sign in sign up
<<
Home , Applied linguistics, Functional linguistics, Geoffrey Sampson, Glossematics, Grammaticality, Levels of adequacy

KEY IN IN IDEAS KEY KEY IDEAS IN LINGUISTICS AND THE OF KEY IDEAS IN LINGUISTICS Edited by Siobhan Chapman and Christopher Routledge AND THE

This book offers introductory entries on eighty ideas that have shaped the study of language up to the present day. Entries are written by experts in the fields of linguistics and the philosophy of language to reflect the full range of approaches LANGUAGE OF PHILOSOPHY THE AND and modes of . Each entry includes a brief of the , an account of its development, and its impact on the field of language study. The book is written in an accessible with clear of technical terms, guides to further , and extensive cross-referencing between entries. A useful additional of this book is that it is cross-referenced throughout with Key Thinkers in Linguistics and the Philosophy of Language (Edinburgh, 2005), revealing significant connections and continuities in the two related disciplines. Ideas covered range from Data, , and , through Generative , Cognitivism, and , to Political Correctness, , and Corpora. Features • The only single-volume book to specifically on ideas from both linguistics and the philosophy of language • Accessibly written for use at all levels, including undergraduate, postgraduate, academic, and other general readers in the fields of linguistics and the philosophy of language • Extensively cross-referenced both within itself and with Key Thinkers in Linguistics and the Philosophy of Language to provide a unique reference

resource Routledge Christopher and Chapman Siobhan by Edited

Siobhan Chapman is Senior Lecturer in at the University of Liverpool. http://siobhanchapman.co.uk Christopher Routledge is a freelance writer and editor. http://chrisroutledge.co.uk

ISBN 978 0 7486 2619 9 Edinburgh University Press Edited by Siobhan Chapman and Christopher Routledge 22 George Square Edinburgh EH8 9LF Edinburgh www.eup.ed.ac.uk Cover illustration: Image © Corbis 2007 Cover design: Michael Chatfield

Pantone 330 Key Ideas in Linguistics and the Philosophy of Language

Key Ideas in Linguistics and the Philosophy of Language

Edited by Siobhan Chapman and Christopher Routledge

Edinburgh University Press © in this edition Edinburgh University Press, 2009 © in the contributions is retained by the authors

Edinburgh University Press Ltd 22 George Square, Edinburgh

Typeset in 10/12pt Sabon by Servis Filmsetting Ltd, Stockport, Cheshire, and printed and bound in Great Britain by MPG Biddles Ltd, King’s Lynn, Norfolk

A CIP record for this book is available from the British Library

ISBN 978 0 7486 2618 2 (hardback) ISBN 978 0 7486 2619 9 (paperback)

The right of the contributors to be identified as authors of this work has been asserted in accordance with the Copyright, Designs and Patents Act 1988. Contents

Preface vii Intuition 103 Acknowledgements ix Language Games 106 Notes on Contributors x Language of Thought 110 Langue/Parole 113 Acceptability/ 1 116 Adequacy 2 Linguistic Variable 119 / 3 Logic 120 4 127 Analytic/Synthetic 11 Logical 128 Artificial Intelligence 14 Mentalism 130 Behaviourism 17 Metaphor 131 Cognitivism 21 Minimalism 134 Compositionality 24 Modality 138 Connotation/ 25 Model-theoretic Semantics 139 Continuity 26 142 Conventional 28 Nonnatural Meaning 144 29 Optimality 145 Corpora 36 Ordinary Language Philosophy 149 Correspondence Theory 42 Performative 155 43 156 Deconstruction 44 Politeness 157 Deduction/Induction 47 Political Correctness 161 Definite Descriptions 54 Port-Royal Logic 163 Descriptivism 57 Semantics 164 (Critical) Analysis 58 Poststructuralism 170 Distinctive Features 65 173 Emic/Etic 66 Private Language 179 / 67 Propositional Attitudes 181 Feminism 74 185 Generative 77 Prototype 186 Generative Semantics 78 Psychoanalysis 189 79 Theory 193 81 199 Ideational 84 Sense/Reference 200 86 and 205 Indeterminacy 92 Situational Semantics 209 Innateness 96 Act Theory 212 Integrationism 97 219 Intentionality 100 Systemic-Functional 225 CONTENTS

Transformational-Generative 247 Grammar 232 Use/Mention 251 Theories 238 Truth Value 244 Index 253 Type/Token 246

vi Preface

The ideas described in this book have been developed in linguistics and the philosophy of language, as well as in some related disciplines such as mathe- matics, logic and . They necessarily represent only a very small - portion of the long tradition of the serious study of language; we have chosen them because of their impact on current work in linguistics and the philosophy of language. These two disciplines are subdivided into many different branches. Linguistics, for example, includes work undertaken in semantics, , phonology, , and many other fields. In general we have not treated these individual fields as key ideas in their own right. You will not, for instance, find an entry here on ‘Pragmatics’, but you will find topics from the field of pragmatics discussed under entries such as ‘Implicature’, ‘’ and ‘ Theory’. The names of different branches of linguis- tics and the philosophy of language do, however, appear in the index. Similarly, we have avoided allocating entries to descriptive categories such as ‘’, ‘’ or ‘conjunction’. The entries are arranged in simple alphabetical order, and aim to elucidate each key idea, offering a succinct definition followed by a more discursive account of the development of the idea and of its impact and current relevance. The book can therefore be used as a stand-alone reference work. However, it is also designed to be used in conjunction with our Key Thinkers in Linguistics and the Philosophy of Language (2005). The coverage of these two volumes is similar: broadly, the study of language in the Western tradition from antiquity to the present-day, with an emphasis on work that has been influential on lin- guistics and the philosophy of language as they are practised in the early part of the twenty-first century. However, the two are complementary in that they arrange and present the material in different ways. Key Thinkers considers dif- ferent ways of thinking about language in the of the work of the particular figures with which they are most closely associated, drawing out con- tinuities and developments of thought in their particular historical and social context. Key Ideas focuses on the development of specific ways of thinking, sometimes across many decades or centuries, considering the influences on these ways of thinking, the relationships between them, and their overall significance. Each entry is cross-referenced both internally within this book and externally to Key Thinkers. The internal cross- (‘See also’) draw attention to other key ideas that are relevant to the particular entry. These may be ideas that are concerned with similar or related issues (such as the reference from ‘Deduc- tion/Induction’ to ‘Empiricism/Rationalism); ideas that form a specific topic within a more general approach (such as the reference from ‘Speech Act Theory’

vii PREFACE to ‘Performative’); or conversely ideas that provide a broader framework for the discussion of the specific issue in question (such as the reference from (‘Ade- quacy’ to ‘Transformational-’). Within each entry, any term that is itself a key idea with its own entry is marked with an on first use. The external cross-references (‘Key Thinkers’) point the reader towards relevant entries in Key Thinkers. The entries are all concerned with encouraging the reader to find out more, and therefore with pointing outside these two volumes to other and more detailed reading. Each entry concludes with a list of suggested , divided into ‘Primary sources’ and ‘Further reading’. Between them these sections give the full references of any works mentioned or quoted in the entry, along with details of other particularly salient works. The section of ‘Further reading’ may also include suggestions of useful introductions to or overviews of the relevant key idea.

viii Acknowledgements

A book of this kind depends on the cooperation and goodwill of a large of people and it has been a privilege to work with all of them. We would like to take this opportunity to thank all of our contributors to this book and its companion volume Key Thinkers in Linguistics and the Philosophy of Lan- guage (2005) for their diligence and professionalism. Two deserve special mention. We would like to thank Sarah Edwards and her team at Edin- burgh University Press for their enduring enthusiasm and support over the course of two lengthy book projects. Jeremy Kaye, whose father Alan S. Kaye was working on entries for this book when he died in May 2007, generously gave his time searching for material that Alan was preparing for us and deliv- ered the one entry his father had managed to complete before becoming too ill to continue. Alan was passionately committed to the study of language and a valuable contributor to these books. We dedicate this volume to his memory.

ix Notes on Contributors

Varol Akman, Chair, Department of Philosophy Bilkent University, Turkey. Salvatore Attardo, Professor, English Department, Youngstown State University, USA. Annalisa Baicchi, Professor of English, Department of Theoretical and , University of Pavia, Italy. Jennifer A. Baldwin, freelance translator (http://jenniferbaldwin.com). Alex Barber, Department of Philosophy, The Open University, UK. Philip Carr, Professor of Linguistics, Department of English, Montpellier University, France. Siobhan Chapman, Senior Lecturer in English Language, University of Liverpool, UK. Billy Clark, Lecturer in English Language, Middlesex University, UK. John Collins, Department of Philosophy, University of East Anglia, UK. Yousif Elhindi, Associate Professor, Department of English, East Tennessee State University, USA. Iván García Álvarez, Lecturer in Linguistics, School of , University of Salford, UK. Agustinus Gianto, Professor of Semitic and Linguistics, Pontifical Biblical Institute, Rome, Italy. Hans Götzsche, Associate Professor and Director, Centre for Linguistics, Aalborg University, Denmark. Eva Herrmann-Kaliner, computational linguist, Germany. Patrick Honeybone, Lecturer in Linguistics and English Language, University of Edinburgh, UK. Asa Kasher, Laura Schwarz-Kipp Professor of Professional Ethics and Philosophy of Practice, University, . Alan S. Kaye (1944–2007), Professor of English, California State University Fullerton, USA. Stavroula-Thaleia Kousta, Fellow, Department of Psychology, University College London, UK. x NOTES ON CONTRIBUTORS

Guy Longworth, Assistant Professor, Department of Philosophy, University of Warwick, UK. Stephen McLeod, Department of Philosophy, University of Liverpool, UK. Marie Nilsenová, Assistant Professor, Department of and , Tilburg University, Netherlands. Ingrid Piller, Professor of Linguistics, Macquarie University, Australia. Geoffrey Poole, Lecturer in Linguistics, School of English Literature, Language, and Linguistics, University of Newcastle upon Tyne, UK. Kanavillil Rajagopalan, Professor of the Semantics and Pragmatics of Natural Languages, State University at Campinas (UNICAMP), Brazil. Christopher Routledge, freelance writer and editor (http://chrisroutledge. co.uk). Karl Simms, Senior Lecturer, School of English, University of Liverpool, UK. Jürg Strässler, Universities of Berne and Zurich, Switzerland. Marina Terkourafi, Department of Linguistics, University of Illinois at Urbana-Champaign. Geoff Thompson, Senior Lecturer in Applied Linguistics, School of English at the University of Liverpool, UK Dominic Watt, Lecturer in Forensic Speech , Department of Language and Linguistic Science, University of York, UK. David Witkosky, Professor of German and , Auburn University, Montgomery, USA.

xi

ACCEPTABILITY/ by a native speaker. A grammatical string is not necessarily meaningful, as GRAMMATICALITY exemplified by Chomsky’s famous sen- tence ‘Colorless green ideas sleep furi- Acceptability is the extent to which a ously’. However, it is assumed that allowed by the rules to be even nonsensical grammatical sen- grammatical is considered permissible tences can easily be pronounced with a by speakers and hearer; grammatical- natural intonation and that speakers ity is the extent to which a ‘string’ of are able to recall them more easily than language conforms with a set of given ungrammatical sentences. Speakers rules. It is assumed that a native also supposedly have intuitions about speaker’s grammar generates gram- grammaticality (or grammaticalness; matical strings and that the speaker these two terms appear to be inter- has the ability to judge a certain string changeable), determined by their com- to be either acceptable or not in her petence (in other , of language. In practice, the two notions a language). are frequently confounded and speak- Acceptability, on the other hand, is ers are typically asked to give their related to speaker’s performance, ‘grammaticality judgements’ instead that is the actual use of her language of ‘acceptability judgements’. in concrete situations. As stressed by Chomsky, acceptability should not See also: Innateness; Mentalism; be confused with grammaticality: Transformational-Generative while an acceptable sentence must Grammar be grammatical, not just any gram- Key Thinkers: Chomsky, Noam matical sentence is necessarily accept- In prescriptive grammar books able. For a sentence to be judged employed in language teaching, a acceptable, it must also appear natu- string is considered to be grammatical ral and appropriate in a given con- if it conforms with a set of prescribed text, be easily understood and, norms. These norms are usually based possibly, be to a certain extent con- on conventionalised rules that form a ventionalised. part of a higher/literary register for a Both grammaticality and accept- given language. For some languages, ability are considered to be gradient these norms are defined and periodi- properties by Chomsky and in the lin- cally updated by an appointed body of guistic literature they are typically experts whose rulings are occasionally expressed with a combination of ‘?’ questioned by the frustrated members and ‘*’ for grammaticality – a sen- of the public. tence marked with ‘**’ being strongly In it has ungrammatical, a sentence marked become customary to utilise the term with a ‘?*’ being questionable, and so ‘grammaticality’ in line with Noam on – and ‘#’ for acceptability. Chomsky’s generative approach. In It has been observed that although generative grammar, a string is gram- speakers may have intuitions about matical – or well formed – if it could be grammaticality, they cannot be generated by a grammar internalised expected to translate their intuitions

1 ADEQUACY into judgements; hence, a categorisa- mathematics known as formal lan- tion task can only rely on their ability guage theory. The idea was that one to judge the acceptability of a string could formulate a finite set of rules (Schütze 1996). This assumption is which generated the set of sentences often violated in the process of data which constituted a specific language, collecting and speakers are typically such as English. The set of sentences encouraged to give grammaticality constituting a language was said to judgements of marginally sounding be infinite. That infinitude is guaran- sentences independently of an actual teed by the recursive nature of the discourse situation. rules. is the phenomenon whereby a given syntactic category Primary sources can be embedded within a category of Chomsky, N. (1957). . the same type. For instance, relative The Hague: Mouton. clauses in English can be embedded Chomsky, N. (1965). Aspects of the Theory within relative clauses, as in the sen- of Syntax. Cambridge, MA: MIT Press. tence ‘I know the woman who shot the man who held up the bank’. The Further reading relative clause ‘who shot the man Schütze, C. (1996). The Empirical Base of who held up the bank’ contains the Linguistics. Chicago and London: The relative clause ‘who held up the University of Chicago Press. bank’. Recursion is widely believed to be a universal feature of human Marie Nilsenová languages. One of the aims of early generative was to generate all and only the sentences which constitute a ADEQUACY specific language. A grammar was said to be observationally adequate if In the theory of transformational- it did this. The aim was to construct generative grammar* generative gram- grammars which did not generate mars are said to be evaluated with sequences which were ill-formed, such respect to three levels of adequacy: as ‘*The house have might been built’, observational adequacy, descriptive where the asterisk denotes ungram- adequacy and explanatory adequacy. maticality. Sequences are judged to be grammatical or ungrammatical on the See also: Acceptability/ basis of the intuitive grammaticality Grammaticality; Intuition; judgements of native speakers of the Transformational-Generative language. In making such judgements, Grammar; Universal Grammar native speakers are said to be access- Key Thinkers: Chomsky, Noam ing a -internal grammar, which is In Syntactic Structures (1957) Noam taken to constitute largely uncon- Chomsky defined a language as a set scious linguistic knowledge, known as of sentences. Generative grammars competence. were taken by Chomsky to be formal A given generative grammar will grammars, based on a branch of generate not only strings of words,

2 but also structural descriptions of AMBIGUITY/ those strings. For instance, one gram- mar might generate a structural VAGUENESS description of the phrase ‘The old man’ in which there are only three Ambiguity and vagueness are related constituents: ‘the’, ‘old’ and ‘man’, in but distinct . Ambiguity, the a flat linear string. Another grammar of having more than one lin- could generate a structural descrip- guistically encoded meaning, is an tion in which there are two interme- important notion in several areas of diate constituents: ‘the’ and ‘old linguistics (including syntax, seman- man’. The two grammars differ in the tics and pragmatics) as well as in phi- structural descriptions they generate. losophy of language. Vagueness, the In Aspects of the Theory of Syntax property of having no uniquely deter- (1965), Chomsky argued that a gen- minable value or interpretation, has erative grammar is justified to the been discussed by since extent that the structural descriptions ancient times and is still a focus of it generates correspond to the intu- attention for philosophers, semanti- itive grammaticality judgements of cists and pragmatists. the native speaker. He also sug- See also: Connotation/Denotation; gested that linguists should seek Conventional Meaning; explanatory adequacy: they should Nonnatural Meaning; Implicature; go further than descriptive adequacy, Indeterminacy; Logic; Logical devising generative grammars which Form; Sense/Reference; Signs and were embedded within a general Semiotics theory of human language. The Key Thinkers: Frege, Gottlob; theory in question was Chomsky’s Grice, H. P.; Peirce, C. S.; Quine, universal grammar*. W. V. O.; Russell, Bertrand; Wittgenstein, Ludwig Primary sources Ambiguity is the property of having Chomsky, N. (1957). Syntactic Structures. more than one linguistically encoded The Hague: Mouton. meaning. Lexical ambiguity is a prop- Chomsky, N. (1965). Aspects of the erty of words (or ) and Theory of Syntax. Cambridge MA: MIT is a property of Press. , clauses or sentences. The English ‘seal’ is an Further reading ambiguous referring, Sampson, G. (1979). ‘What was trans- among other things, to a particular formational grammar?’ Lingua 48: kind of sea creature or a device for 355–78. making sure an opening is fully Smith, N. (2004). Chomsky: Ideas and closed. The sentence ‘I’ve brought the Ideals. Cambridge: Cambridge Univer- seal’ is ambiguous because of the pres- sity Press. ence of this lexical item. A syntactically (or ‘structurally’) Philip Carr ambiguous expression is one which

3 ANALYTIC PHILOSOPHY can be associated with more than one ‘heap’ or ‘painful’ are inherently vague syntactic structure. For example, one since there is no way to say for certain reading of the expression ‘grey seals how much of a particular substance and walruses’ refers to seals which are we need to create a heap nor exactly grey and to walruses which are grey, how much discomfort is required for while on another reading it refers to an experience to be painful. seals which are grey and to walruses regardless of colour. Each of these Primary sources options is associated with a different Williamson, T. (1994). Vagueness. structure which can be represented by London: Routledge. bracketing: Further reading (a) [grey] [seals and walruses] Atlas, J. (1989). Philosophy Without Ambi- (b) [grey seals] and [walruses] guity. Oxford: . Hurford, J. R., B. Heasley, and M. B. A full understanding of human lan- Smith (2007). Semantics: A Course- guage requires an account of knowl- book. Second edition. Cambridge: edge about the ambiguity of linguistic Cambridge University Press. expressions, the processes by which Saeed, J. (2003). Semantics. Second edi- we understand containing tion. Oxford: Basil Blackwell. them and what happens when we fail to disambiguate them. Billy Clark Vagueness is the related but distinct property of failing to determine a unique value or interpretation. Vague- ness can be understood as a very gen- ANALYTIC eral, and fairly informal, notion or as a more systematically understood PHILOSOPHY property of concepts, propositions or utterances. Informally, all linguistic Analytic, or analytical, philosophy expressions are vague since they have can be any of many things. The term more than one possible interpretation. is used to describe philosophy that The expression ‘He has brought the proceeds via analysis, broadly by seal’, for example, could have several seeking to understand the composi- interpretations depending on who is tion of its matter (or concepts understood as the of ‘he’, of that subject matter) out of simple which sense of ‘seal’ is intended and (or simpler) components. In a promi- what is the referent of ‘the seal’. In nent but secondary sense, ‘analytic most contexts, this kind of vagueness philosophy’ applies to most philoso- is relatively easily resolved and a phy carried out in the mainstream of unique interpretation can be deter- Anglo-American university philoso- mined. On the more systematic sense, phy departments together with philos- vagueness is understood as the prop- ophy that bears a suitable family erty of not having a uniquely deter- resemblance to it: work within the minable meaning. Concepts such as so-called analytic tradition.

4 ANALYTIC PHILOSOPHY

See also: Analytic/Synthetic; (knowledge that X is the case) by Compositionality; Definite viewing it as composed of other con- Descriptions; Holism; cepts: , truth, and justification. Indeterminacy; ; Such analysis attempted to explain the Ordinary Language Philosophy; apparent inferential behaviour of the Sense/Reference; Truth Theories; , for example the apparent Truth Values that if someone knows that the Key Thinkers: Arnauld, Antoine; sun is shining then it may be inferred Austin, J. L.; Carnap, Rudolf; that they believe that the sun is shin- Frege, Gottlob; Moore, G. E.; ing and that their belief is both true Pierce, C. S.; Quine, W. V. O.; and justified. The success of the Russell, Bertrand; Ryle, Gilbert; method is dependent both upon the Strawson, P. F.; Wittgenstein, correctness of the analysis and upon Ludwig its capacity to deepen understanding by explaining the less well under- Although the two of ‘analytic stood – the whole – on the basis of philosophy’ described above are quite what is better understood, the simpler different, it is not an accident that the components and the effects of their same expression is used for both. The configuration in the whole. But the type of philosophy that now domi- analysis of the concept of knowledge nates the Anglo-American main- arguably fails on both counts. It has stream began its rise to prominence in been argued that the concept of the very late nineteenth century at the knowledge is a simple concept and so same time as the of analy- not composed of other concepts; and sis as the central method of that type that the concept of justification (and of philosophy. Analysis, broadly con- perhaps also the concept of belief) is strued, has continued to play a large no simpler, or easier to understand, role in much work in that tradition. than is the concept of knowledge. The method of analysis is the The method of analysis played a method of seeking to understand a role in early Greek philosophy (see, for subject matter by coming to under- example, ’s ) and reap- stand its composition. The aim is to peared to take a major part in the early understand the behaviour of a whole modern period, in the work of René by tracing its behaviour to the influ- Descartes and his followers. However, ences of its parts and their organisa- the method began gradually to lose its tion. In philosophy, the aim has often centrality during the rise of German been to understand the inferential , as philosophers responded to behaviour of a concept, to understand the work of by seeking what may be inferred from a judge- to develop grand systematic theories. ment involving the concept and what Although it was a central tool of some a judgement involving the concept other important thinkers in the nine- may be inferred from. For example, teenth century, most notably Franz analytic philosophy has attempted Brentano and C. S. Peirce, its rise to to understand the behaviour of the prominence as the central method concept of propositional knowledge in twentieth-century Anglo-American

5 ANALYTIC PHILOSOPHY philosophy was due mainly to the 1905). A very similar logic was devel- work of G. E. Moore and Bertrand oped in the medieval period, but the Russell. discovery of quantificational logic in In the very late nineteenth century, its modern form is usually attributed Moore began a revolt against German to (1879) and (inde- idealism. There were four main areas pendently) Peirce (1885). This logic of . First, Moore felt that - enables the systematic treatment of ism – according to which mind and the inferential behaviour of a very world are interdependent – was an large range of the statements that can erroneous metaphysical view and be made in (and so that, where possible, there should be a the expressible by the use of return to realism, according to which those statements) as well as the sharp there is an external world that is con- of complex positions and stituted independently of operations arguments. Of special importance of mind (except, of course, where the was Russell’s treatment of definite external world contains individual descriptions* in sentences of the form ). Second, Moore felt that grand ‘The F is G’. Using the example ‘The building should be suppressed present King of France is bald’, Rus- in favour of careful attention to detail sell represented sentences of that form and rigorous argumentation. Third, as conjoining three claims: he objected to what he saw as unnec- essary obfuscation in the of (1) There is at least one F (there is at German idealists. He felt that philo- least one present King of France). sophical theories and arguments for (2) There is at most one F (there is at those theories should be open to most one present King of France). objective assessment and so should be (3) Whatever is F is G (what/who is the stated as clearly and sharply as possi- present King of France is bald). ble. In Moore’s view, work in the sci- ences at the time offered better models In modern logical , the analy- of philosophical and sis becomes: this belief also motivated the fourth source of his dissent, his negative reac- (4) (᭚x) [Fx & (᭙y) (Fy ʛ x = y) & Gy]. tion to the Idealist suppression of the method of analysis. Calling for a By systematising a statement’s inferen- return to the method, Moore wrote tial (or, more broadly, logical) behav- that ‘a thing becomes intelligible first iour, the of the when it is analysed into its constituent statement in a favoured logical system concepts’ (Moore 1899: 182). shows the (or, perhaps, a) logical Inspired by Moore and enamoured form* of the statement. Russell’s in particular with the science of math- treatment provided a model on which ematics, Russell began systematically a definite description that fails to to develop the programme of analysis. apply to exactly one individual may He was helped by his discovery (or be meaningful, and so provided a rediscovery) of modern quantifica- potential solution to old problems tional logic (see, in particular, Russell about the functioning of talk that

6 ANALYTIC PHILOSOPHY

purports to make reference to partic- was the most ular non-existents. On Russell’s famous pupil of Russell and Moore. account, the fact that there is no pres- Wittgenstein’s Tractatus Logico- ent King of France makes the sentence Philosophicus (1922) further devel- ‘The King of France is bald’ false, oped the analysis of statements, and rather than meaningless, because it of representation more generally, in makes false the first clause in his the direction of . analysis. His treatment also made According to logical atomism, the especially evident that the logical form most fundamental level of representa- of a statement might not be obvious tion involves point-by-point connec- from its superficial form. However, tions between simple representational the space between logical and superfi- elements – representational atoms – cial form involved in Russell’s treat- and simple represented elements – ment of ‘The F is G’ via (4) is an represented atoms. This of artefact of Russell’s favoured logic. analysis was driven by the view that An alternative, though slightly less ‘A [i.e. the content of a perspicuous, treatment is given in (5): statement] has one and only one com- plete analysis’ (1922: 3.25). The (5) ([The x: Fx] (Gx).) work’s major conclusion was that there are two types of statement: Russell’s treatment of definite those that represent the world, and so descriptions showed that philosophi- can be either true or false depending cal progress could be made by dis- upon how the world in fact is – the cerning the (or a) logical form of a synthetic and falsehoods – and philosophically problematic range of those that either cannot be true or statements and that some philosophi- cannot be false – the logical or ana- cal disputes are usefully viewed as at lytic truths and falsehoods – that fail least in part concerning how best to to represent the world as being one represent the logical forms of state- way or another. Since genuinely ments involved in those disputes. philosophical statements, derived by Together with the new treatment of analysis, were taken to fall on the quantification more generally, this non-representational (analytic) side became a model for a of of this divide, they were taken to be approaches to philosophical prob- devoid of real content and to have a lems that involved attention to the role other than that of conveying forms of language used in the state- information. And since a core sense of ment of those problems. It supported meaningfulness was identified with the view that philosophical problems representational significance, such can arise due to the misleading super- statements were taken to be – in that ficial forms of the language we use, core sense – meaningless. (For an and it provided a model for how introduction to Wittgenstein’s early problems that arise in that way work, see Mounce 1989. For a dis- might be solved through uncovering cussion that relates logical atomism the true logical forms of the state- with Plato’s Theaetetus, see Burnyeat ments involved. 1990.)

7 ANALYTIC PHILOSOPHY

Members of the – empirical content with a more expan- including especially – sive view of . were inspired by Wittgenstein’s work According to the more expansive and sought to embed its central view, the analysis of statements was to themes in an approach to philoso- include the tracing of their roles phy, known as logical positivism*, within larger systems of language shaped by epistemological concerns. driven by careful attention to the way They replaced Wittgenstein’s distinc- those statements are actually used in tion between statements whose ordinary contexts. The more expan- truth value* depends upon worldly sive view therefore involved a partial and statements whose return to the sort of holistic app - truth value is fixed independently of roach involved in . such contingency with a distinction Wittgenstein’s work within the more between statements that admit of ver- expansive paradigm developed his ification or falsification on the basis of earlier view that the role of philo- experience and statements that cannot sophical analysis should be largely be so verified or falsified. The task of therapeutic: that it should serve the philosophy was taken to be the analy- removal of philosophical perplexity sis of statements into experientially by uncovering and excising the significant components, an analysis sources of confusion in the misleading that would either indicate precisely superficial forms of language. But the course of experience that would many other philosophers working verify or falsify the statement or show within the new paradigm of analysis, it to be beyond verification or falsifi- including J. L. Austin, , cation. In that way, philosophy would and P. F. Strawson – the so-called either show how a statement can be ‘ordinary language’ philosophers – assessed on the basis of scientific returned to the view that some philo- observation, or show the statement to sophical questions might be genuine be (in the Circle’s proprietary sense) and hoped that the analysis of lan- meaningless. Since the only properly guage would deliver answers to those cognitive activity was taken to be the questions. collection of observations, the pro- W. V. O. Quine took a different gramme of the Vienna Circle was path away from the Vienna Circle. shaped by the view that ‘what is left Quine was strongly influenced by the over for philosophy . . . is only a work of Carnap and retained the method: the method of logical analy- Circle’s view that scientific observa- sis’ (Carnap 1932: 77). tion is the only source of cognitive sig- In the early post-1945 period, many nificance. However, he thought that philosophers retained the Vienna Carnap had failed fully to draw out Circle’s animus towards traditional the consequences of that view. Quine but viewed its approach (1953, for example) argued that it to analysis as overly restrictive. They served to undermine the distinction sought to replace what they saw as between statements that can be veri- an empirically unmotivated fixation fied or falsified on the basis of experi- upon a very narrow conception of ence and those that cannot be so

8 ANALYTIC PHILOSOPHY verified or falsified: that is, the Circle’s undermine a distinction between the version of the distinction between analysis of statements and the empiri- statements that are analytic and those cal assessment of those statements. that are synthetic. Quine therefore And the rejection of determinate took the range of statements that are appears to undermine the up for scientific assessment to include assumption that there is such a thing statements in logic and mathematics. as the analysis of a statement. He took our total theory of the world The main effects of Quine’s work to form an interconnected web of have involved the further broadening statements that can only be assessed of the analytic horizon. Many con- as a whole on the basis of the range of temporary analytic philosophers who predictions it makes about the course are influenced by Quine’s rejection of of experience. Quine’s work involved an analytic-synthetic distinction allow in particular a rejection of the goal of their work to be shaped by the find- atomist analysis that formed the impe- ings of empirical science. Similarly, tus for the earliest work in modern many contemporary analytic philoso- analytic philosophy. phers allow that philosophy – even Quine’s rejection of the existence of when approached from the armchair – a category of analytic truths went can be a source of discovery about the hand in hand with a general rejection world. Thus the culmination, through of the philosophical utility of appeals Quine, of the Vienna Circle’s anti- to unreconstructed notions of linguis- metaphysical empiricism has led to tic meaning. In particular, Quine the reinstatement of metaphysics as a argued for the indeterminacy* of legitimate area of cognitive inquiry. translation, the claim that for any Finally, many contemporary philoso- translation from one language into phers have followed Quine in relegat- another (including translation from a ing the study of meaning from its language into itself), there will be central place in the analytic philoso- other that have equally phers’ armoury, and have sought a good empirical credentials. Famously more direct approach to answering he argued that evidence that an philosophical questions. expression is to be translated by the The postwar period of intense focus English expression ‘rabbit’ can be upon language and linguistic meaning equally good evidence that the expres- coincided with Frege’s work becoming sion is to be translated by the English widely available in translation and led expression ‘un-detached part of a to a re-evaluation of his place in the rabbit’ (Quine 1960: 1–79). When development of analytic philosophy. conjoined with Quine’s rejection of a In particular, principled distinction between ana- (1993) has claimed that it is criterial lytic and synthetic statements, accept- of modern analytic philosophy that it ance of Quine’s views has seemed to approaches the study of thought many to undermine the possibility of through a study of the way language is a philosophy based upon methods of used to express thought and that, analysis. The rejection of an analytic- from that perspective, Frege should be synthetic* distinction appears to assigned priority over Russell as its

9 ANALYTIC PHILOSOPHY progenitor. Dummett’s claim is con- (1959), Logical Positivism. Glencoe, IL: troversial for several , not least The Free Press. that it appears to place many contem- Dummett, M. A. E. (1993). Origins of porary philosophers who work within Analytical Philosophy. Cambridge MA: the analytic tradition outside the Harvard University Press. bounds of analytic philosophy proper, Frege, G. (1879). Begriffsschrift: eine der and it also seems to exclude the early arithmetischen nachgebildete Formel- work of Russell and Moore. Second, it sprache des reinen Denkens. Halle. is controversial that Frege would him- Reprinted in Frege (1972), Conceptual self count as an analytic Notation and Related Articles. Trans. by Dummett’s own standard, since and ed. T. W. Bynum. Oxford: Oxford he distrusted natural language and University Press. sought to construct artificial systems Frege, G. (1884). Die Grundlagen der better able to capture the nature of Arithmetik. Eine logisch mathematische thought. Third, although Frege’s work Untersuchung über den Begriff der now occupies a central place in the Zahl. Frege (1950) The Foundation of curriculum of analytic philosophy, Arithmetic. A Logico-mathematical there is little consensus concerning its Enquiry into the Concept of Number. precise role in shaping the initial J. L. Austin. Oxford: Basil Blackwell. development of the modern analytic Moore, G. E. (1899). ‘The nature of judge- tradition. Other philosophers, for ment’. Mind 8: 176–93. example, also played important roles Mounce, H. O. (1989). Wittgenstein’s in the development of Russell, Tractatus: An Introduction. Chicago: Wittgenstein and Carnap. What is un- University of Chicago Press. controversial, however, is that Frege’s Peirce, C. S. (1885). ‘On the algebra of work possesses many of the qualities logic; a contribution to the philosophy associated with analytic philosophy, of notation’. American Journal of including narrow focus, clarity, rigour Mathematics 7: 180–202. Reprinted in and depth, and exhibits all those qual- Peirce (1933), Collected Papers, ed. ities to a very high degree. To that C. Hartshorne and P. Weiss. Cambridge extent, no in analytic phi- MA: Harvard University Press. losophy would be complete that did Quine, W. V. (1953). ‘Two dogmas of not involve a careful study of his empiricism’. In From a Logical Point of work. View. Cambridge, MA: Harvard Uni- versity Press. Primary sources Quine, W. V. (1960). and . Burnyeat, M. F. (1990). The Theaetetus of Cambridge, MA: MIT Press. Plato. Trans. M. J. Levett. Indianapolis: Russell, B. (1903). The Principles of Math- Hackett. ematics. London: Allen & Unwin. Carnap, R. (1932). ‘Überwindung der Russell, B. (1905). ‘On denoting’. Mind Mataphysik durch logische Analyse 14: 479–93. Reprinted in Russell der Sprache’. Erkenntnis 2: 219–41. (1956), Logic and Knowledge, ed. Reprinted as ‘The elimination of meta- R. Marsh. London: Allen & Unwin. physics through logical analysis of lan- Wittgenstein, L. (1922). Tractatus Logico- guage’ trans. A. Pap, in A. J. Ayer (ed.) Philosophicus. Trans. C. K. Ogden and

10 ANALYTIC/SYNTHETIC

F. P. Ramsey. London: Routledge and its use by empiricists as a ‘dogma’. The Kegan Paul. reverberations of Quine’s criticism are still felt today. Further reading See also: Generative Semantics; Arnauld, A. and P. Nicole (1683). La Holism; Logical Positivism; Sense logique ou L’art de penser: contenant Data; Truth Value outre les regles communes, plusieurs Key Thinkers: Carnap, Rudolf; observations nouvelles, propres à Kant, Immanuel; Quine, W. V. O. former le jugement. Second edition. Paris: G. Desprez. Reprinted as The between analytically true Arnauld, A. and P. Nicole (1996). Logic and synthetically true statements is, or The Art of Thinking. Trans. and ed. on the face of it, an intuitive one, J. V. Buroker, K. Ameriks, and D. M. easily conveyed by either definition (as Clarke. Cambridge: Cambridge Univer- above) or example: sity Press. Descartes, R. (1701). Rules for the Direc- (1) ‘All bachelors are unmarried’ (ana- tion of the Mind. Amsterdam: P & J. lytically true). Blaeu. Reprinted in Descartes (1985), (2) ‘Bachelors have a higher average The Philosophical Writings of Descartes, disposable income than spinsters’ vol. 1. Trans. and ed. J. Cottingham, (synthetically true). R. Stoothof and D. Murdoch. Cam- bridge: Cambridge University Press. The terms ‘analytic’ and ‘synthetic’ as Hylton, P. (2007). Quine. London: Rout- they apply to true judgements were ledge. coined by Immanuel Kant (1781). But Hylton, P. (1993). Russell, Idealism and his actual definition, that a true judge- the Emergence of Analytic Philosophy. ment is analytic if its con- Oxford: Clarendon Press. cept is contained within its subject concept and is synthetic otherwise, is Guy Longworth now usually treated as having at most historical interest. Some philosophers argue that statements are analytically true if understanding them requires ANALYTIC/ acknowledgement of their truth. Others define the distinction in two SYNTHETIC stages: first, as it applies to either a belief, proposition, statement, or sen- An analytic truth is a statement (or tence, and second, as it applies, deriv- proposition) that is true solely by ing from this, to some or all other virtue of the meaning of its component members of this group. For example, words (or concepts). A synthetic truth if an analytically true proposition is a on the other hand is a truth deter- proposition true solely by virtue of the mined at least in part by features of the content of its component concepts, world outside of the language system. then an analytically true statement Almost as well known as the distinc- would be a statement that expresses tion is W. V. O. Quine’s description of an analytically true proposition.

11 ANALYTIC/SYNTHETIC

In the twentieth century the such as ‘photon’. Quine thought that analytic/synthetic distinction has been the analytic definitions needed to both important and controversial, bridge the gap between the more theo- particularly because of the weight retical, higher-level synthetic claims placed on it by logical positivists, and and the basic-level observation sen- Quine’s insistence that it could not tences were simply not available. bear that weight. Whether or not The problem, Quine said, was a Quine is right, there may be uses for lack of adequate guidance on how to the distinction in linguistics that are distinguish analytic from synthetic also compatible with his position. claims in practice. Kant’s account of Logical positivists such as Rudolf the distinction, with its talk of con- Carnap had an ambitious goal: to cepts being ‘contained in’ other con- construct a language suited to the cepts, is too metaphorical to be needs of science. Anyone with an helpful here. Similarly, to say that ana- understanding of this language would lytic truths are truths of meaning, or be able to extract empirical conditions truths by definition, leaves us needing from any sentence of it used to express a way of distinguishing truths of a claim. Any properly scientific claim meaning from deeply held convic- could thus be made to wear its confir- tions. Deferring to a merely mation conditions on its sleeve, so to passes the buck up to dictionary writ- speak, making it straightforwardly ers. To fall back on the thought that susceptible to empirical confirmation truths are analytic just in case they are or refutation. logical truths, as Carnap himself did, Quine argued that the logical is to forget that many supposedly ana- positivism* struggled to achieve this lytic sentences, including (1) above, goal of a scientific language was its are non-logically true. Perhaps, Quine reliance on an unworkable distinction conjectures, analytic truths are sup- between analytical and synthetic posed to be stipulations, true by lin- sentences. Synthetic sentences were guistic convention. Against this he supposed to include basic-level obser- claims that we lack criteria for decid- vation sentences such as ‘There is red- ing whether a particular assertion ness to the left of my visual field’ as should be classified as true by conven- well as higher-level theoretical sen- tion rather than as an empirical con- tences such as ‘Photons have wave- jecture. Is the assertion that photons particle duality’, with many levels in have wave-particle duality a stipula- between. Higher-level synthetic sen- tive truth about the word ‘photon’ or tences had to mean the same as (that a substantial empirical assertion? No is, be analytically equivalent to) com- pre-established for the plex basic-level observation sentences, of physics exists that is since the latter expressed the former’s uncommitted to any particular empir- empirical commitments. Logical posi- ical outlook. tivists needed all synthetic sentences to For all that has been said so far, log- reduce to complex basic-level observa- ical positivists could cling to the tion sentences by the repeated applica- thought that the analytic sentences are tion of definitions of theoretical terms simply those that link theoretical

12 ANALYTIC/SYNTHETIC

sentences to their confirmation condi- ‘know’. One motive for this concep- tions, plus those that enable this link tion of their role is that philosophers to be derived. It is up to us to identify realise they are not . What those links. No one ever claimed this they provide must therefore comple- would be easy. Quine’s final objection ment rather than compete with scien- to this faith has, rightly or wrongly, tific knowledge. Another is immodest: probably been the most influential. He they aim to understand the world at an observes that evidence for or against a a priori level, that is, at a level imper- particular clause in some scientific vious to the contingency of experience. theory can, in principle, come from If Quine is right, this model of phi- anywhere. Evidence for a particular losophy may be in trouble. According claim in syntax, for example, may pre- to him, there are no analytic truths in suppose a particular ; evi- any interesting sense. Some support- dence for this morphology may ers of the model insist they are not depend in turn on an with a interested in the analysis of scientific theory of colour vision tied to specific concepts. But they need to defend an assumptions about photons; and so interest in the analyses they offer. on. Given this evidential holism, stat- After all, ‘folk’ ways of thinking are ing the confirmation conditions for often inferior to scientific ones. Good higher-level theoretical claims would analyses of concepts such as ‘causa- require making impossibly detailed tion’ or ‘knowledge’, if Quine is right, and disjunctive reference to all of sci- may be susceptible to improvement as ence. Developing and applying a science develops. Advances in the cog- language that would allow these state- nitive sciences, for example, have ments to be formulated, if possible at shaped our understanding of what all, would be no easier than doing sci- knowledge is. Whereas ‘Knowledge ence as it is done already. must be conscious to the knower’ Interest in and scepticism towards would once have been regarded as the analytic/synthetic distinction thus analytic; it is now commonly regarded grew out of a concern for how scien- as not even true, let alone analytically tific methodology might be optimised. true. Disputes over the status of ana- But the distinction is also central to a lytic truths have, then, broadened out popular conception of what sets phi- into a split between those who see the losophy apart from other disciplines philosophy of x and the science of x as and also to an approach to semantics distinct in both subject matter and conceived as a part of psychology. methodology, and those who see only Many philosophers follow Socrates continuity. in conceiving of their role as being to A very different approach to ana- analyse concepts. There are more or lytic truth emerges from within lin- less sophisticated implementations of guistics itself. It may be that to call a this thought, but a typical illustration sentence analytically true is merely to is the giving of necessary and sufficient give voice to an accidental feature of conditions that approximately match, our language faculty. Certain appar- or perhaps improve on, those for ently analytic entailments, such as the proper application of the word that killing requires causing to die,

13 ARTIFICIAL INTELLIGENCE may be grounded in the semantic fea- Further reading tures of our concepts and/or lexical Boghossian, P. (1996). ‘Analyticity recon- items (see generative semantics*). sidered’. Noûs 30(3): 360–91. This approach to semantics has a che- Jackendoff, R. (1990). Semantic Struc- quered , in part because of the tures. Cambridge MA: MIT Press. difficulty of distinguishing deeply- Jackson, F. (2000). From Metaphysics to held beliefs about something from Ethics: a Defence of Conceptual Analy- semantic features – the same problem sis. Oxford: Oxford University Press. Quine drew attention to in a different guise. That said, appealing to the Alex Barber notion of analyticity in this context would not commit one to logical pos- itivism*. Discovering semantic fea- tures could be interesting for what it ARTIFICIAL tells us about the structure and devel- opment of the human mind, not INTELLIGENCE because of any help it offers in inves- tigating the world outside the mind. The ability of machines to think for The intuitive distinction between the themselves. More generally the sci- analytic and the synthetic has been ence dedicated to exploring and devel- drawn on by philosophers of science in oping the ability of technology – most an attempt to understand how science commonly in the form of a software might be improved, and by other program – to think and/or act ration- philosophers to explain how what they ally or similarly to human capability. are doing complements science. Inter- As it relates to linguistics, the flourish- est in analyticity has also emerged out ing field of artificial intelligence aims of the mentalist approach favoured to develop systems that can under- by some semanticists. The challenge stand, process, and even generate lan- facing all who invoke the notion of guage as well as a human can. analyticity is to ascertain how to dis- See also: Generative Semantics; tinguish in practice between some- Intentionality; Logic; Speech Act thing’s being analytically true and its Theory being profoundly but synthetically Key Thinkers: Chomsky, Noam; true. Montague, Richard; Russell, Bertrand; Searle, John Primary sources Carnap, R. (1928). Der logische Aufbau der The study of language often leads to Welt. Berlin-Schlachtensee: Weltkreis- questions over what defines intelli- Verlag. gence, as exhibited via language com- Kant, I. (1781/1933). A Critique of Pure petence and performance. Artificial Reason, trans. N. Kemp Smith. Second intelligence (AI) provides us with pre- edition. London: Macmillan. cious opportunities to apply what we Quine, W. V. O. (1951). ‘Two dogmas of know about language toward useful empiricism’. Philosophical Review 60: and often valuable purposes, whether 20–43. to serve as a of concept, to

14 ARTIFICIAL INTELLIGENCE

innovate, or to fulfil a specific need. who is interacting with the computer, it The roots of AI extend to the origins would seem that the computer under- of computers and even logic*. From stands Chinese. However, if we replace its foundations in the works of René the computer with a human who does Descartes and Blaise Pascal, through not know Chinese but has been trained the nineteenth-century innovations of to apply rules from a book against Charles Babbage and Ada Lovelace, Chinese language input to arrive at AI flourished in the twentieth century, Chinese language output, we would directly paralleling advancements in never say that the human knows both linguistics and , Chinese. That is, the simple act of through ’s formal applying predetermined rules (as a logic, the Turing test, and contribu- computer does during a Turing test) tions by such innovators as Joseph does not constitute thinking, nor does Weizenbaum and Hans Kamp, to it represent . Neverthe- a few. Not without its share of less, variations of this approach are criticism, AI has resulted in many commonly used to evaluate AI systems practical applications for everyday against a standard of human ability. life, including sys- In order to process language as well tems, software, as a human, an AI system must have and information retrieval systems. the ability both to understand and to Humans are capable of many tasks, generate language. Its means of with language being among the most accomplishing this may differ from complex. Language allows us to com- human mental processes, but the municate abstract thoughts through an effect should be equivalent. At a high infinite combination of words and level, natural language understanding meanings. With this in mind, it is not in an AI system entails accepting surprising that Alan Turing designed language input (whether through his now-famous Turing test around speech recognition or optical charac- language. The Turing test is a way to ter recognition), analysing it (through judge the intelligence of an AI system. and semantic interpretation), It involves a human subject, who car- and disambiguating the message (at ries on a conversation with two enti- the lexical, syntactic, semantic and ties, one being a human and the other pragmatic levels, calling upon outside a computer. If the subject cannot dis- knowledge of the world, as needed). tinguish between the human and the Likewise, natural language generation computer, the computer is said to have requires that the AI system determine passed the test. The best-known coun- a reasonable response, organise its terpoint to the Turing test is John structure, form comprehensible and Searle’s Chinese Room . We natural-sounding sentences, and out- begin this scenario with a computer put them as written text or synthe- that has been programmed to return a sised speech. AI forms an of particular set of Chinese characters as traditional computer science, with a result of applying a set of rules to linguistic AI being commonly cate- another set of Chinese characters gorised as computational linguistics received as input. To a Chinese speaker or natural language processing. These

