commonsense observations.

The study of the progress of change can be carried out in real time by revisiting survey sites at intervals and observing changes in the social distribution of variants from one time to the next. Because changes will not necessarily be completed in the interval but will be continuing, their progress must be calculated by some kind of proportional measure, such as the percentage of the variants, their relative frequency, or their probabilities. The proportional differences from one visit to the next provide a quantitative measure of the progress of the change. Studying change in real time has a number of methodological disadvantages. Most obvious is the time required before attaining a result. Locating subjects on subsequent visits also poses obvious problems because of mobility, cooperation, or death. Furthermore, subsequent visits require the addition of new subjects at the youngest age group each time. This is necessary because of the manner in which linguistic innovations typically diffuse (see diffusion ) throughout the population. Instead of spreading outward from the source and affecting almost everybody in their sphere of influence, as infectious diseases do in epidemics and technological adoptions do when, say, spin dryers replace washboards, linguistic innovations tend to be stratified. Under ordinary circumstances, people acquire their accents and dialects in their formative

Age groups 100 90 80 70 60 50

Figure 1. Percentage of people in different age groups who say snuck, not sneaked, as past tense of sneak in the Golden Horseshoe, Canada (Chambers 2002a, 364–66).

40 30 20 10

years, between 8 and 18, and maintain them throughout their lives. People who grow up saying sneaked as the past tense of sneak tend to use that form all their lives, even after they come to know that younger people in their region say snuck instead. Because of this stratification, the progress of a linguistic change is not measurable in the life span of an individual or one age group but only in comparison between individuals whose formative years are not the same, that is, between different age groups.

Change in Progress Correlating linguistic variants with their use by age groups in the community as the change is taking place is a way of measuring its progress. Inductively, change is evident when a linguistic variant occurs with greater frequency in the speech of younger people than in the speech of their elders. There are some exceptions (such as age grading), but the inference of change can be made with reasonable confidence when the frequency of the variant is stratified from one age group to the next (Labov 2001). That is, linguistic changes are almost never bimodal, with one variant occurring in the speech of younger people and a different one in the speech of older people. Instead, the variants are typically dispersed along the age continuum in a progressive gradation. Figure 1 provides a case study. The variable is the past tense of the verb sneak, with variants sneaked and snuck. The community is the Golden Horseshoe, the densely populated region in southern Ontario, Canada. Figure 1 shows the percentage of people who say snuck, not sneaked, correlated with their age, from octogenarians to teenagers. The correlation shows a progression from 18 percent in the oldest group to 98 percent in the youngest, with gradation in the intermediate age groups (29 percent of 70s, 42 percent of 60s, and so on). Other things being equal, it is possible to draw historical inferences from apparent-time displays like Figure 1. The survey from which the data are drawn took place in 1992. Among people born 80 or more years prior, that is, before 1913, sneaked was the standard variant and snuck had very little

14–19

20–29

30–39

40–49

50–59

60–69

70–79

0ver 80

0

currency. It gained currency steadily thereafter, however, and accelerated most rapidly in the speech of the 50-year-olds, people born in 1933–42, increasing by some 25 percent and becoming the variant used by almost 70 percent of them. In subsequent decades, it was adopted by ever-greater numbers. In the 1980s, the formative years for people born in 1973–8, the teenagers in this survey, snuck virtually eliminated sneaked as a variant. Changes normally take place beneath the level of consciousness. Young people seldom have a sense of the history of the variants. In this case, people under the age of 30 often consider the obsolescent form sneaked to be a “mistake” when they hear someone say it. There is no communal sense that sneaked was the historical standard and accepted form for centuries. Occasionally, changes become self-conscious in the early stages if teachers, writers, or parents openly criticize them. Such criticisms almost never succeed in reversing trends, though they may slow their momentum. When the incoming variant gains enough currency, usually around 20–30 percent, its use accelerates rapidly. It then slows again as it nears completion. The graphic pattern is known as the S-curve in innovation diffusion, with relatively slow (or flat) movement up to 20–30 percent, a rapid rise through the middle stages, and flattening again in the final stages. Lack of communal consciousness of linguistic changes in progress is a consequence of its social stratification (see inequality, linguistic and communicative ). Changes generally progress incrementally, so that differences between the most proximate age groups are small and barely noticeable. In Figure 1, 30-year-olds differ from 40-year-olds by about 10 percent and from 20-year-olds almost not at all. The difference between 30-year-olds and 70-year-olds, by contrast, is over 60 percent. Social relations are closest among age-mates, and the gradation of differences so that proximate age groups are most like one another blunts the perception of generation gaps within the community. By minimizing awareness of changes as they progress, social gradation is a unifying force in communities.

89

Age groups

Apparent-Time Hypothesis Figure 1 draws historical inferences of change based on the behavior of age groups surveyed at the same time. The replacement of sneaked by snuck is not directly observed as it would be in real-time studies, in which researchers go to communities at intervals and track the changes from one time to the next. Instead, the inference of change is based on the assumption that under normal circumstances, people retain the accents and dialects acquired in their formative years. That assumption is known as the apparent-time hypothesis. Common experience tells us that the hypothesis is not without exceptions. People are sometimes aware of expressions that they once used and no longer do, and sometimes they will have changed their usage after their formative years. If such linguistic capriciousness took place throughout the entire community, it would invalidate historical inferences drawn from apparent-time surveys. However, community-wide changes beyond the formative years are rare. Real-time evidence, when available, generally corroborates apparent-time inferences. In the case of sneaked/ snuck, for instance, earlier surveys made in the same region in 1950 and 1972 show proportional distributions of the variants that approximate the apparent-time results. However, inferring linguistic changes from the speech of contemporaneous age groups is not a direct observation of that change. It remains a hypothesis, and its validity must be tested wherever possible.

Age-Graded Changes Deeper understanding of linguistic change in progress should ultimately lead to predictable classes of deviation from the apparent-time hypothesis. One known deviation is age-graded change. These are changes that are repeated in each generation, usually as people reach maturity (Chambers 2009, 200–206). Age-graded changes are usually so gradual as to be almost imperceptible, so that tracking their progress is challenging. As an example, in all English-speaking communities, there is a rule of linguistic etiquette that requires compound subject noun phrases (NPs) to list the first-person pronoun (the speaker) last. Adults say “Robin and I went shopping,” and never say “I and Robin went shopping.” There is no linguistic reason for this rule (that is, the sentences mean the same thing either way), but putting oneself first is considered impolite (see politeness). Children, however, do not know this and tend to say “Me and Robin went shopping.” At some as-yet-undetermined age, children become aware of the rule and change their usage to conform to adult usage. Age-graded changes like these violate the apparent-time hypothesis because the variants used by young people do not persist throughout their lifetimes. Instead, young people change as they reach maturity and bring their usage into line with adults. The occurrence of age-graded changes does not refute the apparent-time hypothesis, but they provide a well-defined exception to it. Failure to recognize them as age-graded can lead to an erroneous inference that change is taking place.

Age and Language Change Introducing age groups into linguistic analysis as an independent variable yielded immediate insights into the understanding

90

Aging and Language of how languages change, who the agents of change are, and how changes diffuse throughout communities. The sociolinguistic perspective on language change as dynamic, progressive, and variable represents an advance in language studies. The principal theoretical construct, the apparent-time hypothesis, provides a comprehensive view of historical sequences from a single methodological vantage point. – J. K. Chambers WORKS CITED AND SUGGESTIONS FOR FURTHER READING Chambers, J. K. 2002a. “Patterns of variation including change.” In The Handbook of Language Variation and Change, ed. J. K. Chambers, Peter Trudgill, and Natalie Schilling-Estes, 349–72. Oxford: Blackwell Publishing. ———. 2002b. “Studying language variation: An informal epistemology.” In The Handbook of Language Variation and Change, ed. J. K. Chambers, Peter Trudgill, and Natalie Schilling-Estes, 3–14. Oxford: Blackwell Publishing. ———. 2009. Sociolinguistic Theory: Linguistic Variation and Its Social Significance. 3rd ed. Oxford: Blackwell. Labov, William. 2001. Principles of Linguistic Change: Social Factors. Oxford: Blackwell Publishing. Sapir, Edward. 1929. “The status of linguistics as a science.” Language 5: 207–14. Weinreich, Uriel, William Labov, and Marvin I. Herzog. 1968. “Empirical foundations for a theory of language change.” In Directions for Historical Linguistics: A Symposium, ed. Winfred P. Lehmann and Yakov Malkiel, 95–188. Austin: University of Texas Press.

AGING AND LANGUAGE It is well-documented that in healthy aging, some aspects of linguistic ability, for example, phonology, syntax, and vocabulary, remain generally preserved into very old age. However, other domains of language, for example, naming, comprehension and spoken discousre, undergo declines, albeit only at very late stages in the adult life span. It is interesting to note that although these linguistic changes occur generally in the older adult population as a group, there are many older adult individuals who do not experience these deficits but, rather, continue to perform as well as younger individuals. Thus, the finding of great inter-individual variability should be considered. This entry presents a detailed review of the three main linguistic domains experiencing decline with age. Additionally, a review of language changes due to cognitive deterioration, for example, dementia, is presented.

Naming A common complaint among the healthy aging population is the increased frequency of word-finding problems in their everyday speech. A subset of these naming problems is often colloquially described as the tip-of-the-tongue (TOT) phenomenon. In terms of cognitive models of lexical access, TOTs are described as a type of word retrieval failure whereby individuals are able to access the conceptual, semantic, syntactic, and even some phonological/orthographic information (e.g., number of syllables or initial sound/letters) of the target word, not enough information, however, to fully phonologically encode the word

Aging and Language for articulation. Research clearly supports the view that there is a general breakdown in phonological encoding (James and Burke 2000). However, since this stage of processing involves various substages (e.g., segment retrieval, syllable encoding) that occur at very fast rates, as per Willem J. M. Levelt, Ardi Roelofs and Antje S. Meyer’s (1999) model, behavioral methods are limited in their ability to identify the locus of processing difficulty. Evidence from priming studies has strongly demonstrated that TOTs are due to a failure in transmission of available semantic information to the phonological system, as explained in the transmission deficit hypothesis (TDH; Burke et al. 1991). This theory proposes that older people are especially prone to word-retrieval problems due to weakened connections at the phonological level. The phonological level, as compared to the semantic level, is particularly vulnerable to breakdowns in retrieval because this level generally has fewer connections (e.g., phoneme-sound), whereas the semantic system has multiple joined connections (e.g. many words/concepts linked to a given word). Factors such as word frequency or recency of use influence the strength of phonological connections; that is, the lower the word frequency and the less recently used the word, the weaker the connections, leading to greater retrieval difficulties. Both younger and older individuals benefit from phonological primes as opposed to unrelated primes during moments of a TOT state, supporting the TDH model. The finding that priming leads to better retrieval is consistent with the claim that priming strengthens the inherently weak phonological connections and thus facilitates resolution of the TOT state. Furthermore, studies using semantic versus phonological cues have demonstrated that in both younger and older people, provision of phonological information was more effective, as a retrieval aid, than semantic cues (Meyer and Bock 1992). This illustrates that in older individuals, semantic information is intact, although there is some degradation in the use of phonological information. In summary, much evidence supports the TDH and the claim that the locus of the breakdown in TOT states is at the phonological stage. The exact phonological substage responsible for this problem still remains unclear; however, there are indications from phonological cuing studies, self-reports of individuals experiencing TOT states (Brown 1991), and an electrophysiological study of lexical retrieval in healthy younger and older adults (Neumann 2007) that the first two substages (segmental and syllabic retrieval) are particular points at which breakdowns occur.

that older adults benefit more from context in noise than do their younger cohorts. Lexical comprehension is one aspect of language that is largely preserved or even enhanced with age. Studies on vocabulary comprehension in older adults using a picture selection task for auditorily presented words show that they are comparable to younger adults in this task (Schneider, Daneman, and PichoraFuller 2002). However, lexical-semantic integration at the level of sentence comprehension may be affected in older adults. Sentence comprehension in the elderly is known to be poor in comparison to younger listeners. A number of reasons, both linguistic and cognitive, have been discussed, including decline in auditory perceptual skills and lexical-semantic and syntactic processing capacity, as well as working memory capacity, speed of processing, and ability to process with competing stimuli and inhibit noise (Wingfield and Stine-Morrow 2000). Studies on syntactic comprehension of language in older adults demonstrate that they are slower at judging sentences that are syntactically complex and semantically improbable (Obler et al. 1991). It has also been found to be relatively more difficult for this population than for younger adults to take advantage of constraining context in a sentence. This leads to difficulties in situations where older adults need to rely on context but are not able to do so skillfully. However, in other aspects of sentence comprehension such as disambiguation, no age-related differences has been reported. It is apparent that sentence comprehension is largely mediated by the ability to hold these sentences in working memory while they are being processed. Working-memory decline has been reported in older adults (Grossman et al. 2002). Moreover, syntactic-processing difficulties have also been attributed to reduction in working-memory capacity (Kemptes and Kemper 1997). Executive functions such as inhibition and task switching have been reported as negatively affected in the elderly. Comprehension problems eventually affect the older adult’s discourse abilities. Often, the elderly find that it gets harder to follow discourse with advancing age. Studies have shown that older adults are significantly poorer than younger adults in fully understanding and inferring complex discourse, and this difficulty is enhanced further with increased perceptual or cognitive load (Schneider, Daneman, and Pichora-Fuller 2002), such as noise and length or complexity of the material. A combination of general cognitive decline and a deterioration of specific linguistic and sensory-perceptual processes contribute to the general slowing observed in the elderly while they engage in conversation and discourse.

Language Comprehension It is relatively well known that comprehension in older adults is compromised in comparison to younger adults. These problems in comprehension may arise from individual or combined effects of decline in their sensory/perceptual, linguistic, or cognitive domains. Current research is focused on disambiguating the effects of these processes on phonological, lexical-semantic, and syntactic aspects of language decline in the elderly. Research shows that speech perception in older adults is especially affected by noise. This means that even with normal hearing, older adults experience difficulties understanding speech (e.g., sentences) under noisy conditions such as a cocktail party or cafeteria. Experiments focusing on sentence perception show

Spoken Discourse Results from many types of tasks requiring sentence production, such as cartoon description tasks (Marini et al. 2005), have indicated that older adults tend to produce grammatically simpler, less informative and more fragmented sentences than do younger adults. Middle-aged and young elderly adults tend to be better and more efficient story constructors than younger or older adults. However, older adults usually produce a larger number of words in their narrative speech, but they can show difficulties in morpho-syntactic and lexico-semantic processing by making more paragrammatic errors and semantic paraphasias than younger adults.

91

Aging and Language Older adults also show a decreased ability to coherently link adjacent utterances in a story. Adults older than 75 use a larger number of missing or ambiguous referents and more units of irrelevant content that affect the coherence of the narratives (Marini et al. 2005). Older people show a huge variation in storytelling abilities, and they can also compensate for their storytelling difficulties due to their greater accumulated life experience, which they can use to combine different themes and to emphasize relevant details. In spoken discourse, some changes in conversational-interaction style can occur in young elderly people (60–74 years), but the most noticeable changes are likely to take place in older elderly people (77–88 years), who show excessive verbosity, failure to maintain topic, poor turn-taking, and unclear referencing (James et al. 1998). Difficulties in grammatical processing in the aging can be attributed to cognitive deterioration involving reduced working memory (Kemper, Herman, and Liu 2004) and inhibitory deficits (James et al. 1998), but they can also be a sign of impaired access to the lemma level during lexical retrieval (Marini et al. 2005). In summary, there are clear differences between younger and older adults in their sentence production, storytelling, and conversational abilities. These differences are manifest as changes in morpho-syntactic and lexico-semantic processing, excessive verbosity, and reduced informativeness and coherence.

Dementia Difficulties in language production, as well as comprehension, become obvious in dementia. Word-finding difficulties are characteristic of mild cognitive impairment (MCI), Alzheimer’s disease (AD), and vascular dementia (VaD). Arto Nordlund and colleagues (2004) indicated that 57.1 percent of the individuals with MCI had significantly lower scores in different language tasks than did typical aging adults. In AD, language-processing difficulties are early signs of the disease. In particular, AD appears to cause a breakdown of the semantic domain of language, which can be reflected in the impaired comprehension and use of semantic relations between words (Garrard et al. 2001) and in the reduced semantic noun and verb fluency performance (Pekkala 2004) in this population. Comparative studies between AD and VaD have indicated that cognitive and linguistic performance cannot clearly differentiate the two types of dementia from each other. Elina Vuorinen, Matti Laine, and Juha Rinne (2000) indicated that both AD and VaD involved similar types of semantic deficits early in the disease, including difficulties in comprehension, naming, and production of semantic topics in narrative speech, while word repetition, oral reading, and fluency of speech output were preserved in both types of dementia. – Yael Neumann, Seija Pekkala, and Hia Datta WORKS CITED AND SUGGESTIONS FOR FURTHER READING Brown, Alan S. 1991. “A review of the tip-of-the-tongue experience.” Psychological Bulletin 109.2: 204–23. Burke, Deborah M., Donald G. MacKay, Joanna S. Worthley, and Elizabeth Wade. 1991. “On the tip of the tongue: What causes word finding failures in young and older adults?” Journal of Memory and Language 30: 542–79.

92

Garrard, Peter, Matthew A. Lambon Ralph, John R. Hodges, and Karalyn Patterson. 2001. “Prototypicality, distinctiveness, and intercorrelations: Analysis of the semantic attributes of living and nonliving concepts.” Cognitive Neuropsychology 18.2: 125–74. Grossman, Murray, Ayanna Cooke, Christian DeVita, David Alsop, John Detre, Willis Chen, and James Gee. 2002. “Age-related changes in working memory during sentence comprehension: An fMRI study.” NeuroImage 15: 302–17. James, Lori E., and Deborah M. Burke. 2000. “Phonological priming effects on word retrieval and tip-of-the-tongue experiences in young and older adults.” Journal of Experimental Psychology: Learning, Memory, and Cognition 26.6: 1378–91. James, Lori E., Deborah M. Burke, Ayda Austin, and Erika Hulme. 1998. “Production and perception of ‘verbosity’ in younger and older adults.” Psychology and Aging 13: 355–67. Kemper, Susan, Ruth E. Herman, and Chiung-Ju Liu. 2004. “Sentence production by young and older adults in controlled contexts.” Journal of Gerontology Series B: Psychological Sciences and Social Sciences 59: 220–4. Kemptes, Karen A., and Susan Kemper 1997. “Younger and older adults’ on-line processing of syntactically ambiguous sentences.” Psychology and Aging 12: 362–71. Levelt, Willem J. M., Ardi Roelofs, and Antje S. Meyer 1999. “A theory of lexical access in speech production.” Behavioral Brain Sciences 22: 1–75. Marini, Andrea, Anke Boewe, Carlo Caltagirone, and Sergio Carlomagno. 2005. “Age-related differences in the production of textual descriptions.” Journal of Psycholinguistic Research 34: 439–63. Meyer, Anje S., and Kathryn Bock. 1992. “The tip-of-the-tongue phenomenon: Blocking or partial activation?” Memory and Cognition 20: 715–26. Morris, John, Martha Storandt, J. Phillip Miller, Daniel W. McKeel, Joseph L. Price, Eugene H. Rubin, and Leonard Berg. 2001. “Mild cognitive impairment represents early-stage Alzheimer’s disease.” Archives of Neurology 58.3: 397–405. Neumann, Yael. 2007. An Electrophysiological Investigation of the Effects of Age on the Time Course of Segmental and Syllabic Encoding during Implicit Picture Naming in Healthy Younger and Older Adults. Publications of the Department of Speech, Language, and Hearing Sciences. New York: City University of New York. Nordlund, Arto, S. Rolstad, P. Hellström, M. Sjögren, S. Hansen, and A. Wallin. 2004. “The Goteborg MCI study: Mild cognitive impairment is a heterogeneous condition.” Journal of Neurology, Neurosurgery, and Psychiatry 76: 1485–90. Obler, Loraine K., Deborah Fein, Marjorie Nicholas, and Martin L. Albert 1991. “Auditory comprehension and aging: Decline in syntactic processing.” Applied Psycholinguistics 12: 433–52. Pekkala, Seija. 2004. Semantic Fluency in Mild and Moderate Alzheimer’s Disease. Publications of the Department of Phonetics 47, University of Helsinki. Available online at: http://ethesis.helsinki.fi/. Reuter-Lorenz, Patricia A., John Jonides, Edward E. Smith, Alan Hartlye, Andrea Miller, Christina Marshuetz, and Robert A. Koeppe. 2000. “Age differences in the frontal lateralization of verbal and spatial working memory revealed by PET.” Journal of Cognitive Neuroscience 12: 174–87. Schneider, Bruce A., Meredith Daneman, and M. Kathleen PichoraFuller. 2002. “Listening in aging adults: From discourse comprehension to psychoacoustics.” Canadian Journal of Experimental Psychology 56: 139–52. Vuorinen, Elina, Matti Laine, and Juha Rinne. 2000. “Common pattern of language impairment in vascular dementia and in Alzheimer’s disease.” Alzheimer Disease and Associated Disorders 14: 81–6. Wingfield, Arthur, and Elizabeth A. L. Stine-Morrow. 2000. “Language and aging.” In The Handbook of Cognition and Aging, ed. Craig and Salthouse, 359–416. Mahwah, NJ: Psychology Press.

Agreement

Agreement Maximization

AGREEMENT

WORKS CITED AND SUGGESTIONS FOR FURTHER READING

Agreement is a form of featural dependency between different parts of a sentence: The morphological shape of a word is a function of particular morphological features of a different, often distant, word. Since the Middle Ages, agreement was taken to be in complementary distribution with government, and, hence, it became important to determine both the context of each type of relation and the reasons why this difference exists (Covington 1984). This rich tradition has survived in generative grammar, all the way up to the minimalist program (see mimimalism), where it is embodied under the notion agree. Depending on the features and occurrences in an expression, agreement can be characterized as external or internal (Barlow and Fergusson 1988). External agreement typically involves person and number features, taking place between verbs and corresponding dependents. We can witness it in you are friendly versus he is friendly. Internal agreement (concord) normally involves gender and number features, and typically takes place internal to nominal expressions, between adjectives or relative clauses and the head noun, freely iterating. Concord is easily observed in modified nominal expressions in Spanish: atractivas damas “attractive ladies” versus atractivos caballeros “attractive gentlemen” (agreeing elements are boldfaced). Genitive agreement internal to nominal expressions normally falls within the external (not the internal) rubric. The principles and parameters system concentrated on external agreement, through the relation (head, specifier) (Aoun and Sportiche 1983). However, since agreement is possible also in situations where no such relation seems relevant, the minimalist program (Chomsky 2000) proposes a relation between a probe and a goal. The probe contains a value-less attribute in need of valuation from a distant feature of the same type, which the probing mechanism achieves. The goal cannot be contained within a derivational cycle (a phase) that is too distant from the probe. When the probe finds an identical category within its complement domain, it attempts to get its valuation from it, thereby sanctioning the relevant agreement. To illustrate, observe the Spanish example in (1); note also the internal agreement manifested within the noun phrase:

Aoun, J. and D. Sportiche. 1983. “On the formal theory of government.” Linguistic Review 2: 211–36. Barlow, M., and C. Barlow, eds. 1988. Agreement in Natural Language. Stanford, CA: CSLI Publications. Boeckx, C., ed. 2006. Agreement Systems, Amsterdam: Benjamins. Chomsky, N. 2000. “Minimalist inquiries: the framework.” In Step by Step, ed. R. Martin, D. Michaels, and J. Uriagereka, 89–155. Cambridge, MA: MIT Press Covington, M. 1984 Syntactic Theory in the High Middle Ages, Cambridge: Cambridge University Press.

(1) Parecen [haber quedado [los locos soldados] en la guarnición] seem-3rd/pl. have remained the-m./pl. crazy-m./pl. soldiers-m./pl. in the garrison (2) Probe1 [… [Goal…]…]] (plus iteration of -os within the nominal)

( ) person

pl. number

( ) number

m. gender

Agreement adds a strange redundancy to the language faculty. In languages where the phenomenon is overt, the extra layer of manifest dependency correlates with differing surface orders. But it is unclear whether that justifies the linguistic emergence of the agreement phenomenon, particularly since the probe/goal mechanism can be present without overt manifestations. This results in much observed variation, from the almost total lack of overt agreement of Chinese to the poly-personal manifestation of Basque. – Juan Uriagereka

AGREEMENT MAXIMIZATION Maximizing, or optimizing, agreement between speaker and interpreter is part of the principle of charity (see charity, principle of) that, according to philosophers of language in the tradition of W. V. O. Quine and Donald Davidson, governs the interpretation of the speech and thought of others and guides the radical interpretation of a radically foreign language. According to Davidson, correct interpretation maximizes truth and coherence across the whole of the beliefs of a speaker. For an interpreter, maximizing truth across the beliefs he ascribes to a speaker necessarily amounts to maximizing agreement between himself and the speaker: He can only go by his own view of what is true. Take one of Davidson’s own examples: Someone says “There is a hippopotamus in the refrigerator,” and he continues: “It’s roundish, has a wrinkled skin, does not mind being touched. It has a pleasant taste, at least the juice, and it costs a dime. I squeeze two or three for breakfast” (1968, 100). The simplest way of maximizing agreement between this speaker and us will probably be interpreting his expression “hippopotamus” as meaning the same as our expression “orange.” Davidson himself, however, soon came to consider maximizing agreement as “a confused ideal” (1984, xvii) to be substituted with “optimizing agreement” (1975, 169). The idea here is that “some disagreements are more destructive of understanding than others” (1975, 169). Very generally, this is a matter of epistemic weight; the more basic a belief is, and the better reasons we have for holding it, the more destructive disagreement on it would be: “The methodology of interpretation is, in this respect, nothing but epistemology seen in the mirror of meaning” (1975, 169). According to Davidson, it is impossible to codify our epistemology in simple and precise form, but general principles can be given: “[A]greement on laws and regularities usually matters more than agreement on cases; agreement on what is openly and publicly observable is more to be favored than agreement on what is hidden, inferred, or ill observed; evidential relations should be preserved the more they verge on being constitutive of meaning” ([1980] 2004, 157). Agreement optimization does not exclude the possibility of error; speakers are to be interpreted as right only “when plausibly possible” (Davidson 1973, 137). In certain situations this prevents the interpreter from ascribing beliefs of his own to the speaker, for instance, perceptual beliefs about objects the speaker is in no position to perceive. Moreover, if the speaker has other beliefs that provide him or her with very good reasons for believing something false, optimizing agreement across all

93

Agreement Maximization

Alliteration

of his or her beliefs might well require ascription of outright mistakes. Optimizing agreement provides an interpreter with a method for arriving at correct interpretations because of the way belief content is determined, Davidson holds. The arguments for this claim have changed over the years; initially, the idea was that “a belief is identified by its location in a pattern of beliefs; it is this pattern that determines the subject matter of the belief, what the belief is about” (1975, 168). Later, however, the role played by causal connections between objects and events in the world and the beliefs of speaker and interpreter becomes more and more prominent: In the most basic, perceptual cases, “the interpreter interprets sentences held true (which is not to be distinguished from attributing beliefs) according to the events and objects in the outside world that cause the sentence to be held true” ([1983] 2001,150). In the later Davidson, the account of content determination underlying the method of charitable interpretation takes the form of a distinctive, social and perceptual meaning externalism: In the most basic cases, the objects of thought are determined in a sort of triangulation as the shared causes of the thoughts of two interacting persons, for instance, a child and its teacher (cf. Davidson [1991] 2001, 2001). According to Davidson, such triangulation is a necessary condition for thought with empirical content; moreover, he derives a quite far-reaching epistemic antiskepticism from it and claims that “belief is in its nature veridical” ([1983] 2001, 146; cf. also [1991] 2001, 211ff). Probably the most influential argument against the idea that any kind of maximizing agreement results in correct interpretation derives from Saul Kripke’s attack on description theories of proper names. According to such theories, the referent of a proper name, for instance, “Gödel” is determined by a description, or cluster of descriptions, held true by the speaker(s), for instance, the description “the discoverer of the incompleteness of arithmetic.” Kripke argued, among other things, that such theories fail because all of the relevant descriptions, all of the relevant beliefs that a speaker, or even a group of speakers, holds about Gödel could turn out to be false (cf. 1972, 83ff). Kripke gave an analogous argument for natural-kind terms such as gold or tiger, and many philosophers today believe that these arguments can be generalized even further. While it is quite clear, however, that most of the descriptions a speaker associates, for instance, with a name could turn out to be false when taken one by one, it is far less obvious that all (or most, or a weighted majority) of them could do so at the same time. According to Davidson, for instance, optimizing agreement amounts to reference determination by epistemically weighted beliefs. A significant number of these are very elementary beliefs like the belief that Gödel was a man, that he was human, that he worked on logic, that he lived on Earth, and so on. If a speaker did not believe any of these things about Gödel, it has been argued with Davidson, it is far less clear that this speaker was in fact talking about Gödel: “Too much mistake simply blurs the focus” (Davidson 1975, 168). – Kathrin Glüer WORKS CITED AND SUGGESTIONS FOR FURTHER READING Davidson, Donald. 1968. “On saying that.” In Davidson 1984, 93–108. ———. 1973. “Radical interpretation.” In Davidson 1984, 125–39.

94

———. 1975. “Thought and talk.” In Davidson 1984, 155–70. ———. [1980] 2004. “A unified theory of thought, meaning, and action.” Problems of Rationality, 151–66. Oxford: Clarendon Press. ———. [1983] 2001. “A coherence theory of truth and knowledge.” Subjective, Intersubjective, Objective, 137–53. Oxford: Clarendon Press. ———. 1984. Inquiries into Truth and Interpretation. Oxford: Clarendon Press. ———. [1991] 2001. “Three varieties of knowledge.” Subjective, Intersubjective, Objective, 205–20. Oxford: Clarendon Press. ———. 2001. “Externalisms.” Interpreting Davidson, ed. P. Kotatko, P. Pagin, and G. Segal, 1–16. Stanford, CA: CSLI. Glüer, Kathrin. 2006. “Triangulation.” The Oxford Handbook of Philosophy of Language, ed. E. Lepore and B. Smith, 1006–19. Oxford: Oxford University Press. Grandy, Richard. 1973. “Reference, meaning, and belief.” Journal of Philosophy 70: 439–52. Kripke, Saul. 1972. Naming and Necessity. Cambridge: Harvard University Press.

ALLITERATION Linguistically, alliteration, also known as initial or head rhyme, is defined as the selection of identical syllable onsets within a specific phonological, morphosyntactic, or metrical domain. It is usually coupled with stress, as in “the three Rs in education: reading, writing, and arithmetic.” Etymologically, the term alliteration (from L. ad- “to” + littera “letter”) includes the repetition of the same letters at word beginnings; its dual association with sounds and letters reflects a common cognitive crisscrossing between spoken and written language in highly literate (see literacy) societies well illustrated by the famous phrase “apt alliteration’s artful aid,” where the alliteration is primarily orthographic. Alliteration based on the sameness of letters is found in visual poetry, advertising, and any form of playful written language. Phonologically based alliteration is a frequent mnemonic and cohesive device in all forms of imaginative language: Examples from English include idioms (beat about the bush), reduplicative word-formation (riffraff), binominals (slowly but surely), catch phrases, refrains, political slogans, proverbs, and clichés. In verse, alliteration serves both as ornamentation and as a structural device highlighting the metrical organization into feet, cola, verses, and lines. Along with rhyme, alliteration is a common feature of folk and art verse in languages as diverse as Irish, Shona, Mongolian, Finnish, and Somali. The most frequent type of alliteration requires identity of the onsets of stressed syllables, which makes it a preferred poetic device in languages with word-initial stress, such as the older Germanic languages. Within the Germanic tradition, metrically relevant alliteration occurs on the stressed syllables of the first foot of each verse (or half-line), where it is obligatory. For Old English, the language of the richest and most varied surviving alliterative poetry in Germanic, the second foot of the first half-line may also alliterate. Alliteration is disallowed on the last stressed syllable in the line. In Old English verse, alliteration appears with remarkable regularity: Only 0.001% of the verses lack alliteration and less than 0.05% contain unmetrical alliteration (Hutcheson 1995, 169). Alliteration is, therefore, a reliable criterion used by

Ambiguity modern editors to determine the boundaries of verses and lines, though no such divisions exist in the manuscripts. The reinvented alliterative tradition of fourteenth-century England also uses alliteration structurally, while its ornamental function is enhanced by excessive verse-internal and run-on alliteration. As a cohesive device in verse, alliteration refers to the underlying distinctions in the language and relies on identity of phonological categories. The interpretation of onset identity for the purpose of poetic alliteration varies from tradition to tradition and can include whole clusters, optionally realized segments, and even the whole syllable up to the coda. In Germanic, all consonants alliterated only with one another, the clusters st-, sp-, sk- could not be split, and all orthographic stressed vowels alliterated freely among themselves, most likely because their identity was signaled by the presence of a prevocalic glottal stop in stressed syllable onsets. – Donka Minkova WORKS CITED AND SUGGESTIONS FOR FURTHER READING Fabb, Nigel. 1997. Linguistics and Literature: Language in the Verbal Arts in the World. Oxford: Blackwell. Hutcheson, Bellenden Rand. 1995. Old English Poetic Metre. Cambridge: D. S. Brewer. Minkova, Donka. 2003. Alliteration and Sound Change in Early English. Cambridge: Cambridge University Press.

AMBIGUITY Ambiguity refers to the potential of a linguistic expression to have more than one meaning. Although many expressions (words, phrases, and even sentences) are ambiguous in isolation, few remain so when used in a particular context. In fact, people typically resolve all ambiguities without even detecting the potential for other interpretations. Ambiguity does not imply vagueness; rather, ambiguity gives rise to competing interpretations, each of which can be perfectly concrete. Although ambiguity is pervasive and unavoidable in natural languages, artificial languages developed for mathematics, logic, and computer programming strive to eliminate it from their expressions. Ambiguity can be lexical, structural, referential, scopal, or phonetic. The examples of these phenomena that follow include well-known classics in English. Lexical ambiguity refers to the fact that some words, as written or spoken, can be used in different parts of speech (see word classes) and/or with different meanings. For example, duck can be used as a noun or a verb and, as a noun, can refer to a live animal or its meat. Structural ambiguity arises when different syntactic parses give rise to different interpretations. For example, in addition to being lexically ambiguous, They saw her duck is also structurally ambiguous: 1. They saw [NP her duck] (the bird or its meat belongs to her) 2. They saw [NP her] [VP duck] (the ducking is an action she carries out)

A common source of structural ambiguity involves the attachment site for prepositional phrases, which can be at the level of the nearest noun phrase (NP) or the clause. In the sentence

Danny saw the man with the telescope, either Danny used the telescope to help him see the man (3) or the man whom Danny saw had a telescope (4). 3. Danny [VP saw [NP the man] [PP with the telescope]] 4. Danny [VP saw [NP the man [PP with the telescope]]]

Referential ambiguity occurs when it is not clear which entity in a context is being referred to by the given linguistic expression. Although deictics (see deixis), such as pronouns, are typical sources of referential ambiguity, full noun phrases and proper nouns can also give rise to it. 5. (at a boy’s soccer game) “He kicked him!” “Who kicked who?” 6. (at a boat race) “That boat seems to be pulling ahead.” “Which one?” 7. (in a university corridor) “I”m off to meet with Dr. Sullivan.” “Chemistry or math?” (There are two Dr. Sullivans in different departments.)

Scopal ambiguity occurs when a sentence contains more than one quantified NP and the interpretation depends on the relative scopes of the quantifiers. For example, Some children saw both plays can mean that a) there exist some children such that each of them saw both plays or b) both plays were such that each, individually, was seen by some children but not necessarily the same children. Phonetic ambiguity arises when a given sound pattern can convey different words, for example, two ~ too ~ to; new deal ~ nude eel. Although people typically do not notice the ambiguities that they effortlessly resolve through context, they are certainly aware of the potential for ambiguity in language. In fact, such awareness is a precondition for getting the joke in Abbott and Costello’s “Who’s on First?” skit or in headlines like “Iraqi Head Seeks Arms.” Whereas ambiguity does not frequently hinder effective communication among people, it is among the biggest hurdles for the machine processing of language. This is not surprising if one considers how much reasoning is required to resolve ambiguity and how much knowledge of language, the context, and the world must underpin such reasoning. As an example of the large scale of the task, consider the short sentence The coach lost a set, which you probably interpreted to mean “the person who is the trainer of some athletic team experienced the loss of a part of a match in an athletic competition” (whether the coach was playing or the team was playing is yet another ambiguity). Other interpretations are also valid, given specific contexts. For example, the person who is the trainer of some team might have lost a set of objects (keys, golf clubs) or a railroad car might have lost a set of objects (door handles, ball bearings). If this sentence were used as input to an English-Russian machine translation system that relied on a standard English-Russian dictionary, that system would have to select from among 15 senses of coach, 11 senses of lose, and 91 senses of set – a grand total of 15,015 combinations, if no further knowledge were brought to bear. Of course, all machine translation systems incorporate some heuristic knowledge, and lexicons developed for natural language processing typically do not permit the amount of sense splitting found in dictionaries for people. On the other hand, it is common

95

Amygdala for sentences to contain upward of 20 words, in which case there is still the threat of combinatorial explosion. – Marjorie J. McShane WORKS CITED AND SUGGESTIONS FOR FURTHER READING Cruse, D. A. 1986. Lexical Semantics. Cambridge: Cambridge University Press. Includes features of and tests for ambiguity. Small, Steven, Garrison Cottrell, and Michael Tanenhaus, eds. 1988. Lexical Ambiguity Resolution: Perspective from Psycholinguistics, Neuropsychology and Artificial Intelligence. San Mateo, CA: Morgan Kaufmann. Zwicky, Arnold M., and Jerrold M. Sadock. 1975. “Ambiguity Tests and How to Fail Them.” In Syntax and Semantics, ed. J. Kimball, IV: 1–36. New York: Academic Press. Discusses tests to distinguish ambiguity from lack of specification.

AMYGDALA Studies in animals have established a clear role for the amygdala in social and emotional behavior, especially as related to fear and aggression (Le Doux 1996). Human studies, including lesion studies, electrophysiology, and functional neuroimaging, have further elucidated the role of the amygdala in the processing of a variety of emotional sensory stimuli, as well as its relationship to behavioral and cognitive responses (Adolphs 2001). These responses not only guide social behavior but also aid in the acquisition of social knowledge. The focus of this entry is on the amygdala and its role in the processing of language, in particular language relevant to social and emotional behavior (see also emotion and language and emotion words). The amygdala is an almond-shaped group of neurons located in the rostral medial temporal region on both left and right sides of the brain (see left hemisphere language processing and right hemisphere language processing). It has reciprocal connections to regions, such as the hypothalamus, that are important for coordinating autonomic responses to complex environmental cues for survival, as well as premotor and prefrontal areas that are necessary for rapid motor and behavioral responses to perceived threat. Visual, somatosensory, and auditory information is transmitted to the amygdala by a series of indirect, modality-specific thalamocorticoamygdalar pathways, as well as by direct thalamoamygdalar pathways. Within the amygdaloid complex, information processing takes place along numerous highly organized parallel pathways with extensive intraamygdaloid connections. The convergence of inputs in the lateral nucleus enables stimulus representations to be summated. Specific output pathways from the central nucleus and amygdalohippocampal area mediate complementary aspects of learning and behavioral expressions connected with various emotional states. The amydgala is thus well positioned to play a role in rapid cross-modal emotional recognition. It is important for the processing of emotional memory and for fear conditioning. In addition, anatomical studies of the primate amygdala demonstrate connections to virtually all levels of visual processing in the occipital and temporal cortex (Amaral 2003). Therefore, the amygdala is also critically placed to modulate visual input, based on affective significance, at a variety of levels along the

96

cortical visual processing stream. Hence, through its extensive connectivity with sensory processing regions, the amygdala is ideally located to influence perception based on emotion.

Language In order to survive in a changing environment, it is especially important for the organism to remember events and stimuli that are linked with emotional consequences. Furthermore, it is important to be vigilant of emotional stimuli in the environment in order to allow for rapid evaluation of and response to these emotional stimuli. In humans, emotional cues are transmitted linguistically, as well as through body posture, voice, and facial expression (see gesture). In the first imaging study to examine language and the amygdala, a modified Stroop task was utilized, along with a high-sensitivity neuroimaging technique, to target the neural substrate engaged specifically when processing linguistic threat (Isenberg et al. 1999). Healthy volunteer subjects were instructed to name the color of words of either threat or neutral valence, presented in different color fonts, while neural activity was measured by using positron emission tomography. Bilateral amygdalar activation was significantly greater during color naming of threat words than during color naming of neutral words (see Color Plate 1). Associated activations were also noted in sensory-evaluative and motor-planning areas of the brain. Thus, our results demonstrate the amygdala’s role in the processing of danger elicited by language. In addition, the results reinforce the amygdala’s role in the modulation of the perception of, and response to, emotionally salient stimuli. This initial study further suggests conservation of phylogenetically older mechanisms of emotional evaluation in the context of more recently evolved linguistic function. In a more recent study that examines the neural substrates involved when subjects are exposed to an event that is verbally linked to an aversive outcome, activation is observed in the left amygdala (Phelps et al. 2001, 437–41). This activation correlated with the expression of the fear response as measured by skin conductance response, a peripheral measure of arousal. The laterality of response may relate to the explicit nature of the fear, as well as to the fact that the stimulus is learned through verbal communication. Fears that are simply imagined and anticipated nonetheless have a profound impact on everyday behavior. The previous study suggests that the left amygdala is involved in the expression of fear when anticipated and conveyed in language. Another study that sought to examine the role of the amygdala in the processing of positive as well as negative valence verbal stimuli also demonstrated activity in the left amygdala (Adolphs, Baron-Cohen, and Tranel 2002, 1264–74). During magnetic resonance (MR) scanning, subjects viewed high-arousal positive and negative words and neutral words. In this study, activity was found in the left amygdala while subjects viewed both negative and positive words in comparison to neutral words. Taken together, these studies suggest that the amygdala plays a role in both positive and negative emotional responses. Furthermore, they suggest that the left amygdala may be preferentially involved in the processing of emotion as conveyed through language. Lesion studies have generally suggested that the amygdala is not essential for recognizing or judging emotional and social

Analogy information from explicit, lexical stimuli, such as stories (see

narrative, neurobiology of) (Amaral 2003, 337–47). However, while recognition of emotional and social information may be relatively preserved in amygdalar damage, the awareness that unpleasant emotions are arousing appears to be lost. In a lesion study, normal subjects judge emotions such as fear and anger to be both unpleasant and highly arousing; however, patient S. M. 046, who sustained early amygdalar damage, judged these same stimuli to be unpleasant but of low arousal. For example, when told a story about someone driving down a steep mountain who had lost the car brakes, she identified the situation as unpleasant but also gave a highly abnormal judgment that it would make one feel sleepy and relaxed. It is interesting to note that S. M. 046 was able to judge arousal normally from positive emotions. The human amygdala is important both for the acquisition and for the online processing of emotional stimuli. Its role is disproportionate for a particular category of emotional information, that pertaining to the evaluation of potential threat in the environment. The amygdala’s role in enhanced, modality-specific processing required for the rapid evaluation and response to threat is highlighted. Furthermore, this review suggests conservation of phylogenetically older limbic mechanisms of emotional evaluation in the context of more recently evolved language. – Nancy B. Isenberg WORKS CITED AND SUGGESTIONS FOR FURTHER READING Adolphs, R. 2001. “The neurobiology of social cognition.” Current Opinions in Neurobiology 11.2: 231–9. Adolphs, R., S. Baron-Cohen, and D. Tranel. 2002. “Impaired recognition of social emotions following amygdala damage.” Journal of Cognitive Neuroscience 14.8: 1264–74. Amaral, D. G. 2003. “The amygdala, social behavior, and danger detection.” Annals of the New York Academy of Sciences 1000 (Dec.): 337–47. Freese, J. L., and D. G. Amaral. 2006. “Synaptic organization of projections from the amygdala to visual cortical areas TE and V1 in the macaque monkey.” Journal of Comparative Neurology 496.5: 655–67. Isenberg, N., D. Silbersweig, A. Engelien, S. Emmerich, K. Malavade, B. Beattie, A. C. Leon, and E. Stern. 1999. “Linguistic threat activates the human amygdala.” Proceedings of the National Academy of Sciences USA 96.18: 10456–9. Le Doux, J. 1996. The Emotional Brain: The Mysterious Underpinnings of Emotional Life. New York: Touchstone. Phelps, E. A. 2006. “Emotion and cognition: Insights from studies of the human amygdala.” Annual Review of Psychology 57: 27–53. Phelps, E. A., K. J. O’Connor, J. C. Gatenby, J. C. Gore, C. Grillon, and M. Davis. 2001. “Activation of the left amygdala to a cognitive representation of fear.” Nature Neuroscience 4.4: 437–41.

ANALOGY Two situations are analogous if they share a common pattern of relationships among their constituent elements, even though the elements are dissimilar. Often one analog, the source, is more familiar or better understood than the second analog, the target (see source and target). Typically, a target situation serves as a retrieval cue for a potentially useful source analog. A mapping, or set of systematic correspondences aligning elements of the source and target, is then established. On the basis of the mapping, it is possible to derive new inferences about the target.

In the aftermath of analogical reasoning about a pair of cases, some form of relational generalization may take place, yielding a schema for a class of situations (Gick and Holyoak 1983).

Psychological Research Within psychology, work in the intelligence tradition focused on four-term or proportional analogies, such as ARM: HAND :: LEG: ? Charles Spearman (1946) reviewed studies that found high correlations between performance in solving analogy problems and the g factor (general intelligence). The ability to solve analogylike problems depends on a neural substrate that includes subareas of the prefrontal cortex (Bunge, Wendelken, and Wagner 2005; see frontal lobe). Although there have been reports of great apes being successfully trained to solve analogy problems, these results are controversial (Oden, Thompson, and Premack 2001). Complex relational thinking appears to be a capacity that emerged in homo sapiens along with the evolutionary increase in size of the frontal cortex. The ability to think relationally increases with age (Gentner and Rattermann 1991). Greater sensitivity to relations appears to arise with age due to a combination of incremental accretion of knowledge about relational concepts (Goswami 1992), increases in working memory capacity (Halford 1993), and increased ability to inhibit misleading featural information (Richland, Morrison, and Holyoak 2006). Analogy plays a prominent role in teaching mathematics (Richland, Zur, and Holyoak 2007). Dedre Gentner (1983) developed the structure-mapping theory of analogy, emphasizing that analogical mapping is guided by higher-order relations – relations between relations. Keith Holyoak and P. Thagard (1989) proposed a multiconstraint theory, hypothesizing that people find mappings that maximize similarity of corresponding elements and relations, structural parallelism, and pragmatic importance for goal achievement. Several computational models of human analogical thinking have been developed. Two influential models are SME (Structure Mapping Engine; Falkenhainer, Forbus, and Gentner 1989), based on a classical symbolic architecture, and LISA (Learning and Inference with Schemas and Analogies; Hummel and Holyoak 2005), based on a neural-network architecture. LISA has been used to simulate some effects of damage to the frontal and temporal cortex on analogical reasoning (Morrison et al. 2004).