15 ARTIFICIAL INTELLIGENCE disciplines call upon generative gram- tion as economically and as efficiently mar and linguistics in order to apply as possible. Statistical AI systems offer what we know about language to promising alternatives within natural technology that may be capable of language processing, incorporating natural language understanding and theory, Bayesian networks, generation. Linguists involved in AI Markov models, and neural networks. maintain a keen interest in advance- The most effective examples of lan- ments within the general field of lin- guage AI, in fact, are hybrid models, guistics, as their goal is to equip combining linguistic knowledge with computers with a comparable sense of statistical . the linguistic knowledge and ability Just as AI draws upon developments we humans innately have, such as in the field of linguistics, it presents grammaticality, phrase structure, gov- valuable opportunities for linguists to ernment and and so on. test their theories in a computational Critics of AI argue that computers environment. An AI system ultimately are incapable of achieving the level of demonstrates cognitive ability that humans exhibit. with its software resources comprising To this point, Searle made a distinction its . Language- between strong AI and weak AI. Strong based AI has many practical applica- AI describes a system that is able to tions in business and leisure. We reason and perform as well as – or encounter AI, for example, in busi- better than – a human. A strong AI ness, where it may be used to forecast system is conscious of itself and oper- market conditions, to analyse large ates beyond a strict adherence to pre- amounts of data, and to save time oth- programmed rules and algorithms. erwise spent on tedious tasks. In our Most AI systems, however, are better homes, we see AI in our appliances, in classified as weak AI, capable of only a our automobiles, and on our personal limited range of skills and lacking in the computers, and we may even play ability to ‘think’ as a human does. chess and other complex games against Hubert Dreyfus, one of the most out- AI systems. Practical uses of AI are all spoken critics of AI, states that com- around us, originating in linguistic and puters simply cannot accomplish the AI circles and ultimately being distrib- complex tasks typically associated with uted around the world for an abun- strong AI. For example, an AI system dance of practical applications. can never learn to love, no matter how AI represents an application of lin- it is programmed. System design is guistic knowledge, and it serves as a rarely flawless and the mere prospect of driving force to learn more, not only developing a fully capable AI program about language, but also about human is downright intractable. Difficulties knowledge and the nature of intelli- that face AI systems encompass the gence itself. It encourages us to revisit and creativity inherent in lan- questions over what defines intelli- guage, such as differences in grammat- gence, what sets apart human capabil- icality across and a heavy use ity, and what we can programme of metaphor*, with the prevailing or train a computer to do. Fictional constraint that a system should func- portrayals of AI systems have created

16 BEHAVIOURISM rather humourous public misconcep- Woods, William A. (1970). ‘Transition tions, from robot uprisings to world network grammars for natural language domination, but also some ideas for analysis’. of the ACM valuable applications of AI, many of Vol. 13: 591–606. which require linguistic AI in particu- lar. In some cases, such as search Further reading engines and some speech recognition Charniak, Eugene (1993). Statistical Lan- systems, language-related AI has guage Learning. Cambridge, MA: MIT become a welcome part of daily life. It Press. brings together many of the things we Hutchins, W. John and Harold L. Somers know about language and tests these (1992). An Introduction to Machine theories, sometimes even in the form of Translation. San Diego, CA: Academic annual competitions. Such challenges Press. lead to innovative approaches and Jurafsky, Daniel and James H. Martin services that would otherwise be unat- (2000). Speech and Language Process- tainable through mere human effort ing. Upper Saddle River, NJ: Prentice and ability. There are endless possibili- Hall. ties for practical applications of AI, Rabiner, Lawrence R. and Biing-Hwang especially where there is consumer Juang (1993). Fundamentals of Speech demand, and it is an important area in Recognition. Upper Saddle River, NJ: which businesses, technological inno- Prentice Hall. vators, and philosophers of language Russell, Stuart and Peter Norvig (1995). can work together. Artificial Intelligence: A Modern Approach. Upper Saddle River, NJ: Primary sources Prentice Hall. Dreyfus, H. (1972). What Computers Can’t Do: The Limits of Artificial Intel- Jennifer A. Baldwin ligence. Cambridge, MA: MIT Press. Kamp, H. (1981). ‘A theory of truth and semantic representation’. In Groe- nendijk, J. A. G., T. M. V. Janssen and M. BEHAVIOURISM B. J. Stokhof (eds), Formal Methods in the Study of Language. Vol. 1: 277–322. The study of behaviour without Schank, Roger C. (1972). ‘Conceptual appeal to mental states. Behaviourism dependency: a theory of natural lan- had a profound effect on linguistics guage understanding’. Cognitive Psy- (especially American structuralist chology Vol. 3: 532–631. linguistics) and philosophy (logical Turing, Alan (1950). ‘Computing machin- behaviourism), but was effectively dis- ery and intelligence’. Mind Vol. 59: placed by the revolu- 433–60. tion in the 1960s. Wilks, Yorick (1973). ‘The Stanford Machine Translation Project’. In See also: Cognitivism; R. Rustin (ed.), Natural Language Empiricism/Rationalism; Processing. New York: Algorithmics Innateness; Mentalism; Press. 243–90. Psychoanalysis; Structuralism;

17 BEHAVIOURISM

Transformational-Generative involuntary behaviour by repeatedly Grammar pairing a neutral stimulus (for exam- Key Thinkers: Bloomfield, ple, a bell sound that does not natu- Leonard; Chomsky, Noam; rally elicit a given response) with an Hockett, Charles; Quine, W. V. O.; unconditioned stimulus (for example, Ryle, Gilbert; Skinner, B. F.; food, which naturally produces a Wittgenstein, Ludwig given response such as salivation in dogs). After repeated pairing, the neu- Behaviourism was the dominant para- tral (now conditioned) stimulus digm in psychology for much of the becomes associated with the (now first half of the twentieth century and conditioned) response: a dog that has until the late 1950s, especially in the undergone this type of training will United States. Despite important dif- salivate when hearing a bell sound, ferences between different schools of even in the absence of food. behaviourism, all behaviourists shared Classical behaviourism had a pro- the conviction that behaviour was the found effect on American structuralist only legitimate object of investigation linguistics, through the work of lin- in psychology, and that behaviour guists such as Leonard Bloomfield, cannot be explained by appealing to and his student Charles Hockett. hypothetical internal states, such as Bloomfield espoused the behaviourist consciousness. As such, behaviourism credo that only behaviour can be the follows in the tradition of British object of scientific study. He advo- empiricism*. cated for the study of linguistics what J. B. Watson is credited as the Watson had advocated for the study founding father of behaviourism. In an of psychology: that it should be article that has come to be known as approached as a natural science, with the ‘behaviourist manifesto’, Watson a firm emphasis on empirical data and (1913) set out the main principles of rigorous description. In his enor- classical behaviourism: psychology is mously influential Language (1933), a natural science; the only legitimate he provided an account of meaning object of psychological study is overt which adapted the classical condition- behaviour; internal mental states and ing to the interpretation of are not behaviour and acts of speech. Language provides a cannot be the causes of behaviour; secondary stimulus-response mecha- behaviour is environmentally deter- nism, pairing a stimulus from the mined; the same principles are applied speaker with a response from the lis- to the study of both animal and human tener. He illustrated this in the now behaviour, which do not differ quali- famous ‘Jack and Jill’ story, in which tatively. In later work, he acknowl- Jill experiences hunger (speaker stim- edged the importance of Ivan Pavlov’s ulus) and asks Jack to fetch an apple work on what came to be known as for her (speaker response). Upon hear- classical conditioning and incorpo- ing the request (hearer stimulus), Jack rated it into the behaviourist agenda. fetches the apple (hearer response). Classical conditioning is a form of Although Bloomfield himself admit- associative learning that modifies ted the difficulties of such an analysis

18 BEHAVIOURISM of meaning, his commitment to complex behaviours, but language behaviourist principles in the study of was not one of them. B. F. Skinner language guided research in theoreti- undertook to fill this gap in Verbal cal and applied linguistics up until the Behavior (1957). In this work, which in the late 1950s took more than twenty years to com- and early 1960s. plete, Skinner asserted that verbal In philosophy and the philosophy behaviour is a complex habit arising of language, behaviourism became as a result of the interaction between known as logical behaviourism. Logi- a speaker’s environment and his/her cal behaviourism, unlike classical past learning history. In his research behaviourism, did not deny the valid- with animals, Skinner had shown how ity of inquiring into mental states. voluntary behaviour could be modi- According to logical behaviourism, fied through operant conditioning: a however, all meaningful – that is veri- food-deprived (‘motivating opera- fiable – statements about mental tion’) rat is placed in a cage and a light phenomena are translatable into is turned on (the ‘discriminative stim- statements which refer to physical ulus’) to signal that food can be concepts only. For instance, to be obtained if the rat performs the appro- angry includes changes in tone of priate action. The rat may initially voice, facial expression, and perform a number of actions, such as so on. For the logical behaviourist, all running around the cage, scratching these physical properties are not just the cage bars, or pressing a lever in the manifestations of ‘anger’, but defini- cage. If by performing one of these tions of the concept ‘anger’. In other actions – for instance, pressing the words, logical behaviourists were lever (‘response’) – the rat obtains committed to the notion that mental food (‘reinforcer’), this particular phenomena can be reduced to behav- behaviour is reinforced. Skinner ioural dispositions. Logical behav- claimed that the same type of func- iourism is mostly linked to the work tional scheme accounts for verbal of Gilbert Ryle, and the later work behaviour as well, with only minor of Ludwig Wittgenstein, but other modifications. For instance, if you are philosophers with behaviourist affilia- hungry (‘motivating operation’), and tions include Rudolf Carnap and later your mother is at home (‘discriminat- W. V. O. Quine. The main criticisms ing stimulus’), you may say ‘Give levied against logical behaviourism me some food’ (‘response’). If your are that several mental states are not mother gives you some food (‘rein- reducible to behavioural dispositions forcer’), she rewards your verbal (for instance, ‘belief’ cannot be under- behaviour and you are more likely to stood independently of ‘desire to say ‘Give me some food’ next time you verify the belief or act on the belief’) are hungry and she is in the house. and that it has nothing to say about In a review of that conceptual relations between mental became more influential than the concepts themselves. book itself, (1959) By the early 1950s behaviourist launched an all-out attack on Skinner’s psychology had tackled a number of approach to language as behaviour. He

19 BEHAVIOURISM claimed that human behaviour and early 1960s, it still survives in some more specifically language are infinitely applied settings (for instance in the more complex than the types of animal treatment of autism) and in certain behaviour Skinner had previously stud- strands of animal learning theory. The ied, and thus principles of animal study of the mind/brain and their role behaviour cannot be applied to verbal in explaining behaviour are now at behaviour. He showed that many of the the forefront of scientific investiga- concepts Skinner used, such as stimu- tion, but the legacy of behaviourism lus, response, and reinforcement, were has not disappeared completely: hopelessly vague and had no explana- behaviourism, for instance, shares tory value in accounting for linguistic with contemporary connectionist knowledge. He argued that Skinner’s approaches to psychology and lan- account is also a poor model of lan- guage the emphasis on associative guage acquisition, since it fails to learning and on the role of experience account for the speed, efficiency and on learning. uniformity with which children acquire language. According to Chomsky, the Primary sources basic of Bloomfield, L. (1933). Language. New point to the existence of an innate pre- York: Henry Holt. disposition to acquire language. This Chomsky, N. (1959). ‘Review of Verbal review, along with work carried out Behavior’. Language 31: 26–58. in computer science, was instrumental MacCorquodale, K. (1970). ‘On Chom- in the ‘demise’ of behaviourism and in sky’s Review of Skinner’s Verbal Behav- effecting the move from studying ior’. Journal of the Experimental behaviour to studying mental processes Analysis of Behavior 13: 83–99. using a computer metaphor (the cogni- Ryle, G. (1949). The Concept of Mind. tive revolution). London: Hutchinson. A direct response to Chomsky’s crit- Skinner, B. F. (1957). Verbal Behavior. icisms never came from Skinner him- New York: Appleton-Century-Crofts. self. Kenneth MacCorquodale, one of Watson, J. B. (1913). ‘Psychology as the Skinner’s former students, published a Behaviourist Views it’. Psychological response (1970) more than ten years Review 20: 158–77. after the review appeared, in which he showed that Chomsky had misunder- Further reading stood several of the points Skinner was Baum, W. M. (2003). Understanding making and that a lot of Chomsky’s Behaviorism: Science, Behavior, and criticisms were not actually criticisms . New York: Harper Collins. of Skinner’s work but of other behav- Skinner, B. F. (1953). Science and Human iourist work popular at the time. This Behavior. New York: Macmillan. response, however, had a very limited Skinner, B. F. (1986). ‘The of impact, partly because a paradigm shift Verbal Behavior’. Journal of the had already taken place and behav- Experimental Analysis of Behavior 45: iourism had given way to cognitivism*. 115–22. Although behaviourism ceased to be in the mainstream of science in the Stavroula-Thaleia Kousta

20 COGNITIVISM significant role to metaphoricity and . The centrality of meaning in cogni- Cognitivism can be defined as the tive grammar is one of the main belief that mediates percep- aspects of the generative semantics tion, unlike in behaviourism. In this heritage in . broad sense, cognitivism is synony- Meaning in CL is taken to be funda- mous with mentalism*. But while mental and directly mapped onto mentalism has found expression in phonological structures. This entails generative grammar, cognitive linguis- the rejection of modularity (the belief tics was initiated as an alternative to that there exist encapsulated modules the generative-transformational para- that process phonological or morpho- digm, with significant influences from logical information without reference the generative semantics* approach, to other levels of language) and the Gestalt psychology and the psychol- adoption of an encyclopedic model of ogy of prototypes. Its main earliest the , that is a belief that all proponents were Ronald Langacker, sorts of encyclopedic information and Len Talmy in the about lexical items are part and parcel early 1980s, but the approach soon of their meaning and correspondingly gathered a broad following in Europe no systematic distinction can be and the United States. drawn between lexical (word-related) and encyclopedic knowledge (knowl- See also: Generative Semantics; edge about the world). Typically, lexi- Mentalism; Metaphor; Prototype; cal knowledge is organised in frames, Transformational-Generative organised complexes of information Grammar, relating to a given concept. Key Thinkers: Sapir, Edward; The centrality of meaning is not Whorf, Benjamin Lee merely axiomatic, but informs the The basic organisations of cognitive entire CL edifice. CL focuses both linguistics (CL) are semantic-phono- on lexical meaning and on grammati- logical mappings and in this respect cal meaning. Generative grammar CL is like all semiotic approaches to ignored the lexicon, seen as a reposi- communication. The significant differ- tory of exceptions, and assumed that ence is that CL is squarely focused on grammar was a formal structure, that meaning and rejects any non-semantic is one without meaning. Conversely, component to the grammar (for exam- grammatical structure, according to ple, empty categories). Other charac- CL, is the result of a process of teristics that distinguish CL from other schematisation, that is of non-cognitive approaches to grammar constructional meaning. So, in effect, are prototypicality, schematicity, and CL claims that all language is mean- perspectivism, as well as more gener- ing, including grammar, and that ally, the rejection of modularity and grammatical meaning is very abstract consequently the adoption of an ency- schematic meaning. Relatedly, CL is clopedic semantics, and the belief that also very keen on constructional cognition is embodied and affords a mean ing, that is the fact that some

21 COGNITIVISM constructions, such as ‘X let alone Y’ Conceptualisation is a central con- have a meaning that cannot be pre- cept in CL. As we have seen, meaning dicted from the sum of its parts. Sig- is the object of CL. Meaning can be nificantly, parts of this meaning would defined as the ‘content’ of an idea, traditionally be considered pragmatic concept, and so on, and its construal, (hence reiterating CL’s rejection of that is the way in which the speaker modularity). looks at the content, the perspective in The notion of prototypicality, elab- which it is seen. For example, the sen- orated by in the 1970s, tences ‘Mary kicked the ball’ and ‘The and originally applied to categories ball was kicked by Mary’ clearly (such as ‘bird’ or ‘furniture’), is also describe the same factual situation applied to grammatical categories, in (content) but with two different per- a way recalling generative-semanticist spectives, one that privileges the active John Ross’s notion of ‘squish’. This role of Mary and another, the passive means that some , to take a clear construction, that privileges the recip- example, are better examples of ient of the action. The same reasoning than others (for example, action applies to ‘downhill/uphill’, which verbs, such as ‘kick’, which involve depends on where the speaker, so to the transfer of force from one entity to speak, puts him/herself. another, are better examples of verbs Conceptualisation is said to be than state-verbs, such as ‘own’). embodied and to play an active role in CL shares with generative seman- the construal of . The embodied tics the programmatic rejection of the nature of conceptualisation refers to core/periphery distinction. In genera- the fact that many metaphorical tive grammar, it is common practice to expressions are based on common focus on the regularities of grammar human characteristics. For example, a (core), rejecting exceptions to the common metaphor is that UP is periphery, where they are accounted GOOD and DOWN is BAD. This is for by pragmatic (the waste-basket seen to be rooted in the upright stance theory of pragmatics) or other factors. of the human body and the elevated CL denies the of this distinc- position of the human . Thus, far tion, and seeks to explain all phenom- from being based on abstract cate- ena of language using the same gories, conceptualisation is based on cognitive principles and hence shows shared physical characteristics of a particular interest in the seeming humans. irregular patters found, for example, The role of language and its cate- in constructions. Moreover, CL incor- gorisation in our of reality porates pragmatic and functional are a significant philosophical tenet of within its conceptual CL, which rejects the idea that reality apparatus, hence rendering a core/ is simply mirrored in language and/or periphery distinction impractical. mental representations, and instead This parallels CL’s rejection of modu- sees an active role of linguistic cate- larity, which presupposes that the gorisation in shaping the way the same principles and mechanisms world is represented linguistically. The apply at all levels of language. embodied nature of conceptualisation

22 COGNITIVISM extends to these factors as well, with show is over’). Blending is a more the assumption that knowledge is recent theory of how mental spaces organised in ways that are influenced can interact (blend) and result in a by the needs, goals, and cultural new mental space with emergent fea- assumptions of the speakers. tures (features which are not present Other concepts closely associated in either of the original spaces in the with CL are mental spaces, which can blend). Blending has been applied to be defined as a conceptualisation metaphors, especially novel meta- informed by the relevant frames by a phors, and to an increasing variety of given speaker of a given situation, phenomena. , and so on. Mental spaces in CL It should be stressed that CL shares are similar to possible worlds in formal with some theories of language such semantics. Mental spaces have been as natural phonology and morphol- widely used, for example, to explain ogy the use of iconicity (the fact that a counterfactuals, but have also been linguistic form reflects the form of the extended to the , referent it describes) as an explana- , and discourse phenomena. tory concept. For example, the com- The study of metaphor* has also parative and superlative forms of the been a central interest of CL. Most adjective are often longer than the linguistic approaches see metaphori- positive form: ‘pretty – prettier – pret- cal meaning as a trope – a figure of tiest’. Relatedly, CL has shown since speech – and would relegate it to the its inception a fondness for visual periphery of language. CL conversely schematic representations that use argues that most meaning is meta - iconic means, such as bolded lines to phorical in nature. Metaphors are indicate focus, that have considerable seen as mappings between two weight in the explication of the con- domains; however, the significant dif- cepts, at least intuitively and pedagog- ference between CL accounts of ically. CL has also always shown a metaphors is that CL sees meta - very distinct interest in space and its phors as grounded in metaphorical linguistic representation: witness the schemata of great generality, such as intensive study of prepositions which ‘argument is war’, which underlie has been one of the great contribu- such metaphors as ‘he shot down tions of CL to grammatical analysis, my argument’ and ‘Mary won the and also the conceptualisation of debate’. Furthermore, CL sees mean- many events as consisting of a figure ing dynamically expanding from lit- and a ground, a trajectory or a land- eral meanings to other metaphorical mark, for example. ones. For example, the meaning of the Finally CL asserts that a good theory preposition ‘over’ is literally related of language is a in to spatial orientation (an object is use. This is mostly a programmatic above another), but that meaning is claim, but recent developments of CL then extended metaphorically to have emphasised the use of corpora* motion over an object, and eventually of naturally occurring utterances, to being past the object, and hence to analyses of conversational data, and an the meaning of finished action (‘the increased attention toward pragmatic

23 phenomena and sociolinguistic varia- COMPOSITIONALITY tion at large.

Primary sources The notion that the meaning of a com- Lakoff, George and Mark Johnson plex expression is a function of the (1980). Metaphors We Live By. meanings of its parts. The ‘Principle Chicago: University of Chicago Press. of Compositionality’ has played an Langacker, Ronald W. (1987). Founda- important role in the work of a large tions of , Volume I, number of philosophers and semanti- Theoretical Prerequisites. Stanford, CA: cists including Gottlob Frege, Donald Stanford University Press. Davidson and . Langacker, Ronald W. (1991). Founda- See also: Conventional Meaning; tions of Cognitive Grammar, Volume II, Language of Thought; Logic; Descriptive Application. Stanford, CA: Logical Form; Mentalism; Model Stanford University Press. Theoretic Semantics; Possible Rosch, Eleanor (1973). ‘Natural Cate- World Semantics; Propositions; gories’, 4: 328–80. Truth Theories Talmy, Leonard (2003). Toward a Cogni- Key Thinkers: Davidson, Donald; tive Semantics – Volume 1: Concept Frege, Gottlob; Montague, Structuring Systems. Cambridge, MA: Richard; Quine, W. V. O. MIT Press. Talmy, Leonard (2003). Toward a Cogni- The Principle of Compositionality is a tive Semantics – Volume 2: Typology guiding principle for many semanti- and Process in Concept Structuring. cists and philosophers, notably Frege, Cambridge, MA: MIT Press. Davidson and Montague. It is some- times referred to as ‘Frege’s Principle’, Further reading although not everyone agrees that Croft, William and D. Alan Cruse (2004). the idea originated with him. Frege Cognitive Linguistics. Cambridge: (1892) states it as follows: ‘The mean- Cambridge University Press. ing of a complex expression is a func- Evans, Vyvyan and Melanie Green (2006). tion of the meanings of its parts and Cognitive Linguistics. An Introduction. their syntactic mode of combination’. Edinburgh: Edinburgh University Press Perhaps the main reason that the Geeraerts, Dirk (ed.) (2006). Cognitive principle is seen as important is the Linguistics: Basic Readings. Berlin/New role it plays in explaining the creativ- York: Mouton de Gruyter. ity* and systematicity of language. Geeraerts, Dirk and Hubert Cuykens (eds) We can explain how we are able to (2007). The Oxford Handbook of Cog- produce and understand an infinite nitive Linguistics. Oxford: Oxford Uni- number of utterances if we assume versity Press. that we know the meanings of indi- Ungerer, Friedrich and Hans-Jörg Schmid vidual expressions and how to com- (1996). An Introduction to Cognitive bine them into larger units. Linguistics. London: Longman. It follows from this principle that the contribution of an expression, Salvatore Attardo say the lexical item ‘red’, to a more

24 CONNOTATION/DENOTATION

complex expression should be the Further reading same in all cases. The contribution of Davidson, D. and G. Harman (eds) ‘red’ should be the same in the noun (1972). Semantics of Natural Language. phrase ‘red wool’ as in ‘red cotton’ Dordrecht: Reidel. and in any other . W.V.O. Dowty, D. R., R. E. Wall. and S. Peters Quine (1960) pointed out some prob- (1981). Introduction to Montague lems with this assumption. For exam- Semantics. Dordrecht: Kluwer. ple, the contribution of ‘red’ in ‘red Thomason, R. (ed.) (1974). Formal Phi- apple’ is not the same as the contribu- losophy: Selected Papers of Richard tion of ‘pink’ in ‘pink grapefruit’, since Montague. New Haven, CT: Yale Uni- a red apple is usually understood to be versity Press. an apple with (mainly) red skin on the outside while a pink grapefruit is usu- Billy Clark ally understood to be a grapefruit with (mainly) pink flesh on the inside. * reports (utterances which contain a statement CONNOTATION/ about an individual’s attitude to a proposition) provide a further test for DENOTATION the principle. For example, we can believe all three of the following: Connotation, from the medieval verb con-noto, refers to an (1) Chris thinks his next-door neigh- implied or accompanying feature, as bour is considerate. the comitative prefix con suggests. (2) Chris thinks the person who reversed Denotation etymologically derives into his bicycle is inconsiderate. from the postclassical Latin com- (3) Chris’s next-door neighbour is the pound verb de-noto which conveys person who reversed into his bicycle. the idea of singling out an entity by way of distinctive features. Despite these problems, most theo- See also: Sense/Reference; Signs rists prefer to retain the Principle and Semiotics of Compositionality rather than Key Thinkers: ; Hjelmslev, attempt ing to develop a new account Louis; Mill, J. S.; Saussure, of the productivity and systematicity Ferdinand de of linguistic knowledge. The notion of denotation can be Primary sources traced back to Aristotle’s classifica- Frege, G. (1892). ‘On sense and meaning’. tion of denominative names (Cate- In and Max Black (eds) gories I) where he labelled ‘paronyma’ (1980), Translations from the Philo- (denominative) the names of entities sophical Writings of Gottlob Frege. which are morphologically derived Oxford: Blackwell, 56–78. First edition from the names of something else, for 1952. example ‘grammaticus’ (‘grammar- Quine, W. V. O. (1960). Word and Object. ian’) derives from ‘gramatica’ (‘gram- Cambridge, MA: MIT Press. mar’). This is tantamount to saying

25 CONTINUITY that the abstract name ontologically directional relationship between a set precedes any morphological variant. of entities in the world and the linguis- This view was spread in the early tic expressions used to refer to them, by St Anselm of Canter- whereas reference/extension is viewed bury who introduced the notion of as a more specific relationship between indirectness. On the one side, denom- one specific entity in the world and the inatives point directly to the property word in a given used to men- (‘grammar’) and indirectly to the sub- tion it. In a similar vein, connotation is stance (‘grammarian’); on the other, often paired with sense/, but denominatives have both meaning while sense encompasses the whole set (the concept of grammar, ‘significa- of properties of an entity, intension tio’) and reference (the person who is refers to the qualities of the entity a grammarian, ‘appellatio’). Two hun- which are implied in a given utterance. dred years later, in the fourteenth cen- In modern linguistics Saussure offered tury, tackled the a clear distinction between the two issue of the semantics of labels. from his semiotic from the perspective and of the , which is con- conceived of a distinction between ceived of a signifier and a signified, ‘absolute’ and ‘connotative’ terms. denotation refers to the meaning of a The former refers either to the names word as is given in a dictionary, the of qualities (albedinem, ‘whiteness’) objective semantic content codified by or of substances (canis, ‘dog’). The the signifier, whereas connotation latter refers to both the names of qual- refers to any other semantic implica- ities and the names of substances. tions that are part of the speaker’s Along the centuries St Anselm’s theory value in a given context. of denotation was progressively aban- doned and it was Ockham’s version Primary sources that became prominent in the nine- Lyons, J. (1995). Linguistics Semantics. teenth century within the philosophi- An Introduction. Cambridge: Cam- cal approach to semantics of John bridge University Press. Stuart Mill. It was Mill who first men- tioned the dichotomy between conno- Further reading tation and denotation in his System of Rigotti, E. and A. Rocci (2006). Denota- Logic (1843), two labels that he used tion versus Connotation. In Ency- in a similar way as sense and refer- clopaedia of Language and Linguistics. ence*, or extension and intension. Elsevier. 436–44. Mill limited Ockham’s ‘absolute’ terms to proper names and to abstract Annalisa Baicchi names for attributes (‘whiteness’). The meaning of the two terms is closely related to the theoretical framework taken as a perspective, CONTINUITY and it varies broadly across theories. In logico-philosophical semantics, A view arising from the theory that deno ta tion corresponds to the bi- human language is the product of a

26 CONTINUITY unique biological and genetically animals. The work of Charles Hock- encoded faculty. The theory of conti- ett in the 1970s and 1980s is central to nuity suggests that human language this effort to describe the key elements must have evolved over time in the of human language, including for same way as other biological and example the use of the vocal-auditory genetically inherited features. Yet channel, semanticity, arbitrariness, there is little hard evidence to suggest and creativity*. But Hockett and that any other animals have a lan- others continually changed their guage faculty with similar capabili- minds about what features should be ties. This is sometimes known as the on the list, and the issue of what con- ‘continuity ’: language must stitutes human language has since have evolved in humans, but there is been recognised as a diversion. no surviving evidence that it did. For One problem for proponents of many thinkers the ‘continuity para- continuity theory is that they often dox’ is simply an unsolved problem presuppose that evolution takes place that, in the absence of any explana- in a linear and hierarchical way, from tion, needs to be accepted. the most ‘primitive’ life forms, to the most ‘sophisticated’, namely humans. See also: Cognitivism; Creativity; However, this view of evolution fails Mentalism; Transformational- to take account of the fact that while Generative Grammar; Universal many animals, including humans and Grammar chimpanzees, may share a common Key Thinkers: Chomsky, Noam; ancestor, they are not necessarily Hockett, Charles directly related. In other words, The theory of continuity holds that human language may have evolved the human language faculty must have after the human and chimpanzee evolved gradually over time, rather ‘branches’ of an evolutionary tree had than appearing fully formed as a split apart. In that case there is no single mutation and that it must be reason why a chimpanzee should ever possible to identify the ‘continuity’ in show human language capabilities. the evolutionary development of lan- Most scientists now believe that they guage in humans. Attempts to prove never will. that language in humans is the result Noam Chomsky has often expressed of a long process of evolution by nat- scepticism about the idea that the lan- ural selection have usually involved guage faculty evolved in the same way animals. Parrots have been trained to as other biological features, but the ‘talk’, while chimpanzees, lacking the lack of evidence for continuity does physical apparatus necessary for not mean it is false. As Stephen Pinker human-like speech, have been taught argues in . One of the problems (1994), the trunk of an elephant is a encountered by such experiments is feature unique to that creature, but deciding what constitutes ‘language’ biologists do not spend time testing in a human sense and establishing other animals to see if they are able to the criteria against which we should use their noses in the same way as measure the linguistic abilities of an elephant uses its trunk. Viewing

27 CONVENTIONAL MEANING human language as a unique adapta- the sounds and the meaning thereby tion, just as an elephant’s trunk is expressed. The stored meaning of unique, does not preclude evolutionary expressions in the lexicon is an exam- continuity. Advances in genetics may in ple of conventional meaning. Conven- the end show how humans came to tional meaning must be learnt, as it have language when other animals do cannot be inferred based on principles not. But in the mean time it is not of rationality. The distinction between unreasonable to assume that, as Pinker conventional meaning (what an ex- says, ‘There were plenty of organisms pression means) and speaker’s mean- with intermediate language abilities, ing (what a speaker means by using a but they are all dead’ (1994: 346). particular expression) is sometimes used to draw the line between seman- Primary sources tics and pragmatics. Aitchison, Jean (1989). The Language See also: Nonnatural Meaning; Instinct: An Introduction to Psycholin- Signs and Semiotics guistics. London: Unwin Hyman. Key Thinkers: Aristotle; Grice, H. Sebeok, T. (ed.) (1968). Animal Commu- P.; Lewis, David; Plato; Saussure, nication: Techniques and Study Results Ferdinand de; Strawson, P. F. of Research. Bloomington: Indiana Uni- versity Press. The question of whether the origin of Pinker, Steven (1994). The Language linguistic meaning lies with natural Instinct: The New Science of Language necessity or human convention is and Mind. Harmondsworth: Penguin. raised in Plato’s , where Socrates expresses the view that lin- Further reading guistic meaning is a matter of habit, a Uriagereka, Juan. ‘The Evolution of Lan- view echoed in Aristotle’s De Inter- guage’, in Seed (http://seedmagazine. pretatione. In contemporary linguis- com), 25 September 2007. http://seed- tics the conventionality of linguistic magazine.com/news/2007/09/the_evolu meaning is acknowledged in Ferdi- tion_of_language.php?page=1(acces sed nand de Saussure’s arbitrary associa- on 30 January 2008). tion of signifier and signified, which engenders the possibility for this asso- Christopher Routledge ciation to shift and change across lan- guages and over time. More recently, H. P. Grice explained linguistic meaning as a type of non- CONVENTIONAL natural meaning*. Contrary to other types of signs which are symptomatic MEANING of particular states of affairs and thus express natural meaning, the meaning The fact that a sequence of sounds of a linguistic sign is dependent on expresses a certain meaning in virtue how speakers use it to bring about par- of a tacit among speakers ticular effects in an audience. Such use at a certain time and place, rather than is in turn more or less constrained by because of any necessary link between the customary use of the sign by

28 CONVERSATION ANALYSIS

speakers in a community, that is by its Grice, H. P. (1989), Studies in the Way conventional meaning. According to of Words. Cambridge, MA: Harvard Grice, ‘the conventional meaning of University Press. 22–40. the words used’ is part of the input to Morgan, Jerry (1978). ‘Two types of con- the derivation of conversational impli- vention in acts’. In P. catures (1967/1989: 31). Nonnatural Cole (ed.), Syntax and Semantics, vol. 9: meaning thus encompasses both Pragmatics. New York: Academic Press. conventional (language-dependent) 261–80. and intentional (speaker-dependent) Traugott, Elizabeth (1999). ‘The role of aspects of meaning. pragmatics in semantic change’. In J. The continuity between conven- Verschueren (ed.), Pragmatics in 1998. tional and intentional aspects of Antwerp: International Pragmatics meaning is the topic of David Lewis’s Association. 93–102. book Convention (1969), where he investigates the rational bases for con- Marina Terkourafi ventions, explaining these as solutions to co-ordination problems. Context plays an important role in sustaining the transition from nonce meaning to CONVERSATION conventional, or coded, meaning. The related notions of conventionalisa- ANALYSIS tion and standardisation have been proposed to account for the grad- The systematic study of all aspects of ual decontextualisation of recurring talk in social interaction. The findings mean ings of linguistic expressions, of this important discipline, started by such that these meanings eventually Harvey Sacks in the mid-1960s, are become part of an expression’s con- significant to all social sciences. ventional meaning. See also: (Critical) ; Speech Act Theory Primary sources Key Thinkers: Austin, J. L.; Grice, Lewis, David (1969). Convention: A H. P.; Sacks, Harvey; Searle, John Philosophical Study. Cambridge, MA: Harvard University Press. According to Anthony Liddicoat, con- Strawson, Peter (1964). ‘ and versation analysis (CA) – also known convention in speech acts’. Philosophi- as ethnomethodology – originated in cal Review 73, 439–60. the work of Harold Garnfikel, an eth- nomethodologist whose research in Further reading the 1960s centred on how members of Bach, Kent (1998). ‘Standardisation revis- a certain community conduct and ited’. In A. Kasher (ed.), Pragmatics: understand social actions and interac- Critical Concepts. Vol. IV. London: tions Liddicoat (2007: 2). These ideas Routledge. 712–22. were developed by Sacks who deter- Grice, H. P. (1967). ‘Logic and Conversa- mined that conversation is an organ- tion’. William James Lectures, Harvard ised activity that has a systematic University typescript. Reprinted in structure. Ironically, Sack’s claim was

29 CONVERSATION ANALYSIS proposed in the mid-1960s, during the connection has been established. This heyday of the Chomskyan paradigm, opening is followed by making an iden- which regarded ‘talk’ as being unfit as tification and exchanging greetings. a source of linguistic data because it is This pattern recurred in the major- riddled with irregularities and flaws. ity of the telephone calls analysed However, Sacks and his collaborators, by Schegloff. However, variations Emanuel Schegloff and , occurred as a result of a number of proved that conversation is an orderly, reasons, including the setting of the collaborative activity characterised by call. If the call is made to a workplace, consistent principles that define what for instance, the answerer would start linguistic forms interlocutors use to by giving the name of the organisation open and close their contributions, and not the . Although how they repair and rephrase their face-to-face are charac- talk, and when they take the floor terised by similar features, there are a during the course of a conversation. In few differences. These result from the addition to these, CA also investigates fact that interlocutors can see each the role of proxemics, or inter-per- other, which would influence the iden- sonal distance and , in tification part of the opening sequence. communication. The principles of CA Members of a have significant implications for a also need to be aware of the way in number of disciplines that include lin- which conversations are closed, or guistics, and com- concluded. Since talk in social interac- munication, to name a few. It is tions is a collaborative process, it is therefore useful to summarise some of important to ensure that a conversa- the areas in which CA has proved tion is not ended in an abrupt manner useful or enlightening. that would deprive any of the partici- Since CA investigates the sequential pants of the right to contribute. It has nature of conversation, analysts have been observed that participants use studied the systematically recurring lin- certain expressions that signal the end guistic units that open these conversa- of the conversation, the most common tions. The majority of these studies among which is ‘goodbye’ or one of have focused on telephone conversa- its synonyms. However, a closing tions because of the limitations result- sequence cannot just be introduced at ing from the absence of physical clues any juncture of the conversation to such as body language, facial expres- signal its conclusion. Before terminat- sions and eye gaze. Schegloff has inves- ing a verbal interaction, a pre-closing tigated opening sequences in telephone sequence alerting the participants to conversations (1979, 1986) and con- the nearness of the conversation end cluded that they are characterised by a has to be introduced. This is meant to predictable sequence in which the ensure participants are given the answerer of the telephone call speaks opportunity to make whatever contri- first, saying ‘hello’. Since making a butions they deem necessary. Sche- call is considered to be a summons gloff and Sacks define a pre-closing that requires a response, the answerer sequence as a turn after the pro- is expected to acknowledge that a duction of which every participant

30 CONVERSATION ANALYSIS

‘declines at least one opportunity to which reflects the turn-taking mecha- continue talking’ (1973: 214). Expres- nism of conversation, this may some sions that could function as pre-clos- times be altered for a certain commu- ing sequences include ‘alright’, ‘right’, nicative purpose. If an interlocutor, and ‘okay’ and they are usually for instance, needs more information marked by falling intonation. If none before he/she answers a question, this of the speakers introduces a new would result in the of the topic, the pre-sequence will be fol- two elements of the adjacency pair. lowed by a closing sequence that Cook gives the following example: would mark the conclusion of the con- versation. However, if a new topic is A: Did you enjoy the meal? (B: Did you? introduced, the conversation will con- A: Yes.) B: So did I. tinue unless one of the participants is unable to proceed because of time Cook explains that the initial adja- constraint or some other reason. cency pair, which is contributor A’s Another phenomenon characteristic first turn and contributor B’s second of conversations is ‘sequence pairs’. turn, has been interrupted by a second These are conversation contributions adjacency pair, the one between paren- by different speakers that are sequen- theses. He calls this interrupting ele- tially placed next to each other. ment an ‘insertion sequence’ (1989: Schegloff and Sacks (1973) proposed 54). It is evident from the example that the term to refer to sequences such speaker B needed more information as greeting-greeting and question- before responding to the question. answer. Since conversation is a collab- Talk in social interaction is a col- orative social act to which participants laborative endeavour in which turn- are expected to contribute, a question taking seems to be a highly structured should be answered and a greeting process. Although interruptions and reciprocated. The failure to do so gaps are not uncommon, conversation would be considered an act of non- participants seem to have tacit knowl- compliance and a violation of social edge of when to ‘take the floor’ and etiquette. Guy Cook (1989) classifies offer their contribution to the conver- responses to into ‘pre- sation. When asked about what clues ferred’ and ‘dispreferred’. According they use to start speaking, people to his framework, agreeing with an often mention pauses and length of assessment is the preferred sequence, contribution as the most reliable while disagreeing with it is dispre- clues. Citing several examples, how- ferred. As for responses to questions, ever, Liddicoat argues that neither of giving the expected answer is the pre- these clues are valid markers of a ferred response, while giving an unex- turn’s end (2007: 54–6). Sacks et al. pected answer is the dispreferred (1974) speculate that speakers know response. Cook adds that a dispre- when to take the floor because they ferred response is usually explicitly have tacit knowledge of turn con- justified or non-verbally marked. structional units. They define these as Although the two parts of an adja- linguistic forms that are not identifi- cency pair occur next to each other, able in terms of syntactic structure. A

31 CONVERSATION ANALYSIS turn constructional unit can be a type of silence may also result in an single word, a phrase, a clause, or a extended lapse in the conversation. longer stretch of discourse. They add Another type of silence, however, that speakers have knowledge of the occurs at a juncture of the conversa- pragmatic function of these context- tion where a specific participant is sensitive units, which enables them to expected to contribute, for example, identify the end of turns. A variety of to answer a question. This type of other proposals have been offered to silence is considered problematic explain what clues participants in because it results in an act of non- conversation use to take the floor. compliance. Strategies used to repair These include utilising knowledge of failure to contribute include repeating syntactic, semantic and suprasegmen- the question or changing the topic. tal features of conversation, in addi- Overlapped speech, on the other tion to using visual clues to determine hand, may be the result of a partici- end of turns. Among these is the para- pant’s misreading of a pause as the tone, which Brown and Yule define as end of the speaker’s turn and a sign of a conversational paragraph whose forfeiting the floor. When that hap- end is marked by a summarising pens, the overlap may persist for some sequence, a significant pause and fall- time before one of the participants dis- ing pitch (1983: 101). Another related continues. Another possible reason area of interest is turn allocation, for overlaps in conversation is when, which investigates whether the at the end of a speaker’s turn, two person holding the floor selects the interlocutors take the floor at the next speaker, or whether the subse- same time. If an overlap occurs at a quent speaker self-selects. In their point that is not a potential juncture 1974 framework, Sacks et al. state for speaker change, it is considered an that a speaker holding the floor may: interruption. O’Grady et al. believe that how overlaps are perceived is cul- (a) directly select a subsequent speaker ture-specific. Communities that speak by using a personal name, a pro- the standard North American English noun, or asking a question; varieties, they argue, adhere to a rigid (b) indirectly select the next speaker by turn-taking mechanism, which results using gaze and establishing eye in very few overlaps. The East Euro- contact; pean Jewish community in New York, (c) that a speaker holding the floor may however, views overlaps favourably not select any one, which leaves it because they reflect the speakers’ open to participants to self-select. interest in the conversation. Hence, they call overlaps in which speakers Despite the tacit knowledge speakers finish one another’s sentences ‘cooper- have of turn-taking rules in conversa- ative overlaps’ (2005: 491). Beside tion, gaps and overlaps occur. Accord- cultural differences, other factors that ing to the framework in Sacks et al., might determine overlaps include the silence may take place at a transition topic, type of conversational interac- relevance place, where any of the par- tion and number of participants. In an ticipants could claim the floor. This setting, for instance, where

32 CONVERSATION ANALYSIS only two people are involved, over- have received specific attention from laps are rare. On the other hand, if a conversation analysts. First, a story large number of people are involved in requires a fairly extended turn, which the discussion of a heated topic, over- may disrupt the orderly turn-taking laps are likely to occur at a high rate. mechanism of conversation; and, Unlike , in the pro- second, in order to maintain coher- duction of which a significant amount ence and topic continuity, a story has of time is spent on composition and to be introduced at the right juncture revision, is planned in the conversational interaction. Jef- in the here-and-now. Consequently, ferson (1978) proposes two factors conversations are marked by ‘repair’, that trigger . The first is a process whereby utterances are mod- when a conversation topic reminds a ified. Liddicoat states that although speaker of the events of a certain expe- repair may target a perceived mistake, rience. When that happens, she or he people often use repair when they fail may elect to relate the story; however, to retrieve a specific item from their the conversation participants have to lexicon, or mental dictionary, in a be alerted by such expressions like ‘by timely manner (2007: 171). Schegloff the way’, or ‘incidentally’. According et al. (1977) have proposed a taxon- to Jefferson, these serve as disjuncts omy for the classification of repairs marking topic transition and provid- that occur in conversation. Their ing some kind of a justification for the categorisation takes into account who . Researchers have also initiates the repair and who makes asserted the interactive characteristics it. This yields four types which are: of story-telling. Contrary to common (1) self-initiated self-repair, which belief, a story is not a monologue that denotes deficiencies detected and is delivered by a single speaker. Not resolved by the speaker; (2) self-initi- only do participants comment on and ated other-repair, which refers to defi- evaluate the narrative after its deliv- ciencies detected by the speaker, but ery, they also frequently ask questions repaired by another; (3) other-initiated about and make remarks on the story self-repair, denoting irregularities while it is being told. Upon the com- pointed out by an interlocutor, but pletion of the narrative, receivers are repaired by the speaker; and (4) other- expected to make some kind of a initiated other-repair, referring to response. That response may be an repairs detected and repaired by some- evaluation of the story, a comment on one other than the speaker. Typically, its plot, or an expression of empathy. Schegloff et al. argue, repairs involving If silence follows the end of a story, grammatical mistakes are initiated by that would be regarded as an act of the speaker, whereas repairs of hearing non-compliance. problems are other-initiated. Citing a Deborah Schiffrin has proposed variety of examples, the researchers a model for the analysis and inter- indicate the different junctures of the pretation of spoken English that conversation at which repair occurs. focuses on what she calls local coher- Because of their unique characteris- ence, or coherence between adjacent tics as a specific of talk, stories units in discourse. This, she ,

33 CONVERSATION ANALYSIS is accomplished by employing dis- Schiffrin also examined the role of course markers that speakers use to ‘well’ in response to requests. She sug- achieve not only semantic, but also gests that in reaction to requests for pragmatic goals. Although Schiffrin action, it is more likely for ‘well’ to be does not give a concrete definition, her used in responses that indicate non- operational definition is that dis- compliance to requests. In addition to course markers are ‘sequentially these contexts, Schiffrin investigates dependent elements which other roles of ‘well’, such as marking units of talk’ (1987: 31). Despite the requests for clarification, self-repairs definition’s limitation, which arises and reported speech. Schiffrin empha- from the difficulty of defining speech sises the need for future research that units, Schiffrin identifies eleven dis- examines the distribution of different course markers: ‘oh’, ‘well’, ‘and’, markers in different conversational ‘so’, ‘now’, ‘then’, ‘I mean’, ‘but’, . Such research, she asserts, ‘because’, ‘y’know’ and ‘or’. Then she would enable us to understand how applies these markers to the analysis coherence is achieved in different of authentic conversation explaining types of discourse. how they are used by participants to Finally, an issue that is of utmost achieve coherence through accom- importance to every conversation plishing a variety of communicative analyst is transcription. Since speech goals. Schiffrin (1985) argues that is transient, it cannot be studied as ‘well’ is the most versatile discourse it unfolds in real time. Hence, marker because it does not have inher- researchers have to tape-record and ent semantic or structural attributes. transcribe it using a variety of con- Its meaning, however, derives from ventions whose purpose is to capture the context in which it occurs. all of the conversational aspects dis- Schiffrin distinguished between occur- cussed above. However, it is not rences of ‘well’ in adjacency pairs easy to come up with a system of (questions/answers, request/compli- transcription that is neutral and ance sequences), and occurrences of objective. Moreover, a transcription ‘well’ which cannot be explained in is not a substitute for the conversa- terms of adjacency pairs. tion. It has been argued that tran- Examining the distribution of ‘well’ scription should be modified as in answers to wh- questions and researchers listen again to the record- yes/no questions, Schiffrin found that ing. Although the objective of the ‘well’ precedes answers to yes/no transcription is to capture as many questions 10 per cent of the time; on features of the conversation as possi- the other hand ‘well’ precedes answers ble, a balance has to be struck be tween to wh- questions 21 per cent of the the comprehensibility of the transcrip- time. She attributed this to the fact tion conventions and the accessibility that yes/no questions delimit the of these conventions to the targeted upcoming answer to either an affir- audience (Liddicoat 2007). mative or a negative response, hence It is widely believed that a tran- ‘well’ is not regarded as a coherent scription should include information marker for answers in these contexts. about the conversation participants,

34 CONVERSATION ANALYSIS time and place. Since the anonymity of Practices, and Applications. Malden, the participants should be maintained, MA: Blackwell Publishing. pseudonyms are usually used. A major Jefferson, G. (1978). ‘Sequential Aspects issue regarding transcriptions is word of Storytelling’. In J. Schenkein (ed.), representation. Using the conven- Studies in the Organization of Conver- tional has the advantage sational Interaction. New York: Aca- that it is a system known to the whole demic. 219–48. literate society. However, since con- Liddicoat, Anthony (2007). An Introduc- ventional orthography is based on a tion to Conversation Analysis. London: standard variety of the language, some Continuum. scholars, for example Duranti (1997), O’Grady, William, John Archibald, Mark argue that using it would undermine Aronoff and J. Rees-Miller (2005). the and of the Contemporary Linguistics: An Intro- transcription. In addition to the con- duction. Fifth edition. Boston, MA: versation, non-linguistic sounds such Bedford/St Martin’s. as laughter and back channelling Sacks, H., E. Schegloff and G. Jefferson forms have to be in cluded. The tran- (1974). ‘A simplest systematics for scription system also needs to indicate the organisation of turn-taking for pauses within a contributor’s turn. conversation’. Language Vol. 50: Moreover, a pause between speakers’ 696–735. turns is also significant because its Schegloff, E. A. (1979). ‘Identification and absence is an indication of overlapped recognition in telephone conversation conversation, which is significant to openings’. In G. Psathas (ed.), Everyday the turn-taking mechanism. Supraseg- Language: Studies in Ethnomethodol- mental features such as and ogy. New York: Irvington. 23–78. intonation have to be transcribed Schegloff, E. A. (1986). ‘The routine because of their interactional signifi- as achievement’. Human Studies Vol. 9: cance. In addition to these features, 2–3, 111–51. the transcription should account for Schegloff, E. and H. Sacks (1973). overlaps, body language and back- ‘Opening up closings’. Semiotica Vol. 7: ground noise. Liddicoat asserts that 289–327. transcripts should be continuously Schegloff, E., G. Jefferson and H. Sacks updated, adding that a transcript is (1977). ‘The preference for self-correc- ‘not an objective account and that it tion in the organization of repair in con- will always be a selective representa- versation’. Language Vol. 53: 361–82. tion of the data itself’ (2007: 50). Con- Schiffrin, Deborah (1987). Discourse versation analysis is an important tool Markers. Cambridge: Cambridge Uni- for understanding language in use, but versity Press. its usefulness to other disciplines, such as and , is also Further reading very significant. Brown, Gillian, and George Yule (1983). Discourse Analysis. Cambridge: Cam- Primary sources bridge University Press. Hutchby, Ian, and Robin Wooffitt (1998). Cook, Guy (1989). Discourse. Oxford: Conversation Analysis: Principles, Oxford University Press.