Analogy and Language Analogy is related to metaphor and similar forms of symbolic expression in literature and everyday language. In metaphors, the source and target domains are always semantically distant (Gentner, Falkenhainer, and Skorstad 1988). Rather than simply comparing the source and target, the target is identified with the source (Holyoak 1982), either directly (e.g., “Juliet is the sun”) or by applying a predicate drawn from the source domain to the target (e.g., “The romance blossomed”). As a domain-general learning mechanism linked to human evolution, analogy offers an alternative to strongly nativist views of language acquisition (Vallauri 2004; see innateness and innatism). Gentner and L. L. Namy (2004) review evidence that analogical comparison plays important roles in speech segmentation, word learning, and possibly acquisition of grammar. – Keith Holyoak

97

Analogy: Synchronic and Diachronic WORKS CITED AND SUGGESTIONS FOR FURTHER READING Bunge, Silvia, C. Wendelken, and A. D. Wagner. 2005. “Analogical reasoning and prefrontal cortex: Evidence for separable retrieval and integration mechanisms.” Cerebral Cortex 15: 239–49. Falkenhainer, Brian, K. D. Forbus, and D. Gentner. 1989. “The Structuremapping engine: Algorithm and examples.” Artificial Intelligence 41: 1–63. Gentner, Dedre. 1983. “Structure-mapping: A theoretical framework for analogy.” Cognitive Science 7: 155–70. Gentner, Dedre, B. Falkenhainer, and J. Skorstad. 1988. “Viewing metaphor as analogy.” In Analogical Reasoning: Perspectives of Artificial Intelligence, Cognitive Science, and Philosophy, ed. D. Helman, 171–7. Dordrecht, the Netherlands: Kluwer. Gentner, Dedre, K. J. Holyoak, and B. N. Kokinov , eds. 2001. The Analogical Mind: Perspectives from Cognitive Science. Cambridge, MA: MIT Press. This book contains survey articles on topics in analogy. Gentner, Dedre, and L.L. Namy. 2004. “The role of comparison in children’s early word learning.” In Weaving a Lexicon, ed. D. Hall and S. Waxman, 533–68. Cambridge, MA: MIT Press. Gentner, Dedre, and M. Rattermann. 1991. “Language and the career of similarity.” In Perspectives on Thought and Language: Interrelations in Development, ed. S. Gelman and J. Byrnes, 225–77. Cambridge: Cambridge University Press. Gick, Mary, and K. J. Holyoak. 1983. “Schema induction and analogical transfer.” Cognitive Psychology 15: 1–38. Goswami, Usha. 1992. Analogical Reasoning in Children. Hillsdale, NJ: Erlbaum. Halford, Graeme. 1993. Children’s Understanding: The Development of Mental Models. Hillsdale, NJ: Erlbaum. Holyoak, Keith. 1982. “An analogical framework for literary interpretation.” Poetics 11: 105–26. Holyoak, Keith, and P. Thagard. 1989. “Analogical mapping by constraint satisfaction.” Cognitive Science 13: 295–355. ———. 1995. Mental Leaps: Analogy in Creative Thought. Cambridge, MA: MIT Press. This book provides a broad introduction to the nature and uses of analogy. Hummel, John, and K. J. Holyoak. 2005. “Relational reasoning in a neurally-plausible cognitive architecture: An overview of the LISA Project.” Current Directions in Cognitive Science 14: 153–7. Morrison, Robert, D. C. Krawczyk, K. J. Holyoak, J. E. Hummel, T. W. Chow, B. L. Miller, and B. J. Knowlton. 2004. “A neurocomputational model of analogical reasoning and its breakdown in frontotemporal lobar degeneration.” Journal of Cognitive Neuroscience 16: 260–71. Oden, David, R. K. R. Thompson, and D. Premack. 2001. “Can an ape reason analogically? Comprehension and production of analogical problems by Sarah, a chimpanzee (Pan Troglodytes).” In Gentner, Holyoak, and Kokinov 2001, 471–97. Richland, Lindsey, R. G. Morrison, and K. J. Holyoak. 2006. “Children’s development of analogical reasoning: Insights from scene analogy problems.” Journal of Experimental Child Psychology 94: 249–71. Richland, Lindsey, O. Zur, and K. J. Holyoak. 2007. “Cognitive supports for analogy in the mathematics classroom.” Science 316: 1128–9. Spearman, Charles. 1946. “Theory of a general factor.” British Journal of Psychology 36: 117–31. Vallauri, Edoardo. 2004. “The relation between mind and language: The innateness hypothesis and the poverty of the stimulus.” Linguistic Review 21: 345–87.

ANALOGY: SYNCHRONIC AND DIACHRONIC Analogy involves two (or more) systems, A and B, which are constituted by their respective parts, that is, a1, a2, a3 (etc.) and b1,

98

b2, b3 (etc.). There is some relation R between a1, a2, and a3, expressed as R(a1,a2,a3), just as there is another such relation S between b1, b2, and b3, expressed as S(b1,b2,b3). For A and B to be analogous, it is required that R and S be exemplifications of the same abstract structure X, as evidenced by a mapping between a1/a2/a3 and b1/b2/b3. This is what is meant by saying that analogy (e.g., between A and B) is a structural similarity, or a similarity between relations (e.g., R and S). It is not a material similarity, or a similarity between things (e.g., a1 and b1). More and more abstract analogies are constituted by similarities between similarities between … relations between things. In its purely synchronic use (see synchrony and diachrony), analogy is understood to be the centripetal force that holds the units of a structure together. To simplify an example given by N. S. Trubetzkoy ([1939] 1958, 60–66), in a structure containing just /p/, /b/, /t/, and /d/, the phoneme /p/ acquires the distinctive features voiceless and bilabial by being contrasted with, respectively, (voiced) /b/ and (dental) /t/. The relation between the pairs /p/ & /b/ and /t/ & /d/ is the same, and so is the relation between /p/ & /t/ and /b/ & /d/, which means – by definition – that there is in both cases an analogy between the two pairs. This type of analogy-based analysis applies to any wellarticulated structure, linguistic or nonlinguistic: A unit is what the other units are not (as /p/ is neither /b/ nor /t/ nor, of course, /d/); and this “otherness” is based on corresponding oppositions (like voiceless vs. voiced and bilabial vs. dental). Synchronic analogy may be characterized as analogy-asstructure. Its counterpart is analogy-as-process,that is, discovery, manipulation, or invention of structural similarity. Traditionally, language acquisition was thought to be based on analogy: “From innumerable sentences heard and understood [the child] will abstract some notion of their structure which is definite enough to guide him in framing sentences of his own” (Jespersen [1924] 1965, 19). After a period of neglect, this traditional view has again become fashionable in some quarters (Pinker 1994, 417; Tomasello 2003, 163–9). Only if analogy-as-process leaves a permanent trace that deviates from the current norm is there reason to speak of language change, the province of diachronic linguistics. Traditionally, the term analogical change was restricted to morphology, or to cases where irregularities brought about by sound change are eliminated so as to achieve, or to approach, the goal of one meaning – one form. However, this same goal is involved in such large-scale changes as have generally been ascribed to a need for harmony or symmetry. In syntactic change, analogy consists in extending a reanalyzed structure to new contexts (Anttila 1989, 102–4). – Esa Itkonen WORKS CITED AND SUGGESTIONS FOR FURTHER READING Anttila, Raimo. 1989. Historical and Comparative Linguistics. Amsterdam: Benjamins. Jespersen, Otto. [1924] 1965. Philosophy of Grammar. London: Allen and Unwin. Pinker, Stephen. 1994. The Language Instinct. New York: Morrow. Tomasello, Michael. 2003. Constructing a Language. A Usage-Based Theory of Language-Acquisition. Cambridge: Harvard University Press.

Analyticity Trubetzkoy, N. S. [1939] 1958. Grundzüge der Phonologie. Göttingen: Vandenhoeck and Ruprecht.

ANALYTICITY Analyticity is a property that a statement has when its truth is in some special way determined by its meaning. Many believe that no such property exists. Led by W. V. O. Quine (1953), philosophers complain that no one has been able to define the concept of analyticity in a way that is precise and that also fulfills the purposes to which it is typically put. To some people, it seems that “all uncles are either married or have a sibling” is true just because of the meaning of its constituent words, most prominently because of the meaning of uncle. But “all uncles are less than eight feet tall” is true not because of meanings but because of how the world has turned out to be. The first sort of statement is said to be analytic, the second synthetic. This distinction has far-reaching interest and application. Empiricists have always had difficulty accounting for the seemingly obvious fact that the truths of logic and mathematics are both necessary (i.e., they could not be false) and a priori (i.e., they are known independently of sensory experience). For most of the twentieth century, it was agreed that analyticity could explain this obvious fact away as a merely linguistic phenomenon. The idea that all necessity could be explained away by analyticity fell out of fashion when S. Kripke (1980) convinced most philosophers that some necessary truths are neither analytic nor a priori (e.g., water = H2O). But as L. BonJour (1998, 28) points out, many still think that the unusual modal and epistemic status of logic and mathematics is due to a special relation between truth and meaning. However we apply the concept, trouble for analyticity begins when we remind ourselves that every truth depends, to some extent, on the meanings of its constituent terms. If the word uncle had meant oak tree, then both previous examples would be false. In response, it is said that an analytic truth is one whose truth depends solely on the meanings of its terms. Our linguistic conventions alone make the first example true, and the second is true party because of meaning and partly because of the way the world is. But how can linguistic convention alone make something true? We can distinguish the sentence “all uncles are either married or have a sibling” from the proposition that this sentence now expresses. Meaning or linguistic convention alone makes this sentence true in the following way. Given that this sentence expresses the proposition that it does (i.e., given our current linguistic conventions), it is true. This cannot be said of our other example. That this sentence means what it does is not sufficient to determine its truth or falsity (see also sentence meaning). The world plays a part. There are three serious problems. First, if this is what it is for a sentence to be true solely in virtue of its meaning, then it is just another way of saying that it expresses a necessary truth, and that tells us nothing about how we know it. Thus, appeal to analyticity cannot explain the necessity and a priority of logic, mathematics, or anything else. Second, the proposition now expressed by our first example would be true no matter how we ended up

expressing it. Thus, our current linguistic conventions do not make it true. Third, our first example certainly does say something about the world. Indeed, it says something about every object in the universe. If it is an uncle, then either it is married or it has a sibling. Some philosophers have called the sort of analyticity discussed so far, where things are said to be true in virtue of meaning, metaphysical analyticity. This is distinguished from epistemic analyticity. A statement is epistemically analytic when understanding it suffices for being justified in believing it (Boghossian 1996). While these considerations make trouble for metaphysical analyticity, they allegedly leave its epistemic counterpart untouched. The purpose of introducing epistemic analyticity is similar to that of its older ancestor. The hope is that mathematical, logical, and conceptual truths can be designated as a priori without postulating a special faculty of reason or intuition. This is done by building certain kinds of knowledge in as preconditions for possessing or understanding concepts. If part of what it is to understand the word uncle is to be disposed to accept that “all uncles are either married or have a sibling,” then it could be argued that once we understand that sentence, we know that it is true. No experience (beyond what is required for understanding) is necessary. The best candidates for epistemically analytic truths are simple truths of logic. But even the most obvious logical truths are not immune to challenge. For example, a few philosophers and logicians have claimed that some statements can be both true and false (Preist 1987) and that modus ponens is invalid (McGee 1985). Yet these sophisticated theoreticians certainly understand the meanings of their own words. Therefore, acceptance of some specific truth of logic is not necessary for understanding any logical concept. And since we might someday have good reason to reject any particular truth of logic while continuing to understand our logical concepts, understanding some logical concept is not sufficient for being justified in believing any particular truth of logic. And if logic is not epistemically analytic, nothing is. These considerations make the existence of analyticity dubious. But there still appears to be a deep difference between the two examples. If there is really a difference, it is not one of true in virtue of meaning versus true in virtue of reality, but one of necessary and a priori versus contingent and empirical. – Michael Veber WORKS CITED AND SUGGESTIONS FOR FURTHER READING BonJour, L. 1998. In Defense of Pure Reason. Cambridge: Cambridge University Press. Boghossian, P. 1996. “Analyticity reconsidered.” Nous 30.3: 360–91. Kripke, S. 1980. Naming and Necessity. Cambridge: Harvard University Press. McGee, V. 1985. “A counterexample to modus ponens.” Journal of Philosophy 82: 462–71. Priest, G. 1987. In Contradiction. Boston: Kluwer. Quine, W. V. O. 1953. “Two dogmas of empricism.” In From a Logical Point of View, 20–46. Cambridge: Harvard University Press. Veber, M. 2007. “Not too proud to beg (the question): Why inferentialism cannot account for the a priori.” Grazer Philosophische Studien 73: 113–31. A critique of Boghossian 1996 and similar views.

99

Anaphora

ANAPHORA Languages have expressions whose interpretation may involve an entity that has been mentioned before: Subsequent reference to an entity already introduced in discourse approximates a general definition of the notion of anaphora (Safir 2004, 4). This works well for core cases of nominal anaphora as in (1) (Heim 1982): (1) a. This soldier has a gun. Will he shoot? b. Every soldier has a gun. ??Will he shoot?

He in (1a) can be interpreted as the same individual as this soldier. In (1b) every soldier is quantificational, hence does not denote an entity he can refer to, which makes anaphora impossible. Possible discourse antecedents are as diverse as soldiers, water, beauty, headaches, dissatisfaction, and so on. In addition to nominal expressions, sentences, verb phrases, prepositional phrases, adjective phrases, and tenses also admit anaphoric relations. Thus, the notion discourse entity must be broad enough to capture all these cases of anaphora yet restrictive enough to separate them from quantificational cases such as “every soldier.” The notion of anaphora is closely related to the notion of interpretative dependency. For instance, in (2), he can depend for its interpretation on every soldier, and here, too, it is said that he is anaphorically related to every soldier. 2.

Every soldier who has a gun says he will shoot.

However, (1) versus (2) shows that two different modes of interpretation must be distinguished: i) directly assigning two (or more) expressions the same discourse entity from the interpretation domain (ID) as a value: co-reference as in (1a), and ii) interpreting one of the expressions in terms of the other by grammatical means: binding (Reinhart 1983). This contrast is represented in (3).

Coreference in (3a) is restricted in terms of conditions on discourse entities, binding in (3b) in terms of grammatical configuration. Expr1 can only bind expr2 if it c-commands the latter (Reinhart 1983). This condition is met in (2), but not in (1b), hence the contrast. Virtually all languages have words and expressions that are referentially defective – they cannot be used deictically (see deixis). In much of the linguistic literature, these are called anaphors, as they appear specialized for anaphoric use. Examples vary from English himself, Dutch zich(zelf), Icelandic sig, Russian sebja, Chinese (ta) ziji, to Georgian tav tavis, and so on. Such expressions cannot be assigned a discourse value directly. Rather, they must be bound, often in a local domain – approximately the domain of the nearest subject – but subject to variation in terms of specific anaphor type and language. Furthermore, under

100

Animacy restrictive conditions, some of these expressions may yet allow a free – logophoric – interpretation. It is an important task to arrive at a detailed understanding of the ways in which languages encode interpretive relations between their expressions and of the division of labor between the components of the language system involved. – Eric Reuland WORKS CITED AND SUGGESTIONS FOR FURTHER READING Chomsky, Noam. 1981. Lectures on Government and Binding. Dordrecht, the Netherlands: Foris. Heim, Irene. 1982. The semantics of definite and indefinite noun phrases. Ph.D. diss., University of Massachusetts at Amherst. Reinhart, Tanya. 1983. Anaphora and Semantic Interpretation. London: Croom Helm. Reuland, Eric. 2010. Anaphora and Language Design. Cambridge: MIT Press. Safir, Ken. 2004. The Syntax of Anaphora. Oxford: Oxford University Press.

ANIMACY Languages often treat animate and inanimate nouns differently. Animacy can affect many aspects of grammar, including word order, and verbal agreement. For example, in Navajo, the more animate noun must come first in the sentence (Hale 1973), and in some Bantu languages, a more animate object must come before a less animate object. Verbs are more likely to agree with more animate nouns (Comrie 1989). Animacy can also affect the choice of case, preposition, verb form, determiner (article), or possessive marker (Comrie 1989). What counts as animate differs cross-linguistically. The grammatical category of animates may include certain objectively inanimate things, such as fire, lightning, or wind. Languages may make additional distinctions among pronouns, proper nouns, and common nouns, or between definites and indefinites, and these are sometimes viewed as part of an animacy hierarchy by linguists (Comrie 1989). person and number distinctions may also be included in an animacy hierarchy in this broader sense. For example, according to Michael Silverstein (1976), subjects with features at the less animate end of the Animacy Hierarchy in (1) are more likely to be marked with morphological case, while the reverse holds of objects. (1)

Animacy Hierarchy 1pl > 1sing > 2pl > 2sing > 3human.pl > 3human.sing > 3anim.pl > 3anim.sing > 3inan.pl > 3inan.sing

Dyirbal exhibits this pattern in that only third person subjects have morphological case, whereas all human objects do. Silverstein (1976) postulates that the function of such differential case marking is to flag less animate subjects and more animate objects to avoid ambiguity. It is interesting to note that the patterns of such differential animacy marking are far more complex and diverse cross-linguistically for objects than for subjects (Aissen 2003). This may be traced to a relation between animacy and object shift, which produces an associated change in case or verbal agreement (Woolford 2000, 2007). The less diverse animacy effects on subject case, which do not affect agreement

Animal Communication and Human Language (Comrie 1991), may be purely morphological, markedness effects (Woolford 2007). – Ellen Woolford WORKS CITED AND SUGGESTIONS FOR FURTHER READING Aissen, Judith. 2003. “Differential object marking: Iconicity vs. economy.” Natural Language and Linguistic Theory 21: 435–83. Comrie, Bernard. 1989. Language Universals and Linguistic Typology. Oxford: Blackwell. ———. 1991. “Form and function in identifying cases.” In Paradigms, ed. F. Plank, 41–55. Berlin: Mouton de Gruyter. Hale, Kenneth L. 1973. “A note on subject-object inversion in Navajo.” In Issues in Linguistics: Papers in Honor of Henry and Renée Kahane, ed. B. B. Kachru et al., 300–309. Urbana: University of Illinois Press. Silverstein, Michael. 1976. “Hierarchy of features and ergativity.” In Grammatical Categories in Australian languages, ed. R. M. Dixon, 112–71. Canberra: Australian Institute of Aboriginal Studies. Woolford, Ellen. 2000. “Object agreement in Palauan.” In Formal Problems in Austronesian Morphology and Syntax, ed. I. Paul , V. Phillips, and L. Travis, 215–45. Dordrecht, the Netherlands: Kluwer. ———. 2007. “Differential subject marking at argument structure, syntax, and PF.” In Differential Subject Marking, ed. H. de Hoop and P. de Swart, 17–40. Dordrecht: Springer.

ANIMAL COMMUNICATION AND HUMAN LANGUAGE An understanding of the communicative capacities of other animals is important on its face both for an appreciation of the place of human language in a broader context and as a prerequisite to discussion of the evolution of language (see, for example, evolutionary psychology). On closer examination, however, the differences between human language and the systems of other animals appear so profound as to make both projects more problematic than they appear at first. In the 1950s and 1960s, ethologists like Konrad Lorenz and Niko Tinbergen revolutionized behavioral biologists’ views of the cognitive capacities of animals, but consideration of animal communication focused on the properties of quite simple systems. A prime example of “communication” in the texts of the time was the stickleback. A crucial component of the mating behavior of this common fish is the pronounced red coloration of the male’s underbelly when he is in mating condition, which furnishes a signal to the female that she should follow him to his preconstructed nest, where her eggs will be fertilized. On this model, communication was viewed as behavioral or other signals emitted by one organism, from which another organism (typically, though not always, a conspecific) derives some information. The biological analysis of communication thus came to be the study of the ways in which such simple signals arise in the behavioral repertoire of animals and come to play the roles they do for others who perceive them. Those discussions make little if any contact with the analysis of human language. In the intervening half century, we have come to know vastly more about the nature and architecture of the human language faculty and to have good reason to think that much of it is grounded in human biology. One might expect, therefore, to find these concerns reflected in the behavioral biology literature. A comprehensive modern textbook on animal communication

within this field (e.g., Bradbury and Vehrenkamp 1998) reveals a great deal about the communicative behavior of many species and its origins, but within essentially the same picture of what constitutes communication in (nonhuman) animals, confined to unitary signals holistically transmitted and interpreted. Little if any of what we have come to know about human linguistic communication finds a place here. Biologists have not, in general, paid much attention to the specifics of linguistic research (though their attention has been caught by the notion that human language is importantly based in human biology) and are often not as sophisticated as one might wish about the complexity of natural language. But the consequences may not be as serious as linguists are inclined to think. In fact, the communicative behavior of nonhumans in general is essentially encompassed within the simple signal-passing model. The complexities of structure displayed by human language are apparently quite unique to our species and may not be directly relevant to the analysis of animal communication elsewhere.

What (Other) Animals Do Communication in the sense of emission and reception of informative signals is found in animals as simple as bacteria (quorum sensing). Most familiar, perhaps, are visual displays of various sorts that indicate aggression, submission, invitations to mate, and so on. In some instances, these may involve quite complex sequences of gestures, reciprocal interactions, and the like, as in the case of the nesting and mating behavior of many birds. In others, a simple facial expression, posture, or manner of walking may provide the signal from which others can derive information about the animal’s intentions and attitudes. These differences of internal structure are, of course, crucial for the correct expression and interpretation of a particular signal, but they play little or no role in determining its meaning. That is, the individual components of the signal do not in themselves correspond to parts of its meaning, in the sense that varying one subpart results in a corresponding variation in what is signaled. Animal signals, however complex in form (and however elaborate the message conveyed), are unitary wholes. An entire courtship dance, perhaps extending over several minutes or even longer, conveys the sense “I am interested in mating with you, providing a nesting place, and care for our offspring.” No part of the dance corresponds exactly to the “providing care” part of the message; the message cannot be minimally altered to convey “I am interested in mating but not in providing care for our offspring,” “I was interested in mating (but am no longer),” and so on. Variations in intensity of expression can convey (continuous) variations in the intensity of the message (e.g., urgency of aggressive intent), but that is essentially the only way messages can be modulated. The most widely discussed apparent exception to this generalization is the dance language of some species of honeybees. The bees’ dance conveys information about a) the direction, b) the distance, and c) the quality of a food source (or potential hive site), all on quasi-continuous scales and each in terms of a distinct dimension of the dance. Although the content of the message here can be decomposed, and each part associated with a distinct component of the form of the signal, there is no element of free combination. Every dance necessarily conveys

101

Animal Communication and Human Language exactly these three things, and it is only the relative value on each dimension that is variable. As such, the degree of freedom available to construct new messages is not interestingly different from that involved in conveying different degrees of fear or aggression by varying degrees of piloerection. Visual displays do not at all exhaust the modalities in which animal communication takes place, of course. Auditory signals are important to many species, including such classics of the animal communication literature as frog croaks and the calls and songs of birds (see birdsong and human language). In some species, portions of the auditory spectrum that are inaccessible to humans are involved, as in the ultrasound communication of bats, some rodents, and dolphins, and the infrasound signals of elephants. Chemical or olfactory communication is central to the lives of many animals, including moths, mice, and lemurs, as well our pet cats and dogs. More exotic possibilities include the modulation of electric fields generated (and perceived) by certain species of fish. In some of these systems, the internal structure of the signal may be quite complex, as in the songs of many oscine songbirds, but the general point still holds: However elaborate its internal form, the signal has a unitary and holistic relation to the message it conveys. In no case is it possible to construct novel messages freely by substitutions or other ways of varying aspects of the signal’s form. In most animals, the relation of communicative behavior to the basic biology of the species is very direct. Perceptual systems are often quite precisely attuned to signals produced by conspecifics. Thus, the frog’s auditory system involves two separate structures (the amphibian papilla and the basilar papilla) that are sensitive to acoustic signals, typically at distinct frequencies. The frequencies to which they are most sensitive vary across species but are generally closely related to two regions of prominence in the acoustic structure of that species’ calls. Mice (and many other mammals) have two distinct olfactory organs, projecting to quite distinct parts of the mouse brain. The olfactory epithelium is responsive to a wide array of smells, but the vomeronasal organ is sensitive primarily to the pheremones that play a major role in communication and social organization. In this case, as in many, many others, the perceptual system is matched to production in ways that optimize the organism’s sensitivity to signals that play a crucial ecological role in the life of the animal. The essential connection between a species’ system of communication and its biology is also manifested in the fact that nearly all such systems are innately specified. That is, the ability to produce and interpret relevant signals emerges in the individual without any necessary role of experience. Animal communication is not learned (or taught) but, rather, develops (in the absence of specific pathology, such as deafness) as part of the normal course of maturation. Animals raised under conditions in which they are deprived of exposure to normal conspecific behavior will nonetheless communicate in the fashion normal to their species when given a chance. Exceptions to this generalization are extremely rare, apart from human language. Vocal learning, in particular, has been demonstrated only to a limited extent in cetaceans and some bats and, more extensively, in 3 of the 27 orders of birds. The study of birds, especially oscine songbirds, is particularly instructive in

102

this regard. In general, their song is learned on the basis of early exposure to appropriate models, from which they in turn compose their own songs. It is interesting to note there appear to be quite close homologies in the neurophysiology of vocal learning – and perhaps even in its underlying genetic basis – between birds and humans, although what is learned in birds is a unitary, holistic signal like those in other nonhuman communication systems, rather than individual lexical items subject to free recombination to produce different meanings. There is much variation across bird species, but a clear generalization emerges: For each, there is a specific range of song structures that individuals of that species can learn. Experience plays a role in providing the models on which adult song is based, but (with the exception of a few very general mimics, such as the lyrebird) this role is quite narrowly constrained by the songlearning system of the individual species.

What Humans Do, and How It Is Different Like the systems of communication of other animals, human language is deeply embedded in human biology. Unlike others, however, it provides an unbounded range of distinct, discrete messages. Human language is acquired at a specific point in development from within a limited range of possibilities, similar to the acquisition of song in birds. Unlike the communicative signals of other species, human language is under voluntary control, with its underlying neurobiology concentrated in cortical structures, as opposed to the subcortical control characteristic of those other species that have been studied in this regard. Human language is structurally a discrete combinatorial system, in which elements from a limited set combine in a recursive, hierarchical fashion to make an unlimited number of potentially novel messages (see recursion, iteration, and metarepresentation). The combinatorial structure of language is governed by two quite independent systems: A small inventory of individually meaningless sounds combine to make meaningful words, on the one hand (phonology), while these words are combined by a quite different system to make phrases, clauses, and sentences (see syntax). These properties (discrete combination, recursive hierarchical organization, and duality of patterning) are not simply idiosyncratic ornaments that could in principle be omitted without affecting the overall communicative capacity of the system. Rather, they are what make large vocabularies practical and unbounded free expression possible. Contrast the unlimited range of potentially novel utterances that any (normal) speaker of a language can produce, and another speaker of the same language comprehend, with the strictly limited range of meaningful signals available to other organisms. No other form of communication found in nature has these properties. Although song in some species of birds does display a limited amount of phonological combinatoriality, there is no analog even here to meaningful syntax. Human language, and especially its syntactic organization, is quite unique in the animal world. Furthermore, efforts to teach systems with these essential properties to other animals have not succeeded. Despite widespread claims to the contrary in the popular literature, there is no evidence that any nonhuman animal is capable of acquiring and using such a system. This should not be seen as particularly surprising. If language is indeed embedded in human biology,

Aphasia there is no reason to expect it to be accessible to organisms with a different biological endowment, anymore than humans are capable of acquiring, say, the echolocation capacities of bats, a system that is equally grounded in the specific biology of those animals.

Conclusion Human language is often considered as simply one more instantiation of the general class of animal communication systems. Indeed, like others it appears to be highly species specific. Although relevant experience is required to develop the system of any particular language, the overall class of languages accessible to the human learner is apparently highly constrained, and the process of language learning is more like genetically governed maturation than like learning in general. The structural characteristics of human language are quite different from those of other communication systems, and it is the freedom of expression subserved by those distinctive properties that gives language the role it has in human life. (See also primate vocalizations and grooming, gossip, and language). – Stephen R. Anderson

emission tomography (PET), and functional MRI (fMRI) have helped define functions of core speech and language areas.

Aphasia Syndromes Aphasia has traditionally been categorized into seven subtypes, including Broca’s, Wernicke’s, global, anomic, conduction, transcortical sensory, and transcortical motor. These aphasia variants are characterized by different patterns of speech fluency, auditory comprehension, repetition, and naming. Although patients may be classified as having one type of aphasia in the early period after a brain injury, this classification may change as language problems resolve with time and treatment. Broca’s aphasia is characterized by a constellation of symptoms, including slow, halting speech with impaired grammar; disturbed auditory comprehension for grammatically complex phrases and sentences; and poor repetition. Word-finding problems and difficulty with reading and writing are common. Motor speech disorders, such as apraxia of speech, a disorder of articulatory planning or coordination, and dysarthria, an impairment in muscle strength, tone, or coordination, very often co-occur. Patients with Broca’s aphasia often talk in a series of nouns and verbs, as is the case in the following sample from a patient describing the picnic scene from the Western Aphasia Battery (WAB):

WORKS CITED AND SUGGESTIONS FOR FURTHER READING Anderson, Stephen R. 2004. Doctor Dolittle’s Delusion: Animals and the Uniqueness of Human Language. New Haven, CT: Yale University Press. Bradbury, J. W., and Sandra Vehrenkamp. 1998. Principles of Animal Communication. Sunderland, MA: Sinauer. Hauser, Marc D. 1996. The Evolution of Communication. Cambridge, MA: MIT Press.

APHASIA Aphasia is a language impairment caused by brain injury that affects speech content, auditory comprehension, reading, and writing to varying degrees. Mild aphasia may result in occasional word-finding problems, while more severe forms can cause profound deficits in all language domains. Aphasia differs from developmental disorders in that it occurs after a brain injury to a person with otherwise normal language skills. Typically, aphasia results from damage to the left hemisphere of the brain due to stroke, traumatic brain injury, tumor, or degenerative neurological disease. Nearly all right-handed individuals and most left-handers are thought to have core linguistic functions – semantics, phonology, syntax, and morphology – lateralized to the left hemisphere, while other aspects of language, specifically prosody and pragmatics, are associated with the right hemisphere. Approximately one million people in the United States are afflicted with aphasia, which is a prevalence rate similar to that of Parkinson’s disease. Roughly 80,000 more acquire aphasia annually. Recent advances in the study of language have provided greater insight into aphasia syndromes. Modern neuroimaging technology has helped refine classic models of language localization, as well as our understanding of aphasia treatment and recovery. In particular, brain-imaging techniques such as magnetic resonance imaging (MRI), computerized tomography (CT), positron

I know it … tree … house … car… man with … uh woman … kid over here with flag … i can know it … nice sun shiny day … [unintelligible word]…

Patients with Broca’s aphasia can participate relatively well in everyday conversation by using single words or phrases, often combined with meaningful gestures and facial expressions. Some patients use writing and drawing to compensate for restricted verbal output. As is true with all aphasia syndromes, there is a wide range of symptom severity. Those with severe Broca’s aphasia may present with such profound verbal deficits that their speech is limited to a single recurrent utterance (e.g., “yeah, yeah”). In these patients, comprehension is usually preserved for short, simple phrases, but is significantly impaired for more complex information. Classic aphasia models assume that lesions to broca’s area (See Figure 1) result in Broca’s aphasia, but research has indicated that this is not always the case. Reports as early as 1870 documented cases that did not support this linear relationship (e.g., Bateman 1870; Marie 1906; Moutier 1908; Dronkers et al. 2007). Modern research has found that chronic Broca’s aphasia typically results from large lesions that encompass left frontal brain regions, underlying white matter, the insula, and the anterior parietal lobe. Lesions restricted to Broca’s area tend to cause transient mutism that spontaneously resolves within days or weeks (Mohr 1976). In some cases, Broca’s aphasia can occur without damage to Broca’s area (e.g., Basso et al. 1985; Mazzocchi and Vignolo 1979). Patients with Wernicke’s aphasia present a reverse pattern of symptoms when compared to those with Broca’s aphasia: While speech is fluent, comprehension is impaired. Patients speak in a normal or rapid rate. However, they often use meaningless words, jargon, or semantic paraphasias (e.g., using a related word, bike for a target word, car). Reading and writing may be similarly disrupted. The following exemplifies the speech

103

Aphasia

Figure 1. Several of the key brain regions affected in aphasia. Areas depicted as typical lesions are derived from patient data obtained at the Center for Aphasia and Related Disorders.

content of a patient with Wernicke’s aphasia describing the WAB picnic scene: And the man and hers and I’ll say … I don’t think she’s working. They’re not doing the thing. Then the ladder … then the tree … and the /let/ [points to kite] and lady here [points to boy] … have to clean that.

In contrast to patients with Broca’s aphasia, those with Wernicke’s aphasia may understand very little in conversation because of their impaired comprehension of single words. In addition, successful communication is made challenging by verbal output that is empty, coupled with an inability to monitor speech content. Using visual information to compensate for comprehension deficits is often beneficial (e.g., providing pictures, drawing, or writing key words during conversational exchanges). Persisting cases of Wernicke’s aphasia are not caused by injury to wernicke’s area alone but, rather, by much larger lesions affecting most of the middle temporal gyrus and underlying white matter (Dronkers, Redfern, and Ludy 1995; see Figure 1). Such damage amounts to a poorer prognosis for recovery. Patients with lesions confined to Wernicke’s area tend to have symptoms of Wernicke’s aphasia that resolve, resulting in milder forms of aphasia, most often conduction aphasia or anomic aphasia, if the lesion spares the middle temporal gyrus. Conduction aphasia is a fluent aphasia characterized by an inability to repeat. Auditory comprehension is relatively preserved, and patients use speech that is largely understandable but may be rife with phonemic paraphasias (substituting sounds in words, e.g., netter for letter). While high-frequency words and short phrases may be repeated accurately (e.g., “the telephone is ringing”), low-frequency items are more difficult (e.g., “first British field artillery”). Patients may retain the meaning of such phrases, owing to their preserved comprehension, but the phonological trace is disrupted, thereby disturbing verbatim repetition. The following typifies the speech of a patient with conduction aphasia, again describing the WAB picnic scene: Well there’s a house near a clearing, evidently it’s on the water.… Further, there’s a stick with a banner in the foreground [referring to the flag]. I don’t know what that’s called … a pier … a tier? There’s a bucket and a /kovel/. It looks like there’s someone playing in the water.

104

Initial reports that conduction aphasia arose from lesions to the arcuate fasciculus (the white matter tract connecting Wernicke’s and Broca’s areas; see Figure 1) have been refined over the years. Modern studies have shown that conduction aphasia results most often from lesions to the posterior superior temporal gyrus (Dronkers et al. 1998), the auditory cortex (Damasio and Damasio 1980), or periventricular white matter underlying the supramarginal gyrus (Sakurai et al. 1998). Global aphasia, the most severe syndrome, is characterized by profound impairments in all language modalities. Speech, auditory comprehension, naming, repetition, reading, and writing are all affected, leaving the patient with very little functional communication. Speech may be limited to single stereotyped or automatic words and phrases (e.g., yes, no, I don’t know). Auditory comprehension may be impaired for even simple yes/ no questions. Such a severe loss of language typically results from a large cortical lesion, encompassing the frontal, temporal, and parietal lobes. Patients often rely on preserved nonverbal skills to aid in communication (e.g., the recognition of pictures and gestures to support auditory comprehension and the ability to draw or gesture to aid in expression). Anomic aphasia, the mildest of the syndromes, results in word-finding deficits (anomia), while other language skills are typically well preserved. When attempting to find a target word, patients with anomic aphasia may describe its function or use a synonym. Speech may be slow and halting, due to anomia, but grammar is unaffected. Anomic aphasia can result from lesions anywhere within the perisylvian region. The transcortical aphasias are rare and characterized by a preserved ability to repeat, despite impairments in other language domains. Transcortical motor aphasia (TCMA) is similar to Broca’s aphasia, in that patients present with nonfluent speech and relatively intact comprehension, but repetition skills are markedly well preserved. Lesions typically spare core language areas, are smaller than those that cause Broca’s aphasia, and are restricted to anterior and superior frontal lobe regions. Although patients may be mute initially, their symptoms tend to resolve quickly, resulting in anomic aphasia. Patients with transcortical sensory aphasia (TCSA) present much like patients with Wernicke’s aphasia, with empty, fluent speech and poor comprehension, but they too retain a striking ability to repeat. Lesions typically involve portions of the posterior temporal

Aphasia

Areal Distinctness and Literature

and parietal regions, but tend to be much smaller than those of Wernicke’s aphasia. Acute symptoms usually resolve to produce an anomic aphasia. While aphasia most often occurs suddenly, as the result of injury, a degenerative form of aphasia was first described over a century ago by Arnold Pick, a Czech neurologist, and later expanded upon by Marsel Mesulam in a landmark paper in which he described six patients who presented with language deficits, in the absence of other behavioral abnormalities (Mesulam 1982). Speech or language deficits remained the only impairment for the first two years in these patients, but as the disease progressed, more generalized dementia emerged. This progressive disorder was distinct from other dementias, such as Alzheimer’s disease, because language problems, rather than memory complaints, were the most salient symptoms. Since then, numerous cases of what is now termed primary progressive aphasia (PPA) have been described, in which patients present with both fluent and nonfluent variants of the disorder (Snowden, et al. 1992; Gorno-Tempini et al. 2004). Neuroimaging typically shows left perisylvian atrophy, encompassing frontal regions in progressive nonfluent aphasia and anterior temporal and temporo-parietal regions in the more fluent semantic dementia and logopenic variants. There are many underlying pathologies that cause the clinical syndrome of PPA, including Pick’s disease, progressive supranuclear palsy, corticobasal degeneration, dementia lacking distinctive pathology, and Alzheimer’s disease.

Treatment for Aphasia Critical reviews of aphasia treatment studies (e.g., Bhogal, Teasell, and Speechley 2003; Holland et al. 1996) have shown that treatment can be effective in improving language skills past the point that might be expected from spontaneous recovery alone. Although it remains difficult to predict the treatment that will result in the greatest amount of change for an individual, there are many options from which to choose. Patients with aphasia are typically referred to speech language pathologists for diagnostic testing aimed at developing treatment goals. Therapy may focus on improving impaired skills or developing compensatory strategies to overcome obstacles to successful communication. Patient-specific factors (e.g., aphasia severity, cognitive ability, general health, and motivation) also influence treatment decisions. Research is inconclusive, however, as to the prognostic weight that these variables contribute to recovery and treatment planning for an individual. – Nina F. Dronkers, Jennifer Ogar WORKS CITED AND SUGGESTIONS FOR FURTHER READING Basso, A., A. R. Lecours, S. Moraschini, and M. Vanier. 1985. “Anatomoclinical correlations of the aphasias as defined through computerized tomography: On exceptions.” Brain and Language 26: 201–29. Bateman, F. 1870. On Aphasia. London: Churchill. Bhogal, S. K., R. Teasell, and M. Speechley. 2003. “Intensity of aphasia therapy, impact on recovery.” Stroke 34.4: 987–93. Damasio, H., and A. R. Damasio. 1980. “The anatomical basis of conduction aphasia.” Brain 103: 337–50. Dronkers, N. F., B. B. Redfern, and C. A. Ludy. 1995. “Lesion localization in chronic Wernicke’s aphasia.” Brain and Language 51: 62–65.

Dronkers, N. F., O. Plaisant, M. T. Iba-Zizen, and E. A. Cabanis. 2007. “Paul Broca’s historic cases: High resolution MR imaging of the brains of Leborgne and Lelong.” Brain 130: 1432–41. Dronkers N. F., B. B. Redfern, C. Ludy, and J. Baldo. 1998. “Brain regions associated with conduction aphasia and echoic rehearsal.” Journal of the International Neuropsychological Society 4: 23–4. Gorno-Tempini, M. L., N. F. Dronkers, K. P. Rankin, et al. 2004. “Cognition and anatomy in three variants of primary progressive aphasia.” Annals of Neurology 55: 335–46. Holland, A. L., D. S. Fromm, F. DeRuyter, M. Stein. 1996. “Treatment efficacy.” Journal of Speech and Hearing Research 39.5: S27–36. Marie P. 1906. “Revision de la question de l’aphasie: La troisieme circonvolution frontale gauche ne joue aucun role special dans la fonction du langage.” Semaine Medicale 26: 241–7. Mazzocchi, F., and L. A. Vignolo. 1979. “Localization of lesions in aphasia: Clinical CT-scan correlations in stroke patients.” Cortex 15: 627–54. Mesulam, M. M. 1982. “Slowly progressive aphasia without generalized dementia.” Annals of Neurology 11: 592–8. Mohr, J. P. 1976. “Broca’s area and Broca’s aphasia.” In Studies in Neurolinguistics. Vol. 1. Ed. H. Whitaker and H. Whitaker, 201–33. New York: Academic Press. Moutier, F. 1908. L’Aphasie de Broca. Paris: Steinheil. Sakurai, Y., S. Takeuchi, E. Kojima, et. al. 1998. “Mechanism of shortterm memory and repetition in conduction aphasia and related cognitive disorders: a neuropsychological, audiological and neuroimaging study.” Journal of Neurological Sciences 154.2: 182–93. Snowden, J. S., D. Neary, D. M. Mann, et al. 1992. “Progressive language disorder due to lobar atrophy.” Annals of Neurology 31: 174–83.

AREAL DISTINCTNESS AND LITERATURE There are two criteria for determining whether a linguistic property is a universal. First, it must occur across languages with a frequency greater than chance. Second, the presence of the property in some of these languages should not have been caused by its presence in other languages. In linguistics, the causal criterion is often operationally specified into two subcriteria – genetic and areal distinctness, which is to say, distinctness in origin and in cross-language interaction. Researchers in literary universals also adopt the preceding criteria. However, literature is different from language in being more readily open to influence. Specifically, the operational criterion of areal distinctness becomes much more difficult to satisfy in the case of literature. Even a single work, transported across continents, may produce significant changes in the recipient literature. There are three ways of responding to this problem. The first is to focus on literary works produced before the period of extensive global interaction. Research of this sort must form the primary basis for any serious study of literary universals. Moreover, such research indicates that there are some significant universals, for example the narrative universals of heroic, romantic, and sacrificial tragicomedy. However, this approach to areal distinctness cannot be as rigorous as one might like. Global interaction extends back through the formation of all the major literary traditions. The second response involves a more nuanced approach for isolating influence from a source tradition to a recipient tradition. Here, we may distinguish between self-conscious and implicit learning. Self-conscious learning can occur with a single

105

Art, Languages of exposure to salient features of a literary work. Implicit learning, however, is likely to require many exposures, commonly while immersed in the culture and language of the source tradition. In isolating literary universals, then, we may take into account the degree to which a particular property is likely to have been transported from one tradition to another by learning of either sort, given the degree of contact between the traditions. For example, the practice of dramatic performance may be transmitted from one tradition to another through limited interaction, as this may be learned through a single exposure. The same point does not hold for background imagery. Finally, we may wish to expand our study of cross-cultural patterns to actual borrowings. Here, too, it is crucial to distinguish different types of influence. We may roughly divide influence into two categories – hegemonic and nonhegemonic. Hegemonic influence occurs when the source tradition has greater economic power (e.g., in the publication and distribution of literary works), more pervasive control of government or education, or a higher level of prestige (due, for example, to military strength), or when it is otherwise in a position of cultural domination over the recipient society. Obvious cases are to be found in colonialism. Common properties that result from non-hegemonic influences are not universals themselves. However, they may tell us something about cross-cultural aesthetic or related propensities. Common properties that result from hegemonic influences, in contrast, may simply reflect the effects of power. – Patrick Colm Hogan WORK CITED AND SUGGESTIONS FOR FURTHER READING Comrie, Bernard. 1981. Language Universals and Linguistic Typology. Chicago: University of Chicago Press. Hogan, Patrick Colm. 2005. “Literary universals and their cultural traditions: The case of poetic imagery.” Consciousness, Literature, and the Arts 6.2. Available online at: http://www.aber.ac.uk/cla/archive/ hogan.html.

ART, LANGUAGES OF Languages of Art, a book by Nelson Goodman (1906–98), was first published in 1968, with a second edition in 1976. The present entry focuses solely on this book, which raises interesting questions about “language” in a general sense and its role in aesthetic experience. This entry does not attempt to contextualize Goodman’s book relative to his philosophy, for which see Daniel Cohnitz and Marcus Rosenberg (2006) and Catherine Z. Elgin (1992); Goodman’s later and related Ways of Worldmaking is also recommended (Goodman and Elgin 1978). By languages (of art), Goodman means more generally symbol systems; natural language is one of the symbol systems, which include, for example, musical notation or the symbol system of cubist painting. Certain characteristics of symbol systems, when used in an artwork, place cognitive demands on its audience, which make the artwork “good to think” (to borrow Claude LéviStrauss’s term). The symbol systems from which artworks are composed enable us to be exploratory, drawing on our cognitive (including emotional) resources. This is because artworks are

106

made from symbol systems that have one or more of the symptoms of the aesthetic: syntactic density, semantic density, syntactic repletenesss, and exemplification. These notions are defined later in this entry. According to Goodman, “A symbol system consists of a symbol scheme correlated with a field of reference” ([1968] 1976,143). Goodman’s primary interest in defining a symbol system is to differentiate the notational from the non-notational schemes, where “notation” is a technical notion to which his Chapter 4 is devoted. His concern about notations follows from a concern about forgeries and fakes and with the fact that some types of art (such as painting) can be faked while others (such as performance of a specific piece of music) cannot. Where a work is defined by compliance to a score (i.e., it has a notation), it cannot be faked; such works are called allographic. Where a work is not defined by compliance to a score, as in the case of a painting, its authenticity can be established only by tracing the history of its production back to its origin, and this permits faking; such works are called autographic. A symbol system is built on a symbol scheme, which consists of characters (and usually modes of combination for these characters). For example, for a natural language, a character is a class of marks, where marks might include anything from single sounds or letters up to whole spoken or written texts, as in the letter P, a character that is the class whose members are all the writings-down of the letter P. Symbol systems are either notations or not notations. If the symbol system is a notation, the characters of which its scheme is comprised must meet two conditions, as follows: (1) For a character in a notation, the members can be interchanged, where different characters can be true copies of one another; this is called the condition of characterindifference and is true, for example, of letters of the English alphabet. (2) Characters in a notation must be finitely differentiated or articulate; for a mark that does not belong to two characters, it must be theoretically possible to determine that it does not belong to at least one of them (this is explained further shortly). (1, 2) are the two syntactic requirements that define a symbol system as notational. Characters in a scheme are correlated with things outside the scheme. For example, the marks that make up a score are correlated with elements in the performance of the score; the mark that is a written word is correlated with a pronunciation of that word; and the mark that is a written word is (also and independently) correlated with that word’s referent. Goodman uses the term “complies” and says that the performance complies with the score, or the referent, or pronunciation, complies with the written word. The set of things that comply with an inscription (e.g., the set of things named that can be denoted by a name) is called the compliance class of the inscription. For the symbol system to be a notation, it must first include a symbol scheme that is notational (i.e., that satisfies the two syntactic conditions), and it must also satisfy three semantic conditions, as follows.

Art, Languages of (3) Notational systems must be unambiguous; it must be clear which object complies with each unique element of the scheme. (4) In a notation, compliance classes must be disjoint; for example, a performance cannot comply with two different scores. (5) A notational system must be semantically finitely differentiated; for an object that does not comply with two characters, it must be theoretically possible to determine that the object does not comply with at least one of them. The notion of “finite differentiation” is important both in the syntax and semantics of notational systems; finite differentiation is articulation, and its lack constitutes density. As we will see, though articulation is important for a notational system, density is more generally important in defining works as aesthetic. Finite differentiation requires gaps between elements in the system (between characters, or between compliants); if between two adjacent elements a third can always be inserted, the scheme lacks finite differentiation. For example, a scheme lacks finite differentiation if it has two characters, where all marks not longer than one inch belong to one character and all longer marks belong to the other, and where marks can be of any length. Between a mark belonging to the character of “marks not longer than one inch” and a mark belonging to the character of “longer marks,” it is always (theoretically) possible to have a third that falls between them (this ever-diminishing between-space is a kind of Derridean “mise-en-abîme”). A symbol system is called a notation if it meets the five conditions. Goodman asks whether various types of symbol systems that have been developed in the arts are notations. (A type of artistic practice may be non-notational just because no notation has been developed for it; in principle, notations might be developed for all of them, but in practice they have not been.) A traditional musical score is a character in a notational system. The compliants of the score are the performances, which collectively constitute the work of music. Similar comments are made for Labanotation, a scoring system developed for dance. A literary work is a character in a notational scheme (but not a character in a notational system): Like the language from which it is composed, it meets the syntactic requirements for a notation, but not the semantic requirements. A painting is a work that is in a symbol system that is not notational. Having developed these notions, Goodman uses them as a way of defining a “representation,” a problem raised and not solved in the first part of the book, where, for example, he argues that we cannot distinguish a representation by criteria such as resemblance to the represented object. Representation for Goodman is distinct from description (i.e., the term representation does not correspond to current cognitive science or linguistic uses, in which propositions or tree structures are representations). A description uses a symbol scheme that is (syntactically) articulate, whereas a representation uses a symbol system that is dense (or uses symbols from a dense part of a symbol scheme). He distinguishes between two types of dense (representational) schemes, differentiating a diagram from a picture. His example is a pair of representations that are visually identical, consisting of a “peaking” line, one an electrocardiogram and the other a picture

of the outline of Mount Fuji. What makes the electrocardiogram a diagram is that not every aspect of its form is relevant; the line can vary in thickness or color without constituting a different character. In contrast, every aspect of the form of the picture is relevant; pictures have much fewer contingent features than diagrams, and pictures are thus said to be relatively (syntactically) replete. The difference between diagram and picture is a matter of degree; repleteness is a relative characteristic. Goodman concludes his book by using these notions to develop four “symptoms of the aesthetic.” Objects have “aesthetic symptoms” when they use symbol systems that are syntactically dense, semantically dense, and syntactically replete. The fourth symptom of the aesthetic is that aesthetic objects exemplify. (In Ways of Worldmaking, Goodman explores the notion of style and proposes that the style of an artwork is one of the referents exemplified by its symbols, where style consists of “those features of the symbolic functioning of a work that are characteristic of author, period, place or school” [1978, 35]. In the same book, he introduces a fifth symptom of the aesthetic, which is multiple and complex reference.) Note that the first three of these symptoms are characteristic of non-notational symbol systems; all three are associated with “density” in some more general sense, which, Goodman says “arises out of, and sustains, the unsatisfiable demand for absolute precision,” thus engaging our interest in aesthetic works ([1968] 1976, 253). Goodman concludes his discussion by asking what gives an aesthetic object its value, both relative to other aesthetic objects and, more generally, to us: What makes us want to know it? He argues that aesthetic objects invite our interest by asking us to understand what they are, including how their symbol systems operate, and what they exemplify; these tasks are made particularly difficult by the four symptoms of the aesthetic, which thus particularly stimulate our interest in aesthetic objects. He summarizes three criteria, drawn from general ideas about aesthetics: Engagement with artworks improves our fitness to cope with the world, just manifests our playfulness (i.e., “homo ludens”), or communicates special kinds of knowledge to us. These are partial insights into the primary purpose of our engagement with aesthetic objects: “The primary purpose is cognition in and for itself; the practicality, pleasure, compulsion, and communicative utility all depend on this” ([1968] 1976, 258). The symbol systems – or “languages” – of art serve this purpose, allowing for the possibility of producing symbolic objects that engage us. Furthermore, the characteristic density of the symbolic systems used in artworks, and the characteristic unparaphrasability of what they express both permit a person to reenter the same artwork and repeatedly to discover new things in it. – Nigel Fabb WORKS CITED AND SUGGESTIONS FOR FURTHER READING Cohnitz, Daniel, and Marcus Rosenberg. 2006. Nelson Goodman. Montreal: McGill-Queen’s University Press. Elgin, Catherine Z. 1992 . “Depiction.” In A Companion to Aesthetics, ed. David Cooper, 113–16. Oxford: Blackwell. ———. 1992. “Nelson Goodman.” In A Companion to Aesthetics, ed. David Cooper, 175–7. Oxford: Blackwell. Goodman, Nelson. [1968] 1976. Languages of Art. 2d ed. Indianapolis: Hackett.

107

Articulatory Phonetics ———. 1988. Reconceptions in Philosophy and Other Arts and Sciences. London: Routledge. Goodman, Nelson, and Catherine Z. Elgin. 1978. Ways of Worldmaking. Indianapolis: Hackett.

ARTICULATORY PHONETICS Articulatory phonetics is that part of phonetics that studies how speech is produced by the lips, tongue, velum (soft palate), larynx, and lungs to alter the air pressures and airflows and turbulent noises in the vocal tract and to create the air spaces that yield the resonances that differentiate speech sounds. The basic vocabulary of articulatory phonetics is used for the taxonomic, that is, classificatory, description of speech sounds. For example, the initial sound in the French word père, “father,” would be described as a voiceless bilabial stop, symbolized using the IPA (International Phonetic Alphabet) symbol [p]. The initial sound in the English word pear would be described as a voiceless aspirated bilabial stop and symbolized using the IPA symbol [ph]. However, besides this essential taxonomic function, articulatory phonetics studies the mechanisms (e.g., muscular, aerodynamic) that produce speech and especially the “how and why” of variability in speech sounds. The following is a brief overview of the subject; for indepth accounts, the readings listed below should be consulted.

History The study of speech articulation and the development of a descriptive terminology has an impressive history, with the first surviving instance being the Aṣṭādhyāyī of the Sanskrit grammarian, Pāṇini (ca. 500 b.p.e.), who gave an articulatory account of the relatively large sound inventory of Sanskrit. Other notable achievements in the characterization of speech sounds were given by many Greek grammarians, notably Dionysius Thrax (first cent. b.p.e.); the Arab and Persian grammarians al Khalil Ibn Ahmad and Sībawaihi of the eighth century, who described the Arabic of their times; the First Grammarian of Iceland (twelfth cent.); and the work commissioned by and credited to the Korean King Sejong (fifteenth cent.), which provided not only an articulatory description of Korean as spoken then but also a transcription, now the official orthography for Korean, hangul, which is partially iconic in its representation of how the sounds are produced. In Europe, the Baroque and modern eras saw dozens of proposals for the description of speech sounds, for example, by John Wilkins, Johan Conrad Amman, William Holder, Francis Lodwick, Alexander J. Ellis, Robert Nares, Ernst Brücke, Richard Lepsius, Alexander Melville Bell, Henry Sweet, and Otto Jespersen. Although there is still some variation in the descriptive terms, works such as Catford (1977) and Maddieson (1984) have helped to standardize the terminology.

The Basics Speech articulations enable communication between speakers and listeners because they create sound; it is the sound transmitted to listeners and the perception of these sounds that are the ultimate goal in speaking. Descriptions of articulation are intended to capture the gestures that create these distinctive elements in the speech code.

108

Sound is short-term variations or disturbances in ambient air pressure. These pressure disturbances are created when air moves from a region of high pressure to a region of low pressure. There are three piston-like articulatory movements that can create such pressure differentials with respect to atmospheric pressure. These, which J. C. Catford calls the “initiation” mechanisms, are pulmonic, glottalic, and velaric. These mechanisms can either create a positive pressure vis-à-vis atmospheric pressure, in which case they are called egressive, or a negative pressure, and then they are called ingressive. Pulmonic egressive initiation is by far the most common. All languages use it and most use it exclusively. The chest cavity, by virtue of decreasing its volume as in normal respiratory expiration, compresses the air in the lungs, thus raising lung pressure above that of the atmospheric pressure. Since speech necessarily involves valves that impede the exiting airflow (e.g., the adducted vocal cords and/or whatever articulations are made in the oral cavity), the pulmonic or subglottal pressures developed in speech are much larger than those seen in quiet respiratory expiration. Because such initiation is so common, it is normally not included in the usual phonetic descriptions; for example, the [p] in French père, which would otherwise be described as pulmonic expiratory voiceless bilabial stop, is usually designated simply as voiceless bilabial stop. Pulmonic ingressive initiation (so-called ingressive voice) is possible and is encountered in many cultures, notably in Scandinavia and France, where short interjections, ja, oui, non, can be uttered on ingressive voice (usually with some breathiness), but although some sociolinguistic or pragmatic contrast may be associated with this trait, no language documented so far uses pulmonic ingressive initiation to make lexical contrasts. Ingressive phonation may also be encountered as a (not very effective) vocal disguise, and it is universally encountered as a kind of coda to very young babies’ cries where the vocal cords are still approximated but the respiration has shifted from expiratory to inspiratory. If the vocal cords are tightly closed and the larynx as a whole is raised, acting like a piston, while there is a complete closure in the oral cavity (and with the velum raised), a positive pressure may be developed. Such sounds, glottalic egressives or ejectives, are not uncommon, being found in various African languages (from different language families), in some languages of South and Central America and the Pacific Northwest (in the Americas), and in the Caucasus. For example, Quechua “bread” is [t’anta]. Glottalic ingressives or implosives involve the larynx – most commonly when the vocal cords are in voicing position – being lowered during the stop closure, thus creating a negative pressure in the oral cavity or at least moderating the buildup of positive pressure. Historically, such stops often come from voiced, especially geminated (long), stops, for example, Sindhi /paɓuni/ “lotus plant fruit” < Prakrit *pabːa. Enlarging the oral cavity helps to maintain a positive pressure drop across the glottis, which favors voicing. Although ejective fricatives are attested, there are no implosive fricatives – probably because the noise of a fricative is generated when the air jet expands after leaving the narrow constriction. Such expansion would occur inside the vocal tract if made implosively, and the sound would be attenuated by the oral constriction.

Articulatory Phonetics If an air pocket is trapped between the tongue and palate or the tongue dorsum and lips, and the tongue is lowered, a large negative pressure can be generated, which, when released, can create quite a loud sound. Such sounds, so-called clicks have velaric ingressive initiation. They are common in cultures all over the world as interjections, signals to animals, and so on. However, they are used as speech sounds to differentiate words only in a few languages of southern and East Africa. They are very common in the Khoisan languages and a few neighboring Bantu languages. For example, Khoi “one” is [|ui] where the [|] is a dental affricated click (the sound often symbolized in Western orthographies as “tsk” or “tut”). In the Khoisan languages, clicks can be freely combined with pulmonic and/or glottalic egressives either simultaneously or in clusters. Velaric egressives, where a positive pressure is created by upward and forward movement of the tongue, are not found in any language’s lexicon but are used in some cultures as a kind of exaggerated spitting sound, where the positive pressure creates a brief bilabial trill. In general, after the mechanism of initiation is specified for speech sounds, there are three main categories of terms to further characterize them: place, manner, qualifiers. For example, the Russian [bratj] “to take” has in word-final position a voiceless palatalized dental stop. The manner is “stop,” the place is “dental,” and “voiceless” and “palatalized” are qualifiers.

Manners There are two fundamental categories of manners, obstruent and sonorant, each with subcategories. An obstruent is a sound that substantially impedes the flow of air in the vocal tract to a degree that turbulent noise is generated either as continuous frication or as a noise burst. Obstruents may be stops or fricatives. (Ejectives, implosives, and clicks are inherent stops.) Sonorants, which do not impede airflow, are subdivided generally into laterals, glides, approximants, nasals, and vowels. Stops present a complete blockage of the airflow, for example, the glottal stop in the name of the Hawai’ian island O’ahu [oʔahu]. A special subclass of stops are affricates, which are stops with a fricative release, as in the initial and final consonants of English judge [ʤ͡ʌʤ͡]. Another subclass, often categorized in other ways, is comprised of trills, for example, the initial sound of Spanish roja “red” [roxa]. Fricatives present a partial blockage, but with noise generated due to air being forced through a relative narrow constriction, for example, the velar fricative in Dutch groot “large” [xʁot]. Fricatives may be further divided into sibilants (s-like fricatives) made by an apical constriction on or near the alveolar ridge. The essential characteristic of this class, as opposed to other nonsibilant fricatives, is relatively loud high-frequency noise that exploits the small downstream resonating cavity, for example, English schist “a category of rock” [ʃɪst] which contains two different sibilant fricatives. Nonsibilant fricatives either have no downstream resonating cavity (e.g., the bilabial fricative in Japanese Fuji “name of the famous mountain” [ϕuʤ͡ i]) or, like the velar fricative, are made further upstream of the alveolar region and so have a longer downstream resonating cavity and, thus, lower frequencies. Presence or absence of voicing also affects fricatives’ loudness: Even if paired voiced and voiceless fricatives have the same degree of constriction, voiceless fricatives will have more

intense frication noise, because with no resistance to the airflow at the glottis, the velocity of airflow will be greater at the oral constriction, and that also affects the degree and loudness from the air turbulence. Subcategories of sonorants include laterals, where the constriction is on one side of the palate, the other being open, for example, the medial geminate (or long) alveolar lateral in Hindi palla “loose end of a sari used as a head covering” [pʌlːa]. Nasals are consonants made with a complete closure in the oral cavity (at any place farther forward of the uvular region) but with a lowered velum, for example, Tswana [ŋku] “sheep” with an initial velar nasal. Glides and approximants have nonlateral oral constrictions that are not sufficient to generate turbulence, for example, the labial-velar glide at the beginning of the English word [wɚd]. Vowels are considered to have the least constriction (descriptive terms follow).

Place The primary places of articulation of speech sounds, proceeding from the farthest back place to the farthest forward: glottal, pharyngeal, uvular, velar, palatal, alveolar, dental, and labial. Some of these places have already been illustrated. Finer place distinctions can easily be made if necessary by appending the prefixes pre- and post-, and multiple simultaneous constrictions can be differentiated by concatenating these terms as was done with the labial-velar glide [w]. In most cases, these anatomical landmarks on the upper side of the vocal tract are sufficient; if necessary, an indication of the lower (movable) articulator can be specified, for example, the voiced labial-dental fricative [v] as in French voir “to see” [vwɑʁ] (as opposed to, say, the voiced bilabial fricative [β] in Spanish cerveza “beer” [sɛɾβesa]).

State of the Glottis In most cases, specifying whether the vocal cords are apart and not vibrating (voiceless), lightly approximated and vibrating (voiced), or tightly pressed together (glottal stop) is sufficient. However, voicing itself occasionally needs to be further differentiated as breathy (a more lax type of voicing lacking energy in the higher harmonics), tense (rich in higher harmonics), or creaky (irregular staccato type of phonation, also with much energy in the higher frequencies, though since it is irregular, one cannot clearly identify harmonics as such). Many of the Indic languages employ a distinctive breathy voice associated with voiced stops, for example, Hindi bhāshā “language” [b̤ɑ:ʃɑ:], and many languages and many speakers’ voiced phonation changes to creaky at a point of low F0 in intonation. (Low F0 is a low rate of vibration of the vocal cords due to lesser tension giving rise to low pitch or “note” of the voice.) Creaky voice is also a common variant of glottal stop.

Vowels The descriptors for vowels deviate from those for consonants. An imaginary quadrilateral space in the mouth (seen sagittally) is posited, and vowels are said to have the high point of tongue at regions in this space whose dimensions vertically are high – mid – low and horizontally, front – central – back. (These may also have further qualifiers. In French the high front unrounded vowel contrasts with a high front rounded vowel [i] and with a

109

Articulatory Phonetics high back rounded vowel [u], e.g., dit “(s/he) said” [di] vs. du “of the” [dy] vs. doux “soft” [du].) Although ostensibly referring to anatomical features, it is now accepted that these descriptors actually correspond to acoustic-auditory features of vowels: The height dimension correlates inversely with their first formant, the front dimension correlates directly with their second formant, and unrounded-rounded correlates roughly with their third formant. It is still technically possible to apply the traditional anatomical-physiological descriptors to vowels, in which case [i] would be a close palatal vowel with spread lips, and [u] a close velar vowel with lip rounding, and [ɑ] a pharyngeal vowel with a widely open mouth shape. Other vowels would just be variants of these with either less constriction or intermediate places of constriction. There is merit in applying the anatomicalphysiological labels, for example, to explain the Danish dialectal variant pronunciations [bi] ~ [biç] “bee.” The latter variant with the voiceless palatal fricative can arise simply from the vowel terminus being devoiced (since, it will be recalled, the degree of turbulence is determined not only by the degree of closure but also by the velocity of the airflow, which, in a voiceless vowel, is high to generate turbulence at the point of constriction).

Secondary Articulations or Modifications There are dozens of ways a speech sound can be qualified. Typically, these are additional modifications or lesser constrictions that can be done simultaneously with the primary constriction, or are in such close temporal proximity to or invariably linked to it that they are considered inherent to it. The label breathy voiced is an example. Some additional examples (where the italicized term is the qualifier): voiceless aspirated velar stop, as in English key [khi]; the English phoneme /ʃ/, often phonetically a voiceless post-alveolar labialized fricative as in ship [ʃwip]; the nasalized mid-front vowel as in French faim “hunger” [f].

Prosody, Tone, Intonation The terminology describing distinctive uses of voice pitch and relative timing of sounds (and, perhaps, different voice qualities) is still relatively nonstandardized except in the case of tones. The International Phonetic Association’s transcription recognizes a variety of possible tone shapes, for example, Thai [kha\|] (with falling tone) “servant” versus [kha/|] (with rising tone) “leg.” Here, the vertical line is supposed to represent the range of voice pitch characteristic of the speaker, the sentence context, and so on, and the attached line the range and direction of the distinctive pitch modulation.

Beyond Taxonomy The conventional descriptions of speech just reviewed form the basis for scientific investigations of considerable sophistication and with applications in fields as diverse as medicine (especially speech pathology), man-machine communication, first (and subsequent) language learning, and phonology. These investigations involve study of more than just the anatomical-physiological character of speech sounds, but also, as was hinted at in the preceding discussion, speech aerodynamics, speech acoustics, speech perception, and neurophonetics. Space allows just one example in the area of phonology: Medial stops emerge seemingly out of nowhere in words such as glimpse < gleam + s (nominalizing

110

Artificial Languages suffix), dempster “judge” < deem + ster, Thompson > Thom + son, youngster [jʌŋkstɚ] < young [jʌŋ] + ster. One has to ask where the medial stop came from in these nasal + fricative clusters, neither element of which is a stop. The answer emerges when one considers that these speech sounds are opposite in the state of the oral and velic exit valves. The nasal has all oral exit valves closed and the velic valve open whereas the fricative has an oral valve open, and the nasal valve closed. If in the transition between the nasal and fricative the velic valve should close prematurely, then all exit valves will be closed and thus a brief epiphenomenal stop will emerge. (For more examples, see Ohala 1997.) – John Ohala WORKS CITED AND SUGGESTIONS FOR FURTHER READING Browman, C. P., and L. Goldstein. 1986. “Towards an articulatory phonology.” In Phonology Yearbook 3, ed. C. Ewan and J. Anderson, 219–52. Cambridge: Cambridge University Press. Catford, J. C. 1977. Fundamental Problems in Phonetics. Bloomington, IN: Indiana University Press. Goldstein, L., and C. P. Browman. 1986. “Representation of voicing contrasts using articulatory gestures.” Journal of Phonetics 14: 339–42. Hardcastle, W. J., and J. Laver. 1999. The Handbook of Phonetic Sciences. Oxford: Blackwells. Huffman, M. K. and R. Krakow, eds. 1993. Nasals, Nasalization and the Velum: Phonetics and Phonology, Vol. 5. San Diego, CA: Academic Press. Ladefoged, P. 1964. A Phonetic Study of West African Languages: An Auditory-Instrumental Survey. Cambridge: Cambridge University Press. Ladefoged, P., and I. Maddieson. 1995. The Sounds of the World’s Languages. Oxford: Blackwell. MacNeilage, P. F., ed. 1983. The Production of Speech. New York: SpringerVerlag. Maddieson, I. 1984. Patterns of Sounds. Cambridge: Cambridge University Press. Ohala, J. J. 1990. “Respiratory activity in speech.” In Speech Production and Speech Modelling, ed. W. J. Hardcastle A. Marchal, 23–53. Dordrecht, the Netherlands: Kluwer. ———. 1997. “Emergent stops.” Proc. 4th Seoul International Conference on Linguistics [SICOL] 11–15 Aug 1997: 84–91. Rothenberg, M. 1968. The Breath-Stream Dynamics of Simple-ReleasedPlosive Production. Basel: Karger. Silverman, D. 2006. A Critical Introduction to Phonology: Of Sound, Mind, and Body. London and New York: Continuum International Publishing Group. Solé, M-J. 2002. “Aerodynamic characteristics of trills and phonological patterning.” Journal of Phonetics 30: 655–88.

ARTIFICIAL LANGUAGES An artificial language can be defined as a language, or languagelike system, that has not evolved in the usual way that natural languages such as English have; that is, its creation is due to conscious human action. However, this definition leaves open some questions. For one thing, what do we mean by “language or language-like system”? Among the systems of communication that could be, and have been, called artificial languages are systems of logic, for example, predicate calculus, and computer languages, such as BASIC. However, the functions of these languages are different from the function of natural languages,

Artificial Languages which is communication among humans. I, therefore, focus on artificial languages that have this latter function, for example, Esperanto. Under the heading of artificial languages, one might also include languages that have been made up in connection with novels, films, television programs, and so on, for example, Klingon (fictional or imaginary languages), or as part of some other imaginary world, or those that have been created for the enjoyment of their designer (personal languages). Some languages (philosophical languages) were designed to reflect the real world better than natural languages. Some of the earliest known ideas on artificial languages, from the seventeenth century, involve this type. The terms constructed language (or conlang) and planned language are roughly equivalent to artificial language (although one could point out that some natural languages have undergone a degree of planning), while (international) auxiliary language covers only those languages intended for international communication (of course, some natural languages are also used for this); many, if not most, artificial languages have been created for this purpose. Another question concerns our notions of artificial and natural. On the one hand, many (arguably all) natural languages have been subjected to some human manipulation. Consider, for example, the long line of English prescriptivists who have tried to eliminate some constructions of the language, or organizations such as the French Academy, which has attempted to keep some English words out of French. Although many of these manipulations have not completely succeeded, they have had some effect, and therefore one could argue that English and French are partly artificial. On the other hand, many consciously created languages were built from elements of one or several natural languages and could thus be considered not entirely artificial. Therefore, the boundary between natural and artificial languages is not entirely clear. In fact, a common classification of artificial languages is in terms of whether they are based on natural languages: a posteriori languages are, while a priori languages are not (the philosophical languages belonging to the second group). That is, a priori languages are (supposedly) built “from scratch,” not taking anything from natural languages. This is a simplification, as few, if any, languages are entirely a priori; many contain both a posteriori and a priori components. Therefore, the distinction should, rather, be seen as a spectrum, with languages at different points having differing ratios of a priori and a posteriori components. Artificial languages consisting of substantial proportions of both types are called mixed languages. Esperanto stands far above other artificial languages in terms of success – it has vastly more speakers than any others (and even some native speakers). It has been claimed to have more than a million speakers, though some would disagree with such a large number, and of course, the question hinges on how one defines a speaker. Only a relatively small number of artificial languages have achieved much of a community of speakers. These include Volapük, Interlingua, and Ido, the latter being a modified Esperanto. Many artificial languages were not used by anyone other than their designer and perhaps several other people. In fact, a large number of artificial languages were never fully developed, with only incomplete descriptions having been published. Let us now see some examples of sentences in artificial languages. Because the a priori languages do not (intentionally) use

Aspect elements from any natural languages, on the surface they may seem rather strange, as shown by the following examples from the language Oz (which, in spite of its name, was a serious project): (1)

ap if-blEn-vOs he HABITUAL-seldom-study “he seldom studies”(Elam 1932, 20)

(2)

ep ip-Qks ap I PAST-see him “I saw him” (ibid.)

However, one could assert that since even a priori languages are human creations, they cannot be that different from natural languages. A posteriori languages can draw from several languages or from just one. In the latter case, they are usually or always simplifications of the language. There have been many such simplifications of Latin, some of English, and some of other languages. Following is a Latin sentence and its equivalent in SPL (or SIMPLATINA), an artificial language created from Latin. (3)

Nuntium audiverat antequam domum venit. Fin audit núntium ántequam fin venit in domus. “He had heard the news before he came home.” (Dominicus 1982, 21)

One might be surprised to learn that there are a thousand or more artificial languages, even excluding fictional ones (but including languages that were not fully elaborated). It might also be unexpected that people have continued to devise new artificial languages for international communication, given how many have already been proposed and not achieved their goal. The existence of the Internet may have served as an impetus, since it is now easy for language creators to present their languages to a wide audience. The number of artificial languages will probably keep increasing, though with none of them achieving the status of a universal second language. – Alan Reed Libert WORKS CITED AND SUGGESTIONS FOR FURTHER READING Albani, Paolo, and Berlinghiero Buonarroti. 1994. Aga Magéra Difúra. Bologna: Zanichelli. An encyclopaedia with many entries for artificial languages, including fictional languages, and their creators. Dominicus, Richardius. 1982. SPL. Wisconsin: Dominicus Publishing House. Elam, Charles Milton. 1932. The Case for an A Priori Language. Cincinnati, OH: The Open Sesame Press. Large, Andrew. 1985. The Artificial Language Movement. Oxford: Basil Blackwell. Pei, Mario. 1968. One Language for the World. New York: Biblo and Tannen.

ASPECT Situations unfold over time. When we talk about them, we often specify how they unfold over time (or not). There are many ways in which language conveys such temporal information. While tense specifies location of an event in relation to other points in time (e.g., past, present, future, pluperfect), aspect specifies

111

Aspect internal temporal structure of a situation (e.g., whether it is ongoing or completed). This is important information to convey for our linguistic communication to be successful, and many languages convey it by various means – lexical, grammatical, and/ or pragmatic. English grammatically marks tense (-ed, -s, will), while Chinese does not, relying instead on lexical and pragmatic means. English also grammatically marks aspect (progressive “be V-ing”), while Hebrew does not. Grammatical marking of aspect, often encoded in auxiliaries and inflections, is known as grammatical aspect, or viewpoint aspect. It is called viewpoint aspect because it signifies the speaker’s viewpoint. When one chooses to say “He ran a mile,” one is viewing the situation from outside, disregarding its internal structure (perfective aspect), while if one says “He was running a mile,” the beginning and end of this situation are disregarded and one is focusing on the internal structure of this situation (imperfective aspect) (Comrie 1976; Smith 1997). The former is often used to push the narrative storyline forward (foreground), while the latter is associated with background information (Hopper 1979). Equally important in conveying aspectual information is lexical aspect – also known as inherent (lexical) aspect, situation aspect (or situation type), aktionsart, event type, and so on. This is defined by the temporal semantic characteristic of the verb (and its associated elements) that refers to a particular situation. Although there are numerous proposals, the most well known is the classification proposed by Zeno Vendler (1957): Achievement: that which takes place instantaneously, and is reducible to a single point in time (e.g., recognize, die, reach the summit) Accomplishment: that which has dynamic duration, but has a single clear inherent endpoint (e.g., run a mile, make a chair, walk to the store) Activity: that which has dynamic duration, but with an arbitrary endpoint, and is homogeneous in its structure (e.g., run, sing, play, dance) State: that which has no dynamics, and continues without additional effort or energy being applied (e.g., see, love, hate, want) Lexical aspect has proved to be important in linguistic analysis, acquisition, and processing of aspect. In linguistic analysis, Carlota S. Smith (1997) proposed the two-component theory, a system in which the aspectual meaning of a sentence is determined by the interaction between lexical aspect and grammatical aspect. For example, imperfective aspect (e.g., progressive in English) takes the internal view, and, therefore, it is compatible with durative predicates of activity and accomplishment and yields progressive meaning. In contrast, achievement, since it is nondurative, is not so compatible with imperfective aspect, and such pairing is often anomalous (e.g., *He is noticing the error) or results in preliminary stages meaning (e.g., He is dying). In acquisition, this interaction of lexical and grammatical aspect has been observed since the 1970s. Cross-linguistically, when children acquire (perfective) past tense marking, they show strong association between telic verbs (achievements and accomplishments), between general imperfective marking (such

112

as French imparfait) and atelic verbs (states and activities), and between progressive (i.e., dynamic imperfective) marking and activity verbs (see Li and Shirai 2000 for a review). Psychologists and linguists alike have tried to explain this observation. One important proposal relies on innateness (the language bioprogram hypothesis, Bickerton 1981), while an alternative proposal is based on input frequency (Shirai and Andersen 1995). The notion of compatibility is crucial when we discuss the interaction of lexical and grammatical aspect since some combinations are more natural, prototypical, and frequent. Telic verbs are more compatible with perfective aspect, while activities are most naturally associated with progressive marking. This is reflected in frequency distribution cross-linguistically (Andersen and Shirai 1996). For example, about 60 percent of past tense markers in child-directed speech in English were attached to achievement verbs, while almost 95 percent of past tense forms in children’s speech were used with achievement verbs (e.g., broke, dropped) when children started using them (Shirai and Andersen 1995). This frequency effect is not yet well recognized in the area of language processing. Carol J. Madden and Rolf A. Zwaan (2003) and Todd Feretti, Marta Kutas, and Ken McRae (2007) found the strong effect of grammatical aspect in their experiments on aspectual processing, but they did not manipulate the effect of lexical aspect. Although Madden and Zwaan (2003) found a facilitating effect of perfective aspect on processing but not of imperfective aspect, their experiments used only accomplishment verbs, which are telic and more compatible with perfective aspect (i.e., past tense in English). Foong Ha Yap and colleagues (2009) replicated facilitating effects of perfective aspect with accomplishments and, in addition, of imperfective (progressive) aspect with activities in Cantonese. Thus, the interaction of lexical and grammatical aspect is pervasive and cannot be ignored in any research involving aspectual phenomena. – Yasuhiro Shirai WORKS CITED AND SUGGESTIONS FOR FURTHER READING Andersen, Roger W., and Yasuhiro Shirai. 1996. “Primacy of aspect in first and second language acquisition: The Pidgin/Creole connection.” In Handbook of Second Language Acquisition, ed. W. Ritchie and T. Bhatia, 527–70. San Diego, CA: Academic Press. Bickerton, Derek. 1981. Roots of Language. Ann Arbor, MI: Karoma. Comrie, Bernard. 1976. Aspect. Cambridge: Cambridge University Press. Ferretti, Todd R., Marta Kutas, and Ken McRae. 2007. “Verb aspect and the activation of event knowledge.” Journal of Experimental Psychology: Learning, Memory, and Cognition 33.1: 182–96. Hopper, Paul J. 1979. “Aspect and foregrounding in discourse.” In Syntax and Semantics. Vol. 12: Discourse and Syntax. ed. T. Givon, 213–41. New York: Academic Press. Li, Ping, and Yasuhiro Shirai. 2000. The Acquisition of Lexical and Grammatical Aspect. Berlin: Mouton de Gruyter. Madden, Carol. J., and Rolf A. Zwaan. 2003. “How does verb aspect constrain event representation?” Memory & Cognition 31: 663–72. Shirai, Yasuhiro, and Roger W. Andersen. 1995. “The acquisition of tenseaspect morphology: A prototype account.” Language 71: 743–62. Smith, Carlota S. 1997. The Parameter of Aspect. 2d ed. Dordrecht, the Netherlands: Kluwer. Vendler, Zeno. 1957. “Verbs and times.” Philosophical Review 66: 143–60.

Auditory Processing Yap, Foong Ha, Patrick Chun Kau Chu, Emily Sze Man Yiu, Stella Fay Wong, Stella Wing Man Kwan, Stephen Matthews, Li Hai Tan, Ping Li, and Yasuhiro Shirai. 2009. “Aspectual asymmetries in the mental representation of events: Role of lexical and grammatical aspect.” Memory & Cognition 37: 587–95.

AUDITORY PROCESSING Auditory processing refers to the cognitive processes that enable a listener to extract a message from the raw material of a speech signal. The study of auditory processing draws upon a range of sources within the linguistic sciences: most notably cognitive psychology, discourse analysis, phonetics, phonology and neurolinguistics (see brain and language). It is distinct from theories of empathetic listening (what makes a “good listener”) in areas such as counseling. It was not until the 1960s that a significant body of listening research developed with the advent of more sophisticated recording equipment and the increased availability of spectrograms to display the physical characteristics of the speech signal (see acoustic phonetics). The many advances in our understanding of the skill since then include early work on phoneme perception by the Haskins Laboratories; the recognition that processing occurs on line rather than waiting for an utterance to be completed; and insights into how listeners identify word boundaries in connected speech. Evidence of the extent to which listeners have to build a message for themselves on the basis of inference has resulted in a sharp move away from a view of listening as a “passive” skill and the recognition that a listener actively engages in a process of meaning construction. Auditory processing falls into two closely linked operations: Decoding, where acoustic stimuli are translated into linguistic units; Meaning construction, which embellishes the bare meaning of the utterance by reference to knowledge sources outside the signal. It also requires listener decisions as to the importance of what has been said and how it is linked to the discourse that preceded it. The listener thus draws upon four information sources. The first is perceptual, based on the signal reaching the listener’s ear. The second is linguistic, consisting of the listener’s stored knowledge of the phonology, lexis, and syntax of the language being spoken. The third is external: drawing upon the listener’s knowledge of the world, the speaker, the topic, and the type of situation. A final component is the listener’s ongoing model of what has been said so far.

Decoding Decoding is principally a matching operation in which evidence from the signal is mapped onto stored representations in the listener’s mind of the phonemes, words, and recurrent chunks of a language (see mapping). The process was once represented as taking place in a linear and bottom-up way, with phonemes shaped into syllables, syllables into words, and words into clauses. In fact, listeners appear to draw upon several levels of representation at once, with their knowledge of higher-level units,

such as syllables, words, and formulaic chunks, influencing their decisions as to what has been heard. Decoding is also assisted by context, in the form of world knowledge, knowledge of the speaker, and recall of what the speaker has said so far. There has been much discussion as to whether these pieces of information combine in the mind of the listener during decoding or whether they are handled separately. The argument behind the first (interactive) view is that all the evidence can be considered simultaneously; the argument behind the second view (modularity) is that the processor operates more rapidly if it employs localized criteria specific to the phoneme, the word, or the context. Decoding can be discussed at three levels. The first is phoneme recognition (see speech perception). Translating acoustic evidence into the sounds of the target language does not involve simple one-to-one matching. There is, firstly, an issue of noninvariance: researchers have not succeeded in finding clusters of cues that uniquely identify individual phonemes. Indeed, they have discovered that the same set of cues may be interpreted differently according to the phonemes that precede and follow them. There is also an issue of nonlinearity: The phonemes within a word are not clearly bounded units but blend into each other in a process known as co-articulation. A further complication is speaker variation: A listener has to adjust to differences between individual speakers in pitch of voice, accent, speech rate, and so on. One solution holds that listeners employ a more reliable unit of analysis than the phoneme. They might map direct from acoustic stimuli to words stored in their minds, or they might use the syllable as their principal perceptual unit. Another solution views phoneme recognition as the outcome of cue trading, where the listener weighs competing evidence until a particular candidate emerges as the most likely. These accounts tend to assume that we have in our minds a set of idealized templates for the sounds of a language and match imperfect real-life examples to them by normalization – by editing out features that are nonstandard or irrelevant. An alternative approach shifts the focus away from processing and onto how the listener represents the sounds. A variant of the template notion suggests that a phoneme may be represented in an underspecified way that stores only a few features essential to its recognition. A second possibility is that phonemes are stored as prototypes with a range of permissible variation associated with each. But there is now increasing support for a view that we do not construct a central representation of a phoneme but instead store multiple examples of the words we hear, uttered in their variant forms. This exemplar account accords with evidence that the human mind is better at storing massive amounts of information than was previously supposed. It explains how we are able to adjust to unfamiliar accents in our native language and to recognize the same word uttered in a range of different voices. The second level is word recognition. The listening process takes place on line, with a listener able to shadow (repeat) what a speaker says at a delay of about a quarter of a second. Cohort theory (Marslen-Wilson 1987) postulates that a listener retrieves a bank of possible word matches when the initial phonemes of a word are uttered, then gradually narrows them down

113

Auditory Processing as more of the word is heard. The correct item is identified when the word’s uniqueness point is reached and the cohort is reduced to one possible match. However, many words do not have a uniqueness point (the sequence man might be a complete word or the first syllable of manner or manager), and there are no consistent gaps between words in connected speech to mark where boundaries fall. Locating boundaries (lexical segmentation) is unproblematic when one is listening to languages that bear fixed stress on the first, penultimate, or last syllable of the word but becomes an issue when processing languages with variable stress. Research suggests that listeners exploit certain prosodic features of these languages in order to establish the most likely points for words to begin or end. In English, they take advantage of the fact that the majority of content words in running speech are monosyllabic or begin with a strong syllable (Cutler 1990). A further problem for lexical recognition is that many words in connected speech (particularly function words) occur in a reduced form. They might be brief, of low saliency, and very different from their citation forms. Using the gating method, which presents connected speech in gradually increasing segments, researchers have demonstrated that word identification in listening is sometimes a retrospective process, with listeners unable to identify a word correctly and confidently until two or more syllables after its offset. There have also been attempts to model lexical recognition using connectionist computer programs that analyze spoken input in brief time slices rather than syllables. Early matches are continually revised as evidence accumulates across slices (see connectionist models, language structure, and representation). The most well known of these programs is TRACE (McClelland and Elman 1986). The speed with which a word is identified by a listener is subject to variation. High-frequency words are more rapidly retrieved than low-frequency ones and are said to be more easily activated. Recognition is also faster when the listener has recently heard a word that is closely associated with the target. Thus, encountering a word like doctor facilitates (or primes; see priming, semantic ) later recognition of nurse, patient, and hospital. This process, known as spreading activation, is highly automatic and is distinct from the normal effects of context. Syntactic parsing is the third level. Though speech is received linearly, syllable by syllable, a listener needs to build larger-scale syntactic structures from it. Listeners appear to retain a verbatim record of the words heard until a major syntactic boundary. A wrap-up process then turns the words into an abstract proposition, and they cease to be available to report. Intonation contours frequently coincide with syntactic units and assist listeners in locating clause and sentence boundaries. Where there is ambiguity early in an utterance, a listener has to carry forward parallel hypotheses about how the utterance will end, with one prioritized and the others held in reserve. Researchers employ garden path sentences (example: The lawyer questioned … by the judge apologized) to establish the criteria that influence the preferred interpretation. Listeners appear to be swayed by multiple factors, including syntactic simplicity, semantic probability, and argument structure.

114

Decoding in one’s first language (L1) is highly automatic – which is why decoding skills in a second language are difficult to acquire. Studying vocabulary and syntax is not sufficient; a listener needs to recognize the relevant linguistic forms as they occur in connected speech. The listener is likely to perceive the sounds of the second language by reference to the phoneme categories of the first and may also transfer processing routines, such as the L1’s lexical segmentation strategy or the relative importance it accords to word order, inflection, and animacy. Second language listeners often find themselves heavily dependent upon contextual cues in order to compensate strategically for failures of decoding.

Meaning Construction The outcome of decoding is an abstract proposition, which represents the literal meaning of the utterance independently of the context. A listener has to build a more complex meaning representation (or mental model), which (a) adds to and contextualizes the proposition; (b) links it conceptually to what has gone before. This operation takes place locally as well as at a discourse level. Listeners need to make local connections between ideas, associating pronouns with their referents and recognizing logical connectives (in addition, however). But they also have to carry forward a developing representation of the whole discourse so far. Meaning construction embraces several different processes. Many of them are more cognitively demanding in listening than in reading because the listener is entirely dependent upon the mental representation that he/she has built up and cannot look back to check understanding. ENRICHMENT. The listener adds depth and relevance to the proposition by drawing upon external information: knowledge of the world, the topic, the speaker, and the current situation. Understanding is also deepened by relating the proposition to the current topic and to the points made so far by the speaker. INFERENCE. Listeners supply connections that the speaker does not make explicitly. They might employ scripts to provide default components for common activities. If a speaker mentions going to a restaurant, the listener takes for granted a waiter, a menu, and a conventional procedure for ordering. SELECTION. Listeners do not simply record facts; they select some, they omit some, and they store some in reduced form. The same utterance may result in differently constituted messages in the minds of different listeners. One important factor is the listener’s perception of the intentions of the speaker. Another is the listener’s own purpose for listening. A further consideration is redundancy: Spoken discourse is often repetitive, with the speaker reiterating, rephrasing, or revisiting information that has already been expressed. INTEGRATION. Listeners integrate the incoming information into what has been heard so far. Heed is paid to whether it extends an established topic or whether it initiates a new one. SELF-MONITORING. Listeners check to see whether incoming information is consistent with the meaning representation built

Autism and Language up so far. If it is not, one or the other needs to be adjusted, or a comprehension check needs to be made. DISCOURSE STRUCTURE. Listeners impose an argument structure upon the meaning representation, with major points distinguished from minor. Here, they may be assisted by analogy with previous speech events. – John Field WORKS CITED AND SUGGESTIONS FOR FURTHER READING Brown, Gillian. 1995. Speakers, Listeners and Communication. Cambridge: Cambridge University Press. Discourse-based account of meaning construction in listening. Cutler, Anne. 1990. “Exploiting prosodic possibilities in speech segmentation.” In Cognitive Models of Speech Processing, ed. G. Altmann, 105–21. Cambridge, MA: MIT Press. Garrod, Simon, and Martin Pickering, eds. 1999. Language Processing. New York: Psychology Press. Papers on lexical, syntactic, and discourse processing. Marslen-Wilson, William. 1987. “Functional parallelism in spoken wordrecognition.” Cognition 25: 71–102. McClelland, J. L., and J. L. Elman. 1986. “The TRACE model of speech perception.” Cognitive Psychology 18: 1–86. Miller, Joanne L., and Peter D. Eimas, eds. 1995. Speech, Language and Communication. San Diego, CA: Academic Press. Papers by L. Nygaard et al. and Anne Cutler review issues in perception. Pisoni, David B., and Robert E. Remez. 2005. The Handbook of Speech Perception. Oxford: Blackwell. Papers covering most major issues in decoding.

AUTISM AND LANGUAGE Autism is a neurodevelopmental disorder that is among the most prominent of disorders affecting language. While its causes are unknown, research has focused on cognitive, neurological, and genetic explanations. Autism affects more than one domain of functioning, with language and communication as primary deficits. Since Leo Kanner published the first account of children with autism in 1943, widening diagnostic criteria have increased the identification of cases. There have also been dramatic changes in classification: Autism is no longer regarded as an isolated disorder but includes Asperger syndrome and atypical autism under the rubric autism spectrum disorder (ASD) in order to reflect that variability in expression. Diagnoses along the spectrum are characterized by a common set of features: impairments in social communication, restricted interests, and repetitive activities, with behaviors varying at different ages as well as different levels of functioning (DSM-IV, American Psychiatric Association 1994). Autism occurs in at least 0.2 percent of the population, affecting three times more males than females, while the other disorders on the spectrum are estimated to affect another 0.4% (Fombonne et al. 2006). The social communication problems in ASD vary widely. Parents of young children later diagnosed with ASD often observe an absence of simple communicative behaviors, such as shared attention (e.g., pointing to something to share interest) and make-believe play. Although many children with autism never acquire functional speech, others develop speech that

differs remarkably from that of age-matched peers. Speech characteristics typical of autism include pronoun reversal (referring to self as “you”); unvaried or atypical intonation; neologisms (Volden and Lord 1991); the use of stereotyped, repetitive, and idiosyncratic language; and echolalia. Barry Prizant and Judith Duchan (1981, 246) suggest that echolalia may serve important communicative and cognitive functions, such as turn-taking for people with autism. Significantly, social communication in ASD often fails even in the presence of apparently intact grammatical skills. This can be seen in Asperger syndrome, where language skills can be advanced, vocabulary extensive, and syntax formally correct, even “bookish.” The speech of individuals with Asperger syndrome is often pedantic, exhibiting unvaried, stereotyped phrases and expressions associated with contexts or registers not presupposed by the immediate situation of talk. The speech patterns associated with ASD are part of the broader spectrum of impaired reciprocal social interaction. Conversation may be the most difficult area of communication for people with ASD. Conventional rules of turn-taking are often ignored. Speakers may fail to sustain conversation beyond yes/no answers or speak at length on circumscribed interests, and they may resist attempts to shift topic. Speakers may also fail to attend to the conversational needs of listeners and may have difficulty applying contextual and cultural knowledge in conversation. They may thus encounter problems interpreting deictic references, as the following example illustrates: Speaker 1: What did you do on the weekend? Speaker 2: What weekend? Here, the conventional response to the question posed would be to interpret the weekend as the one that had just passed. Such problems with relevance appear to be related to the tendency in ASD toward an overliteral understanding of communication, including difficulties interpreting indirect requests and metaphor (Happé 1993). A number of cognitive theories are currently being explored to explain the core features of ASD. Executive dysfunction is one widely accepted cognitive explanation for some behavior difficulties in ASD. This refers to decision-making processes that are necessary for performing goal-directed activities, which are thought to originate in the frontal lobes (Russell 1997). Weak central coherence theory posits a detail-oriented processing style at the expense of global and contextual information and alludes to poor connectivity between brain regions (Happé and Frith 2006). Intriguingly, this information-processing style can often lead to superior performance on certain tasks, such as the Embedded Figures Task (Witkin et al. 1971), underscoring the fact that ASD is not merely a set of impairments but involves unique ways of processing information. The theory most frequently cited to explain communication difficulties in ASD is theory of mind (ToM) (Baron-Cohen 1995). ToM explains these difficulties in terms of a cognitive mechanism underlying the ability to recognize others’ mental states. Many of the pragmatic impairments that are known to occur in ASD can be linked to a lack of intuitive mentalizing ability, for example, difficulties understanding pretense, irony, deception, and nonliteral language. The ToM hypothesis does not preclude the presence of assets and islets of ability as suggested by weak central coherence theory. Cognitive theories and hypothesized

115

Autonomy of Syntax neural correlates with respect to facial and emotion information processing in the amygdala have so far provided the most compelling explanations for the communication impairments seen in ASD. Research into genetic causes appears promising, since some of the strongest genetic effects in autism seem related to language abilities. – Jessica de Villiers WORKS CITED AND SUGGESTIONS FOR FURTHER READING American Psychiatric Association. 1994. Diagnostic and Statistical Manual of Mental Disorders (DSM-IV). 4th ed. Washington, DC: American Psychiatric Association. Baron-Cohen, Simon. 1995. Mindblindness: An Essay on Autism and Theory of Mind. Cambridge, MA: MIT Press. Fombonne, Eric, Rita Zakarian, Andrew Bennett, Linyan Meng, and Diane McLean-Heywood. 2006. “Pervasive developmental disorders in Montreal, Quebec, Canada: Prevalence and links with immunizations.” Pediatrics 118.1: 139–50. Frith, Uta. 2003. Autism: Explaining the Enigma. 2d ed. Oxford: Blackwell. Happé, Francesca. 1993. “Communicative competence and theory of mind in autism: A test of relevance theory.” Cognition 48.2: 101–19. Happé, Francesca, and Uta Frith. 2006. “The weak coherence account: Detail-focused cognitive style in autism spectrum disorders.” Journal of Autism and Developmental Disorders 36.1: 5–25. Kana, Rajesh K, Timothy A. Keller, Vladimir L. Cherkassky, Nancy J. Minshew, and Marcel Adam Just. 2006. “Sentence comprehension in autism: Thinking in pictures with decreased functional connectivity.” Brain 129: 2484–93. Kanner, Leo. 1943. “Autistic disturbances of affective contact.” Nervous Child 2: 217–50. Prizant, Barry, and Judith Duchan. 1981. “The functions of immediate echolalia in autistic children.” Journal of Speech and Hearing Disorders 46: 241–9. Russell, James, ed. 1997. Autism as an Executive Disorder. Oxford: Oxford University Press. Volden, Joanne, and Catherine Lord. 1991. “Neologisms and idiosyncratic language in autistic speakers.” Journal of Autism and Developmental Disorders 21.2: 109–30. Witkin, H., P. Oltman, E. Raskin, and S. Karp. 1971. A Manual for the Embedded Figures Test. Palo Alto, CA: Consulting Psychologists Press.

AUTONOMY OF SYNTAX Autonomy of syntax refers to what in recent times has been the dominant assumption concerning the formulation of syntactic regularities: syntax is determined independently of phonological realization or semantic interpretation. The formal properties of syntax are manipulated purely formally. Such an assumption is familiar to modern students of linguistics from numerous textbook presentations, such as Andrew Radford’s (1988, 31): autonomous syntax principle. No syntactic rule can make reference to pragmatic, phonological, or semantic information.

Some such assumption is already in place in Noam Chomsky (1957, 17): “I think that we are forced to conclude that grammar is autonomous and independent of meaning.” And in a later espousing of the assumption, Chomsky traces the idea back to what he refers to as “structural linguistics” (1972, 119):

116

A central idea of much of structural linguistics was that the formal devices of language should be studied independently of their use. The earliest work in transformational-generative grammar took over a version of this thesis, as a working hypothesis. I think it has been a fruitful hypothesis. It seems that grammars contain a substructure of perfectly formal rules operating on phrase-markers in narrowly circumscribed ways. Not only are these rules independent of meaning or sound in their function…

This passage is very pertinent in seeking to understand the assumption and its status and origins (and cf. Chomsky 1975, 18–22; 1977, 38–58). But let us observe some things about origins that it doesn’t entirely convey. Firstly, Chomsky’s “much of structural linguistics” should not be taken to include most of the work done in early structural linguistics in Europe, even by self-declared autonomists (see Anderson 2005). It is notably the followers of Leonard Bloomfield (1926) and proponents of transformational grammar who insist on the autonomy of syntax from meaning. And, even for the “post-Bloomfieldians,” syntax is far from autonomous from phonology. Perhaps more significantly, we should fully register the extent to which the autonomy assumption is an innovation (Anderson 2005). Gram mar or syntax before structuralism was not conceived of as autonomous; syntactic rules and principles can refer to semantically and/or phonologically defined categories. Consider as an example of this earlier tradition Otto Jespersen’s description of the syntax of the SPEECH-ACT category of “question”: “[T]he formal means by which questions are expressed, are (1) tone; (2) separate interrogative words … ; (3) word-order” (1924, 305). Syntax (and intonation) is expressive of meaning. However, autonomy requires that elements that participate in purely syntactic regularities be syntactic themselves. Thus, the feature “Q for interrogative clauses” of Chomsky (1995, §4.5.4) is part of syntax. It is “interpretable,” but its interpretation is not pertinent to syntax (see illocutionary force and sentence types). But from a traditional point of view, it is the categorial meaning of “Q” that, as with other syntactic elements, drives its syntax. Its status as (prototypically) a request for information is what demands, for instance, the presence in the sentence of an “open” element, marked, for example, by wh-, or by intonation or some such indication of openness of the truth of the sentence itself. The autonomy hypothesis is falsifiable only if there is an independent notion of what constitutes syntax; otherwise, any apparent counterexample can be relegated to interaction between syntax and some other module (see modularity). An unfalsifiable assumption of autonomy defines a research program, rather than constituting an empirical hypothesis: It is methodological rather than ontological. The program, as well as the hypothesis, is based on the premise that it is “fruitful” to operate as if syntax is autonomous, in contrast with the more traditional view that nonreference by a syntactic regularity to interpretation is exceptional, involving demotivation (grammaticalization) within a syntax whose central concern is with the role of sound and structure as expressive of meaning. Opponents of the autonomy assumption, whatever its status, tend to interpret it in the absolute form described here (as

Babbling Langacker 1987,1–55). Chomsky, however, envisages autonomy theses “of varying degrees of strength” (1977, 43), whereby syntax is not necessarily exhausted by the “substructure of perfectly formal rules” (1972, 119) of his formulation. Thus, “the significant question with regard to the autonomy thesis may not be a question of ‘yes’ or ‘no,’ but rather of ‘more’ or ‘less,’ or more correctly, ‘where’ and ‘how much’” (Chomsky 1977, 42). Certainly, provided again that we have an independent characterization of syntax, the extent of autonomy and its accommodation are in themselves interesting empirical questions, with consequences for modularity, universal grammar, and the autonomy of language itself. And seeking to answer them might be more comprehensible to opponents of a strong interpretation of autonomy. Work within the autonomist program(s), whatever the status of the assumption, has undoubtedly had important results, but there is room for debate as to how fruitful has been the pursuit of the autonomy assumption as such. And in addition to the question of how it relates to independent notions of what syntax is, a major difficulty in evaluating the assumption, and its contribution to these results, is the changing nature of the grammatical enterprise(s) in which autonomy has been invoked, as well as the varying degrees of emphasis with which it has been put forward or denied. – John M. Anderson WORKS CITED AND SUGGESTIONS FOR FURTHER READING Anderson, John M. 2005. “Structuralism and autonomy: From Saussure to Chomsky.” Historiographia Linguistica 32: 117–48. Bloomfield, Leonard. 1926. “A set of postulates for the science of language.” Language 2: 153–64. Chomsky, Noam. 1957. Syntactic Structures. The Hague: Mouton. ———. 1972. “Some empirical issues in the theory of transformational grammar.” In Goals of Linguistic Theory, ed. Stanley Peters, 63–130. Englewood Cliffs, NJ: Prentice-Hall. ———. 1975. The Logical Structure of Linguistic Theory. New York: Plenum Press. ———. 1977. Essays on Form and Interpretation. Amsterdam: NorthHolland. ———. 1995. The Minimalist Program. Cambridge, MA: MIT Press. Jespersen, Otto. 1924. The Philosophy of Grammar. London: George Allen and Unwin. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar, I: Theoretical Prerequisites. Stanford, CA: Stanford University Press. Radford, Andrew. 1988. Transformational Grammar: A First Course. Cambridge: Cambridge University Press.