35 CORPORA

Duranti, A. (1997). Linguistic Anthropol- the empiricist J. R. Firth observed, ogy. Cambridge: Cambridge University ‘You shall know a word by the com- Press. pany it keeps.’ A corpus provides not Psathas, George (1995). Conversation only such company (technically known Analysis: The Study of Talk-in- as concordance), but also many other Interaction. Thousand Oaks, CA: Sage linguistic insights. Linguists have Publications. numerous pre-existing corpora avail- Schiffrin, D. (1985). ‘Conversational able to them, or they may choose to Coherence: The Role of “Well” ’. Lan- assemble their own, such as when they guage Vol. 61: 640–67. seek to analyse or test criteria or char- acteristics not adequately represented Yousif Elhindi in existing corpora. An effective corpus is a representa- tive sample of natural language taken from the population being studied. CORPORA Linguists must match their needs against the specific criteria of a corpus Collections of written texts or spoken such that the results of their research utterances representative of a particu- and analysis may transfer to the real lar language, , population or world what the corpus is meant to style. Linguists use corpora to provide reflect. There are a number of charac- real-world context in language analy- teristics by which corpora vary from sis and language processing. one another. For instance, some cor- pora contain textual data in the form See also: Behaviourism, of written words, sentences and punc- Conversation Analysis, (Critical) tuation, while others contain speech Discourse Analysis, data in form of utterances, fragments Empiricism/Rationalism and filled pauses. Another major Key Thinkers: Firth, J. R.; Sinclair, factor is whether the corpus is raw or John annotated. A raw corpus has not been Linguistics calls upon language data to modified from its original form, as in provide evidence to support the exis- the case of an online book or maga- tence of particular linguistic phenom- zine. By contrast, an annotated corpus ena or to challenge theories and has been marked up with tags to iden- algorithms. Linguists may derive such tify certain elements, like parts of evidence from experience or personal speech. A corpus may focus on a par- knowledge, or they may refer to a writ- ticular subject area or population, or ten or recorded set of external data in it may represent a variety of subjects the form of a corpus. A corpus – from or even genres. Corpora also vary by the Latin for ‘body’, as in ‘body of time period, geographical region and texts’ – provides perspective and con- size, and they may even use different text for language data. Although languages or dialects. Some corpora modern corpora typically exist as com- are multilingual, containing text or puter databases or files, the idea is far speech in multiple languages. When from new. In the mid-twentieth century those different languages express the

36 CORPORA same content, this is known as a par- Times. A more modern version of the allel corpus. Existing corpora, avail- corpus is the Associated Press World able through a number of sources, Stream English, published in 1998 including the popular Linguistic Data and containing approximately 143 Consortium (LDC), may be available million words. It combines the North for a fee or free of charge. American News with Among the first widely-available English-language text written in all corpora are those in the the Brown parts of the world, usually not in corpus, named for Brown University, American English. This corpus is where it was created in the early marked up in Standard Generalised 1960s. First published as the Standard Markup Language (SGML) to make Corpus of Present-Day American analysis easier. English by Henry Kucera and Nelson Newswire data is also used in the Francis, it contains 500 written texts Wall Street Journal (WSJ) corpus, and totals more than one million which exists in several forms. An early words of running text, averaging over version of it, called CSR-I, contains 2,000 words per text. Although it has Wall Street Journal news articles read seen a number of proofreadings and aloud in the early 1990s as part of revisions since its creation, the Brown an effort to support research on corpus remains an admirable repre- large-vocabulary Continuous Speech sentation of 1961 written American Recognition (CSR) systems. This was English, spanning a wide range of soon revised as CSR-II, containing styles and varieties of prose, including 78,000 utterances, or seventy-three both informative and imaginative hours of speech. There are also writ- prose, such as text from newspapers, ten text variations of Wall Street Jour- novels, non-fiction and academic pub- nal data, containing thirty million lications. Kucera and Francis con- words parsed in a -style (cap- ducted a thorough analysis of the turing syntactic structure) with part of corpus, a landmark in modern corpus speech tagging. linguistics. Another version of the Combining both written and spoken Brown corpus now exists with tagging text from a wide range of British Eng- for parts of speech. lish sources, the British National Another important corpus includes Corpus (BNC) contains 100 million texts published by the Associated words and has been updated several Press (AP), aptly known as the North times since its birth in 1991. Its mas- American News Text Corpus, or more sive size is roughly equivalent to 1,000 simply the AP corpus. There are actu- average paperback books. The written ally multiple versions of this corpus. A portion of the BNC corpus represents 1987 version of it contains fifteen mil- 90 per cent of the total and contains lion words, and a 1988 version con- text from British newspapers, periodi- tains thirty-six million. The content cals, journals, non-fiction and fiction came from North American authors, books, letters and essays. The remain- news stories in American Eng- ing 10 per cent consists of spoken lish for the Los Angeles Times, the utterances by British English speakers Washington Post and the New York in the form of informal conversation,

37 CORPORA business or government meetings and All of the corpora discussed above radio shows. Its goal is to represent a are monolingual, representing only cross-section of written and spoken the English language. By contrast, the British English, produced by a variety CELEX corpus is multilingual corpus, of sources and intended for a mixture including not only English, but also of ages and backgrounds, representa- German and Dutch. It was developed tive of the British population itself. The as a joint effort by the University of BNC corpus includes Nijmegen, the Institute for Dutch Lex- tagging, as well as built-in identifica- icology in Leiden, the Max Planck tion of its structure, such as headings, Institute for in paragraphs, and lists. Nijmegen, and the Institute for Per- An important corpus for spoken ception Research in Eindhoven. This language data is the SWITCHBOARD corpus is a database of lexical word corpus, which consists of about 2,430 forms, with detailed information on spontaneous telephone conversations orthography (variations in spelling, by over 500 diverse native speakers of hyphenation), phonology (phonetic all major dialects of American English transcriptions, variations in pronunci- between the ages of twenty and sixty. ation, syllable structure, primary The conversation topics come from a stress), morphology (derivational predetermined list of topics, and the and compositional structure, inflec- speakers did not already know one tional ), syntax (word , another at the time of their recorded word class-specific subcategorisa- conversations. Developed at Texas tions, argument structures) and word Instruments in 1991, SWITCH- frequency (summed word and lemma BOARD contains a total of over 240 counts, based on recent and represen- hours of data. Unlike the corpora tative text corpora). mentioned above, the spoken lan- A very popular multilingual corpus guage within SWITCHBOARD is very is the European Parliament Pro- natural, not scripted from written text. ceedings Parallel (Europarl) corpus, However, a written version of the assembled from 1996 to 2003. Ex - corpus does exist, produced later via tracted from the proceedings of the transcription of the speech data. For European Parliament, it includes par- each transcription, a time alignment allel text in eleven European lan- file indicates the beginning time and guages: Romanic (French, Italian, duration of each word, information Spanish and Portuguese), Germanic that is particularly useful for studying (English, Dutch, German, Danish and the phonetic characteristics of the Swedish), Greek and Finnish. It con- speech, as is common in developing tains roughly twenty million words in and evaluating automatic speech 740,000 sentences per language, with recognition systems. Another advan- each language aligned by sentence or tage of the corpus is that it comes with by document. This corpus makes it background information about each possible to align 110 language pairs, speaker along with details surround- including languages (particularly ing each specific conversation, includ- Greek and Finnish) that were poorly ing the nature of the telephones used. represented in previous corpora. For

38 CORPORA this reason, Europarl is frequently corpus to have multiple instances, used within the field of statistical each formatted according to a differ- machine translation to train and eval- ent standard. One such standard is a uate systems to an extent that was pre- treebank, where each sentence is viously not possible. annotated as a tree structure based The CALLHOME corpora is a col- upon its syntactic structure, which lection of speech data, originally cre- itself is often annotated through part ated to support research in large of speech tagging. A treebank may vocabulary conversational speech represent phrase structure (as in recognition, which was suffering due the case of the Penn Treebank) or to a shortage of speech data for lan- dependency structure (as in the case guages other than English. Its creators of the Prague Dependency Treebank). offered its speakers free telephone Another formatting standard, used calls to their native country in specifically for bilingual corpora, is a exchange for permission to record bilingual knowledge bank (BKB), their conversations. The result is a which organises a bilingual corpus series of corpora, arranged by lan- into translation units, or aligned guage (English, Spanish, Arabic, ‘bitexts’. German, Mandarins and Japanese), Corpora play a supporting role in containing unscripted telephone con- several fields of research. The most versations, up to thirty minutes in obvious of these is , length and originating in North Amer- which focuses its study of language ica. All speakers are native speakers of specifically on corpora and real-world their respective languages, and many text. To accurately represent language interactions are between family mem- as it is actually used, a corpus must be bers or close friends. The CALL- very large in volume. Corpus linguis- HOME corpora were thus named for tics thus relies heavily upon auto- their collection method. mated methods for conducting proper The corpora discussed here are analysis of data. Technology has given good examples of the range of cor- us new analysis techniques that were pora available to linguists, who select previously not possible for reasons of a corpus based upon their specific sheer quantity of data, particularly needs in analysing data or testing the- when the corpus being studied is ories or natural language processing annotated with information about the (NLP) systems. Data within a corpus text structure, parts of speech, seman- may be textual data or speech data, tics, and so on. Statistical NLP also and even speech data may be tran- uses data from corpora in order to scribed such that it, too, exists as count words and compute word prob- text. Text corpora are commonly for- abilities based on word frequencies matted to identify key elements or within a training corpus, even taking structure within the data, and the for- into account such factors as concor- matting conventions must be clear to dance and discourse features. Speech those who use the corpus so that they corpora are particularly useful in can properly interpret the data it con- speech recognition technology, which tains. It is even possible for a single may involve training software using

39 CORPORA speech corpora data as well as testing of the language used by the popula- systems against an existing corpora of tion of interest. When selecting a natural, unscripted speech. It may corpus or collecting data to build a even play a role in assembling a con- new corpus, a linguist must identify trolled language to be used for a what type of text a corpus must reflect speech recognition system, identifying as well as whether the results of work the words and phrase structures that with the corpus will transfer to the regular people may use when interact- domain of interest. That is, a corpus ing with the system. Machine transla- serves as a powerful tool for linguists, tion (MT) technology also uses but it cannot wholly substitute for the corpora for example-based systems real world. In general, the larger a working on the assumption that trans- corpus, the more useful it is and the lation involves finding or recalling more likely it is to capture actual lin- analogous examples of words and guistic use. Yet even the largest corpus phrases. Corpus-based approaches to available is only as useful as the extent MT may be trained using a subset of to which it reflects the target domain texts within a particular corpus and and offers a usable format for analy- then tested against other texts within sis. In addition to size, the breadth of that same corpus. Finally, corpora are a corpus also affects its usefulness and also useful for exploring and inter- applicability for meaningful research. preting historical documents. For A valuable corpus for a linguist study- example, researchers working on ing general language use, for example, deciphering an ancient text could use should contain a variety of text a corpus of texts from the same his- sources, such as newspapers, text- torical period and geographical loca- books, popular writing, fictions and tion to find clues to better understand technical material. Another potential the text they are studying. drawback specific to speech corpora is A key criticism of corpus analysis is that they lack linguistic cues that are that it supports statistical methods otherwise available in a text corpus. that disregard our knowledge of lan- Speech corpora include both frag- guage. Within NLP, systems fall along mented, incomplete words and filled a spectrum, with purely rule-based pauses, such as ‘uh’ and ‘um’. While approaches at one end and corpus- these elements reflect how people based (statistical) approaches at the actually speak, which is crucial for other. Evaluations of competing NLP speech recognition systems, they systems show that hybrid systems, understandably make it harder to incorporating a combination of lan- work with the data. guage rules and statistical methods, The concept of a corpus presents consistently outperform systems that an interesting circular relationship use only one of the two methodolo- between linguistics and society in gen- gies. Even when a corpus-based (or eral. A corpus captures a subset of hybrid) approach is undisputed, a par- real-world language data, while lin- ticular corpus itself might not fit the guists use this data in a format that task. A corpus fails its job when it is enables them to study it, to analyse its not an accurate representative sample contents, and to develop and support

40 CORPORA theories or NLP systems. These NLP Knowing the importance of size, systems, developed and tested against linguists are using technological theoretically real-world corpora, in advancements to make increasingly turn can help their real-world users to large corpora to better represent the carry out tasks. For example, a speech real world. They also take advantage corpus, like the SWITCHBOARD of statistically-based techniques for corpus or CALLHOME corpus dis- automatically deriving cussed above, originates from real- concerning word counts, frequencies world speech data collected via and concordance. As long as linguists unscripted telephone conversations. continue to propose theories, they will Linguists may use such a corpus to need corpora to support them, and as learn about how people communicate long as they continue to design and over the telephone and thereby build NLP systems, they will need cor- develop a speech recognition system pora to test them. that can function properly with prac- tical, real-world speech data. The Primary sources quality of the final product is a direct Godfrey, John, Edward Holliman and reflection of the quality of the speech Jane McDaniel (1992). ‘SWITCH- corpus by which it was trained and BOARD: Telephone for evaluated. A poor-quality corpus research and development’. Proceedings would result in a speech recognition of the IEEE Conference on , system that proves to be an inade- Speech, and Signal Processing. San quate fit for its users’ needs. Likewise, Francisco: IEEE. a high-quality corpus would result in Koehn, Philipp (2005). ‘Europarl: A Paral- a speech recognition system that lel Corpus for Statistical Machine Trans- better anticipates the types of utter- lation.’ MT Summit 2005. Unpublished. ances its users may give it as input. A Kucera, Henry and Nelson Francis (1967). corpus aims to reflect real-world lan- Computational Analysis of Present-Day guage so as to support theories and American English. Providence, RI: systems that, too, reflect real-world Brown University Press. language. Marcus, Mitchell, Beatrice Santorini and A single corpus cannot be a perfect Mary Ann Marcinkiewicz (1993). fit for every linguistic endeavour, nor ‘Building a large annotated corpus of can it exhibit all aspects of language English: the Penn Treebank’. Computa- use. This is why there are so many tional Linguistics Vol. 19: 313–30. existing corpora available to linguists. Quirk, Randolf (1960). ‘Towards a Nevertheless, linguists often create description of English usage’. Transac- new corpora tailored to a particular tions of the Philological Society 40–61. language use scenario, geographical region, demographic group or format. Further reading Corpora will continue to play an Biber, Douglas, Susan Conrad and Randi important role in linguistics and the Reppen (1998). Corpus Linguistics: study of language, and automated Investigating Language Structure and methods will improve our ability to Use. Cambridge: Cambridge University collect, format and analyse their data. Press.

41 CORRESPONDENCE THEORY

Edwards, Jane (1993). ‘Survey of elec- phy, and its contemporary version is tronic corpora and related resources for mostly influenced by G. E. Moore, language researchers’. In J. Edwards Bertrand Russell, Ludwig Wittgen- and M. Lampert (eds), Talking Data: stein and Alfred Tarski. For many Transcription and Coding in Discourse people outside philosophy, including Research. Hillside, NJ: Lawrence Erl- linguists and lay-persons, the cor- baum Associates. respondence solution seems intu- Morley, Barry (2006). ‘WebCorp: a tool for itively correct: in our daily lives we online linguistic information retrieval have to assume that there is a relation and analysis’. In A. Renouf and A. Kehoe between what people say and the (eds), The Changing Face of Corpus Lin- things they talk about. But whether guistics. Amsterdam: Rodopi. what they say is true or false, we need the concept of truth because without Jennifer A. Baldwin it we would not be able to describe what happens in the world. One modern way of formulating the correspondence theory solution is to CORRESPONDENCE say that a sentence (a proposition) is true iff (if and only if) it corresponds to THEORY some fact, or iff it corresponds to some state of affairs. It is not hard to see the A theory that tries to solve the prob- problem: in order to appreciate the lem of what constitutes truth as a notion of correspondence one has to property of sentences by claiming that grasp the notions of ‘sentence’, ‘fact’ there is a relation of correspondence and ‘state of affairs’ respectively. Intu- between the meaning of true sentences itively we would say that the truth of a and the way the world is at a certain sentence has to do with the meaning of place and time. the sentence, and this may be differen- tiated from the state of affairs, which See also: Language of Thought; seems to be what is out there, in Logic; Sense/Reference; Truth reality. But can ‘sentence meaning’ be Theories; Truth Value separated from ‘fact’? Some sentences Key Thinkers: Aristotle; Kant, – for example, ‘Open the window’ – Immanuel; Moore, G. E.; Russell, do not express facts, but in general Bertrand; Tarski, Alfred; sentences are interpreted as expressing Wittgenstein, Ludwig facts of some kind. So, if a sentence Correspondence theory is the most meaning conveys true information, prominent of several theories trying to what is the between the solve the problem of what truth is. It meaning of a sentence and the facts originates in Greek philosophy, specif- that it conveys? In other words, if ‘sen- ically in the work of Aristotle, and tence meaning’ and ‘fact’ are the same appears in the medieval period in thing, how can they not correspond? the work of Thomas Aquinas and Dictionary definitions on the notion William of Ockham. It was later a of truth usually take up the nature of principle in Immanuel Kant’s philoso- the relation between language and

42 CREATIVITY reality. Some term for the concept of Semantics, . Indi- relation seems inescapable and so it is anapolis: Hackett. 152–278. in modern and philoso- Wittgenstein, Ludwig (1922). Tractatus phy of science. Thus the standard defi- Logico-Philosophicus. London: Rout- nition of knowledge is ‘justified true ledge. belief’ (Dancy 1985: 23), meaning that there must be a relation between Hans Götzsche beliefs and what the beliefs are about. If scientific knowledge is to count as trustworthy, it must be true (see Popper 1963: 215–50); there must be CREATIVITY an agreement between what science says and how things happen out there. The ability of the ordinary use of lan- Kant expressed the correspondence guage to be innovative and free from principle in the most transparent way: stimulus . As associated most ‘Wahrheit ist die Übereinstimmung der recently with Noam Chomsky Erkenntnis mit ihrem Gegenstand’, (though with ultimate roots in Carte- ‘truth is the correspondence of knowl- sian rationalism), it is claimed that edge with its object’ (Kant 1781/1787, this creativity points to the central 1924/1927: 992), even though he did place of language in the study of not consider it a problem, while Tarski human nature. offered a technical and rather com - See also: Behaviourism; Metaphor; plicated clarification of the corres- Universal Grammar pondence relation based on predicate Key Thinkers: Chomsky, Noam; logic*. Although it is commonly Descartes, René; Humboldt, accepted that some idea of correspon- Wilhelm von; Skinner, B. F. dence is a necessary requirement in truth theory*, it is also commonly In one sense, linguistic creativity has a recognised that the nature of this rela- narrow meaning, referring to the abil- tion is the source of much controversy. ity of human beings to innovate within the lexico-semantic domain. Speakers, Primary sources writers and poets can use the elements Kant, Immanuel (1781[A]/1787[B]. Pub- of their language to draw attention lished by Raymund Schmidt 1924/ to surprising and interesting aspects 1927). Kritik der reinen Vernunft. Wies- of the world through metaphor*, baden: VAM-Verlag. and the like. However, there is Popper, Karl R. (1963/1989). Conjectures another, broader sense of linguistic and Refutations. Fifth edition. London: creativity, most commonly attributed Routledge. in the modern era to Chomsky, which applies not to the activities of gifted Further reading individuals, but rather to the ordinary Dancy, Jonathan (1985). Contemporary use of language by everyday speakers. Epistemology. Oxford: Blackwell. This creativity underlines the funda- Tarski, Alfred (1935). ‘The concept of mental role that language plays in truth in formalized language’, in Logic, understanding the essential nature of

43 DECONSTRUCTION human beings and is often referred to . Therefore, according to by Chomsky as ‘the creative aspect of the Cartesians, human linguistic capac- language use’. ities provided evidence for the exis- The most general aspect of creativ- tence of a second substance ‘mind’ (as ity in this broad sense, and one which distinct from mere ‘body’). Although connects Chomsky’s work to the ear- metaphysical dualism no longer forms lier rationalist traditions of Wilhelm part of standard scientific assumptions, von Humboldt and René Descartes is these observations regarding linguistic the observation that human beings creativity nonetheless do suggest that, can produce (and understand with no in studying language, we are studying sense of novelty) an infinite number one of the foundations on which our of sentences which may be new in humanity rests. their experience or even new in the history of the language. This cre- Primary sources ativity is thus intimately connected Chomsky, Noam (1959). ‘Review of with the ‘generative’ nature of gener- Verbal Behavior by B. F. Skinner’. Lan- ative grammar, and forms part of the guage 35: 26–58. key motivation for universal gram- Chomsky, Noam (1966). Cartesian Lin- mar*. guistics: A Chapter in the History of As noted in some detail by Chom- Rationalist Thought. New York: sky (1959), human language use is Harper & Row. Reprinted (2002) also creative in the sense of being free Christchurch, New Zealand: Cyberedi- from identifiable stimulus control. It is tions Corporation. appropriate to a situation, but it is not caused by it. A traditional Geoffrey Poole behaviourist account of the response ‘Dutch’ to a painting on the wall would be to say that the speaker was under the stimulus control of the DECONSTRUCTION painting’s ‘Dutchness’. However, as Chomsky observes, a speaker could Deconstruction is an approach to have just as easily responded ‘It’s doing philosophy that subverts just tilted’, ‘I thought you liked abstracts’ about everything mainstream philoso- or an infinite number of other things. phy has stood for through the cen- Chomsky (1966) notes that this free- turies. It is associated with the name dom from stimulus control was one of of the controversial Algerian-born the Cartesian arguments for the exis- French philosopher tence of mind. Machines, once the who was strongly influenced by, internal arrangement of the parts among others, and the external conditions are and Martin Heidegger. specified, behave in a completely pre- dictable manner(or randomly). How- See also: Analytic Philosophy; ever, human linguistic behaviour, like Poststructuralism; Structuralism; human thought and action more gener- Truth Theories ally, transcends simple ‘mechanical’ Key Thinkers: Derrida, Jacques

44 DECONSTRUCTION

The very mention of the word ‘decon- There is a grain of truth in the struction’ causes jitters among many claim that deconstruction engages in academic philosophers and calls forth textual analysis. But it is important to derision from others. It is probably add that, as Derrida himself once true to say that no other term in phi- remarked, there is nothing outside the losophy has recently been the target of text. What Derrida was trying to press such tireless tirade and misrepresenta- home in his trademark claim was that, tion. In common parlance, it is often contrary to conventional thinking on erroneously and dismissively used as the matter, the reading of a text does though it were just a fancy word for not consist in pitting the text against destruction – thus proving the old something that is essentially of some adage that one can give a dog a bad other order – say, ideas or name and hang it. When Jacques Der- in the mind of the speaker/writer or a rida first employed the word, scarcely putative reality ‘out there’. Instead, a could he have imagined that it would reading of the text is an extension of soon be at the epicentre of so much that very text. That is to say, no text controversy and acrimony. From the comes out unaffected by its successive 1960s on, when Derrida took the readings. Each new reading of the text world of philosophy by storm, decon- adds something to that text and is struction became a catchword among fully incorporated into the text’s academics across a wide spectrum of ‘meaning’ so that the meaning of the disciplines, and his influence is felt in text – to the extent there is such a areas as diverse as literary criticism, thing – may be described as that which linguistics, sociology, and even such is constantly undergoing change, unlikely disciplines as and despite the illusory sensation that it is . the self-same object that one is dealing Both Derrida and the movement he with on successive occasions. started have understandably been the So what on earth does Derrida target of scathing criticism, especially mean by deconstruction and exactly from those scholars who, speaking how does it work? It has been on behalf of the philosophical ‘estab- remarked that it is easier to ‘define’ lishment’, rightly sensed a potential deconstruction by saying what it is threat to the way they went about not, rather than what it is. In a famous their routine business. However, to text called ‘Letter to a Japanese call the approach sceptical or down- Friend’, Derrida wrote that ‘decon- right nihilistic, or to characterise it as struction is neither an analysis, nor a just plain philosophical dilettantism, critique’ and, furthermore, it is ‘not a as some critics have done, is to dismiss method and cannot be transformed it without a fair hearing. To regard into one’. In fact, as it turns out, the deconstruction as a form of textual very quest for definitions is antitheti- analysis is to give it a left-handed com- cal to the spirit of deconstruction. pliment and, in effect to relegate it to This is so because the idea that mean- the realm of literary criticism – ings can be captured and encapsulated another way of saying it is anything in neat definitions is the very hallmark but philosophical. of philosophy in its traditional sense.

45 DECONSTRUCTION

Recall, for instance, Socrates’s irritat- ness of the reader to tease it out of spe- ing habit of asking his interlocutors to cific texts. provide him with a definition of, say, The process of unravelling a text’s ‘piety’ (as repeatedly shown in Plato’s putative meaning begins the moment Dialogues). Therefore, to the extent one notices that the that that deconstruction seeks to, as it a text’s author posits in order to con- were, ‘call the bluff’ of philosophy in struct his arguments are actually rela- respect of its time-honoured preten- tions of hierarchies, often ingeniously sions, it is only to be expected that it camouflaged in order to be presented should stubbornly resist every attempt as symmetric. In other words, it is to enclose it in a neat definition. invariably the case that, in a dichoto- The idea that one can define one’s mous pair, one side is privileged to the terms once and for all rests on the detriment of the other. In his book Of assumption that there are such things (1967) Derrida illus- as meanings that can remain stable trated this by showing how one of the over a period of time and which could founding dichotomies of linguistics – be captured and ‘imprisoned’ in the namely, the one that opposes meaning form of rigorous definitions. Well, and the linguistic object (say, the among other things, deconstruction text)–is itself the result of a ‘textual is, as we have seen, concerned with prestidigitation’ at the hands of Ferdi- reminding us that there are no such nand de Saussure, the discipline’s stable meanings to begin with, and so founding father, who postulated it. the quest for definitions is a wild goose Initially presented as a symmetric pair, chase. Rather, the meaning of a given the distinction between signified and text is precisely what is iterable – a signifier reveals itself, upon closer coined by Derrida, meaning inspection, to be an hierarchical one. that which comes out different every This is because the very stability of time one repeats it, as bizarre as this that sign relation (the sign being, for might indeed appear at first blush. It is, Saussure, a relation of bond between to use another of Derrida’s coinages, a signifier and a signified) is predi- determined by différance, an amalgam cated upon there being one side of the of deferral and difference. relation (namely, the signified) not Derrida maintained throughout dependent on the other and thus capa- that deconstruction is a form of close ble of ‘standing alone’ – unlike the sig- reading, where all that one needs to nifier whose contribution to the sign mobilise is the text’s own internal relation is always that of ‘pointing to’ logic. In other words, texts decon- the other, namely the signified, con- struct themselves as it were when sub- ceived of as ‘auto-sufficient’ in that it jected to relentless close reading. All is the meaning, period. that a reader needs to do is to press Derrida extended his deconstructive ahead with its own internal logic. analysis to other prized dichotomies of That logic, Derrida insisted, is built such as speech around the notion of logocentrism, versus writing. He argued that the lin- the belief that there are stable mean- guists’ preference for speech to the ings out there and that it is the busi- detriment of writing (the ‘wandering

46 DEDUCTION/INDUCTION outcast of linguistics’) is yet another relentlessly and with a rigour it is demonstration of discipline’s complic- not used to, will ultimately turn ity with ’s trade- against itself. Derridean philosophy of mark logocentrism, which in the case anti-philosophy has impacted such of linguistics takes the form of ‘phono- movements as poststructuralism* and centrism’. Using a very subtle line of . But it is important to reasoning, Derrida maintained that register that it has, so to speak, been the argument frequently used to bol- ‘hijacked’ by both the political left and ster up the case of speech as the the right to serve their own respective privileged object of scientific linguis- agendas. tics – its spontaneity and authenticity – crumbles as soon as one realises that Primary sources it is based on the implicit idea that Culler, Jonathan (1982). On Deconstruc- speech is closer to the speaker’s tion: Theory and Criticism after Struc- intended meaning than writing is. turalism. Ithaca, NY: Cornell University Aristotle condemned writing for being Press. the imitation of what is already an imi- Derrida, Jacques (1967). Of Grammatol- tation. This means, Derrida insisted, ogy. Trans. Gayatri Spivak. Baltimore: that speech is always already being The Johns Hopkins University Press. conceived of along very much the Derrida, Jacques (1991). A Derrida same lines as writing (namely, the rep- Reader. Between the Blinds. Ed. Peggy resentation of something else), so that Kamuf. Hertfordshire: Harvester the very distinction between speech Wheatsheaf. and writing may be said to be based on Norris, Christopher (1982). Deconstruc- an unacknowledged notion of ‘arch- tion: Theory and Practice. London: writing’. Furthermore, the so-called Methuen. intended meanings are themselves being treated as though they were Further reading inscribed in the speaker’s mind/brain, Wood, David (ed.) (1992). Derrida: A ready to be scrutinised any time Critical Reader. Oxford: Blackwell. thanks to what is credited with being the distinguishing feature of writing, Kanavillil Rajagopalan its permanence over a period of time. Using analogous reasoning, Derrida argued that such hallowed dichotomies as nature versus culture, philosophy DEDUCTION/ versus literature, science versus myth, reason and unreason, and so forth – the INDUCTION bedrock of much of Western thought – would not survive deconstructive Deduction is a form of reasoning in scrutiny. This has, as he rightly fore- which one proceeds from general prin- saw, devastating implications for the ciples or to specific cases. Induc- very enterprise of philosophy. tion is a form of reasoning in which Deconstruction is a constant re- one arrives at general principles or minder that reason, when pursued laws by generalising over specific cases.

47 DEDUCTION/INDUCTION

See also: Empiricism/Rationalism; that these latter types of proposition Metaphor; Propositions; Type/ are less certain. He claimed that induc- Token; Universal Grammar tive generalisations are not well- Key Thinkers: Chomsky, Noam; founded in logic*: from the fact that Hume, David.; Locke, John; Mill, the sun has come up on all the preced- J. S.; Peirce, C. S.; Popper, Karl; ing days of our lives, it does not follow Russell, Bertrand; Sapir, Edward; logically that it will come up tomor- Whorf, Benjamin Lee row, or, more generally, that there is a valid general law which states that it The status and role of these two forms comes up every day. A clear example of of reasoning/ in human the uncertain nature of inductive gen- thought and language have been much eralisations is the case of the generali- discussed in the history of both philos- sation that all swans are white. From ophy and linguistics. No-one doubts the fact that all swans one has previ- that human beings are capable of both ously observed were white, it does not kinds of reasoning, but there has been follow that the next swan one will much debate as to exactly what role observe will be white. The discovery of they play in establishing human black swans in Australia showed that knowledge, including linguistic knowl- this particular inductive generalisa- edge. Central to these debates has been tion, once held to be true, was in fact the role of inductive generalisations. A false. simple example of an inductive gener- These features of inductive general- alisation is the generalisation that the isations make sun comes up every day. We arrive at different in kind from deductive rea- such generalisations by observing spe- soning, since deduction is founded in cific events, such as the sun coming up logic: in the syllogism ‘All men are on a specific day, and then generalising mortal; Peter is a man; therefore Peter over other events which are thought is mortal’, the conclusion that Peter is by the observer to count as instances of mortal follows logically as a deduc- the same type of event, for example, tion from the . Whether the the sun coming up on subsequent days. premises are well-founded is another The eighteenth-century philosopher matter: the argument is logically distinguished between (a) sound. The deduction in the follow- propositions concerning states of ing argument is equally well-founded: affairs that are certain (such as the (a) ‘All rabbits are carnivores’, proposition that 2 + 2 = 4), and (b) (b) ‘Rupert is a rabbit’, therefore propositions which are unlike such (c) ‘Rupert is a carnivore’. In stating mathematical propositions in that that inductive generalisations have no they rely on observed objects and basis in logic, one need not deny that events. The latter sorts of proposition such generalisations are well-founded include propositions based on induc- psychologically, and Hume’s theory of tive generalisations arising from the induction was indeed a psychological perception of those objects and events theory: he argued that repeated obser- (such as the proposition that the sun vation of the ‘same’ event causes us to comes up every day). Hume argued arrive at inductive generalisations. He

48 DEDUCTION/INDUCTION took that to be a central feature of is functioning on the basis of cate- human psychology. gories, and these are said by many to Hume’s theory of induction was play a central role in human percep- later re-examined in the twentieth tion, including . century by the philosopher Karl Central to the formation of inductive Popper (1963 and elsewhere), who generalisations and the formation of argued that no two (or more) events categories is the notion of similarity: (say, the observation of a flash of light- in order to take two events or objects ning) will ever be exactly the same, to be instances of the same thing, we and that we therefore need to appeal need to be able to perceive similarity to the idea of two (or more) events between those instances. That capac- counting as ‘the-same-for-us’ (or ity to perceive similarities may well be indeed, the same for a member of innate, and is arguably at the heart of another species, which may not be much of human cognition. It appears quite the same thing as it is for our to underlie our capacity to establish species). Popper argued that human similes and metaphors* which lin- beings are born with an innate guists such as George Lakoff take to propensity to seek out regularities, lie at the heart of human language. and that it is this propensity that There has been much discussion of underlies our capacity to form induc- the role of deduction and induction in tive generalisations. This is similar, if the field of child language acquisition. not identical, to Bertrand Russell’s Chomsky’s approach to the child’s lin- claim that induction is ‘incapable of guistic development is founded on being inferred from experience or his linguistic rationalism (otherwise from other logical principles’ (1946: known as linguistic nativism): he pos- 647). Popper’s claim is not to be con- tulates innate linguistic knowledge, fused with the doctrine of innate often referred to as an innate language ideas, associated with the work of, module or faculty, and also referred to among others, René Descartes in the by many as universal grammar*. The seventeenth century and Noam child is said by Chomsky to be born Chomsky in the twentieth century. with an innate set of universal linguis- The propensity to seek out regular- tic principles which are not, by defini- ities is, arguably, intimately connected tion, acquired by the child in his/her with our capacity to form categories, interaction with the mind-external and our capacity to take a specific world. The input (otherwise known as event on a specific occasion to be a ‘the stimulus’) to which the child is token of a type, in the sense invoked exposed is said by Chomsky and his by the twentieth-century American followers to be ‘impoverished’ in cer- philosopher Charles Saunders Peirce. tain ways: it is said to be full of If one takes, say, a specific flash of hesitations, false starts, utterances of lightning to be a token of the type sentences which are never completed, ‘flash of lightning’, or a specific speech and so on. Chomsky argues that, sound token (say, a [t] uttered on a given the sheer complexity of adult specific occasion) to be a token of the linguistic knowledge, it could not type ‘[t]’, then one’s perceptual system have been acquired on the basis of this

49 DEDUCTION/INDUCTION

‘impoverished’ stimulus alone. Put the course of that development. Since another way, adult linguistic knowl- Chomsky adopts a version of natural- edge is underdetermined by the input. ism, in which a postulated biologi- This is the cally-endowed module of mind, on argument for an innate linguistic exposure to the input, undergoes bio- module of mind. Chomsky has also logical growth, linguistic development appealed to what he calls the logical is said not to be something that the problem of language acquisition. The child does; rather, it is something that problem is said to be based on the dif- happens to the child. It is important to ficulty of arriving at fully-fledged adult note that this is an entirely passive linguistic knowledge on the basis of conception of the child’s linguistic the supposedly impoverished input/ development. The importance of this stimulus. The problem is said to be a lies in the fact that rationalists often logical one since the child is born in depict empiricist approaches to child of knowledge of the nature linguistic development as a passive of language, in the form of general lin- conception of first language acquisi- guistic principles and a set of parame- tion, a depiction that is at odds with ters, and is said to deduce the much present-day empiricism. Note properties of the ambient language. An too that appeal by rationalists to the example of a parameter is the position act of setting parameters and engaging of the verb in sentence structure. The in deduction runs counter to the claim child is said by Chomskyans to deduce that the acquisition process does not the parameter settings in universal amount to the child actively doing grammar, thus yielding the specific lan- anything to acquire a first language. guage he/she is being exposed to. Those who oppose Chomskyan Deduction is thus central to Chom- rationalism typically emphasise the sky’s vision of child linguistic develop- role of inductive and analogical gen- ment. Chomskyans, such as Smith eralisation in child language acquisi- (2004), argue that general learning tion. Examples of such generalisations mechanisms play a very limited role in are child expressions such as ‘Three that development. The capacity for sheeps comed’ (‘Three sheep came’), inductive generalisation forms part of in which the irregular ‘sheep’ our general learning mechanisms: we and the irregular ‘came’ are can form inductive generalisations in regularised. This suggests that the any cognitive domain, from knowing child has arrived at the regular rules about the seasons to knowing what for plural and past-tense formation the traffic might be like at a given in English via induction based on point in the day. Because Chomskyans repeated exposure to regular forms, do not take the child’s linguistic followed by analogical extension of development to constitute a kind of the inductive rule to irregular learning, they play down the role of and verbs. The emphasis on analogy inductive generalisation in child lin- and induction is central to the work of guistic development. Equally, they linguists such as , who play down the role of our capacity for seek to re-establish the importance of forming analogical generalisations in usage (performance, in Chomskyan

50 DEDUCTION/INDUCTION terms) in understanding the nature of he/she must acquire knowledge of a linguistic knowledge. Bybee (2001) set of social conventions, including argues that frequency of occurrence of linguistic conventions (such as know- specific words in language usage plays ing that the subject precedes the verb a major role in the synchronic state of in English sentences, or knowing that a given language, and in its histori- regular of nouns consist of the cal development. She distinguishes singular form with a specific kind of between token frequency, which is the suffix). The child can acquire these extent to which specific word-forms conventions by observing the conven- are uttered, and type frequency (see tional word orders used by other below). She argues that words with human beings in his/her linguistic high token frequency, such as the past- environment. Crucial to Tomasello’s tense form ‘kept’ in English, are more view of child language acquisition is resistant than words with lower token the social, interactive nature of the frequency (such as the verb ‘weep’) to child’s world, in particular the child’s the kind of analogical changes that coming to attribute intentions to the would yield regularised past-tense people that he/she is interacting with, forms such as ‘keeped’: lower-token along with the child’s ability to have frequency ‘weep’ is likely to regu- intentions towards those intentions. larise, via analogy, to ‘weeped’, Not all inductive reasoning is con- whereas higher-token frequency scious. While adults and children alike ‘keep’ is much less likely to be regu- can engage in explicit learning, involv- larised, via analogy, to ‘keeped’. Type ing conscious reasoning, there is a frequency is defined as the frequency form of implicit learning which even of occurrence of a specific pattern, very young infants (in the first year of such as the English past-tense ‘-ed’ life) can engage in. This kind of learn- pattern, which applies to the vast ing rests on the human capacity to majority of English verbs. High type extract probabilities from speech frequency is said to determine the pro- input, such as probabilities about ductivity of a specific pattern. In this likely sequences of words and speech case, the ‘-ed’ pattern is highly likely sounds. For instance, children in the to be applied to novel word forma- first year of life have been shown to tions (such as the past-tense verb form prefer made-up words which conform ‘googled’) and to borrowed verbs to the phonotactic constraints of the (such as ‘nuanced’). Type frequency, language they are being exposed to. like token frequency, is based on By ‘phonotactic constraints’ is meant inductive and analogical . the constraints on sequences of Some twenty-first-century child lan- sounds in the syllable structure of guage researchers, such as Michael words. The sequences /pr/, /pl/, /tr/, Tomasello, argue that induction plays /kr/, /kl/ are all permissible sequences a central role in child language acqui- at the beginning of words in English, sition. For Tomasello (2001), the child but the sequences /pn/, /ps/, /tl/ and possesses no innate linguistic knowl- many others are not. Infants in the edge at birth. Rather, the child is said first year of life prefer made-up words to be born into a world in which which contain the permissible, rather

51 DEDUCTION/INDUCTION than the non-permissible, sequences. form the basis This is because they have the capacity for Sampson’s empiricism. Interest- to tune in to the high frequency of the ingly, Sampson and Chomsky both permissible sequences and to extract stress the role of deduction in the the phonotactic patterns. Infants are child’s linguistic development, but for capable of this kind of probabilistic very different reasons, and in different (otherwise known as ) ways. Note that earlier claims by learning without conscious effort. The Chomskyans that the child is a little existence of this kind of unconscious , whose hypotheses are con- inductive learning has been used by strained by a postulated innate lan- opponents of Chomskyan nativism to guage acquisition device, run counter show that the stimulus which the child to the present-day view of Chomsky is exposed to is rich in information, that the child’s linguistic development which the child can access, rather than is entirely passive. impoverished, as claimed by support- Appeal to the hypothetico- ers of linguistic nativism. deductive method in the child’s con- One of Chomsky’s fiercest critics, ceptual and linguistic development the British linguist Geoffrey Sampson, can be found in current work by adopts an empiricist approach to Alison Gopnik. She proposes a child language acquisition. Sampson, ‘theory theory’ of : like Tomasello, and unlike Chomsky, the child is said to be actively con- argues that children learn language, structing theories about the surround- and that they do so in the same way as ing world, including the ambient they learn anything else: using general language. Gopnik (2001) reports on learning mechanisms. The most cen- experiments which seem to show that tral part of child language acquisition, certain stages in the child’s conceptual for Sampson, is the hypothetico- development, such as the capacity to deductive method: the child is said to grasp the idea of a means towards an come up with hypotheses as to the end, come on stream just ahead of the nature of the language he/she is being relevant child’s expressions denoting exposed to, and then uses the capacity such notions. This work seems to sug- for deduction to deduce testable gest that, for such stages in concep- claims about the structure of that lan- tual development, it is the concept guage. These are then said to be tested which comes first, followed by an against the data the child is exposed utterance-type for that concept. to. The child can then modify or aban- Gopnik also argues for a version of don his/her hypotheses about the the Sapir-Whorf hypothesis, accord- structure of that language, and thus ing to which aspects of the structure come to learn the language. This is the of the ambient language will induce view that the child is ‘a little scientist’ the child to conceptualise certain in his/her attempts at coming to grasp kinds of events and objects in specific the structure of his/her native lan- ways. For instance, in a language guage. This emphasis on the hypo- such as Korean, ellipsis of nouns is thetico-deductive method is central to much more common than in Eng- the work of Popper, whose ideas in the lish. Because of this, a child of fifteen