B BABBLING Babbling can be defined as infant vocal production that is broadly adultlike in phonetic shape but that lacks any identifiable adult model or intended meaning. The formal criterion – broadly adultlike – limits study to the period that follows the child’s first temporally regular or rhythmic C-V (Consonant-Vowel)-syllable production, also known as canonical babbling (Oller 1980); these

vocal forms include both a complete or nearly complete supraglottal closure and a transition to a recognizable vocalic nucleus, for example, [dada], [babababa], [ŋaŋaŋa]. Prior to that, the child vocalizes in more primitive ways that are not thought to be directly related to language. The lack of an adult model is a question of interpretation since the first word forms are highly similar in form to concurrent babbling. In fact, there is in most cases a gradual shift from a predominance of unidentifiable vocalizations, beginning with the emergence of canonical babbling (at 6 to 10 months), to a predominance of word use identifiable in the situational context (16 to 22 months). The extent to which the shift is relatively abrupt or gradual is highly individual. Finally, there are vocal forms that do not appear to be based on an adult model but that are nevertheless used with consistent broad communicative meaning such as request, rejection, or interest; these transitional forms or protowords (Vihman and Miller 1988) should thus be distinguished from babbling, which lacks any apparent communicative goal. Babbling is a “largely self-directed process of exploration” (Elbers 1982, 45).

Brief Modern History THE “CONTINUITY ISSUE”: THE RELATIONSHIP OF BABBLE TO WORDS. Roman Jakobson ([1941] 1968) was the first linguist to pay serious theoretical attention to babbling – if only to deny its relevance to language learning. On the basis of the diary accounts available to him, Jakobson developed the (“discontinuity”) view that babbling was merely random sound production, expressing the full range of human phonetic possibility but unrelated to the more “austere” or constrained repertoire of the first words. Jakobson saw the latter as reflecting a well-ordered universal scheme for the emergence of phonological oppositions, such that the low vowel /a/ is primary, with contrast with /i/ following, while anterior stops are the first consonants produced (labial /b/ or dental /d/), followed by nasals and only later by other places and manners of articulation. This impressively articulated universalist theory held sway for many years but was challenged in the 1970s when diary data began to be supplemented by planned recordings of infants (typically in free interaction with an adult). Charles A. Ferguson and Olga K. Garnica (1975) and Paul Kiparsky and Lise Menn (1977) were among the first to raise objections to Jakobson’s ideas of gradual phonemic differentiation, which disregarded the effect of word position on order of segment acquisition and which would be difficult to defend on the basis of the very few words produced in a child’s earliest lexical period. On the other hand, Jakobson’s claims regarding the phones that occur in the first words were, on the whole, quite accurate, based as they were on decades of diary records provided by linguists and psychologists. What was not supported by later studies was the strong separation required by Jakobson’s theory between words (or phonology) and babble (or phonetic production). Far from babbling being unrelated to word production, later studies have established that the first words draw their phonetic resources from the particular inventory of sound patterns developed by the individual child through babbling (Vihman et al. 1985; continuity has also been reported for “babbled” gesture and first

117

Babbling signs: Cheek et al. 2001). For example, a French child whose prelinguistic babbling made considerable use of liquids (mainly [l]) was found to develop several first words with [l(j)], which is uncommon in early phonology: allo “hello” (on the telephone) [ailo], [hailo], [haljo], [alo]; lolo “bottle” (babytalk term) [ljoljo]; donne (le) “give (it)” [də], [dlə], [ldɛ], [heldo] (Vihman 1993). BABBLING DRIFT: THE EFFECT OF PERCEPTION ON PRODUCTION. A second issue that has aroused interest for half a century is that of possible drift in babbling toward the sounds of the native language (Brown 1958). The issue has generated considerable heat and is important since it concerns the extent to which infants can be taken to be capable of translating their perceptual experience of the sound patterns of the ambient language into their limited production repertoire. That is, any identifiable ambient language influence on prelinguistic vocalizations means that infants have both perceived the typical sounds of their language and adjusted their vocal production accordingly. Many studies, from Atkinson, MacWhinney, and Stoel (1968) to Engstrand, Williams, and Lacerda (2003), have used adult perceptual judgments of recorded vocalizations to determine whether infants’ language of exposure can be identified, as that would provide evidence of drift; the findings remain inconclusive, however. Meanwhile, Bénédicte de BoyssonBardies and colleagues, using acoustic analyses of vowels (1989) and tallies of transcribed consonant types (BoyssonBardies and Vihman 1991), established significant prelinguistic adult language influence, although the mechanism for such an effect remained unclear. More recent work demonstrating the extent of early implicit or distributional learning (Saffran, Aslin, and Newport 1996) suggests that infants are capable of registering dominant patterns of their language within the first year. Thus, the mechanism needed to account for drift may be the effect of implicit perceptual learning on production: Those vocalizations that, as the producing child perceives them, activate perceptual responses already familiar from input patterning would strengthen perceptuomotor connections, leading to their repeated use.

Theoretical approaches FRAME AND CONTENT: THE ARTICULATORY BASIS OF BABBLING. The most widely accepted current model of babbling is that of Peter F. MacNeilage and Barbara L. Davis (1990 and Davis and MacNeilage 1990, 1995 (for a review of competing ideas, see Chen and Kent 2005). The articulatory basis of babbling is claimed to be frame dominance, meaning that the patterns produced largely reflect mandibular oscillation without independent control of lip and tongue movement. The result is strong C-V associations, such that alveolars are followed by front vowels, labials by central vowels (the pure frames, requiring no particular tongue setting), and velars by back vowels. Furthermore, the model predicts that changes in the mandibular cycle will result in height changes for vowels and manner changes for consonants in variegated babbling sequences. The work of this team and collaborators investigating a range of other languages (e.g., Dutch, French, Romanian, Turkish: Davis et al. 2005) have largely supported the predictions and have demonstrated a tendency for adult languages to show the C-V associations as well (MacNeilage and Davis 2000) – but

118

Chen and Kent (2005) report an association of labials with back vowels in their extensive Mandarin data, both child and adult. The balance of ambient language (perceptual) influence versus universal (physiological or motoric) tendencies thus remains controversial. Any early C-V associations can be expected to fade with lexical growth as infants follow their individual paths toward segmental independence (freeing the “content” from the frame). VOCAL MOTOR SCHEMES AND THE EFFECT OF PRODUCTION ON PERCEPTION. Lorraine McCune and Marilyn M. Vihman (2001) introduced the concept of “vocal motor schemes” (VMS), or “generalized action patterns that yield consistent phonetic forms” (p. 673), identified on the basis of repeated high-frequency production of one or more consonants over the course of several recordings. VMS index emergent stability in consonant production, a reliable predictor of lexical advance. Vihman’s “articulatory filter” model (1993) posits that an infant’s babbling patterns will effectively “highlight” related forms in the input. Once one or more VMS are established, it is possible to test the model by measuring infants’ attentional response to a series of short sentences featuring nonwords that do or do not include that child’s VMS. Capitalizing on wide infant variability in the timing and nature of first vocal forms (within the limits of the strong universal contraints), Rory A. DePaolis (2006) established an effect of infant production on the perception of speech. His findings support the idea that the first words, typically produced in priming situations (“context-limited” words: McCune and Vihman 2001), are based on infant experience of a rough match between vocal forms established through babbling practice and words heard frequently in input speech (Vihman and Kunnari 2006). Such “selection” of words to attempt based on the vocal forms available for matching would account for the relative “accuracy” of first words (Ferguson and Farwell 1975), their constrained shapes (e.g., one or two syllables in length, with little variegation across word position or syllables), and their strong rootedness in the biomechanical basis of babbling as established by Davis and MacNeilage. It also explains the difficulty of distinguishing words from babble (continuity) and the subtlety of the ambient language effect on babbling and early words (drift). – Marilyn Vihman WORKS CITED AND SUGGESTIONS FOR FURTHER READING Atkinson, Kay, Brian MacWhinney, and Carol Stoel. 1968. “An experiment on the recognition of babbling.” Language Behavior Research Laboratory Working Paper 14. University of California, Berkeley. Boysson-Bardies, Bénédicte de, Pierre Hallé, Laurent Sagart, and Catherine Durand. 1989. “A crosslinguistic investigation of vowel formants in babbling.” Journal of Child Language 16: 1–17. Boysson-Bardies, Bénédicte de, and Marilyn M. Vihman. 1991. “Adaptation to language.” Language 67: 297–319. Brown, Roger. 1958. Words and Things. Glencoe, IL: Free Press. Cheek, Adrianne, Kearsy Cormier, Ann Repp, and Richard P. Meier. 2001. “Prelinguistic gesture predicts mastery and error in the production of early signs.” Language 77: 292–323. Chen, L. M., and Raymond D. Kent. 2005. “Consonant-vowel cooccurrence patterns in Mandarin-learning infants.” Journal of Child Language 32: 507–34.

Babbling Davis, Barbara L., and Peter F. MacNeilage. 1990. “Acquisition of correct vowel production.” Journal of Speech and Hearing Research 33: 16–27. ———. 1995. “The articulatory basis of babbling.” Journal of Speech and Hearing Research 38: 1199–1211. Davis, Barbara L., Sophie Kern, Dilara Koçbaş, and Inge Zink. 2005. “Vocalizations in canonical babbling.” Paper presented at Symposium, 10th International Congress of the Association for the Study of Child Language, Berlin. DePaolis, Rory A. 2006. “The influence of production on the perception of speech.” In Proceedings of the 30th Boston University Conference on Language Development, ed. D. Bamman, T. Magnitskaia, and C. Zaller, 142–53. Somerville, MA: Cascadilla Press. Elbers, L. 1982. “Operating principles in repetitive babbling.” Cognition 12: 45–63. Engstrand, Olle, Karen Williams, and Francisco Lacerda. 2003. “Does babbling sound native?” Phonetica 60: 17–44. Ferguson, Charles A., and Carol B. Farwell. 1975. “Words and sounds in early language acquisition.” Language 51: 419–39. Ferguson, Charles A., and Olga K. Garnica. 1975. “Theories of phonological development.” In Foundations of Language Development, ed. Eric H. Lenneberg and Elizabeth Lenneberg, 153–80. New York: Academic Press. Jakobson, Roman. [1941] 1968. Child Language, Aphasia, and Phonological Universals. The Hague: Mouton. Eng. translation of Kindersprache, Aphasie und allgemeine Lautgesetze, Uppsala. Kiparsky, Paul, and Lise Menn. 1977. “On the acquisition of phonology.” In Language Learning and Thought, ed. John Macnamara, 47–78. New York: Academic Press. MacNeilage, Peter F., and Barbara L. Davis. 1990. “Acquisition of speech production: Frames, then content.” In Attention and Performance. Vol. 13: Motor Representation and Control. Ed. Marc Jeannerod, 453–75. Hillsdale, NJ: Lawrence Erlbaum. ———. 2000. “On the origin of internal structure of word forms.” Science 288: 527–31. McCune, Lorraine, and Marilyn M. Vihman. 2001. “Early phonetic and lexical development.” Journal of Speech, Language and Hearing Research 44: 670–84. Oller, D. Kimbrough. 1980. “The emergence of the sounds of speech in infancy.” In Child Phonology. Vol. 1: Production. Ed. Grace Yenikomshian, James F. Kavanagh, and Charles A. Ferguson, 93–112. New York: Academic Press. ———. 2000. The Emergence of the Speech Capacity. Mahwah, NJ: Lawrence Erlbaum. This book provides a thorough review of babbling studies conducted with hearing, hearing-impaired, premature, and low SES (socioeconomic status) infants, as well as provocative ideas about the evolution of language based on evidence from ontogeny. Saffran, Jenny R., Richard N. Aslin, and Elissa L. Newport. 1996. “Statistical learning by 8-month-old infants.” Science 274: 1926–8. Vihman, Marilyn M. 1993. “Variable paths to early word production.” Journal of Phonetics 21: 61–82. ———. 1996. Phonological Development. Oxford: Blackwell. This book provides an overview of research in infant speech perception and production and their interactions, as well as of theories of phonological development, early word patterning, and the nature of the transition into language. Vihman, Marilyn M., and Sari Kunnari. 2006. “The sources of phonological knowledge.” In Recherches Linguistiques de Vincennes 35: 133–64. Vihman, Marilyn M., Marlys A. Macken, Ruth Miller, Hazel Simmons, and James Miller. 1985. “From babbling to speech: A re-assessment of the continuity issue.” Language 61: 397–445. Vihman, Marilyn M., and Ruth Miller. 1988. “Words and babble at the threshold of lexical acquisition.” In The Emergent Lexicon, ed. Michael D. Smith and John L. Locke, 151–83. New York: Academic Press.

Basal Ganglia

BASAL GANGLIA Nothing in biology makes sense except in the light of evolution. Dobzhansky 1973

The basal ganglia are subcortical structures that can be traced back to frogs and are traditionally associated with motor control. However, current studies show that complex behaviors generally are regulated by neural circuits that link local processes in different parts of the brain. In humans, the basal ganglia play a critical role in neural circuits regulating cognitive processes, including language, as well as motor control and emotion. The capacities that differentiate humans from other species, such as being able to talk, forming and comprehending sentences that have complex syntax, and possessing cognitive flexibility, devolve from neural circuits that link activity in different regions of the cortex through the basal ganglia. The neural bases of human language thus involve the interplay of processes that regulate motor control, other aspects of cognition, mood, and personality. Given the involvement of multiple regions of the brain that are involved in many activities, it is difficult to see how any “organ” of the brain could be specific to language and language alone, such as the “narrow faculty of language” that, according to Marc D. Hauser, Noam Chomsky, and W. T. Fitch (2002) yields the recursive properties of syntax. Evidence from experiments-in-nature that attempt to link specific behavioral deficits with damage to a particular part of a patient’s brain led to the traditional Broca-Wernicke theory. This traditional theory claims that linguistic processes are localized in these two regions of the neocortex, the outermost part of the brain. However, evidence from brain-imaging techniques, such as computer augmented tomography (CT scans), demonstrated that aphasia, permanent loss of language, never occurs in the absence of subcortical damage (Stuss and Benson 1986). Subsequent findings from techniques such as functional magnetic resonance imaging (fMRI – see neuroimaging) that indirectly map neural activity show that although Broca’s area and Wernicke’s area are active when neurologically intact subjects perform various linguistic tasks; these areas are elements of complex neural circuits that link activity in other cortical regions and subcortical structures (Kotz et al. 2003). Studies of neurodegenerative disorders, such as Parkinson’s disease (Lieberman, Friedman, and Feldman 1990; Grossman et al. 1992), revealed the role of the basal ganglia in regulating speech and language. Speech production and the comprehension of distinctions in meaning conveyed by syntax deteriorated when basal ganglia function was impaired. Basal ganglia dysfunction is implicated in seemingly unrelated conditions, such as obsessive-compulsive disorder, schizophrenia, Parkinson’s disease, and verbal apraxia – a condition in which orofacial, laryngeal, and respiratory control during speech is impaired (Lieberman 2006)

Neural Circuits These syndromes follow from the basal ganglia activity in different neural circuits. Neural circuits that link activity in different parts of the brain appear to be the bases for most, if not all, complex mammalian behaviors. In humans, a class of neural circuits that links activity in different regions of the cortex through the

119

Basal Ganglia basal ganglia and other subcortical structures appears to play a key role in regulating aspects of human linguistic ability, such as talking and comprehending the meaning of a sentence, as well as such seemingly unrelated phenomena as decision making, walking, attention, and emotional state. To understand the nature of neural circuits, we must take account of the distinction that exists between local operations that are carried out within some particular part of the brain and an observable behavior that results from many local operations linked in a neural circuit. Complex brains, including the human brain, perform local operations involving tactile, visual, or auditory stimuli in particular regions of the brain. Other neural structures perform local operations that regulate aspects of motor control or hold information in short-term (working) memory, and so on. The basic computational elements of biological brains are neurons. Local operations result from activity in an anatomically segregated, population (a group) of neurons. A given part of the brain many contain many distinct anatomically segregated neuronal populations that each carry out similar local operations. But these local operations do not constitute observable behaviors. Each anatomically segregated neuronal population projects to anatomically distinct neuronal populations in other regions of the brain, forming a neural circuit. The linked local operations performed in the circuit constitute the neural basis of an observable aspect of behavior, such as striking the keys of a computer keyboard.

Basal Ganglia Operations Research that initially focused on Parkinson’s disease, a neurodegenerative disease that affects the operation of the basal ganglia, largely sparing cortex, demonstrated their role in motor control, syntax, and cognitive flexibility. In their review article, C. D. Marsden and J. A. Obeso (1994) noted that the basal ganglia constitute a “sequencing engine” for both motor and cognitive acts. The basal ganglia regulate routine motor acts by activating and linking “motor pattern generators” that each constitute an instruction set for a submovement to the frontal regions of the brain involved in motor control. As each submovement reaches its goal, the pattern generator for the next appropriate submovement is activated. Therefore, motor control deficits characterize neurodegenerative diseases such as Parkinson’s that degrade basal ganglia operations. The basal ganglia have other motor functions; in changing circumstances, they can switch to a set of motor pattern generators that constitute a better fit to the changed environment constituting adaptive motor control. Basal ganglia operations involving “cognitive pattern generators” (Graybiel 1997) account for the subcortical dementia associated with Parkinson’s disease. Afflicted individuals perseverate: They are unable to switch to a new train of thought when circumstances change. On cognitive tests such as the Wisconsin Card Sorting Test (WCST), they have difficulty switching to a new cognitive criterion. For example, a subject who has been successfully sorting cards by their color will have difficulty switching to sorting them by the number of symbols printed on each card. Neurophysiologic studies that trace the linkages between the segregated neuronal populations of the basal ganglia and cortex confirm circuits that project from the basal ganglia to regions of the brain that are implicated in

120

cognitive as well as motor acts. Brain imaging studies reveal increased basal ganglia activity in syntactically complex sentences, as well as at the points where a person must switch from one criterion to another, as is the case in studies using tests of cognition such as the WCST (Monchi et al. 2001). Thus, basal ganglia dysfunction arising from neurodegenerative diseases, lesions, or the effects of oxygen deprivation (Lieberman et al. 2005) also can result in an inability to comprehend distinctions in meaning conveyed by complex syntax. Afflicted individuals appear to have difficulty switching the cognitive pattern generators that code syntactic operations at clause boundaries or that in sentences depart from a simple canonical form. These subjects typically have difficulty sequencing motor acts, including those involved in speech. Their motor acts are slower, resulting in longer vowel durations, and those subjects have difficulty rapidly sequencing the tongue, lip, and laryngeal maneuvers necessary to differentiate stop consonants, such as [b] for [p], or [d] from [t].

Motor Control and Syntax Linguists have long realized that the syntactic operations (i.e., the rules that they use to describe the structure of a sentence) yield hierarchical structures. In describing the syntax of the sentence John saw the cat, the words the cat are part of a constituent that includes the verb saw. The rules that can be used to describe seemingly simple motor acts such as walking also yield hierarchical structures. Both motor control and syntax involve selectional constraints that result in hierarchical structures. For example, the motor pattern generator for heel strike cannot be activated before or much after your foot meets the ground. This yields a hierarchical tree diagram similar to those commonly used to convey the grammatical structure of a sentence. The syntactic tree diagram for a square dance in which swing your partner occurred again and again would not differ in principle from that of a sentence having embedded relative clauses. (For more on the similarities between motor control rules and those of generative syntax, see Lieberman 2006).

Genetic Findings Studies of the regulatory gene FOXP2 provide a starting point for understanding the evolution of the cortical-striatal-cortical circuits that confer human linguistic ability (see genes and language). Other genes undoubtedly are involved and FOXP2 is not a “language gene.” FOXP2 governs the embryonic development of the basal ganglia, other subcortical structures, and lung tissue and other structures. Its discovery resulted from a long-term study of an extended family in which many individuals are marked by a genetic anomaly. A syndrome, a suite of speech and orofacial movement disorders, and cognitive and linguistic deficits mark these individuals. They are not able to protrude their tongues while closing their lips, cannot repeat two word sequences, and have difficulty comprehending distinctions in meaning conveyed by syntax (Vargha-Khadem et al. 1998). On standardized intelligence tests, they have significantly lower scores than their nonafflicted siblings. MRI imaging shows that the caudate nucleus (a basal ganglia structure) is abnormal. fMRI imaging, which provides a measure of neural activity, shows underactivation in the putamen (the principal basal ganglia input structure), Broca’s

Basal Ganglia

Basic Level Concepts

area, and its right homolog (Watkins et al. 2002; Liegeois et al. 2003). These structures are connected by neural circuits through the striatum (Lehericy et al. 2004). The behavioral deficits of afflicted individuals are similar to those seen in Parkinson’s disease and oxygen deprivation (cf. Lieberman 2006 for details). The role of FOXP2 during early brain development in humans and of the mouse version ( foxp2) in mice was established by C. S. Lai and colleagues (2003). The gene governs the expression of other genes during embryonic development. In both the human and mouse brain, the gene is active in the interconnected neural structures that constitute the cortical-striatal-cortical circuits regulating motor control and cognition in humans, including the caudate nucleus and putamen of the basal ganglia, the thalamus, inferior olives, and cerebellum. Despite the high degree of similarity, the mouse and human versions are separated by three mutations. The chimpanzee and human versions are separated by two mutations. W. Enard and colleagues (2002), using the techniques of molecular genetics, estimate that the human form appeared somewhere in the last 200,000 years, in the time frame (Stringer 1998) associated with the emergence of anatomically modern Homo sapiens. The appearance of human speech anatomy 50,000 years ago presupposes the prior appearance of this neural substrate (see speech anatomy, evolution of). In short, the basal ganglia are neural structures that were initially adapted for one function – motor control. In the course of evolution, the human basal ganglia were modified, taking on additional cognitive and linguistic tasks. – Philip Lieberman WORKS CITED AND SUGGESTIONS FOR FURTHER READING Dobzhansky, Theodosius. 1973. “Nothing in biology makes sense except in the light of evolution.” American Biology Teacher 35: 125–9. Enard, W., M. Przeworski, S. E. Fisher, C. S. Lai, V. Wiebe, T. Kitano, A. P. Monaco, and S. Paabo. 2002. “Molecular evolution of FOXP2, a gene involved in speech and language.” Nature 41: 869–72. Graybiel, Ann M. 1997. “The basal ganglia and cognitive pattern generators.” Schizoprenia Bulletin 23: 459–69. Grossman, Murray, S. Carvell, S. Gollomp, M. B. Stern, G. Vernon, and H. I. Hurtig. 1992. “Sentence comprehension and praxis deficits in Parkinson’s disease.” Neurology 41: 1620–8. Hauser, Marc D., N. Chomsky, and W. T. Fitch. 2002. “The faculty of language: What is it, who had it, and how did it evolve?” Science 298: 1569–79. Kotz, Sonia A., M. Meyer, K. Alter, M. Besson, D. Y. von Cramon, and A. D. Frederica. 2003. “On the lateralization of emotional prosody: An fMRI investigation.” Brain and Language 86: 366–76. Lai, C. S., D. Gerrelli, A. P. Monaco, S. E. Fisher, and A. J. Copp. 2003. “FOXP2 expression during brain development coincides with adult sites of a pathology in a severe speech and language disorder.” Brain 126: 2455–62. Lehericy, S. M., M. Ducros, P. F. Van de Moortele, C. Francois, L. Thivard, C. Poupon, N. Swindale, K. Ugurbil, and D. S. Kim. 2004. “Diffusion tensor tracking shows distinct corticostriatal circuits in humans.” Annals of Neurology 55: 522–9. Lieberman, Philip. 2006. Toward an Evolutionary Biology of Language. Cambridge: Harvard University Press. Lieberman, Philip, J. Friedman, and L. S. Feldman. 1990. “Syntactic deficits in Parkinson’s Disease.” Journal of Nervous and Mental Disease 178: 360–5.

Lieberman, Philip, A. Morey, J. Hochstadt, M. Larson, and S. Mather S. 2005. “Mount Everest: A space analogue for speech monitoring of cognitive deficits and stress.” Aviation, Space and Environmental Medicine 76: 198–207. Liegeois, F., T. Baldeweg, A. Connelly, D. G. Gadian, M. Mishkin, and F. Varhgha-Khadem, 2003. “Language fMRI abnormalities associated with FOXP2 gene mutation.” Nature Neuroscience 6: 1230–7. Marsden, C. D., and J. A. Obeso.1994. “The functions of the basal ganglia and the paradox of stereotaxic surgery in Parkinson’s disease.” Brain 117: 877–97. Monchi, O., P. Petrides, V. Petre, K. Worsley, and A. Dagher. 2001. “Wisconsin card sorting revisited: Distinct neural circuits participating in different stages of the task identified by event-related functional magnetic resonance imaging.” Journal of Neuroscience 21: 7733–41. Stringer, Christopher B. 1998. “Chronological and biogeographic perspectives on later human evolution.” In Neanderthals and Modern Humans in Western Asia, ed. T. Akazawa, K. Abel, and O. Bar-Yosef, 29–38. New York: Plenum. Stuss, Donald T., and D. F. Benson. 1986. The Frontal Lobes. New York: Raven Vargha-Khadem, Faraneh, K. E. Watkins, C. J. Price, J. Ashburner, K. J. Alcock, A. Connelly, R. S. Frackowiak, K. J. Friston, M. E. Pembrey, M. Mishkin, D. G. Gadian, and R. E. Passingham. 1998. “Neural basis of an inherited speech and language disorder.” PNAS USA 95: 12695–700. Watkins, Kate, F. Vargha-Khadem, J. Ashburn, R. E. Passingham, A. Connelly, K. J. Friston, R. S. Frackiwiak, M. Miskin, and D. G. Gadian. 2002. “MRI analysis of an inherited speech and language disorder: Structural brain abnormalities.” Brain 125: 465–78.

BASIC LEVEL CONCEPTS A concept is a mental representation that allows people to pick out a group of equivalent things or a category (see categorization). For example, people use their concept of dog to pick out members of category of things that are called dogs. Concepts are also organized into hierarchical taxonomies, or sequences of progressively larger categories, in which each category includes all the previous ones. For example, an object driven on a highway with four wheels and a top that folds back can be called a convertible, a car, or a vehicle. The category car is more general than convertible because it includes other objects (e.g., station wagons) as well as the members of convertible. The category vehicle is more general than convertible and car because it contains other objects (e.g., trucks) as well as the members of these categories. Strong evidence from cognitive psychology (Rosch et al. 1976) and anthropology (Berlin 1992) suggests that one level of such hierarchies is cognitively privileged. Eleanor Rosch and colleagues (1976) used a wide range of converging methods that singled out the basic level as playing a central role in many categorization processes. For example, the category level represented by chair and dog is typically considered the basic level, in contrast to more general superordinate concepts, such as furniture and animal, and more specific subordinate concepts, such as recliner and labrador retriever. Basic level concepts have advantages over other concepts. Pictures of objects are categorized faster at the basic level than at other levels (Jolicoeur, Gluck, and Kosslyn 1984). As noted by Rosch and her colleagues, people primarily use basic level names in naming tasks, and the basic level is the highest level

121

Basic Level Concepts

Bilingual Education

for which category members have similar overall shape (cf. car versus vehicle). Children learn basic level concepts sooner than other concepts (Brown 1958; Horton and Markman 1980). Basic level advantages are found in many other domains, including environmental scenes (Tversky and Hemenway 1983), social categories (Cantor and Mischel 1979), and actions (Morris and Murphy 1990). One explanation for the advantages of basic level categories over other categories is that they are more differentiated (Rosch et al. 1976; Murphy and Brownell 1985). Members of basic level categories have many features in common. These features are also distinct from those of other categories at this level. In contrast, although members of more specific, subordinate categories (e.g., sports car) have slightly more features in common than do those of basic level categories, many of these features are not distinctive. That is, members of a subordinate category share their features with other subordinates (e.g., members of sports car share a number of features with other subcategories of car). In contrast, the members of more general, superordinate categories (e.g., vehicle) have few common features. Differentiation explains the basic level advantage because it reflects a compromise between two competing functions of concepts. Categories should be informative so that one can draw inferences about an entity on the basis of its category membership. Emphasizing this function leads to the “formation of large numbers of categories with the finest possible discriminations between categories” (Rosch et al. 1976, 384). However, the formation of categories should only preserve important differences between them that are practical: “It is to the organism’s advantage not to differentiate one stimulus from others when that differentiation is irrelevant to the purposes at hand” (ibid.). This function counteracts the tendency to create large numbers of categories and reflects the principle of cognitive economy (Rosch et al. 1976). Overall, basic level categories have an advantage because they are relatively general and informative, whereas superordinate categories, though general, are not informative and subordinate categories, though informative, are not general. The basic level may change with expertise in a way that is consistent with the differentiation explanation. For example, James Tanaka and Marjorie Taylor (1991) investigated expertise effects on the basic level in expert dog breeders and birdwatchers. Using a number of tasks, they tested each expert in both the dog and bird domains. For instance, in a speeded categorization task, experts in their novice domain were fastest at the basic level and slowest at the subordinate level (as Rosch et al. 1976 found). However, in their area of expertise, categorization was equally fast at the basic and subordinate levels. For more detailed reviews of this literature, see Lassaline, Wisniewski, and Medin 1992; Murphy and Lassaline 1997.) – Edward Wisniewski WORKS CITED AND SUGGESTIONS FOR FURTHER READING Berlin, Brent. 1992. Ethnobiological Classification: Principles of Categorization of Plants and Animals in Traditional Societies. Princeton, NJ: Princeton University Press. Brown, Roger. 1958. “How shall a thing be called?” Psychological Review 65: 14–21.

122

Cantor, Nancy, and Walter Mischel. 1979. “Prototypes in person perception.” In Advances in Experimental Social Psychology, ed. L. Berkowitz, 4–52. New York: Academic Press. Horton, Marjorie, and Ellen Markman. 1980. “Developmental differences in the acquisition of basic and superordinate categories.” Child Development 51: 708–15. Jolicoeur, Pierre, Mark Gluck, and Steven Kosslyn. 1984. “Pictures and names: Making the connection.” Cognitive Psychology 16: 243–75. Lassaline, Mary, Edward Wisniewski, and Douglas Medin. 1992. “Basic levels in artificial and natural categories: Are all basic categories created equal?” In Percepts, Concepts, and Categories: The Representation and Processing of Information, ed. B. Burns, 328–80. North Holland: Elsevier. Morris, Michael, and Gregory Murphy. 1990. “Converging operations on a basic level in event taxonomies. Memory and Cognition 18: 407–18. Murphy, Gregory, and Hiram Brownell. 1985. “Category differentiation in object recognition: Typicality constraints on the basic category advantage.” Journal of Experimental Psychology: Learning, Memory and Cognition 11: 70–84. Murphy, Gregory, and Mary Lassaline. 1997. “Hierarchical structure in concepts and the basic level of categorization.” In Knowledge, Concepts, and Categories, ed. K. Lamberts and D. Shanks, 93–131. London: Psychology Press Rosch, Eleanor, Carolyn Mervis, Wayne Gray, David Johnson, and Penny Boyes-Braem. 1976. “Basic objects in natural categories.” Cognitive Psychology 8: 382–439. Tanaka, James, and Marjorie Taylor. 1991. “Object categories and expertise: Is the basic level in the eye of the beholder?” Cognitive Psychology 23: 472–82. Tversky, Barbara, and Kathy Hemenway. 1983. “Categories of environmental scenes.” Cognitive Psychology 15: 121–49.

BILINGUAL EDUCATION In principle, bilingual education is just the use of two languages in instruction in a school setting. However, in practice, it covers a wide array of programs. Bilingual education programs range from high-status schools promoting international education through prestige languages, such as English and French, to highly marginalized schools devoted to the bare-bones schooling of immigrant children. In its “weak” form, bilingual education may involve transitional or subtractive bilingualism, leading to monolingualism (e.g., teaching Spanish-speaking children English to ensure their assimilation and integration into mainstream America). In its “strong” form, bilingual education aims at maintaining the language of a minority child in addition to the learning of a majority language, thus leading to additive bilingualism. Heritage bilingual schools often practice an ideal version of additive bilingualism, which stresses bicultural education in addition to bilingualism. One of the central concerns of bilingual education is to address the educational needs/performance of minority children by maintaining their mother tongue. The proponents of maintaining the mother tongue claim that such maintenance is critical for linguistic and cognitive growth of the child, school performance, psychological security, ethnic and cultural identity (see ethnolinguistic identity), self-esteem, and many other positive personal and intellectual characteristics. The supporters of transitional bilingualism claim that only transitional

Bilingualism, Neurobiology of bilingualism is capable of saving children from poor academic performance and allowing for assimilation. Bilingual education has been steadily gaining strength around the globe since the era of decolonization. It is further fueled by the growth of ethnic awareness and the movement to prevent the extinction of languages of the world. Many countries, particularly in Europe, that earlier fostered monolithic ideology have begun to recognize their diversity as a source of social and economic capital, thus marking a new era of bilingual/multilingual education (e.g., in the United Kingdom, France, and Spain, among others; see language policy). Multilingual countries of Asia and Africa continue to nurture a long tradition of bilingual education. Since 1956, India, for example, has had as official policy a three-language formula in education. This formula calls for multilingual education. In addition to learning the two national languages – Hindi and English – students are expected to learn a third or a fourth language. Because of its deep-rooted association with immigrants, bilingual education in the United States is particularly notable for its turbulent history. On June 2, 1998, the people of California voted to end a tradition of bilingual education by passing Proposition 227, which gives immigrant children just one year to learn English before they enroll in regular classes. Many school systems in other states are waiting either to put in place severe restrictions on bilingual instruction or to eliminate it completely by passing “English only” policies (for details, see Genesee 2006). While bilingual education is often associated with the education of minority students (e.g., in the United States), the Canadian “immersion” programs in French devoted to the majority Anglophones serve as a model of bilingual education for majority students (for details, see Genesee 2006). – Tej K. Bhatia WORKS CITED AND SUGGESTIONS FOR FURTHER READING Crawford, James. 2004. Educating English Learners. Los Angeles: Bilingual Educational Services. Genesee, Fred. 2006. “What do we know about bilingual education for majority-language students?” In Handbook of Bilingualism, ed. Tej K. Bhatia and William C. Ritchie, 547–76. Oxford: Blackwell.

BILINGUALISM, NEUROBIOLOGY OF Neurobiology of bilingualism refers to the study of the cerebral organization of multiple languages in the human brain. From early accounts of selective loss and recovery in bilingual aphasia (i.e., loss of language due to a brain lesion) to recent electrophysiological and functional neuroimaging studies, issues inherent to the bilingual brain have inspired researchers for more than a century. Investigations into the neural basis of bilingualism focus not only on how two languages (L1 and L2) are represented in the brain (i.e., the anatomical location) but also on how these languages are processed. Indeed, the main assumption is that a “weaker” L2 may be processed through brain mechanisms that may differ from those underlying L1 processing. After a brief historical overview, I illustrate findings inherent to the representation of languages, followed by a section focusing on language processing.

Historical Overview From a historical standpoint, the first approach to studying brain organization for bilingualism was the study of bilingual aphasics. Several clinical aphasia studies have shown that bilingual aphasics do not necessarily manifest the same language disorders with the same degree of severity in both languages. In some cases, L1 is recovered better than a L2. In other cases, the converse obtains. Since the landmark 1895 study of the French neurologist Albert Pitres, who was the first to draw attention to the relative frequency of differential language recovery following aphasia in bilinguals, many different recovery patterns have been described: from selective recovery of a given language (i.e., one language remains impaired while the other recovers); parallel recovery of both languages, successive recovery (i.e., after the recovery process of one language, the other language recovers); alternating recovery (i.e., the language that was first recovered will be lost again due to the recovery of the language that was not first recovered); and alternating antagonistic recovery (i.e., on one day the patient is able to speak in one language while on the next day only in the other); to the pathological mixing of two languages (i.e., the elements of the two languages are involuntarily mixed during language production). The study of bilingual aphasia is important because it indicates the cortical regions necessary for performance of a linguistic task (e.g., speaking in L1). Clinical case reports indicate a set of relevant factors and have led to theoretical conjectures. However, at present we lack a causal account for the various recovery patterns and cannot predict clinical outcomes. Concerning the possible factors involved, no correlation has been found between the pattern of recovery and neurological, etiological, experiential, or linguistic parameters: not site, size or origin of lesion, type or severity of aphasia, type of bilingualism, language structure type, or factors related to acquisition or habitual use. Theoretical conjectures arising from the study of bilingual aphasia developed along two distinct lines, a more traditional approach and a more dynamic approach. The more traditional localizationist view argued, for instance, that the specific loss of one language would occur because the bilingual’s languages are represented in different brain areas or even in different hemispheres, and hence, a focal brain lesion within a languagespecific area may alter only that specific language, leaving the other language intact. In contrast, according to the dynamic view selective recovery arises because of compromise to the language system, rather than to damage to differential brain representations. A selective loss of a language arises because of increased inhibition, that is, of a raised activation threshold for the affected or lost language, or even because of an imbalance in the means to activate the language due to the lesion. It is worth underlining that Pitres himself proposed a dynamic explanation of language recovery in bilingual aphasics: Language recovery could occur only if the lesion had not entirely destroyed language areas but temporarily inhibited them through a sort of pathological inertia. In Pitres’s opinion, the patient generally first recovered the language to which she/he was premorbidly more exposed (not necessarily the native language) because the neural elements subserving “the more exposed language” were more strongly associated.

123

Bilingualism, Neurobiology of The dynamic view not only explains the so-called selective recovery of a language but can also explain many reported recovery patterns in bilingual aphasia. As outlined by M. Paradis (1998), a parallel recovery would then occur when both languages are inhibited to the same degree. When inhibition affects only one language for a period of time and then shifts to the other language (with disinhibition of the prior inhibited language) a pattern of alternating antagonistic recovery occurs (see Green 1986). Selective recovery would occur if the lesion permanently raised the activation threshold for one language, and pathological mixing among languages would occur when languages could no longer be selectively inhibited. In general, the aphasia data have provided a rich source of evidence on the range of language disorders and language recovery patterns in bilinguals. However, there are limitations to the generalizability of such data to neurologically healthy individuals. Concerns about the lesion-deficit approach include the inability to determine whether specific language deficits are the result of damage to a specialized language component at the lesion site, or if the damaged area is simply part of a larger neural network that mediates a given component of language. Likewise, aphasia data do not allow one to separate the effects of injury from those of neural plasticity or a reallocation of healthy cortical tissue for the mediation of language functions lost as a result of brain injury. Nevertheless, studying the effects of brain damage on linguistic function in bilinguals has led to a number of interesting observations about the nature and course of language impairment and recovery, which in turn has stimulated researchers to apply functional neuroimaging techniques to the investigation of bilingual language processing.

The Neural Representation of L2 Since its inception, neuroimaging work on bilinguals has been motivated by the same localizationist questions that run through the bilingual aphasia literature: whether multiple languages are represented in overlapping or separate cerebral systems. In addition, neuroimaging and neurophysiological data on this issue have often been influenced by possible biases, such as lack of information on the age of acquisition and degree of proficiency in the experimental subjects. Both these variables indeed exert profound influences on the brain organization of L2. According to psycholinguistic evidence grounded on the concept of universal grammar, the age of L2 acquisition is expected to be crucial for grammatical processing. In fact, grammatical processing may be particularly deficient when L2 is learned later in life. On the other hand, lexical-semantic processing seems to be less affected by age of acquisition than to depend on the degree of L2 proficiency. It is likely that other factors, such as usage and exposure to a given language, can affect brain plasticity mechanisms, leading to modifications of the neural substrate of language. I consider separately how these variables may influence L2 processing. An ongoing issue in neurobiology concerns the fact that the acquisition of language seems to depend on appropriate input during a biologically based critical periods. It has also been suggested that L2 learning may be subject to such crucial timelocked constraints. However, L2 can be acquired at any time in life, although L2 proficiency is rarely comparable to that of L1 if

124

L2 is acquired beyond the critical periods. The dependence of grammatical processing upon these age effects was confirmed by early event-related potentials (ERP) studies (Weber-Fox and Neville 1996) and by recent functional brain imaging studies (Wartenburger et al. 2003). In particular, I. Wartenburger and colleagues reported no differences in brain activations for grammar in L1 and L2 in very early (from birth) highly proficient bilinguals. On the other hand, late highly proficient bilinguals were in need of additional neural resources in order to achieve a comparable nativelike performance in grammatical tasks. The same did not apply to lexical-semantic processing, for which the only difference in the pattern of brain activity in bilinguals appeared to depend upon the level of attained proficiency. As mentioned, the degree of language proficiency seems to exert a more pervasive influence on the lexical-semantic level of L2. According to psycholinguistics, during the early stages of L2 acquisition there may be a dependency on L1 to mediate access to meaning for L2 lexical items. As L2 proficiency grows, this dependency disappears. Higher levels of proficiency in L2 produce lexical-semantic mental representations that more closely resemble those constructed in L1. According to D. W. Green’s “convergence hypothesis” (2003), any qualitative differences between native and L2 speakers disappear as proficiency increases. The convergence hypothesis claims that the acquisition of L2 arises in the context of an already specified or partially specified system and that L2 will receive convergent neural representation within the representations of the language learned as L1. Whether word or sentence production and word completion were used as experimental tasks, neuroimaging studies reported common activations in the left hemisphere when the degree of L2 proficiency was comparable to that of L1. This happened irrespective of the differences in orthography, phonology and syntax among languages. Conversely, bilinguals with low proficiency in L2 engaged additional brain activity, mostly in the left prefrontal cortex. Similar results were found in studies that did not directly address lexical retrieval, but employed judgment tasks in the lexical-semantic domain. It is worth underlining that the activity found in the left prefrontal cortex is located anteriorily to the classical language areas and, thus, not directly linked to language functions but rather linked to other cognitive functions, such as cognitive control and attention. Crucially, the engagement of the left prefrontal cortex was reported for bilinguals with a low degree of L2 proficiency and/or exposure. One may conclude that the differences found between high and low proficient bilinguals are not due to anatomical differences of L2 brain representations but instead reflect the cognitive dynamics of processing a weaker L2 as compared to L1.

Neural Aspects of L2 Processing One of the most salient aspects and one specific to bilingual language processing is language control. Language control refers to the fact that there may be competition between languages and that this competition is resolved by actively inhibiting the socalled non-target language. Consider that individuals can perform different actions on the same stimulus. For instance, a bilingual can name a presented word in L1 or translate it into L2. The task

Bilingualism, Neurobiology of goal must be maintained in the face of conflicting goals, and the various actions required to perform the task must be coordinated (e.g., retrieve or compute the word’s phonology from its spelling or retrieve the meaning of the word and select its translation). Once a given task is established, however (e.g., speaking in L2), competition with alternative possible tasks (speaking in L1) may be resolved more automatically. Where individuals wish to alter their goal (for example, to switch from speaking in one language to speaking in another), they must disengage from the current goal and switch to the new goal. Lexical concepts matching the intended language must be selected and produced, while those not matching the intended language must be inhibited through language control mechanisms. For instance, in word production studies, language control would inhibit potential interferences from the non-target language. Psycholinguistic evidence points to the fact that such interference is more common during production in a language that is mastered to a lower degree of proficiency, for example, a weak L2. In that case, for example, when asked to name a picture in L2, the bilingual speaker has to inhibit L1 in order to prevent a prepotent interference from L1. Functional neuroimaging studies using experimental tasks like picture naming, switching, translating, and so on have elegantly shown that these tasks are paralleled by the activation of a set of brain areas that are not directly linked to language representation, such as the brain activity within the left prefrontal cortex, the left caudate nucleus, and the anterior cingulate cortex. The engagement of these areas is even more relevant when subjects have to process a weak L2. The functions generally ascribed to the prefrontal cortex comprise working memory, response inhibition, response selection, and decision making, while the left caudate was reported to be imported for language selection and set switching. The anterior cingulate cortex is related to such functions as conflict monitoring, attention, and error detection. It becomes clear that the engagement of these structures provides a cerebral testimony of the cognitive processes inherent to bilingual language processing: competition and control between languages.

Conclusions Extensive reviews focusing on the bilingual brain as studied with functional neuroimaging are available in the literature to which the reader is referred (Abutalebi, Cappa, and Perani 2005; Perani and Abutalebi 2005; but see also Paradis (2004) for a critical viewpoint). In broad outlines, functional neuroimaging has shed new light on the neural basis of L2 processing and on its relationship to native language (L1). First of all, the long-held assumption that L1 and L2 are necessarily represented in different brain regions or even in different hemispheres in bilinguals has not been confirmed. On the contrary, functional neuroimaging has elegantly outlined that L1 and L2 are processed by the same neural devices. Indeed, the patterns of brain activation associated with tasks that engage specific aspects of linguistic processing are remarkably consistent among different languages, which share the same brain language system. These relatively fixed brain patterns, however, are modulated by a number of factors. Proficiency, age of acquisition, and exposure can affect the cerebral representations of each language, interacting in a complex way with the modalities of language performance.

Bilingualism and Multilingualism Consider as an example the complex process of L2 acquisition. This process may be considered as a dynamic process, requiring additional neural resources in the early stages of L2 acquisition. These additional neural resources are mostly found within the left prefrontal cortex (more anteriorily to the classical language areas), the left basal ganglia, and the anterior cingulated cortex and seem to be associated with the greater control demand when processing a “weaker” L2. However, once the L2 learner gains sufficient L2 proficiency, the neural representation of L2 converges to that of L1, at least at the macroanatomical level. At this stage, one may suppose that L2 is processed in the same fashion as L1, as psycholinguistic evidence points out (Kroll and Stewart 1994). This latter point is an important one because many functional neuroimaging studies did not take into consideration linguistic and psycholinguistic evidence (Paradis 2004). Yet evidence from neuroimaging should be integrated with the psycholinguistic findings to the mutual advantage of both research traditions. Integrating these findings with the psycholinguistic theory may allow us to demonstrate the biological consistency of different models, organize and consolidate existing findings, and generate novel insights into the nature of the cerebral organization of bilingualism. – Jubin Abutalebi WORKS CITED AND SUGGESTIONS FOR FURTHER READING Abutalebi J., S. F. Cappa and D. Perani. 2005. “Functional neuroimaging of the bilingual brain.” In Handbook of Bilingualism: Psycholinguistic Approaches, ed. J. F. K Kroll and A. De Groot, 497–515. Oxford: Oxford University Press. Green, D. W. 1986. “Control, activation and resource.” Brain and Language 27: 210–23. ———. 2003. “The neural basis of the lexicon and the grammar in L2 acquisition.” In The Interface Between Syntax and the Lexicon in Second Language Acquisition, ed. R. van Hout, A. Hulk, F. Kuiken, and R. Towell, 197–218. Amsterdam: John Benjamins. Kroll, J. F., and E. Stewart. 1994. “Category interference in translation and picture naming: Evidence for asymmetric connections between bilingual memory representations.” Journal of Language and Memory 33: 149–74. Paradis, M. 1998. “Language and communication in multilinguals.” In Handbook of Neurolinguistics, ed. B. Stemmer and H. Whitaker. San Diego, CA: Academic Press, 417–430. ———. 2004. A Neurolinguistic Theory of Bilingualism. Amsterdam and Philadelphia: John Benjamins. Perani, D., and J. Abutalebi. 2005. “Neural basis of first and second language processing.” Current Opinion of Neurobiology 15: 202–6. Wartenburger, I., H. R. Heekeren, J. Abutalebi, S. F. Cappa, A. Villringer, and D. Perani. 2003. “Early setting of grammatical processing in the bilingual brain.” Neuron 37: 159–70. Weber-Fox, C. M., and H. J. Neville. 1996. “Maturational constraints on functional specialization for language processing: ERP and behavioral evidence in bilingual speakers.” Journal of Cognitive Neuroscience 8: 231–56.