52 DEDUCTION/INDUCTION to twenty-one months acquiring remains a central topic of research Korean monolingually will be delayed and debate in contemporary language in the ‘naming explosion’ stage, and study. on categorisation tasks, in compari- son with a child of the same age Primary sources acquiring English monolingually. But Bybee, Joan (2001). Phonology and Lan- these Korean-acquiring infants are guage Use. Cambridge: Cambridge Uni- advanced on means-ends abilities in versity Press. comparison with English-acquiring Chomsky, Noam (2000). New Horizons in infants of the same age. Gopnik the Study of Language and Mind. Cam- claims that these differences in devel- bridge: Cambridge University Press. opment are the result of the child Gopnik, Alison (2001). ‘Theories, lan- using the hypothetico-deductive guage and culture: Whorf without winc- process to arrive at conclusions about ing’. In M. Bowerman and S. Levinson the shape of the ambient language (eds), Language Acquisition and Con- and the nature of the world they ceptual Development. Cambridge: inhabit. Cambridge University Press. Current work by Annette Hume, D. (1748). An Enquiry Concerning Karmiloff-Smith (1998), who works Human Understanding. Reprinted in A. in the tradition established by Jean Flew (ed.) (1962), David Hume On Piaget, adopts a version of construc- Human Nature and The Understand- tivism with respect to child develop- ing. London: Collier Macmillan. ment. She argues that there is no Karmiloff-Smith, Annette (1998). ‘Devel- innate language module, but there opment itself is the key to understand- may well be innate in different ing developmental disorders’. In Trends cognitive domains (such as the recog- in Cognitive Sciences 2. Reprinted in nition of familiar faces). These biases M. Tomasello and E. Bates (eds) are said to develop with training on (2001), : The the environment: the child’s percep- Essential Readings. Oxford: Blackwell, tual input is said to shape those biases 331–50. into domain-specific cognitive capaci- Popper, Karl R. (1963). Conjectures and ties which appear modular in nature Refutation. London. Routledge and (such as the face recognition module). Kegan Paul. Inductive generalisations play a role in Russell, Bertrand (1946). The History of this approach to child development, Western Philosophy. London: Rout- since the child’s initial biases are said ledge (2004). to become increasingly richly devel- Sampson, Geoffrey (1997). Educating oped as a result of training on specific Eve: The ‘Language Instinct’ Debate’. sorts of input, where repeated expo- London: Cassell. sure to specific kinds of sensory event Sapir, Edward (1921). Language: An results in the forming of inductive gen- Introduction to the Study of Speech. eralisations in that domain. New York: Harcourt Brace. The issue of the respective roles of Tomasello, Michael (2001). ‘Perceiving induction and deduction in the intentions and learning words in the construction of linguistic knowledge second year of life’. In M. Tomasello and

53 DEFINITE DESCRIPTIONS

E. Bates (eds), Language Development: Key Thinkers: Frege, Gottlob; The Essential Readings. Oxford: Black- Russell, Bertrand; Strawson, P. F. well. 132–58. In 1905 Bertrand Russell published an Further reading article drawing attention to what he Pinker, S. (1994). The Language Instinct. saw as some specific and philosophi- London: Penguin. cally interesting properties of definite Pullum, G. K. and B. Scholz (2002). descriptions. Phrases introduced by ‘Empirical assessment of stimulus ‘the’ appear to indicate not only the poverty arguments’. The Linguistic existence but also the uniqueness of Review 19: 9–50. some entity. In this respect they can be Sampson, G. (2002). ‘Exploring the rich- contrasted with indefinite descrip- ness of the stimulus’. The Linguistic tions; ‘the leader of the free world’ Review 19: 73–104. conveys the idea that there is only one Scholz, B. and G. K. Pullum (2002). such person, while ‘a leader of the free ‘Searching for arguments to support lin- world’ does not. According to Frege’s guistic nativism’. account of meaning, definite descrip- 19: 185–223. tions should be treated as names*, Smith, Neil (2004). Chomsky: Ideas and having both a sense and a reference. In Ideals. Second edition. Cambridge: opposition to this analysis, Russell Cambridge University Press. draws attention to the problem of def- inite descriptions that do not have any Philip Carr actual reference in the world; ‘a phrase may be denoting, and yet not denote anything’ (1905: 471). Russell discusses this problem in DEFINITE relation to his now-famous example ‘the present king of France is bald’. DESCRIPTIONS According to Frege’s account, this sen- tence expresses a simple proposition Denoting phrases that are introduced of subject-predicate form. Since the by the definite article ‘the’, or by a subject fails to refer, the proposition as possessive such as ‘my’ or ‘Sherlock a whole should fail to have a truth Holmes’s’. Bertrand Russell’s logical value*; it should be . Russell analysis of sentences containing argues against this reading on the definite descriptions has been highly grounds that ‘it is not nonsense, since influential, and highly controversial, it is plainly false’ (1905: 484). This throughout the past century. can be explained, he claims, once we realise that definite descriptions do See also: Analytic Philosophy; not operate in the same way as names. Connotation/Denotation; Logic; In fact, it is not appropriate to dis- Logical Form; Names; Ordinary cuss their meaning in isolation; the Language Philosophy; meaning of a definite description can Presupposition; Propositions; be considered only in relation to the Sense/Reference sentence in which it appears. The

54 DEFINITE DESCRIPTIONS presence of the definite description that where necessary the messy and ensures that the logical form of the imperfect constructions of natural sentence is a complex set of proposi- language must be ‘translated’ into log- tions that concern existence and ically correct form before they can uniqueness as well as baldness. The provide suitable subject matter for logical form of ‘the present king of philosophical analysis. This has been France is bald’ could be paraphrased the cause of a number of responses to as ‘there is an entity x such that x is the Russell. Perhaps the most significant present king of France, no entity that of these has been that put forward by is not equivalent to x is the present Peter Strawson in his 1950 article king of France, and x is bald’, or more ‘On referring’. Coming from a back- informally ‘there is one, and only one, ground in ordinary language philoso- king of France, and he is bald’. phy* (OLP), Strawson criticises This analysis allowed Russell to Russell for being obsessed with logic explain his conviction that the sen- and mathematics, at the expense of tence is false. The simple proposition attending to the of natural that a present king of France exists is language. Russell was overlooking false. This is sufficient to make the important facts about the ways in logical form* of the sentence, which is which speakers actually use expres- a coordination of this and two other sions containing definite descriptions; propositions, false also. The analysis as Strawson puts it, ‘“mentioning”, or also allowed Russell to maintain a “referring”, is not something an classical bivalent logic* for language. expression does; it is something that If ‘the present king of France is bald’ someone can use an expression to do’ is false, then logic dictates that ‘the (1950: 326). present king of France is not bald’ Strawson claims that someone must be true. This is indeed the case if encountering an utterance of ‘the the is read as having present king of France is bald’ would over the whole of the logical form: as not be likely to reply ‘that’s false’, as Russell paraphrases it, the negative Russell’s account would seem to pre- sentence can be read as ‘it is false that dict. Rather they would be stuck for a there is an entity which is now king of response of any kind, feeling that France and is bald’. The negative sen- there was something badly wrong tence is ambiguous, however. On with this statement; the question of another reading it can be interpreted whether it is true or false would just as ‘there is an entity which is now king not arise. For Strawson, the proposi- of France and is not bald’, and on this tion that there exists a present king of reading it is as false as its positive France is not part of the logical form equivalent. of the statement, but is a presupposi- Russell’s theory is very much a tion* attached to its use. It is neces- product of his analytic approach to sary for this presupposition to be philosophy. It draws attention to what fulfilled for the sentence to be either he sees as a discrepancy between the true or false. Hence both ‘the present grammatical form of a sentence and king of France is bald’ and ‘the pres- its logical form. The implication is ent king of France is not bald’ must be

55 DEFINITE DESCRIPTIONS categorised as neither true nor false. chairman of the teetotallers’ society is By allowing that some statements can informed that someone is drinking a lack a truth value* in certain contexts cocktail at their Christmas party, he of use, Strawson was arguing that nat- might ask ‘Who is the man drinking a ural language did not conform to the martini?’ in order to establish the usual laws of logic; it should be stud- identity of the culprit. In this case the ied and analysed on its own terms. attribute ascribed to the individual is Russell strongly rejected this conclu- all-important. If it turned out that the sion, arguing in a response to Straw- individual in this case was drinking son that ‘common speech’ must be only water, then there is no possi- modified before it is fit for philosophy. ble right answer to the chairman’s He points out that ‘my theory of question. Donnellan also discusses descriptions was never intended as examples of definite descriptions an analysis of the state of mind of introduced not by ‘the’, but by pos- those who utter sentences containing sessives: examples such as ‘Smith’s descriptions’ (1957: 388). murderer is insane’. He argues that, Keith Donnellan (1966) also criti- for different reasons, neither Russell’s cises Russell for being unable to nor Strawson’s account can ade- account for the intentions and pur- quately handle the distinction poses with which people actually use between referential and attributive definite descriptions in everyday lan- uses of definite descriptions. guage. He draws attention to a dis- Despite criticisms such as these, tinction between two different ways in Russell’s theory of definite descrip- which definite descriptions are used, tions has been widely regarded as an the referential and the attributive uses. exemplary application of analytic phi- Used referentially, definite descrip- losophy*. It is a measure of the con- tions pick out an individual in order to tinuing importance of ‘On denoting’ say something about that individual. in the philosophy of language that a A person who uses a definite descrip- special edition of Mind was devoted tion attributively ‘states something to commemorating the centenary of about whoever or whatever is the so- its publication. At the start of his and-so’ (1966: 285). If someone sees a contribution to this edition, Stephen distinguished-looking man drinking Schiffer comments that Russell’s from a martini glass at a party and article not only contributed to defin- asks ‘Who is the man drinking a mar- ing a new philosophy of language, but tini?’, that person is using the definite also remains the dominant theory of description referentially. In this case definite descriptions in the present the particular attribute ascribed to the day. individual is not very important; the speaker would still successfully have Primary sources referred to the distinguished-looking Donnellan, Keith (1966). ‘Reference and man even if it turned out that he had definite descriptions’. Philosophical only water in his glass. The same defi- Review 75: 281–304. nite description, in the same sentence, Frege, Gottlob (1892). ‘On sense and could also be used attributively. If the meaning’. In Peter Geach and Max

56 DESCRIPTIVISM

Black (eds) (1980), Translations from characterising – not judging or criticis- the Philosophical Writings of Gottlob ing – language produced within a par- Frege. Oxford: Blackwell. (First edition, ticular speech community. 1952.) 56–78. See also: Langue/Parole; Russell, Bertrand (1905). ‘On denoting’. Structuralism; Mind 14: 479–99. Transformational-Generative Russell, Bertrand (1957). ‘Mr. Strawson Grammar on referring’. Mind 66: 385–9. Key Thinkers: Bloomfield, Strawson, P.F. (1950). ‘On referring’. Leonard; Boas, Franz; Brugmann, Mind 59: 320–44. Karl; Chomsky, Noam; Halliday, M. A. K.; Hockett, Charles; Labov, Further reading William; Pike, Kenneth; Rask, Bezuidenhout, A. and M. Reimer (eds) Rasmus; Sapir, Edward; Saussure, (2004). Descriptions: Semantic and Ferdinand de Pragmatic Perspectives. Oxford: Oxford University Press. Descriptivism became an important Devitt, Michael and Kim Sterelny (1999). trend in linguistics after 1900. Its the- Language and Reality. Second edition. ories and principles support an open Cambridge, MA: MIT Press. Chapter 3. attitude toward language and linguis- Evans, Gareth (1982). The Varieties of tic study. It is opposed to prescrip- Reference. Oxford: Oxford University tivism, the term most often used to Press. refer to a linguistic school of thought Ostertag, G. (1998). Definite Descrip- in which individuals seek to promote tions: A Reader. Cambridge, MA: MIT one particular variety of a language, Press. formulate its rules, and enforce adher- Schiffer, Stephen (2005). ‘Russell’s theory ence to those rules. Descriptive lin- of definite descriptions’. Mind 114: guists strive to present a picture of 1135–83. language as complete as possible as it actually exists at a specific point in Siobhan Chapman time and place; they first describe observable facts about a particular spoken and written language and note generalisations about that informa- DESCRIPTIVISM tion. They then draw conclusions about that language and tie their Primarily a synchronic method of lin- conclusions to an analysis of human guistic analysis in which the structure language in general. and variation of written and spoken Nineteenth-century European lin- language are portrayed dispassion- guists such as Karl Brugmann and ately and non-judgementally. Advo- moved the study of cates of descriptivism, borrowing from language away from ‘fanciful’ notions methodology employed by linguists of philosophy or folklore to a more in the nineteenth century, furthered disciplined analysis of written lan- the development of linguistics as a sci- guage. Although these scholars con- ence and focused on cataloguing and centrated on diachronic studies of

57 (CRITICAL) DISCOURSE ANALYSIS

European languages, their systematic, Primary sources scientific approach appealed to early Bloomfield, Leonard (1933). Language. twentieth-century linguists such as New York: Henry Holt. , and Ferdi- Hockett, Charles Francis (1958). A nand de Saussure, who refined the Course in Modern Linguistics. New methodology and used it to present York: Macmillan. synchronic analyses of various lan- guages, including those of North Further reading American Indian tribes. Descriptivism Harris, Zellig (1951). Methods in Struc- is especially associated with Leonard tural Linguistics. Chicago: University of Bloomfield, Charles Hockett, William Chicago Press. Labov and Kenneth Pike, who estab- Hudson, R. A. (1980). Sociolinguistics. lished procedures for examining Cambridge: Cambridge University phonology, morphology, syntax and Press. semantics, and who insisted that lin- Newmeyer, Frederick J. (1996). Genera- guists eliminate value judgements tive Linguistics. London: Routledge. from their studies and avoid social, Pedersen, Holger (1931). Linguistic Sci- cultural and moral grandstanding. ence in the Nineteenth Century: Meth- They believed that linguists should ods and Results. Trans. John Webster observe actual speech in linguistic Spargo. Cambridge, MA: Harvard Uni- communities, form hypotheses about versity Press. common trends in language and inter- relationships between various parts of David V. Witkosky language, and test their hypotheses. The heyday of descriptivism span- ned the years 1930 to 1960. After- wards, an increasing number of (CRITICAL) linguists found the approach too lim- DISCOURSE iting. For example, in the 1960s Noam Chomsky and supporters of ANALYSIS transformational-generative linguis- tics expressed the view that scholars A term applied to a broad and hetero- should investigate speakers’ uncon- geneous range of approaches to scious awareness and knowledge of language which share a family their language’s abstract system of resemblance in their focus on the lin- rules. In the 1970s M. A. K. Halliday guistic characteristics, organisational and other sociolinguists asserted that patterns and communicative func- language could not be studied and tions in context of naturally-occurring explained without examining the text, be it spoken or written (or structure of the society in which it is signed). It has been argued that the spoken. Regardless of reservations or centre of gravity of linguistic study criticisms, the openness, inclusiveness has shifted over the past thirty years and rigour of descriptivism continue from the sentence to the text, that is, to provoke and inform linguistic from syntax to discourse; this shift debate into the twenty-first century. reflects an increasing acceptance that

58 (CRITICAL) DISCOURSE ANALYSIS linguistics should be concerned with remains a central focus. One con - describing and explaining language in sequence of this openness to other use. disciplines, however, is that it is sometimes difficult to trace the limits See also: Conversation Analysis; of what is and is not included in Corpora; Implicature; discourse analysis. Integrationism; Langue/Parole; This indeterminacy is com- Politeness; Relevance Theory; pounded by the inherently multifac- Systemic-Functional Grammar; eted nature of discourse, which can be Speech Act Theory approached at a micro level, such as Key Thinkers: Austin, J. L.; exploring how each utterance in turn Bakhtin, Mikhail; Cameron, relates to the next in an interaction, at Deborah; Grice, H. P.; Halliday, a macro level, for example investigat- M. A. K; Labov, William; Sacks, ing the discourse of medicine from the Harvey; Sinclair, John; Tannen, perspective of its role as a form of reg- Deborah ulative control within a society, or Discourse analysis is often defined as anything in between. It is reflected in the analysis of language beyond the the number of terms which can be sentence. This is over-simple, but it seen as alternative labels (although has the merit of making clear how it many of the practitioners would stress differs from traditional linguistics the differences): discourse studies, centred around syntax, since the sen- text analysis, , register tence is accepted as the maximum analysis, genre analysis and so on. It is domain within which syntactic struc- also reflected in uncertainty over the tures operate. A further crucial dif- boundaries between pragmatics and ference is that, whereas theoretical discourse analysis, and over whether linguistics has, since Ferdinand de approaches such as speech act Saussure, insisted on the theory* or conversation analysis* of linguistics and attempted to isolate come under the umbrella of discourse it from other fields by strictly cir- analysis, or are compatible but sepa- cumscribing its domain, discourse rate fields of inquiry (the plurality of analysis is strongly oriented towards views can be seen simply by typing . It makes contact, ‘discourse analysis’ into an Internet and partly overlaps, with a wide search engine such as Google and spectrum of other social sciences opening the first five or six links in the including social and cognitive psy- list). It is therefore perhaps best to chology, sociology, , think of discourse analysis less as a anthropology, pedagogy and com- unified approach and more as a pre- munication studies. In turn, it has disposition on the part of the linguist been applied in those other fields to aim to relate language forms to (and in some cases it has then lost their contexts of use. What is striking almost all characteristics that relate is that this predisposition has come to it to linguistics as a discipline). A dominate in a way core feature of ‘linguistic’ discourse that would have seemed unlikely in analysis is that the language itself the early days when the structural

59 (CRITICAL) DISCOURSE ANALYSIS

linguistics of Bloomfield, Chomsky (‘non-structural’ in the sense that they and others was exerting its hegemony. cannot be adequately described in The first use of the term ‘discourse terms of clause structures such as analysis’ is generally ascribed to Zellig subject-verb-object). These include Harris in a 1952 paper. However, his conjuncts (for example, ‘However’), paper did not set an agenda that dis- reference items (for example ‘These’ course analysis has followed since. at the start of the present sentence Although he focused on structure referring back to ‘links’ in the preced- above the sentence, his aim was to ing sentence), and repetition (for extend existing methods for analysing example, the phrase ‘language beyond sentence structures, based purely on the sentence’ above is repeated from distribution with no recourse to an earlier paragraph to signal the con- meaning, to the analysis of texts. A tinuity of topic). Halliday and Hasan’s pioneer whose contribution is recog- definition of ‘text’ (1976: 2) marked a nised as more relevant is T. F. Mitchell conscious departure from the con- who, in a 1957 article, identified func- cerns of structural linguistics: ‘A text tional stages in interactions between is best regarded as a semantic unit: a buyers and sellers in a particular cul- unit not of form but of meaning.’ tural context. This study had certain Although much work was done key features that came to be charac- from the late 1960s on that laid the teristic of much discourse analysis: it foundations for discourse analysis, analysed naturally-occurring text; it the first explicit introduction to the worked at text level; it explored how field was Malcolm Coulthard’s 1977 discourse is organised to carry out the book, which was soon followed by communicative purposes of the inter- other ground-breaking volumes by actants, particularly by highlighting Beaugrande and Dressler (1981), the ways in which each stage served Brown and Yule (1983) and Stubbs different functions but all contributed (1983). There are naturally areas of to the achievement of the interactants’ overlap, but the differences between goals; it demonstrated that such inter- these four books reflect some of the actions are not randomly recreated most important directions in which each time but follow conventions that discourse analysis developed. For are accepted as appropriate by, and Coulthard, whose treatment was reflect the norms of behaviour of, inspired by the findings of an earlier members of that culture; and it identi- study that he and John Sinclair had fied linguistic signals that distin- carried out into the classroom lan- guished each stage of the interaction. guage of teachers and pupils, ‘dis- One of the most significant start- course’ essentially refers to spoken ing points in the study of language interaction: he draws on work in beyond the sentence was the publica- speech act theory* and other areas tion of and Ruqaiya of pragmatics, and in conversation Hasan’s 1976 book on cohesion. This analysis*, as well as work by Hymes explored the non-structural cohesive (1974) and others on the ethnography links which play a central part in of speaking, and by Labov (1972) and tying sentences together into a text others on sociolinguistics. Coulthard

60 (CRITICAL) DISCOURSE ANALYSIS is particularly interested in the methodological issues of collecting, structure of exchanges (multi-part transcribing and analysing data. Such stretches of conversation that form issues are generally of little relevance functional units, such as question- to theoretical linguists, who rely on answer-acknowledgement), and the intuition; but they are crucial for dis- ways in which utterances constrain course analysts, for whom data is the what can follow them, and provide starting point of inquiry. the context in which the following Between them, the four books thus utterances make sense. Beaugrande cover ‘text’-oriented aspects of speech and Dressler, on the other hand, focus (how utterances function in relation on the general characteristics of texts, to each other) and writing (what with more attention to written text. makes a text hang together as a text), They outline seven ‘standards of - ‘process’-oriented aspects (how pro- tuality’ which distinguish texts from ducers and receivers construct non-texts. They argue that, to be meaning collaboratively) and ‘con- accepted as a text, a stretch of dis- text’- oriented aspects (how discourse course must have qualities such as reflects and constructs social contexts cohesion (largely in Halliday and of use). These four lines of inquiry, Hasan’s sense), coherence (recognis- sometimes pursued separately but able unity of content) and situational- often combined in the same study, ity (recognisable appropriacy to the were to provide rich ground for dis- communicative situation). Brown and course analysis over the following Yule establish a distinction between decades. text-as-product and discourse-as- At around the same time as these process: that is, they see ‘text’ as the volumes were appearing, further record of a process of communication, impetus for the focus on discourse while ‘discourse’ is the process itself. came from applied linguistics and This leads them to place emphasis more specifically, from the teaching of on the psycholinguistic processing English for Specific Purposes (ESP), aspects of discourse: the ways in which expanded dramatically from which producers organise their mes- the 1980s on. Teachers trying to help sages so as to guide receivers to the their students cope with, for example, intended meaning, and receivers use writing academic articles in English, their background knowledge and soon realised that it was not enough to powers of inference to construct teach them to produce grammatically meaning. Finally, as the subtitle of his correct sentences. They needed to be book (‘the sociolinguistic analysis of shown how to write texts in the ways natural language’) indicates, Stubbs is that were conventionally accepted in concerned with the ways in which dis- their field. This involved, among course can only be fully analysed if it many other things, knowing how to is situated in its sociocultural context. organise their text following accepted Like Coulthard, he explores speech patterns, such as the IMRAD (intro- acts and exchange structure, but duction, methods, results and discus- he also highlights the ethnographic sion) format of scientific research dimension, and devotes much space to articles, and how to deploy language

61 (CRITICAL) DISCOURSE ANALYSIS choices appropriately in different successfully and in order for this stages of the text – for example, in an process to be implemented effectively, academic paper, using tense in the lit- the genres need to be analysed and erature review to show the degree of described in a way that simultane- centrality of cited work to their own ously uncovers the functions that lan- research, or using modality* to adjust guage serves in a particular genre and the strength of claims in the discussion shows the place of that genre in rela- section. Since did tion to other genres in the culture. not cover such topics, many ESP The stages (or, in Swales’s terms, the teachers became discourse analysts in moves) of a genre are identified in order to work out what their students terms of their function in achieving needed to know: they were enthusias- the overall communicative aim. The tic consumers of the growing research more delicate description of the lan- in the field, and they applied it and fed guage choices in the stages then their insights back into the discipline, involves register analysis. A register is enriching and extending it. Probably seen as a variety of language associ- the best-known example of the ESP ated with particular communicative discourse analysis tradition is John contexts which members of a culture Swales’s 1990 book Genre Analysis, recognise as recurring in accepted pat- based on research going back to the terns: for example, classroom interac- late 1970s, but this is only one repre- tions, school textbooks, recipes, news sentative of the huge body of work broadcasts, research articles and so arising from the educational needs of on. All registers use the same language learners of English. system (for example, research articles At the same period, the educational and recipes in English are recognis- context in Australia inspired a slightly ably in the same language), but the different kind of discourse analysis, probabilities of occurrence of particu- also under the label ‘genre analysis’. lar configurations of grammatical and Jim Martin and his colleagues (for lexical choices from the system vary example, in Martin 1985/1989) from register to register. To take a applied systemic-functional gram- simple example, recipes (in the mar* in investigating the genres that instruction stage) have a markedly pupils and students are required to high number of imperatives of verbs master in their . Their of action (‘mix’, ‘add’, and so on); work has a very strong sociocultural research articles, on the other hand, orientation: Martin’s definition of have many fewer imperatives, and the genre is that it is ‘a staged, goal- verbs in these cases are from a small oriented, purposeful activity in which conventional set referring to mental speakers engage as members of our processes (‘see’, ‘consider’, and so on). culture . . . Culture seen in these terms Martin and Rose (2003) is an accessi- can be defined as a set of generically ble introduction to this view of genre interpretable activities’ (1985: 25). and register. Education is in essence the process of By the mid-1980s, discourse analy- helping children to extend the range sis had already developed much of of genres in which they can operate its all-encompassing nature. It was

62 (CRITICAL) DISCOURSE ANALYSIS

certainly losing overall coherence tity and norms (that is, discourse does through the proliferation of often- not simply reflect existing external conflicting methodologies and ‘facts’, it plays a central role in con- assumptions about language, society structing the world as we see it). How- and the main goals of the analysis, but ever, CDA goes further in focusing at the same time it was gaining especially on the role of discourse in strength through the sheer range of constructing and reproducing social topics that could be covered and inequalities as embodied, for exam- through the increasing sophistication ple, in racial and sexual discrimina- of the analytical methods on offer. A tion. Certain forms of discourse are sense of the diversity can be gained by available to those with power, while inspecting the spectrum of areas cov- other forms of discourse construct the ered in surveys of discourse analysis users as (relatively) powerless. To take such as those edited by van Dijk a simple example, each time someone (1985) and by Schiffrin et al. (2003). fills in a tax return, providing the The diversity can also be seen by information required in the format comparing publications such as (to specified, they are accepting and take just two examples fairly ran- thereby reaffirming the authority of domly) Georgakopoulou and Goutsos the controlling their behav- (1997), which keeps to a firmly text- iour. This seems completely natural, linguistic approach and highlights and most people do not take a resist- especially the linguistic and organisa- ant attitude, whatever their feelings tional differences between narrative about paying tax. The discourse and non-narrative discourse, and embodied in tax forms and the like Wetherell et al. (2001), a reader has become so naturalised that we no design ed to introduce discourse analy- longer notice the implications, in sis to students of the social sciences terms of reinforcing the unequal dis- and oriented strongly towards ethno- tribution of power, of doing as we are graphic and socio-cultural issues. The told – that is, of playing our assigned discipline is also served by an increas- roles in a discourse event. The set of ingly wide selection of journals naturalised assumptions about the including Discourse Studies, Dis- ‘state of things’ with which language course and Society, Text & Talk and users work, and which are reflected in others. the language they use, are seen as One of the more recent manifesta- making up the users’ . The tions of discourse analysis that is more inequality is built into one’s ide- worth special mention has been in the ology – that is, the less obvious it is – form of critical discourse analysis the more difficult it is to resist or even (CDA), as developed by scholars to question. CDA practitioners set such as (1995), themselves the task of using various Teun van Dijk and . This forms of analysis to bring to light the starts from the basic constructionist ideological assumptions underlying assumption in most discourse analysis any discourse, particularly those that discourse inherently functions to which embody dominant forces in the construct and reproduce social iden- society.

63 (CRITICAL) DISCOURSE ANALYSIS

As the above outline makes clear, it Martin, J. R. (1985). Factual Writing: is difficult in the scope of this article to Exploring and Challenging Social Real- provide a straightforward overview of ity. Geelong, Australia: Deakin Univer- what discourse analysis is. The lack of sity Press. Republished by Oxford a clearly-defined domain has been the University Press 1989. focus of much criticism over the years, Schiffrin, Deborah, and together with the complaint (however Heidi E. Hamilton (2003). The Hand- elegantly expressed) that discourse book of Discourse Analysis. Oxford: analysis is not ‘real linguistics’ but Blackwell. simply impressionistic . It Stubbs, Michael (1983). Discourse Analy- is certainly true that the centrality of sis: The Sociolinguistic Analysis of Nat- theoretically-grounded analysis of ural Language. Oxford: Basil Blackwell. language is sometimes overlooked: Swales, John (1990). Genre Analysis: Eng- Antaki et al. (2002) is both a useful lish in Academic and Research Settings. summary of the problems and a Cambridge: Cambridge University reminder that they need to be avoided. Press. However, the achievements of dis- Van Dijk, Teun A. (ed.) (1985). Handbook course analysis have proved its of Discourse Analysis (four volumes). robustness: they have shown that it is London: Academic Press. as rigorous as other social sciences in Wetherell, Margaret, Stephanie Taylor its study of language in its contexts of and Simeon J. Yates (2001). Discourse use, and that this can provide Theory and Practice: A Reader. uniquely valuable insights into how London: Sage. we conduct our lives through lan- guage. Further reading Antaki, Charles, Michael Billig, Derek Primary sources Edwards and Jonathan Potter (2002). Beaugrande, Robert de and Wolfgang ‘Discourse analysis means doing analy- Dressler (1981). Introduction to Text sis: a critique of six analytic shortcom- Linguistics. London: Longman. ings’. Discourse Analysis Online 2, Brown, Gillian and George Yule (1983). http://www.shu.ac.uk/daol/articles/ope Discourse Analysis. Cambridge: Cam- n/2002/002/antaki2002002-paper.html bridge University Press. (accessed 11 October 2007). Coulthard, Malcolm (1977). An Introduc- Harris, Zellig (1952). ‘Discourse analysis’. tion to Discourse Analysis. London: Language 28: 1–30. Longman. Second edition 1985. Hymes, Dell (1974). ‘Ways of speaking’. Fairclough, Norman (1995). Critical Dis- In R. Baumann and J. Sherzer (eds) course Analysis. Harlow: Longman. Explorations in the Ethnography of Georgakopoulou, Alexandra and Diony- Speaking. Cambridge: Cambridge Uni- sis Goutsos (1997). Discourse Analysis: versity Press. 433–52. An introduction. Edinburgh: Edinburgh Labov, William (1972). Sociolinguistic University Press. Patterns. : University of Halliday, Michael A. K. and Ruqaiya Pennsylvania Press. Hasan (1976). Cohesion in English. Martin, J. R. and David Rose (2003). London: Longman. Working with Discourse: Meaning

64 DISTINCTIVE FEATURES

Beyond the Clause. London and New developed these ideas to propose what York: Continuum. became, through further reworking Mitchell, T. F. (1957). ‘The language of by Noam Chomsky and , buying and selling in Cyrenaica’. Hes- the standard model. peris 44: 31–71. Although the notions were implicit in earlier phonetic description, Nikolai Geoff Thompson Trubetzkoy (1939) focused attention on phonology’s subsegmental level. Trubetzkoy described ‘oppositions’ be- tween , invoking the idea that DISTINCTIVE phonemes are characterised by the contrasts they exhibit in languages. FEATURES American structuralists also spoke of pho nemes’ features, principally to op- The smallest units of linguistic struc- pose the distinctive and non- distinctive ture, from which larger units are built, properties of languages’ in divisible sometimes seen as the attributes by phonemes. Jakobson revised these which phonemes* can differ. The idea largely language-specific notions and is fundamental in phonology, where sought a small language-universal set of many generalisations are standardly features, which exist independently of stated in terms of features. the segments that they compose. Trubetzkoy focused on distinctive See also: Generative Phonology; oppositions – those which signal Phoneme; phonological contrast. These could be Key Thinkers: Chomsky, Noam; privative (a marked property is either Jakobson, Roman; Trubetzkoy, present or absent), equipollent (both N. S. members are of equal status) or grad- One of the few areas of phonological ual (with several gradations of one consensus is that segments are com- property). All Jakobson’s features were posed of features. Features play cru- arguably equipollent, with two values, cial phonological roles, being used (1) each characterising a definite property to express how segments contrast (for example, tense/lax, nasal/oral). with each other and (2) what groups Jakobson’s work (some collaborative, of segments (‘natural classes’) for- clearly expressed in Jakobson, Fant mally have in common, (3) to model and Halle 1952) based features prima- what changes in phonological pro- rily on segments’ acoustic properties. cesses in generative phonology*, and Chomsky and Halle (1968) redefined (4) in the formulation of constraints in features using principally articulatory optimality theory*. The Prague Lin- definitions, and used them in phono- guistic Circle provided the first logical rules, setting the scene for detailed expression of the ideas standard generative phonology. They behind feature theory, and relevant reinforced features’ binary nature, ideas occur in American structuralist using plus and minus values (for exam- work. , seen as the ple, [+back], [–nasal]) in underlying father of theory, representations and virtually all rules.

65 EMIC/ETIC

Phonologists have since refined the people producing that behaviour, set of features, or sought structure in while etic refers to categorising behav- their organisation (to account for iour from the perspective of an out- group behaviour in processes). Fea- sider in ways that are applicable to tures are now linkable to multiple seg- different systems and can be used to ments, and sometimes reinterpreted as compare them. Since the 1950s the exclusively privative (and given differ- terms have gained currency in the ent names, such as ‘autosegments’, and social sciences, espe- ‘components’ or ‘elements’). Theory is cially in the fields of anthropology and fundamental in phonology: certain cross-cultural psychology. currents now push for a less categori- See also: Phoneme cal notion of feature, but these small- Key Thinkers: Pike, Kenneth; est linguistic units are in no danger of Sapir, Edward being split further. The emic/etic distinction is part of Primary sources Pike’s theory of tagmemics, according Chomsky, Noam and Morris Halle to which verbal and nonverbal aspects (1968). The Sound Pattern of English. of behaviour are inseparable and must New York: Harper and Row. be studied in tandem, for the signifi- Jakobson, Roman, Gunnar Fant and cance of the one frequently can only Morris Halle (1952). Preliminaries to be discovered through reference to the Speech Analysis. Cambridge, MA: MIT. other. Pike saw the compartmentalisa- Trubetzkoy, N. S. (1939). Grundzüge der tion of levels of analysis prevailing in Phonologie. Baltaxe. Trans. Principles linguistics as producing only etic of Phonology, Berkeley: University of analyses, or what is worse, disguising California Press, 1969. emic categories as etic – for example, as when asking the natives for judge- Further reading ments of sameness/difference necessi- Anderson, Stephen (1985). Phonology in tating recourse to meaning, while the Twentieth Century. Chicago: Uni- explicitly rejecting meaning as a cate- versity of Chicago Press. gory external to the linguistic system. By proposing the emic/etic distinction, Patrick Honeybone inspired by Edward Sapir, Pike aimed to increase awareness of the pitfalls of projecting one’s own emic categories onto an alien system, and to empha- EMIC/ETIC sise the need to keep apart the two perspectives, of the native speaker and Terms coined by Kenneth Pike, from of the analyst. (phon)emic and (phon)etic respec- Pike did not, however, perceive a tively, to refer to two complementary rigid dichotomy between these two ways of analysing behavioural data. perspectives. Rather he recognised a Emic refers to categorising behaviour progression from emic to etic, once from the perspective of the insider in emic units discovered via analysis of ways that are meaningful to the one language are applied to analysis

66 EMPIRICISM/RATIONALISM of another. To the extent that they are Psychology. Amsterdam: Swets & relevant to the latter, these units con- Zeitlinger. 55–63. stitute part of an etic inventory that is no longer internal to any single lan- Marina Terkourafi guage. The International Phonetic is an example of such an inventory, while the of individual languages correspond to EMPIRICISM/ the emic systems from which the former is abstracted. RATIONALISM An inventory of etic units is created by the analyst prior to the analysis of Put simply, empiricism is the view that the particular language to which it is all knowledge derives from experi- applied. Etic units correspond to raw ence; rationalism is a contrasting view observational data, often measurable in which knowledge comes from rea- by instrumental means. Since they soning. In epistemology, the branch of do not combine into a system of philosophy concerning theories of meaningful contrasts, they are pre- knowledge, empiricism and rational- structural. Conversely, emic units result ism are types of position that have when the raw data are interpreted by been taken about the sources of natives, though they are not necessarily knowledge, in particular in discus- consciously known (named) by them. sions about what is required for a They are thus structural and valid for state to count as knowledge. In psy- only one language. Combining emic chology and its philosophy, empiri- and etic perspectives yields ‘a kind of cism and rationalism concern the “tri-dimensional understanding” of sources of psychological states and human behaviour instead of a “flat” capacities that may include, but are etic one’ (Pike 1954: 12). not confined to, states of knowledge. See also: Analytic/Synthetic; Primary sources Behaviourism; Pike, Kenneth (1954). Language in Rela- Deduction/Induction; Holism; tion to a Unified Theory of the Structure Innateness; Linguistic Relativity; of Human Behaviour. Glendale, CA: Logical Positivism; Mentalism; Summer Institute of Linguistics. Universal Grammar Key Thinkers: Berkeley, George; Further reading Chomsky, Noam; Descartes, René; Headland, Thomas, Kenneth Pike and Frege, Gottlob; Hume, David; (eds) (1990). Emics and Locke, John Etics: The Insider/Outsider Debate. Frontiers of Anthropology. Vol. 7. In a general sense, a position is a London: Sage. form of empiricism insofar as it holds Jahoda, Gustav (1977). ‘In pursuit of the that knowledge (or other psychologi- emic-etic distinction: can we ever cal states and capacities, indicated capture it?’. In Y. H. Poortinga (ed.), henceforth: (etc.)) about some partic- Basic Problems in Cross-Cultural ular subject matter, S, derives from

67 EMPIRICISM/RATIONALISM

experience of that subject matter; and meaning (although issues surrounding a position is a form of rationalism the structure of language and the pos- insofar as it holds that knowledge sibility of a perfect language – a lan- (etc.) about some particular subject guage designed perfectly to reflect the matter, S, derives from the use of structure of reason – have also been of reason or, more generally, from our some importance). Thus, the empiri- rational nature(s). Traditionally, these cist David Hume argued that ideas – positions, and disputes among their that is concepts, including the mean- proponents, have concerned knowl- ings of words – must derive ultimately edge (etc.) about mind-independent from impressions – experiential input. subject matters, about how the world On the basis of this constraint, Hume is independent of particular views that argued that we can have no idea cor- we might take about it. responding with genuine causation – In the twentieth century a central that is, no idea whose content out- debate between empiricists and strips our experience of one type of rationalists concerned our knowl- occurrence regular following another edge of mathematics. Gottlob Frege type. Similarly, in the twentieth cen- attempted to show, first, that our tury, logical positivists like Ayer knowledge of mathematics is substan- argued that, in order for a sentence to tive knowledge – knowledge of an be meaningful, it must be possible to independent subject-matter and, verify (or falsify) it – to determine second, that mathematics derives whether it is true or false on the basis from logic* so that our knowledge of of experience. Opponents of the logi- mathematics can be seen to derive cal positivists argued in turn that this from the use of (pure) reason. Frege’s Verification Principle was not itself empiricist predecessors, like J. S. Mill, verifiable by appeal to experience. had attempted to treat mathematical Some of these opponents took them- knowledge as deriving from experi- selves to be pushing for an empiricism ence. In the face of Frege’s critique of that was even more extreme than that that move, his empiricist successors, of the positivists: they would argue like A. J. Ayer, attempted to treat that this shows that the Verification mathematical knowledge as derivative Principle is itself meaningless and that from knowledge of meaning and so we need to construct a different form (they argued) not really a substantive of empiricism. Other opponents form of knowledge. This debate in argued that knowledge of some very turn raised questions about the status general claims must derive from the of our knowledge of language: is this use of reason; the latter form of oppo- knowledge derived from reason (more sition may reasonably be viewed as a generally, an aspect of our natures as form of rationalism. rational beings), or is it rather a form Three major traditional points of of experiential knowledge? dispute between empiricists and ratio- With respect to the study of lan- nalists centre on the following three guage, major disputes between characteristic rationalist theses: (1) forms of empiricism and rationalism knowledge of a particular subject have focused upon our knowledge of matter is underwritten by intuition (or

68 EMPIRICISM/RATIONALISM rational insight) and deductive rea- its lack of specificity. First, articulating soning, rather than by experience of the precise content of a form of that subject matter; (2) knowledge of empiricism is dependent upon further a particular subject matter is innate specification of the notion of experi- (very roughly, determined by nature ence employed in the approximate rather than, for example, by the par- account. For instance, one form of ticular course of experience); and (3) empiricism holds that knowledge the concepts or ideas that constitute (etc.) about a particular subject matter our abilities to think about a particu- derives solely from sense experience of lar subject matter are innate. Ratio- that subject matter – visual, auditory nalists about knowledge (etc.) about a or tactile experience – and not from particular subject matter characteristi- other forms of experience – introspec- cally endorse at least one of (1)–(3) tive experience or religious revelation. with respect to that subject matter. Further specification of that form of Empiricists about knowledge (etc.) of empiricism would be dependent upon a particular subject matter - further specification of the boundaries istically reject (1)–(3) with respect to of sensory experience. Correlatively, that subject matter. Since epistemolog- articulating the precise content of a ical forms of empiricism and rational- form of rationalism is dependent upon ism concern the justification or further specification of the extent of warrant required for a state to count reason or our rational nature(s). as knowledge, and not the sources of Whether a particular form of ration- psychological states and capacities in alism is in dispute with a particular general, it is possible to adopt psycho- form of empiricism depends upon the logical forms of each type of position details of such further specification. In without also adopting epistemological particular, it depends upon whether forms. For instance, one might hold the type(s) of experience to which that a particular belief is innate – and appeal is made in characterising so be a psychological rationalist with the particular form of empiricism respect to the belief – and also hold includes, or excludes, the outputs of that in order to be justified or war- the type(s) of reason, or our rational ranted the innate belief must be sup- nature(s) to which appeal is made in plied with experiential support – and characterising the particular form of so be an epistemological empiricist rationalism. with respect to the belief. Alterna- Second, someone who holds that tively, one might hold that a particular there are sources of knowledge in belief is only acquired on the basis of addition to experience and reason, or experience, but that the justification our rational nature(s), might reject or warrant for the belief derives from empiricism (or rationalism) about a reason. In that case, one would be a subject matter without endorsing psychological empiricist and an epis- rationalism (or empiricism) about that temological rationalist. subject matter. For instance, that Two points are worth noting about person might hold that knowledge the approximate account of the two about some subject matters depends types of position, both pertaining to upon aspects of our non-rational

69 EMPIRICISM/RATIONALISM nature(s) that are not sense- sometimes known as a ‘poverty of the perceptual; or that it depends upon stimulus’ or ‘poverty of evidence’ con- non-sensory experience. sideration – is that our knowledge There are other dimensions along (etc.) about the target subject matter which particular forms of empiricism could not have been acquired or rationalism might vary. But we can through, for example, sense experi- see already that classifying a position ence, so that empiricism is ruled out. as a form of empiricism or rationalism The second type of consideration is an is quite unrevealing. And we can also account of how the rationalists’ see that there is no such thing as the favoured source – reason or our dispute between empiricism and rational nature(s) – could have under- rationalism; rather, there are – or written our acquisition of knowl- could be – various disputes each edge (etc.) of the subject matter. For taking place between particular forms instance, a rationalist about our of each broad type of position. knowledge of ethics might attempt to Finally, we can see that care is account for our knowledge about required in classifying individual what we ought to do in a particular thinkers as empiricists or rationalists, case by appeal to our possession of for careless adherence to such a broad innate knowledge of general ethical scheme of classification can serve to principles from which our knowledge disguise differences between particu- about particular cases is derived. lar members of one of the groups and In response to the first type of con- similarities between members of the sideration, empiricists will attempt different groups. René Descartes, to argue that experiential resources Benedictus de Spinoza and Gottfried suffice to explain the knowledge that Leibniz are often aligned as paradig- we in fact possess. Empiricists will matic rationalists, the so-called Conti- develop their response either by nental Rationalists. They are often attempting to provide an account of seen in opposition to , how the knowledge (etc.) could have and David Hume, been acquired on the basis of experi- the so-called British Empiricists, who ence, or by arguing that we do not in are often treated as paradigmatic fact possess the knowledge that ratio- empiricists. However, it is important nalists claim we do. For instance, an to look beyond that preliminary clas- empiricist about our knowledge of sification in coming to a proper appre- mathematics might attempt to argue ciation of the works of those that this knowledge is supported by important thinkers. induction from our experiences of Where a dispute arises between a groups of objects. And an empiricist particular form of empiricism and a about our putative knowledge of particular form of rationalism, the ethics might claim that we do not dispute characteristically takes the fol- really have such knowledge but only lowing general form. The rationalists various feelings about particular characteristically offers two types of courses of action. Since such disputes consideration in favour of their posi- typically concern knowledge (etc.) tion. The first type of consideration – about a particular mind-independent

70 EMPIRICISM/RATIONALISM subject matter, the empiricist might The most prominent contemporary attempt to argue that we have the defender of a form of rationalism is knowledge that the rationalist claims Noam Chomsky. Chomsky, together we do, but that it is knowledge about with numerous co-workers in linguis- the operations of our own minds, or tics, psychology and philosophy, has relations among our concepts or used ‘poverty of stimulus’ considera- ideas, rather than about mind- tions in support of the thesis that independent reality. human knowledge of natural lan- In response to the second type of guage has a significant innate compo- consideration, the empiricist will nent. And Chomsky, again together attack the rationalist account of how with co-workers in a variety of disci- we come to have the knowledge (etc.) plines, has developed an increasingly that rationalists claim we possess. For detailed account of the development instance, the empiricist might follow of our knowledge of particular natu- Locke in attempting to argue that the ral languages – for example, particu- claim that a piece of knowledge (etc.) is lar dialects of English or particular innate is either false or consistent with – that makes appeal empiricism. If the rationalist claims to innate structures and capacities. that a piece of knowledge (etc.) is Of course, Chomsky accepts that innate only if it is possessed by every- experience plays some role in the one at birth, then the empiricist will acquisition of knowledge of language, point to the absence of that piece of since it would otherwise be a mystery knowledge (etc.) in the very young or that children typically acquire knowl- the dysfunctional. Alternatively, if the edge of language that enables them rationalist opts for looser requirements to communicate with those whose on innateness by counting a piece of speech they experienced during acqui- knowledge (etc.) as innate if we are sition. But he holds that the role of born with a capacity to acquire it, then experience is primarily to select from the empiricist will also be willing to amongst the child’s innate repertoire accept that all knowledge is innate in the bits that will be operative in the that very thin sense. As Locke puts it: competence that they come to employ. Chomsky’s form of rationalism has If the capacity of knowing, be the natu- been subjected to both sides of the ral impression contended for, all the standard empiricist critique. First, ver- truths a man ever comes to know, will, sions of Locke’s objection have been by this account, be every one of them, pressed, according to which Chom- innate; and this great point will amount sky’s appeal to innate psychological to no more, but only an improper way states or capacities either fails to dis- of speaking; which whilst it pretends to tinguish his position from empiricism assert the contrary, says nothing differ- or is easily falsified. Because Chomsky ent from those, who deny innate princi- aims to provide a detailed, predictive ples. For nobody, I think, ever denied, account of the course of acquisition of that the mind was capable of knowing particular languages, his position several truths. (Locke, 1690: book I, appears to avoid the second horn of chapter II, section 5, p. 61) Locke’s dilemma by going beyond the

71 EMPIRICISM/RATIONALISM bland claim that we are predisposed rationalism. That should not be sur- to acquire language. And because he prising now that we have recognised provides an account according to the variety of possible rationalist posi- which our initial state of knowledge is tions. But some of the apparent differ- shaped by experience in the course of ences are especially striking and have normal development, his account been taken by some thinkers to under- avoids falsification by the fact that mine Chomsky’s classification as a small children and subjects of abnor- rationalist, or the bearing of his work mal development lack ordinary on the standing of more traditional knowledge of language. forms of rationalism. However, important questions First, although Chomsky talks of remain concerning the precise content knowledge of language, it is not clear of the claim that a basic component of that he thinks of this knowledge as the human linguistic capacity or state is sort of propositional knowledge – innate. And the fact that many theo- knowledge that such-and-such – that rists who align themselves with either is of concern to epistemologists. Some empiricism or rationalism – including philosophers have thought that the Chomsky – agree that both innate knowledge is really knowledge-how – and experiential factors play a role in that is, practical knowledge – like shaping knowledge of language knowledge how to ride a bicycle. tends to undermine the utility of their Although Chomsky rejects that inter- classification as empiricists or ratio- pretation, it remains an open question nalists. Second, empiricists have whether he is right to do so. And attempted to provide accounts of lan- Chomsky admits other reasons for guage acquisition that make more lim- thinking that the sort of knowledge ited appeals to innate psychological in question differs from the sort states and capacities. Some empiricist that concerns epistemologists. For approaches agree with the rationalist instance, he does not think that assessment of what the child acquires knowledge of language is justified or and attempt to provide accounts of warranted and he does not think that how the child might acquire it more or we are typically conscious of possess- less solely on the basis of experience. ing it; he thinks of it as tacit knowl- Other empiricist approaches involve edge. Moreover, some of these reasons an attempt to show that the child might also be grounds for thinking acquires less than the rationalist has that knowledge of language is not claimed, so that the task of accounting really a psychological state, so under- for their acquisition is made easier for mining Chomsky’s classification even the empiricist. Thus far, no empiricist as a psychological rationalist. account has been provided that has Second, Chomsky does not think anything approaching the depth, that knowledge of language is knowl- detail and coverage of rationalist edge about a mind-independent sub- accounts. ject matter. Rather, on Chomsky’s Chomsky’s form of rationalism view, facts about an individual’s appears to differ in certain respects language are constituted by facts from some more traditional forms of about the individual’s psychology, in

72 EMPIRICISM/RATIONALISM particular facts about their knowledge Primary sources of the language. If he were right about Berkeley, G. (1975). Philosophical Works, this, then it would form another dif- Including the Works on Vision. Ed. ference between knowledge of lan- M. R. Ayers, Everyman edition. guage and ordinary propositional London: J. M. Dent. knowledge. Third, although Chomsky Chomsky, N. (1988). Language and Prob- thinks that knowledge of language is lems of Knowledge. Cambridge, MA: determined as a part of human nature, MIT Press. he does not appear to think that it is Descartes, R. (1641/1984). Meditations determined as a part of our rational on First Philosophy. In The Philosophi- nature(s). See also the entry on uni- cal Writings of Descartes, vol. 2. Trans. versal grammar*. For a readable J. Cottingham, R. Stoothoff and D. introduction to some considerations Murdoch. Cambridge: Cambridge Uni- that support Chomsky’s position, see versity Press. Pinker (1994). Hume, D. (1748/1999). An Enquiry Con- To sum up, let’s return to the earlier cerning Human Understanding. Ed. T. twentieth-century debate concerning L. Beauchamp. Oxford: Oxford Univer- the tenability of empiricist restrictions sity Press. on the possible meanings of sentences. Leibniz, G. (c. 1704). New Essays on Recall that the logical positivists Human Understanding. In G. H. R. argued that meaningfulness coincides Parkinson (ed.) (1973), Leibniz: Philo- with the possibility of verification (or sophical Writings. Trans. M. Morris falsification) on the basis of experi- and G. H. R. Parkinson London: J. M. ence. The empiricist position here was Dent and Sons. that there could be nothing in our Locke, J. (1690/1997). An Essay on knowledge of language, including our Human Understanding. Ed R. Wool- knowledge of meaning, that was not house. London: Penguin Books. put there by experience. Chomsky’s work casts doubt on that view. For if Further reading Chomsky is right, then our knowledge Carruthers, P. (1992). Human Knowledge of language is shaped as much by our and Human Nature. Oxford: Oxford as by our experience. More- University Press. over, Chomsky’s work makes plain Cottingham, J. (1984). Rationalism. that views about possible limits to our London: Paladin Books. knowledge must engage with (experi- Kenny, A. (ed.) (1986). Rationalism, ence-based) work in the sciences Empiricism and Idealism. Oxford: concerned with human nature or con- Oxford University Press. stitution. Hence, even those who seek Loeb, L. (1981). From Descartes to Hume: a more empiricist view about our Continental Metaphysics and the Devel- acquisition of knowledge of language opment of Modern Philosophy. Ithaca, typically seek to support that view by NY: Cornell University Press. appeal to work in those sciences. Pinker, S. (1994). The Language Instinct. Needless to say, the various disputes London: Penguin Books. in this area are far from being Spinoza, B. de (1677/1985). The Ethics. In resolved. The Collected Works of Spinoza, vol. 1.