BILINGUALISM AND MULTILINGUALISM Growing recognition that bilingualism/multilingualism is not an exception or irregular phenomenon but is, in fact, a

125

Bilingualism and Multilingualism growing global phenomenon marks a new challenge and a shift for linguistic research. For instance, the traditional domain of psycholinguistic research, which has been the monolingual child, is now shifting to the bilingual child and multilingual language processing (see, e.g., de Bot and Kroll 2002, 133). What is bilingualism and who is a bilingual? The questions of identification and measurement that are considered irrelevant in the context of monolingualism become more pertinent and urgent in the context of bilingual language acquisition, production, comprehension, and processing.

Bilingualism/Multilingualism: Two Conceptual Views Is a bilingual a composite of two monolinguals? Does the bilingual brain comprise two monolinguals crowded into a limited space? For some researchers, the answer to these questions has traditionally been affirmative. Such a view of bilingualism is termed the fractional view. According to this view, monolingualism holds a key to the understanding of bilingualism. However, a more balanced and accurate picture of bilingualism emerges from the holistic view of bilingualism. According to this view, neither is a bilingual person the mere sum of two monolinguals nor is the bilingual brain a composite of two monolingual brains. The reason for this position is that the cooperation, competition, and coexistence of the bilingual’s two languages make a bilingual a very complex and colorful individual (for details, see Grosjean 1989).

Defining and Measuring Bilingualism: Input Conditions and Input Types Defining and measuring bilingualism is a very complex and uphill task due to the number and types of input conditions. For instance, while a monolingual child receives input from his or her parents only in one language in all settings, a bilingual child is provided input at least in two separate languages (e.g., oneparent one-language input; one-place one-language input) in addition to a code-mixed input in a variety of environments. In addition, biological (age of acquisition), sociopsychological, and other nonlinguistic factors lead to a varying degree of bilingual language competencies. Therefore, it is natural that no widely accepted definition or measure of bilingualism exists. Instead, a rich range of scales, dichotomies, and categories are employed to describe bilinguals. A bilingual who can speak and understand two languages is called a productive bilingual, whereas a receptive bilingual is an individual who can understand but cannot speak a second language. A child who has acquired two languages before the age of five at home (natural setting) is called a simultaneous or early bilingual, whereas those who learn a second language after the age of five, either at home or in school setting, are described as late or sequential bilinguals. Other labels and dichotomies, such as fluent versus nonfluent, balanced versus nonbalanced, primary versus secondary, and partial versus complete, are based upon different types of language proficiency (speaking, writing, listening) or on an asymmetrical relationship between the two languages. Compound versus coordinate bilingualism refers to the differential processing of language in the brain. Compound bilinguals process two languages using a common conceptual system, whereas coordinate bilinguals keep language separation at both

126

conceptual and linguistic levels (see bilingualism, neurobiology of). These labels and dichotomies demonstrate the complex attributes of bilingualism that make the task of defining and measuring bilinguals a daunting one. A working definition of bilingualism is offered by Leonard Bloomfied ([1933] 1984, 53), who claimed that a bilingual is one who has “native-like” control over two languages (i.e., balanced bilingual).

Bilinguals’ Language Organization Bilinguals’ organization of a verbal repertoire in the brain is also very different from that of monolinguals. When a monolingual decides to speak, his/her brain does not have to make complex decisions concerning language choice as does the bilingual. Such a decision-making process for a monolingual is restricted at best to the choice of a variety/style (informal vs. formal) selection. It is inconceivable for monolinguals to imagine that a multilingual person, such as this author, has to make a choice from among four languages and their varieties while communicating within his family in India. The language choice is not a random one but is unconsciously governed by a set of factors. The author is a speaker of Multani, Punjabi, Hindi, and English. Normally, he used Multani to talk with his brothers and parents while growing up. He speaks Punjabi with two of his sisters-in-law, Hindi with his nephews and nieces, and English with his children. In short, each language in his brain is associated with a well-defined domain. A violation of such a domain allocation has serious implications not only for communication mishaps but also for interpersonal relationships. In addition to the language-person domain allocation, other factors such as topics and emotions determine his language choice. While discussing an academic topic, he switches from Multani to English with his brothers and from English to Hindi with his children if the context is emotive. In short, the determinants of language choice are quite complex among bilinguals, and this, in turn, presents evidence that bilinguals’ organization of their verbal repertoire is quite different from monolinguals’ organization.It is interesting to note that language choice (or language negotiation) is a salient feature of bilingual linguistic competence and performance. The complexity of language choice and its unconscious determinants pose a serious challenge for the psycholinguistic theory of bilingual language production.

Individual, Societal, and Political Bilingualism Bilingualism can be viewed from individual, societal, and political perspectives. In a bilingual family, not all members are always bilinguals. Parents may be monolingual, while children may be bilinguals or vice versa. Societal factors such as the overt prestige of a language (or a presence of a majority language) often leads to individual or family bilingualism. However, individual or family bilingualism can persist even without societal support. Such bilingualism can be termed covert prestige bilingualism, which is often motivated by the consideration of group identity. In those societies of Asia or Africa where bilingualism exists as a natural phenomenon as the result of a centuries-long tradition of bilingualism, an ethnic or local language becomes a Low variety, that is, it is acquired at home and/or in an informal setting outside school (e.g., on a playground), whereas a language of wider communication or a prestige language functions as a High

Bilingualism and Multilingualism variety, which is learned formally in schools. In a diglossic society, a single language develops two distinct varieties, the L- and the H-variety. People become bilingual for a wide variety of reasons: immigration, jobs, marriage, or religion, among others. These factors create a language contact situation but do not always lead to stable bilingualism. For instance, it is well known that immigrant communities in the United States often give up their mother tongue in favor of English and become monolingual after a brief period of bilingualism. The classification of countries as monolingual, bilingual, or multilingual often refers to the language policies of a country, rather than to the actual incidence of bilingualism or multilingualism. Canada is a bilingual country in the sense that its language policies are receptive to bilingualism. It makes provision for learning French in those provinces that are Anglophone. Such a provision is called territorial bilingualism. However, it does not mean that everybody in Canada is a bilingual, nor does it mean that the country guarantees individual bilingualism (personality bilingualism) outside territorial bilingualism. In multilingual countries such as India, where 20 languages are officially recognized, the government language policies are receptive to multilingualism. India’s three-language formula is the official language policy of the country. In addition to learning Hindi and English, the conational languages, schoolchildren can learn a third language, spoken outside their state.

Bilingual Verbal Behavior: Language Separation and Language Integration Language separation and language integration are the two most salient characteristics of bilinguals and thus of the bilingual brain. Whenever deemed appropriate, bilinguals can turn off one language and turn on the other language. This enables them to switch from one language to another with the ease of a driver of a stick-shift car shifting into different gears whenever necessary. The fractional view of bilingualism can account for such a verbal behavior of bilinguals. In addition to keeping the two linguistic systems separate, bilinguals can also integrate the two systems by mixing two languages. Language mixing is a far more complex cognitive ability than language separation. The holistic view of bilingualism can account for these two types of competencies. Language mixing comes naturally to bilinguals. Therefore, it is not surprising that such mixed languages as Spanglish, Hinglish, Japlish, and Germlish are emerging around the globe. Contrary to the claims of earlier research, the grammar of language mixing is complex yet systematic. The search for explanations of cross-linguistic generalizations about the phenomenon of code mixing (particularly, code mixing within sentences) in terms of independently justified principles of language structure and use has taken two distinct forms. One approach is formulated in terms of the theory of linguistic competence, for example, Jeff MacSwan (2005). The other approach – as best exemplified by the Matrix Language Frame (MLF) model (Myers-Scotton and Jake 1995; see codeswitching) – is grounded in the theory of sentence production, particularly that of M. Garrett (1988) and W. Levelt (1989) (see Bhatia and Richie 1996, 655–7 for discussion). For further development of these ideas and a critique, see Bhatia and Ritchie (1996) and MacSwan (2005).

Effects of Bilingualism/Multilingualism What is the effect of bilingualism/multilingualism on an individual, particularly on a child? The research on this question is fundamentally driven by two hypotheses: the linguistic deficit hypothesis and the linguistic augmentation hypothesis. According to the former, bilingual children show serious linguistic and cognitive adverse effects of bilingualism. Exposure to two languages leads to semilingualism; that is, they become deficient in both languages, which in turn leads to other disabilities (e.g., stuttering) and cognitive impairments (low intelligence, mental retardation, and even schizophrenia). Such a hypothesis has become obsolete in light of the findings of the research driven by the linguistic augmentation hypothesis. Solid on theoretical and methodological grounds, research by Elizabeth Peal and Wallace E. Lambert (1962) put to rest such negative and frightening effects of bilingualism. Their research and the findings of the succeeding research provide ample evidence that the negative conclusions of the earlier research were premature and misguided due to the theoretical and methodological flaws. Contrary to the findings of the previous research, bilingual children exhibit more cognitive flexibility than do monolinguals and perform better on verbal and nonverbal measures. Peal and Lambert’s study, which was conducted in Montreal, revolutionized research on bilingualism and multilingualism by highlighting a positive conception of bilinguals. Their research has been replicated in many countries, confirming the augmenting rather than subtracting effect of bilingualism. Beyond this research, the economic, communicative (intergenerational and cross-cultural), and relational (building relations) advantages of bilingualism are inarguable.

Conclusion In short, bilingualism/mulitilingualism is a global phenomenon that continues to gain further momentum in the age of globalization. It is a by-product of a number of biological, sociopsychological, and linguistic factors. These factors lead to individuals with varying degree of language competencies. Therefore, it is not surprising that defining and measuring bilingualism/multilingualism continues to be a challenging task. Bilinguals are complex and colorful in the way they manage and optimize their linguistic resources. For that reason, they are not a sum of two monolinguals. Language mixing and shifting are two defining characteristics of bilinguals. Current socio- and psycholinguistic research attempts to account for these two salient properties of the bilingual brain. – Tej K. Bhatia WORKS CITED AND SUGGESTIONS FOR FURTHER READING Bhatia, Tej K., and William C. Ritchie, eds. 2006. The Handbook of Bilingualism. Oxford: Blackwell. This book presents a multidisciplinary and comprehensive collection of state-of-the-art research on bilingualism and multilingualism. Chapters 7 and 8 deal with bilingual production models, including the discussion of Garrett 1988 and Levelt 1989. Bhatia, Tej K., and William C. Ritchie. 1996. “Bilingual language mixing, Universal Grammar, and second language acquisition.” In Handbook of Second Language Acquisition, ed. W. C. Ritchie and T. K. Bhatia, 627–82. San Diego, CA: Academic Press. Bloomfield, Leonard. [1933] 1984. Language. Chicago: University of Chicago Press.

127

Binding de Bot, Kees, and Judith F. Kroll. 2002. “Psycholinguistics.” In An Introduction to Applied Linguistics, ed. Norbert Schmitt, 133–49. London: Arnold. Edwards, John. 2006. “Foundations of bilingualism.” In Bhatia and Richie 2006, 7–31. Garrett, M. E. 1988. “Process in sentence production.” In The Cambridge Linguistic Survey. Vol. 3. Ed. F. Newmeyer, 69–96. Cambridge: Cambridge University Press. Grosjean, Francois. 1989. “Neurolinguists, beware! The bilingual is not two monolinguals in one person.” Brain and Language 36: 3–15. Hakuta, Kanji. 1986. Mirror of Language. New York: Basic Books. This work offers an excellent multidisciplinary account of bilingualism in general and bilingualism in the United States in particular. Among other topics, it presents an excellent account of the linguistic deficiency and linguistic augmentation hypotheses. Levelt, W. 1989. Speaking: From Intention to Articulation. Cambridge, MA: MIT Press. MacSwan, Jeff. 2005. “Remarks on Jake, Myers-Scotton and Gross’s response: There is no ‘Matrix Language.’” Language and Cognition 8.3: 277–84. Myers-Scotton, Carol and J. Jake. 1995. “Matching lemmas in a bilingual language competence and production model: Evidence from intrasentential code switching.” Linguistics 33: 981–1024. Peal, Elizabeth, and Wallace E. Lambert. 1962. “Relation of bilingualism to intelligence.” Psychological Monographs 76: 1–23. Ritchie, William C., and Tej K. Bhatia. 2007. “Psycholinguistics.” In Handbook of Educational Linguistics, ed. Bernard Spolsky and Francis Hult, 38–52. Oxford: Blackwell.

logic: The bindee must be contained in the sister constituent to the binder, a relation usually called c(onstituent)-command (see c-command). For movement relations, this amounts to the ban on downward or sideways movement, the proper binding condition, which is pervasive across languages. For quantifierpronoun relations, it blocks sideways binding as in (4a) (neither noun phrase [NP] c-commands the other), and upward binding as in (4b) (the putative binder is c-commanded by the pronoun); note that in both examples the pronouns have to be interpreted as referentially independent of no one/actress: (4) a. If no one is here, he’s elsewhere. b. Her calendar showed that no actress had left early.

A systematic class of exceptions to the c-command requirement is found in so-called indirect binding, for example (5), where the object can be bound from within the subject (sideways): (5)

Unlike semantic binding, coreference among two NPs does not require c-command: (6)

(1)

Every cat chased its tail.

(2) a. Sue hopes that she won. b. Edgar spoke for himself. c. Wesley called PRO to apologize. (3) a. Which book did Kim read t? b. Antonia was promoted t.

Semantically, only (1) and (3) are clear instances of binding (the pronouns/traces are interpreted like variables, and their antecedents are often nonreferring), yet coreference is almost universally subsumed under the binding label in linguistics. All three binding relations are frequently represented by coindexing the binder (or antecedent) and the bound element (e.g., Every cat6 chased its6 tail), though richer, asymmetrical representations have been proposed and are arguably required for semantic interpretation. Semantic binding relations are subject to a structural constraint, to a first approximation, the same as in quantified

128

His/Jacques’s teacher said that he/Jacques failed.

Yet certain prohibitions against coreference, for example, that nonreflexive pronouns in English cannot corefer with expressions in the same finite clause, only regard NPs that c-commands the pronoun, (7) (similarly for nonpronominal NPs): Your mother

BINDING In quantified logic, binding names the relation between a quantifier and one or more variables, for example, ∀x and x in ∀x[P(x) → Q(x)]. In linguistics, the term has been used in at least three domains: first, for the relation between quantified expressions and pronouns that referentially depend on them, (1); second for coreference, the relation between two referring expressions with the same referent, (2), including hypothesized empty pronouns, (2c); and third, in theories that assume transformations, for the relation between a dislocated phrase and its trace, (3) (see government and binding theory):

Somebody from every city likes its beaches.

(7)

defended you.

*You Likewise, reflexive pronouns in English need an antecedent that is not just within the same finite clause but also c-commands them: She (8)

defended herself.

*Her mother These conditions on the distribution of reflexive and nonreflexive restrict binding by quantified nominals as well and are indiscriminately referred to as binding conditions. While c-command seems relevant in binding conditions cross-linguistically, other aspects, such as the number of morphological classes (reflexives, nonreflexives, etc.) or the size of relevant structural domains, vary widely. – Daniel Büring WORKS CITED AND SUGGESTIONS FOR FURTHER READING Büring, Daniel. 2005. Binding Theory. Cambridge Textbooks in Linguistics. Cambridge: Cambridge University Press. Chomsky, Noam. 1981. Lectures on Government and Binding. Dordrecht, the Netherlands: Foris Publications. Dalrymple, Mary. 1993. The Syntax of Anaphoric Binding. Stanford, CA: CSLI. Kuno, Susumo. 1987. Functional Syntax – Anaphora, Discourse and Empathy. Chicago: Chicago University Press.

Biolinguistics

BIOLINGUISTICS Biolinguistics is the study of the biology of language. The modern biolinguistic program was initiated by Noam Chomsky in the 1950s (Chomsky 2006), although it has much earlier historical antecedents (see cartesian linguistics). It investigates the form and function of language, the development (ontogeny) of language, and the evolution of language (phylogeny), among other topics. Biolinguists study such questions as the following: (1) What is knowledge of language? (2) How does knowledge of language develop in the child? (3) How does knowledge of language evolve in the species? To answer the question of what knowledge of language is (1), biolinguists have proposed various generative grammars, that is, explicit models of the faculty of language. The study of generative grammars draws from a variety of areas, including syntax, semantics, the lexicon, morphology, phonology, and articulatory and acoustic phonetics. In addition, the biolinguist investigates the neurological mechanisms underlying the faculty of language (see

syntax, neurobiology of; semantics, neurobiology of; morphology, neurobiology of; phonetics and phonology, neurobiology of). Such studies of brain and language include studies of expressive and receptive aphasia, split brain patients, neuroimaging, and the electrical activity of the brain. The biolinguist also studies performance models (language processing), including parsing, right hemisphere lan-

guage processing, left hemisphere language processing, and speech perception. To answer the question of how knowledge of language develops in the child (2), one may visualize this as the study of the language acquisition device: experience → ? → language (English, Japanese, etc.)

where the box represents what the child brings to language learning. We ask how the child maps experience (primary linguistic data) to a particular language. It is posited that the child moves through a number of states from an initial state, corresponding to the child’s genetic endowment, to a final state, corresponding to a particular language. For each subarea discussed, biolinguistics studies the development or growth of language, often referred to as language acquisition (e.g., syntax, acquisition of; semantics, acquisition of; and phonology, acquisition of). The initial state may be characterized by a universal grammar, which is a set of general principles with parameters that are set by experience, thus accounting for the variation across languages. For example, there are general principles of word order that permit some variation; for example, the verb precedes the object (English) or the verb follows the object (Japanese) (see x-bar theory). Such a theory is referred to as a principles and parameters theory. (For specific subareas, see syntax,

universals of; semantics, universals of; morphology, universals of; and phonology, universals of.) For some different parametric proposals, see the microparametric approach

of Richard Kayne’s (2004) and Charles Yang’s (2004) work on competitive theories of language acquisition. (From other perspectives, see universals, nongenetic; absolute and statistical universals, implicational universals, and typological universals.) In addition to comparative grammar (see also morpholog-

ical typology), universals of language change, syntactic change, semantic change, pidgins, and creoles provide additional evidence for the nature of universal grammar and language acquisition. Moreover, the study of genetic language disorders, as well as familial and twin studies, has been very fruitful for the study of language acquisition (see genes and language; specific language impairment; see also the extensive literature on the FOXP2 gene [Marcus and Fisher 2003]). Studies of language-isolated children provide information about the critical period for language learning. The study of sign languages has been invaluable for investigating language outside the modality of sound (see also sign language, acquisition of; sign languages, neurobiology of). Finally, the study of linguistic savants has been quite useful for delineating the modularity of the language faculty as distinct from other cognitive faculties. To answer the question of how knowledge of language evolves in the species, (3), biolinguists integrate data from a variety of areas, including comparative ethology (see Hauser, Chomsky, and Fitch 2002; see also animal communication and human language; speech anatomy, evolution of), comparative neuroanatomy, and comparative genomics. Since evolution of language took place in the distant past, mathematical modeling of populations of speaker-hearers has recently attracted much interest in work on dynamical systems (see self-organizing systems). Such studies have proven useful not only for the study of evolution but also for the study of language acquisition and change. (For some hypotheses on origins of language, see origins of language; grooming, gossip, and language.) Questions (1) to (3) might be called the what and how questions of biolinguistics. One can also ask why the principles of language are what they are, a deeper and more difficult question to answer. The investigation into why questions is sometimes referred to as the minimalist program or minimalism. In addition, there is the related question of how the study of language can be integrated with the other natural sciences, a problem that Chomsky calls the “unification problem” (see Jenkins 2000). All of these questions are certain to continue to fascinate investigators of the biology of language for decades to come. (For more information on other explicit models of the language faculty, see transformational grammar; standard

theory and extended standard theory; categorial grammar; head-driven phrase structure grammar; lexical-functional grammar; optimality theory; role and reference grammar; cognitive grammar; connectionism and grammar; construction grammars). – Lyle Jenkins WORKS CITED AND SUGGESTIONS FOR FURTHER READING Chomsky, Noam. 2006. “Biolinguistics and the human capacity.” In Language and Mind, 173–85. Cambridge: Cambridge University Press.

129

Birdsong and Human Language Hauser, M. D., N. Chomsky, and W. Tecumseh Fitch. 2002. “The faculty of language: What is it, who has it, and how did it evolve?” Science 298: 1569–79. Kayne, Richard. 2004. “Antisymmetry and Japanese.” In Variation and Universals in Biolinguistics, ed. Lyle Jenkins, 3–5. Amsterdam: Elsevier. Jenkins, Lyle. 2000. Biolinguistics: Exploring the Biology of Language. Cambridge: Cambridge University Press. Marcus, G. F., and S. E. Fisher. 2003. “FOXP2 in focus: What can genes tell us about speech and language?” Trends in Cognitive Sciences 7.6: 257–62. Yang, Charles. 2004. “Toward a theory of language growth.” In Variation and Universals in Biolinguistics, ed. Lyle Jenkins, 37–56. Amsterdam: Elsevier.

BIRDSONG AND HUMAN LANGUAGE Language is often claimed to be uniquely human (see animal communication and human language ). This belief has discouraged efforts to identify potential animal models of language, even though animal models have been essential in ascertaining the neurobiology of other cognitive functions. It is conceivable, however, that useful homologies or analogies exist between human language and the communicative systems of other species, even if language is unique in some respects. One particularly interesting homology might exist between human language and birdsong. Songbirds rely on a specialized frontal lobe–basal ganglia loop to learn, produce, and perceive birdsong (Brenowitz and Beecher 2005) (see also broca’s area). Disruptions to this circuit disrupt the sensorimotor learning needed to acquire song, and also the sequencing skills needed to produce and properly perceive it. Recent work has revealed a remarkable homology in this circuit between birds and mammals (Doupe et al. 2005). The homologous circuit in human and nonhuman primates involves loops connecting many regions in the frontal cortex to the basal ganglia. Afferents from the frontal cortex densely innervate the striatum of the basal ganglia, which also receives inputs from many other areas of the cortex. The striatum seems to control behavioral sequencing in many species (Aldridge and Berridge 1998). Spiny neurons, the principal cells of the striatum, have properties that make them ideal for recognizing patterned sequences across time (Beiser, Hua, and Houk 1997). Damage to this loop in primates produces problems with motor and cognitive skills that require planning and manipulating patterns of sequences over time (Fuster 1995). These observations lend plausibility to the notion that the frontal cortex–basal ganglia circuit might play a role in the syntax of human language. If so, then it is probably not coincidental that the acquisition of human language and birdsong have compelling parallels (Doupe and Kuhl 1999). Humans and songbirds learn their complex, sequenced vocalizations in early life. They similarly internalize sensory experience and use it to shape vocal outputs, by means of sensorimotor learning and integration. They show similar innate dispositions for learning the correct sounds and sequences; as a result, humans and some species of songbird have similar critical periods for vocal learning, with a much greater ability to learn early in life. These behavioral

130

parallels are what one would expect if both species rely on a similar neural substrate for learning and using their communicative systems. Relevant genetic evidence is also available. The much-discussed FOXP2 gene is similarly expressed in the basal ganglia of humans and songbirds (Teramitsu et al. 2004; Vargha-Khadem et al. 2005). A FOXP2 mutation in humans results in deficits in language production and comprehension, especially aspects of (morpho)syntax that involve combining and sequencing linguistic units (Marcus and Hisher 2003; Vargha-Kadham et al. 2005). One of the neurobiological effects of the mutation is a notable reduction in the gray matter of the striatum (Vargha-Kadham et al. 2005). Perhaps, then, the combinatorial aspects of human language were enabled by the preadaptation of an anterior neural circuit that has been highly conserved over evolutionary time and across species, and by a genetic mutation in this circuit that increased its computational space. Finally, some birdsong, like human language, is compositional; songbirds learn units and rules of combination (Rose et al. 2004), although the rules of combination are obviously far less sophisticated than those that characterize human language. A skeptic might argue that the syntax of human language is too complex (too highly structured, too recursive, too creative; see recursion, iteration, and metarepresentation ) to be modeled as a simple patterned sequence processor that relies on associative learning mechanisms. In fact, the explanatory burden placed on rule-based, recursive syntax has diminished over recent decades. Modern grammars tend to be lexicalist in nature; that is, much of the knowledge relevant to sentence structure is stored in the lexicon with individual words, rather than being computed by abstract phrase structure rules (see lexical-functional grammar ). Recursion, while clearly a characteristic of human language, is much more limited in actual language usage than would be predicted given the standard model. And, because conceptual knowledge (see semantics) has its own structure (Jackendoff 1990), it seems plausible that some of the burden for structuring the input rests with the conceptual stream (Jackendoff 2002), rather than entirely with the syntax. Birds and humans are fundamentally different in many ways, as are their systems of communication. Nonetheless, birds and humans are two of only a handful of vocal learners, and recent work points to communication-relevant homologies and similarities. It is not unreasonable to think that a comparative approach might provide important clues to how language evolved and, perhaps, to the nature of language itself. – Lee Osterhout WORK CITED AND SUGGESTIONS FOR FURTHER READING Aldridge, J. Wayne, and Kent C. Berridge. 1998. “Coding serial order by neostriatal neurons: A ‘natural action’ approach to movement sequence.” Journal of Neuroscience 18: 2777–87. Beiser, David G., Sherwin S. Hua, and James C. Houk. 1997. “Network models of the basal ganglia.” Current Opinion in Neurobiology 7: 185–90. Brenowitz, Eliot, and Michael D. Beecher. 2005. “Song learning in birds: Diversity and plasticity, opportunities and challenges.” Trends in Neurosciences 28: 127–32.

Blended Space

Blindness and Language constructs in the conceptual blending framework. Some researchers use blended space and blend interchangeably to refer to the particular kind of mental space described here (e.g., Fauconnier and Turner 1994). Elsewhere blend is used to describe the entire integration network, as in double-scope blend (e.g., Núñez 2005), or the process of generating such a network, as in running the blend (e.g., Fauconnier and Turner 2002, 48). Where the use may be ambiguous, blended space provides maximal clarity.

Doupe, Allison J., and Patricia Kuhl. 1999. “Birdsong and human speech: Common themes and mechanisms.” Annual Review of Neuroscience 22: 567–631. Doupe, Allison J., David J. Perkel, Anton Reiner, and Edward A. Stern. 2005. “Birdbrains could teach basal ganglia research a new song.” Trends in Neurosciences 28: 353–63. Fuster, Joaquin M. 1995. Memory in the Cerebral Cortex: An Empirical Approach to Neural Networks in the Human and Nonhuman Primate. Cambridge, MA: MIT Press. Jackendoff, Ray. 1990. Semantic Structures. Cambridge, MA: MIT Press. ———. 2002. Foundations of Language: Brain, Meaning, Grammar, Evolution. New York: Oxford University Press. Lieberman, Philip. 2000. Human Language and Our Reptilian Brain. Cambridge: Harvard University Press. Marcus, Gary F., and Simon E. Fisher. 2003. “FOXP2 in focus: What can genes tell us about speech and language?” Trends in Cognitive Sciences 7: 257–62. Rose, Gary, Franz Goller, Howard J. Gritton, Stephanie L. Plamondon, Alexander T. Baugh, and Brendon G. Cooper. 2004. “Species-typical songs in white-crowned sparrows tutored with only phrase pairs.” Nature 432: 753–8. Teramitsu, Ikuku, Lili C. Kudo, Sarah E. London, Daniel H. Geschwind, and Stephanie A. White. 2004. “Parallel FOXP1 and FOXP2 expression in songbirds and human brain predicts functional interaction.” Journal of Neuroscience 24: 3152–63. Vargha-Khadem, Faraneh., David G. Gadian, Andrew Copp, and Mortimer Mishkin. 2005. “FOXP2 and the neuroanatomy of speech and language.” Nature Reviews: Neuroscience 6: 131–8.

Fauconnier, Gilles, and Mark Turner. 1994. “Conceptual projection and middle spaces.” UCSD Department of Cognitive Science Technical Report 9401. ———. 1998. “Principles of conceptual integration.” In Discourse and Cognition, ed. Jean-Pierre Koenig, 269–83. Stanford, CA: CSLI Publications. ———. 2002. The Way We Think: Conceptual Blending and the Mind’s Hidden Complexities. New York: Basic Books. Grady, Joseph, Todd Oakley, and Seana Coulson. 1999. “Conceptual blending and metaphor.” In Metaphor in Cognitive Linguistics, ed. Raymond W. Gibbs, Jr., and Gerard J. Steen, 101–24. Amsterdam and Philadelphia: John Benjamins. Núñez, Rafael E. 2005. “Creating mathematical infinities: Metaphor, blending, and the beauty of transfinite cardinals.” Journal of Pragmatics 37: 1717–41.

BLENDED SPACE

BLINDNESS AND LANGUAGE

A blended space is one element of the model of meaning construction proposed by conceptual blending theory. In this framework, mental representations are organized in small, selfcontained “conceptual packets” (Fauconnier and Turner 2002, 40) called mental spaces, which interconnect to form complex conceptual networks. In a conceptual integration network, or blend, some mental spaces serve as “input spaces” that contribute elements to a new, blended mental space (Fauconnier and Turner 1994, 1998, 2002). The minimal conceptual integration network connects four mental spaces: two inputs, a generic space that contains all the structures that the inputs seem to share, and a blended space. The conventional illustration of this prototypical network, sometimes called the “Basic Diagram” (Fauconnier and Turner 2002, 46–7), shows four circles marking the points of a diamond, with the circle representing the generic space at the top and the blended space at the bottom. However, this four-space model is only the minimal version of the integration network; in conceptual blending theory, networks can contain any number of input spaces. Blended spaces can also serve as inputs to new blends, making elaborate “megablends” (Fauconnier and Turner 2002, 151–3). What makes a blended space special is that it contains newly emergent structure that does not come directly from any of the inputs. For example, understanding This surgeon is a butcher involves selective projection from inputs of butchery and surgery, but the inference that the surgeon is incompetent arises only in the blended space. There is some potential for confusion regarding the terminology used to distinguish blended spaces from other theoretical

Reading by Touch

– Vera Tobin WORKS CITED AND SUGGESTIONS FOR FURTHER READING

Blind people achieve literacy by reading braille, a tactile coding system for reading and writing. Coding is based on raised dots arranged in rectangular cells that consist of paired columns of three dots each. Patterns of one or more dots represent letters, numbers, punctuation marks, or partial and whole word contractions (Figure 1). Initially, braille was coded for the Latin alphabets of French or English. For languages with non-Latin alphabets, braille patterns are assigned according to a transliteration of the Latin alphabet. For example, the third Greek letter gamma has the dot pattern for the third Latin letter c. Chinese and other Asian languages use phonetic adaptations of braille. Chinese braille codes syllables into one, two, or three patterns for, respectively, an initial consonant sound, a final vowel sound, and a word tone. There are no braille patterns for individual Chinese ideograms. Japanese orthography is more complex, as it includes a combination of Kanji (ideograms imported from China), Kana (phonograms), Western alphabet, and Arabic numerals. Kanji is converted to Kana first before translation to braille. While alphabet represents a single sound, Kana represents a syllable (a consonant and a vowel). Standard braille in English and many European languages is usually read in a contracted form (Grade II) in which selected single patterns signify commonly used words, part-words, or syllables. Hence, many words require only one, two, or three braille cells and spaces, which reduce reading effort and space for text. The same braille pattern can represent a letter or a contraction, depending on context, thus expanding 63 to 256 interpretable dot patterns in Grade II English braille. Although all alphabet-

131

Blindness and Language

Figure 1. American standard braille cell patterns for alphabet, punctuation marks, some contractions, and whole words.

based languages use the same braille patterns, associated contractions vary. Thus, multilingual reading requires the learning of language-unique contractions. During reading, scanning movement across text evokes intermittent mechanical stimulation from contacting successive braille cells, which activates most low-threshold cutaneous mechanoreceptors found in the fingertip (Johnson and Lamb 1981). A spatial-temporal transformation of the evoked peripheral activity indicates an isomorphic reproduction of braille cell shapes across a population of mechanoreceptors (Phillips, Johansson, and Johnson 1990). Through connecting sensory pathways, these physiological representations of braille cell shape are conveyed to primary somatosensory cortex (Phillips, Johnson, and Hsaio 1988) in the parietal lobe. Despite the expected isomorphic representation of braille cell shapes in somatosensory cortex, we do not know whether tactile reading in fact relies on holistically discriminating shape. Braille cell patterns also differ in the density of dot-gaps, which is perceived as variations in texture (Millar 1985). These texture changes produce a dynamically shifting lateral mechanical shearing across the fingertip as it moves over braille text in fluent reading. Good braille readers attend to these temporally extended stimulation patterns, as opposed to global-holistic spatial shapes (Millar 1997, 337). In addition, top-down linguistic content drives perceptual processing in skillful readers, for whom the physical attributes of the text are subservient to lexical content. In other words, they do not puzzle out words, letter by letter; instead, they recognize them due in part to their physical properties but also to semantic context, the familiarity of words stored in their mental lexicon, and so on. Less accomplished readers trace shape by making more disjointed trapezoidal finger movements over individual cells, a strategy that fluent readers utilize when asked to identify particular letters, which is a shape-based task (Millar 1997, 337).

132

Braillists generally prefer bimanual reading (Davidson, Appelle, and Haber 1992), with each hand conveying different information. While one hand reads, the second marks spatial position in the text (e.g., lines, locations within lines, spaces between braille cells or words). Photographic records reveal skin compression of only one fingertip even during tandem movements across text; there is no coincident reading of different braille cells by multiple fingers (Millar 1997, 337). Text and spatial layout are tracked simultaneously in bimanual reading; there is no “best hand” (Millar 1984). Some individuals read an initial line segment with the left and a final segment with the right hand (Bertelson, Mousty, and D’Alimonte 1985). Despite bimanual reading, the left hemisphere is generally dominant for language even in left-handed braillists (Burton et al. 2002a).

Visual Cortex Contribution to Language Blindness requires numerous adjustments, especially for language. These adjustments appear to involve substantial reorganization of the visual cortex (occipital lobe), which in sighted people is dominated by visual stimulation. In blind people, the visual cortex responds more readily to nonvisual stimulation and especially contributes to language processing. A clinical case study of a congenitally blind, highly fluent braille reader is particularly salient. Following a bilateral posterior occipital ischemic stroke, she lost the ability to read braille (Hamilton et al. 2000). However, auditory and spoken language were unimpaired, and she retained normal tactile sensations on her braille reading hand despite a destroyed visual cortex. A similar but transient disruption in tactile reading occurs in congenitally blind people following repetitive transcranial magnetic stimulation (rTMS) to occipital cortex (Hamilton and Pascual-Leone 1998; Pascual-Leone et al. 2005). The obvious explanation for these observations is that the visual cortex reorganizes after blindness. But things are more complex. First, occipital cortex normally processes some tactile

Blindness and Language information in sighted people, especially following short periods of visual deprivation. Thus, blindfolding sighted people for five days, during which they train to discriminate braille letters, leads to visual cortex activity to tactile stimulation and sensitivity to disrupting braille letter discrimination by occipital rTMS (Pascual-Leone and Hamilton 2001). Even without visual deprivation, occipital rTMS impairs macrogeometric judgments of raised-dot spacing in sighted people (Merabet et al. 2004). These findings indicate that visual cortex normally contributes to some tactile discrimination. Brain-imaging studies have dramatically revealed the role of visual cortex in language for blind people. For example, generating a verb to an offered noun activates visual cortex in blind people (see Color Plate 2), irrespective of whether the noun is read through braille (Burton et al. 2002a) or heard (Burton et al. 2002b). In early blind individuals, verb generation engages both lower-tier (e.g., V1, V2, VP) and higher-tier (e.g., V7, V8, MT) visual areas (Color Plate 2). Similar adaptations occur in late blind individuals, though fewer areas are affected. The semantic task of discovering a common meaning for a list of heard words also evokes extensive visual cortex activation in the early blind and a smaller distribution in the late blind (Color Plate 2). Similar distributions of visual cortex activity occur when the early blind listen to sentences with increased semantic and syntactic complexity (Röder et al. 2002) and during a semantic retrieval task (Noppeney, Friston, and Price 2003). A phonological task of identifying a common rhyme for heard words activates nearly all visual areas bilaterally in early blind but few in late blind people (Color Plate 2). The sublexical task of identifying block capital letters translated passively across a fingertip activates only parts of visual cortex in early and late blind people (Color Plate 2). In general, semantic language tasks activate a greater extent of visual cortex than lower-level language and most perceptual tactile or auditory processing tasks. The functional relevance of occipital cortex to semantic processing is demonstrated when rTMS over left visual cortex transiently increases errors in verb generation to heard nouns in the early blind without interrupting the articulation of words (Amedi et al. 2004). Of course, performing any semantic task depends on retrieving word associations. Thus, two studies report a stronger relationship between retention performance and visual cortex activity, predominantly in V1, than with verb generation. V1 response magnitudes correlate more positively with verbal retention (Amedi et al. 2003) and the accuracy of long-term episodic memory (Raz, Amedi, and Zohary 2005) in congenitally/early blind participants. As in sighted people, blind individuals still utilize traditional left-lateralized frontal, temporal and parietal language areas (Burton et al. 2002a, 2002b; Burton, Diamond, and McDermott 2003; Noppeney, Friston, and Price 2003; Röder et al. 2002). Thus, the visual cortex activity represents an addition to the cortical language-processing areas. Visual cortex activity distributes bilaterally in all blind people for semantic tasks. However, left visual cortex is more active in early blind individuals (Color Plate 2). In contrast, right hemisphere responses predominate in late blind participants when they read braille with the right hand but are more symmetrically bilateral when verbs are generated to heard nouns (Color Plate 2). It is currently unknown whether reorganized visual cortex contains specific language domains.

Neuroplasticity has been observed in visual cortex of blind individuals at all ages of blindness onset. Such observations garner surprise only when primary sensory areas are viewed as unimodal processors that funnel computations into a cascade of cortical areas, including multisensory regions. Activation of reorganized visual cortex by nonvisual stimulation most parsimoniously reflects innate intracortical connections between cortical areas that normally exhibit nonvisual and multisensory responsiveness in sighted people. The demanding conditions of blindness possibly alter and expand the activity in these connections and thereby reallocate visual cortex to language processing. – Harold Burton WORKS CITED AND SUGGESTIONS FOR FURTHER READING Amedi, A., N. Raz, P. Pianka, R. Malach, and E. Zohary. 2003. “Early ‘visual’ cortex activation correlates with superior verbal memory performance in the blind.” Nature Neuroscience 6: 758–66. Amedi, A., A. Floel, S. Knecht, E. Zohary, and L. G. Cohen. 2004. “Transcranial magnetic stimulation of the occipital pole interferes with verbal processing in blind subjects.” Nature Neuroscience 7: 1266–70. Bertelson, P., P. Mousty, and G. D’Alimonte. 1985. “A study of braille reading: 2. Patterns of hand activity in one-handed and two-handed reading.” Quarterly Journal of Experimental Psychology A: Human Experimental Psychology 37: 235–56. Burton, H., J. B. Diamond, and K. B. McDermott. 2003. “Dissociating cortical regions activated by semantic and phonological tasks to heard words: A fMRI study in blind and sighted individuals.” Journal of Neurophysiology 90: 1965–82. Burton, H., D. G. McLaren, and R. J. Sinclair. 2006. “Reading embossed capital letters: A fMRI study in blind and sighted individuals.” Human Brain Mapping 27: 325–39. Burton, H., A. Z. Snyder, T. E. Conturo, E. Akbudak, J. M. Ollinger, and M. E. Raichle. 2002a. “Adaptive changes in early and late blind: A fMRI study of braille reading.” Journal of Neurophysiology 87: 589–611. Burton, H., A. Z. Snyder, J. Diamond, and M. E. Raichle. 2002b. “Adaptive changes in early and late blind: A fMRI study of verb generation to heard nouns.” Journal of Neurophysiology 88: 3359–71. Davidson, P. W., S. Appelle, and R. N. Haber. 1992. “Haptic scanning of braille cells by low- and high-proficiency blind readers.” Research in Developmental Disabilities 13: 99–111. Hamilton, R., J. P. Keenan, M. Catala, and A. Pascual-Leone. 2000. “Alexia for braille following bilateral occipital stroke in an early blind woman.” Neuroreport 11: 237–40. Hamilton, R., and A. Pascual-Leone. 1998. “Cortical plasticity associated with braille learning.” Trends in Cognitive Sciences 2: 168–74. Johnson, K. O., and G. D. Lamb. 1981. “Neural mechanisms of spatial tactile discrimination: Neural patterns evoked by braille-like dot patterns in the monkey.” Journal of Physiology (London) 310: 117–44. Merabet, L., G. Thut, B. Murray, J. Andrews, S. Hsiao, and A. PascualLeone. 2004. “Feeling by sight or seeing by touch?” Neuron 42: 173–9. Millar, S. 1984. “Is there a ‘best hand’ for braille?” Cortex 20: 75–87. ———. 1985. “The perception of complex patterns by touch.” Perception 14: 293–303. ———. 1997. Reading by Touch. London: Routledge. Noppeney, U., K. J. Friston, and C. J. Price. 2003. “Effects of visual deprivation on the organization of the semantic system.” Brain 126: 1620–7. Pascual-Leone, A., A. Amedi, F. Fregni, and L. B. Merabet. 2005. “The plastic human brain cortex.” Annual Review of Neuroscience 28: 377–401. Pascual-Leone, A., and R. Hamilton. 2001. “The metamodal organization of the brain.” Progress in Brain Research 134: 427–45.

133

Bounding (3) *[Handsomei though [S I believe [NP the claim that [S Dick is ti]]], I’m still going to marry Herman.

Phillips, J., R. Johansson, and K. Johnson. 1990. “Representation of braille characters in human nerve fibers.” Experimental Brain Research 81: 589–92. Phillips, J. R., K. O. Johnson, and S. S. Hsiao. 1988. “Spatial pattern representation and transformation in monkey somatosensory cortex.” Proceedings of the National Academcy of Sciences (USA) 85: 1317–21. Raz, N., A. Amedi, and E. Zohary, E. 2005. “V1 activation in congenitally blind humans is associated with episodic retrieval.” Cerebral Cortex 15: 1459–68. Röder, B., O. Stock, S. Bien, H. Neville, and F. Rosler. 2002. “Speech processing activates visual cortex in congenitally blind humans.” European Journal of Neuroscience 16: 930–6. Van Essen, D. C. 2004. “Organization of visual areas in macaque and human cerebral cortex.” In The Visual Neurosciences, ed. L. Chalupa and J. S. Werner, 507–21. Cambridge, MA: MIT Press.

To correct this undesirable effect of subjacency, Chomsky hypothesized that long-distance movement proceeds in short steps, passing through successive cycles. In particular, he postulated that movement can stop by at the edge of the clause (S’or COMP; the modern complementizer phrase [CP] area). In other words, instead of moving long distance in one fell swoop, movement first targets the closest clausal edge and from there proceeds from clausal edge to clausal edge, typically crossing only one S(/IP)-node at a time:

BOUNDING

(5) [Handsomei though [S I believe [S’ t’i that [S Dick is ti]]], I’m still going to marry Herman.

For all its modernity and insights into the fundamental workings of language,Noam Chomsky’s early writings (1955, 1957) contain a curious gap: They do not contain any explicit discussion of locality. One does not even find extensive discussion of the fact that movement appears to be potentially unbounded. This gap is all the more curious from our current perspective where locality and long-distance dependencies are arguably the major area of study in theoretical syntax. We owe our modern interest in locality to John R. Ross’s ([1967] 1986) seminal work in which the concept of island was introduced. Ross’s thesis is full of examples of long-distance dependencies like (1a and b). (1) a. Handsome though Dick is, I’m still going to marry Herman. b. Handsome though everyone expects me to try to force Bill to make Mom agree that Dick is, I’m still going to marry Herman.

Ross systematically investigated the fact that seemingly minute manipulations dramatically affected the acceptability of sentences. Witness (2a and b). (2) a. Handsome though I believe that Dick is, I’m still going to marry Herman. b. *Handsome though I believe the claim that Dick is, I’m still going to marry Herman.

Ross’s thesis contains a list of contexts, technically known as islands, which disallow certain types of dependencies. Chomsky (1973) set out to investigate what the various domains identified by Ross as islands have in common. Thus began the modern study of locality and, in many ways, the nature of current linguistic theorizing. Chomsky’s central insight in 1973 is that movement is subject to the subjacency condition, a condition that forbids movement from being too long. Specifically, his notion of subjacency prevented movement from crossing two bounding nodes. For Chomsky, the bounding nodes were the top clausal node (S for sentence; our modern inflectional phrase [IP]) and NP (noun phrase; our modern DP [determiner phrase]). The condition correctly captured the unacceptability of (2b), but wrongly predicted (2a) to be out (see [3] and [4]).

134

(4) [Handsomei though [S I believe that [S Dick is ti]], I’m still going to marry Herman.

Successive cyclicity may at first seem like a patch, an exemption granted to fix a bad problem (without it, the theory would wrongly rule out acceptable constructions). But subsequent research has uncovered a wealth of data, reviewed in Boeckx (2007), that converge and lend credence to the successive cyclic movement hypothesis, making it one of the great success stories of modern generative grammar. It appears to be the case that long-distance, unbounded dependencies are the result of the conjunction of small, strictly bounded steps. Currently, our most principled explanation for the phenomenon of successive cyclicity is that it is the result of some economy condition that requires movement steps to be kept as short as possible (see Chomsky and Lasnik 1993; Takahashi 1994; Boeckx 2003; Bošković 2002). As Željko Bošković (1994) originally observed, some additional condition is needed to prevent this economy condition requiring movement steps to be kept as short as possible from forcing an element that has taken its first movement step to be stuck in creating infinitesimally short steps. Put differently, some condition is needed to prevent chain links from being too short. The idea that movement that is “too short” or superfluous ought to be banned has been appealed to in a variety of works in recent years, under the rubric of anti-locality (see Grohmann 2003 for the most systematic investigation of anti-locality; see also Boeckx 2007 and references therein.) The anti-locality hypothesis is very desirable conceptually. It places a lower bound on movement, as Chomsky’s subjacency condition places an upper bound on movement. Since it blocks vacuous movement, it is also an economy condition (don’t do anything that is not necessary), on a par with the underlying force behind subjacency. We thus arrive at a beautifully symmetric situation, of the kind promoted by the recently formulated minimalist program for linguistic theory (Chomsky 1995 and Boeckx 2006, among others): Long-distance dependencies, which are pervasive in natural languages, are not taken in one fell swoop. The patterns observed in the data result from the conjunction of two economy conditions: movement must be kept as short as possible, but not too short. – Cedric Boeckx

Brain and Language WORKS CITED AND SUGGESTIONS FOR FURTHER READING Boeckx, Cedric. 2003. Islands and Chains. Amsterdam: John Benjamins. ———. 2006. Linguistic Minimalism: Origins, Methods, Concepts, and Aims. Oxford: Oxford University Press. ———. 2007. Understanding Minimalist Syntax: Lessons from Locality in Long-Distance Dependencies. Oxford: Blackwell. Bošković, Željko. 1994. “D-structure, θ-criterion, and movement into θ-positions.” Linguistic Analysis 24: 247–86. ———. 2002. “A-movement and the EPP.” Syntax 5: 167–218. Chomsky, Noam. 1955. “The logical structure of linguistic theory.” Manuscript, Harvard/MIT. Published in part in 1975 by Plenum, New York. ———. 1957. Syntactic Structures. The Hague: Mouton. ———. 1973. “Conditions on transformations.” In A Festschrift for Morris Halle, ed. S. Anderson and P. Kiparsky, 232–86. New York: Holt, Rinehart, and Winston. ———. 1995. The Minimalist Program. Cambridge, MA: MIT Press. Chomsky, Noam, and Howard Lasnik. 1993. “Principles and parameters theory.” In Syntax: An International Handbook of Contemporary Research, ed. J. Jacobs, A. von Stechow, W. Sternefeld, and T. Vennemann, 506–69. Berlin: de Gruyter. Grohmann, Kleanthes K. 2003. Prolific Domains. Amsterdam: John Benjamins. Ross, John R. [1967] 1986. “Constraints on variables in syntax.” Ph.D. diss., MIT. Published as Infinite Syntax! Norwood, NJ: Ablex. Takahashi, Daiko. 1994. “Minimality of movement.” Ph.D. diss., University of Connecticut.

BRAIN AND LANGUAGE The brains of humans have developed to control our articulators and our sensory systems in ways that permit human language. Our knowledge of how the brain subserves language was cursory in the early millennia of recorded history; in the late 1800s, developments in neurology in Europe gave us the tools to form a more precise understanding of how brains support language. Subsequent advances in neurolinguistic knowledge arose when clinician-scientists abstracted patterns and other researchers developed technical tools (of neuropathology, linguistics, psychology, psycholinguistics, and brain localization via imaging) that permitted both groups to understand the complexity of brain-behavior relationships at ever-finer-grained levels.

Clinical Observation: The Behavioral Sequelae of Brain Damage and the Brain Structures Underlying Them As neurology developed in France, psychiatric patients were distinguished during their lives from aphasics (those with language problems but not emotional ones), and post mortem dissection and advances in staining techniques permitted localization of the brain areas that could be linked to the language behaviors recorded prior to the patients’ death. The developing understanding of French neurologist Paul Broca that not only the frontal lobe (1861) but also the dominant, left, hemisphere (1865) was linked to articulated language was extended by German neurologist Carl Wernicke (1874). Wernicke suggested an additional region, farther back in the brain, that was responsible not for articulated speech but, rather, for comprehension of it. In his paper, moreover, Wernicke proposed a model of centers

for language that predicted another type of aphasia that could be, and was, found: conduction aphasia. In conduction aphasia, it was not centers but connections between them that were impaired: Neither broca’s area of the brain nor wernicke’s area was itself damaged, but the link between them was; the production of speech and comprehension of it were not impaired, but, rather, repetition of auditory input became problematic. The model postulated by Wernicke in his paper showed centers for speech, comprehension, and ideas overlaid on the image of a right [sic] hemisphere, and his colleague Ludwig Lichtheim abstracted this localizationist model away from a picture of underlying brain, expanding it to include reference to centers for reading and writing. In England, John Hughlings Jackson took exception to this localizationist/connectionist approach, taking the holist position. He pointed out that even in patients with extensive damage to the dominant hemisphere, some language remained (e.g., a subset of emotional words, often curse words, as had been the case with Broca’s first patient), suggesting, by way of the subtractionist logic these researchers employed, that the nondominant hemisphere also participated in language in the healthy individual. In France, the debate between those who believed in multiple types of aphasia, each associated with brain damage in a different area, and those who believed in a unitary aphasia associated with a single location continued in a series of debates (analyzed in English by Lecours et al. 1992). Neuropathologist Auguste Déjérine and her neurologist husband Jules led proponents for the multiple-connected-centers position, whereas Pierre Marie argued for a unitarist one. He asserted that what we now call Marie’s quadrilateral, a region near the insula that has only recently been seriously implicated in language again, was the seat of all language. In addition to discussions of localization, bilingualism earned a place in explanations of neuropsychological phenomena among students of the neurologist Jean-Martin Charcot in the later nineteenth century, as Sigmund Freud, in his 1891 book On Aphasia, and Albert Pitres, in his 1895 article on bilingual aphasia, respectively championed the first-learned versus the best-known language in predicting patterns of differential recovery from aphasia. Into the early twentieth century, European neurologists continued developing their careful clinical examination of patients, which they then followed with an examination of their brains, via advanced staining techniques, post mortem. In 1906, the German neurologist Alois Alzheimer isolated a type of disease among those theretofore housed in psychiatric institutions when he discovered distinctive cellular changes in specific levels of cortex associated with what we now call Alzheimer’s dementia. His extended descriptions of the communication problems of his patients are models of the careful observation of the semantic and conversational breakdown associated with this disease.

Rehabilitation The next major step forward in neurolinguistics lay in developing rehabilitation techniques for those with language impairment. One important group had been identified by the ophthalmologist James Hinshelwood (1902), who described the case of a child who had particular difficulty learning to read despite normal

135

Brain and Language intelligence and vision. The American neurologist Samuel Orton, who examined an increasing number of such children through referrals, published his 1937 book Reading, Writing and Speech Problems in Children and Selected Papers [sic], classifying their problems. He worked with Anna Gillingham to develop a multisensory, systematic, structured system for training which, like others that have been derived from it, enables children whose brains do not naturally pick up reading – today termed dyslexics or developmental dyslexics (see dyslexia) – to learn to do so. The recognition that dyslexics might have normal (or better than normal) intelligence but have substantial difficulty learning to read confirmed a second point that Wernicke had included in his 1874 article, that language and intelligence were dissociable. It may also be seen as the earliest vision of individual differences in brain organization that went beyond the donation of their brains by the Paris Anthropology Society to determine whose was bigger and phrenology’s assertion that differing sizes of cortical regions, as evidenced by differences in skull configuration, explained personality differences. This focus on rehabilitation resulted in the initiation of the field of speech therapy, today speech-language pathology in North America and logopedics elsewhere. The seminal work of Hildred Schuell, James Jenkins, and E. Jimenez-Pabon (1964) classified the language disorders resulting from injury to the brain in adulthood according to the primary impairment of either comprehension or production. A more holist approach developed alongside this one, that of Bruce Porch, whose system of classification showed a set of language abilities clustering together.

Lateral Dominance In the 1950s and 1960s, American psychology was also developing more rigorous methods of studying behavior, and brain, though not necessarily linking them yet. Broca’s late-nineteenth-century observation that aphasia tended to arise primarily from damage to the left hemisphere of the brain, rather than the right, took on a new life as the techniques of dichotic listening and tachistoscopic presentation evolved to study lateral dominance in nonbrain-damaged individuals. In dichotic listening, series of three or so pairs of words are presented simultaneously, one to each ear, and participants are asked to recall as many words as they can. Because the primary connections between ear and brain are

to the brain hemisphere opposite a given ear (the contralateral one), participants are better able to recall more words from the ear contralateral to their language-dominant hemisphere, that is, the right ear for language stimuli. Tachistoscopic presentation permitted a visual analogue to dichotic listening: When visual information is flashed so that it is visible only to a single visual field, that information is processed first by the brain hemisphere contralateral to the visual field. The eyes cannot turn quickly enough to see the stimulus in the central visual field, which projects to both hemispheres. Thus written language, but not non-language visual information such as pictures, is processed faster and better by the language-dominant hemisphere. From such laterality studies we came to understand the dominant importance of the left hemisphere for processing auditory and written language for most humans, and the link between this lateralized brain dominance and handedness. For a period, such techniques were used as well to determine if bilinguals’ brains employed relatively more right hemisphere in language processing than monolinguals’ did, following up the suggestions of a number of researchers that bilingualism might be more bilaterally organized, or that early less-proficient language abilities might rely relatively more on right hemisphere contributions. They hypothesized this possibility because they thought they saw a disproportionately large number of instances of crossed aphasia, that is, aphasia resulting from right hemisphere damage rather than the more usual left hemisphere damage. Today it appears, instead, that crossed aphasia is no more frequent among bilinguals than among monolinguals. During this same midcentury period, behavioral neurologist Norman Geschwind and neuropathologist Marjorie LeMay undertook post mortem studies of sizable numbers of brains, demonstrating that the apparent symmetry of the hemispheres is misleading. Rather, they demonstrated precisely that the cortical region around the Sylvian fissure (the perisylvian region; see Figure 1) that was understood to be crucial for language differed markedly, with a steeper rise of the Sylvian fissure in most right hemispheres corresponding to more temporal lobe cortex available for language on the left. (In a small percentage of brains, presumably those from left-handers for the most part, the two hemispheres were indeed identical in this regard; in another small set the cortical asymmetries were reversed.)

Figure 1. Brain and language (cortical structures).

136

Brain and Language

Lenticular nucleus (putamen and globus pallidus)

BASAL GANGLIA OF FORK-BRAIN

The Return of Localization Geschwind brought a localizationist approach back to aphasiology in his work at the Boston Veterans Administration Hospital during the mid-1960s. He anonymously translated Wernicke’s 1874 article into English, and himself published a seminal work (1965) on disconnection syndromes, reminding readers of the particular pair of brain-damaged sites required for alexia without agraphia, that is, a difficulty in reading but not in rewriting resulting from brain damage in adults who had previously been literate. With his colleagues at the Aphasia Research Center of the Boston VA Hospital, Edith Kaplan and Harold Goodglass, he developed this approach into the aphasia classification schema behind the Boston Diagnostic Aphasia Exam (Goodglass and Kaplan, 1972), which includes categories for Wernicke’s and Broca’s aphasias, as well as conduction aphasia, anomia, and the transcortical aphasias. (This test and classification system is quite similar to that of their student Andrew Kertesz, the Western Aphasia Battery). In the 1970s, aphasia grand rounds at the Boston VA Hospital were structured as a localizationist quiz: Neurologists, neuropsychologists, and speech-language pathologists who had tested a patient would first present their findings; then Geschwind or Goodglass would test the patient in the front of the room. After the patient had left, those gathered would guess what the angiography and, later, CT scan results would demonstrate; then Naeser would report the lesion location. Perhaps precisely because CT scans permitted inspection of brain damage beneath the cortex, an understanding of the subcortical aphasias was developed from the linking of patients who were neither fluent (as Wernicke’s aphasics are) nor nonfluent (as Broca’s aphasics are in that the speech they produce consists largely of substantives, with functor grammatical words and affixes omitted or substituted for). Early work distinguished the cortical aphasias from the subcortical ones (e.g., Alexander and Naeser 1988), while more recent work has distinguished aphasias in ever more discrete subcortical regions (e.g., thalamus and globus pallidus; see basal ganglia and Figure 2). With further advances in neuroimaging technology such as magnetic resonance imaging (MRI) to study brain regions and, more recently, diffusion tensor imaging (DTI) to study pathways,

Figure 2. Schematic representation of the chief ganglionic categories. Adapted from Gray, Henry. Anatomy of the Human Body. Edited by Warren H. Lewis. Philadelphia: Lea & Febiger, 1918.

more precision has become available to distinguish areas of the brain that are damaged when language is impaired. In addition, as functional tools are developed (e.g., functional MRI, also called fMRI), it is no longer necessary to rely on the traditional neurolinguistic premise (if area X is damaged and language problem Y results, then X must be where Y is localized, or, at least, X is crucial for normal performance of Y). It is interesting to note that these advances in technology point to a number of regions outside the traditional perisylvian language area of cortex that appear linked to language – areas such as prefrontal cortex, the supplementary motor area, and the like. J. Sidtis (2007) points out the logical problem that arises as one would want to reconcile the aphasia data, which suggest a relatively delimited area of the dominant hemisphere that serves language, with the imaging data that indicate areas beyond the perisylvian region in the dominant hemisphere, as well as many subcortical and non-dominant-hemisphere regions (usually, counterparts to those of the dominant hemisphere) that appear to be involved in language processing.

Linguistic Phenomena Driving Neurolinguistic Study Alongside developments in tools for measuring brain regions involved in language during the last quarter of the twentieth century and the beginning of the twenty-first, developments in linguistics both within and beyond Chomsky’s (itself protean) school have permitted refinements of the questions asked in neurolinguistics and, thus, the answers received. Sheila Blumstein opened up the study of phonology within aphasiology, researching, first, the regression hypothesis of her mentor Roman Jakobson and demonstrating, consistent with the literature on speech errors in non-brain-damaged individuals, that errors tended to differ from their targets by fewer, rather than more, distinctive features. She and her students have studied suprasegmental phonology as well as segmental phonology, demonstrating differences in intonation patterns used exclusively for languages from those that are not. Indeed, other speech scientists have turned to tone languages, such as Thai and Chinese, to demonstrate the parallels there: Brain areas of the dominant hemisphere process those features that one’s

137

Brain and Language language treats as linguistic, even if one’s language is a signed one (sign language), where phonemic elements are not, strictly speaking, “phonemic.” Phonology of lexical items is treated, in some classificatory schemata, as a part of semantics. Lexical-word-shape, however, can be divorced from word-meanings, as patients with the syndrome of word-deafness – the ability to recognize that they know a lexical item without knowing what it means – demonstrate. Studies of aphasic errors have demonstrated the psychological reality of such concepts as phoneme, syllable, word stress, and the like. Moreover, priming studies show that lexical items are organizationally linked to others in their phonological (or spelling or both) neighborhoods. morphology has been studied via aphasiology with particular reference to agrammatism, that syndrome associated with Broca’s aphasia in which the language that patients produce is telegraphic, consisting largely of substantive nouns and, to a lesser extent, verbs, and relatively devoid of functor grammatical elements, including inflectional and derivational affixes. A number of theories have been developed to account for this interesting phenomenon, and it is clear that the salience of the omissible elements plays a role, as their production can be induced in a number of ways, suggesting that they are not lost but, rather, costly to produce in Broca’s aphasia. Evidence for salience varying across different languages can be found in the reports on agrammatism in 14 languages in Menn and Obler (1990) and in the work of E. Bates and her colleagues (see Bates and Wulfeck 1989). Compounding, too, has recently gained attention in neurolinguistics, and can be seen to pose problems for agrammatic patients. Agrammatic patients are particularly pertinent in studying syntax as well, since not only does their production suggest that they minimize syntactic load, but their comprehension is arguably impaired syntactically as well. Whether this is because traces are nonexistent in such individuals or because of more general processing deficits associated with brain damage, perhaps half of the patients with agrammatism have difficulty in processing passive sentences, suggesting that the brain areas impaired in Broca’s aphasia are required for full syntactic function. In non-brain-damaged individuals, fMRI studies suggest that substantial regions beyond the traditional perisylvian language areas – regions such as prefrontal cortex, linked to general “control” functions of brain activity – subserve comprehension (e.g., Caplan et al. 2007). Semantics is better studied in patients with dementing diseases such as Alzheimer’s disease, in whom it breaks down, than in patients with aphasia, in whom it appears better spared, at least for nonfluent (Broca’s) and anomic aphasics. Nevertheless, there have been indications that patients with severe, global aphasia have difficulty with aspects of semantic processing. Psycholinguistic studies of lexical priming are also useful in studying semantics in non-brain-damaged individuals. When they have to judge if nurse is a word or not in English, they are faster if they have seen doctor previously than the word horse, suggesting semantic networks in our lexica. Event-related potential (ERP) measures, moreover, demonstrate that we process top-down for semantic consistency. When we are presented with a sentence that includes a word that is semantically anomalous,

138

a characteristic electrical response around 400 msec after that word indexes our surprise. Study of pragmatics in brain-damaged individuals rarely focuses on aphasics, as their pragmatic abilities tend to be remarkably spared. Rather, it is patients with damage to the right hemisphere who evidence problems with pragmatic abilities, such as verbal humor appreciation, inferencing, conversational coherence, and the like. Patients with the dementias show an interesting combination of the sparing of some basic pragmatic abilities (e.g., eye contact during communication, use of formulaic language) and deficits in higher-level pragmatic behaviors, including, among those who are bilingual, the appropriate choice of whether to address their interlocutors in one or the other, or both, of their two languages. Written language, of course, is not studied differently from auditory language by linguists, and rarely even by psycholinguists, but the literature on brain-damaged individuals offers numerous examples of selective impairment of one of these modes of input and/or output. Historically, such studies focused on alexia and agraphia, that is, disturbances of reading and/or writing in previously literate individuals who had difficulties with these skills as the result of adult-onset brain damage. Currently, information about brain–language links for reading and writing comes from the study of dyslexia, which, of course, is not linked to frank brain damage but has been shown to co-occur with unusual distribution of certain cellular types in language-related brain areas (e.g., Galaburda and Kemper, 1979). Psycholinguistic and brain-imaging studies of both groups of individuals have shown differences, as well as similarities, between the processing of written and spoken language. The same can be said for signed languages, as evident from those who are bilingual speakers of a signed and a spoken language.

Conclusion: Language in Humans’ Brains In many branches of science, pendulum swings are evident between a focus on overarching patterns achieved by ignoring details of individual differences and a focus on those individual differences. In neurolinguistics, the latter can show the full range of human brains’ substrates for humans’ language abilities. We assume that all individuals (except those with specific language impairment) learn their first language in pretty much similar fashion, though we are well aware that even in the first year of life, some of us start talking sooner and others later. In adulthood, too, we acknowledge certain individual differences that are linked to language performance: Some of us are detail oriented and others more “big picture” in cognitive style; some of us are good second-language learners postpubertally and others less so, some of us naturally good spellers and others not, and some of us slow readers and others of us fast. Indeed, we can assume that at many levels, from the cellular to brain regional, from the electrophysiological to fiber connectivity, differences subserve our human ability to communicate via language. From the first century of work primarily in Europe to the late-twentieth-century addition of North American contributions, centers worldwide now participate in moving the field of neurolinguistics forward. – Loraine K. Obler

Brain and Language

Broca’s Area

WORKS CITED AND SUGGESTIONS FOR FURTHER READING

BROCA’S AREA

Alexander, M. P., and M. A. Naeser. 1988. “Cortical-subcortical differences in aphasia.” In Language, Communication and the Brain Research Publications: Association for Research in Nervous and Mental Disorders. Vol. 66. Ed F. Plum, 215–28. New York: Raven Press. Bates L. and B. Wulfeck. 1989. “Comparative aphasiology: A crosslinguistic approach to language breakdown.” Aphasiology 3: 111–42. Caplan, D., G. Waters, D. Kennedy, N. Alpert, N. Makris, G. DeDe, J. Michaud, and A. Reddy. 2007. “A study of syntactic processing in aphasia II: Neurological aspects.” Brain and Language 101: 151–77. Freud, S. [1891] 1953. Zur Auffassung der Aphasien. Trans. E. Stengel as On Aphasia. New York: International University Press. Galaburda, A., and T. Kemper. 1979. “Cytoarchitectonic abnormalities in developmental dyslexia: A case study.” Annals of Neurology 6: 94–100. Geschwind, N. 1965. “Disconnexion syndromes in animals and man.” Brain 88: 585–644. Goodglass, H., and E. Kaplan. 1972. The Assessment of Aphasia and Related Disorders. Philadelphia: Lea & Febiger. Goodglass, H., and Wingfield, A. 1997. Anomia: Neuroanatomical and Cognitive Correlates. San Diego, CA: Academic Press. Lecours, A.R., F. Chain, M. Poncet, J.-L. Nespoulous, and Y. Joanette. 1992. “Paris 1908: The hot summer of aphasiology or a season in the life of a chair.” Brain and Language 42: 105–52. Menn, L., and L. K. Obler, eds. 1990. Agrammatic Aphasia: A CrossLanguage Narrative Sourcebook. Vol. 3. Amsterdam: John Benjamins. Obler, L.K., and K. Gjerlow. 1999. Language and the Brain. Cambridge: Cambridge University Press. Orton, S. [1937] 1989. Reading, Writing and Speech Problems in Children and Selected Papers. Repr. Austin, TX: International Dyslexia Association. Pitres, A. 1895. “Etude sur l’aphasie des polyglottes.” Rev. Méd. 15: 873–99. Schuell, H., J. Jenkins, and E. Jimenez-Pabon. 1964. Aphasia in Adults, New York: Harper & Row. Sidtis, J. 2007. “Some problems for representations of brain organization based on activation in functional imaging.” Brain and Language 102: 130–40.

In 1861, Pierre Paul Broca presented findings from two patients who had difficulty speaking but relatively good comprehension (Broca 1861a, 1861b, 1861c). At autopsy, he determined that both of these patients, Leborgne and Lelong, suffered from injury to the inferolateral frontal cortex. He concluded, “The integrity of the third frontal convolution (and perhaps of the second) seems indispensable to the exercise of the faculty of articulate language” (1861a, 406). Four years later, Broca realized that these and subsequent cases all had lesions to the left inferior frontal gyrus, thus making the association between language and the LEFT hemisphere (Broca 1865). This assertion proved to be a landmark discovery that laid the groundwork not only for the study of language but also for modern neuropsychology. The region of left hemisphere cortex described by Broca subsequently came to be known as Broca’s area and the speech disorder, Broca’s aphasia. Today, Broca’s area is generally defined as Brodmann’s cytoarchitectonic areas (BA) 44 and 45, corresponding to the pars opercularis and pars triangularis, respectively. These regions make up the posterior part of the inferior frontal gyrus (see Figure 1). Recent investigations have suggested important differences between BA 44 and 45, both in anatomical asymmetries and in function. However, Broca himself never designated the region so specifically, considering the posterior half of the inferior frontal gyrus to be most crucial for the speech disturbance he described (Dronkers et al. 2007). Although Broca’s area is widely described as a critical speech and language center, its precise role is still debated. Broca originally thought of this region as important for the articulation of speech (Broca 1864). More recently, a large body of research has discussed its potential role in processing syntax. This premise first arose from behavioral studies of patients with Broca’s aphasia in the early part of the twentieth century. It was noted that patients with Broca’s aphasia produced agrammatic speech, often omitting functor words (e.g., a, the) and morphological markers (e.g., -s, -ed). The following is an example of such

Journals worth checking: Behavior and Brain Science, Brain, Brain and Language, Journal of Cognitive Neuroscience, Journal of Neurolinguistics, Nature Reviews Neuroscience, and NeuroReport.

Figure 1. Three-dimensional MRI reconstruction of the lateral left hemisphere of a noraml brain in vivo, showing the pars operculars (Brodmann's area 44, anterior to the precentral sulcus) and the pars triangularis (Brodmann's area 45, between the ascending and horizontal limbs of the sylvian fissure; see perisylvian cortex). Reprinted with permission from Brain (2007), 130, pg. 1433, Oxford University Press.

139

Broca’s Area

Figure 2. Lesion reconstruction of a patient with Broca’s aphasia who does not have a Broca’s area lesion (left) and a patient with a Broca’s area lesion without Broca’s aphasia (right).

Figure 3. Photographs of the brains of Leborgne (A) and Lelong (C), Paul Broca’s first two aphasic patients, with close-ups of the lesion in each brain (B and D). Reprinted from N. F. Dronkers, O. Plaisant, M. T. Iba-Zizen, E. A. Cabanis (2007), “Paul Broca’s historic cases: High resolution MR imaging of the brains of Leborgne and Lelong,” Brain, 130.5: 1436, by permission of Oxford University Press.

telegraphic speech in a patient with Broca’s aphasia, describing a drawing of a picnic scene by a lake: O, yeah. Det’s a boy an’ a girl … an’ … a … car … house … light po’ (pole). Dog an’ a … boat. ‘N det’s a … mm … a …coffee, an’ reading. Det’s a … mm … a … det’s a boy … fishin’.

During the 1970s, a number of studies reported that comprehension of complex syntactic forms was also disrupted in this patient group. A seminal study by A. Caramazza and E. B. Zurif (1976) reported that Broca’s aphasics had particular difficulty understanding semantically reversible versus irreversible sentences (e.g., The cat that the dog is biting is black vs. The apple that the boy is eating is red). This study concluded that Broca’s area mediated syntactic processes critical to both production and comprehension. By extension, Broca’s area came to be associated with syntactic processing. Although many subsequent studies supported this general notion, several others pointed out the need for caution. For example, M. C. Linebarger, M. Schwartz, and E. Saffran (1983) showed that agrammatic patients could make accurate grammaticality judgments, which would seem to challenge the notion of Broca’s area being broadly involved in syntactic processing. Second, many patients with fluent aphasia and lesions outside of Broca’s area have been found to exhibit grammatical deficits that

140

overlap with those of Broca’s aphasics (Caplan, Hildebrandt, and Makris 1996). In addition, individual patients with Broca’s aphasia do not always show the same pattern of grammatical deficit but, rather, vary in the types of errors they make (Caramazza et al. 2001). Cross-linguistic studies also put a damper on the agrammatic theory of Broca’s aphasia. E. Bates and colleagues have shown that in other languages such as German, where grammatical markers are critical for conveying meaning and semantic content, patients with Broca’s aphasia do not omit morphemes as they do in English (Bates, Wulfeck, and MacWhinney 1991). Finally, recent studies have shown that grammatical errors can be induced in normal participants through the use of degraded stimuli or stressors, such as an additional working memory load (Dick et al. 2001). These findings would argue against a grammar center but, rather, suggest that competition for resources could also underlie deficits observed in Broca’s aphasia.

Lesion Studies of Broca’s Area As with the previous claims, it is often assumed that all patients with Broca’s aphasia have lesions in Broca’s area, and thus that the deficits in Broca’s aphasia equate to dysfunction in Broca’s area. However, many studies making claims about Broca’s area and its functions did not actually verify lesion site. In fact, lesion studies have shown that Broca’s aphasia typically results from

Broca’s Area a large left hemisphere lesion that extends beyond Broca’s area to include underlying white matter, adjacent frontal cortex, and insular cortex (Alexander, Naeser, and Palumbo 1990; Mohr et al. 1978). Color Plate 3 shows a lesion overlay map of 36 patients with chronic Broca’s aphasia persisting more than one year. As can be seen, the region of common overlap (shown in dark red) is not Broca’s area but rather medial regions (i.e., more central), namely, insular cortex and key white matter tracts. Indeed, only 50–60 percent of patients with lesions extending into Broca’s area have a persistent Broca’s aphasia. Lesions restricted to Broca’s area tend to cause a transient mutism followed by altered speech output, but not a chronic Broca’s aphasia (Mohr et al. 1978; Penfield and Roberts 1959). These findings would suggest that Broca’s area proper might be more involved in later stages of speech production (e.g., articulation). Even early studies reported contradictory cases, namely, patients with Broca’s area affected but no Broca’s aphasia or patients with Broca’s aphasia but no lesion in Broca’s area (e.g., Marie 1906; Moutier 1908). Figure 2 shows an example of an individual with Broca’s aphasia whose lesion spares Broca’s area (left) and one with word-finding problems but no Broca’s aphasia after a lesion to Broca’s area (right). Although Broca deduced that the critical region for his patients’ articulation disturbance was the inferior frontal gyrus, he realized that his patients’ lesions most likely extended more medially. However, he wanted to maintain the brains for posterity and chose not to dissect them (see Figure 3). Recently, N. F. Dronkers and colleagues (2007) had the opportunity to acquire three-dimensional MRI images of the brains of Broca’s two original patients (Leborgne and Lelong), which are kept in a Paris museum. They found that the lesions in both patients extended quite medially, involving underlying white matter, including the superior longitudinal fasciculus. Moreover, one of the patients’ brains (Leborgne’s) had additional damage to the insula, basal ganglia, and internal and external capsules. With respect to the extent of frontal involvement, Leborgne’s lesion affected the middle third of the inferior frontal gyrus to the greatest extent, with only some atrophy in the posterior third. In Broca’s second patient, Lelong, the lesion spared the pars triangularis, affecting only the posterior portion of the pars opercularis. Thus, even what is commonly referred to as Broca’s area (BA 44, 45) is not exactly the region affected in Broca’s original patients.

Functional Neuroimaging of Broca’s Area More recently, functional neuroimaging techniques, such as functional magnetic resonance imaging (fMRI) and positron emission tomography (PET), have opened up new avenues for the study of brain areas involved in language and cognition. Consistent with the behavioral studies described previously, a number of functional neuroimaging studies with normal participants have suggested a link between Broca’s area and syntactic processing (e.g., Caplan et al. 2000; Friederici, Meyer, and von Cramon 2000). However, Broca’s area has also been linked to a number of other nonsyntactic cognitive processes in the left hemisphere, such as verbal working memory, semantics, frequency discrimination, imitation/mirror neurons (see mirror systems, imitation, and language ), tone perception (in tonal languages), and phonological processing. It is

possible that subregions of what is now interpreted as Broca’s area (i.e., BA 44, 45) may be functionally distinct, which could explain the heterogeneity of functions associated with this area. An alternative explanation is that functional activations of Broca’s area may be due to task demands involving articulation and/or subvocal rehearsal. These ideas remain to be explored. A number of theories have arisen to try to reconcile the early lesion studies of agrammatism and newer functional neuroimaging findings. For example, it has been suggested that Broca’s area is crucial not for syntactic processing but for aspects of on-line storage (i.e., verbal working memory) that in turn may underlie the ability to process complex grammatical forms (Stowe, Haverkort, and Zwarts 2005). C. J. Fiebach and colleagues (2005) showed that BA 44 was active when participants processed sentences with a large working-memory load, and many neuroimaging studies have suggested that Broca’s area is involved in verbal working memory, in particular (e.g., Awh, Smith, and Jonides 1995). Such findings would suggest that Broca’s area may be crucial for the understanding of complex syntactic forms due to a basic role in subvocal rehearsal (but see Caplan et al. 2000). In sum, though the term Broca’s area has persisted for more than a hundred years, the precise anatomical demarcation of this brain region, along with its exact role in speech and language processing, are still being debated. At a minimum, Broca’s area plays a role in end-stage speech production but is unquestioningly a significant part of a larger network that supports speech and language functions in the left hemisphere. Further work is needed to determine whether it plays a direct or indirect role in a number of other cognitive processes that have been suggested and whether these relate to neighboring regions within the inferior frontal gyrus or distinct functional subregions within the territory known as Broca’s area. – Nina F. Dronkers and Juliana V. Baldo WORKS CITED AND SUGGESTIONS FOR FURTHER READING Alexander, M., M. Naeser, and C. Palumbo. 1990. “Broca’s area aphasias: Aphasia after lesions including the frontal operculum.” Neurology 40: 353–62. Awh, E., E. Smith, and J. Jonides. 1995. “Human rehearsal processes and the frontal lobes: PET evidence.” Annals of the New York Academy of Sciences 769: 97–117. Bates, E., B. Wulfeck, and B. MacWhinney. 1991. “Crosslinguistic research in aphasia: An overview.” Brain and Language 41: 123–48. Broca, P. 1861a. “Nouvelle observation d’aphémie produite par une lésion de la troisième circonvolution frontale.” Bulletins de la Société d’anatomie (Paris), 2e serie, 6: 398–407. Broca, P. 1861b. “Perte de la parole: Ramollissement chronique et destruction partielle du lobe anterieur gauche du cerveau.” Bulletins de la Société d’anthropologie, 1re serie, 2: 235–8. ———. 1861c. “Remarques sur le siège de la faculté du langage articulé, suivies d’une observation d’aphémie (perte de la parole).” Bulletins de la Société d’anatomie (Paris), 2e serie, 6: 330–57. ———. 1864. “Sur les mots aphémie, aphasie et aphrasie; Lettre à M. le Professeur Trousseau.” Gazette des hopitaux 23. ———. 1865. “Sur le siège de la faculté du langage articulé.” Bulletin de la Société d’Anthropologie 6: 337–93. Caplan, D., N. Alpert, G. Waters, and A. Olivieri. 2000. “Activation of Broca’s area by syntactic processing under conditions of concurrent articulation.” Human Brain Mapping 9: 65–71.

141

Cartesian Linguistics Caplan, D., N. Hildebrandt, and N. Makris. 1996. “Location of lesions in stroke patients with deficits in syntactic processing in sentence comprehension.” Brain 119: 933–49. Caramazza, A., E. Capitani, A. Rey, and R. S. Berndt. 2001. “Agrammatic Broca’s aphasia is not associated with a single pattern of comprehension performance.” Brain and Language 76: 158–84. Caramazza, A., and E. B. Zurif. 1976. “Dissociation of algorithmic and heuristic processes in language comprehension: Evidence from aphasia.” Brain and Language 3: 572–82. Dick, F., E. Bates, B. Wulfeck, M. Gernsbacher, J. A. Utman, and N. Dronkers. 2001. “Language deficits, localization, and grammar: Evidence for a distributive model of language breakdown in aphasics and normals.” Psychological Review 108: 759–88. Dronkers, N. F., O. Plaisant, M. T. Iba-Zizen, and E. A. Cabanis. 2007. “Paul Broca’s historic cases: High resolution MR imaging of the brains of Leborgne and Lelong.” Brain 130.5: 1432–41. Fiebach, C. J., M. Schlesewsky, G. Lohmann, D. Y. von Cramon, and A. D. Friederici. 2005. “Revisiting the role of Broca’s area in sentence processing: Syntactic integration versus syntactic working memory.” Human Brain Mapping 24: 79–91. Friederici, A. D., M. Meyer, and D. Y. von Cramon. 2000. “Auditory language comprehension: An event-related fMRI study on the processing of syntactic and lexical information.” Brain and Language 74: 289–300. Linebarger, M. C., M. Schwartz, and E. Saffran. 1983. “Sensitivity to grammatical structure in so-called agrammatic aphasics.” Cognition 13: 361–93. Marie, P. 1906. “Revision de la question de l’aphasia: La troiseme circonvolution frontale gauche ne joue aucun role special dans la function du langage.” Semaine Medicale 26: 241–7. Mohr, J., Pessin, S. Finkelstein, H. H. Funkenstein, G. W. Duncan, and K. R. Davis. 1978. “Broca aphasia: Pathologic and clinical.” Neurology 28: 311–24. Moutier, F. 1908. L’aphasie de Broca. Paris: Steinhell. Penfield, W., and L. Roberts. 1959. Speech and Brain Mechanisms. Princeton, NJ: Princeton University Press. Stowe, L. A., M. Haverkort, and F. Zwarts. 2005. “Rethinking the neurological basis of language.” Lingua 115: 997–1042.

C CARTESIAN LINGUISTICS This term began as the title of a 1966 monograph by Noam Chomsky. It has become the name of a research strategy for the scientific study of language and mind that Chomsky in other works calls “rationalist” or “biolinguistic,” which he contrasts to an empiricist strategy (see biolinguistics). Cartesian Linguistics illuminates these strategies by focusing on contrasting assumptions concerning mind and language and their study that are found in the writings of a selection of philosophers and linguists from the late sixteenth century through 1966. The rationalists include Descartes, the Port-Royal Grammarians, Humboldt, Cudworth, and – clearly – Chomsky himself in 1966 and now. The empiricists include Harris, Herder, the “modern linguists” (L. Bloomfield, M. Joos, etc.), and – again clearly – behaviorists, connectionists, and others attracted to the idea that children “learn” languages rather than “growing” them.

142

Case Rationalists adopt a nativist (see innateness and innatism) and internalist approach to the study of language. Support for nativism is found in poverty-of-the-stimulus observations. To take these observations seriously, rationalists believe, someone constructing a theory of language is advised to assume that much of linguistic structure and content is somehow latent in the infant’s mind: Experience serves to “trigger” or “occasion” structure and content, not form and constitute them. Descartes himself was a rationalist and appealed to poverty facts to support his views of “innate” and “adventitious” (but not “made up”) concepts/ideas. Until taken up by his Port-Royal followers, however, there was little attention to the innateness of the structure of language itself. Descartes’s greater contribution to the strategy with his name lies in less discussed but equally important observations concerning the “creative aspect of language use” (see creativity in language use). Encapsulated in Descartes’s Discourse V, these note that speakers can on occasion produce any of an unbounded set of expressions (unboundedness), without regard to external and internal stimulus conditions (“stimulus freedom”), sentences that nevertheless are appropriate and coherent with respect to discourse context (“appropriateness and coherence”). Taking these seriously suggests a scientific research strategy that focuses not on linguistic action/behavior (“language use”) itself, for that is in the domain of free human action, but on an internal system, the “language faculty.” A science of linguistic action would not only have to take into account the speaker’s understanding of reasons for speaking and the “job” that an utterance is understood to perform but would also have to say what a person will utter on a specific occasion. No extant science can do that, and likely none ever will. Given stimulus freedom, independently specifiable conditions for utterance are unavailable, and there is no upper bound on sentences appropriate to a speaker’s understanding of “discourse circumstance.” The scientist of language should focus on language as an internal system, on competence, not on what people do by using the “tools” their systems offer. Science can say what a language can yield; a theory of competence does that. But it likely cannot say what will be said, when it will be said, or whether it is appropriate and, if so, why. Generally speaking, the rationalist strategy treats languages as natural (native) systems in the head, not artifacts. The empiricist strategy (not empirical science) treats languages as artifacts in the head, as socially constituted sets of practices or behaviors, mastery of which (learning) requires training and “negative evidence.” Cartesian Linguistics emphasizes that there are few, if any, linguistic practices and that children “grow” languages. – James McGilvray WORKS CITED AND SUGGESTIONS FOR FURTHER READING Chomsky, Noam. 1966. Cartesian Linguistics. New York: Harper and Row. ———. 2003. Cartesian Linguistics. 3rd ed. Ed. and introd. J. McGilvray. Cambridge: Cambridge University Press.

CASE This term has been traditionally employed to designate the type of morphological ending that indicates the syntactic function of the

Categorial Grammar noun phrase that bears it. In Latin, for instance, the word meaning “girl” may surface as puella (nominative case), puellam (accusative case), or puellae (genitive case), depending on whether it is the subject of the sentence, is the object of a transitive verb, or stands in a possessor relation with respect to another noun. Languages vary with respect to the morphological case distinctions they make. Languages such as Latin have six case distinctions (nominative, genitive, dative, accusative, vocative, and ablative) whereas languages such as Chinese have none. One may also find languages like English in which only a subset of nominal elements display case distinctions as in he (nominative), him (accusative), and his (genitive). On the basis of a suggestion by Jean-Roger Vergnaud (Rouveret and Vergnaud 1980), Noam Chomsky (1981) developed a theory of abstract case (annotated as Case). According to this theory, it is a property of all languages that noun phrases can only be licensed in a sentence if associated with Case. Whether or not the abstract Cases get morphologically realized as specific markings on (some) noun phrases is a language-specific property. Research in the last two decades has been devoted to identifying i) which elements are Case-licensers, ii) which structural configurations allow Case licensing, and iii) what the precise nature of such licensing is. The contrast between John/he/*him/*his sang and *John/*he/*him/*his to sing, for example, has led to the conclusion that in English, the past tense may license nominative Case, but the infinitival to is not a Case-licenser. In turn, the contrast between John/he/*him/*his was greeted and was greeted *John/*he/*him/*his indicates that nominative is licensed by the past tense if the relevant noun phrase occupies the subject, but not the object, position. When a given Case only encodes syntactic information, it is referred to as structural Case. Nominative and accusative cases in English are prototypical examples of structural Case. In John greeted her and she was greeted by John, for example, the pronoun bearing the thematic role of patient has accusative Case in the first sentence, but nominative in the second. On the other hand, when a given Case also encodes thematic information, it is referred to as inherent Case (Chomsky 1986; Belletti 1988). The preposition of in English, for example, has been analyzed as a marker of inherent Case, for it only licenses a noun phrase that is the complement of the preceding noun: It may license that country in the invasion of that country but not in *the belief of that country to be progressive because that country is the complement of invasion but not of belief. – Jairo Nunes WORKS CITED AND SUGGESTIONS FOR FURTHER READING Belletti, A. 1988. “The Case of unaccusatives.” Linguistic Inquiry 19: 1–34. Chomsky, N. 1981. Lectures on Government and Binding. Dordrecht: Foris. ———. 1986. Knowledge of Language: Its Nature, Origin and Use. New York: Praeger. Rouveret, A., and J. R. Vergnaud. 1980. “Specifying reference to the subject: French causatives and conditions on representations.” Linguistic Inquiry 11: 97–202.

project word order and build logical forms via general rules of grammar. First proposed by Kazimierz Ajdukiewicz (1935), it is arguably the oldest lexicalized formalism. CG is not itself a theory of natural language grammar, but its use has many linguistically relevant ramifications: lexicalization, flexible constituency, semantic transparency, control over generative capacity, and computational tractability.

Basic Principles of Categorial Grammar In Ajdukiewicz’s system, words are assigned categories, which are atomic types like np (noun phrase) and s (sentence) or complex types like s\np, which indicate the arguments that a function (such as the category for an intransitive verb) subcategorizes for. Words and phrases combine with others by cancellation of subcategorized arguments through a general operation akin to multiplication. That is, just as 4 × (3 ÷ 4) = 3, there is a grammatical correlate: np . (s\np) = s. Given that the word Olivia has the type np and that sleeps has the type s\np, their concatenation Olivia sleeps has the type s via cancellation of the np argument of sleeps. CG assumes compositionality: The global properties associated with a linguistic expression are determined entirely by the properties of its component parts. Linguistic expressions are multidimensional structured signs containing phonological/ orthographic (ϕ), syntactic (σ), and semantic (μ) specifications for the expression. CG is distinguished in that it uses categories as syntactic types, such as those mentioned. Complex categories encode both subcategorization and linear order constraints, using the leftward slash \ and the rightward slash /. Some (simplified) lexical entries are given in the following format ϕ := σ : μ.

The transitive verb category (s\np)/np seeks an object noun phrase to its right and then a subject noun phrase to its left; after these arguments are consumed, the result is a sentence. Semantics are given as λ-calculus expressions that reduce to predicate and argument structures after syntactic combination. The λ-calculus is a standard system used in CG (and many other frameworks) for representing semantics derived by syntax, where variables in the λ-terms are bound to corresponding syntactic arguments. For the category for likes, the x variable is bound to the object (the /np argument), and the y is bound to the subject (the \np argument) – an example of how this works in a derivation follows. These semantics (and the categories themselves) are obviously simplified and are intended here only to demonstrate how the correct dependencies are established. Leftward and rightward slashes project directionality via two order-sensitive, universal rules of function application:

CATEGORIAL GRAMMAR Categorial grammar (CG) is a family of formalisms that model

syntax and semantics by assigning rich lexical categories that 143

Categorial Grammar

Combinatory Categorial Grammar and Categorial Type Logics

Figure 1.

In words, the forward rule states that “a category of type X/Y can combine with one of type Y found to its right to produce a category of type X.” The symbol > is an abbreviation used in derivations (as in the next example). The function is applied similarly for the backward rule. When these rules are used to combine two syntactic categories, their semantic components are also combined via function application in the λ-calculus (indicated in the rules as f a). For example, the result of applying the function λx.λy.like(y,x) to the argument chocolate is λy.like(y,chocolate). This lock-step syntactic-semantic combination underlies the transparent syntax-semantics interface offered by CG. With these rules and lexicon, the derivation can be given in Figure 1. The subcategorized arguments of the verb are consumed one after the other, and the semantic reflexes of the syntactic rules are carried out in parallel. This derivation is isomorphic to a standard phrase structure grammar (PSG) analysis of such sentences. Derivational steps can be viewed as instantiations of rules of a PSG written in the accepting, rather than producing, direction (e.g., np s\np ⇒ s instead of s → np vp).

Type Dependency CG with just function application is weakly equivalent to standard context-free phrase structure grammar. Nonetheless, the approach is radically different: Syntactic well-formedness in CG is type dependent rather than structure dependent, and derivation is an artifact rather than a representational level. Also, categories labeling the nodes of categorial derivations are much more informative than the atomic symbols of constituent structure produced by PSGs. Subcategorization is directly encoded in categories like s\np, (s\np)/np, and ((s\np)/np)/ np, rather than with stipulated nonterminal symbols such as V-intrans, V-trans, and V-ditrans that have no transparent connection to their semantic types. Furthermore, there is a systematic correspondence between notions such as intransitive and transitive: After the transitive category (s\np)/np consumes its object argument, the resulting category s\np is that of an intransitive verb. More importantly, type dependency shifts the perspective on grammar – shared with tree-adjoining grammar and headdriven phrase structure grammar – away from a topdown one in which phrase structure rules dictate constituent structure into a bottom-up one in which lexical items project structure through the non-language-specific rules (i.e., CG’s universal grammar). Recent developments in the transformational grammar tradition, such as minimalism, have also incorporated such a lexically driven perspective.

144

CG moves further from PSGs and other frameworks by incorporating other rules that provide new kinds of inference over categories. These rules are responsible for the type-driven flexible constituency for which CG is well known. The two main branches of CG can be broadly construed as rule based, exemplified by combinatory categorial grammar (CCG) (Steedman 2000), and deductive, exemplified by categorial type logics (CTL) (Moortgat 1997; Oehrle in press). We discuss both of these briefly. CCG adds a small set of syntactic rules that are linear counterparts of the combinators from combinatory logic. Two combinators, composition (B) and type-raising (T), lead to the following rules, among others:

The rules are guaranteed to be semantically consistent. Composition of categories leads to composition of the semantic functions in the λ-calculus. Type-raising turns an argument into a function over functions that seek that argument. See the following for an example of both of these rules in a derivation. CTL is a family of resource-sensitive linear logics, complete with hypothetical reasoning. This approach began with Joachim Lambek (1958) recasting basic CG as a logical calculus in which slashes are directionally sensitive implications; for example, the English transitive category (s\np)/np is (np → s) ← np. As such, categories are provided sound and complete model theoretic interpretations. The application rules given earlier are then just leftward and rightward variants of modus ponens. Additional abstract rules may be defined that allow structured sequents of proof terms to be reconfigured to allow associativity and permutativity. One result is that many rules can be derived as theorems of a given CTL system. For example, any expression with the category np can be shown to also have the category s/(s\np), among others. This is an instance of type-raising; similarly, CCG’s composition rules (as well as others) can be show to follow from CTL systems that allow associativity. With CTL, such rules follow from the logic, whereas rule-based systems like CCG tend to incorporate a subset of such abstract rules explicitly based on empirical evidence. As an example of how CCG’s rules engender flexible constituency, the sentence Olivia likes chocolate in Figure 2 has an alternative derivation using composition and type-raising. This derivation involves a nontraditional constituent with category s/np for the string Olivia likes, which then combines with chocolate to produce the same result as the previous derivation. A similar analysis can be given with CTL. Whether through CCG’s rules or through CTL proofs, semantically coherent interpretations for a wide variety of non-traditional constituents can be created. This forms the core of accounts of extraction and coordination, as well as intonation and information structure and incremental processing in CG. For example, subject relative pronouns like who have the category (n\n)/(s\np), which seeks an intransitive verb type to produce a post-nominal modifier, while object relativizers like whom have the category

Categorial Grammar

Figure 2. (n\n)/(s/np), which seeks types which are missing objects – such as Olivia likes and Olivia gave Finn, which both have the type s/ np. Extraction is thus handled without appeal to movement or traces. long-distance dependencies in object extraction are captured because forward composition allows the unsaturated argument to be successively passed up until it is revealed to the relativizer, as in Figure 3. Under the standard assumption that coordination combines constituents of like types, then right-node raising is simply constituent coordination: [Kestrel heard]s/np and [Finn thinks Olivia saw]s/np the plane flying overhead.

The compositional semantic terms, omitted here for brevity, are guaranteed to be consistent with the semantics projected by the lexical entries because the composition and type-raising rules themselves are semantically consistent. These processes conspire in other contexts to create constituents for argument clusters that allow similarly straightforward analyses for sentences like Kestrel gave Finn comics and Olivia books. This phenomenon has been called nonconstituent coordination, reflecting the difficulty in assigning a meaningful phrase structure that groups indirect objects with direct objects. From the CG perspective, it is simply treated as type-driven constituent coordination. One of the key innovations CTL has brought to CG is the incorporation of a multimodal system of logical reasoning (including unary modes) that allows selective access to rules that permit associativity and permutation. It is thus possible for a grammar to allow powerful operations (like permutative ones needed for scrambling) without losing discrimination (e.g., engendering a collapse of word order throughout the grammar), while enjoying a (quite small) universal rule component. Other rules – be they CCG-style rules or CTL’s structural rules – support analyses for phenomena such as nonperipheral extraction, heavy-NP shift, parasitic gaps, scrambling, ellipsis, and others. The commitment to semantic transparency and compositionality remains strong throughout; for example, Pauline Jacobson (2008) tackles antecedent-contained deletion in a directly compositional manner with CG and variable-free semantics. See Wood (1993) for a balanced overview of many CG approaches and analyses. Steedman and Baldridge (in press) gives a more recent introduction to and overview of work in CCG. Vermaat (2005) provides a clear and concise introduction to CTL (including pointers to connections between CTL and minimalism), as well as an extensive cross-linguistic account of wh-questions using a very small set of universal structural rules.

Figure 3.

Current Applications and Developments While CG is well known but not widely practiced in mainstream linguistics, it has considerable uptake in both mathematical logic and computational linguistics. Computational implementations of CCG are used for parsing and generation for dialog systems. Current probabilistic CCG parsers, trained in the CCGbank corpus of CCG derivations for newspaper texts, are among the fastest and most accurate available for identifying deep syntactic dependencies. A useful aspect of CG for such implementations that also has implications for its relevance for psycholinguistics is that the competence grammar is used directly in performance. Despite a long divergence between the rule-based and deductive approaches to CG, recent work has brought them into greater alignment. CCG has adopted CTL’s multimodal perspective. This connection allows efficient rule-based parsing systems to be generated from CTL grammars – the lexicon remains the same regardless of the approach. CCG itself can be viewed as the caching out of an underlying definition given in CTL; as such, CTL can be seen as providing metatheories for ruled-based CGs like CCG. Work in CTL explores fine-grained control over grammatical processes within the space of sound and complete logics; work in CCG focuses on cross-linguistic, wide-coverage parsing and computational grammar acquisition. Researchers in both of these traditions continue to expand the range of languages and syntactic phenomena receiving categorial treatments. – Jason Baldridge WORKS CITED AND SUGGESTIONS FOR FURTHER READING Ajdukiewicz, Kazimierz. 1935. “Die syntaktische Konnexität.” In Polish Logic 1920–1939, ed. Storrs McCall, 207–31. Oxford: Oxford University Press. Translated from Studia Philosophica 1: 1–27. Jacobson, Pauline. 2008. “Direct compositionality and variable-free semantics: The case of antecedent contained deletion.” In Topics in Ellipsis, ed. Kyle Johnson, 33–68. Oxford: Oxford University Press. Lambek, Joachim. 1958. “The mathematics of sentence structure.” American Mathematical Monthly 65: 154–70. Moortgat, Michael. 1997. “Categorial type logics.” In Handbook of Logic and Linguistics, ed. Johan van Benthem and Alice ter Meulen, 99–177. Amsterdam: Elsevier; Cambridge, MA: MIT Press. Oehrle, Richard. “Multi-modal type-logical grammar.” In NonTransformational Syntax: A Guide to Current Models, ed. Robert Borsley and Kersti Börjars. Malden, MA: Blackwell. In press. Steedman, Mark. 2000. The Syntactic Process. Cambridge, MA: MIT Press.

145

Categorization Steedman, Mark, and Jason Baldridge. “Combinatory categorial grammar.” In Non-Transformational Syntax: A Guide to Current Models, ed. Robert Borsley and Kersti Börjars. Malden, MA: Blackwell. In press. Vermaat, Willemijn. 2005. “The logic of variation. A cross-linguistic account of wh-question formation.” Ph.D. diss., Utrecht University. Wood, Mary McGee. 1993. Categorial Grammar. London: Routledge.