73 FEMINISM

Ed. and trans. E. Curley. Princeton, NJ: were silenced and interrupted by men. Press. Another influential study of the area is Stich, S. (ed.) (1975). Innate Ideas. Pamela’s Fishman’s 1978 analysis of Berkeley, CA: California University the home conversations of American Press. couples: she found that women did a lot of the interactional ‘shitwork’ – as Guy Longworth she termed it – that is mundane but sustains relationships, in the same way that they do other mundane work that sustains the family, such as house- work. Some of the researchers of the time FEMINISM did not only argue that male domi- nance led to gender differences in Feminism is a hotly contested term speech but to female deficiencies. The but it can broadly be described as a ‘deficit approach’ as an extension of range of social movements and theo- the ‘dominance approach’ was carried ries that have discrimination on the over from an earlier period where men basis of gender as their key concern. were seen as ‘default human beings’ Feminist linguistics explores the inter- and everything that deviates from the relationship between language and male was seen as deficient. Thus gender. speaking like a woman was not only seen as feminine but, at the same time, See also: Conversation Analysis; as deviating from ‘normal’ – that is Deconstruction; (Critical) male – speech. It is one of the lasting Discourse Analysis; Political achievements of this early work to Correctness; Poststructuralism identify the ‘double bind’ in which Key Thinkers: Cameron, Deborah; women as speakers found themselves: Tannen, Deborah if they talked ‘like a lady’, they were Feminist linguistics, or ‘language and considered less than a full human gender’ as the field is more widely being; if they talked like a man, they known, emerged in the context of were considered insufficiently femi- the second feminist movement of the nine. 1960s and 1970s. Linguists of the By the 1980s the political climate time who were also feminists started had changed and while the students of to ask what part language played in these pioneers were still interested in the widespread discrimination against researching the way men and women women. This early work found a focal spoke differently, they were no longer point in Robin Lakoff’s 1975 book ready to see those differences as an Language and Woman’s Place. The expression of female subjugation and author argued that men and women male dominance. Feminism had spoke differently and that those dif- turned into a social movement that ferences were evidence of male domi- was more focused on celebrating nation: women were described as less female difference. Consequently the confident speakers, for instance, who ‘dominance’ and ‘deficit’ approaches

74 FEMINISM of the 1960s and 1970s started to be evidence that women talked more than replaced by the ‘difference approach’. men and equally solid evidence that The key exponent of this approach men talked more than women. At the is best-selling linguist Deborah same time, the universalist underpin- Tannen. In books such as That’s Not nings of the feminist movement started What I Mean (1986) and You Just to be questioned by women of colour, Don’t Understand! (1990), she argues the queer movement and others who that communication between men and did not see themselves represented women could best be understood in the stereotypical straight white through applying the framework of middle-class US–American woman intercultural communication – for who seemed to have been taken to rep- example, the way people who grew up resent the prototypical female experi- in different parts of the world may ence up until then. Both the sometimes create communication dif- ‘dominance approach’ and the ‘differ- ficulties – but men and women, she ence approach’ came in for substantial argues, have communication difficul- criticism on a number of grounds, par- ties because they operate with differ- ticularly for treating ‘men’ and ent values. Men are apparently driven ‘women’ as homogeneous categories; by competition and women by coop- for a lack of attention to context; for eration. These different values are an unsophisticated understanding of expressed in different communicative power relations; and for insensitivity styles, which lead to misunderstand- to ethnic, racial, social, cultural and ings and fights between the sexes. linguistic diversity. Representatives of the difference As a result, poststructuralist approach were keen to stress that approaches to while men and women had different started to emerge. Deborah communicative styles, these were Cameron’s book Feminism and Lin- equally valid. guistic Theory (1985) is often consid- By the late 1980s language and ered a foundational text for this new gender had changed from an almost approach. Poststructuralism* does esoteric interest of some female lin- not start from the assumption that guists to become a thriving research men and women speak differently – it field which was institutionally based in does not even start from the assump- university departments and curricula. tion that men and women naturally However, this high level of research exist as meaningful categories. Rather activity also had a surprising side- it is language that calls the (gendered) effect that led to a complete conceptual identity of speakers into existence. rethinking of the field: as more and Gender is thus no longer treated as a more researchers tested the various given but the linguistic concern is now claims about female and male speech with the way in which gendered sub- empirically, differences no longer jectivities are constituted in language. appeared as clear-cut as they had Once the question ‘how do men and seemed to exponents of both the dom- women talk differently?’ had become inance and the difference approach. obsolete, feminist linguistics was in a For instance, there was empirical position to focus on new questions.

75 FEMINISM

One such new question is related to such as applied linguistics, feminism the construction of gendered identities and language and gender are only in interaction. Such studies have slowly finding their way into language shown that gender is not some form of education (Pavlenko et al. 2001). Irre- static identity but changes from con- spective of the approach, it is a key text to context: in some contexts characteristic of most feminist linguis- gender may not matter at all, in others tics that it is committed to a political we may highlight or downplay certain cause, namely emancipation in vari- aspects of our gender identities. ous forms. Since the inception of the Another new question that has been field, feminist linguists have not only particularly taken up by scholars aimed at describing linguistic prac- working in (critical) discourse analy- tices. They have also been committed sis* centres around the representation in various degrees to engagement with of particular types of femininity or ‘the real world’ and to challenging masculinity in the media. Media dis- practices that disadvantage groups of courses have a central influence on speakers. many facets of identity in contempo- rary societies from gendered risk- Primary sources taking to gendered parenting. There Cameron, Deborah (1992). Feminism and have also been inquiries into the ways Linguistic Theory. London: Macmillan. in which gender structures access to First edition 1985. linguistic resources. For instance, Fishman, Pamela (1978). ‘What do cou- immigrant women often find it more ples talk about when they’re alone?’. In difficult to learn the majority language Douglas Butturff and Edmund L. as they may have less access to the Epstein (eds), Women’s Language and public spaces where it is taught and Style. Akron, OH: L&S Books. 11–22. used. In other contexts, girls may find Lakoff, Robin (1975). Language and it easier to access Woman’s Place. New York: Harper and learning because it may be discur- Row. (A revised and expanded edition sively constructed as a ‘girlie’ thing to with commentaries by contemporary do while boys are discouraged from feminist linguists was edited by Mary language learning because of these Bucholtz and published by Oxford Uni- very associations. versity Press in 2004.) What all these questions and Tannen, Deborah (1986). That’s Not approaches have in common is that What I Meant! How Conversational they do not treat gender as given and Style Makes or Breaks Relationships. static but as emergent in context and New York: Ballantine Books. as being comprised of a range of per- Tannen, Deborah (1990). You Just Don’t formances, some of which may be Understand: Women and Men in Con- more hegemonic than others. Gender versation. New York: Ballantine Books. thus turned from a noun into a verb. Language and gender today has a Further reading place in probably every sociolinguis- Pavlenko, Aneta, Adrian Blackledge, tics undergraduate class in the world. Ingrid Piller and Marya Teutsch- Despite being highly relevant in areas Dwyer (eds) (2001). ,

76 GENERATIVE PHONOLOGY

Second Language Learning, and and only the surface forms of a natu- Gender. Berlin and New York: Mouton ral language, focusing on its speech de Gruyter. sounds. As with any such set of rules, the measure of its success was the ade- Ingrid Piller quacy* with which it was consistent with underlying linguistic knowledge, known as universal grammar*. Chomsky and Halle’s book The GENERATIVE Sound Pattern of English (1968), known as ‘SPE’, is widely regarded as PHONOLOGY the defining text of generative phono - logy. In it they attempted to specify the A branch of generative grammar that phonological rules underlying the aims to establish a set of rules, princi- speech sounds of native English ples or constraints capable of produc- speakers. SPE established the standard ing the surface phonetic forms of a framework for this type of theory. language and of modelling the inter- That framework is as follows: that nalised linguistic knowledge of the there are abstract rules determining native speaker. Generative phonology the actual acoustic output of speech; was a central idea in linguistic that the rules apply sequentially to research throughout the 1960s and produce a series of derivations result - although it has undergone reforms ing in an abstract repre sentation of the and changes in subsequent decades, it phonetic representation; that the continues to be the dominant frame- consists of a series of work for many developments in segments that could be exhaustively phonological theory. defined in terms of sets of binary features; that the rules are strictly See also: Adequacy; Distinctive ordered. Features; Optimality Theory; By the 1970s SPE had become a Transformational-Generative benchmark against which most other Grammar; Universal Grammar work in phonology was measured. Key Thinkers: Bloomfield, Leonard; But critics began to find problems Chomsky, Noam; Jakobson, with some of its basic assumptions. Roman; Trubetzkoy, N. S. For instance, SPE’s focus on abstract Generative phonology originated with rules, rather than detailed phonetic the work of Noam Chomsky and analysis, and the formal complexity of Morris Halle at the Massachusetts many of its proposals, became stum- Institute of Technology (MIT) in the bling blocks for many linguists. These late 1950s. It built on N. S. Trubet- criticisms led to developments such zkoy’s idea of phonemic oppositions as ‘Natural Generative Phonology’, and Roman Jakobson’s later work on which attempted to establish rules distinctive features*. More specifi- that were more psychologically plau- cally it drew on the general aspiration sible than the of SPE. of generative grammar to stipulate a Later work on optimality theory* did set of rules capable of producing all away with rigid sequential derivation

77 GENERATIVE SEMANTICS in favour of an algorithm for selecting half of the 1960s and the 1970s. It surface forms from a set of possible opposed the approach dubbed ‘inter- alternatives. pretive semantics’ favoured by Noam In the twenty-first century, phonol- Chomsky and others. Among its major ogy might appear to have moved some proponents were George and Robin way from the rigid model imposed in Lakoff, Paul Postal, John (Haj) Ross, SPE. Nevertheless the formal and and James McCawley. Generative principled approach of generative semantics is credited for bringing atten- phonology and the wider generative tion to meaning in linguistics and for project remains an important founda- ushering in interest in pragmatics, cog- tion for much work in the field. nitive linguistics and some aspects of sociolinguistics. Primary sources See also: Cognitivism; Prototype; Chomsky, Noam (1964). Current Issues in Transformational-Generative Linguistic Theory. The Hague: Mouton. Grammar Chomsky, Noam and Morris Halle Key Thinkers: Austin, J. L.; (1968). The Sound Pattern of English. Chomsky, Noam; Grice, H. P.; New York: Harper and Row. Searle, John Halle, Morris (1962). ‘Phonology in Gen- erative Grammar’. Word 18, 54–72. Chomsky’s standard theory (1965) Prince, Alan and postulates the presence of formal rules (1993/2004). Optimality Theory: Con- which generate syntactic sentence straint Interaction in Grammar. Rutgers skeletons which are then filled by lex- University and University of Colorado ical insertion rules to create deep at Boulder. Oxford: Blackwell. structures. These are then turned by transformations into questions, pas- Further reading sives and so on. According to Chom- Carr, Philip (1993). Phonology. Hound- sky, transformations do not affect mills: Macmillan Press. meaning. Chomsky and many of his Kenstowicz, M. and C. Kisseberth (1986). collaborators took the position that Generative Phonology: Description and had to take place Theory. New York: Academic Press. after deep structures had been generated, relegating the role of lexi- Christopher Routledge and Siobhan cal semantics to the interpretation of Chapman structures that had already been gen- erated, just like phonology (hence the name given to this position: interpre- tive semantics). The generative GENERATIVE semanticists took a different position, that lexical insertion had to happen SEMANTICS both in the deep structure and after some transformations, but most An approach to the treatment of seman- significantly that transformations tics within transformational-generative affected meaning, hence the name grammar* popular between the second ‘generative semantics’.

78 GLOSSEMATICS

The debate between the two camps the study of meaning and its relations was fierce and highly technical, to social interaction. revolving, for example, on the correct decomposition of the verb ‘kill’ in Primary sources ‘cause to become not alive’. The Harris, Randy Allen (1993). The Linguis- details are beyond an introductory tics Wars. New York and Oxford: account (but see Harris 1993) but the Oxford University Press. result of the controversy was that the generative semanticists, while essen- Further reading tially correct, did not present a cohe- Chomsky, Noam (1965). Aspects of the sive research programme and moved Theory of Syntax. Cambridge, MA: on to other fields, including gender MIT Press. research (Robin Lakoff), pragmatics Rosch, Eleanor (1975). ‘Cognitive repre- (Georgia Green), cognitive linguistics sentations of semantic categories’. Jour- (George Lakoff, Ronald Langacker), nal of Experimental Psychology: General or developed other theories of syntax 104: 192–233. (Paul Postal, James McCawley). Ross, John Robert (Haj) (1973). ‘Nouni- In general, generative semantics is ness’. In Paul Kiparsky, John Robert to be credited with the attention to Ross, James D. McCawley and Osamu meaning that characterised the fields Fujimura (eds), Three Dimensions of of pragmatics and cognitive linguistics Linguistic Theory. Tokyo: TEC Corpo- in the last quarter of the twentieth ration. 137–257. century and the first decade of the Zadeh L. A. (1965). ‘Fuzzy sets’. Informa- twenty-first. In particular, generative tion and Control 8: 338–53. semantics’ interest in broad data, often resulting in odd, whimsical or Salvatore Attardo arcane examples, led to an interest in notions that defied the rigid categori- sation of , such as Ross’s ‘squish’, Lotfi Zadeh’s GLOSSEMATICS ‘fuzzy logic’, and Eleanor Rosch’s ‘prototypicality’, which are at the core A structuralist approach to the study of cognitive linguistics. Generative of language that attempts to establish semantics was also instrumental in a formal and abstract theory of lan- bringing J. L. Austin’s, H. P. Grice’s guage equivalent to the exactness of and ’s ideas to the fore- theories in the natural sciences by set- front of linguistic theory and con- ting up a of description tributing significantly to the creation based on a elementary unit called a of the discipline of pragmatics. ‘glosseme’. The theory was developed Generative semantics can then in the 1930s by the Danish scholar be seen as the precursor of many of in collaboration with the most significant contemporary Hans Jørgen Uldall and is the most approaches to the study of linguistic prestigious outcome of the works of meaning and as having brought about the ‘Copenhagen Linguistic Circle’. a significant shift in paradigm toward Any list of linguistic theories mentions

79 GLOSSEMATICS glossematics, but apart from accounts stance’, which may best be understood of the phonematic systems of some as unformed matter. dialects, the theory has not been The most fundamental work in adopted by many linguists. glossematics was written in Danish (Omkring sprogteoriens grundlæggelse See also: Logical Positivism; (English translation: Prolegomena to a Phoneme; Signs and Semiotics; Theory of Language (1953)) and it is Structuralism significant that in Paul L. Garvin’s Key Thinkers: Hjelmslev, Louis; review of the translation by Francis J. Greimas, Algirdas; Jakobson, Whitfield, he says that ‘The Prolegom- Roman; Saussure, Ferdinand de ena are probably among the most The point of departure in glossematics unreadable books in linguistics’. Partly consists in a few axiomatic claims: lan- due to this and partly due to the fact guage is one of the semiotic systems that, apart from writings in Danish, the employed by humans to think and bulk of Hjelmslev’s works was written communicate, the linguistic system is in French, the details of the theory are a an immanent, self-contained, structure challenge for the non-initiated, and that should be described without although the intellectual achievements any ‘metaphysical’ or ‘psychological’ of glossematics are widely acknowl- claims, and the formal system of edged, the approach has had little glossematics is the adequate tool for impact on empirical studies. Its most that. The theory is a hierarchically- prominent status is as an inspiration for ordered set of terms conceiving ‘func- linguists and as a beacon of scholarly tion’ (in a non-mathematical sense) as rigour. the key-concept on which the other terms are elaborated. Basically it Primary sources denotes dependence relations between Garvin, Paul L. (1954). ‘Review of Prole- other entities, called ‘functives’. gomena to a Theory of Language’. According to the intrinsic logic of American Anthropologist, vol. 56, no. glossematics, it presents a complicated 5, part 1: 925–6. of ‘functions’ and ‘func- Hjelmslev, Louis (1943). Omkring sprogte- tives’, the most important of which are oriens grundlæggelse. Copenhagen: Uni- the terms ‘constant’, a ‘functive whose versity of Copenhagen. Reprinted presence is a necessary condition for Akademisk Forlag 1966. Trans. Francis the presence’ of another ‘functive’, and J. Whitfield (1953), Prolegomena to a ‘variable’ which is a ‘not necessary’ Theory of Language. Baltimore: Waverly condition. The apparatus is utilised on Press. both sides of the linguistics sign, Hjelmslev, Louis (1975). Résumé of a ‘expression’ and ‘content’, implying Theory of Language. Copenhagen: that minimal units of the same nature, Nordisk Sprog-og Kulturforlag. ‘glossemes’, can be found both in the ‘expression form’ and the ‘content form’. Form as a concept should Further reading be conceived of as a synonym for Hjelmslev, Louis (1928). Principes de structure, and it is opposed to ‘sub- grammaire générale. Copenhagen: Høst

80 HOLISM

and Søn. Reprinted Copenhagen: and ‘long’, the second of which itself Munksgaard, 1968. depends for its meaning on ‘space’, Hjelmslev, Louis (1932). Etudes baltiques. and so on until the meanings of all Copenhagen: Levin & Munksgaard. words are seen to belong to an inter- Hjelmslev, Louis and Hans Jørgen Uldall connected web. The mentalist form (1957). Outline of Glossematics. Copen- holds that the content of concepts hagen: Nordisk Sprog-og Kulturforlag. (the components of propositions*) depends on that of all other concepts Hans Götzsche in the thinker’s repertoire. The pros and cons of are usu- ally unaffected by which form it takes. A word’s meaning is, after all, given HOLISM by the concept it expresses. Early dis- cussions of holism tended to be Holism about meaning, or semantic framed linguistically. Latterly it has holism, is the idea that there are no been framed in terms of the content of independent units of meaning smaller the language of thought*. than the entire representational Most arguments for semantic system, that is, the language. While holism have a two- pattern. some have happily embraced semantic The first premise is that some property holism, others have claimed that its other than meaning is holistic. The allegedly unpalatable consequences second is that semantic facts are undermine otherwise seemingly grounded in this other property. attractive theories in the philosophy Meaning therefore inherits the holism of language or psychology. of the other property. The three instances of this form of argument See also: Analytic/Synthetic; presented below centre on evidence, Descriptivism; Indeterminacy on translation, and on psychological Key Thinkers: Dummett, Michael; explanation respectively. Fodor, Jerry; Quine, W. V. O. Holism about evidence (‘confirma- ‘Holism’ has several different mean- tion holism’) is the thesis that whether ings, according to what is being said some empirical discovery confirms a to be holistic (as opposed to ‘atom- given proposition depends on the con- istic’). Some other forms of holism firmation status of a good number of will be mentioned here, but the focus other propositions, and ultimately of is on semantic holism, which is a every other proposition. Science stands pivotal idea in the philosophy of or falls as a whole, since it consists of language. mutually supporting theses, not eviden- Semantic holism itself has both a tially isolated ones. If we accept this, linguistic and a mentalistic guise. The and also hold (more contentiously) that linguistic form holds that what one the meaning of a sentence is consti- word in a language means depends on tuted by its confirmation conditions, what all the other words mean. For semantic holism results. Tying meaning example, perhaps ‘giraffe’ depends for to confirmation conditions can have its meaning on the meaning of ‘neck’ different motives. Logical positivists,

81 HOLISM for example, wished to streamline sci- waiting in the cupboard, and so on. ence by making this link. Others think Ultimately, it seems, an ’s acts that if a sentence is independent in prin- emerge out of a whole sea of rationally ciple to confirmation or disconfirma- coherent propositional attitudes* and tion, then it would never be rationally related mental states. According to assertible and its meaning could never cognitive holists, this deeper truth is be learnt. disguised by the way we highlight just Translation holism is the thesis that one propositional attitude in our ordi- how we translate one word can nary explanatory practices. To get depend on how we translate any other from this alleged truth to semantic word in the same language. Consider holism (of the mentalistic kind in this how we might complete the transla- case) we must make a further assump- tion from of the following tion: that a mental state’s content is Saxon law: grounded in the state’s explanatory potential. Since this potential is holis- (1) No man shall kill hwyđer except in tic, mental content is holistic. the presence of two or three wit- Having sketched three popular nesses; and then he shall keep his arguments for semantic holism, let us skin for four days. turn to objections, not to the argu- ments, but to semantic holism itself. If Clearly we need to know something successful, these objections can tell us about what skinned creature it might more about meaning than that it is have been impermissible (in the culture non-holistic. They also show that, for of the utterer) to kill save in the pres- each of the three arguments just con- ence of witnesses. But we know that sidered, either the underlying prop- this is what we need to figure out only erty is not holistic after all, or that because we have already translated the meaning is not constituted out of con- other words in the sentence. Semantic firmation conditions, translatability holism follows if we add the further or other such aspects. assumption that the meaning of a word One common objection to semantic is constituted by its translatability. This holism is that to understand any word further claim is not entirely implausi- or concept one would have to instan- ble. After all, communicability is a fun- taneously understand every word in damental aspect of word meaning, and the language, or every concept in an depends on our translating (or inter- entire repertoire. Even if this is not preting) one another’s words. impossible, it certainly seems not to be Holism about psychological expla- how we do achieve semantic compe- nation (‘cognitive holism’) is best tence. One response is to insist that approached by example. Suppose we children acquire concepts one by one, explain why someone walked towards but these morph, becoming gradually their kitchen by saying that they distinct and more sophisticated con- desired food. This only explains their cepts, reflecting the beliefs they behaviour if we assume they think that acquire on the way to adulthood. A there is food in their kitchen, that its more radical strategy is to accept that door is not locked, that no assassin lies we do indeed acquire concepts en

82 HOLISM masse, but to express this more plau- istic alternatives. These latter usually sibly as the view that we become identify the meaning of a word or proper subjects for the attribution of with a causal propositional attitudes in a kind of relation of some kind. It has proven Gestalt event. immensely difficult to get straight on A different objection centres on the what kind of causal relation could commitment of holists to a fine- work. The debate here often collapses grained conception of meaning. The into that between descriptivist theo- nature of a person’s concept is read off ries of reference (popular with holists) from the sum of beliefs that person and direct theories of reference (popu- has involving that concept, since a lar with atomists). belief is just a linking of one concept Even so, it is a simplification to treat to other concepts. A change in belief the debate as a choice between seman- could change the concept’s identity. tic holism and semantic atomism. Ultimately, a single discrepancy in There is an intermediate position – beliefs held would mean that two semantic molecularism – and a hybrid people had no concepts in common. position, either of which may evade This is, to say the least, counterintu- the difficulties faced by the purer itive: we could neither understand, views. In semantic molecularism a learn from, nor disagree with one word or concept’s meaning is from its another, or with our previous selves connection to some proper subset of after we have changed our minds, for the other words or concepts in the lan- that matter. Semantic holists try to guage or conceptual system, rather reply by saying that while, strictly than being either independent from all speaking, we have no concepts in these connections or dependent on all common, our concepts are similar of them. The difficulty here, as critics enough to allow understanding, learn- have pointed out, is how to distin- ing, disagreement and so on. guish between ancillary connections Finally, some commentators think and constitutive connections. that holism is incompatible with com- The hybrid position involves adopt- positionality. The meaning of a phrase ing an atomistic theory for some pur- like ‘pet fish’ ought to be a function of poses and a holistic theory for others. the meanings of ‘pet’ and ‘fish’. But if For example, ‘two factor’ theories of semantic holism is correct, the meaning mental content say that, for purposes of ‘pet fish’ ought to be associated with of cognitive explanation, a holistic that of ‘golden’, even though there is account is better, while an atomistic nothing in the meaning of either ‘pet’ or theory is better suited to the purpose ‘fish’ to predict this association. A pos- of understanding the conditions under sible response, here, is to insist that which a belief is true. Each hybrid semantic holism applies only to simple theory needs to be evaluated on its phrases or concepts; the meaning of merits, but a general problem attaches complex phrases or concepts is deter- to the strategy: how to fuse the two mined by compositionality*. elements of the theory. Semantic holists respond in kind by Holism about meaning is distinct pointing out problems with the atom- from other forms of holism but it is

83 common to argue for semantic holism IDEATIONAL from holism of these other kinds by making a constitutive assumption THEORIES about semantic facts. Equally, the alleged incoherence of semantic Ideationalists hold that the meaning holism has led others to go in the other of words is inherited from the mean- direction and reject these constitutive ing of mental entities (ideas or con- assumptions. Nevertheless, arguments cepts) rather than the other way about holism touch on many other around. Two classic statements of areas of the study of meaning. ideationalism have had an enduring influence, the first by John Locke and Primary sources the second by H. P. Grice. Block, N. (1986). ‘Advertisement for a See also: Nonnatural Meaning; semantics for psychology’. In P. A. Private Language French (ed.), Midwest Studies in Philos- Key Thinkers: Grice, H. P.; Locke, ophy, vol. X. Minneapolis: University of John; Putnam, Hilary; Minnesota Press. 615–78. Wittgenstein, Ludwig Davidson, D. (1975). ‘Thought and talk’. In S. Guttenplan (ed.), Mind and Lan- Locke sought to answer the question: guage. Oxford: Oxford University Press. ‘[How can] the thoughts of men’s Devitt, M. (1996). Coming to Our Senses. minds be conveyed from one to Cambridge: Cambridge University Press. another?’ (1706/1997: III.1.2). Having Dummett, M. (1976). ‘What is a Theory a thought, according to him, was a of Meaning?’ In G. Evans and J. matter of having ‘ideas’ – roughly, per- Mcdowell (eds), Truth and Meaning: ceptual and abstract concepts – in one’s Essays in Semantics. Oxford: Claren- mind. The existence and content of an don Press. 67–137. idea was supposed to be private, that Dummett, M. (1991). The Logical Basis of is, directly accessible only to its posses- Metaphysics. Cambridge, MA: Harvard sor. Language allows this privacy to be University Press. overcome so that knowledge can be Fodor, J. (1981). ‘The present status of the pooled. Words ‘stand as marks for the innateness controversy’. In J. Fodor, ideas within [the speaker’s] own mind, Representations. Brighton: Harvester. whereby they might be made known to Fodor, J. and E. Lepore (1991). Holism: A others’. That is to say, spoken words Shopper’s Guide. Cambridge, MA: have a proxy content that is inherited Blackwell. from the speaker’s ideas and, with Quine, W. V. (1960). Word and Object. luck, appropriately decoded by the Cambridge, MA: MIT Press. hearer. Locke treated the principle of Further reading ideationalism as obvious, and focused Fodor, J. A. and E. Lepore (1993). Holism: instead on confusions that arise when A Consumer Update. Amsterdam: we ignore how imperfect a vehicle for Rodopi. ideas language can be. The first influ- ential criticism of ideationalism did Alex Barber not emerge until two centuries later

84 IDEATIONAL THEORIES when Ludwig Wittgenstein criticised itself subject to extra-cranial influ- the notion of a private bearer of con- ence, which is hence passed on to the tent – a Lockean idea. On one reading content of utterances. To make this of his private language* argument, move would, however, undermine the Wittgenstein is suggesting that Lock- classification of Grice as an ideation- ean ideas could never be held mistak- alist. Ideationalists insist that words enly, and hence that it is also inherit their meaning from mental meaningless to talk of their being held states. To allow that external factors, correctly. If content is a matter of cor- including characteristics of the lin- rectness conditions, Lockean thoughts guistic community, can influence the lack content and so cannot be the content of our mental states is to source of our words’ content. allow that the inheritance relation Ideationalists need not be wedded may also run in the other direction. to Locke’s specific model of human These reflections suggest that psychology. Grice, for example, ideationalism is a tendency rather argued in 1957 that utterances inherit than a specific thesis. It is difficult to their meaning from the peculiar inten- deny that the meaning of linguistic tions with which they are performed. expressions and our utterances of If an utterance ‘Snow is cold’ means them depends in some way on lan- that snow is cold, it does so because it guage users having mental states. But was intended to cause the hearer to how that dependency is spelled out, believe that snow is cold, and to do so and whether the dependency ever runs through the hearer’s recognising this in the other direction, are questions very intention. Nothing here commits whose answer is as yet unsettled. Grice to Lockean ideas. In a second major potential problem for ideation- Primary sources alism, in 1975 devel- Grice, H. P. (1957). ‘Meaning’. Philosoph- oped a famous ical Review 66: 377–88. involving ‘Twin Earth’ to argue Putnam, H. (1975). ‘The Meaning of against the view, then prevalent, that a “Meaning”’. In Gunderson, K. (ed.), word’s content depends entirely on Minnesota Studies in the Philosophy of the internal mental states of the Science VII: Language, Mind, and person uttering it. Our intuitions Knowledge. Minnesota: University of appear to show that the content of an Minnesota Press. utterance of ‘water’ or ‘gold’ can Locke, J. (1706/1975). An Essay Concern- depend in part on factors outside the ing Human Understanding. Ed. P. Nid- utterer’s skull, such as the physical ditch. Oxford: Clarendon Press. composition of the colourless liquid in local lakes or the opinions of experts. Further reading Grice could embrace Putnam’s con- Davis, Wayne A. (2006). Nondescriptive clusion by allowing that the intentions Meaning and Reference: An Ideational from which our utterances inherit Semantics. New York: Oxford Univer- their content are not internal mental sity Press. states but external ones. That is, per- haps the content of our intentions is Alex Barber

85 IMPLICATURE weakest of the Achaeans was it that had smitten [Hector]’ has been usually taken to convey, when uttered by the When a sincere performance of a author of The Iliad (15.1) about Ajas, speech act takes place in a certain con- that Ajas was rather the strongest of text of utterance, what is conveyed by the Achaeans. The usage of such the performance, under the circum- expressions of understatement was stances, beyond what is then being lit- marked in early rhetorical studies as erally said by it, is an implicature of it. ‘litotes’ (Hoffmann 1987), but the theoretical problem of delineation and See also: Nonnatural Meaning; problem of derivation of what is con- Relevance Theory veyed but not said remained open Key Thinkers: Grice, H. P. until the second half of the twentieth ‘I meant what I said, and I said what I century. meant’, said an elephant called It was the philosopher H. P. Grice Horton in Horton Hatches the Egg, a who made the first crucial contribu- book for children by an author and tions to a theory of implicature that illustrator called Theodore Seuss purports to delineate whatever is con- Geisel, otherwise known as Dr Seuss. veyed beyond what is said when a cer- What Horton said about himself tain speech act is sincerely performed makes sense, because what Horton in a certain context of utterance and meant is not necessarily what Horton to show how to derive what is con- said. Generally speaking, when a veyed but not said from the speech act person performs a speech act in some and its context of utterance. Grice’s context of utterance, a full report of major ideas were made public during what the person did under the cir- the 1960s, first in a 1961 paper on the cumstances will have to include an causal theory of perception, and then answer to the question, ‘What was in the 1967 Harvard University conveyed by the utterance made by William James lectures, parts of which the speaker under the circumstances?’ were distributed and published in It is a major insight into the nature of various forms, and then in a revised language use that the required report version. They appeared in full in is dividable into two separate parts: Grice (1989), which includes also one, reporting what was said, and an related ‘Prolegomena’ and ‘Retrospec- additional one, reporting what was tive epilogue’. conveyed beyond what was said. Grice’s theory is about contexts of Interesting examples abound of utterance that involve a speaker, a speech acts that convey more than speech act, which includes a sentence what they say, but much less so sys- or other expressions, and a hearer. tematic depictions of what is con- Such a context of utterance is one veyed but not said by a certain speech in which the speaker performs a act in a certain context of utterance speech act intending to have some and systematic understanding of how communicative effect on the hearer. a speech act conveys what it does not Grice’s theory introduces two major say. The Homeric expression ‘not the ideas about communicative, ‘conver-

86 IMPLICATURE sational’ contexts of utterance. (Quality supermaxim) Roughly speaking, the first idea is that Try to make your contribution one that speech acts performed in conversa- is true. tional contexts of utterance are gov- erned by certain principles and (Quality maxims) maxims of conversation; the second (1) Do not say what you believe to be idea is that what is conveyed but not false. said is what follows from the speaker’s (2) Do not say that for which you lack observing those maxims and what fol- adequate evidence. lows from the speaker’s seemingly flouting those maxims. It is clear that (Relation maxim) such ideas can serve in solving the Be relevant. delineation problem of what is con- veyed but not said. It will become (Manner supermaxim) clear in the sequel that the same ideas Be perspicuous. are of much significance in an attempt to solve the derivation problem as (Manner maxims) well. We turn now to a brief presenta- (1) Avoid obscurity of expression. tion of these two ideas. (2) Avoid ambiguity. Speech acts that are performed (3) Be brief (avoid unnecessary prolix- in conversational contexts of utter- ity). ance are governed, according to (4) Be orderly. Grice’s theory, first and foremost, by what Grice labelled the Cooperative ‘and one might need others’. (Grice Principle: 1989: 27)

(CP) Make your conversational contri- Grice’s and bution such as is required, at the stage supermaxims and maxims apply to at which it occurs, by the accepted pur- conversations in different ways. The pose or direction of the talk exchange in following distinctions are crucial for which you are engaged. a proper understanding of Grice’s theory. For the sake of brevity, we will The fundamental (CP) gives rise to use the term ‘the norms’ for the Coop- supermaxims and maxims of conver- erative Principle, supermaxims and sation, classified by Grice into four maxims. groups, ‘echoing Kant’, entitled in terms of the categories of Quality, Conversational by Quantity, Relation and Manner: simple observation of the norms versus. conversational implicatures (Quantity maxims) by dramatic observation of the (1) Make your contribution as inform- norms ative as is required (for the current Given an ordinary context of utter- purposes of the exchange). ance, one often, or even usually, (2) Do not make your contribution assumes that the speakers who partici- more informative than is required. pate in the conversation observe the

87 IMPLICATURE norms. Conversational implicature by However, as every fluent speaker of the simple observation of the norms is language knows, the speech act of implicature that follows from the saying ‘I saw the movie’ under such cir- assumption that the norms have been cumstances counts as an appropriate observed, under the given circum- reaction to the posed question, being stances, in which there is no indication cooperative and informative, though – real or apparent – to the contrary. If a in an indirect and implicit way. First, speaker says, in an ordinary context of assuming the speaker does observe the utterance, ‘It is raining here’, we norms, we understand the speaker as assume the person is observing the holding that the movie is sufficiently norms, in particular Quality super- similar to the book, to the extent that maxim and maxim 1, given no indica- a person who is acquainted with the tion to the contrary. A conversational movie can be regarded as a person implicature by simple observation acquainted with the book. Notice that would be that the speaker believes, at such an understanding rests on the that context of utterance, that it is assumption that in reacting the way raining there. This explains the so- the speaker did, the speaker observed called Moore’s Paradox, which the Relation maxim and reacted in a emerges when a persons says: ‘It is rain- relevant way. Secondly, assuming the ing here, but I don’t believe it is’ or a speaker observed Quantity maxim 1 similar expression. The conversational and Quality maxim 1, we have a pos- implicature that follows from the first sible explanation for the speaker not part of the assertion contradicts what is answering the posed question in the said in the second part of the same affirmative. Hence, by assuming the assertion. speaker observed the norms, we reach Conversational implicatures by the conversational implicature that the simple observation abound and are of speaker had not read the book. Here much significance in ordinary interac- the question arises: If the speaker had tion. More interesting however, at least not read the book, why didn’t the from a theoretical point of view, are speaker admit it and simply answer conversational implicatures that are ‘No’? The response would be in terms drawn in contexts of utterance in of two insights we gain by trying to which the norms are actually observed apply the norms to the conversation though there are seeming indications under discussion. Notice, first, that the to the contrary. When a speaker answer ‘No’ would have been less answers the question, ‘Have you cooperative and less informative than read the present top best-selling book’, the reaction ‘I saw the movie’. The by saying, ‘I saw the movie’, the speak- latter helps the person who posed the ers apparently flouts the Cooperative question to make the next step in the Principle, because strictly speaking the conversation, pursuing the goal he reaction is not an answer. The question tried to reach, had the answer been is of the form called ‘yes’/‘no’-question ‘Yes’. If acquaintance with the movie is and the speaker’s reaction does not enough, the conversation will go on as seem to be the required one, because it planned for that stage of it. Moreover, includes neither a ‘yes’ nor a ‘no’. by assuming the speaker observed

88 IMPLICATURE

Quantity maxim 2, we draw the con- not all of my colleagues read fiction. versational implicature that the Such a conversational implicature is speaker believes that, for the sake of called ‘generalised’. the present conversation, it is sufficient to point out the fact that he had seen Conversational implicatures versus the movie, conveying but not admit- conventional implicatures ting that the speaker had not read the An implicature in general is what is book. conveyed by a performance of a speech The distinction between conversa- act beyond what is said when the tional implicature by simple derivation speech act is performed. According to from the norms and conversational Grice’s theory, it is possible to convey implicature by what Grice called ‘dra- by a performance of a speech act what matic’ derivation from the norms will is beyond what is said independently have to be manifest in each attempted of the norms of conversation which improvements on Grice’s theory. pro duce conversational implicature. There is a class of implicatures Grice Particularised conversational called ‘conventional implicatures’, implicatures versus generalised which do not rest on the norms of con- conversational implicatures versation but rather on conventions Our discussion of the conversational that govern the use of certain expres- implicatures that can be drawn from a sions. Consider an ordinary use of speech act of saying ‘I saw the movie’ ‘but’, such as ‘What they did was applied to the particular context of legally permissible, but ethically utterance, in which that speech act wrong’. On the level of what is said, served as a response to a particular one may replace ‘but’ with ‘and’ with- question about the speaker, ‘Have you out changing the propositional signifi- read the present top best-selling cance of the assertion and its truth- book?’. We can easily imagine con- value. However, the use of ‘but’ con- texts of utterance in which the same veys that there is some significant con- speech act appears, even in reference trast between the two parts of the to the same movie, made by the same assertion. The significance of the con- speaker, where none of the above- trast varies, but not its conveyed mentioned conversational implica- appearance. tures can be drawn. Hence, those Grice himself thought the nature of conversational implicatures are called conventional implicatures should be ‘particularised’ ones. better explained before ‘any free use However, not all conversational of it, for explanatory purposes’. Some implicatures are particularised. If a scholars have tried to reduce all con- speaker uses the expression ‘most’ in ventional implicatures to other phe- an ordinary speech act, a conversa- nomena of language use, such as tional implicature is going to be usu- conversational implicature, presuppo- ally drawn in terms of ‘not all’. Thus, sition* or what is said. Related theo- for example, an ordinary usage of retical debates are still ongoing. ‘Most of my colleagues read fiction’ Grice’s theory lends itself to three would conversationally implicate that types of study, related to the following

89 IMPLICATURE theoretical problems. First, what is the able from (R) without any resort to nature of the norms and what justifies (CP). each of the maxims? Second, do To move to our second question, speakers usually observe the norms of several types of empirical studies of conversation? Third, what alternative Grice’s theory should be mentioned. theories of language use would serve One type includes surveys of linguistic as improvements upon Grice’s theory, behaviour that result in apparent explaining the facts better and more counter-examples, such as people broadly? regularly making vague comments The first attempt to show the foun- or references in conversation. Such dation of Grice’s system of norms, dis- examples can, however, be shown to regarding debatable details, was be compatible with and even Kasher (1976), further developed in explained by Grice’s theory, where the what appears as an appendix in ‘at least cost’ part of (R) involves yet Kasher (ed.) (1998). The major idea of another parameter, such as commit- the explanation is that speech activity ment or guilt on grounds of being is ideally rational, that is, it acts shown to be mistaken, which should according to the Rationality Principle: be minimised. Empirical studies of other types are (R) Given a desired end, one is to related to different theoretical ways of choose that action which most effec- explaining data about what is con- tively, and at least cost, attains that end, veyed but not said (Bezuidenhout and everything else being equal. Morris, in Noveck and Sperber 2006), to the psychological processes of com- When the best means at one’s disposal puting conversational implicatures are the verbal ones, then one follows (Chierchia et al., in Noveck and Sper- (R) by performing a certain speech ber 2006), to the operation of impli- act. It is not difficult to explain the cature production devices in the brain ‘most effectively’ requirement of (R) (Kasher et al. 1999), and to child when a desired end is given, but it is acquisition of such devices (Noveck, much more difficult to elucidate the in Noveck and Sperber 2006). Such ‘at least cost’ requirement. A simple studies are related to the issue of mod- reading of Grice’s norms shows what ularity (in the sense of Fodor 1983), he took to be some of the costs, such where much remains to be studied. as verbal effort, related to number of Finally, let us consider three theo- words uttered and the time it takes to ries of implicature that have been voice them and similar attributes of developed in the footsteps of Grice’s speech. Other ‘at least cost’ parame- theory. The first two have been labeled ters include inaccuracies (avoided by ‘neo-Gricean’. All of them have fun- expressions of the Henry James style) damental principles of the above- and hurt feelings (avoided by using mentioned form ‘most effectively and politeness manifestations). Kasher at least cost’. (1976) includes arguments against Laurence Horn suggested (see Horn (CP) and arguments that show that 1984 and 2004) a replacement of supermaxims and maxims are deriv- Grice’s maxims by two principles:

90 IMPLICATURE

(1) The Q-Principle (where Q stands dominant ideas in pragmatics in the for quantity): make your contribu- 1980s and early 1990s. tion sufficient; say as much as you can, given the R-principle. Primary sources (2) The R-principle (where R stands for Fodor, Jerry (1983). The Modularity of relation): make your contribution Mind. Cambridge, MA: MIT Press. necessary; say no more than you Grice, H. P. (1989). Studies in the Way of must, given the Q-principle. Words. Cambridge, MA: Harvard Uni- versity Press. The Q-principle gives rise to Horn Hoffmann, Maria E. (1987). Negatio scales, such as . A Contrarii: A Study of Latin Litotes. sincere speech act that portrays a Assen: Van Gorcum. thesis as a good one has a Q-implica- Horn, Laurence R. (1984). ‘Toward a new ture that the thesis is not an excellent for pragmatic inference: one. Q-based and R-based implicature’. In Stephen Levinson (2000) suggested D. Schiffrin (ed.), Meaning, Form and an additional fundamental principle Use in Context: Linguistic Applications. of a different nature. The M-principle Washington, DC: Georgetown Univer- for the speaker (where M stands for sity Press. 11–42. manner): do not use a marked ex - Horn, Laurence R. (2004). ‘Implicature’. pression without reason. A marked In Laurence R. Horn and Gregory Ward expression indicates an unusual situa- (eds) (2004). 3–28. tion, while an unmarked expression Horn, Laurence R. and Gregory Ward indicates a usual one. A sincere speech (eds) (2004). The Handbook of Prag- act that describes a certain action as matics. Oxford: Blackwell. ‘not unreasonable’ has an M-implica- Kasher, Asa (1976). ‘Conversational ture that the action is not perfectly maxims and rationality’. In Asa Kasher reasonable. (ed.), Language in Focus: Foundations, Relevance theory*, as developed by Methods and Systems. Dordrecht: Sperber and Wilson (see Sperber and Reidel. 197–216. Republished with an Wilson 1987 and Wilson and Sperber appendix, ‘Gricean inference revisited’, 2004) and their followers, rests on in Kasher (ed.) (1998), vol. IV: 181–214. their principle of relevance, according Kasher, Asa, Gila Batori, Nahum Soroker, to which human cognition is ideally David Graves and Eran Zaidel (1999). geared to maximising relevance. Here ‘Effects of right- and left hemisphere a major shift from Grice’s theory is damage on understanding conversa- manifest. Whereas Grice was inter- tional implicatures’. Brain and Lan- ested in rational speech activity, guage 68: 566–90. relevance theory is about cognitive Levinson, Stephen C. (2000). Presumptive effects and processes. According to Meanings: The Theory of Generalized relevance theory, relevance is a func- Conversational Implicature. Cambridge, tion of cognitive effects, which should MA: MIT Press. be maximised, and of processing Noveck, Ira A. and (eds) efforts, which should be minimised. (2006). Experimental Pragmatics. Bas- Relevance theory was one of the ingstoke: Palgrave Macmillan.

91 INDETERMINACY

Sperber, Dan and (1987). ings of expressions and utterances Relevance: Communication and Cogni- precisely because there are no mean- tion. Oxford: Blackwell. ings about which to be definite. Mean- Wilson, Deirdre and Dan Sperber (2004). ing does not exist as an autonomous ‘Relevance theory’. In Laurence R. entity but only in observable behav- Horn and Gregory Ward (eds), The iours within a language community. Handbook of Pragmatics. Oxford: Writers starting from quite diverse Blackwell. 607–32. sets of initial assumptions have found themselves advocating this sceptical Further reading and, to many thinkers, counterintu- Carston, Robyn (2004). ‘Relevance theory itive thesis. and the saying/implicating distinction’. See also: Behaviourism; Holism In Laurence R. Horn and Gregory Ward Key Thinkers: Kripke, Saul; (eds), The Handbook of Pragmatics Quine, W. V. O.; Wittgenstein, Oxford: Blackwell. 633–56. Ludwig Davis, Wayne A. (1998). Implicature: Intention, Convention, and Principle in Indeterminacy in linguistics is most the Failure of Gricean Theory. Cam- commonly associated with W. V. O. bridge: Cambridge University Press. Quine’s 1960 discussion of translation. Gazdar, Gerald (1979). Pragmatics: Impli- Quine alleged that translation is inde- cature, Presupposition and Logical terminate because words themselves Form. London: Academic Press. do not have meaning; all that linguists Grandy, Richard E. and Richard Warner can do is observe patterns of behaviour (eds) (1986). Philosophical Grounds of in relation to particular language use. Rationality. Oxford: Clarendon Press. Quine’s ‘meaning scepticism’ finds an (See, in particular, ‘: a view of echo in a puzzle raised by . his work’ by the editors and the reply by Kripke’s puzzle purports to show that Grice.) we do not follow any determinate Kasher, Asa (ed.) (1998). Pragmatics: Crit- meaning-rule in applying a word in ical Concepts. Six volumes. London: novel circumstances. Routledge. (See in particular vol. IV In his 1960 discussion Quine is not which includes seventeen classical offering practical advice but scrutinis- papers on implicature.) ing the notion of meaning. For this reason he asks us to imagine we are Asa Kasher what he calls ‘radical translators’, developing a ‘translation manual’ – in other words, a recipe for going from sentences in some entirely unfamiliar INDETERMINACY language to sentences of our own lan- guage. To develop this manual we ‘Indeterminacy’ in a linguistic context may use only a narrow range of clues. has various interpretations, but in In particular we may not rely on the most senses it is taken to refer to the two languages having a common idea that linguists cannot make defi- origin and we certainly may not use a nite pronouncements about the mean- dictionary. The only evidence we may

92 INDETERMINACY use is behavioural: dispositions on the part of the same whole as’ rather than part of native speakers to assent or ‘is the same as’; the alternative manual dissent in varying circumstances to would then correctly predict assent. utterances of sentences of their lan- Quine is confident that, after comple- guage. Quine limits himself to this evi- mentary juggling of different clauses, dence on the grounds that sentence alternative but equivalent translation meanings, in so far as there are such manuals will always be available. things (and he ends up doubting there Translation, he insists, is a matter of are), must be constituted by trans- fitting a complete manual to all the latability on these terms. Allowing sentences of the other language (see other evidence would be presupposing holism*); Quine treats as myth the rather than coming to understand the view that successful translating is a nature of a sentence’s meaning. Inde- matter of revealing the one true mean- terminacy of translation is the thesis ing of individual words and sentences. that, in the context of radical transla- Some critics have argued that it will tion, there will always be at least two always be possible to isolate at most acceptable but incompatible transla- one translation scheme compatible tion manuals available. with assent/dissent patterns. There is Quine gives an example. Suppose certainly more room for manoeuvre that, as a rabbit scurries by, a native with Quine’s ‘Gavagai’ example than speaker utters the one-word sentence, he initially realised. In 1970 Quine ‘Gavagai’. You guess it means ‘Lo, a expressed regret at the focus on this rabbit’. Tests on future occasions particular example. He tried to show, using assent and dissent to this same quite generally, how his controversial sentence bear out your hypothesis. indeterminacy thesis follows from a Before adopting this as ‘the’ correct widely accepted underdetermination translation, however, you must rule thesis. According to this, empirical out alternatives. But this cannot be evidence rarely forces any particular done. For example, why not take scientific theory upon us. His infer- ‘Gavagai’ to mean ‘Lo, undetached ence from underdetermination of part of a rabbit’? In those few contexts theory to indeterminacy of translation where the difference between this and has failed to convince the majority of your translation shows up, Quine commentators. Other critics have holds that it will be possible to tweak questioned whether indeterminacy of other parts of the manual to accom- translation, even if correct, generates modate the discrepancy. Imagine you trouble for the notion of meaning. point to one part of a rabbit, then Most target the behaviourism* built another part of the same rabbit, and into his statement of the evidence solicit assent or dissent from ‘This gav- available to a radical translator. agai is the same as that gavagai’ – or Kripke’s 1982 indeterminacy puzzle rather, ‘Shiz gavagai sumo shaz gava- is inspired by his reading of Wittgen- gai’, as it may be. Native-speaker stein’s Philosophical Investigations assent here will not automatically rule (1953). To set up the puzzle, Kripke out the alternative translation manual. uses, for illustrative purposes, the We could take ‘sumo’ to translate as ‘is expression ‘+’ ; or, in spoken English,

93 INDETERMINACY

‘added to’. Suppose you are asked present question is ‘125’ rather than ‘What is 68 + 57?’ Suppose further that ‘five’. This is bad enough. But the this is a question you have never before sceptic also denies the possibility of had occasion to address (there will knowing whether you meant plus by always be some such question). You ‘+’in the past, and so mean plus now. answer with confidence: ‘125’. You are It is not a matter of ignorance as to assuming that the expression ‘+’ refers which you mean: there is simply no here to what it has always referred fact of the matter as to which you to, namely, the plus-function. But a meant in the past, or which you mean sceptic challenges you: ‘How do you now. know that ‘+’ has always referred to The puzzle can be generalised in the plus function?’ Perhaps it has various ways. First, it extends to always referred to the quus (pro- expressions other than ‘+’. Suppose nounced ‘qwus’) function, shown in you ask yourself, right now, ‘Is this a Figure 1. book I am reading?’ Your answer depends on what you mean by ‘book’. Perhaps ‘book’ refers now, as 5 if x = 68 and y = 57 in the past, to quooks, defined in x quus y = Figure 2. x plus y otherwise The conclusion we are invited to accept is that whenever we come to Figure 1 apply a familiar word in a novel con- text, there is nothing in virtue of which our application is correct or The only way you could ascertain incorrect, since nothing in previous which of the two functions ‘+’ referred uses determines that the word to earlier is to apply the expression in means one thing rather than another. the context of figuring out the sum of Moreover, since every context of a fifty-seven and sixty-eight. But by word being used was novel once, the hypothesis you could not have done puzzle applies to every use of any this. word. Finally, the puzzle is not just Kripke’s imagined sceptic has thus about communication. It applies even far claimed to establish only that you when we silently ask ourselves a do not at present know whether ear- question, so it is a puzzle about lier uses of ‘+’ referred to plus rather thought itself, not just about the than quus; and hence that you do not exchange of thoughts through spoken know if the correct answer to the language.

true if x is a book not called Key Ideas in Linguistics and the Philosophy of Language; x is a quook

false otherwise.

Figure 2

94 INDETERMINACY

The quus and quook readings be to say that there is no determinate cannot be dismissed on the grounds function from sentences and contexts that ‘quus’ and ‘quook’ are defined in of utterance onto propositions*. This terms of the more familiar ‘plus’ and apparent indeterminacy lies at the ‘book’, and are hence less simple. heart of the question of how we ought After all, it is equally possible to define to distinguish between words’ contri- the latter in terms of the former. Nor bution to an utterance’s meaning does it help to observe that, since we (broadly, semantics) and the utterance are disposed to reply ‘125’ rather than context’s contribution (broadly, prag- ‘five’, we must mean plus rather than matics). See Recanati (2004) and Cap- quus. If what we mean by a word is a pelen and Lepore and (2005) for product of how we are disposed to opposing views. apply it, it would be impossible ever to We have seen that challenges to make an erroneous application and determinate meaning can take differ- that would be absurd. Kripke himself ent forms. The interest attaching to tentatively offers what he calls a these challenges often lies not in sceptical solution. This seeks to the scepticism they generate about understand the role of meaning-state- meaning but in the resources devel- ments within discursive practice with- oped to meet them, and the clarity out allowing that they assert anything that emerges from reflecting on the true or false. Rather, they show some- assumptions that give rise to them. thing about the statement-maker’s willingness to use a term like ‘+’ or Primary sources ‘book’ in a particular way. Few have Cappelen, H. and E. Lepore (2005). Insen- embraced Kripke’s solution, but there sitive Semantics. Oxford: Blackwell. is no consensus over where his scepti- Kripke, S. (1982). Wittgenstein on Rules cal argument breaks down. and Private Language. Oxford: Black- Quine’s and Kripke’s discussions well. both threaten the widespread assump- Quine, W. V. O. (1960). Word and Object. tion that words and sentences have a Cambridge, MA: MIT Press. Chapter 2. particular meaning. A more recent lit- Quine, W. V. O. (1970). ‘On the reasons erature (albeit on a very old topic) also for indeterminacy of translation’. Jour- calls semantic determinacy into ques- nal of Philosophy, 67: 178–83. tion. Suppose I were to ask you ‘Have Recanati, F. (2004). Literal Meaning. Cam- you had breakfast?’. Am I asking you bridge: Cambridge University Press. whether you have had breakfast ever, Wittgenstein, L. (1953). Philosophical or just whether you have had breakfast Investigations. Oxford: Blackwell. this morning? It is tempting to appeal to the context. But after setting aside Further reading formal indexicals (‘I’, ‘here’, ‘tomor- Miller, A. (1998). Philosophy of Language. row’, and so on), context’s contribu- London: UCL Press. Chapters 4–6. tion to utterance meaning is difficult to systematise, even for ‘Have you had Alex Barber breakfast?’. To say that its contribu- tion is impossible to systematise would

95 INNATENESS (information about grammatical sen- tences), yet they successfully acquire language. The claim that some aspects of lin- From this basis one has to conclude guistic competence are genetically that humans are genetically hard- specified rather than learnt through wired for language: children are born experience. This claim has been driv- equipped with a universal grammar* ing research in generative linguistics (Chomsky 1981), containing informa- and language acquisition since the late tion about linguistic universals 1950s. enabling them to form hypotheses about the structure of the language See also: they are learning. The proposed exis- Acceptability/Grammaticality; tence of a critical period for language Continuity; Mentalism; (typical language acquisition is not Transformational-Generative possible after a certain age) and stud- Grammar; Universal Grammar ies of deaf children who sponta- Key Thinkers: Chomsky, Noam; neously develop sign language have Descartes, René; Plato been used as further evidence in sup- Noam Chomsky has proposed that port of linguistic nativism. humans possess domain- and species- Opponents of the Chomskyan view specific knowledge of the structure of claim that the richness of the data possible languages, which enables available to children is vastly underes- human young to acquire language timated; the stimulus argument fails with speed, efficiency and uniformity. only given the generative definition of This view can be traced back to Carte- what language is. One of the strongest sian cognitivism and Platonic philoso- challenges to nativism comes from phy. Opponents claim that language Connectionist psychology (Elman et acquisition is innately constrained but al. 1996) where artificial neural net- only by mechanisms that underlie gen- works have been trained to reproduce eral cognitive ability. In other words, linguistic behaviours such as recursion, it is currently uncontroversial that previously thought to be possible only language acquisition is innately con- through innate pre-programming. strained; what is the object of heated Research on innateness has spurred debate is exactly what is innate. research with animals and in genetics. The argument that has most force- In 2001 Cecilia Lai and colleagues sug- fully been used in support of the lin- gested that a mutation in the FOXP2 guistic innateness position is that the gene is causally involved in language stimulus argument, in which children disorders. Despite widespread enthusi- learn language by experience, is seri- asm in the popular press about the ‘lan- ously flawed (Chomsky 1980: 34). guage gene’, the exact role of FOXP2 For example, since language is a com- in relation to language development is plex system, it could only be acquired far from clear. through experience if negative evi- Although the innateness question dence was available. Children only is still unresolved, interdisciplinary ever have access to positive evidence research that straddles the gap

96 INTEGRATIONISM between linguistics, anthropology, challenge what they characterise as artificial intelligence, genetics and neu- ‘segregational’ linguistics, which treats roscience has opened up novel, excit- language as an autonomous object of ing ways in which the question can be study separate from non-linguistic addressed. aspects of communication, and argue forcefully that the attempt to decon- Primary sources textualise language leads to a pro- Chomsky, N. (1965). Aspects of the found distortion. Theory of Syntax. Cambridge, MA: See also: Acceptability/ MIT Press. Grammaticality; Conversation Chomsky N. (1980). Rules and Represen- Analysis; Creativity; (Critical) tations. Oxford: Blackwell. Discourse Analysis; Indeterminacy; Chomsky, N. (1981). Lectures on Govern- Systemic-Functional Grammar; ment and Binding: The Pisa Lectures. Structuralism; Holland: Foris Publications. Transformational-Generative Grammar Further reading Key Thinkers: Chomsky, Noam; Elman, J. L., E. A. Bates, M. H. Johnson, Firth, J. R.; Halliday, M. A. K.; A. Karmiloff-Smith, D. Parisi and K. Malinowski, Bronislaw; Saussure, Plunkett (1996). Rethinking Innateness: Ferdinand de A Connectionist Perspective on Devel- opment. Cambridge, MA: MIT Press. The dominant tradition in Western Hauser, M. D., N. Chomsky and W. T. linguistics over the twentieth century Fitch (2002). ‘The faculty of language: and into the twenty-first century has What is it, who has it, and how did it insisted on the autonomy of linguistics evolve?’ Science 298: 1569–79. as a science. This has meant defining Lai, C. S. L., S. E. Fisher, J. A. Hurst, and delimiting the subject in such a F. Vargha-Khadem and A. P. Monaco way that language can be investigated (2001). ‘A forkhead-domain gene is in isolation from other phenomena mutated in severe speech and language’. which might be thought to impinge on Nature 413: 519–23. it. It is accepted that areas of contact with the non-linguistic world may be Stavroula-Thaleia Kousta worth exploring, through sociolin- guistics, speech act theory*, seman- tics, discourse analysis* and so on; but these are all seen as in some way INTEGRATIONISM less ‘pure’ extensions from the investi- gation of the essential linguistic core, An approach to linguistics which is which is autonomous syntax. Integra- radically opposed to the assumptions tionism, which grew from a series on which much of mainstream linguis- of books published by Roy Harris tics has been based since the time of from 1980 on, rejects this stance , and insists that completely. Following Harris’s lead, language must be seen as action integrationists have set out to demol- embedded in context. Integrationists ish what they see as the suspect

97 INTEGRATIONISM

foundations on which the linguistic obscurity and ambiguity* in order to edifice has been constructed. There maintain the smooth flow of commu- have been integrational studies of nication. Speakers are often uncon- areas such as fictional dialogue, , cerned with avoiding these apparent legal arguments and so on, but the pri- hindrances, secure in the knowledge mary aim has been to force main- that the hearer will, all things being stream linguists to at least question, if equal, collaborate in making enough not revise, many assumptions that sense of the utterances for communi- they have taken for granted. cation to take place satisfactorily. A The integrationist critique high- valid integrationistmodel of linguistic lights three key interrelated problems behaviour cannot relegate this to ‘per- with the linguistic tradition founded formance factors’ which do not affect on Saussure’s work. The first is what the underlying ‘competence’: it is a Harris in his 1981 book terms the key distinguishing feature ofcom - myth of ‘telementation’. This is the munication which must be taken into image (introduced in a famous dia- account. gram in Saussure 1916/1922 of two The second point on which integra- heads with arrows passing between tionists take issue with the mainstream them representing ideas being trans- is the determinacy of the linguistic ferred) of language as serving to sign. If, as in the telementation model, convey the thoughts of the speaker ideas were transferred undistorted, this into the mind of the hearer. The ideas would have to rely on signs (words) are assumed to be transferred undis- having determinate meanings which torted, fully comprehensible and fully are known to, and accepted by, both comprehended. Integrationists argue interactants. But, as Harris and others that, even when mainstream linguists have stressed, if this were the case, claim publicly not to work with this lawyers would be completely unneces- model of communication, their prac- sary, since much of their work involves tice indicates that they do. A funda- negotiating over the precise applica- mental flaw in the model is that there tion of laws to individual cases – that is in fact no way of knowing to what is, exactly what the law ‘means’ in a extent, or even whether, the meanings particular context. This seems odd, understood by the hearer are those since legal language is designed to be as intended by the speaker. It also exact and determinate as possible. If restricts ‘meaning’ to ‘ideation’, thus indeterminacy* occurs here, it is even ignoring interpersonal or other aspects more likely to be inherent in everyday of meaning (compare systemic- uses of language. For integrationists, functional grammar*, where these interactants are engaged in a constant other types of meaning are fully inte- process of making sense of communi- grated in the model). Experience cation in particular contexts. They can shows that the telementation model draw on their experience of previous does not correspond to reality: very situations which they perceive as simi- often hearers do not understand every- lar, but these of similarity thing that they hear, and are content are not dependent on some kind of to cope with degrees of vagueness, communally held ‘dictionary’ which is

98 INTEGRATIONISM the final arbiter – they are the result ers (as represented by the linguist of general problem-solving abilities him/herself), and around this is a applied (unconsciously and instanta- periphery of variation in forms, neously, because of years of practice) including the usages of individuals or to seeing what function utterances groups (such as dialects), which have need to serve in context in order to be an uncertain relationship to the code an appropriate contribution to the but which can be left to sociolinguists communicative event. Signs are inher- and other scholars with ‘fringe’ inter- ently indeterminate, and are only ests to deal with. Integrationists rendered determinate (to whatever would see this as merely ducking the degree is judged necessary by the inter- issue (and, in the process, taking an actants) in particular contexts. Taking sweepingly arrogant view of society). this to the next level, segregationist For them, language is in a state of con- approaches are underpinned by the stant flux, geographically and histori- view that a language is a fixed code, so cally but also moment-by-moment as that sentences can be deemed unequiv- speakers negotiate and create mean- ocally to be ‘grammatical’ (part of the ing in context; and generalisations code) or ‘ungrammatical’ (not part of about the characteristics of a particu- the code). However, for the integra- lar ‘language’ will necessarily exclude tionist this is an equally untenable dis- the variability that is fundamental to tortion: this could only be true of a the way that language operates. grammar held in the mind of the ‘ideal A charge that has been levelled speaker-hearer in a homogeneous against integrationism is that its speech community’ – an imaginary robust attacks on mainstream linguis- entity that does not, and cannot, exist. tics have not been balanced by any The third point of contention clear programme or methodology to involves a claim which seems counter- replace current practices. Michael intuitive, but which follows logically Toolan counters this by arguing that from the points above: ‘language’ such a complaint ‘reflects a way of exists but ‘languages’ do not. Speakers thinking deeply committed to the pre- do, of course, find it useful to refer vailing post-Saussurean cognitivist- to ‘languages’ such as English and mechanistic paradigm’ and that the French, but the integrationist sees that term integrational linguistics ‘names a as a convenient fiction which is unnec- principle rather than a method’ (1996: essary and misleading as a basis for 23). He notes that much of the work linguistics. ‘A language’ implies deter- in ethnography, conversation analy- minate signs and a fixed code which sis* and discourse analysis is in prac- are, at least potentially, shared by all tice congruent with integrationism, speakers of that language. But who and the way forward will probably decides which parts of ‘English’ involve adapting methods from those belong to the code and which do not? approaches. A more theoretical Some linguists would attempt to get counter-attack from a ‘segregationist’ round this question by arguing that position (though they would reject the there is a fixed core that is recognised label) can be found in Borsley and and accepted by all (educated) speak- Newmeyer (1997) who insist that

99 INTENTIONALITY mainstream linguistics does not in fact Muddle: Roy Harris and Generative work with a telementation model. Grammar’. In G. Wolf and N. Love (eds), This paper, and the response in Toolan Linguistics Inside Out. Amsterdam. (1998), give a flavour of the debate. Philadelphia: John Benjamins. 42–64. It is hard to judge how far thein - Saussure, Ferdinand de (1922/1983). tegrationist critique has altered the Course in General Linguistics. Second course of linguistics as its proponents edition. Trans. Roy Harris. London: would wish. An increasing number Duckworth. of approaches have independently Toolan, Michael (1998). ‘A few words on arrived, by different routes, at rather telementation’. In R. Harris and G. similar conclusions, and key aspects of Wolf (eds), Integrational Linguistics. the integrationist position (such as the Oxford: Pergamon. 68–82. indeterminacy of the sign) have them- selves become mainstream, though Geoff Thompson generally under different labels. Inte- grationism can claim to have con- tributed to weakening the hegemony of a particular view of linguistics,but INTENTIONALITY there is as yet little sign that it has brought about its demise. Directedness or aboutness; the prop- erty of mental states whereby they are Primary sources about, or directed towards, states of Davis, Hayley G. and Talbot. J. Taylor affairs in the world, typically (eds) (1990). Redefining Linguistics. expressed in language through such London: Routledge. ‘intentional verbs’ as ‘believe’, Harris, Roy (1981). The Language Myth. ‘desire’, ‘know’, and ‘intend’. Inten- London: Duckworth. tionality is an important concept in Harris, Roy (1998). Introduction to Inte- both and ethical grational Linguistics. Oxford: Perga- theory as a way of recognising respon- mon. sibility towards other beings, and in Harris, Roy and George Wolf (1998). distinguishing humans and other Integrational Linguistics: A First higher animals from computers. Reader. Oxford: Pergamon. See also: Artificial Intelligence; Toolan, Michael (1996). Total Speech: An Nonnatural Meaning; Integrational Linguistic Approach to Phenomenology; Speech Act Language. Durham/London: Duke Uni- Theory versity Press. Key Thinkers: Husserl, Edmund; Wolf, George and Nigel Love (eds) (1997). Searle, John Linguistics Inside Out: Roy Harris and His Critics. Amsterdam/Philadelphia: Intentionality is an important concept John Benjamins. in both continental and analytic phi- losophy. It was introduced by the Further reading German philosopher Franz Brentano, Borsley, Robert and Frederick, J. and developed by his pupil Edmund Newmeyer (1997). ‘The Language Husserl, for whom it is one of the

100 INTENTIONALITY foundational and essential concepts of around his room, Husserl describes phenomenology. Intentionality is also the experience of perceiving a sheet of recognised as Husserl’s most impor- white paper in a dim light, to draw a tant influence on the analytic tradition distinction between perceptual experi- of philosophy, in which it is especially ence and something perceived. The associated with the speech act theory* concrete experience of the paper of John Searle. appearing from a particular angle, The term ‘intentionality’ was first with lack of clarity because of the used in its modern sense by Brentano light, is a conscious experience, and in his Psychology from an Empirical description of that experience is Standpoint (1973, first published phenomenological description, as 1874), although the idea is ultimately opposed to a scientific description of derived from Aristotle’s concept of something perceived. A description ‘mental inexistence’, the notion that of consciousness and a description when one thinks of an object, one has of phenomenal experience are for an object ‘in mind’, but the object in Husserl one and the same thing: ‘the mind does not ‘exist’ in the same way basic character of intentionality [is] that the object in the world exists. the property of being a “conscious- According to Brentano (1973: 88), ness of something”’ (Husserl 1931: ‘every mental phenomenon is charac- 120). Moreover, ‘the pregnant mean- terised by . . . the intentional (or ing of the expression cogito’ (as in mental) inexistence of an object, and Descartes’ cogito ergo sum, ‘I think what we might call . . . reference to a therefore I am’), is ‘I have conscious- content, direction toward an ness of something’, or ‘I perform an object, . . . or immanent objectivity’. act of consciousness’ (Husserl 1931: Brentano’s purpose here is to distin- 118). Thus for Husserl the Cartesian guish the mental from the physical, cogito is also an intentional act, and and in turn to define the mental and to Descartes is unwittingly the true characterise mental states. According father of intentionality. to Brentano, all mental phenomena More recently, intentionality has are intentional; indeed, that some- become of interest to philosophers thing is intentional defines it as a working in the analytic tradition. In mental phenomenon. This is what has his ‘Minds, Brains and Programs’ become known as an ‘irreducibility (1980), John Searle demonstrates that thesis’: the mental cannot be reduced a programmed digital computer to the physical. cannot have cognitive states such as Brentano’s biggest influence was on understanding. He does this through a Husserl, who in his Ideas (1913) famously elegant thought experiment extends the concept of intentionality called the ‘Chinese Room’. Searle to all experiences of thought, feel- imagines himself locked in a room and ing or will: intentionality and phe- given a batch of Chinese writing. He is nomenal experience are for Husserl then given a second batch of Chinese inextricably linked. In a celebrated writing, with a set of rules (in English) pastiche of René Descartes’ descrip- correlating the second batch with the tion of the experience of looking first. Finally, he is given a third batch

101 INTENTIONALITY of Chinese writing, with instructions, conversely’ (ibid.). Nevertheless, the again in English, for correlating ele- speech-acts distinction between ments of the third batch with the first propositional content and illocution- two batches. Unknown to Searle, who ary force has a parallel in intentional understands no Chinese, the people states, where it is expressed as a dis- giving him the batches of Chinese tinction between representative con- writing call the first a ‘script’, the tent and psychological mode. For second a ‘story’, and the third ‘ques- example, a sentence containing a tions’. Eventually Searle becomes so speech-act verb, such as ‘I predict that good at following the instructions for you will leave the room’, can be manipulating the that, from analysed into its illocutionary force of the point of view of someone outside predicting, and its propositional con- the room, his answers to questions are tent that you will leave the room. A indistinguishable from those of a sentence containing an intentional native Chinese speaker. However, we verb, meanwhile, such as ‘I believe remember that Searle still understands that you will leave the room’, can be no Chinese. Unlike the native Chinese analysed into its psychological mode speaker, Searle in his room is merely of belief, and its representative con- manipulating symbols. This shows tent that you will leave the room. Fur- that although a computer might in ther, just as there is a range of ‘fits’ of theory be programmed with the sum speech acts – they may be true or false, total of all information there is to be obeyed or disobeyed, kept or broken, had about the world, this would not for example, depending on whether mean that the computer understood the acts are assertive, directive, com- the world. Understanding is an missive and so on – so there is a range intentional phenomenon, along with of ‘fits’ of intentional sentences. perceiving, acting and learning. Com- Beliefs are true or false, while inten- puters merely manipulate symbols, tions are complied with or not com- and hence are incapable of being plied with, and desires are fulfilled or attributed with these intentional phe- unfulfilled. Moreover, an intentional nomena. state is a sincerity condition of its type Searle elaborates this position in of speech act; ‘the performance of the two subsequent essays, ‘What is an speech act is eo ipso an expression of Intentional State?’ (1982) and Inten- the corresponding intentional state’ tionality: An Essay in the Philosophy (Searle 1982: 263), so that it sounds of Mind (1983), which connect inten- odd to assert, for example, ‘Congrat- tional states with speech acts. Accord- ulations on winning the prize, but I ing to Searle, ‘intentional states am not glad you won the prize’. And represent objects and states of affairs finally, speech acts and intentional in exactly the same sense that speech states are linked by ‘conditions of - acts represent objects and states of isfaction’ or ‘conditions of success’. affairs’ (1982: 260). This is not to say Just as a statement is satisfied if and that intentionality is essentially lin- only if it is true, a promise is satisfied guistic: on the contrary, ‘language is if and only if it is kept, and so on, so a derived from intentionality, and not belief is satisfied if and only if it is true,

102 INTUITION a desire is satisfied if and only if it is Searle, John R. (1983). Intentionality: An fulfilled, and so on. In short, inten- Essay in the Philosophy of Mind. Cam- tional states, like linguistic entities, bridge: Cambridge University Press. are not things, but representations – they have logical properties, not onto- Further reading logical ones. Dreyfus, Hubert L. (1992). What Com- The concept of intentionality con- puters Can’t Do: A Critique of Artificial tinues to be of relevance today, since it Reason. Cambridge, MA and London: is recognised as essential to the ascrip- MIT Press. tion of consciousness to beings. The Harney, Maurita J. (1984). Intentionality, question of how intentionality is pro- Sense and the Mind. The Hague: Nijhoff. duced, and what the criteria are for having a mental state, is therefore Karl Simms important to computer scientists and philosophers of cognitive science. (1987, 1993), for example, takes a contrary view to INTUITION Searle, being prepared to ascribe intentionality to computers and to Speakers’ introspective judgements various animals if they manifest what about aspects of their language. A is perceived to be intentional behav- source of evidence for theories of lan- iour. guage and cognition which has been used extensively in recent decades, Primary sources particularly under the influence of Brentano, Franz (1973). Psychology from Noam Chomsky. an Empirical Standpoint. Trans. A. C. See also: Acceptability/ Rancurello, D. B. Terrell and L. L. Grammaticality; Adequacy; McAlister. London: Routledge. Corpora; Creativity; Dennett, Daniel C. (1987). The Inten- Empiricism/Rationalism; tional Stance. Cambridge, MA and Innateness; Mentalism; London: MIT Press. Transformational-Generative Dennett, Daniel C. (1993). Consciousness Grammar; Universal Grammar Explained. London: Penguin. Key Thinkers: Austin, J. L.; Husserl, Edmund (1931). Ideas: General Chomsky, Noam; Greenberg, Introduction to Pure Phenomenology. Joseph; Katz, J. J.; Sinclair, John Trans. W. R. Boyce Gibson. London: Allen & Unwin. Informant intuitions about language Searle, John R. (1980). ‘Minds, brains, have been a major source of data for and programs’. The Behavioral and linguists since the work of Chomsky Brain Sciences 3: 417–57. was first published in the late 1950s. Searle, John R. (1982). ‘What is an inten- Chomsky argued that there was clear tional state?’. In H. L. Dreyfus with H. evidence that speakers share system- Hall (ed.), Husserl, Intentionality, and atic intuitions about their languages Cognitive Science. Cambridge, MA and and that these should be used as London: MIT Press. data in studying language. Chomsky’s

103 INTUITION work led to major changes in linguis- consistent intuitions about this and so tics and in psychology (often referred this constitutes good data about their to as ‘the Chomskyan revolution’) underlying system of linguistic knowl- and a large amount of work in these edge. A large amount of work in lin- disciplines is now based on data from guistics and psychology has been intuitions. based on evidence from intuitions like Chomsky’s work changed linguis- this. tics in several ways. He introduced a From the outset, there was consid- new way of thinking about what was erable discussion of issues associated the object of study (mental grammars) with the use of intuitions as data. and also about the methods to be used Chomsky himself pointed out that not in studying it. Before Chomsky’s all intuitions are as clear, or as clearly work, the most common methods shared, as the intuitions about exam- used by linguists were ‘discovery pro- ples like (1)–(2), and he suggested that cedures’, which were techniques for theories should be based as much as ‘discovering’ facts about the language possible on the clearer cases. Other being studied. Chomsky argued that linguists have questioned the useful- these techniques must be based on a ness of intuitions at all, arguing that it number of factors, including intuition, is better to look at naturally-occurring and also that speakers’ intuitions pro- data, such as data gathered in a vide evidence for the existence of an corpus. Chomsky has always rejected internalised language system which this view, pointing out what he sees as should be the object of study for lin- serious problems with such naturally- guistics. He suggested that the best occurring data. One problem is that method for studying the system was to there is an element of luck in whether explore the intuitions. One famous a corpus will provide relevant exam- data set discussed by Chomsky con- ples. Related to this is the fact that a cerns the examples in (1) and (2): corpus cannot provide negative evi- dence; it cannot show us what is not (1) (a) John is easy to please. possible in a language. We cannot, for (b) It is easy to please John. example, use corpus data to find out (2) (a) John is eager to please. that sequences like ‘It is eager to (b) It is eager to please John. please John’ (with the same reading as (2a)) or ‘Which student did Sylvia tell Chomsky pointed out that (2b) is not the teacher who taught to go home’ a possible way of expressing the same are not possible in English. thing as is expressed by (2a) (it is a But the reasons why these are not possible utterance, of course, but only possible are important if we aim to if we take ‘it’ to refer to a particular understand the nature of our language entity capable of being eager to please system. Another reason is that Chom- someone). There is no logical reason sky sees the object of study as an why (2b) could not be a way of saying internalised system, which he origi- the same thing as (2a) by analogy with nally termed ‘competence’ (Chomsky the pattern in (1a) and (1b), but this is 1965) and distinguished from not possible in English. Speakers have language in actual usage, which he

104 INTUITION termed ‘performance’. A large num - sun and observe directly what is hap- ber of irrelevant factors affect per- pening there. But we can observe light formance, for example interruptions, and heat coming from the sun and reformulations, the physical and make inferences about what must be mental state of the speaker. Any causing what we observe. Similarly, corpus shows many examples of we cannot look directly at the compe- ‘ungrammatical’ utterances involving tence of speakers but we can make overlap, false starts, hesitation, repeti- inferences based on their intuitions tion, reformulation, unfinished utter- about particular examples. ances and so on. Chomsky argued Birdsong (1989) and Schütze that what we should do is look for (1996) considered methodological sophisticated ways of finding data issues with the use of intuitions, and which reveal facts about the underly- used the term ‘metalinguistic perform- ing system of competence. The best ance’ to describe what speakers are data, he argued, come from the intu- doing when they make these judge- itions of speakers. ments. This term makes clearer what What are the intuitions exactly? the judgements are and highlights the They are often referred to as ‘gram- fact that this approach is using per- maticality judgements’ but this is formance data to explore competence. slightly misleading, particularly if we One major issue which all re- take seriously the distinction between searchers need to address is that the ‘grammaticality’ and ‘acceptability’. intuitions of speakers vary. While Chomsky argues that mental gram- most speakers will agree on (1)–(2) mars are systems of ‘tacit’ knowledge: above, a large number of examples are speakers know that (2b) does not judged differently by different speak- work like (1b) but they cannot explain ers. One source for the variation in why. There are also a number of well- judgements is that all speakers have known examples which speakers slightly different language back- reject but which are nevertheless con- grounds and experiences resulting not sidered to be grammatical, and vice just in ‘dialectal’ variation (variation versa. Strictly speaking, then, speak- based on where the speaker lives or ers cannot be relied on to make judge- has lived) but also ‘idiolectal’ varia- ments about ‘grammaticality’ but only tion reflecting the speaker’s individual about ‘acceptability’, that is about linguistic history. While there is sig- whether particular examples ‘sound nificant agreement about examples OK’ or whether they can imagine like (1)–(2), there would be variation hearing or saying them. in responses to (3)–(4): This means that intuitions are quite far removed from the system of gram- (3) He and Sylvia will go. mar which is the object of study. (4) All of the staff are pleased to work Chomsky (1980: 189–92) acknowl- here. edged this remoteness, making an analogy with the study of thermonu- In fact, Chomsky has suggested that clear reactions inside the sun. We language is ‘an individual phenome- cannot set up a laboratory inside the non’ and that ‘no two individuals

105 LANGUAGE GAMES share exactly the same language’ data should be used instead. Others, (1988: 36). like Schütze, envisage research carry- Given this variation, linguists need ing on using a range of kinds of to make idealisations about the lan- data. For methodologically eclectic guage they are studying. They need linguists, the range of possible either to abstract away from the vari- sources of data includes elicitation ation and imagine a group of speakers (engaging speakers in conversations who do share a language, or to focus which are likely to elicit utterances of on describing the competence of just the forms being considered), inter- one speaker. In practice, for many lin- views, questionnaires, experiments guists, the object of study is their own and corpora. mental grammar. But even individual speakers do not always make the same Primary sources judgements about particular exam- Chomsky, N. (1965). Aspects of the ples. So linguists need to be careful not Theory of Syntax. Cambridge, MA: to accept too readily the judgements MIT Press. of informants. A number of studies Schütze, C. T. (1996). The Empirical Base have compared actual recorded usage of Linguistics: Grammaticality Judge- of speakers with what they report ments and Linguistic Methodology. about themselves and found signifi- Chicago: Chicago University Press. differences. Schütze (1996) makes a number of Further reading suggestions to help address these diffi- Birdsong, D. (1989). Metalinguistic Per- culties. He suggests that it is legitimate formance and Interlinguistic Compe- to continue to use intuitions as data tence. New York: Springer-Verlag. but suggests that certain ‘precautions’ Chomsky, N. (1957). Syntactic Structures. should be followed and that linguists The Hague: Mouton. should aim to develop other tech- Chomsky, N. (1980). Rules and Represen- niques, in particular experimental tations. Oxford: Basil Blackwell. work. As well as a number of specific Chomsky, N. (1988). Language and Prob- suggestions about materials and the lems of Knowledge: The Managua Lec- procedure of gathering intuitions, one tures. Cambridge, MA: MIT Press. of the main precautions urged by Schütze is that subjects who provide Billy Clark intuitions should never themselves be linguists since there is a risk that their training will interfere with their judgements. LANGUAGE GAMES Chomsky has been resolute in rejecting alternative kinds of data, The idea that language use can be suggesting for example that gathering compared to a game, where conver- corpus data is unlikely to lead to sational participants are the players significant insights. Other linguists and the goal of their conversation suggest that intuitions are unreliable can be reached if they perform and that ‘naturally-occurring’ corpus certain types of moves within the

106 LANGUAGE GAMES

context of publicly-known rules. work of game-theoretical semantics, The philosophical outline was most an account closely related to Paul famously provided by Ludwig Lorenzen’s dialogue games (Lorenzen Wittgenstein and further developed 1955, Lorenzen and Lorenz 1978). In by a number of researchers in logic*, this approach, semantic games linguistics, artificial intelligence* and between a verifier (the player who is computation. trying to prove that the statement is true) and a falsifier (proving that the See also: Artificial Intelligence; statement is false) are used to evaluate Implicature; Speech Act Theory logical formulas in a negation normal Key Thinkers: Grice, H. P.; Lewis, form. For example, a formula in David; Wittgenstein, Ludwig which the outmost component is an The use of games as paradigms for existential quantifier would be inter- dialectical situations involving a ques- preted as a game in which the verifier tioner and an answerer goes back to has the first move, selecting a witness Aristotle. The idea that all conversa- to the truth of the formula. The veri- tion can be understood in terms of a fier thus substitutes a variable with a language game, however, is com- proper name referring to an entity monly ascribed to Ludwig Wittgen- which substantiates the property or stein who utilises the term to refer to relation contained in the formula. the complex of a particular language Similarly, the presence of a disjunction and the actions in which it is prompts a verifier’s move in that she anchored. In Philosophical Investiga- chooses which disjunct she will sup- tions, Wittgenstein strives to under- port. Universal quantifier and con- stand how we acquire the meanings of junction, on the other hand, are words. He notes that simply knowing moves appointed to the falsifier. A how to name a concept is not suffi- player is said to have a winning strat- cient for us to be able to use the word egy (a proof) if she can win disregard- in a conversation; we must also know ing the choices of the opposing player. about the conventions or rules gov- The semantic game can become more erning its use. In this sense, language complex if we impose imperfect can be compared to a game of chess; if knowledge on the players, for exam- we were to teach someone how to play ple by making them ‘forget’ some of chess, it would not be sufficient to give their earlier moves (imperfect recall). them the names of the chess figures, Hintikka’s game-theoretical semantics we would also have to explain to them has received important applications in how the figures can move on the chess logic and on fragments of natural board. Depending on the context, the language. use of a word or a sentence that exists Other recent developments in the in the language – a move – can be field of language and games have pri- viewed as more or less appropriate; marily been based on research in game non-existing language elements are theory. Classical game theory describes not moves at all. the behaviour of interacting rational Jakko Hintikka further extends agents who take into account their Wittgenstein’s notion in his frame- knowledge of each other’s preferences;

107 LANGUAGE GAMES it makes use of tools employed by For example, in Prashant Parikh’s utility and decision theory. Evolution- game-theoretic model of com- ary game theory includes a description munication (an extensive cooperative of the changes in players’ strategies as game with partial information), the they dynamically adapt to their envi- hearer disambiguates a speaker’s ronment. utterance by reasoning about its Various lines of research have been alternatives and by taking into based on the idea that language can account the cost associated with pro- be compared to a coordination ducing (and processing) complex game (Lewis 1969; Parikh 2001), in expressions. accordance with H. P. Grice’s view According to other accounts, a con- of a conversation as an interaction versation can more appropriately be between rational agents with a described as a game of opposing inter- shared goal. In a coordination game, ests (a zero-sum game) or a bargaining players’ ‘payoffs’ are aligned – that is, (mixed-motive) game, in which the unlike in other sorts of games exam- interests of the players are only par- ined in the context of game theory, tially aligned (Merin 1994). According the players do not have opposing to Arthur Merin’s Algebra of Elemen- interests. Based on their prior beliefs tary Social Acts, the general purpose and preferences, the agents would of the game is to establish the con- interact in a way that leads them to a tent of the conversational common good decision as to which actions to ground. The players must have strictly perform (problem-solving games) or, opposing preferences regarding the in a special case of problem solving, adoption of the proposition under dis- which propositions to consider to be cussion, since otherwise there would true (inquiries). The shared knowl- be no point debating it. The commu- edge of the purpose of the game helps nicating agents also have different bar- them to code and interpret meaning gaining powers in the game, expressed efficiently and successfully. For by the dominance parameter. Different example, in David Lewis’s signalling states of the game – actually, different games, a signalling system (in fact, a types of dialogue acts (claim, conces- language) evolves between a sender sion, denial and retraction) – are char- and a receiver if the receiver reacts acterised as sets of values for the appropriately to the sender’s signal dominance parameter, as well as the by choosing an action that matches actor role, initiator role and preference the state of the world. The speaker with respect to the proposition under and the receiver can be compared to discussion. two people travelling in a boat; the As observed in a number of studies, role of the sender is to signal possi- game-theoretical approaches to com- ble danger ahead; and the role of munication offer the possibility to the receiver is to steer the boat account for the construction of form- accordingly. A signalling system will meaning pairs in terms of solution evolve if the interaction results in suc- concepts. A well-known solution con- cess above chance level, depending cept is the Nash Equilibrium – a pro- on the strategy the agents employ. file with the players’ best response to

108 LANGUAGE GAMES the choices of other players, all of metaphor rather than as a vehicle for whom are trying to maximise precise formalisation. their payoff. Dekker and Van Rooy (2000) propose to use Nash Equilib- Primary sources ria to characterise winning candi- Benz, A., G. Jaeger and R. Van Rooij dates in two-dimensional optimality (2005). Game Theory and Pragmatics. theory, an extension of optimality Basingstoke: Palgrave Macmillan. theory* employed on the semantics- Hintikka, J. (1973). Logic, Language pragmatics interface. Under the Games and Information: Kantian two-dimensional optimality theoretic Themes in the . interpretation, a form-meaning pair is Oxford: Clarendon Press. optimal if it satisfies both the speaker’s Lewis, D. (1969). Convention. Cam- and the hearer’s communicative goals, bridge, MA: Harvard University Press. originally expressed by the Gricean Merin, A. (1994). Algebra of Elementary maxims and later subsumed under the Social Acts. In Tshatzidis, S. (ed), Foun- Q- and R-principle (Horn 1984). The dations of Speech Act Theory. London: principles capture the intuition that Routledge. the speaker’s and the hearer’s attempt Parikh, P. (2001). The Use of Language. to minimise communicative effort Stanford, CA: CSLI Publications. leads to opposing preferences (for the Wittgenstein, L. (1958). Philosophical speaker, to say as little as possible; for Investigations. Trans. G. E. M. the hearer, to maximise the informa- Anscombe. Oxford: Blackwell. tion value of the message). If the speaker acted only on the basis of her Further reading own preferences, our language would Dekker, P. and R. Van Rooy (2000). ‘Bi- consist only of expressions involving directional optimality theory: an appli- low articulatory effort with a high cation of game theory’. Journal of degree of homonymy. A division of the Semantics 17: 217–42. pragmatic labour results in the use of Horn, L. R. (1984). ‘Towards a new tax- unmarked forms in unmarked situa- onomy for pragmatic inference: Q- tions and marked forms in marked sit- based and R-based implicatures. In D. uations. Schiffrin (ed.), Meaning, Form, and Use Other recent accounts of pragmatic in Context. Washington: Georgetown phenomena in which the authors University Press. 11–42. make use of game-theoretical results Lorenzen P. (1955). Einführung in die concern indirect speech acts and operative Logik und Mathematik. underspecification, credibility, ques- Berlin: Springer. tion-answer pairs and grounding Lorenzen, P. and K. Lorenz (1978). Dialo- (Benz, Jaeger and van Rooij 2005). gische Logik. Darmstadt: Wis- The game jargon has also been utilised senschaftliche Buchgesellschaft. in computational analyses of dia- Osborne, M. J. and A. Rubinstein (1994). logue, where dialogue acts are charac- A Course in Game Theory. Cambridge, terised as ‘moves’ appropriate for MA: MIT Press. different game types. Here, the idea of a language game serves as a loose Marie Nilsenová