CATEGORIZATION William Labov (1973, 342) stated, “If linguistics can be said to be any one thing, it is the study of categories: that is, the study of how language translates meaning into sound through the categorization of reality into units and sets of units.” Labov is here addressing the relation between linguistic expressions and the things and situations to which the expressions are used to refer. The circumstances of the world are limitless in their variety; linguistic resources are finite. Since it is not possible to have a unique name for every entity that we encounter or a special expression for every event that happens, we need to categorize the world in order to speak about it. We need to regard some entities, and some events, as being “the same” as others. The relation between a word and a referent is not direct but is mediated by the word’s meaning. It is in virtue of its meaning that a word can be used to refer. A word’s meaning can be thought of as a concept, and a concept, in turn, can be thought of as a principle of categorization. To know the word mug (to take one of Labov’s examples) is to have the concept of a mug, which in turn means being able to use the word appropriately, namely, for things that are called mugs. This goes not only for names of concrete things like mugs but also for names of abstract entities and for words of other syntactic categories. To state that X is on Y is to categorize the relation between X and Y as an on-relation rather than an in- or an at-relation. We can make similar claims for other elements in a language, such as markers of tense and aspect. To describe an event in the present perfect as opposed to the past simple or to use progressive as opposed to non-progressive aspect is to categorize the event in a manner consistent with the concept designated by the morpho-syntactic elements. On many counts, therefore, linguists need a theory of categorization. The theory must provide answers to two related questions. On what basis are entities assigned to a category? And why do we categorize the world in just the way that we do? According to what has come to be known as the classical or Aristotelian theory, a category is defined in terms of a set of necessary and sufficient conditions; it follows that things belong in a category because they exhibit each of the defining features. There are many problems associated with this view. First, it often is just not possible to list the defining features. What, for example, are the defining features of mug as opposed to cup? Then there is the question of the features themselves. Each feature will itself define a category, which in turn must be defined in terms of its necessary and sufficient features. Unless we are prepared to postulate a set of primitive features out of which all possible categories are constructed, we are faced with an infinite regress. Finally, the classical theory makes no predications about why we should have the categories that we do. Any conceivable combination of features could constitute a valid category.

146

A major landmark in the development of a nonclassical theory of categorization was the work of psychologist Eleanor Rosch (1978). She argued that categories have a prototype structure, that is, are centered around good examples, and that things belong to the category in virtue of their exhibiting some similarities with the prototype. The members of a category, therefore, do not need to share the same set of features. Moreover, some members can be “better” or more representative examples of the category than others. Rosch addressed not only the internal structure of categories but also the question of what makes a good category. Good categories – the ones that people operate with, and which are likely to be encoded in human languages – are those that deliver maximum information to the user with minimal cognitive effort. We can approach this matter in terms of the interplay of cue validity and category validity. Cue validity means that having observed that an entity exhibits a certain feature, you can assign the entity, with a fair degree of confidence, to a certain category. Category validity means that having learned that an entity belongs to a certain category, you have expectations about the likely properties of the entity. In this way, we can infer quite a lot about the things that we encounter, on the basis of minimal information about them. Categories also need to be studied against broader conceptual and cultural knowledge having to do with human intentions and purposes, presumed causal relations between things and events, and beliefs about how the world is structured. We can imagine all kinds of hypothetical categories – say, a category comprising things that are yellow, weighing under five kilograms, and manufactured in 1980. Such a category is unlikely to be lexicalized in any human language. It displays very low cue and category validity and would, therefore, not be useful to its users. It would also have no role to play in any broader knowledge system. As already recognized by Labov, the issue of categorization applies not only to the categories we use in talking about the world but also to the analysis of language itself. The very terminology of linguistic description is replete with names of categories, such as phoneme, noun, direct object, word, dialect, and so on, and practical linguistic description involves assigning linguistic phenomena to these various categories. Although the classical approach to categorization is still very strong among linguistic theoreticians, it is increasingly recognized that the categories of linguistics may have a prototype structure and are to be understood, in the first instance, in terms of good examples. As Labov remarked, linguistics is indeed the study of categories! – John R. Taylor WORKS CITED AND SUGGESTIONS FOR FURTHER READING Labov, William. 1973. “The boundaries of words and their meanings.” In New Ways of Analyzing Variation in English, ed. C. J. Bailey and R. W. Shuy, 340–72. Washington, DC: Georgetown University Press. Lakoff, George. 1987. Women, Fire, and Dangerous Things: What Categories Reveal About the Mind. Chicago: University of Chicago Press. Murphy, Gregory. 2002. The Big Book of Concepts. Cambridge, MA: MIT Press.

Causative Constructions Rosch, Eleanor. 1978. “Principles of categorization.” In Cognition and Categorization, ed. E. Rosch and B. Lloyd, 27–48. Hillsdale, NJ: Lawrence Erlbaum. Taylor, John R. 2003. Linguistic Categorization. Oxford: Oxford University Press.

CAUSATIVE CONSTRUCTIONS Causative construction (CC) is defined as form-meaning mapping that encodes a causative situation (CS) in which an entity (typically human), or causer, acts upon another entity (typically human), or causee, to induce some action. Cross-linguistically, this mapping is known to operate using periphrastic syntactic construction, valence-increasing morphology, or lexical verb. Three structural types of CCs thus identified are a) the syntactic causative, which employs a periphrastic causative verb like make in English (1); b) the morphological causative, which employs a causative affix like -(s)ase- in Japanese (2); and the lexical causative, wherein a transitive verb with inherently causative meaning is employed (3), e.g. kiseru “to put (clothes) on (someone)” in Japanese. (1)

Mary made him read the book. (Syntactic)

(2)

Mary-ga musuko-ni huku-o ki-sase-ta. (Morphological) nom son-dat clothes-acc wear-caus-past “Mary made her son put on clothes.”

(3)

Mary-ga musuko-ni huku-o kise-ta. (Lexical) nom son-dat clothes-acc put (clothes) on (someone)-past “Mary put clothes on her son.”

Grammatical and Semantic Hierarchies of the Causee Nominal Case Marking CCs can involve the adjustment of case marking of a causee NP, accompanied by the increased valence of a causer NP. As shown in Japanese examples (2)–(3), with the causer/subject NP Mary assigned the nominative case marking -ga, the causee NP “John” is demoted or deranked to lower case marking, in this case the dative –ni, since the accusative –o is already assumed by another NP/direct object hon (“book”). The deranking order of a causee NP reflects a hierarchy of grammatical relations established cross-linguistically (4): (4)

Subject > Direct Object > Indirect Object > Oblique (Whaley 1997, 193)

Functional-typological studies (Shibatani 1976a; Givón 1980; Cole 1983; see functional linguistics, typology) have noted that differential case marking indexes the differing degrees of control that a causee can exercise over his or her action relative to the causer. Consider a Japanese example (5). (5)

Mary-ga John-{o/ni} Tokyo-e ik-ase-ta. nom acc/dat to go-caus-past “Mary {made/let} John go to Tokyo.”

The accusative case marker –o indexes a lesser degree of control retained by the causee than the dative case marker -ni. This semantic difference between accusative (patient-marking) case and dative (experiencer-marking) case is captured by the

semantic hierarchy (6) proposed by Peter Cole (1983). It reflects the greater-to-lesser degree of control retained by a causee NP. (6)

Agent > Experiencer > Patient

Further Semantic/Pragmatic Dimensions of CCs It is not unusual for a language to have more than one type of CC, for example, Japanese morphological and lexical causatives (3) and (2). In lexical causative (3), causative and causativized verbs are completely fused, while in morphological causative (2) they are separated by morpheme boundary. Crucially, as demonstrated by John Haiman (1983), the differential degrees of fusion semantically correlate with the differing degrees of directness involved in causing an event. For instance, while a lexical causative (3) encodes the causer’s nonmediated action of putting clothes on the causee, a morphological causative (2) can express the situation where the causee put on his clothes upon the causer’s request (e.g., verbal command). Languages can employ two causatives in a single sentence, as in a Korean example ( 7), to encode a sequence of CSs. (7)

John-i Tom-eykey Mary-lul cwuk-i-key ha-ess-ta. nom dat acc die- caus-caus-past-decl “John made Tom kill Mary.” (Ishihara, Horie, and Pardeshi 2006, 323)

Double CCs do not always encode a sequence of CSs and can serve some pragmatic function instead. For instance, as observed by J. Okada (2003), the Japanese double causative occurs most frequently in highly conventionalized benefactive expressions indexing a speaker’s expression of humbleness toward his/her own action, as well as politeness toward the addressee, such as -(s)ase-sase-te itadaku (“to have someone allow one to do something”), as in (8). (8)

Otayori yom-as-ase-te itadaki-masu. letter read- caus-caus-conj humbly receive-pol:nonpast “Allow me to read this letter.” (Okada 2003, 29)

In this instance, as contrasted with its single causative counterpart yom-ase-ite itadaku (read-caus-humbly receive), the double causative serves to reinforce the speaker’s expression of humbleness and politeness. CCs have also been productively investigated by more formally oriented linguists (e.g., Kuroda 1993, Miyagawa 1998). – Kaoru Horie WORKS CITED AND SUGGESTIONS FOR FURTHER READING Cole, Peter. 1983. “The grammatical role of the causee in universal grammar.” International Journal of American Linguistics 49: 115–33. Comrie, Bernard. 1976. “The syntax of causative constructions: Crosslanguage similarities and divergences.” In Shibatani 1976b, 261–312. ———. 1985. “Causative verb formation and other verb-deriving morphology.” In Language Typology and Syntactic Description. Vol. 3: Grammatical Categories and Lexicon. Ed. Timothy Shopen, 309–48. Cambridge: Cambridge University Press. Givón, Talmy. 1980. “The binding hierarchy and the typology of complements.” Studies in Language 4: 333–77. Haiman, John. 1983. “Iconic and economic motivation.” Language 59: 789–811.

147

C-Command

Cerebellum

Haspelmath, Martin. 1993. “More on the typology of inchoative/causative verb alternation.” In Causatives and Transitivity, ed. Bernard Comrie and Maria Polinsky, 87–120. Amsterdam: John Benjamins. Ishihara, Tsuneyoshi, Kaoru Horie, and Prashant Pardeshi. 2006. “What does the Korean ‘double causative’ reveal about causation and Korean? A corpus-based contrastive study with Japanese.” In Japanese/Korean Linguistics. Vol. 14. Ed. Vance, Timothy, 321–30. Stanford, CA: CSLI. Kemmer, Susanne, and Ariel Verhagen. 1994. “The grammar of causatives and the conceptual structure of events.” Cognitive Linguistics 5: 115–56. Kuroda, Shige-Yuki. 1993. “Lexical and productive causatives in Japanese: An examination of the theory of paradigmatic structure.” Journal of Japanese Linguistics 15: 1–81. Miyagawa, S. 1998. “(S)ase as an elsewhere causative and the syntactic nature of words.” Journal of Japanese Linguistics 16: 67–110. Okada, J. 2003. “Recent trends in Japanese causatives: The sa-insertion phenomenon.” In Japanese/Korean Linguistics. Vol. 12. Ed. McClure, William. 28–39. Stanford, CA: CSLI. Shibatani, Masayoshi. 1976a. “The grammar of causative constructions: A conspectus.” In Shibatani 1976b, 1–40. Shibatani, Masayoshi, ed. 1976b. Syntax and Semantics. Vol. 6, The Grammar of Causative Constructions. New York: Academic Press. ———. 2002. The Grammar of Causation and Interpersonal Manipulation. Amsterdam: John Benjamins. Shibatani, Masayoski, and Prashant Pardeshi. 2002. “The causative continuum.” In Shibatani 2002, 85–126. Song, Jae Jung. 1996. Causatives and Causation. London: Longman. Whaley, Lindsay. 1997. Introduction to Typology. The Unity and Diversity of Language. New York: Sage Publications.

C-COMMAND An enduring and fundamental hypothesis within syntactic theory is that the establishment of most, if not all, syntactic relations (agreement, binding, case, control structures, movement, etc.) requires c-command. Tanya Reinhart (1979) provides the following definition of c-command (see also Edward Klima’s (1964) “in construction with”): (1)

α c-commands β if and only if a. The first branching node dominating α dominates β, and b. α does not dominate β, and c. α does not equal β.

To illustrate, consider (2). (2)

Sentence Noun Phrase

Verb Phrase

Noun Phrase

Noun

Verb

Noun Phrase

Mary’s

mother

criticizes

herself

Does the noun phrase Mary’s mother c-command herself in (2)? The first branching node dominating Mary’s mother is sentence, which dominates herself. Also, Mary’s mother does not dominate or equal herself. Since c-command obtains, in this sentence Mary’s mother corefers with herself (i.e., herself must mean Mary’s mother). By contrast, since Mary’s fails to c-command

148

herself, Mary’s and herself are unable to enter into such a relation (see anaphora and binding). Although Reinhart’s pioneering definition is formally explicit and strongly supported empirically, questions arise regarding explanatory depth, as with any definition. In this respect, S. Epstein and colleagues (1998) ask: (i) Why should this formal relation, and not any definable other, constrain syntactic relations? (ii) Why is (first) branching relevant? (iii) Why must α not dominate or equal β? Epstein and colleagues argue that (i–iii) receive natural answers under a bottom-up derivational approach with recursive application of the binary operation merge, as independently motivated in minimalism (Chomsky 1995). C-command is then arguably an emergent property of this structure-building process and is expressible in terms of merge, as in (3): (3)

α c-commands all and only the terms of the category β with which α was merged in the course of the derivation.

Under (3), c-command is not defined on assembled trees, but emerges as a consequence of the merger process by which trees are built. It then follows that only the first branching node is relevant for computing what α c-commands, since this branching node is precisely the syntactic object resulting from merging α with another syntactic category. It also follows that α must not dominate β, since dominance entails non-merger. Finally, because a category α cannot merge with itself, α does not c-command itself. – Gerardo Fernández-Salgueiro, and Samuel David Epstein WORKS CITED AND SUGGESTIONS FOR FURTHER READING Aoun, Joseph, and Dominique Sportiche. 1983. “On the formal theory of government.” Linguistic Review 2: 211–35. Brody, Michael. 2000. “Mirror theory: Syntactic representation in perfect syntax.” Linguistic Inquiry 31.1: 29–56. Chomsky, Noam. 1995. The Minimalist Program. Cambridge, MA: MIT Press. Epstein, Samuel David, Erich M. Groat, Ruriko Kawashima, and Hisatsugu Kitahara. 1998. A Derivational Approach to Syntactic Relations. Oxford: Oxford University Press. Kayne, Richard. 1984. Connectedness and Binary Branching. Dordrecht: Foris. Klima, Edward. 1964. “Negation in English.” In The Structure of Language, ed. Jerry Fodor and Jerrold Katz, 246–323. Englewood Cliffs, NJ: Prentice-Hall. Langacker, Ronald. 1969. “On pronominalization and the chain of command.” In Modern Studies in English, ed. D. Reibel and S. Schane, 160–86. Englewood Cliffs, NJ: Prentice-Hall. Lasnik, Howard. 1976. “Remarks on coreference.” Linguistic Analysis 2: 1–22. Reinhart, Tanya. 1979. “Syntactic domains for semantic rules.” In Formal Semantics and Pragmatics for Natural Languages, ed. F. Guenthner and S. Schmidt, 107–30. Dordrecht: D. Reidel Publishing Company.

CEREBELLUM The cerebellum is a brain structure located underneath the posterior part of the cerebral hemispheres. Three anatomical loops

Cerebellum connect the cerebellum to various parts of the nervous system (Ramnani 2006). The cerebro-cerebellar loop has been of particular interest to language researchers because this pathway supports anatomical connections between the cerebellum and the cortex, potentially including language-related cortical regions in the contralateral cerebral hemisphere. Recent advances in neuroimaging techniques, clinical testing, and anatomical methods provide evidence that strongly implicates the cerebellum in a broad range of language-related tasks, including those involving speech production and perception, single-word reading, and higher-level language processing.

Historical Perspectives Historically, the functions of the cerebellum were thought to be limited to motor processes, such as motor control, performance, and skill acquisition. Basic neuroscience research has led to different proposals regarding its role in motor processes such as errordriven learning (Marr 1969) and internal timing (Ivry 1996). In the late 1980’s, H. C. Leiner, A. L. Leiner, and R. S. Dow (1986) proposed that the cerebellum is not exclusively involved in motor functions but that it also contributes to cognitive processes. Specifically, they argued for a putative role in language because the evolutionary development of the cerebellum paralleled a similar evolution of cortical areas associated with linguistic functions (e.g., broca’s area) (Leiner, Leiner, and Dow 1993). Based on the homogeneity of cerebellar cellular organization, a similar role was attributed to the cerebellum across both motor and non-motor domains. Empirical work providing support for the claims proposed by Leiner, Leiner, and Dow began to emerge in the late 1980s and 1990s (Desmond and Fiez 1998). A positron emission tomography (PET) study conducted by S. E. Petersen and J. A. Fiez (1993) showed increases in cerebellar activity during a verb generation task. This neuroimaging finding was consistent with a follow-up case study of a patient with a lateral cerebellum lesion (Fiez et al. 1992). The patient showed particularly poor performance on verb generation despite the fact that other neuropsychological assessments were within the normal range.

Current Perspectives The cerebellum has been implicated in a broad range of languagerelated tasks. The majority of the work, however, can be related to one of three domains: 1) speech production and perception, 2) reading, and 3) higher-level word processing. In order to account for the cerebellum’s function in language, investigators have made reference to the timing and error correction functions that have been attributed to the cerebellum in the motor literature. For a review on how these may be general mechanisms that contribute to both motor and non-motor processes, see Ivry and Spencer (2004) and Doya (2000). SPEECH PRODUCTION AND PERCEPTION. During speech production, the control, coordination, and timing of movements are essential. Not surprisingly, clinical findings demonstrate profound speech and motor deficits associated with lesions to the cerebellum. One common speech disorder resulting from damage to the cerebellum is dysarthria, which is characterized by distorted and slurred speech that is often monotonic and of a slower rate (Duffy 1995). neuroimaging studies provide further

evidence for the involvement of the cerebellum in speech production. In a recent study, participants performed a syllable repetition task in which speech rate was varied. The results showed increases in cerebellar activity that corresponded with increases in speech rate (Riecker et al. 2006). Cerebellar contributions to speech extend to the domain of perception. Lesions to the cerebellum produce deficits in temporal duration discrimination and impair categorical perception for consonants that differ in the onset of voicing (Ackermann et al. 1997). This clinical evidence is consistent with neuroimaging data that show increases in right cerebellar activity during a duration discrimination task for linguistic items (Mathiak et al. 2002). Other neuroimaging results suggest that the cerebellum is also involved in learning new perceptual distinctions, such as the non-native /r/-/l/ phonetic contrast for Japanese speakers (Callan et al. 2003). As in the motor literature, many of these studies provide evidence that the cerebellum may be important for coordination and timing in the production as well as the perception of speech (for a discussion, see Ivry and Spencer 2004). Other research also draws upon knowledge from the motor literature to emphasize a potential role of the cerebellum in error correction. The fluency of normal speech has led many models of speech production to incorporate a mechanism for monitoring and correcting speech errors (Postma 2000). More detailed computational work has mapped certain processes in speech production to specific brain regions and defined the cerebellum as an important component in monitoring (Guenther, Ghosh, and Tourville 2006). Similar ideas have emerged in models of verbal working memory. Specifically, J. E. Desmond and colleagues (1997) suggest that a rehearsal process that relies on inner speech to maintain verbal items in working memory may also implement an error correction process. In their model, inputs from frontal and parietal cortex into superior and inferior regions of the cerebellum are used to calculate and correct discrepancies between phonological and articulatory codes in order to improve memory performance. READING. Data from neuroimaging studies consistently show cerebellar activation during single-word reading tasks (Fiez and Petersen 1998; Turkeltaub et al. 2002). In addition, individuals with developmental reading disorders show some of the same symptoms that are often seen in patients with cerebellar damage, such as poor duration discrimination and impaired gross motor functions (Nicolson, Fawcett, and Dean 1995). These observations led R. Nicolson, A. Fawcett, and P. Dean (2001) to propose a relationship between cerebellar deficits and developmental reading disorders. Consistent with this idea, anatomical findings have reported smaller right anterior lobes of the cerebellum in children diagnosed with developmental dyslexia (Eckert et al. 2003). This work in developmental reading disorders has focused on the importance of cerebellar involvement in coordination and timing. Integrating the neuroanatomical findings with behavioral work on dyslexia will be key for establishing a specific role for the cerebellum in reading. Recent neuropsychological research provides mixed findings on the causal relationship between lesions to the cerebellum in adult skilled readers and reading difficulties. One study found that patients with lesions to the cerebellar vermis had more errors in single-word reading when compared to controls

149

Cerebellum (Moretti et al. 2002). On the other hand, a study of native English speakers with lesions to the lateral cerebellar hemispheres did not find any reading difficulties at the level of single words or text (Ben-Yehudah and Fiez 2008). These seemingly inconsistent findings may be due to differences in the site of the cerebellar lesions in the two patient groups. HIGHER-LEVEL LANGUAGE. There is accumulating evidence that higher-level language processes may also involve the cerebellum, although this level has received less attention. A meta-analysis of the neuroimaging literature conducted by P. Indefrey and W. J. M. Levelt (2004) reveals increased activity in the cerebellum for tasks that require higher-level word processing; such tasks include picture naming and verb generation (Indefrey and Levelt 2004) or internal generation of semantic word associations (Gebhart, Petersen, and Thach 2002). It is important to note that higher-level language processes seem to recruit more lateral areas, often in the contralateral right hemisphere of the cerebellum (Indefrey and Levelt 2004). Neuropsychological studies have observed impairments in higherlevel language processing (Silveri, Leggio, and Molinari 1994; Riva and Giorgi 2000), including poor performance on a grammaticality judgment task relative to controls (Justus 2004).

Summary In summary, these data collectively provide strong support for cerebellar involvement in many aspects of language, including speech processing, reading, and higher-level language processing. They also suggest that there may be different regions of the cerebellum that are involved in different types of language tasks. This observation is consistent with an emerging concept that distinct cerebro-cerebellar loops support cerebellar interactions with cortex, thus potentially enabling the cerebellum to apply one or more of its suggested functions (e.g., error correction) to separate input-output loops (Kelly and Strick 2003). According to this view, language tasks that rely on different cortical regions would engage distinct cerebro-cerebellar loops that recruit specific cerebellar regions. – Sara Guediche, Gal Ben-Yehudah, and Julie A. Fiez WORKS CITED AND SUGGESTIONS FOR FURTHER READING Ackermann, H., S. Graber, I. Hertrich, and I. Daum. 1997. “Categorical speech perception in cerebellar disorders.” Brain and Language 60: 323–31. Ben-Yehudah, G., and J. Fiez. 2005. “Impact of cerebellar lesions on reading and phonological processing.” Annals of the New York Academy of Sciences 1145: 260–74. Callan, D. E., K. Tajima, A. M. Callan, R. Kubo, S. Masaki, and R. AkahaneYamada. 2003. “Learning-induced neural plasticity associated with improved identification performance after training of a difficult second-language phonetic contrast.” NeuroImage 19: 113–24. Desmond, J. E., J. Gabrieli, A. Wagner, B. Ginier, and G. Glover. 1997. “Lobular patterns of cerebellar activation in verbal working memory and finger-tapping tasks as revealed by functional MRI.” Journal of Neuroscience 17.24: 9675–85. Desmond, J. E., and J. A. Fiez. 1998. “Neuroimaging studies of the cerebellum: Language, learning and memory.” Trends in Cognitive Sciences 2.9: 355–358. This article reviews neuroimaging evidence suggesting that the cerebellum is involved in cognitive tasks, including those that involve learning, memory, and language.

150

Doya, K. 2000. “Complementary roles of basal ganglia and cerebellum in learning and motor control.” Current Opinion in Neurobiology 10: 732–9. This paper suggests that the cerebellum is part of a more general supervised learning system that is guided by error signals. Duffy, J. R. 1995. Motor Speech Disorders. St. Louis, MO: Mosby. Eckert, M., C. Leonard, T. Richard, E. Aylward, J. Thomson, and V. Berninger. 2003. “Anatomical correlates of dyslexia: Frontal and cerebellar findings.” Brain 126 (Part 2): 481–94. Fiez, J. A., and S. E. Petersen. 1998. “Neuroimaging studies of word reading.” Proc Nat Acad Sci USA 95: 914–21. Fiez, J. A., S. E. Petersen, M. K. Cheney, and M. E. Raichle. 1992. “Impaired non-motor learning and error detection associated with cerebellar damage.” Brain 115: 155–78. Gebhart, A. L., S. E. Petersen, and W. T. Thach. 2002. “Role of the posterolateral cerebellum in language.” Annals of the New York Academy of Sciences 978: 318–33. Guenther, F. H., S. S. Ghosh, and J. A. Tourville. 2006. “Neural modeling and imaging of the cortical interactions underlying syllable production.” Brain and Language 96: 280–301. Indefrey, P., and W. J. M. Levelt. 2004. “The spatial and temporal signatures of word production components.” Cognition 92: 101–44. Ivry, R. B. 1996. “The representation of temporal information in perception and motor control.” Current Opinion in Neurobiology 6: 851–7. Ivry, R. B., and R. M. Spencer. 2004. “The neural representation of time.” Current Opinion in Neurobiology 14: 225–32. This article discusses timing processes and their potential neural correlates. A review of the evidence from many different methods is provided, with an emphasis on the potential contributions made by the cerebellum and basal ganglia. Justus, T. 2004. “The cerebellum and English grammatical morphonology: Evidence from production, comprehension, and grammaticality judgments.” Journal of Cognitive Neuroscience 16.7: 1115–30. Kelly, R. M., and P. L. Strick. 2003. “Cerebellar loops with motor cortex and prefrontal cortex of a nonhuman primate.” Journal of Neuroscience 23.23: 8432–44. This article shows cerebellar regions that receive input from the same cerebral cortical regions they project to. They hypothesize closed cerebro-cerebellar loops for the basis of the interactions between cerebellum and cortex. Leiner, H. C., A. L. Leiner, and R. S. Dow. 1986. “Does the cerebellum contribute to mental skills?” Behavioral Neuroscience 100.4: 443–54. ———. 1993. “Cognitive and language functions of the human cerebellum.” TINS 16.11: 444–7. Marr, D. 1969. “A theory of cerebellar cortex.” J Physiol 202.2: 437–70. Mathiak, K., I. Hertrich, W. Grodd, and H. Ackermann. 2002. “Cerebellum and speech perception: A functional magnetic resonance imaging study.” Journal of Cognitive Neuroscience 14.6: 902–12. Moretti, R., A. Bava, P. Torre, R. M. Antonello, and G. Cazzato. 2002. “Reading errors in patients with cerebellar vermis lesions.” Journal of Neurology 49: 461–8. Nicolson, R., A. Fawcett, and P. Dean. 1995. “Time estimation deficits in developmental dyslexia: Evidence of cerebellar involvement.” Proc Biol Sci 259.1354: 43–7. ———. 2001. “Developmental dyslexia: The cerebellar deficit hypothesis.” Trends in Neurosciences 24.9: 508–11. Petersen, S. E. ,and J. A. Fiez. 1993. “The processing of single words studied with positron emission tomography.” Annual Reviews of Neuroscience 16: 509–30. Postma, A. 2000. “Detection of errors during speech production: A review of speech monitoring models.” Cognition 77: 97–131. Ramnani, N. 2006. “The primate cortico-cerebellar system: Anatomy and function.” Nature Reviews Neuroscience 7: 511–22. This review provides a brief description of cerebellar anatomy, and stresses integrating anatomical, computational, and experimental knowledge.

Charity, Principle of Riecker, A., J. Kassubek, K. Groschel, W. Grodd, and H. Ackermann. 2006. “The cerebral control of speech tempo: Opposite relationship between speaking rate and BOLD signal changes at striatal and cerebellar structures.” NeuroImage 29: 46–53. Riva, D., and C. Giorgi. 2000. “The cerebellum contributes to higher functions during development: Evidence from a series of children surgically treated for posterior fossa tumours.” Brain 123: 1051–61. Silveri, M., M. Leggio, and M. Molinari. 1994. “The cerebellum contributes to linguistic production: A case of agrammatic speech following a right cerebellar lesion.” Neurology 44.11: 2047–50. Turkeltaub, P., G. Eden, K. Jones, and T. Zeffiro. 2002. “Meta-analysis of the functional neuroanatomy of single-word reading: Method and validation.” NeuroImage 16.3 (Part 1): 765–80.

CHARITY, PRINCIPLE OF A charity principle is a principle governing the interpretation of the speech and thought of others. It says that the correct interpretation of certain kinds of expressions, areas of discourse, or whole languages maximizes truth and rationality across the (relevant) beliefs of its subject. According to Donald Davidson, the main defender of a principle of charity, its validity derives from the essentially rational and veridical nature of belief and thought. Principles of charity are of central importance in discussions of radical interpretation or radical translation. In W. V. O. Quine’s version, charity governs the translation of the logical constants (cf. Quine 1960, 59). According to Donald Davidson, charity governs the radical interpretation of all expressions of a language. In an early formulation, it tells the radical interpreter to “optimize agreement … by assigning truth conditions to alien sentences that make native speakers right when plausibly possible” (Davidson [1973] 1984, 137). To “make native speakers right” is to interpret them as having beliefs that are largely true and coherent with each other. Later, Davidson distinguished explicitly between these two aspects of charity: The Principle of Coherence prompts the interpreter to discover a degree of logical consistency in the thought of the speaker; the Principle of Correspondence prompts the interpreter to take the speaker to be responding to the same features of the world that he (the interpreter) would be responding to under similar circumstances. Both principles can be (and have been) called principles of charity: One principle endows the speaker with a modicum of logical truth, the other endows him with a degree of true belief about the world. Successful interpretation necessarily invests the person interpreted with basic rationality. (Davidson [1991] 2001, 211)

Coherence restricts belief ascription in terms of the logical relations among the beliefs of a speaker. Correspondence restricts the ascription of empirical beliefs to a speaker in terms of their truth. Since this can only be done according to the interpreter’s own view of what is true, following the principle of correspondence amounts to agreement maximization between speaker and interpreter. Here, Davidson more and more emphasized a causal element; in the most basic perceptual cases, the principle of correspondence calls for the ascription of beliefs shared by speaker and interpreter. The objects of these beliefs are determined as the shared, external causes of

these beliefs: “Communication begins where causes converge” (Davidson [1983] 2001, 151). In later years, Davidson liked to use the metaphor of “triangulation” for this three-way interaction among speaker, interpreter, and external object (cf. Davidson [1991] 2001). The principle of charity does not exclude the possibility of error; speakers are to be right “when plausibly possible” (Davidson [1973] 1984, 137). Charity, thus, in certain situations actually prevents the interpreter from ascribing beliefs of his or her own to the speaker, for instance, perceptual beliefs about objects the speaker cannot perceive from his or her position in space, or beliefs it would be irrational for the speaker to hold on the basis of other beliefs. If something false follows rather directly from other beliefs the speaker holds, charity might even call for ascribing outright mistakes. The rationality induced by the principle is of a minimal, subject-internal character. For Davidson, the principle of charity plays a double role: On the one hand, it provides the method for the radical interpreter, but it does so because it, on the other hand, is the principle metaphysically determining meaning (and belief content): “What a fully informed interpreter could learn about what a speaker means is all there is to learn; the same goes for what the speaker believes” (Davidson [1983] 2001, 148). This is a kind of supervenience: According to Davidson, meaning (and content) supervene on (dispositions to) observable behavior in observable circumstances. That is, there cannot be a difference in meaning (or content) without a (potential) difference in behavior. This can be called a weak semantic behaviorism, but according to Davidson, meaning (and content) cannot be reduced to behavior. That meaning is determined by charity leaves room for a certain indeterminacy, according to Davidson, but does not lead to antirealism or skepticism about meaning or thought content. Because of the role that external objects, as shared causes, play in the determination of content for basic perceptual beliefs, he thought of his own position as a kind of externalism (cf. Davidson 2001; see meaning externalism and internalism). The principle of charity has been widely discussed. Not only have questions of its exact formulation and of its truth or validity been raised but also the question of what kind of a truth it is, if any. What is its epistemic status – a priori or a posteriori? And what is its metaphysical status – necessary or contingent? Davidson mostly thought of charity as a principle constitutive of thought and meaning, an a priori truth of conceptual necessity. Many commentators have claimed that radical interpretation is supposed to provide an (a priori) argument for charity: If radical interpretation is possible, charity is valid (see, for example, Lepore and Ludwig 2005, 204 ff). But according to others, the direction of argument can only be the opposite: If charity holds, radical interpretation is possible (Davidson 1994, 122; Glüer 2006, 344). Then, Davidson would be seen as arguing for charity from considerations regarding the nature of thought content, its holism and externalist determination (cf. Davidson [1991] 2001; 1999, 343; 2001). Partly against Davidson, it has been argued that charity can only be an a posteriori necessity (cf. Føllesdal 1982; Glüer 2006) and that it, like other nomological principles, can be justified by the principles of empirical science (cf. Pagin 2006). – Kathrin Glüer

151

Children’s Grammatical Errors WORKS CITED AND SUGGESTIONS FOR FURTHER READING Davidson, Donald. [1973] 1984. “Radical interpretation.” In Inquiries into Truth and Interpretation, 125–39. Oxford: Clarendon Press. ———. [1983] 2001. “A coherence theory of truth and knowledge.” In Subjective, Intersubjective, Objective: 137–53. Oxford: Clarendon Press. ———. [1991] 2001. “Three varieties of knowledge.” In Subjective, Intersubjective, Objective, 205–20. Oxford: Clarendon Press. ———. 1994. “Radical interpretation interpreted.” Philosophical Perspectives 8: 121–8. ———. 1999. “Reply to Andrew Cutrofello.” In The Philosophy of Donald Davidson, ed. L. Hahn, 342–4. Chicago: Open Court. ———. 2001. “Externalisms.” In Interpreting Davidson, ed. P. Kotatko, P. Pagin, and G. Segal, 1–16. Stanford, CA: CSLI. Føllesdal, Dagfinn. 1982. “The status of rationality assumptions in interpretation and action explanation.” Dialectica 36: 301–16. Glüer, Kathrin. 2006. “The status of charity I: Conceptual truth or aposteriori necessity?” International Journal of Philosophical Studies 14: 337–60. Lepore, Ernest, and K. Ludwig. 2005. Donald Davidson: Meaning, Truth, Language, and Reality. Oxford: Clarendon Press. Pagin, Peter. 2006. “The status of charity II: Charity, probability, and simplicity.” International Journal of Philosophical Studies 14: 361–84. Quine, Willard V. O. 1960. Word and Object. Cambridge, MA: MIT Press.

CHILDREN’S GRAMMATICAL ERRORS Language learners make errors. This observation, easily verified, is not confined to children in thrall to their first encounter with language. It applies equally well to adults acquiring a further language (see second language acquisition). And it applies to cases of both typical and atypical language development (see specific language impairment). There is, then, nothing abnormal about speech errors. They are an intrinsic feature of language acquisition and do not mark out special cases of learning but, rather, constitute the norm. In this vein, one might argue that the very notion of language development almost inevitably implies the occurrence of errors. Perfect speech could not readily be ascribed to a language learner, after all. It is not surprising, therefore, to find that linguistic errors have featured prominently in research on language development. As a universal feature of language acquisition, errors provide not only evidence that learning is taking place but also, in some cases, evidence of how that learning occurs. Children’s speech errors range over every level of language: phonological (see phonology, acquisition of); lexical (see lexical acquisition); morphological (see morphology, acquisition of); and grammatical (see syntax, acquisition of). Grammatical errors are of special interest because they are germane, in the field of language acquisition, to the nature-nurture controversy. Barbara C. Scholz and G. K. Pullum (2006, 60) usefully encapsulate the nativist credo: “[M]ost of the acquisition of natural languages by human beings depends on unacquired (or acquired but unlearned) linguistic knowledge or language-specialized cognitive mechanisms.” Child grammatical errors present a problem, therefore. If the bulk of what is acquired is unlearned, why is there a protracted period in a young child’s life (several years) during which language is manifestly imperfect? At the very least, grammatical errors throw into sharp relief the messiness of the data which nativists must

152

grapple with. Howsoever powerful the child’s innate mechanisms might be, they do not equate to an attribute (language) that comes into the world fully formed at birth. Instead, there is a bridge to be crossed from what Noam Chomsky (1980) has called the child’s initial state (the genetic endowment for language, present at birth) to the steady state (the mature knowledge of grammar finally attained). Several explanations are available to deal with this problem (see innateness and innatism). Of note here is the simple point that such explanations are required by nativists and, inevitably, muddy the waters both theoretically and empirically. On the nurture side of the nature-nurture fence, speech errors (grammatical and otherwise) again present a vexing issue that needs to be addressed. In particular, the behaviorist approach to language acquisition has been castigated for an excessive reliance on operant conditioning as a mechanism of language learning. B. F. Skinner (1957) argued that one of the key processes in language development was the shaping of the child’s verbal behavior through reward. On this view, child utterances are rewarded according to their proximity to the adult models provided. But this is problematic. In a celebrated demolition of Skinner’s thesis, Chomsky (1959, 42) pointed out that operant conditioning cannot tell the whole story since “a child will be able to construct and understand utterances which are quite new, and are, at the same time, acceptable sentences in his language.” Thus, operant conditioning cannot account for novelty. Similarly, imitation cannot account for the child’s speech, particularly errors. Although not mentioned by Chomsky (1959), grammatical errors do, in fact, present the most striking demonstration that language acquisition is not largely based on imitation. The reason is that children are exposed to very few grammatical errors in the input they receive from parents. Hence, there are very few faulty models for children to copy. For example, Elissa Newport, H. Gleitman, and L. R. Gleitman (1977) report just one instance of parental ungrammaticality in a corpus of 1,500 utterances directed toward young children. In consequence, one cannot easily blame the parents for child errors. A further critical point is that the child cannot imitate grammar, only the products of grammar (sentences). Perhaps not surprisingly, since the demise of behaviorism, several other theories of language acquisition have been promulgated that do not rely on operant conditioning or imitation as their mainstay. Beyond their relevance for the nature-nurture issue, child errors have been studied because of the insights they furnish about the processes of language acquisition. For example, an error like i thought they were all womans reveals that the child has extracted the regular suffixation rule for forming plurals of nouns in English. That is, the child knows to add -s to singular nouns in order to mark plurality. The child’s error lies in mistaking woman for a regular noun. This kind of error is commonly described as an overregularization, since the plural rule for regular forms (add –S) has been applied beyond its conventional confines to an irregular noun. Thus, errors of this kind illuminate the child’s ability to extract and generalize a morphological rule. We know that the child is indeed “adding -s” to make plurals, even in the case of regular plurals like coconuts. This latter fact has been established even though it is conceivable that the child has simply heard the form coconuts in the input and

Children’s Grammatical Errors stored it whole, entirely unaware that the word can be parsed into coconut and the plural marker -s. Jean Berko Gleason (1958) invented nonsense words (including wug) to denote birdlike creatures (also invented), in one of the first experiments in the field of child language. Children were shown a picture of one of these creatures and heard “This is a wug.” They were then shown a picture with two of these creatures and heard “Now there are two of them. There are two … ” The pronunciation of two is left hanging in the air, inviting the child to complete the sentence. And, indeed, children four years of age will often declare “There are two wugs.” Since the child has never before encountered the word form wugs in the input, we can be sure that this word has been assembled on-line using the new word form wug and prior knowledge of the plural suffix -s. What is almost always overlooked, or possibly just taken for granted, in research on children’s errors is the fact that “error” is an intrinsically relative concept. In the case of language learners (young and old), utterances can be judged against the standard of an “expert,” typically a parent or other native speaker. When my four-year-old son said, “What’s the man who the forest doing?” I registered an error, based on the dictates of how I would have said the same sentence (possibly, “What’s the man who is in the forest doing?”). But the intuitions of a parent are not sufficient proof that a given child sentence is ungrammatical. Parental intuitions, as a form of evidence, are neither objective nor decisive. Nevertheless, linguists continue to rely on intuitions as a primary source of data on grammaticality (Smith 2004). It is argued that the intuitions of an adult native speaker constitute direct evidence for mental grammar, that is, the knowledge of grammar residing in the head of an individual human being. However, the judgment of what is and is not grammatical is embedded in social convention. Whatever rule happens to be mentally represented by a given individual, and whatever intuitions that rule gives rise to, its acceptance as part of the grammar for a given language is judged in comparison with the intuitions of other persons. Thus, the grammaticality of a child utterance will be judged against the intuitions of the parent or, in some cases, a passing child language researcher. The social nature of this process is rooted in the appointment (or self-appointment) of one or more people as arbiters over the grammaticality of any given utterance. Evidently, decisions about when an error really is an error are not entirely straightforward. And even when one has made that judgment (on whatever basis), one is then faced with a further difficult issue that has, hitherto, received scant attention. In short, does a given error arise from an immature or incomplete knowledge of grammar? Or is the underlying knowledge base entirely adultlike, but somehow an error has slipped out, owing to a technical hitch in production? In this vein, Chomsky (1965) distinguished between competence and performance. Competence refers to the speaker-hearer’s tacit knowledge of his or her language. Performance, on the other hand, comprises the use of this knowledge in producing speech. The utterances we produce arise from both our linguistic competence and other intervening performance factors, including the limitations of short-term memory, motor control over the execution of speech plans, and even the effects of anxiety or alcohol. Cognitive factors of this kind can cause errors to creep into our speech output

despite the fact that our linguistic knowledge (competence) may be flawless. Adult speech (in particular, speech directed toward other adults) may be laden with false starts, hesitations, unnecessary repetitions, and slips of the tongue. The default assumption about adult errors is that they are the product of faulty performance. Child speech errors, on the other hand, are more likely ascribed to an immature competence. However, all the factors that apply to adults as causes of performance errors apply equally well to children. At the same time, the task of distinguishing errors of competence from errors of performance is empirically fraught. And tellingly, it is a task that researchers have not even begun to tackle with any serious purpose (though, see Jaeger 2005 for work on children’s slips of the tongue). With regard to adult errors, there is also a scarcity of evidence to support the assumption that they are, unfailingly, the product of performance factors. It may well turn out, on closer inspection, that adults vary in terms of their grammatical competence. As noted, theories of grammar acquisition tend to assume that immature competence lies at the root of grammatical errors. A notable exception is found in the study of children’s past tense errors. The so-called words and rules theory suggests that when children learn an irregular past tense form (e.g., broke), it automatically blocks the application of the regular suffixation process (break + -ed →breaked). In this way, errors are avoided (Marcus et al. 1992). Of course, young children do produce errors from time to time. To explain these errors, Gary F. Marcus and colleagues (1992) suggest that young children’s memory retrieval system is immature and sometimes lets them down. In consequence, the child may occasionally fail in an attempt to retrieve an irregular form like broke. This failure then triggers the default regular process to produce breaked. Hence, the explanation for child errors is based on limitations in performance, not competence. In support of this idea, it is argued that overregularization rates are generally very low, something like 4 percent (Marcus et al. 1992). This rarity lends itself to a performance-based explanation for what prompts the child’s errors. In the event, error rates may be considerably higher than initial estimates might indicate (Maslen et al. 2004). Sampling limitations may mask brief periods of very high error rates, especially for high-frequency verbs. A further problem is that there is no empirical support for the speculation that errors are caused by failures in memory retrieval. Very little is known about retrieval processes in young children, especially in connection with language. Whatever the merits of the words-and-rules account of past tense errors, it does at least raise awareness that children’s grammatical errors may not necessarily stem solely from an immature competence. As noted, the fact that children produce errors in the course of language acquisition is uncontroversial. Where controversy does arise is in the attempt to explain how children expunge errors and move toward a more adultlike system of grammar. The obvious solution to the child’s problem is for parents and others to supply corrections. Corrections for grammatical errors are often referred to as negative evidence, that is, evidence that some structure is not permitted by the target grammar. However, opinions differ sharply as to whether negative evidence is available to children. Roger Brown and C. Hanlon (1970) demonstrated that parents do not overtly disapprove of their children’s

153

Children’s Grammatical Errors grammatical errors. Thus, they do not reliably mark grammatical errors with injunctions like Don’t say that or No, that’s wrong. This finding has exerted an enormous influence in the field of child language, being hailed by Steven Pinker as “one of the most important discoveries in the history of psychology” (1988, 104). Undoubtedly, Pinker overstates the case. But his enthusiasm stems from the perception that a crucial aspect of linguistic knowledge could not have arisen in the child’s mind through the mediation of the environment. That is, if children receive no help or information in the input from parents concerning what is or is not grammatical, then one must conclude that the child’s knowledge in this respect is innate. Observe that the reach of this conclusion is extensive, since it could, conceivably, encompass each and every rule or principle of grammar in a language. Nativist enthusiasm for what is known as the no negative evidence assumption is tempered by numerous empirical studies that challenge this assumption. Beginning with Kathy HirshPasek, R. Treiman, and M. Schneiderman (1984), researchers have noted that the markers of disapproval examined by Brown and Hanlon (1970) do not constitute the only possible form of corrective input. More recent research has focused on the frequent contrasts between erroneous child usage and correct adult models that figure in child–adult discourse (Chouinard and Clark 2003; Saxton, Backley, and Gallaway 2005). The following example is an exchange between my four-year-old son and myself (emphases highlight the contrast in linguistic forms, not pronunciation stress). Child: I thinked about it with my brain. Adult: You thought about it.

To function as a form of corrective input, contrasts of this kind would have to be interpreted by the child as not simply modeling a correct form. The child would also have to regard them as signals that their own previous usage was ungrammatical (for evidence consistent with this view, see Saxton 1997 and Strapp and Federico 2000). Curiously, Brown and Hanlon themselves remarked on the corrective potential of contrastive discourse, observing that “repeats of ill-formed utterances usually contained corrections and so could be instructive” (1970, 43). However, this observation was entirely overlooked for many years, leading to a considerable distortion of the empirical facts. At the same time, though, and as noted previously, the fact that contrastive discourse is abundantly available to children does not entirely resolve the matter. It still remains to be demonstrated decisively that children actually perceive such contrasts as a form of negative evidence and that they exploit that information in shedding errors and arriving at a mature system of grammar. To conclude, children’s grammatical errors demand the attention of language scientists for two reasons. First, and most obvious, errors stand out. They attract our attention like brightly colored flags, flapping above the parapet. And, second, the investigation of errors reveals much about the processes of language acquisition. They provide the paradigm demonstration that language develops. The fact that errors occur at every level of language, both in abundance and for extended periods of time,

154

Chirographic Culture provides a strong stimulus for language scientists to seek explanations for how and why language learners differ from fully competent native speakers. – Matthew Saxton WORKS CITED AND SUGGESTIONS FOR FURTHER READING Berko Gleason, Jean. 1958. “The child’s learning of English morphology.” Word 14: 150–77. Brown, Roger, and C. Hanlon. 1970. “Derivational complexity and order of acquisition in child speech.” In Cognition and the Development of Language, ed. J. Hayes, 11–53. New York: John Wiley. Chomsky, Noam. 1959. “Review of B. F. Skinner’s Verbal Behavior.” Language 35: 26–58. ———. 1965. Aspects of the Theory of Syntax. Cambridge, MA: MIT Press. ———. 1980. Rules and Representations. New York: Columbia University Press. Chouinard, Michelle M., and E. V. Clark. 2003. “Adult reformulations of child errors as negative evidence.” Journal of Child Language 30: 637–69. Hirsh-Pasek, Kathy, R. Treiman, and M. Schneiderman. 1984. “Brown & Hanlon revisited: Mothers’ sensitivity to ungrammatical forms.” Journal of Child Language 11: 81–8. Jaeger, Jeri J. 2005. Kids’ Slips: What Young Children’s Slips of the Tongue Reveal about Language Development. Mahwah, NJ: Lawrence Erlbaum. Marcus, Gary F., S. Pinker, M. Ullman, M. Hollander, T. J. Rosen, and F. Xu. 1992. Overregularization in Language Acquisition. Monographs of the Society for Research in Child Development, serial no. 228. Maslen, Robert J. C., A. L. Theakston, E. M. V. Lieven, and M. Tomasello. 2004. “A dense corpus study of past tense and plural overregularization in English.” Journal of Speech, Language, and Hearing Research 47: 1319–33. Newport, Elissa, H. Gleitman, and L. R. Gleitman. 1977. “‘Mother, I’d rather do it myself’: Some effects and non-effects of maternal speech style.” In Talking to Children: Language Input and Acquisition, ed. C. Snow and C. Ferguson, 109–149. Cambridge: Cambridge University Press. Pinker, Steven. 1988. “Learnability theory and the acquisition of a first language.” In The Development of Language and Language Researchers: Essays in Honor of Roger Brown, ed. F. Kessel, 97–119. Hillsdale, NJ: Lawrence Erlbaum. Saxton, Matthew. 1997. “The contrast theory of negative input.” Journal of Child Language 24: 139–61. Saxton, Matthew, P. Backley, and C. Gallaway. 2005. “Negative input for grammatical errors: Effects after a lag of 12 weeks.” Journal of Child Language 32: 643–72. Scholz, Barbara C., and G. K. Pullum. 2006. “Irrational nativist exuberance.” In Contemporary Debates in Cognitive Science, ed. R. Stainton, 59–80. Oxford: Basil Blackwell. Skinner, Burrhus F. 1957. Verbal Behavior. New York: Appleton-CenturyCrofts. Smith, Neil. 2004. Chomsky: Ideas and Ideals. 2d ed. Cambridge: Cambridge University Press. Strapp, Chehalis M., and A. Federico. 2000. “Imitations and repetitions: What do children say following recasts?” First Language 20.3: 273–90.