109 LANGUAGE OF fore, in which LOTH is an alternative to the theory of intentionality*. THOUGHT LOTH is predicated on a number of theses. The first is so-called ‘represen- The Language of Thought Hypothesis tational realism’, or the holding of a is the hypothesis that thinking and Representational Theory of Mind thought are conducted in a mental (RTM). According to this theory, language (‘mentalese’) that is innate, there is a unique, distinct, dedicated distinct from all natural languages, psychological relation for each propo- universal among all thinking beings, sitional attitude, and each thought and physically realised in the brain. incorporating an attitudinal verb is a The language of thought is anim - token of this mental type. Thinking is portant concept in the attempt to thoughts joined up – in other words, find a cognitive and neurological causal sequences of tokenings of this explanation for language and con- mental representation of the attitude, sciousness. be it belief, desire, or whatever. It fol- lows that there is a strong rationalist See also: Innateness; Intentionality; in LOTH: thinking is defined as Mentalism; Private Language; rational thought, the ratiocinative Propositional Attitudes; process consisting of thoughts in Propositions; Universal Grammar causal sequence. Key Thinkers: Chomsky, Noam; The second LOTH thesis is that Fodor, Jerry these mental representations belong The Language of Thought Hypothesis to a representational or symbolic (LOTH) was first postulated by Jerry system that is itself language-like. This Fodor in 1975. According to this is the part of the hypothesis that is hypothesis, thought and thinking are most closely influenced by Noam done in a mental language, ‘men- Chomsky: roughly speaking, this sym- talese’. It is a bold hypothesis insofar bolic system (often called ‘mentalese’) as this language is held to be innate, corresponds to what Chomsky calls distinct from all natural languages, ‘competence’ or ‘deep structure’. and physically realised in the brain. Mentalese has its own grammar, and LOTH is derived from work on one which is, moreover, universal. ‘propositional attitudes’*, which are LOTH, despite being concerned with described by sentences of the form propositional attitudes rather than ‘Mary believes that pigs might fly’. propositional content, is thus situated The general form of such sentences is in the tradition of linguistic thought ‘SAsthat P’, where S is the subject stretching back to Port-Royal logic*, who holds the attitude, A is an ‘atti- which sees the purpose of language as tudinal’ verb such as ‘believe’, ‘desire’, being predicative, that is to make ‘hope’, ‘intend’ and P is any sentence. propositional statements about the It may be seen that ‘attitudinal’ verbs world which are demonstrably true or coincide with what elsewhere in the false. LOTH is incompatible with the philosophy of mind are called ‘inten- ‘meaning is use’ tradition of linguistic tional’ verbs: there is a sense, there- philosophy promulgated by Ludwig

110 LANGUAGE OF THOUGHT

Wittgenstein and developed by such capacity. Innateness is necessary to the philosophers of language as J. L. hypothesis in order to guard against Austin, H. P. Grice and John Searle. an infinite regress: if mentalese is Indeed, Fodor (1975: 55–97) explic- innate, then there is no need to postu- itly argues, contra Wittgenstein, that a late a language that a being would private language* is not impossible, need in order to learn a language in but necessary, and that mentalese is order to learn a language, and so on, just that private language. since mentalese is not learned. A third LOTH thesis is that the lan- As well as being a nativist thesis, guage of thought is distinct – that is, it LOTH is also a naturalist one. It is not identical to any spoken lan- attempts to answer the ‘hard question’ guage. This is again a strengthening of of consciousness, which in Fodor’s a Chomskian position: in LOTH, an own words is, ‘How could any- attempt is made not only at grammat- thing material have conscious states?’ ical (syntactic) transformations from (1991: 285). LOTH answers this deep structure to surface structure, question by collapsing mind into but also at semantic (representational) brain: contrary, for example, to psy- transformations from mentalese to choanalysis*, it has no need of psychic spoken language. processes. Fourth, LOTH is a nativist or According to LOTH, propositional innatist hypothesis: the language of thought cannot be accounted for thought is held to be universal in exclusively through mental images, but humans, and to be genetically deter- requires a syntax that combines sen- mined. LOTH is therefore also an tences. It follows that, finally, LOTH is empirical hypothesis: it is believed predicated on a theory of ‘semantic that the language of thought can be completeness’; in other words, any discovered as a material reality, as can predicate that has meaning in any the biological mechanism that gener- spoken language is also expressible in ates it. While this is a development of the language of thought. Chomsky’s theory, again it is more It is the second LOTH thesis – that radical: for Chomsky, language is mental representations belong to a uniquely human, whereas for LOTH, language-like representational or it is at least partly possessed by all sym bolic system, mentalese – which is species which have cognitive pro - the most original feature of the cesses. And while Chomsky holds that hypothesis, and also the most con- a being needs a language acquisition tentious. LOTH holds that each device in order to learn language, propositional attitude (belief, desire LOTH holds that a being needs a lan- and so on) is a function in the mathe- guage in order to learn a language – matical or computational sense. and that language is mentalese. The LOTH is often characterised as a language of thought is the hypothesis species of ‘functionalist materialism’. used to explain how it is that certain According to this view, that a being animals, and pre-linguistic infant chil- should have a propositional attitude is dren, can be seen to be thinking, even enabled (some would say, caused) by though they have no spoken language physical properties of that being,

111 LANGUAGE OF THOUGHT

especially neurological properties. despite the controversy it continues to These physical properties realise generate, is an important hypothesis mental representations through per- in cognitive psychology and neuro- forming computational operations, science. It is especially important as a expressed as functions, on them. conceptual tool in attempts to make Moreover, each function is specific the findings of cognitive science to a certain attitude: belief is ex- accord with folk psychology (people’s pressed through a belief-function, commonsensical beliefs about other desire through a desire-function and people’s propositional attitudes), as so on. Thus the general form of the seen, for example, in the work of language of thought, ‘S As that P’ may Daniel Dennett. be read as ‘S performs a computa- tional operation on P’. According to Primary sources this view, subjects do not directly Carruthers, Peter (1996). Language, believe that, for example, it is raining, Thought and Consciousness: An Essay but rather believe the proposition ‘it is in Philosophical Psychology. Cam- raining’ to be true. bridge: Cambridge University Press. This computational theory of lan- Dennett, Daniel C. (1987). ‘Reflections: guage and mind implies that a com- the language of thought reconsid- puter could be constructed with ered’. In The Intentional Stance. Cam- propositional attitudes, and that con- bridge, MA and London: MIT Press, versely, a certain kind of computer 227–35. could serve as a model of the human Fodor, Jerry A. (1975). The Language of mind-brain. This once again pits Thought. Hassocks: Harvester. LOTH against speech act theorists Fodor, Jerry A. (1987). Psychosemantics: such as Searle, who in his Chinese The Problem of Meaning in the Philos- Room thought experiment sought to ophy of Mind. Cambridge, MA and demonstrate that a digital computer London: MIT Press. could not have intentionality*. Fodor, Jerry A. (1990). A Theory of Con- Searle’s answer to the question of tent and Other Essays. Cambridge, MA what intentionality ‘is’ was that it is and London: MIT Press. not a thing, but a logical property. Fodor, Jerry A. (1991). ‘Replies’. In B. LOTH’s answer to the same ques- Loewer and G. Rey (eds), Meaning in tion, on the other hand, is that inten- Mind: Fodor and his Critics. Oxford: tionality is reducible to a specific Blackwell. 255–319. computational operation. In this view, ‘intentionality’ becomes a misnomer: Further reading propositional attitudes are proposi- Cain, M. J. (2002). Fodor: Language, tional functions of mental processes. Mind and Philosophy. Cambridge: Interactions between thoughts and Polity. concepts are merely computations, Loewer, Barry, and Georges Rey (eds) and ‘meaning’ is merely the internal (1991). Meaning in Mind: Fodor and his interaction of thought with concept. Critics. Oxford: Blackwell. As an empirical, naturalistic theory of how the mind works, LOTH, Karl Simms

112 LANGUE/PAROLE habits of any one member of a given speech community, Saussure sought to examine language in general and to Referring to two aspects of language identify the systems or rules and con- examined by Ferdinand de Saussure ventions according to which language at the beginning of the twentieth cen- functions. Saussure’s views on lan- tury, langue denotes a system of inter- guage influenced linguistics during the nalised, shared rules governing a twentieth century, and his imprint can national language’s vocabulary, be found in theoretical works dis- grammar, and sound system; parole cussing phonetics, phonology, mor- designates actual oral and written phology, syntax, pragmatics and communication by a member or mem- especially semantics. Indeed, the dis- bers of a particular speech commu- tinction between langue and parole nity. Saussure’s understanding of the forms an important part of the theo- nature of language and his belief that retical basis of structuralism*. scholarship should focus on investi- A popular lecturer at the University gating the abstract systematic princi- of Geneva, Saussure suggested ideas ples of language instead of researching and concepts that fascinated his stu- and language philosophy dents, yet he did not personally write led to a revolution in the field of an authoritative guide to his views. linguistics. Two colleagues of his, Charles Bally and Albert Sechehaye, collected and See also: Phoneme; Signs and edited student notes from three occa- Semiotics; Structuralism; sions during 1906–11 when he deliv- Transformational-Generative ered his lectures, publishing the Grammar assembled remarks under the title Key Thinkers: Bakhtin, Mikhail; Cours de Linguistique Générale in Barthes, Roland; Bloomfield, 1916. In the 1990s newly-edited ver- Leonard; Boas, Franz; Chomsky, sions of student notes based on Saus- Noam; Jakobson, Roman; Pike, sure’s lectures, along with translations Kenneth; Sapir, Edward; Saussure, into English, appeared. At the begin- Ferdinand de ning of the twenty-first century, there The discussion concerning langue and is still disagreement about a number parole was first suggested by Ferdi- of Saussure’s statements, and prob- nand de Saussure and popularised in lems surrounding the fragmented his Cours de Linguistique Générale nature of some of the student notes (Course in General Linguistics), a have not been fully resolved. series of Saussure’s university lectures Through Cours de Linguistique collected by his students and published Générale, Saussure’s views concerning posthumously in 1916. Abandoning language and the study of language the mindset, goals and objectives of were introduced to scholars through- , Saussure advo- out the world. Saussure rejected cated a synchronic examination of lan- the nineteenth-century notion that guage. Not interested in studying a linguistics should be primarily histor- particular language or the linguistic ical and comparative, and disagreed

113 LANGUE/PAROLE

vigorously with the idea that substan- speaker’s ingenuity or experimenta- tial effort should be made to identify, tion. Moreover, he felt that an indi- codify and promote the standard form vidual’s potential influence on of any national language; he felt it was language is minimal. An individual more worthwhile to focus attention might create a memorable turn of on describing language as it exists at a phrase, but that person is unable to given point in time, and believed that affect the overall structure or sound this activity could be conducted in an system of a given language. Finally, impartial manner. speakers can manipulate language in For Saussure, three aspects of lan- minor ways, but language imposes its guage could be potential objects of rules, order and possibilities on all consideration in linguistic study, and speakers without exception. he used the French words langage, As part of their intuitive knowledge langue and parole to designate these of langue, members of a speech com- aspects. Langage refers to the anatom- munity share possession and compre- ical ability and psychological need or hension of a body of signs (signes). urge of humans to create a system of According to Saussure, a sign consists linguistic signs for expressing ideas. of two components: a signifier (signifi- Langue represents a system of rules, ant) and a signified (signifié). Linguis- usages, meanings and structures that tic signs can encompass words, units of are products of the human ability to grammar, and expressions. The signi- create language and are shared by fier is a sound or series of sounds, and members of a specific speech commu- the signified is the meaning that the nity. Parole is often equated with sounds represent. Saussure was careful speech. It is the concrete realisation of to note that signs are actually linked to a collectively-internalised system and clusters of meanings or associations also reflects the personality, creativity and not to specific things. For example, and physiological capabilities of an the word ‘house’ does not refer to a individual speaker. specific object in the world but rather Overall Saussure paid little atten- to a concept involving images and tion to langage, considering it the sub- associations that speakers have in ject matter of other fields of inquiry, mind when they say or write the word. and he regarded parole as too idio- Furthermore, the connection between syncratic. Instead, he believed that lin- the series of sounds and the cluster of guistics should study langue in order images and emotions is arbitrary. The to gain a picture of the comprehen- words ‘girl’, ‘Mädchen’, and ‘niña’ sive, complex, ordered assemblage of might all refer to a female child, but sounds, words and syntactical units. there is no direct connection between Making use of a concept suggested in the sounds of each word and the mean- the writings of the French sociologist ing. Even so, speakers form a strong and philosopher Émile Durkheim, connection in their minds between Saussure viewed language as a social sounds and meaning. fact. According to Saussure, language Saussure stated that langage, is acquired through the socialisation the psychological and physiological process; it is not created through a faculty to produce meaningful lan-

114 LANGUE/PAROLE guage, does not manifest itself solely phonology. On occasion, agreement or in the creation of individual sounds, disagreement with Saussure’s beliefs words or units of meaning, and he can be traced back to an individual’s stressed that parole, individual com- political and philosophical leanings. munication within a speech commu- The Marxist linguist nity, does not take on the form of a disapproved of Saussure’s efforts to string of unrelated utterances. Lan- distinguish individual production of gage becomes a reality in langue – and language (parole) from collective ultimately in parole – through the knowledge and linguistic awareness rules governing the use and organisa- (langue), a division that, to Bakhtin’s tion of signs. These linguistic conven- way of thinking, isolates an individual tions are expressed in the form of from society; he was much more in syntagmatic and paradigmatic rules, favour of a theory of language that two types of systems that enable lan- portrays speech as dependent on, and a guage to convey messages by organis- product of, a specific social context. ing and sequencing the building Stimulated by Saussure’s discussion of blocks of sound and meaning. Syntag- the sign and its two components – the matic relationships refer to the limita- signified and the signifier – Roland tions governing sequences of sounds, Barthes investigated the contrast parts of words, and complete words between the message of our speech and offered by a given national language its form and articulation, and Kenneth to create meaning. Paradigmatic rela- Pike advanced his system of tag- tionships concern the existence of memics, a type of grammatical analysis words of similar meaning or gram- developed in the 1950s. Noam Chom- matical form that can substitute for sky, too, responded to Saussure’s ideas each other in a given context. when he transformed Saussure’s con- Saussure’s views concerning langue cepts of langage, langue and parole and parole, as well as his understand- into ‘language capacity’, ‘competence’ ing of the purpose and goals of linguis- and ‘performance’, and achieved a new tics, have exerted immense influence understanding of the Saussurean con- on linguists in Europe and North cepts. Twenty-first-century linguists America. Leonard Bloomfield, Franz remain attracted to Saussure’s concept Boas and Edward Sapir adopted Saus- of the dual nature of language and to sure’s method of objective, synchronic his theory of meaning. language study as the basis for their descriptive analyses of various North Primary sources American Indian languages. Bloom- Saussure, Ferdinand de (1996). Premier field also incorporated elements of Cours de Linguistique Générale (1907): Saussure’s innovative teachings into d’après les cahiers d’Albert Riedlinger. his writings, most notably Language French ed. Eisuke Komatsu. English (1933). Roman Jakobson and other ed. and trans. George Wolf. Oxford: members of the Prague School of Lin- Pergamon. guistics were inspired by Saussure as Saussure, Ferdinand de (1997). Deuxième they investigated sound systems and Cours de Linguistique Générale (1908– developed theories of phonetics and 1909): d’après les cahiers d’Albert

115 LINGUISTIC RELATIVITY

Riedlinger et Charles . French ed. limits of the native speaker’s language Eisuke Komatsu. English ed. and trans. are the limits of his/her world. In the- George Wolf. Oxford: Pergamon. ories of linguistic relativity, the vocab- Saussure, Ferdinand de (1993). Troisième ulary and linguistic structure of one’s Cours de Linguistique Générale (1910– native language limits or influences 1911): d’après les cahiers d’Émile Con- one’s Weltanschauung or world view. stantin. French ed. Eisuke Komatsu. See also: Feminism; Political English ed. and trans. Roy Harris. Correctness; Structuralism Oxford: Pergamon. Key Thinkers: Humboldt, Wilhelm Saussure, Ferdinand de (1966). Course in von; Peirce, C. S.; Sapir, Edward; General Linguistics. Cours de Linguis- Saussure, Ferdinand de; Whorf, tique Générale. Trans. Wade Baskin. Benjamin Lee New York: McGraw-Hill. First French edition 1916. The idea that the native language colours the speaker’s world view has Further reading been in the forefront of linguistic sci- Chomsky, Noam (1964). Current Issues in ence since the time of Wilhelm von Linguistic Theory. The Hague: Humboldt, and has found advocates Mouton. and critics from various disciplines. Harris, Roy (1987). Reading Saussure: A The American philosopher Charles Critical Commentary on the ‘Cours de Sanders Peirce postulated that the Linguistique Générale’. London: Duck- symbolic universe could only make worth. sense through language, which he Harris, Roy (2004). Saussure and His defined as semiotic, a system of signs. Interpreters. Edinburgh: Edinburgh Ferdinand de Saussure, in the Cours University Press. de Linguistique Générale, stated that: Koerner, E. F. K. (1973). Ferdinand de ‘No ideas are established in advance, Saussure: The Origin and Development and nothing is distinct, before the of His Linguistic Thought in Western introduction of linguistic structure’ Studies of Language. Amsterdam: Ben- (1916: 155). However, the notion of jamins. linguistic relativity has largely become Sanders, Carol (ed.) (2004). The Cam- associated with , bridge Companion to Saussure. Cam- who along with Edward Sapir, his lin- bridge: Cambridge University Press. guistic mentor at , used modern linguistic concepts to advo- David V. Witkosky cate the position that language limits, or at least influences, the way a speech community conceives of its world view and reality. LINGUISTIC Part of the groundwork for this hypothesis was laid by Whorf’s work RELATIVITY as a fire insurance investigator. During his career, he had the opportunity to The idea central to the Sapir-Whorf analyse many reports as to why fires hypothesis, which states that the broke out in factories. He found that

116 LINGUISTIC RELATIVITY workers would use extreme caution esis’. When viewed in terms of output, when around ‘full’ drums of gasoline. one could counter that a more appro- Just as one would expect, workers priate label would be the ‘Whorf- were careful not to smoke around Sapir hypothesis’. ‘full’ drums. Yet, these same workers A rather interesting development in when around ‘empty’ drums of gaso- this debate over giving credit where line would often toss lit cigarettes credit is due has been the attempt to nearby. This caused a violent explo- disassociate Sapir from the hypothesis sion because an ‘empty’ drum entirely. Desirous of preventing the (unknown to the smoker) still con- image of the great maestro Sapir from tained volatile gasoline vapour; an being tarnished by the taint of contro- ‘empty’ drum was really much more versy, some, most notably Alfred L. of a threat than a ‘full’ one. Using Kroeber, have claimed that Edward these data, Whorf concluded that the Sapir’s views were not really pro- meanings of certain words had an Whorfian. This viewpoint is not borne effect on a person’s behaviour. out by an examination of Sapir’s own It was the research of both Sapir writings. For example, as one can and Whorf into the grammatical sys- plainly see in the following passage, tems of many American Indian lan- there can be no doubt that Sapir’s guages, however, that proved to have position was fundamentally one that the greatest impact on this hypothesis. equated language with culture and By predicating their insights into the thinking. In Sapir’s words: interrelationships of language and culture on what they had learned from Language is a guide to ‘social real- the structures of these so-called ity’ . . . it powerfully conditions all our ‘exotic’ languages, the basic idea of thinking about social problems and language shaping the perceptions of processes. Human beings do not live in its speakers and providing for them a the objective world alone, nor alone in vehicle so that their experiences and the world of social activity as ordinar- emotions can be placed in significant ily understood, but are very much at categories, was given scientific under- the mercy of the particular language pinnings. Generally, Sapir is credited which has become the medium of with giving the problem of establish- expression for their society. It is quite ing the link between language and cul- an illusion to imagine that one adjusts ture its initial formulation, continuing to reality essentially without the use of in the tradition of Johann Gottfried language and specific problems of Herder and Humboldt. Whorf is hon- communication or reflection . . . No oured as the one who took this idea two languages are ever sufficiently sim- and developed it into a bona fide ilar to be considered as representing hypothesis. Hence, the resultant sup- the same social reality . . . We see and position is commonly given the desig- hear and otherwise experience very nation the ‘Whorfian hypothesis’. largely as we do because the language Pointing to Sapir’s pre-eminent stature habits of our community predispose as a linguist, some writers prefer the certain choices of interpretation. appellation the ‘Sapir-Whorf hypoth- (1929: 209)

117 LINGUISTIC RELATIVITY

There are really two different yet what I have called the ‘linguistic relativ- related versions of the Whorfian ity principle’, which means, in informal hypothesis, which is understandable terms, that users of markedly different when one considers that Whorf did all grammars are pointed by their gram- of his professional writing in the mars toward different types of observa- rather short period between 1925 and tions and different evaluations of 1941. His ideas, quite naturally, were externally similar acts of observation, continuously developing. The strong and hence are not equivalent as version of the hypothesis, which is observers but must arrive at somewhat called linguistic , holds different views of the world. (In Carroll that language determines thinking, or 1956: 221) as Stuart Chase writes in the foreword to Whorf’s collected works: ‘All Perhaps the most incontrovertible higher levels of thinking are depend- piece of evidence in favour of linguis- ent on language’ (in Carroll 1956: vi). tic relativity comes from the realm of This position is most difficult to and numerals. There are lan- defend primarily because translation guages, such as Hottentot, also known between one language and another is as Nama, which only have words for possible, and ‘thinking’ can take place the numerals for ‘one’ and ‘two’ and a without language at all, as evidenced word roughly translatable as ‘many’ by fine art. for three or more. A few languages Mirroring Sapir’s thoughts as men- have no numerals whatsoever and tioned above, Whorf notes in his 1940 their speakers are consequently unable article ‘Science and Linguistic’: to undertake even basic arithmetic. In Whorf’s undated manuscript ‘A We dissect nature along lines laid down Linguistic Consideration of Thinking by our native languages . . . We cut in Primitive Communities’, it is nature up, organize it into concepts, shown that the problem of ‘thinking’ and ascribe significances as we do, by so-called ‘primitive’ peoples is largely because we are parties to an ‘approachable through linguistics’ (in agreement to organize it in this way – an Carroll 1956: 65). Further, as linguists agreement that holds throughout our have come to fully appreciate only speech community and is codified in the fairly recently, Whorf maintained that patterns of our language. (In Carroll ‘linguistics is essentially the quest of 1956: 213) MEANING’ (in Carroll 1956: 73). Example after example is given of The milder version of the Whorfian things which are relatively easy to say hypothesis is labelled ‘linguistic rela- in Hopi but awkward or clumsy to say tivity’. This states that our native in such Standard-Average-European language influences our thoughts or (SAE) languages as English, Spanish perceptions. In fact, it was Whorf and German. The term SAE was of who coined the phrase ‘linguistic Whorf’s own invention. relativity’. In the article ‘Linguistics In the classic ‘An American Indian as an Exact Science’, Whorf com- Model of the Universe’, Whorf argues mented: that since there is neither an explicit

118 nor an implicit reference to time in the LINGUISTIC and thus no tenses for its verbs, according to the Hopi view VARIABLE of the world ‘time disappears and space is altered’ (in Carroll 1956: 58). A descriptive unit defined as a cate- Whorf’s basic contention is that Hopi gory of two or more linguistic alter- metaphysics, which underlies its cog- natives co-varying with one another in nition, is different from our own. In one of three ways: in a categorical way other words, the Hopi calibrate the (the variation always occurs given cer- world differently because their lan- tain circumstances); in a quasi- guage defines experience differently predictable or probabilistic way (in for them. line, for example, with another lin- As more information has surfaced guistic variable or a social variable); about Hopi, some of Whorf’s specific or in an apparently unpredictable, grammatical points have not with- random way (‘free variation’). The stood the test of time. Although most central idea is that there are ‘multiple linguists today dismiss many of ways of saying the same thing’, Whorf’s claims, it should be pointed though debate continues as to out that Whorf’s basic idea of linguis- whether two different linguistic struc- tic relativity – that the structure and tures are ever exactly equivalent in vocabulary of one’s mother tongue function and/or meaning, and about influences one’s world view – the extent to which individual lan- although not proven to be correct, guage users are able to exercise con- also has not been proven to be wrong. scious choice over which alternative to select from the range available. Primary sources See also: Acceptability/ Carroll, John B. (ed.) (1956). Language, Grammaticality; Conversation Thought and Reality: Selected Writings Analysis; Corpora; (Critical) of Benjamin Lee Whorf. Cambridge, Discourse Analysis; Descriptivism; MA: MIT Press. Emic/Etic; Mandelbaum, David G. (ed.) (1949). Empiricism/Rationalism; Intuition; Selected Writings of Edward Sapir in Speech Act Theory; Type/Token Language, Culture, and Personality. Key Thinkers: Cameron, Deborah; Berkeley: University of California Press. Halliday, M. A. K.; Pike, Kenneth; Sapir, Edward (1929). ‘The status of Labov, William; Milroy, Lesley; linguistics as a science’. Language 5: Sapir, Edward 207–14. Variables operate at all levels of lin- Further reading guistic structure. Grammatical vari- Saussure, Ferdinand de (1916/1966). ables capture some aspect of Course in General Linguistics (Cours de optionality in the occurrence of mor- Linguistique Générale). Trans. Wade phological or syntactic forms. For Baskin. New York: McGraw-Hill. instance, the use of double modal constructions like ‘might could’ (+ Alan S. Kaye main verb) is not obligatory in those

119 LOGIC

varieties of English which permit machinery of variationist sociolin- it. The study of lexical variables, guistics, enabling quantitative analy- traditionally the province of nine- sis of language data. teenth-and twentieth-century dialect geographers, involves examination of Primary sources alternative words or phrases for the Watt, Dominic (2007). ‘Variation and the same object or concept in a language’s variable’. In C. Llamas, P. Stockwell different dialects or sociolects. Dis- and L. Mullany (eds), The Routledge course variables, such as tag questions Companion to Sociolinguistics. and the focus marker ‘like’, are more London: Routledge. 3–11. difficult to handle given that it is not Wolfram, Walt (1991). ‘The linguistic always clear that utterances in which variable: fact and fantasy’. American they occur can necessarily be viewed Speech. Vol. 66:1, 22–32. as functionally equivalent to those in which they could occur but do not. Further reading Problems involving incomplete syn- Cornips, Leonie and Corrigan, Karen onymy of syntactic structures, dis- (eds) (2005). Syntax and Variation: Rec- course variables and lexical variables onciling the Biological and the Social. tend not to arise with phonological Amsterdam: John Benjamins. variables, as the equivalence of two Guy, Gregory (1997). ‘Competence, per- alternative is generally formance and the generative grammar unambiguous. For example the use of of variation’. In F. Hinskens, R. van [␻] rather than [w] in the word Hout and L. Wetzels (eds), Variation, ‘which’ indicates a difference in Change and Phonological Theory. Ams- two individuals’ geographical and/or terdam: John Benjamins. 125–43. social origins, rather than a difference in the word’s meaning. Dominic Watt Knowing that there are different ways of expressing the ‘same thing’ is almost certainly an intrinsic compo- nent of untrained native speakers’ LOGIC knowledge of their language(s). Awareness of the variable’s utility as a Logic is the study of argument. Logic means of formally accounting for lin- is central to philosophy, linguistics guistic alternations is apparent in the and many other fields. grammar of Pānini (seventh See also: Logical Form; Port-Royal to fourth centuries bc) which contains Logic; Truth Value variable rules allowing for differing Key Thinkers: Aristotle; Frege, outputs. The notion is implicit in the Gottlob; Russell, Bertrand historical linguistics and dialectology of more recent centuries, for example An argument is an attempt to per- in the work of William Jones on suade using reasoning. Arguments are genetic affiliations between Indo- composed of statements. As standard, European languages, and it forms a each statement has a truth value*: it is central element of the theoretical either true or false, but not both. Nor-

120 LOGIC mally, an argument has at least one being valid.) To appreciate that it is premise and exactly one main conclu- possible, rather than actual, truth sion. The main conclusion of an argu- values that matter to validity, consider ment is the statement it ultimately the following arguments embodying 2 aims to support. The premises are the and 4: statements that are meant to support the conclusion. An inference is a step Argument 2* in reasoning. Logic is primarily concerned with P1 All US presidents are communists. deductive arguments. These aim to be P2 George W. Bush is a US president. valid. An argument is valid if and only C George W. Bush is a communist. if the truth of the premises would guar- antee the truth of the conclusion. If it Argument 4* is consistent to assert an argument’s premises and deny its conclusion, the P1 All US presidents before 2007 were argument is invalid. The validity or male. invalidity of an argument does not P2 Eugene McCarthy was a US presi- depend on the actual truth values of its dent before 2007. parts: it depends on their possible C So, Eugene McCarthy was male. truth values. As Figure 3 shows, four permutations of truth-values are pos- In 2*, P1 is false and C is false. How- sible. ever, if we assume, for the purposes of The only permutation of truth testing for validity, that P1 and P2 are values that debars an argument from both true, then we cannot deny C being valid is 3. This is because valid without being committed to a contra- inferences are truth-preserving: if diction. So, the argument is valid: if its the inference from a given set of premises were all true, then its con- premises to a given conclusion is clusion would be true. In 4*, P2 is valid, then if the premises are all false. If the premises of 4* were all true, the conclusion must also be true, then the conclusion would true. (Embodying 1, 2 or 4 does not have to be true as well, on pain of con- make an argument valid, but nor tradiction. So again we have a valid does it disqualify the argument from argument.

13 All the premises are true. All the premises are true. The conclusion is true. The conclusion is false. 24 At least one premise is false. At least one premise is false. The conclusion is false. The conclusion is true.

Figure 3

121 LOGIC

To assess an argument for validity Argument F by informal means, first suppose that Every human is mortal. all its premises are true. Then ask Every philosopher is human. whether, under this supposition, the So, every philosopher is mortal. conclusion must also, on pain of , be considered true as By using term variables, Aristotle is well. If so, the argument is valid. If able to discuss the logical forms of var- not, it is invalid. Consider the follow- ious arguments in abstraction from the ing argument: general terms (for example, ‘human’, ‘mortal’, ‘philosopher’) those argu- Argument T ments employ in natural language. Formal logic goes beyond the informal P1 If the war in Iraq was a war for oil, method of testing for validity. The then it was an unjust war. formal logician codifies deductive rea- P2 The war in Iraq was not a war for soning to help distinguish valid argu- oil. ment forms from invalid ones. C The war in Iraq was not an unjust Aristotle is a formal logician in our war. sense: the term variables he uses enable him to discern various patterns Suppose, for the test, that P1 and P2 of inference valid by virtue of their of Argument T are both true. Would structures. As well as being a term this then guarantee that C is true? No. logic, Aristotle’s logic is syllogistic. A P1 does not assert that the only way in syllogism is an argument with two which the war could have been unjust premises and a conclusion, in which is by being a war for oil. Nothing in both premises and the conclusion are the content of the premises rules out general sentences. The sorts of general that the war was unjust by virtue of sentence Aristotle is concerned with some other factor, so the conclusion are what he calls ‘universal’ and ‘par- does not follow. ticular’ sentences (De Interpretatione, Aristotle founded the discipline of 17a–17b): sentences that use expres- logic. He defines a proposition as ‘a sions equivalent to ‘all’, ‘some’ and statement denying or affirming some- ‘no’. The following types of general thing of something’ (Prior Analytics, sentence can feature in a syllogism: 24a). Propositions* are of subject- predicate form, where a subject term Every A is B. (Universal affirmative) picks something out and a predicate No A is B. (Universal negative) term says something about it. Aris- Some A is B. (Particular affirmative) totle uses letters to serve as term vari- Some A is not B. (Particular negative). ables, that is, to stand proxy for any terms whatever. For example, ‘If A The Stoics were also ancient pioneers predicated of every B, B every C, A be of logic. Chrysippus (c. 280–c. 207 predicated of every C’ (Prior Analyt- BC) was the most important Stoic logi- ics, 26a) depicts a valid argument cian. He regarded as valid such forms form. Here is an instance of that as ‘If the first, then the second; but the form: first; therefore the second’ and ‘If the

122 LOGIC

first, then the second; but not the ‘logical’ and the ‘non-logical’ vocabu- second; therefore not the first’ (see lary of the lexicon. Let us consider a Kneale and Kneale 1962: 162–3). language of propositional logic, called ‘The first’ and ‘the second’ here are ‘PL’ (after Tomassi 1999). The lexicon place-markers for complete ‘assert- of PL consists of: ibles’. Assertibles are the meanings of declarative sentences (see Bobzien i. The sentence letters: ‘P’, ‘Q’, ‘R’, 2003). Stoic logic is a type of proposi- and so on, which symbolise atomic tional (or ‘sentential’) logic. The sentences – that is sentences with no Stoics put what are now called the other sentences as parts. ‘logical operators’ of propositional ii. The propositional operators: ‘~’, logic at the centre of their investiga- ‘&’, ‘v’, ‘→’, ‘↔’, at least roughly tions. Their English equivalents are equivalent to ‘it is not the case ‘If . . . then . . .’, ‘ . . . or . . .’, ‘ . . . and that . . .’, ‘. . . and . . .’, ‘. . . or . . .’, . . . ‘ and ‘It is not the case that. . .’. ‘If . . . then . . .’, and ‘. . . if and These operators are used to form only if . . .’. complex statements: that is, state- iii. The parentheses: ‘(’and ‘)’, used as ments with other statements as parts. a type of punctuation. In symbolic logic, artificial, ‘formal’ languages are constructed and em - A PL formula is any string of symbols ployed, in contrast with the mixture of from the lexicon of PL. A well-formed artificial symbols and natural lan- formula (WFF) of PL is a formula con- guage used by earlier formal logicians. structed in observance of the follow- Symbolic logic in this sense developed ing formation rules: with the work of Gottlob Frege and of Bertrand Russell and A. N. White- 1. Every sentence letter is a WFF. head, following earlier nineteenth- 2. Prefixing a WFF with ‘~’ gives a century innovations. Frege’s work is WFF. of unparalleled influence and impor- 3. For any WFFs containing A B, put- tance. ting them either side of ‘&’ or ‘v’ or Among Frege’s most valuable con- ‘→’ or ‘↔’ and putting parentheses tributions to logic were the inven- round the resulting formula gives a tions of modern predicate logic (also WFF. known as ‘quantificational logic’ and 4. Nothing else is a WFF. ‘predicate calculus’) and the first formal system, both originally The propositional operators are the log- included in his Begriffsschrift (1879). ical vocabulary of PL. In propositional While Frege’s notation is obsolete, logic, the quantifiers ‘some’, ‘all’ and contemporary logic is built upon his ‘no’ (and equivalent expressions) are predicate logic and his systematisa- not treated as logical vocabulary. Thus, tion of the notion of proof. not every valid argument is proposi- A consists of a lex- tionally valid. For example, the validity icon (a set of symbols) and a syntax of a valid syllogism, like Argument F (a set of rules for using the symbols). above, depends on expressions other Logicians distinguish between the than the propositional operators.

123 LOGIC

After Aristotle and Frege, predicate Argument A logic treats these quantifiers as logical Socrates is a philosopher. vocabulary. A language of predicate Therefore, someone is a philosopher. logic both retains and supplements the logical and non-logical vocabulary of To translate ‘Socrates is a philosopher’ a language of propositional logic. Let into propositional logic, we assign a us specify such a language of predicate sentence letter, such as ‘P’ to the logic called ‘QL’ (for ‘Quantificational sentence. Since ‘Someone is a philoso- Logic’) broadly after Tomassi (1999). pher’ is a different sentence, we assign The lexicon of QL consists of the lex- it a different sentence letter, such as icon of PL (that is, i–iii above) plus: ‘Q’. With these assignments, Argu- ment A comes out as: iv. Individual constants: ‘a’, ‘b’, ‘c’, and so on. These are lower-case letters Argument A* from the beginning of the alphabet, used to symbolise proper names Premise P (such as ‘Socrates’, ‘London’). Conclusion Q v. Predicate letters: ‘F’, ‘R’, and so on, used to symbolise predicative The problem with this is that no logi- expressions (such as ‘. . . is a man’, cal form is revealed, since there are no ‘. . . loves . . .’). logical operators in play. So, A* vi. The quantifiers ‘᭚’ (the existential cannot be an instance of a valid logical quantifier, for ‘some’) and ‘᭙’ (the form of PL and cannot capture the universal quantifier, for ‘all’). intuitive validity we assign to A. Pred- vii. Individual variables: ‘x’, ‘y’, ‘z’, icate logic solves this problem. Logical and so on. form within atomic sentences is viii. The identity sign: ‘=’. revealed. The way A is dealt with in ix. The : ‘[’, ‘]’, which are QL is as follows. another form of punctuation. First, we specify a domain of quan- tification, D. This is the set of entities Though formation rules are specifi- we are quantifying over. Then we able for QL, let us instead note two specify the meanings of the non-logi- features of QL. First, its logical vocab- cal vocabulary we are going to use in ulary includes the logical operators of translating from English to QL. So, we PL plus the quantifiers, the individual can begin to translate Argument A by variables and the identity sign. specifying the following. Second, the non-logical vocabulary of QL, specifically the individual con- D{human beings} stants and the predicate letters, allows a Socrates us to formalise, at a sub-sentential F . . . is a philosopher level, sentences that do not contain any propositional operators. Both We can now translate the premise of points can be illustrated using a valid Argument A as: argument that is neither proposition- ally nor syllogistically valid: Fa

124 LOGIC

We read this as ‘a is F’. By convention, tional logic is first-order when it we put a predicate letter before a quantifies over individuals but not name to which it applies. over properties. ‘Higher-order’ sys- The existential quantifier ‘᭚’ and tems quantify over properties. Such the universal quantifier ‘᭙’ are used, systems include predicate variables with the individual variables, to among their vocabulary. The bearing express ‘some’ and ‘all’. Thus, we can of the first-order/higher-order distinc- translate the conclusion of Argument tion cannot be explored here. A as follows: The term ‘classical’ in ‘’ does not refer to the ancient ᭚x[Fx] world, but to some basic features shared by all classical systems. These We read this as ‘There is an x that x is include, among others, adoption of F’, or ‘For some x, x is F’. (‘x’ is an the principle of bivalence (see the individual variable.) So, the transla- entry on truth value*) and an account tion of Argument A is: of according to which a given conclusion follows Argument A** from a given set of premises if and Premise Fa only if the negation of the conclusion Conclusion ᭚x[Fx] is inconsistent with the premises. Non-classical logic is obtainable by The validity of A** (and thus of A) is extending or revising classical logic. formally demonstrable in predicate What is it to extend classical logic? logic. (For example, the argument can A system of logic S* is an extension of be proven by using an inference rule a system of logic S if S* supplements called ‘Existential Introduction’.) the language and derivational appara- Modern predicate logic has greatly tus of S. By supplementing the lan- superior expressive power to its pre - guage of S, we mean keeping all the decessors: it captures all the logical form symbols and formation rules of S and they could capture and logical form that adding further symbols and formation eluded them. This is a legacy from Frege rules. The inference rules of S* will (see Noonan 2001, ch. 2). also supplement those of S. Thus, Frege also presented the first formal every S-argument will be S*-valid, but systems of propositional and predi- not every S*-valid argument will be cate logic. Minimally, a formal system S-valid. But why extend classical consists of a formal language plus a logic? The usual motive is that classi- method of proof. The latter is a pro- cal logic does not capture enough cedure for demonstrating the validity validity. For example, classical logic of valid arguments in a formal lan- provides us with no formal means of guage. In working with this idea of a counting as valid such inferences as: formal system, contemporary logi- Argument H cians follow Frege. The sort of logic normally studied It is necessary that Socrates is human. on an introductory course is first-order Therefore, it is not impossible that classical logic. A system of quantifica- Socrates is human.

125 LOGIC

Supplementing the vocabulary and Ernest is brave and Ernest is not brave. rules of inference of a classical system So Ernest is a mountaineer. with logical operators standing for ‘It is necessary that’ and/or ‘It is pos- The premise here is a contradiction, so sible that’, enables the development it cannot be true. So, we cannot be in of systems of . A formal a situation in which the premise is true system S is said to ‘under-generate’ and the conclusion false. Thus we have relative to natural language if there a classically valid argument. Logicians are valid arguments of natural lan- in the field hold that arguments like guage that are not S-valid. Modal Read’s example show that the classical logicians hold that, in failing for- account of validity over-generates. mally to account for the validity of Linguists and philosophers of lan- such arguments as H, classical logic guage investigate the relationship under-generates. between logic and natural language, What is it to revise classical logic asking, for example, about the rela- and why do so? A system of logic S* tionship between logical and gram- is a revision of a classical system of matical form and the extent to which logic S if and only if S* discards one semantic theory for natural language or more of the inference rules of S. A can be modelled on the formal seman- formal system is said to ‘over- tics for logic. Logic is also interesting generate’ if there are arguments that for linguists because the syntax of a ought to be considered invalid but formal language is finitely specifiable which come out as valid within the by reference to a set of recursive for- system. Some logicians object to the mation rules (for example, the PL classical account of logical conse- formation rules set out above). An quence on the basis that it results in analogous set of rules for a natural over-generation. This objection some- language would be explanatory with times stems, as in the case of intu- respect to a speaker’s capacity to form itionist logic, from rejection of a sentences never previously encoun- principle of classical semantics (see tered. the entry on truth value). If there are Analytic philosophy* of language, classically valid arguments which inaugurated by Frege and Russell, has ought not to be considered valid at all, logic at its heart, and some of the then at least one of the inference rules foundational works of modern of classical logic must be rejected or symbolic logic were its impetus. In lin- restricted. guistics, formal semanticists and con- Revisions to classical logic include temporary syntactic theorists use logic intuitionist, relevant, fuzzy and in their analyses of natural language. dialethic . The relevant logician, A solid grounding in logic goes a long for example, notes a seemingly odd way for the student of linguistics or feature of classical validity: that from the philosophy of language. contradictory premises any conclu- sion whatever can validly be inferred. Primary sources Read (1995: 55) gives the following Aristotle. De Interpretatione. Trans. J. L. example: Ackrill. In Jonathan Barnes (ed.)

126 LOGICAL FORM

(1984), The Complete Works of Aris- why it logically entails other sentences totle, vol. 1. Princeton: Princeton Uni- that it entails and why it is entailed by versity Press. 25–38. other sentences that logically entail it. Aristotle. Prior Analytics. Trans. A. J. Questions about logical form are Jenkinson. In Jonathan Barnes (ed.) important to the characterisation of (1984), The Complete Works of Aris- logic and to its relationship with nat- totle, vol. 1. Princeton: Princeton Uni- ural language. versity Press. 39–113. See also: Definite Descriptions; Frege, Gottlob. Begriffsschrift (1879/ Logic; Logical Positivism 1967). Trans. Stefan Bauer-Mengelberg. Key Thinkers: Aristotle; Chomsky, In Jean van Heijenoort (ed.) (1967), Noam; Russell, Bertrand From Frege to Gödel: A Sourcebook in Modern , 1879– Bertrand Russell aimed to put a con- 1931. Cambridge, MA: Harvard Uni- ception of logical form at the centre of versity Press. 5–82. the practice of philosophy. Logical form in this sense is distinct from the Further reading notion called ‘LF’ by linguists after Allwood, Jens, Lars-Gunnar Andersson Noam Chomsky. The investigation of and Östen Dahl (1977). Logic in Lin- logical form begins with the inquiry guistics. Cambridge: Cambridge Uni- into initiated by versity Press. ancient logicians. For example, Aris- Bobzien, Susanne (2003). ‘Stoic logic’. In totle noted that any argument that is Brad Inwood (ed.), Cambridge Com- an instance of the pattern below is panion to the Stoics. New York: Cam- deductively valid: bridge University Press. 85–123. Noonan, Harold W. (2001). Frege: A Crit- Some As are Bs. ical Introduction. Cambridge: Polity. All Bs are Cs. Kneale, William and Martha Kneale So, some As are Cs. (1962). The Development of Logic. Oxford: Clarendon. Statements themselves have logical Read, Stephen (1995). Thinking about forms: ‘Some philosophers are logi- Logic. Oxford: Oxford University cians’ is an instance of the pattern Press. ‘Some As are Bs’. Tomassi, Paul (1999). Logic. London: With the development of modern Routledge. symbolic logic*, a view arose that grammatical form can be misleading Stephen McLeod as to logical form. For example, on Russell’s account, sentences that use definite descriptions* have logical fea- LOGICAL FORM tures not evident at the level of surface syntax. In ‘The present King of France is bald’, none of the words is a logical On one central conception, the logical operator. Russell claims, however, form of a sentence consists in those of that the sentence is to be analysed as its structural features which clarify employing covert logical operators.