CHIROGRAPHIC CULTURE Writing and script are systems of graphic marks that represent words, syllables, or individual sounds (phonemes)

Chirographic Culture of a language. Chirography shares the same definition with the added meaning of writing by hand. The term thus applies to all the writing systems or scripts that followed the first invention of writing in Mesopotamia, circa 3200 b.c., and before Gutenberg’s invention of the printing press about 1437. Chirography is generally viewed as the gateway to complex literate societies while leaving behind the archaic oral cultures. The nature and extent of the divide between oral and chirographic cultures has long been a matter of debate. In the fifth century b.c., in Phaedrus and Letter VII, the Greek philosopher Plato expressed his concerns with the impact of chirography on human cognition. He warned that writing would weaken human memory and threaten scholarship by allowing the ignorant to fake omniscience. As discussed in Khosrow Jahandarie’s volume Spoken and Written Discourse (1999), the present consensus is less critical. Wherever chirography emerged, in Mesopotamia about 3200 b.c., in China about 1250 b.c. and in Mesoamerica circa 650 b.c., it is held as a productive supplement of speech. This is based on the facts that, first, the human voice can be heard only by a small audience but written documents can be sent to any destination, and, second, speech disappears instantaneously, leaving no trace, while texts can be preserved over time. It is, therefore, generally agreed that chirography extends the network of human communication from culture to culture and makes it possible to trace the roots of cultures and their evolution in history. Moreover, by reducing to order and clarity a myriad of details, chirography is credited with revolutionizing record keeping. Registers allow for administering communities and keeping track of entries and expenditures, profits, and losses of businesses. Finally, writing is recognized for creating data banks far beyond the power of human memory, resulting in turn in the accumulation and articulation of an unlimited quantity of complex data regarded as instrumental to the formulation of significant syntheses and the creation of new cognitive skills. In other words, literacy is viewed as enhancing the possibilities for socially distributed cognition, allowing civilization to grow more complex with the administration of organizations of greater dimensions and larger political units, the creation of more extensive economies, the development of complex sciences and technologies, and a more accurate knowledge of the past. The major priority of the twentieth century, however, has been to investigate the impact of chirography on the human mind. With his famous adage “the medium is the message,” Marshall McLuhan emerged as a popular champion of literacy. In his books The Gutenberg Galaxy (1962) and Understanding Media (1964), the Canadian media critic advocated that media were not passive conduits of information but, rather, vortices of power restructuring human perceptions. McLuhan argued that by translating sounds into a visual code, writing exchanged “an eye for an ear” and that “[p]honetic culture endows men with the means of repressing their feelings and emotions when engaged in action” (1964, 84, 88); he therefore claimed that literate humans develop the power of acting with detachment from emotional involvement. He further emphasized the impact of the linear format of writing, pointing out that civilized societies acquire a

Figure 1. Cuneiform tablet featuring a list of goods. Courtesy of the Texas Memorial Museum, The University of Texas at Austin.

lineal conception of time, that they view events linearly, with a beginning, a middle, and an end. And, in his view, the resulting systematic sequential presentation of arguments translated into a more rigorous logic. Also in the 1960s, Eric A. Havelock analyzed how the adoption of the Semitic alphabet in Greece influenced the organization of ideas, abstraction, and consciousness. Havelock, as well as McLuhan, dealt primarily with alphabetic scripts. In contrast, Walter J. Ong, S.J., university professor of humanities at St. Louis University, and Jack Goody, anthropologist at Cambridge University, included in their analyses prealphabetic chirographic systems, such as the Mesopotamian cuneiform script and non-alphabetic oriental writing systems. Among its many important contributions, Ong’s book Orality and Literacy ([1982] 1983) makes the case that the nonliterate relates to the world in a concrete, situational way, downplaying generalization and abstraction. Relying on Aleksandr R. Luria’s psychological field work, Ong argued that the illiterate does not name geometric figures abstractly as “circles” or “squares” but as “moons” or “plates,” and “doors” or “mirrors.” Furthermore, the nonliterate avoids self-analysis, which requires abstracting the self from the surrounding world. Goody, the author of The Domestication of the Savage Mind (1977) and The Interface Between the Written and the Oral (1987), investigated how the series of shifts involved in the development of writing restructured human thought. In particular, he analyzed how the first Mesopotamian texts that consisted exclusively of lists changed thought processes. He suggested that the lists of goods generated by the Mesopotamian administration (Figure 1) or the sign lists compiled by scribes encouraged scrutiny by selecting which items to include and which to

155

Chirographic Culture

Figure 2. Tokens from Uruk, Iraq, ca. 3300 B.C. Courtesy Vorderasiatisches Museum Berlin, Germany.

exclude. Moreover, he argued that the lists segmented reality by breaking down the perceptual world. For example, a list of tree signs abstracted the trees from the forests to which they belong. In other words, according to Goody, lists decontextualize data but also regroup elements, ordering them by type, shape, size, number, and so on. Consequently, lists reorganize the world, transforming it into an ideal form and creating a new reality upon which the literate is forced to reflect at a new level of generality. Among other twentieth-century authors who considered that writing affected the human mind, David R. Olson, professor of applied cognitive science at the Ontario Institute for Studies in Education, emphasized in The World on Paper (1994) the importance of writing for reflecting upon ourselves. On the other hand, Bruno Latour, an anthropologist, is among the scholars who disagree with the proposition that writing created new cognitive skills. In an article in 1990, he proposed that it is the combination of images and writing in maps, charts, graphs, photos, and diagrams that create better tools to allow scientists to foray into new ideas. Others credit schooling, rather than writing, for increasing rationality and abstract thought. These seminal studies of the 1960–80s must now be updated by taking into account the archaeological discovery that the cuneiform script, the earliest chirographic system, was derived from an earlier visual code. As described by Denise

156

Schmandt-Besserat in Before Writing (1992) and How Writing Came About (1996), a system of tokens was used to keep track of goods in the Near East, starting about 7500 b.c. The tokens were modeled in clay in multiple shapes, such as cones, spheres, disks, and cylinders. Each token shape stood for a unit of an agricultural product: A cone was a small measure of grain, a sphere stood for a large measure of grain, and a cylinder for an animal. Four thousand years later, in the urban period circa 3300 b.c., the tokens had evolved into a complex accounting device with a repertory of about 300 shapes, some including additional incised or punched markings, to record the various units of goods manufactured in workshops (Figure 2), such as wool, textiles and garments. The fact that, in the Near East, the first script was preceded by a visual code – a system of arbitrary symbols to represent words – sheds new light on chirography’s contribution to society. In particular, some of the merits formerly attributed to writing have to be credited to the token system. For example, tokens, not chirography, shifted “an eye for an ear.” Like texts, the tokens were permanent and, therefore, could be transported or stored. The clay artifacts symbolized units of real goods and as such handled data in abstraction. Finally, like written lists, tokens could organize information in successive lines in the most concise way, allowing scanning, evaluating, scrutinizing, and analyzing a budget. As a result, the token system stretched human cognition by making it possible for the neolithic oral cultures of the Near East to handle large amounts of complex information. Once appropriate recognition is given to the token system, the revolutionary contributions of chirography become very clear. First, chirography abstracted numbers (Figure 3). It should be well understood that the tokens were used in one-toone correspondence, which means that one, two, or three small measures of grain were shown by one, two, or three cones. But numerals – signs that represent numbers abstractly – appeared about 3100–3000 b.c., after the three-dimensional tokens were replaced by their images – pictographs – traced onto clay tablets. At this point, 10 jars of oil and 10 large units of grain were no longer shown by 10 ovoid tokens and 10 spherical tokens but by one sign standing for ten, followed by a sign for the goods in question. Second, chirography abstracted the sounds of speech. Whereas the tokens merely acted as logograms – or signs standing for a concept – chirography created phonetic syllabic signs to write personal names (see Figure 3). In sum, compared to tokens that stood for concrete merchandise, the chirographic numerals symbolized oneness, twoness, and abstract constructs of the mind, and the chirographic phonetic signs symbolized the immaterial and evanescent sounds of the voice. By creating numerals and phonetic signs, chirography, therefore, raised human cognition to far higher levels of abstraction. In the twentieth century, research on the impact of chirography on cognition was confined to issues of interest to the humanities. In the 21st century, however, the debate was extended to the field of art. In When Writing Met Art (2007),

Chirographic Culture

Figure 3. Pictographic tablet from Uruk, Iraq, ca. 3000 B.C. Courtesy Deutsches Archaeologisches Institut, Berlin, Germany. The tablet features a list of goods. In the upper cases, the units of merchandise are indicated by pictographs or images of tokens traced in clay. Numerals are shown with impressed signs. On the lower case, phonetic signs indicate the name of the recipient or donor.

ancient Near East. The preliterate lines of a repeated motif were apprehended at a glance, but the narrative compositions of chirographic cultures were read analytically, sequentially. It is generally assumed that the Neolithic motifs, such as triangles or ibexes, were symbols – like the dove is a symbol of peace in our culture. Thus, the preliterate Near Eastern art probably evoked ideas – perhaps profound ideas – but only the art compositions of chirographic cultures could tell complex stories. On the basis of these recent findings, the immense legacy of the first handwritten texts can now be assessed with greater clarity. Art demonstrates that chirography created a paradigm that can be successfully implemented in other communication systems. Archaeology shows that compared to its archaic token precursor, chirography meant leaps in abstraction with the creation of numerals and phonetic signs. By inventing symbols to express such numbers as 1, 10, and 60, chirography laid the foundation for the development of mathematics. By representing the sounds of the voice by phonetic signs, chirography set the stage for writing to become a universal system of communication emulating speech. – Denise Schmandt-Besserat

WORKS CITED AND SUGGESTIONS FOR FURTHER READING

Schmandt-Besserat argued that ancient Near Eastern art compositions – the way images are organized – changed with the advent of chirography in 3100–3000 b.c. She showed that preliterate painted or carved compositions consisted of the mere repetitions of one motif as many times as necessary to cover a surface. For instance, the same triangle or ibex was replicated around the body of the vessel. In contrast, by borrowing the strategies of writing, compositions of the chirographic cultures were able to organize multiple figures into a narrative. To illustrate this concept, large and small signs of writing denoted greater or lesser units of goods and similarly, the size of images indicated status. Gods were represented as larger than kings, and the images of kings were shown larger than those of their fellow citizens. Just as signs changed value by being placed to the right or the left, above or below other signs, the heroes’ actions and interactions were indicated by their orientation, order, and direction. For instance, one figure standing in front of another was understood as being more important than one standing behind it. From writing, art also acquired ways of loading images with information by using symbols akin to determinatives – signs denoting a general class. For instance, the horned tiara indicated the divine status of a figure in the same way the star-shaped sign dingir did in Sumerian cuneiform texts. As a result, “reading” art became akin to reading a text. In sum, art, which is, in at least certain respects, a mirror of culture, signals a conceptual change in design compositions that coincides with the advent of chirography in the

Coulmas, F., ed. 1999. The Blackwell Encyclopedia of Writing Systems, Oxford: Oxford University Press. ———. 2003. Writing Systems: An Introduction to Their Linguistic Analysis. Cambridge: Cambridge University Press. Goody, Jack. 1977. The Domestication of the Savage Mind. Cambridge: Cambridge University Press. ———. 1987. The Interface between the Written and the Oral. Cambridge: Cambridge University Press. Havelock, Eric. A. 1963. Preface to Plato. Cambridge: Belknap Press of Harvard University Press. Houston, Stephen D., ed. 2004. The First Writing. Cambridge: Cambridge University Press. Jahandarie, Khosrow. 1999. Spoken and Written Discourse, A Multidisciplinary Perspective. Stamford, CT: Ablex Publishing. Latour, Bruno. 1990. “Drawing things together.” In Scientific Practice, ed.M. Lynch and S.Woolgar, 19–68. Cambridge, MA: MIT Press. McLuhan, Marshall. 1962. The Gutenberg Galaxy: The Making of Typographic Man. Toronto: University of Toronto Press. ———. 1964. Understanding Media: The Extensions of Man. New York: New American Library. Niditch, S. 1996. Oral World and Written Word. Louisville, KY: Westminster John Knox Press. Olson, David R. 1994. The World on Paper. Cambridge: Cambridge University Press. Ong, Walter J., S. J. [1982] 1983. Orality and Literacy. London: Methuen. Plato. 1973. Phaedrus and Letters VII and VIII. Trans. Walter Hamilton. Harmondsworth: Penguin Books. Schmandt-Besserat, Denise. 1992. Before Writing. Austin: University of Texas Press. ———. 1996. How Writing Came About. Austin: University of Texas Press. ———. 2007. When Writing Met Art. Austin: University of Texas Press.

157

Clitics and Cliticization

CLITICS AND CLITICIZATION The unusual properties of “little words” have attracted the attention of generations of linguists. This is especially true of the items known to traditional grammar as enclitics and proclitics and to modern linguistics simply as clitics. Different theoretical frameworks have highlighted different characteristics of what has often been seen as a unitary class of elements, with the result that two quite distinct sorts of unusual behavior have not always been carefully distinguished.

Two Senses of Clitic The etymology of the word clitic (from Greek kli:no “lean”) brings out what seemed most distinctive to an earlier generation of scholars, their tendency to “lean on” or form part of a prosodic unit with a preceding or following word, linked to their typical lack of autonomous accent. Such attachment may give rise to phonological words containing syntactically unrelated material as a result of linear adjacency. In addition to standard cases in Greek, Latin, and Sanksrit, a wellknown example of this is furnished by the Wakashan language Kwakw’ala, where determiner elements at the left edge of nominal expressions (among other clitics) attach phonologically to the rightmost word of a preceding constituent. In a sentence like məx̣̣’id – ida bəgWanəma-x̣a gənanəma-sa kWix̣ayu “hit-Det man-Det child-Det club, The man hit the child with a club,” the words bəgWanəma-x̣a “man-Obj” and gənanəma-sa “child-Inst” end in determiner elements that are syntactically linked not to the phrase of which they are phonologically a part but, rather, to the argument that follows in the sentence. In this case, the anomaly is a phonological grouping that does not appear to reflect syntactic organization, and this can plausibly be attributed to the prosodic weakness of the clitic (here the determiners). In other instances, however, something else must be at work. The pronominal object clitics in a French sentence like Je le lui donne “I give it to him” appear preceding the main verb, a position in which objects cannot otherwise occur and which requires reference to ordering principles outside of the language’s normal syntax. Elements appearing in unusual positions in this way are generally also phonologically weak, and this has led to a tendency to conflate the two sorts of exceptionality, taking prosodic weakness (including lack of stress) as diagnostic of clitics and then proposing distinctive ordering for the items so identified. In fact, however, some prosodically weak elements appear only in positions that are quite normal syntactically (e.g., the reduced forms of is and has in Fred’s sleeping and Fred’s lost his dog), while some elements that are positioned unusually along with other clitics are nonetheless prosodically full words, such as Tagalog tayo/ natin “we (incl.)” and a number of other pronouns and particles in this language. Such facts suggest that we should recognize two distinct dimensions of unusual behavior of clitics, one phonological (associated with prosodic weakness) and the other syntactic (associated with unusual positioning of a restricted sort, elements commonly called special clitics). The two often, but not always,

158

coincide (as with French object le, which is both unaccented and distinctively positioned). An association between them has been noted at least since the classic work of Jakob Wackernagel (1892), who pointed out that phonetically weak elements in the ancient Indo-European languages tended to cluster in second position within the sentence, a position that was quite anomalous from the point of view of the rest of the syntax. Much later literature has treated the connection between phonological weakness (and especially a lack of autonomous accent) and unusual positioning as essential, although the two turn out to be quite separate characteristics, not only logically but also empirically. The essential distinction between them was pointed out by Arnold Zwicky (1977) and further developed in much later literature, including Anderson (2005).

Accounts of Clitic Behavior Once we realize that it follows from the prosodically impoverished nature of the elements concerned, the phonological dimension of clitic behavior finds a natural account in more general theories of prosody and stress, as demonstrated in work such as that of Elizabeth Selkirk (1995). The (morpho-) syntactic dimension is somewhat more controversial, however. Special clitics appear in a limited range of positions. Any given clitic can be associated with some syntactic domain, and within this domain it may appear initially, finally, postinitially (in second position), prefinally, or adjacent to the head of the domain. Syntacticians have tended to see special clitics as filling normal syntactic positions and then displaced to their surface position under the influence of distinctive movement principles within the syntax. One difficulty with this is the fact that in some languages (of which certain forms of SerboCroatian are the most discussed), the element defining second position for the location of clitics is a unit in phonological terms (a prosodic word) but not necessarily a unit that ought to be accessible to the syntax. An alternative that avoids this difficulty is to note the close parallel between possible positions for clitics and for affixes and to treat special clitics as a class of affixes introduced directly into the surface form of phrases by principles closer to those of morphology than to syntax.

Conclusion The analysis of clitics and the principles by which they find their place in sentence structure and prosody involve an intricate interplay among all of the major components of grammatical structure, including syntax, phonology, morphology, and even semantics. These elements have been invoked by scholars in all of these areas as evidence for fundamental claims about linguistic structure, and an assessment of those claims is only possible on the basis of a clearer understanding of the subdivisions among clitics, and the appropriate mechanisms for accommodating their specific properties, than is often found in the existing literature. – Stephen R. Anderson

Codeswitching WORKS CITED AND SUGGESTIONS FOR FURTHER READING Anderson, Stephen R. 2005. Aspects of the Theory of Clitics. Oxford: Oxford University Press. Selkirk, Elizabeth. 1995. “The prosodic structure of function words.” In Papers in Optimality Theory, 439–70. University of Massachusetts Occasional Papers in Linguistics 18. Wackernagel, Jakob. 1892. “Über ein Gesetz der indogermanischen Wortstellung.” Indogermanische Forschungen 1: 333–436. Zwicky, Arnold. 1977. On Clitics. Bloomington: Indiana University Linguistics Club.

CODESWITCHING Introduction Codeswitching (CS) is defined as the use of two or more language varieties in the same conversation, not counting established borrowed words or phrases. Two general types of structural configurations occur. 1) Intersentential CS, switches for one sentence or many, is generally studied for its social implications (1). 2) Intrasentential or intraclausal CS is more studied for its grammatical configurations (2–4). (1)

(Policeman to heckler in Nairobi crowd, switching from English to Swahilisentences) How else can we restrain people from stealing except by punishment? Wewe si mtu ku-tu-ambia vile tu-ta-fanya kazi – tu-na sheria yetu. Swahili translation: “You aren’t a person to tell us how to do our work – we have our laws.” (Myers-Scotton 1993, 77)

(2)

(A clause in French embedded in a Brussels Dutch clause) [‘t is dat ][que j’ai dit à madame]. “That is what that I told the lady.” (Treffers-Daller 1994, 30)

(3)

(Single English content morpheme in a Hungarian-framed clause) játsz-ok school-ot play-s.pres school-acc “I’m playing school.” (Bolonyai 1998, 34).

(4)

(English verb stem with Swahili affixes in Swahili frame)… father a-li-m-buy-i-a vi-tabu a-ka-potez-a vy-ote s-past-obj-buy-appl-fv cl-book s-consec-losefv cl-all “… father bought for him books and he lost all [of them]” (Myers-Scotton 1997, 87)

CS and Its Social Meanings CS is a means of presenting a particular persona or negotiating interpersonal relationships in a given interaction, making it a major research topic for some sociolinguists and linguistic anthropologists. A starting point is John J. Gumperz’s (1982) notion that CS is one of the possible “contextualization cues” of the speaker’s pragmatic intentions. Also, researchers often mention E. Goffman’s concept of “footing,” and M. Bakhtin’s concept of speakers’ “multiple voices” that are echoes of earlier utterances.

Many studies remain at the descriptive level, but at least two models offer explanations for why CS occurs within a discourse. conversation analysis (CA) analysts emphasize a switch’s sequential positioning in conversation, claiming that it provides vital information about its sociopragmatic message (Auer 1998 inter alia; Li 2005). In contrast, the markedness model emphasizes that speakers use CS as a tool to present a certain persona; they exploit participants’ sense of the indexicality of each code (see indexicals) and of the contrast between the social import of codes in a given context (Myers-Scotton 1993 inter alia). Some analysts, such as B. Rampton, C. Stroud, and J. Gafaranga, emphasize CS as exemplifying the speaker’s creative agency. CS researchers agree on two points: 1) To engage in CS is largely an unconscious move, and 2) speakers seldom intend a single, specific meaning; potentially ambiguous or multiple meanings are part of the pragmatic message. Two overlapping generalizations capture differences in various approaches. First, the meaning of strategy, with its implication that CS carries messages of intentionality, divides analysts. Second, analysts differ on the role of community values and participants’ own sociolinguistic profiles, as well as a variety’s multiple associations, as they relate to a speaker’s motivation for making a switch.

CS and Its Grammatical Structure Most analysts agree that CS has a principled grammatical structure, but the principles they propose to constrain sentence/clause structure vary. Many early studies employed a linear-based framework; for example, Shana Poplack (1980) argues that possible switching depends on surface-level syntactic equivalences across participating languages. Some place importance on distinguishing borrowing from CS through quantitative analyses (e.g., Budzakh-Jones 1998). In contrast, The matrix language frame model links CS at abstract levels to psycholinguistic models of language production (Myers-Scotton 1997, 2002). Asymmetries between the structuring roles of the participating languages are stressed. Also, languages do not supply morpheme types equally. The 4-M model and a uniform structure principle explain different morpheme distributions with more precision (cf. Myers-Scotton and Jake 2009). Still other researchers argue that current syntactic theory of mainstream generative grammar, though intended for monolingual data, can explain CS parisomiously (MacSwan 2000). Although CS involves bilingual data (see bilingualism and multilingualism), researchers claim that no dominant or matrix language is needed. This conclusion is debated (cf. MacSwan 2005a, 2005b and Jake, Myers-Scotton, and Gross 2002, 2005). CS as a vehicle in convergence in grammatical patterns is also studied (e.g., Muysken 2000; Clyne 2003; Backus 2005). – Carol Myers-Scotton WORKS CITED AND SUGGESTIONS FOR FURTHER READING Auer, Peter, ed. 1998. Code-switching in Conversation: Language, Interaction and Identity. London: Routledge. Backus, Ad. 2005. “Codeswitching and language change: One thing leads to another?” International Journal of Bilingualism 9: 337–40. Bolonyai, Agnes. 1998. “In-between languages: Language shift/maintenance in childhood bilingualism.” International Journal of Bilingualism 2: 21–43.

159

Cognitive Architecture Budzhak-Jones, Svitlana. 1998. “Against word-internal codeswitching: Evidence from Ukrainian-English bilingualism.” International Journal of Bilingualism 2: 161–82. Clyne, Michael. 2003. Dynamics of Language Contact. Cambridge: Cambridge University Press. Gumperz, John J. 1982. Discourse Strategies. Cambridge: Cambridge University Press. Jake, Janice, Carol Myers-Scotton, and Steven Gross. 2002. “Making a minimalist approach to codeswitching work: Adding the matrix language.” Bilingualism, Language & Cognition 5: 69–91. ———. 2005. “A response to MacSwan (2005): Keeping the matrix language.” Bilingualism, Language, and Cognition 8: 271–6. Li, Wei. 2005. “‘How can you tell?’: Towards a common sense explanation of conversational code-switching.” Journal of Pragmatics 37: 375–89. MacSwan, Jeff. 2000. “The architecture of the bilingual language faculty: Evidence from intrasentential code switching.” Bilingualism, Language, and Cognition 3: 37–54. ———. 2005a. “Making a minimalist approach to codeswitching work: Adding the matrix language.” Bilingualism, Language, and Cognition 5: 69–91. ———. 2005b. “Remarks on Jake, Myers-Scotton and Gross’s response: There is no ‘matrix language.’” Bilingualism, Language and Cognition 8: 277–84. Muysken, Pieter. 2000. Bilingual Speech, A Typology of Code-mixing. Cambridge: Cambridge University Press. Myers-Scotton, Carol. 1993. Social Motivations for Codeswitching: Evidence from Africa. Oxford: Oxford University Press. ———. 1997. Duelling Languages, Grammatical Structure in Codeswitching. 2d ed. Oxford: Oxford University Press. ———. 2002. Contact Linguistics: Bilingual Encounters and Grammatical Outcomes. Oxford: Oxford University Press. ———. 2006. Multiple Voices: An Introduction to Bilingualism. Oxford: Blackwell. Chapters 6 and 9 deal with codeswitching for advanced undergraduates. Myers-Scotton, Carol, and Janice Jake. 2009. “A universal model of codeswitching and bilingual language processing and production.” In The Cambridge Handbook of Linguistic Code-Switching, ed. B. Bullock and A. Toribio, 336–57. Cambridge: Cambridge University Press. Poplack, Shana.1980. “‘Sometimes I’ll start a sentence in English Y TERMINO EN ESPAÑOL’: Toward a typology of code-switching.” Linguistics 18: 581–618. Treffers-Daller, Jeanine. 1994. Mixing Two Languages: French-Dutch Contact in a Comparative Perspective. Berlin: Mouton de Gruyter. Winford, Donald. 2003. An Introduction to Contact Linguistics. Cambridge: Cambridge University Press. Winford provides a comprehensive overview of codeswitching designed for beginning graduate students.

As digital computers evolved, so, too, did the notion of computer architecture. Computer designers had to pay attention not only to the needs of the user but also to additional constraints that arose with the development of high-level programming languages and with the invention of new hardware technologies. Brooks’s more modern definition of architecture reflects these developments: “The architecture of a computer system we define as the minimal set of properties that determine what programs will run and what results they will produce. The architecture is thus the system’s functional appearance to its immediate user” (Blaauw and Brooks 1997, 3). The key element here is that a computer’s architecture describes the information-processing capacities of a device without appealing to its specific hardware properties. In short, a computer’s architecture is a description of its logical and abstract information-processing properties (Dasgupta 1989).

The Cognitive Architecture The concept cognitive architecture is the direct result of applying the notion of computer architecture to human cognition. Cognitive scientists assume that cognition is information processing (e.g., Dawson 1998). Cognition – as information processing – must therefore be characterized by a fundamental set of logical and abstract properties (e.g., a primitive set of symbols and operations). By identifying this set of properties for human cognition, one specifies the cognitive architecture. For example, Z. W. Pylyshyn isolates “the basic operations for storing and retrieving symbols, comparing them, treating them differently as a function of how they are stored (hence, as a function of whether they represent beliefs or goals), and so on, as well as such basic resources and constraints of the system, as a limited memory. It also includes what computer scientists refer to as the ‘control structure,’ which selects which rules to apply at various times” (1984, 30). It is no accident that this account emphasizes symbols and primitive operations for their manipulation. This is because Pylyshyn wants to ensure that the architecture is indeed cognitive, which for him means that it must be representational: “It [the cognitive architecture] is the level at which the system is representational, and where the representations correspond to the objects of thought” (1991,191). There may be other levels of system organization above and below the cognitive architecture, but researchers like Pylyshyn would argue that these levels are not cognitive.

Architecture and Explanation COGNITIVE ARCHITECTURE The Architecture of a Computer The cognitive sciences have developed in large part from the application of concepts that initially arose in computer science. One such concept is that of architecture. The term computer architecture was originated by Frederick P. Brooks, Jr., a pioneering force in the creation of IBM’s early computers. For Brooks, “computer architecture, like other architecture, is the art of determining the needs of the user of a structure and then designing to meet those needs as effectively as possible within economic and technological constraints” (1962, 5).

160

Brooks’s (1962) original notion of computer architecture was driven by the goals of computer design: The architecture served as a functional account of capabilities to be used as a blueprint by hardware engineers in order to bring a computer into being. In the study of cognition and language, the information processor already exists. Why, then, is there a need for the cognitive architecture? The answer is that an architectural account converts a cognitive description into a cognitive explanation. Architectural components convert descriptions into explanations by providing a bridge between functional and physical accounts. They can do this because components of the cognitive architecture must be both cognitive and physical (e.g., Haugeland 1985, 100).

Cognitive Architecture These “two lives” are important because the predominant research methodology used by cognitive scientists is functional analysis (Cummins 1983). In functional analysis, a researcher attempts to account for some complex function by decomposing it into an organized system of subfunctions. Each subfunction often becomes the subject of its own functional analysis; this methodology is intrinsically iterative. However, if this were all that there was to this methodology, functional analysis would fall victim to an infinite regress and generate an infinite variety of unexplained functions (Ryle 1949). To avoid Ryle’s regress, functional analysis also attempts to progressively simplify the proposed subfunctions: “The highest level design breaks the computer down into a committee or army of intelligent homunculi with purposes, information and strategies. Each homunculus in turn is analyzed into smaller homunculi, but, more important, into less clever homunculi” (Dennett 1978, 80). At some point, the “less clever” homunculi become so simple that they can be replaced by physical devices that carry out the desired function. At this point, the functional description is physically subsumed (Cummins 1983) and becomes explanatory. From this perspective, the cognitive architecture can be described as the set of primitive functions that have been subsumed in a functional analysis. Their functional – or cognitive – role defines how these components mediate complex information processing. Their physical role defines how such processing can be physically implemented and explained.

Architecture and Language To the extent that language is mediated by information processing, an explanation of language must be grounded in a cognitive architecture. In many respects, linguistics provides prototypical examples of architectural accounts of language. However, it can also be argued that the cognitive architecture holds an uneasy position within some linguistic theories. On the one hand, dominant theories in linguistics appear to provide architectural accounts of language. We have already seen that a cognitive architecture requires a set of functions to be established as primitives by subsuming them as neural implementations. From its inception, the standard Chomskyan approach to language appears to strive toward this kind of architectural account. First, the specification of a generative grammar provides a detailed account of a set of complex tokens, and the rules for their manipulation, that are required to assign structural descriptions to sentences. Furthermore, this grammar is intended to describe (at least in part) cognitive processing: “Every speaker has mastered an internalized a generative grammar that expresses his knowledge of his language” (Chomsky 1965, 8). In short, one purpose of a generative grammar is to describe the functional properties of an internalized set of symbols and rules. Second, the Chomskyan tradition presumes a strong link between generative grammar and the brain. This link is included in the general view that the language faculty is a biological organ (Hauser, Chomsky, and Fitch 2002). According to Chomsky, “The human brain provides an array of capacities that enter into the use and understanding of language (the language faculty); these seem to be in good part specialized for that function

and a common human endowment” (1995, 167). Complete accounts of human language must appeal to these biological underpinnings. Thus, in the Chomskyan tradition, an architectural account would include the specification of a generative grammar, as well as additional processes that are necessary and sufficient for mediating language. Furthermore, this account would be biologically grounded. On the other hand, the Chomskyan tradition takes strong positions that conflict with the general notion of cognitive architecture sketched earlier. Two of these positions require special mention here. The first is that a theory in linguistics should focus on competence, and not performance. The second is that the language faculty is modular in the sense of J. A. Fodor (1983). These positions are important because both have been used to exclude certain concepts from linguistic study that are critical components of the cognitive architecture. For example, memory has not been deemed to be properly part of the linguistic domain. That is, while memory might impact language production or comprehension (e.g., by limiting the number of embedded clauses that can be processed), some researchers would argue that memory is not part of the language faculty proper. For some, memory limitations are viewed as being important to a theory of language performance, but not to a theory of language competence (e.g., Chomsky 1965). Furthermore, memory limitations are related to a general cognitive resource, which by definition therefore cannot be solely part of a language faculty (Hauser, Chomsky, and Fitch 2002). More recent variations of the Chomskyan approach provide a more flexible view of the competence/performance distinction and, as a result, lead to theories that make strong proposals about cognitive architecture as construed earlier: “A theory that allows us to readily relate competence to performance ought to be favored over one that creates hard boundaries between the two” (Jackendoff 2002, 197). Jackendoff’s parallel architecture (2002) is one such theory. He assumes that syntax is not the only system responsible for the combinatorial nature of language. Instead, he proposes a parallel architecture in which three separate levels (phonology, syntax, and semantics) are independent, each having their own primitive operations and combinatorial principles. Though independent, the three levels are linked by interface constraints. Thus, all three levels cooperate to produce the generative structure of language. Furthermore, this theory of linguistic structure can be mapped to a parallel architectural theory in which each level is a modular processor, but there are interfaces between the three processors that share a common linguistic memory. The preceding discussion of language and the cognitive architecture has emphasized theories that adopt a so-called classical perspective that emphasizes the rule-governed manipulation of symbols. It is important to realize that alternative architectures for language have also been explored. For instance, classical researchers have argued that artificial neural networks are not capable of modeling the generative properties of language (Fodor and Pylyshyn 1988). However, empirical and theoretical analyses would indicate that this criticism is not valid (Hadley and Hayward 1997; Siegelmann 1999). As a result, many examples

161

Cognitive Grammar exist in which neural networks have been used to model a variety of language phenomena (Mammone 1993; Sharkey 1992). – Michael R. W. Dawson WORKS CITED AND SUGGESTIONS FOR FURTHER READING Blaauw, G. A., and F. P. Brooks. 1997. Computer Architecture: Concepts and Evolution. Reading, MA: Addison-Wesley. Brooks, F. P. 1962. “Architectural philosophy.” In Planning a Computer System – Project Stretch, ed W. Buchholz, 5–16. New York: McGrawHill. Chomsky, N. 1965. Aspects of the Theory of Syntax. Cambridge, MA: MIT Press. ———. 1995. The Minimalist Program. Cambridge, MA: MIT Press. Cummins, R. 1983. The Nature of Psychological Explanation. Cambridge, MA: MIT Press. Explores the role of the architecture in providing explanatory power to functional analyses. Dasgupta, S. 1989. Computer Architecture: A Modern Synthesis. New York: Wiley. Dawson, M. R. W. 1998. Understanding Cognitive Science. Oxford: Blackwell. Dennett, D. 1978. Brainstorms. Cambridge, MA: MIT Press. Fodor, J. A. 1983. The Modularity of Mind. Cambridge, MA: MIT Press. Fodor, J. A., and Z. W. Pylyshyn. 1988. “Connectionism and cognitive architecture.” Cognition 28: 3–71. Hadley, R. F., and M. B. Hayward. 1997. “Strong semantic systematicity from Hebbian connectionist learning.” Minds and Machines 7: 1–37. Haugeland, J. 1985. Artificial Intelligence: The Very Idea. Cambridge, MA: MIT Press. Hauser, M. D., N. Chomsky, and W. T. Fitch. 2002. “The faculty of language: What is it, who has it, and how did it evolve?” Science 298: 1569–79. Jackendoff, R. 2002. Foundations of Language: Brain, Meaning, Grammar, Evolution. Oxford: Oxford University Press. Provides an overview of the state of linguistics that pays special attention to architectural issues. Mammone, R. J. 1993. Artificial Neural Networks for Speech and Vision. New York: Chapman and Hall. Contains many examples of artificial neural networks applied to specific areas of speech and language. Pylyshyn, Z. W. 1984. Computation and Cognition. Cambridge, MA: MIT Press. A detailed examination of the role of cognitive architecture in cognitive science. ———. 1991. “The role of cognitive architectures in theories of cognition.” In Architectures for Intelligence, ed. K. VanLehn, 189–223. Hillsdale, NJ: Lawrence Erlbaum. Ryle, G. 1949. The Concept of Mind. London: Hutchinson and Company. Sharkey, N. E. 1992. Connectionist Natural Language Processing. Dordrecht: Kluwer Academic Publishers. Siegelmann, H. T. 1999. Neural Networks and Analog Computation: Beyond the Turing Limit. Boston: Birkhauser.

COGNITIVE GRAMMAR Cognitive Grammar (CG) refers to the theory of language articulated most comprehensively in Ronald W. Langacker (1987, 1991), two mutually dependent volumes that are best read together. Langacker (1988) provides a succinct chapter-length overview of his theory, while Taylor (2002) and Evans and Green (2006, 553–640) are highly recommended as student-oriented introductions to the theory. CG is wide ranging in its scope and provocative in its approach to an understanding of linguistic

162

structure. It has played a key role in the history of cognitive linguistics. Fundamental to CG is the idea that language is an integral part of human cognition and cannot be properly understood without reference to cognitive abilities. A pervasive feature of CG is the determination to reconcile accounts of linguistic structure with what is known about cognitive processing in domains other than language. CG contrasts in this respect with models that insist upon a discrete, autonomous grammar module and the autonomy of syntax. The cognitive orientation of CG is apparent from a reliance on notions such as sensory imagery, perspective, mental scanning, attention, and figure versus ground asymmetry in accounting for linguistic phenomena. In broad terms, grammatical structure is explained as conventional imagery, with alternate structures reflecting alternate construals of the conceived situation. Not surprisingly, the cognitive notions underlying CG assume a relatively abstract interpretation when applied to some aspects of linguistic structure. For example, cognitive processes such as registration of contrast, scanning of a field, and perception of a boundary are all deemed relevant for explicating the notion of a count noun, understood as a “bounded region in some domain” in Langacker (1987, 189–203). Such processes may be obvious factors in the conceptualization of nouns with clear spatial boundaries (e.g., cup, pencil), but a more abstract interpretation of these processes is clearly required in other domains. Body-part nouns (e.g., waist, shoulder, side) must be explicated in terms of a virtual boundary that does not correspond to any visible, objectively identifiable demarcation. Likewise, the notions of figure and ground familiar from the study of perception are seen as underpinning various relational asymmetries in language. These notions have most obvious relevance in the case of words relating to the spatial domain, such as the contrasting pair above and below, where there is a figure-ground reversal of the figure and the conceptual reference point. The terms trajector (an adaptation of the notion of figure) and landmark (an adaptation of the notion of ground) are used to refer to the specifically linguistic manifestation of the perceptual notions of figure and ground, such that the book is the trajector and the table is the landmark in the book under the table. Conversely, the table is the trajector and the book is the landmark in the table over the book. More abstractly still, the traditional syntactic contrast between subject and object is construed in terms of relative salience, such that the subject is a primary clausal figure, or trajector, and the object is a secondary clausal figure, or landmark. At the heart of CG is the concept of a symbolic unit, consisting of semantic structure standing in correspondence to a phonological structure. Consistent with the idea that language is part of conceptual structure, semantic structure is understood as “conceptualization tailored to the specifications of linguistic convention” (Langacker 1987, 99; see Talmy 2000, 4 for a similar view of semantic structure). CG takes the notion of symbolic unit (similar to, but not to be equated simply with, the Saussurean sign) as fundamental and applicable at all levels of representation, including lexical items, grammatical classes, and grammatical constructions. The lexical item tree, for example, consists of a semantic unit [tree] and a corresponding phonological unit [tri], which combine to form the symbol for tree, [[tree]/[tri]].

Cognitive Grammar The same apparatus is applicable for defining a word class such as a noun, abbreviated by Langacker as [[thing]/[…]], indicating a schematic semantic specification of a thing but without any specific content phonologically. A morphologically more complex lexical item such as trees is represented as a composite structure integrating two symbolic units representing the noun tree and the plural [z]: [[[tree]/[tri]]-[[pl]/[z]]]. Grammatical constructions are in principle no different from a lexical item like trees in terms of the descriptive apparatus required to capture all the relevant detail, with each of the component structures of a construction represented by a symbolic unit. Grammatical morphemes appearing in a construction, such as of, are treated as symbolic units in their own right, with semantic structure (of, for example, specifies a part–whole relation). The integration of any two symbolic units goes hand in hand with distinguishing the dependent and autonomous parts of the composite structure. As far as semantic structure is concerned, [tree] is autonomous, while [pl] is dependent, requiring an elaboration by a noun to complete the structure. In terms of phonological structure, [tri] is pronounceable as a whole syllable and can be considered autonomous, while the single consonant [z] is dependent. A striking feature of CG is the detail provided for in the integration of structures into larger composite structures. The analysis of the English passive construction in Langacker ([1991] 2001, 101–47) illustrates the theoretical notions relevant to a detailed grammatical description and is recommended as a prime example of a full-blown CG account of a construction type. Briefly, and consistent with the foregoing remarks, each morpheme in the passive (including by and the auxiliary verbs) has its own symbolic representation, giving rise to the overall semantic structure, just as the active counterpart has its own compositional structure and resulting semantic structure. Passive clauses do not derive from active clauses in this view, nor do they derive from some abstract structure underlying actives and passives. Rather, passive clauses exist in their own right as instantiations of a construction type with its own distinctive way of integrating symbolic units, reflecting a particular construal of the event. While phonological structure can be fruitfully explored within CG (see Langacker 1987, 328–48, 388–400; Taylor 2002, 78–95), it is semantic structure that has received most attention and for which most theoretical apparatus has been developed. Fundamental to semantic structure is the idea of a network that is employed to represent polysemy relationships and to provide motivation for conventional and novel extensions. Each node of the semantic network, together with the associated phonological structure, represents a semantic variant of the lexical item. Two types of relationships figure prominently in these networks: schematicity, whereby one node of the network expresses a meaning fully contained in another node (see schema), and extension, understood as a relationship between semantic nodes of a lexical item involving a conflict in semantic specifications. The word head, for example, can be assigned a sense [part of a whole which controls the behavior of the whole] that is schematic relative to finer-grained elaborations, such as [part of the human body where thinking is located] and [person who manages an administrative unit]. In some cases, a highest-level node or superschema can be proposed, encompassing all lower-level senses in the network, though such superschemas

are not feasible for every network. The extensive polysemy of head, for example, makes one single superschema covering such diverse senses as “head of a lettuce,” “head of a bed,” “head of a university department,” and so on unlikely. Semantic extension holds between the more basic sense of “human head” and the sense of “head of an administrative unit.” The node that is the source of the extension constitutes a local prototype (with respect to the extended sense); where one node is experienced as representative of the whole category, as is likely in the case of the “human head” sense of head, we speak of a global prototype. There is clearly variation among speakers in their judgments about nodes and relationships within the network, including their ability to identify relatedness of senses and to extract schematic meanings. This variation poses challenges for description but does not negate the need to acknowledge the reality of such networks. CG adopts a nonreductionist or maximalist stance in its analysis of linguistic structure, contrasting with prevailing reductionist, minimalist approaches in contemporary linguistics. The nonreductionist approach of CG explicitly provides for the listing of highly specific patterns alongside the statement of more general patterns, rather than recognizing only the most general rules and schemas. The existence of a general rule of plural formation in English suffixing /s/~/z/~/əz/ to a noun, for example, does not mean that certain instantiations of the rule, such as cats, dogs, horses, and so on, have no place in a grammatical description. On the contrary, where such instantiations have gained unit status and are activated directly by the speakers, it is appropriate to recognize them alongside other symbolic units, grammar and lexicon forming a continuum of types of symbolic elements. Even when particular instantiations conform to a general rule, they may acquire unit status in their own right, for example, through high frequency of use. Acknowledging low-level, highly specific instantiations runs counter to deeply entrenched practices in contemporary linguistics, which has been preoccupied with higher-level generalizations and the principle of economy in description. Langacker has repeatedly emphasized the desirability of both general and particular statements in linguistic description, referring to the assumption that a phenomenon is to be accounted for in a mutually exclusive way as either a rule or a list as the rule/list fallacy (Langacker 1987, 40–2). Grammar, in CG terms, amounts to “a structured inventory of conventional linguistic units” (Langacker 1987, 73). The units, so conceived, may be semantic or phonological; they range from the symbolic units consisting of a single morpheme to larger composite symbolic units at the clause level, and they include highly specific, as well as highly schematic, units. This conception of grammar makes CG comparable to construction grammars, which are also inventory-based (cf. Evans and Green 2006, 475–83), particularly radical construction grammar (Croft 2001). By including quite specific syntagmatic patterns within a grammatical description, CG is able to comfortably accommodate phenomena that have been largely neglected in linguistic theorizing, for example, the collocational patterning of great idea, absolutely fabulous, and so on involving combinations of particular words. The greater emphasis on specific patterning makes CG highly compatible with the methodology of

163

Cognitive Linguistics and Language Learning corpus linguistics and other approaches that focus on language in use whereby actual usage, including frequency of occurrence and patterns of co-occurrence, can be observed and used as a basis for extracting patterns of varying generality (see also the entries treating connectionism). Fully general, exceptionless rules are seen as atypical, and while it is valid to seek out such rules, it would be misguided in this approach to attend only to the most general patterns. Finally, a word on notation employed in CG. There is an array of notational devices used by Langacker, who employs a distinctive and highly original geometric style of representation (in his earlier publications, he used the term “Space Grammar” to refer to his approach). To some extent, the notation is intuitive: A circle is used to denote a [thing] entity; thicker, darker lines represent the profile, that is, the designated thing or relation in the semantic structure of a morpheme. A full appreciation of the notation, however, requires careful study. Of course, not all the detail needs to be represented all the time, and CG ideas can be effectively incorporated into linguistic analyses simply in prose or with a minimum of notation (as in Taylor 2002). – John Newman WORKS CITED AND SUGGESTIONS FOR FURTHER READING Croft, William. 2001. Radical Construction Grammar: Syntactic Theory in Typological Perspective. Oxford: Oxford University Press. Evans, Vyvyan, and Melanie Green. 2006. Cognitive Linguistics: An Introduction. Edinburgh: Edinburgh University Press. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar. Vol. 1. Theoretical Prerequisites. Stanford, CA: Stanford University Press. ———. 1988. “An overview of cognitive grammar.” In Topics in Cognitive Linguistics, ed. Brygida Rudzka-Ostyn, 3–48. Amsterdam and Philadelphia: John Benjamins. ———. 1991. Foundations of Cognitive Grammar. Vol. 2. Descriptive Application. Stanford, CA: Stanford University Press. ———. [1991] 2001. Concept, Image, and Symbol: The Cognitive Basis of Grammar. Berlin and New York: Mouton de Gruyter. The chapters in this volume cover key areas of grammar (grammatical valence, case, passive, etc.) and can be read more or less independently of one another – a good balance between CG theory and application to data. Newman, John. 2004. “The quiet revolution: Ron Langacker’s fall quarter 1977 lectures.” In Imagery in Language: Festschrift in Honour of Professor Ronald W. Langacker, ed. Barbara LewandowskaTomaszczyk and Alina Kwiatkowska, 43–60. This chapter gives a firsthand account of an early presentation of the material that eventually became Langacker (1987). Talmy, Leonard. 2000. Toward a Cognitive Semantics. Vol. 1. Concept Structuring Systems. Cambridge, MA: MIT Press. Taylor, John. 2002. Cognitive Grammar. Oxford: Oxford University Press.

COGNITIVE LINGUISTICS AND LANGUAGE LEARNING A fundamental challenge for any theory of language is to provide a convincing account of how the prelinguistic child becomes a competent member of his or her linguistic community. To be convincing, the developmental account should be consistent with the general model of language posited for the adult system. Michael Tomasello (2003, 4–5) calls this the “how do we get there (to the adult language system) from here (the pre-linguistic infant)” problem. A cognitive linguistic approach to the adult

164

language system holds that language is a reflection of human cognition and that language can be accounted for by the interaction of the complex set of cognitive capabilities with which humans are endowed. Language is understood to emerge from contextualized use; that is, it is usage based. The language system to be learned is an inventory of linguistic constructions (units with a phonological pole, i.e., the construction’s form, and a semantic pole, i.e., the construction’s meaning) which range in size from lexical items and morphemes to syntactic and even discourse (see discourse analysis [linguistic]) patterns. The challenge for language acquisitionists working within a cognitive linguistic framework is to account for the rapid learning of this vast array of linguistic patterns, drawing solely on general cognitive processes. Over the past 25 years, developmental psychologists have developed a large body of observational and experimental evidence that begins to do just that. Elizabeth Bates famously said, “Language is a tool. We use it to do things” (1976, 1). Cognitive linguists hold that the primary thing for which we use language is communication. Humans are highly social by nature. We also have the unique ability to think about entities and events that are not in the immediate environment. Since we cannot communicate using mental telepathy, we have to somehow externalize our internal conceptualizations to make them available to others. Language is the major tool we have developed to accomplish this task. Cognitive linguists hold that when children learn a language, what they are learning is constructions, of varying sizes and degrees of abstractness, as they engage in using language in context. This learning process takes place over a rather extended period of time, with most of the pieces in place by the time the child is six or seven. According to the developmental psychologist Jean Mandler (2004), children begin forming rudimentary concepts, many of which form the basic semantic frames from which language is constructed, in early infancy. Very early on, the infant begins a cognitive process of reformatting raw perceptional information into simple, schematized spatial representations that express fundamental experiences, such as self-motion, caused motion, and containment (see schema ). Experimental evidence shows that by three months, the infant can distinguish typical human motion from mechanical motion. The infant learns that certain entities start up under their own power while others do not. The same entities that generate self-motion can also cause other entities to move or change; the entities that do not generate self-motion tend to be acted upon, and so on.