127 LOGICAL POSITIVISM

On one view, logical form is con- lenges’. Philosophy Compass 1/3, strued as relative to a formal language. 303–316, 10.1111/j.1747–9991.2006. A sentence in a natural language then 00017.x. has a certain logical form in virtue of its being translatable into an instance Stephen McLeod of a given sort of formula in the formal language. Alternatively, logical form can be conceived as already present in natural language itself, independently LOGICAL of a given formal language’s attempts to represent it. On this second concep- POSITIVISM tion, a sentence’s logical form is not a matter of it having a certain relation to A movement in twentieth-century phi- some construction in a formal lan- losophy whose adherents argued that guage. Rather, the aim of the formal philosophically legitimate discussion language is to capture intrinsic struc- must be limited to statements that tural features of sentences of natural could be assigned a determinate truth language. A related question concerns value*. Logical positivism triggered whether logical form is distinct from, some important debates about the or instead a level of, grammatical form nature of meaning and the appropri- (see Bach 2002). ate methods and for the A central criticism concerns how analysis of language. any division of the vocabulary of a See also: Analytic Philosophy; natural language into logical and non- Analytic/Synthetic; logical terms is to be justified (see Deduction/Induction; Jackson 2006). That is, which words Empiricism/Rationalism; Logic; in a natural language are logical oper- Truth Value ators and which are not? Scepticism Key Thinkers: Ayer, A. J.; Carnap, about logical form may start from the Rudolf; Russell, Bertrand claim that there is no entirely non- arbitrary answer to this question. Logical positivism was practised in various forms in Europe and the USA Primary sources in the early and mid-twentieth cen- Russell, Bertrand (1914). ‘Logic as the tury, but it is most closely associated essence of philosophy’. In Our Knowl- with the work of the Vienna Circle. edge of the External World. Chicago: This was a group of philosophers led Open Court. 33–59. by Moritz Schlick and including Rudolf Carnap, and Further reading Friedrich Waismann who held meet- Bach, Kent (2002). ‘Language, logic, and ings at the University of Vienna during form’. In Dale Jacquette (ed.), A Com- the 1920s and 1930s. They were com- panion to . Oxford: mitted to establishing a rigorously sci- Blackwell. 51–72. entific mode of philosophical inquiry Jackson, Brendan (2006). ‘Logical form: where positive, or empirically justi- classical conception and recent chal - fied, knowledge was expressed in

128 LOGICAL POSITIVISM

logically coherent language. Members physics, of and, most of the Vienna Circle published a controversially of all, of ethics and number of monographs and journal religion. articles in German, but their ideas Logical positivism was beset with were disseminated more widely, and various problems, for instance to do in particular became available to an with the reliability of empirical evi- English- speaking audience, when A. J. dence and the method of induction* Ayer published Language, Truth and on which it depended, and was grad- Logic in 1936, after a attending a ually modified or abandoned. How- number of meetings of the Circle on ever, its influence has continued to be an extended visit to Vienna. felt in the philosophy of language and Logical positivism was not prima- subsequently in linguistics, largely rily concerned with the analysis of because of the alternative ideas about natural language; in fact Carnap and meaning that its critics suggested. For others were openly dismissive of instance, W. V. O. Quine advocated everyday usage. They saw it as impre- holism* in an account of meaning, cise and illogical; its statements were rather than expecting each statement in need of translation into a logically in a language to be justified by an acceptable form before they could be independent process of verification. the subject of serious discussion. J. L. Austin argued that ‘true’ and Rather, the impact of logical posi- ‘false’ were not the only philosophi- tivism on language study was due cally interesting labels that could be largely to its criterion of meaningful- applied to uses of language, a posi- ness. To be counted as meaningful and tion that led to his development of therefore admitted into philosophical speech act theory*. This in turn has discussion, a statement must be capa- had a significant impact on how ble of being classified as either true or meaning is discussed in present-day false. Analytic statements, a class linguistics. taken by the logical positivists to include the statements of mathematics Primary sources and logic*, were meaningful because Ayer, A. J. (1946). Language Truth and they were true by virtue of their own Logic. Second edition. Harmonds- internal properties. Synthetic state- worth: Pelican (1971). First edition ments could be admitted as meaning- London 1936. ful if they were capable of being Ayer, A. J. (ed.) (1959). Logical Posi- subjected to an identifiable process tivism. Glencoe, IL: The Free Press. of verification: that is, if it was possi- Carnap, Rudolf (1932). ‘The elimination ble to establish what sort of empirical of metaphysics through logical analysis evidence a statement could be evalu- of language’. In A. J. Ayer (1959), Log- ated against in order to establish ical Positivism. Glencoe, IL: The Free whether it was true or false. All other Press. 60–81. statements – synthetic statements that Schlick, Moritz (1930). ‘The turning point were not amenable to verification – in philosophy’. In A. J. Ayer (ed.) (1959) were simply meaningless. This class Logical Positiviism. Glencoe, IL: The included the statements of meta- Free Press. 53–9.

129 MENTALISM

Further reading investigations on objectively observ- Gower, Barry (ed.) (1987). Logical Posi- able data. Behaviourists see language tivism in Perspective. London: Croom acquisition as a conditioning process, Helm. while mentalists propagate an inborn Stadler, Friedrich (ed.) (1993). Scientific device according to which language is Philosophy: Origins and Developments. acquired systematically. Dordrecht: Kluwer Academic Publishers. The first mentalists were psycholo- gists such as Edward Thorndike and Siobhan Chapman Edward Bradford Titchener, whose aim was to study the mind scientifi- cally using methods of association and introspection. Their objects of inquiry MENTALISM were ‘mental facts’, and they were convinced that by replacing specula- A notion that can be applied to any tion with experimentation, the mind linguistic theory that deals with the could be analysed into the compo- relationship between language and nents from which complex thoughts the mind. It explores the relationship are constructed. between language, thought and real- Another kind of mentalist linguistic ity, and describes the internal lan- theory is rooted in the study of Native guage devices which explain the American languages. Edward Sapir creativity in language acquisition and (1921) and Benjamin Lee Whorf the processes involved in thinking, (1956) claimed that every language speaking and understanding. has its own descriptive categories through which the world is seen. Lan- See also: Artificial Intelligence; guage shapes thought and is not just Behaviourism; Cognitivism; an expression of thought; thus it is evi- Innateness; Language of Thought; dence of how people think. Linguistic Relativity Mentalism was revived and devel- Key Thinkers: Bloomfield, oped by Noam Chomsky (1965, Leonard; Chomsky, Noam; 1968), based on René Descartes’ Descartes, René; Sapir, Edward; rationalism, as a reaction to Leonard Skinner, B. F.; Whorf, Benjamin Bloomfield’s structuralistic approach Lee and B. F. Skinner’s radical behav- Mentalism dates back to the emer- iourism. In contrast to these theories, gence of scientific psychology in the which concentrated on observable early twentieth century. It has been in surface structures, Chomsky postu- constant rivalry with behaviourism* lated an underlying deep structure as ever since John B. Watson’s (1919) well as an innate language acquisition reaction to the mentalist methods of device that enables human beings to introspection and descriptions of feel- develop their linguistic competence. ings and thoughts. Whereas for men- Cognitive linguistics investigates talists the main concern is the question the mental processes involved in the of whether the mind is accessible for acquisition and use of language and of introspection, behaviourists base their knowledge in general. It is closely

130 METAPHOR linked to cognitive psychology and to Lakoff and Mark Johnson (1980) artificial intelligence*, which tries to affirm that metaphors are deeply make machines carry out tasks requir- ingrained in our thoughts, actions and ing intelligence and in doing so exam- everyday language. ines how humans perform such tasks. See also: Ideational Theory; The theory that mental states and Linguistic Relativity; Possible processes are independent of but can World Semantics; Signs and explain behaviour has had its impact Semiotics on linguistics and related fields Key Thinkers: Derrida, Jacques; throughout the twentieth century. Saussure, Ferdinand de; Whorf, Although criticised heavily by the Benjamin; Wittgenstein, Ludwig behaviourists, it remains a concept to be considered in any approach to lan- Lakoff and Johnson (1980) state that, guage and its applications. contrary to common belief, a meta- phor is not a rhetorical device Primary sources employed to embellish literary lan- Chomsky, Noam (1965). Aspects of the guage. They argue that metaphors Theory of Syntax. Cambridge, MA: MIT constitute the foundation of our con- Press. ceptual system and influence our Whorf, Benjamin L. (1956). Language, thoughts, actions and communica- Thought, Reality: Selected Papers. Ed. tion. In addition to their description John B. Carroll. New York: Wiley. and classification of metaphors, Lakoff and Johnson discuss their sig- Further reading nificance in both philosophy and lin- Chomsky, Noam (1968). Language and guistics. Zoltán Kövecses (2002) Mind. New York: Harcourt, Brace and proposes a detailed framework of cog- World. nitive metaphors and their role in lin- Sapir, Edward (1921). Language: An guistics, literary analysis, ethics and Introduction to the Study of Speech. politics. In his (2005) publication, New York: Harcourt, Brace and World. Kövecses focuses on the diversifica- Watson, John B. (1919). Psychology from tion of metaphors as a result of cul- the Standpoint of a Behaviorist. Philadel- tural differences. He explores the phia: Lippincott. correlations between metaphoric and cultural variations. Murray Knowles Jürg Strässler and Rosamund Moon (2006) give a comprehensive account of how meta- phors pervade a diverse number of disciplines, including semantics, liter- METAPHOR ature, religion, cinema and music. Criticising philosophers and lin- A figure of speech in which a term that guists for their failure to appreciate is usually associated with a certain the significant role metaphors play in entity is used to describe another, as in our perception and communication, ‘the dawn of history’. In their seminal Lakoff and Johnson (1980) point out work Metaphors We Live By, George numerous everyday expressions that

131 METAPHOR are metaphoric in nature. These nonphysical entities as physical ones. expressions include such conceptual Examples include the quantification metaphors like argument as war of abstract entities, like speaking (Lakoff and Johnson use small capi- about someone as having ‘a lot of tals to denote conceptual metaphors patience’, or describing the world as and differentiate them from meta- being ‘filled with intolerance’. The phoric expressions). To support their most prominent type of ontological hypothesis they provide a multitude of metaphors, according to Lakoff and expressions associated with this and Johnson, is personification, which is other metaphorical concepts. These giving a nonhuman entity a human include our speaking about ‘argu- quality or attribute. This is evident in ment’ as if it were ‘war’ when we use expressions like talking about infla- such expressions as ‘attacking or tion ‘eating up’ someone’s savings, or defending a position’, ‘winning or a disease ‘catching up’ with some- losing an argument’, or talking about body. claims as being ‘indefensible’. Another Another figure of speech that is is time as metaphoric in nature is , money, which generates a number of which is employing an entity or fea- expressions that include ‘saving’, ture to make reference to another. ‘wasting’ or ‘investing’ time. Examples of this include making ref- In addition to this type, which they erence to restaurant clients by the regard as complex and call ‘structural meals they order, for example ‘The metaphors’ because ‘one concept is chicken lasagna left a big tip’, or ‘The metaphorically structured in terms of beef burrito spilled his drink’. A major another’ (1980: 14), Lakoff and John- category of metonyms, according to son identify a number of simple types Lakoff and Johnson’s framework, of metaphors, including orientational involves using the part as a whole. and ontological metaphors. The This phenomenon, which has tradi- majority of the belonging tionally been called synecdoche, is to the former type are related to spa- exemplified by expressions such as tial orientation. They speculate that ‘wheels’ to make reference to ‘cars’, or these metaphors may have resulted ‘fresh blood’ as a reference to ‘new from our cultural and physical experi- people’ in a work-place setting. ence. The most commonly used of Lakoff and Johnson refute the claim these is the up-down metaphor. In that metonyms are purely referential Western culture ‘up’ is associated with in nature when compared with positive concepts like ‘happiness’ and metaphors. They argue that the ‘part’ ‘health’, whereas ‘down’ is related to used to make the reference would usu- negative ones such as ‘sadness’ and ally have a vital role in determining ‘illness’. This gives rise to such the significance of the utterance. metaphoric expressions as ‘high spir- In Philosophy in the Flesh, Lakoff its’, ‘feeling up’, ‘falling ill’ and ‘being and Johnson argue that ‘the banish- depressed’. Ontological metaphors, ment of metaphor from the realm of on the other hand, include diverse truth explains why metaphor has tra- expressions most of which refer to ditionally been left to and

132 METAPHOR

literary analysis’ (1999: 120). Chal- nutrient, has been that the target lenging this – among other – Western domain ‘food’, a physical entity, is philosophical principles, they assert understood in terms of the source metaphors form the basis of concep- domain ‘love’, a nonphysical entity. tualisation. Without them, therefore, Such a simplistic view of the relation- the discussion of science, morality or ship between the source and target philosophy would not be possible. domains, Kövecses argues, is not Even a universal concept like ‘time’ is sufficient to explain the various metaphorically anchored because it metaphorical expressions based on is perceived of, and spoken about, the complex relationship between the metaphorically, not temporally. Citing two domains. He argues that the intri- examples from English, they discuss a cate mappings between ‘love’ and number of expressions related to the ‘nutrient’, and our ability to highlight various aspects of ‘time’. Some of different aspects of them enable us to these are metaphoric expressions that use metaphors like ‘love’ as ‘food’, indicate the passage of ‘time’ as ‘the desire for love’ as ‘hunger’, and ‘approaching’, ‘arriving’, ‘running’ ‘consequences of love’ as ‘effects of and ‘flying’. This metaphorisation of nourishment’. These complex map- time, they believe, is culturally specific pings, Kövecses argues, enable speak- as reflected in different languages. In ers to highlight specific source-target English, for instance, we ‘look for- relationships that would make it pos- ward’ to future events and regard past sible for them to express subtle mean- ones as being ‘behind us’ whereas in ings (2002: 79–92). Aymara, a language spoken in the Since its inception as an independ- Chilean Andes, the future is ‘behind’. ent, vital discipline, metaphor theory In this culture, the metaphorisation of has seen a number of developments. future events as being behind indicates One of these is the universality of the unforeseeable nature of such metaphors and their variation in dif- events (1999: 141). ferent . Although Lakoff and Lakoff and Johnson’s pioneering Johnson (1999) alluded to this aspect research on metaphor sparked interest of metaphors, Kövecses (2005) is in the field and paved the way for a credited with its development. Kövec- multitude of subsequent publications. ses predicates the potential universal- Reiterating the basic principles of ity of many conceptual metaphors on their framework, Kövecses (2002) the similarity of human physiological argues that the formula proposed to and conceptual experiences. Anger in explain conceptual metaphors, which humans, for instance, results in many states that a target domain is under- physiological changes, including an stood in terms of a source domain, is increase in body temperature and insufficient. Using expressions such as blood pressure. Kövecses believes this ‘someone is starved for love’ or accounts for diverse cultures utilis- ‘hungry for affection’, he argues that ing figures of speech based on the the traditional explanation of such fig- pressurised container conceptual ures of speech, which are based on metaphor. English metaphoric expres- the conceptual metaphor love as a sions like ‘boiling blood’, ‘simmering

133 MINIMALISM down’ and ‘blowing one’s top’, for Cambridge: Cambridge University example, have parallels in languages Press. as diverse as Chinese, Japanese, Lakoff, George and Mark Johnson Hungarian, Wolof, Zulu and Polish (1980). Metaphors We Live By. (2005: 39). However, Kövecses adds, Chicago: University of Chicago Press. there are a number of variations in Lakoff, George and Mark Johnson metaphoric expressions within the (1999). Philosophy in the Flesh: The same culture and between different Embodied Mind and its Challenge to ones. He attributes this variation to Western Thought. New York: Basic regional, social and experiential dif- Books. ferences. Other developments in metaphor Further reading theory include work on metaphor and Aitchison, J. (2002). Words in the Mind: the brain, acquisition of metaphors by An Introduction to the . children, and nonverbal metaphors. Oxford: Blackwell. Despite the inconclusiveness of the Cameron, L. and G. Low (eds) (1999). evidence, research suggests that Researching and Applying Metaphor. metaphors are processed by the right Cambridge: Cambridge University hemisphere of the brain. This hypoth- Press. esis is consistent with lateralisation Kövecses, Z. (2000). Metaphor and Emo- studies that characterise the left hemi- tion: Language, Culture and Body in sphere as being specialised in holistic Human Feeling. Cambridge: Cam- types of processing that are creative bridge University Press. and nonliteral. Knowles and Moon Lakoff, George and M. Turner (1989). cite research showing that patients More Than Cool Reason: A Field Guide who had right-hemisphere aphasia to Poetic Metaphor. Chicago: Univer- found it difficult to process metaphors sity of Chicago Press. (2006: 62). Since figurative compe- Whorf, B. L. (1956). Language, Thought tence is acquired relatively late, it has and Reality. Cambridge, MA: MIT been determined that children acquire Press. metaphoric processing between the ages of ten and twelve. Finally, non- Yousif Elhindi verbal metaphors in cinema, music and pictorial representation have also been areas of interest. MINIMALISM Primary sources Knowles, Murray and Rosamund Moon A cover term for ideas related to the (2006). Introducing Metaphor. New , an approach to York: Routledge. the study of the human language fac- Kövecses, Zoltán (2002). Metaphor: A ulty chiefly associated with Noam Practical Introduction. Oxford: Oxford Chomsky. It is driven by a radical con- University Press. ceptual and technical parsimony and Kövecses, Zoltán (2005). Metaphor in explores the possibility that the design Culture: Universality and Variation. of the human computational system

134 MINIMALISM for language optimally satisfies con- approaching the same order of magni- straints imposed only by the need to tude of complexity as the phenomena service a sensori-motor interface (PF) themselves. Similar concerns moti- and a conceptual-intentional interface vated the transition from the Revised (LF). Extended Standard Theory to Govern- ment-Binding Theory in the late See also: Adequacy; Logical Form; 1970s. Additionally, various principles Transformational-Generative began to emerge which had a ‘least Grammar; Universal Grammar effort’ flavour, or were computation- Key Thinkers: Chomsky, Noam ally more simple: namely local metrics The Minimalist Program, the succes- for determining the domain of certain sor framework to Noam Chomsky’s syntactic effects. As a result of consid- Government-Binding Theory, was ini- erations such as these, the strong min- tially articulated in the book of the imalist thesis began to emerge in the same name (Chomsky 1995) which mid-1990s. brought together several earlier Not surprisingly, this thesis has far papers with some new material. It was reaching implications for every aspect expanded upon principally in Chom- of the computational system, and sky (2000, 2001, 2004, 2005). The much of the organisation and technol- leading idea, often called by Chomsky ogy of Government-Binding Theory the ‘strong minimalist thesis’, is that has undergone revision. If PF and LF language is a ‘perfect solution’ to the are the only linguistic levels with any problems imposed by virtual concep- significance (being the only two seem- tual necessity. In other words, there ingly mandated by virtual conceptual are certain inescapable constraints on necessity, though see below), then the computational system of the other linguistic levels, such as D-struc- human language faculty. If it is to ture and S-structure, must be elimi- relate sound and meaning, the com- nated. Linguistic principles stated in putational system must interface with terms of these levels must be reformu- at least a sensory-motor system (Pho- lated. netic Form/PF) and a semantic/ Structure-building has also under- conceptual-intentional system (Logi- gone radical revision, returning to a cal Form/LF). The strong minimalist version of generalised transforma- thesis hypothesises that the only con- tions from Chomsky’s early work in straints are those imposed by these the 1950s. Rather than building up interfaces, and that computational syntactic objects in a top-down system satisfies these constraints in an fashion through recursive phrase- optimal fashion. structure rules, the so-called As with previous radical changes in operation builds structure from the framework developed by Chomsky, bottom up, combining two syntactic questions of , and ultimately objects of arbitrary complexity: that the logical problem of language acqui- is, individual lexical items or larger, sition, were central. By the early 1990s previously created syntactic objects. there was within Government-Binding Consistent with the goal of appealing Theory a sense that explanations were only to virtual conceptually necessary

135 MINIMALISM mechanisms, it is claimed that Merge Merge. As Chomsky notes, this view ‘comes for free’, insofar as language is of structure-changing entails the undeniably hierarchically recursive. copy theory of movement. The two As language is hypothesised to be elements of the chain have to be an ‘optimal’ solution to the conditions identical, rather than one being an imposed by the two interfaces, ques- inclusiveness-violating ‘trace’, be - tions of ‘economy’, both of represen- cause Merge cannot change what it tation and derivation, have taken on a operates on. What would tradition- new prominence. Economy of repre- ally be referred to as the head and the sentation prohibits ‘superfluous’ sym- tail of the chain can still be distin- bols. This entails that the interface guished, however, by examining the levels of PF and LF may only be com- context in which the copies appear. posed of symbols which are inter- Insofar as it is features, rather than pretable at that interface. This is the lexical items themselves, that are rele- principle of Full Interpretation. Fea- vant to the PF and LF interfaces, these tures uninterpretable at a given inter- become the focus of the syntactic face must be eliminated prior to that derivation within the Minimalist Pro- interface. The syntactic computation gram. As alluded to above, a distinc- therefore becomes principally driven tion is made between interpretable by the need to eliminate these uninter- and uninterpretable features. Certain pretable features. Economy of repre- features are plausibly interpretable at sentation also entails the principle of LF, though only on certain elements. Inclusiveness, which states that the Consider ‘agreement features’, such as computational system may not itself person, number and gender (usually introduce any symbols into the deri- referred to as phi-features). Whether a vation. As a result many mechanisms given Noun Phrase (NP) is singular or in previous Chomskyian frameworks, plural, second-person or third-person, such as traces, binding-theoretic affects the way in which the item indices and the like, do not exist. Syn- is interpreted. However, although tactic principles which make reference person and number features can to them must be eliminated or refor- appear on functional categories such mulated. Economy of derivation stip- as Tense (reflected ultimately in sub- ulates that the syntactic derivation ject-verb agreement), they seem not to itself proceed in an optimal fashion. have the same interpretive implica- In another radical change from pre- tions for Tense at LF. Thus the phi- vious Chomskyan approaches, the features on NPs are an instance of Minimalist Program claims that struc- ‘interpretable’ features at LF, while ture-building and structure-changing phi-features on Tense are uninter- are essentially the same operation. pretable at LF and must be eliminated. Movement is simply an instance of In the earliest minimalist literature Merge where the object merged at the of the early-mid 1990s, movement or comes from inside the existing Internal Merge was the mechanism by syntactic object rather than from out- which uninterpretable syntactic fea- side; Chomsky usually distinguishes tures were eliminated. However, in Internal Merge (Move) from External the early 2000s these operations

136 MINIMALISM became separated. The feature-check- object over to the sensory-motor inter- ing operation is Agree. It consists of a face and the conceptual-intentional probe, a syntactic element which has interface, preserving a strong version an uninterpretable feature (say, phi- of the strict cycle. On this view, the features on Tense), and a goal, an ele- difference between overt and covert ment which has an interpretable movement lies in whether an element version of this feature (say, an NP). is displaced before or after deciding to The Agree-relation deletes the unin- spell out the phase. terpretable feature of the probe and Cyclic spell-out and interpretation provides a ‘value’ for morphological entails a Phase Impenetrability Condi- purposes based on the appropriate tion. Since material within a phase has feature of the goal. This recreates the already been spelled out and inter- traditional claim that it is tensed verbs preted, it is no longer accessible to the that agree with their subjects, not vice syntactic computation. However, so versa. Displacement itself is triggered as not to completely rule out succes- by an edge feature (often also referred sive-cyclic movement in any form, it is to as an EPP feature) which particular hypothesised that material at the edge functional categories possess and of a phase (a specifier or of a which can act as a probe. Edge fea- phase head) and the phase head itself tures further underline the essential are accessible to further operations. identity of Merge and Move within Potentially one of the most interest- the Minimalist Program, as it is also ing things about the minimalist Pro- the edge feature of a lexical item gram, as noted by Chomsky (2004), is which permits it to be merged with its contribution to the longstanding another syntactic object (its comple- question of adequacy*. The highest ment). level of adequacy, explanatory ade- One of the most important mini- quacy, is attained when there is a gen- malist developments relating to struc- eral theory of grammars that provides ture-building and changing concerns the basis for selecting the most the concept of the phase, first intro- descriptively adequate grammar from duced in Chomsky (2001). In a bid to a selection of possible grammars. reduce computational complexity, However, explanatory adequacy may Chomsky suggested that derivations be just description at a higher level, to were inspected for legitimacy at par- the extent that the general theory of ticular key points; specifically, after grammars merely incorporates what- the Merge of a phase head: vP (the ever features the most descriptively sister to VP which is impli- adequate grammars happen to pos- cated in ) or CP. sess. Minimalism allows one to go Chomsky (2004) notes that even beyond explanatory adequacy, in the levels of PF and LF can be elimi- theory at least. Not only can the gen- nated under the assumption that spell- eral theory of grammars select the out and interpretation take place most descriptively adequate grammar, cyclically, at the level of the phase. but if language truly is a ‘perfect’ solu- When a phase is completed, the oper- tion given the constraints imposed, ation Transfer hands the syntactic then we can explain why a minimalist

137 MODALITY grammar ends up being the most former is usually called epistemic descriptively adequate. modality and the latter, . In formal semantics and Primary sources modal logic, modality concerns the Chomsky, Noam (1995). The Minimalist possibility or necessity of the predica- Program. London: MIT Press. tion and its negation. Chomsky, Noam (2000). ‘Minimalist See also: Speech Act Theory; inquiries: the framework’. In R. Martin, Politeness; Propositions D. Michales and J. Uriagereka (eds), Key Thinkers: Aristotle; Austin, Step by Step: Essays on Minimalism in J. L.; Searle, John Honor of Howard Lasnik. Cambridge, MA: MIT Press. 89–155. A proposition such as ‘They go home Chomsky, Noam (2001). ‘Derivation by at this hour’ can be communicated phase’. In M. Kenstowicz (ed.), Ken with different attitudes marking its Hale: A Life in Language. Cambridge, modality. There are two basic types: MA: MIT Press. 1–52. (1) , which is about Chomsky, Noam (2004). ‘Beyond explana- the truth of the proposition, such as tory adequacy’. In A. Belletti (ed.), Struc- guessing (‘They may have gone home tures and Beyond – The Cartography of at this hour’) and (2) deontic modal- Syntactic Structure, vol. 3. Oxford: ity, which is about the realisability of Oxford University Press. 104–131. the proposition by some agent, such Chomsky, Noam (2005). ‘Three factors in as granting permission (‘They may go language design’. Linguistic Inquiry home at this hour’). Closely related to 36.1: 1–22. deontic modality is dynamic modality which expresses ability (or inability) Further reading to put the assertion into practice inde- Hornstein, Norbert, Jairo Nunes and pendently of the judgement or will of Kleanthes K. Grohmann (2005). Under- the speaker, such as ‘They can go standing Minimalism. Cambridge: home; they still have the return tick- Cambridge University Press. ets’. Often held as part of epistemic Boškovic, Željiko and Howard Lasnik modality is – that is the (2007). Minimalist Syntax: The Essen- warrant the speaker has for the asser- tial Readings. Oxford: Blackwell Pub- tion, such as in ‘The show seems to be lishing. a success – whereby the speaker relies on some evidence, for example, Geoffrey Poole applause heard, rather than subjective judgement. Epistemic modality is also called proposition-modality, or speaker-ori- MODALITY ented modality. Deontic modality is also known as event-modality since it The expression of the speaker’s is about the realisability of the state of attitude concerning the truth of a affairs expressed in the proposition. In proposition or the realisability of a some recent literature, deontic modal- proposition by some agent. The ity is also described as agent-oriented

138 MODEL-THEORETIC SEMANTICS modality because it characterises how Discourse. Amsterdam and Philadel- some agent will carry out the proposi- phia: Benjamins. tion. Lyons, J. (1977). Semantics. Two vols. Modality is expressed lexically or Cambridge: Cambridge University grammatically or a combination of Press. both. Lexical expressions include such as ‘possibly’, ‘maybe’, Agustinus Gianto ‘probably’, ‘presumably’, ‘suppos- edly’, for example. Grammatical expressions make use of the modal verbs ‘can’, ‘may’, ‘must’, ‘shall’, MODEL-THEORETIC ‘will’. In a number of languages, the morphology of the verbal predicate SEMANTICS also indicates modality, hence the tra- ditional term mood, such as state- One of the mainstream approaches ment, question, imperative. Note that to the study of meaning in natural the illocutionary force of a statement languages. Model theory is a branch may generate modality, such as ‘This of mathematical logic concerning room is a bit dark’ can be interpreted the description of the semantics of as a request to switch on the light, or artificial languages. A model- a question like ‘Have we finished?’ theoretic approach to the semantics may mean an invitation to leave. of natural languages was first put for- ward by the Polish logician Alfred Primary sources Tarski in 1935. Following the Palmer, F. R. (2001). Mood and Modality. Tarskian tradition, current model- Second edition. Cambridge: Cambridge theoretic approaches in linguistics University Press. usually embrace a truth-conditional Von Wright, G. H. (1951). An Essay theory of semantics. Model-theoretic in Modal Logic. Amsterdam: North- semantics provides a mathematically Holland. rigorous and elegant way of describ- ing and explaining extremely intricate Further reading semantic phenomena. Hoye, Leo (1997). Adverbs and Modality See also: Analytic Philosophy; in English. London and New York: Compositionality; Logic; Longman. Logical Form Palmer, F. R. (1990). Modality and the Key Thinkers: Frege, Gottlob; English Modals. Second edition. Montague, Richard; Tarski, Alfred London and New York: Longman. Bybee, J. L.,R. D. Perkins and W. Pagliuca The most influential application of (1994). The Evolution of Grammar: model theory to natural language Tense, Aspect and Modality in the Lan- semantics can be found in the work of guages of the World. Chicago: Univer- the American logician and philoso- sity of Chicago Press. pher of language Richard Montague. Bybee, J. L. and S. Fleischman (eds) Analytic philosophers advocated the (1995). Modality in Grammar and systematic study of formal, idealised

139 MODEL-THEORETIC SEMANTICS languages as a means of better under- Montagovian model-theoretic seman- standing slippery notions like entail- tics, and its influence in linguistic ment (logical consequence) and semantics.) contradiction. They also believed that Central to the model-theoretic by approaching such semantic phe- approach to semantics is the notion of nomena through the analysis of artifi- ‘truth with respect to a model’. In cial languages, one could avoid the other words, the specification of the , vagueness and truth conditions of sentences in a lan- that are frequently found in natural guage is not provided in a vacuum, languages. but rather relative to an abstract Montague was of the same opinion. of some state of He famously held the view that natu- affairs in the world, which we could ral languages are not fundamentally think of informally as a ‘snapshot’ of different from formal languages and reality. The construction of idealised that, consequently, the same approach models of certain aspects of reality is can be employed fruitfully in their a common practice in many scientific study (Montague 1970). Thus Mon- contexts, as these can provide the tague extended the mathematical basis for successful explanations of techniques that were traditionally difficult phenomena in the natural sci- used in the semantics of formal lan- ences. guages to the study of meaning in nat- Model-theoretic semantics provides ural languages. interpretations for sentences in a lan- One such technique was model guage by first specifying the entities theory. Model theory was used by that exist in the state of affairs in the logicians to provide a precise specifi- world being modelled. Thus models cation of the truthconditions of sen- can be very complex and they can tences in artificial languages. Starting comprise abstract as well as real enti- with Tarski’s seminal essay ‘The Con- ties. For example, models usually con- cept of Truth in Formalized Lan- tain two special abstract entities guages’ (1935), the concept of truth referred to as True (or 1) and False (or has played a fundamental role in the 0), the truth values, which are often semantics of formal languages. Tarski assumed to be the semantic values of suggested that describing the interpre- true and false sentences, respectively. tation of a given sentence in a formal Models are, in effect, idealised repre- language involved giving, in a meta- sentations of what sentences in a lan- language previously understood, a guage can be about. It is important to precise specification of the conditions point out, however, that logical under which that sentence is true. models of natural language are not Montague’s model-theoretic seman- meant to capture our knowledge of tics for natural language is truth-con- language, and therefore should not be ditional, as it provides a theory of construed epistemologically. truth for a fragment of English. Having specified the of the (Partee (1975, 1996, 1997) provides a model, the task of a model-theoretic comprehensive overview of the main semantics is to assign interpretations ideas and historical development of to the basic expressions of the object

140 MODEL-THEORETIC SEMANTICS language by associating them with tions that must be obtained for entities in the model. Of course, dif- sentences in the to be ferent expressions may denote, or true relative to the model under refer to, different kinds of things in the consideration. Bach (1989), Cann model. For example, a proper name (1993), Dowty, Wall and Peters such as ‘Mary’ denotes a member of (1981) and Gamut (1991) provide the set of individuals in the model, excellent introductions to model- whereas the adjective ‘clever’ is taken theoretic semantics as it is currently to denote a subset of that set, namely practised in linguistics. the set of individuals in the model who Perhaps the most important advan- are clever. A sentence like ‘Mary is tage of the notion of ‘truth with clever’ is assigned a truth value*: true respect to a model’ is that it allows us if the sentence is actually true – that is, to define, in a rigorous and formally if Mary is in fact a member of the set precise way, key semantic properties of clever individuals in the model – or of natural language sentences like false otherwise. As we can see, models validity and contradictoriness, as well have internal structure. In order to as important meaning relations that have a systematic way of describing exist between them such as these distinctions, models are often or logical equivalence. Equipped with structured into different domains this notion, we can quantify over the according to the types of things that class of possible models and say, for each domain comprises. For example, example, that a sentence S1 entails a the three expressions mentioned sentence S2 just in case every model in above denote members of different which S1 is true is a model in which S2 domains in the model: respectively the is true as well; or that any two sen- domain of individuals, the domain of tences S1 and S2 of our object lan- sets of individuals and the domain of guage are logically equivalent just in truth values. case S1 is true in exactly the same A set of recursive syntactic rules models in which S2 is true and in no defining the class of well formed others; or that a sentence S1 is contra- sentences of the language must be pro- dictory just in case it is false with vided, together with a precise specifi- respect to every possible model. A cor- cation of how the various basic rect understanding of these properties expressions can be combined syntacti- and relations is at the heart of any sci- cally to yield complex expressions. entific theory of natural language For each of these syntactic rules, there semantics. is a corresponding semantic rule An issue that causes disagreement which determines the interpretation of between different theories of model- the complex expressions on the basis theoretic semantics is the syntax- of the interpretation of their parts, in semantics interface. For example, accordance with the principle of com- some theories postulate a separate level positionality* of meaning (sometimes of linguistic representation mediating also referred to as ‘Frege’s Principle’). between natural language syntax and Ultimately, a model-theoretic seman- model-theoretic semantics. This addi- tics must explicitly define the condi- tional level of representation, which is

141 NAMES derived from other linguistic levels by (eds), Handbook of Logic and Lan- transformational operations in the guage. Amsterdam and Cambridge, MA: syntax, is called logical form* (May Elsevier and MIT Press. 5–91. 1985). On the other hand, the strict Tarski, A. (1935). ‘The concept of truth in compositionality constraint between formalized languages’. In J. Corcoran form and meaning built into Mon- (ed.) (1983), Logic, Semantics, Meta- tague’s original theory precludes the mathematics. Indianapolis, IN: Hackett need for such an additional level of Publishing. 152–278. syntactic representation. Model-theoretic semantics can pro- Further reading vide interpretations for (semantic rep- Bach, Emmon (1989). Informal Lectures resentations of) linguistic expressions on Formal Semantics. Albany: SUNY in a way that accurately captures our Press. pre-theoretical semantic judgements Cann, Ronnie (1993). Formal Semantics: of such expressions. It is currently the An Introduction. Cambridge: Cam- focus of a great deal of research in lin- bridge University Press. guistics, but much of this research also Dowty, David R., R. E. Wall and S. Peters unifies and expands various other dis- (1981). Introduction to Montague ciplines, such as philosophy, . Dordrecht: D. Reidel Pub- science, artificial intelligence*, and lishing. theoretical computer science. Work in Gamut, L. T. F. (1991). Logic, Language model-theoretic semantics has made and Meaning (Volume 2): Intensional significant contributions to our under- Logic and Logical Grammar. Chicago: standing of truth and meaning in nat- University of Chicago Press. ural languages, for example in areas Iván García Álvarez as diverse as tense and aspect, gener- ics, negation, and plurality.

Primary sources NAMES May, Robert (1985). Logical Form: Its Structure and Derivation. Cambridge, In philosophy the term has been used MA: MIT Press. sometimes to describe any word or Montague, Richard (1970). ‘Universal phrase that can refer, and sometimes grammar’. Theoria 36: 373–98. in the more restricted sense of a word Partee, Barbara H. (1975). ‘Montague that uniquely identifies an individual. grammar and transformational gram- In linguistics, names or proper nouns mar’. Linguistic Inquiry 6: 203–300. are recognised as a separate class of Partee, Barbara H. (1996). ‘The develop- linguistic signs, distinct from common ment of formal semantics in linguistic nouns. The field of name studies, or theory’. In S. Lappin (ed.), The Hand- onomastics, is increasingly moving book of Contemporary Semantic Theory. from language-internal criteria to Oxford: Blackwell Publishing. 11–38. contextual ones. Partee, Barbara H. (with Herman L. W. Hendriks) (1997). ‘.’ See also: Definite Descriptions; In J. van Benthem and A. ter Meulen Sense/Reference; Use/Mention

142 NAMES

Key Thinkers: Frege, Gottlob; because of a chain of communication Kripke, Saul; Mill, J. S.; Russell, established within a community of Bertrand; Searle, John speakers. In linguistics, unique reference is The nature of names has exercised usually taken to be the key distinction philosophers of language for a very between names and common nouns. long time. The best known early However, this distinction is not as treatise is probably Plato’s Cratylus clear-cut as it may seem. First, some Dialogue which centres around the common nouns such as ‘sun’ or ‘hell’ question of whether names have have unique reference but they are an intrinsic relation to their referent not considered names. Second, the or are as arbitrary as other linguistic of many names have cer- signs. The discussion of names in tain things in common: all referents modern philosophy was initiated of the name ‘Mary’, for instance, are by John Stuart Mill (1867), who female, born to English-speaking distinguished between ‘connotative’ parents (or they are learners of Eng- names, such as ‘The father of lish as a who are Socrates’, which refer to an individ- using an English name in addition to ual by means of describing that their native one), and, given that individual, and ‘non-connotative’ most personal names are subject to names, such as ‘Sophroniscus’, which fashion, one may even make an edu- simply refer. In contrast, Gottlob cated guess about the age of many Frege (1892) argued that all names, Marys. including personal names, have The syntactic criterion that distin- both a reference and a sense. Accord- guishes proper nouns from common ing to Bertrand Russell (1919), per- nouns in English is the absence of a sonal names can be used to refer . However, not all common directly to an individual who is nouns can take the full range of deter- immediately present, but in most miners: ‘a music’, for example, is not instances are used as ‘abbreviated a possible expression in English. The descriptions’. There are various non-translatability of names has been problems for Russell’s account, put forward as another distinction including the fact that, since different criterion. However, counter-examples speakers may identify an individual are easy to find such as German by means of different descriptions, ‘Schwarzwald’, which becomes ‘Black the meaning of a name would seem to Forest’ in English. vary depending on who is using it. In Thus, it is obvious that proper and one attempt to address this problem, common nouns form a gradient, with John Searle (1958) developed his prototypical cases at either end and ‘cluster theory’ account. In perhaps many fuzzy ones in between. Ulti- the most influential philosophical mately, the intention and perception discussion of names since Russell, of language users is the only way to Saul Kripke (1972) argued that distinguish between common nouns names are ‘rigid designators’ that and proper nouns. However, this dis- succeed in referring to individuals tinction is no longer systematic and

143 inherent in language but a matter of NONNATURAL naming practices in context. There can be no doubt that the rela- MEANING tionship between a name and its refer- ent has more psychological reality for A type of meaning, which includes lin- language users than that pertaining guistic meaning, in which there is no between any other linguistic signs and necessary link between a sign and their referents. Modern linguistics and what it represents. In his influential onomastics, however, have tended to account of nonnatural meaning, H. P. dismiss this widely held assumption Grice emphasised the importance of a as ‘primitive’ or ‘superstitious’. This speaker’s intentions and a hearer’s stance has led to an ever-widening recognition of these intentions. chasm between limited academic See also: Conventional Meaning; interest in names and naming and Implicature; Intentionality; Signs widespread popular interest. In the and Semiotics; Use/Mention philosophy of language, the relation- Key Thinkers: Grice, H. P.; Peirce, ship between names and their refer- C. S.; Saussure, Ferdinand de; ents remains an enduring focus of Searle, John; Strawson, P. F. interest. The idea that linguistic meaning can Primary sources be distinguished from natural mean- Frege, Gottlob (1892). ‘On sense and ing is long standing. It can be traced meaning’. In Peter Geach and Max back to classical philosophy, and in Black (eds) (1980), Translations from more recent times can be found in the Philosophical Writings of Gottlob Peirce’s account of words as ‘sym- Frege. Oxford: Blackwell. 56–78. First bols’ and in Saussure’s discussion of Edition 1952. the arbitrary nature of the sign. How- Kripke, Saul (1972). ‘Naming and neces- ever, Grice’s short article ‘meaning’ sity’. In D. Davidson and G. Hartman attempted a more precise characteri- (eds), Semantics of Natural Language. sation. Dordrecht: Reidel. Grice draws attention to two differ- Mill, J. S. (1867). A System of Logic. ent ways in which the verb ‘mean’ is London: Longman. Chapter 2. used, exemplified by ‘Those spots Russell, Bertrand (1919). Introduction to mean measles’ and ‘Those three rings Mathematical Philosophy. London: on the bell (of the bus) mean that the George Allen and Unwin. bus is full’. The first case involves nat- Searle, John (1958). ‘Proper names’. Mind ural meaning; the spots simply are a 67: 166–73. symptom of the disease. In the second case, there is no necessary connection Further reading between the three rings and the mean- Evans, Gareth (1982). The Varieties of ing, but rather someone (the bus con- Reference. Oxford: Oxford University ductor) meant something by the rings; Press. we can say that the rings mean ‘The bus is full’. Unlike in the case of natural Ingrid Piller and Siobhan Chapman meaning, the rings are not a guarantee

144 OPTIMALITY THEORY of the truth of this statement; the bus Searle, John (1969). Speech Acts. Cam- conductor may be mistaken or deliber- bridge: Cambridge University Press. ately trying to deceive. Grice labelled Strawson, P. F. (1964). ‘Intention and con- this second type of meaning ‘nonnat- vention in speech acts’. The Philosophi- ural meaning’, for which he coined the cal Review 73: 439–60. abbreviation ‘meaningnn’. Linguistic meaning is a type of meaningnn. Further reading For Grice, meaningnn is determined Avramides, Anita (1989). Meaning and by a speaker’s intention to communi- Mind. Cambridge, MA: MIT Press. cate something, and by a hearer’s Chapman, Siobhan (2005). Paul Grice: recognition of that intention. If A is a Philosopher and Linguist. Basingstoke: speaker and x an utterance, then ‘“A Palgrave. meantnn something by x” is roughly equivalent to “A intended the utter- Siobhan Chapman ance of x to produce some effect in an audience by means of the recognition of this intention”’ (1957: 385). Grice raises but does not fully develop the OPTIMALITY idea that linguistic meaning (‘what x THEORY meansnn’) may itself be determined by speakers’ intentions, and hence that conventional meaning is to be defined A framework in theoretical linguistics in terms of psychology. He also hints used to formalise analyses in phonol- that what speakers intend to commu- ogy, and less frequently other areas of nicate in specific contexts may some- linguistics. Its core is the assumption times go beyond what their words that linguistic generalisations should actually meannn. be described using a set of violable ‘Meaning’ has largely been constraints on surface representations favourably received by philosophers which are ranked in terms of their and linguists alike, although some importance. have put forward criticisms that sug- See also: Generative Phonology; gest flaws in Grice’s theory (Strawson Universal Grammar 1964; Searle 1969; Schiffer 1972). Key Thinkers: Chomsky, Noam ‘Meaning’ also introduced Grice’s interest in the distinction between Optimality theory (OT) took centre what our words mean and what we stage in theoretical linguistics during mean by using our words, which he the 1990s. Its basic tenets were both explored further in his work on con- familiar and revolutionary and this versational implicature*. has doubtless contributed to its suc- cess. It developed ideas long present Primary sources in linguistics, but gave them new Grice, Paul (1957). ‘Meaning’. The Philo- characteristics, considerably chang- sophical Review 66: 377–88. ing the understanding of the gram- Schiffer, Stephen (1972). Meaning. mar. Despite some influence from Oxford: Clarendon Press. neural networks, OT was essentially

145 OPTIMALITY THEORY conceived as a development of gener- tial, setting out the OT approach ative phonology*, so it maintains the (along with work by McCarthy and basic competence/performance dis- Prince). tinction, and that between underlying OT analyses are formulated in a and surface levels of representation, ‘tableau’, exemplified below (see although these are reinterpreted. One Table 1) for German Final Obstruent central OT assumption is that only Devoicing (FOD) in its standard constraints should be used to charac- description: syllable-final voiced terise linguistic generalisations. There obstruents are devoiced, as in Bund are no substantive rules or transfor- [bʊnd] ‘federation’ (a contentious but mations. A set of potential surface standard assumption here is that [d] forms (‘candidates’) are evaluated by in German is specified for [voice], the grammar and one is chosen as while [t] is unspecified). The tableau’s the optimal candidate (the ‘output’) top row shows the input (the under- because it violates a language’s con- lying representation), and then the straints in the least bad way. This is constraints, in ranked order from left possible thanks to another key char- to right. The first column shows the acteristic: all constraints are violable, set of candidates, one of which is and languages rank them in order of chosen as the output (indicated by a importance, so it is worse to violate a pointing finger). The other columns high-ranked constraint than a low- show constraint violations, each ranked one. As the standard assump- receiving one asterisk. The ‘fatal’ vio- tion is that the set of constraints lation (which rules candidates out) is (known as Con) is universal, lan- indicated by an exclamation mark. guage learners must work out how There are two basic types of con- constraints are ranked in the lan- straint: and faithfulness guages they acquire. constraints. Markedness constraints OT provided linguists with new penalise candidates which contain ways to work and new theoretical marked structures: the analysis of problems to crack, although some FOD uses *FinalObstruent/Voice have rejected it as fundamentally mis- (*FOV), which dictates that final guided. It was introduced by Alan obstruents may not be specified for Prince and Paul Smolensky and has [voice]. Faithfulness is a crucial inno- been developed by many others, most vation of OT: markedness constraints notably John McCarthy. OT was cre- can exert pressure for an output to ated with phonological problems in differ from its input, but faithfulness mind, and is still most popular constraints do the opposite, requiring among phonologists, but is also used identity between input and output. in syntax and other areas. After early Individual constraints regulate par- presentations in 1991, Prince and ticular aspects of faithfulness: Smolensky distributed the manu- Ident() that the value of [voice] be script Optimality Theory: Constraint the same in input and output, and Interaction in Generative Grammar Max requires everything in the input in 1993. Although not published to have some correspondent in the until 2004, this was highly influen- output, ruling out deletion.

146 OPTIMALITY THEORY

/bʊnd/ *FOV Max Ident(voice) bʊnd !* ☞bʊnt☞ * bʊn!* bʊ !**

Table 1

Fully faithful [bʊnd] is rejected indistinct: thus it does not matter because it violates high-ranked *FOV. whether or not stops are aspirated An unlimited number of candidates underlyingly in English, as these seg- are in fact produced (by the function ments do not contrast and, assuming Gen,which simultaneously performs that aspiration is enforced by a high- every possible process on the input to ranked Aspiration constraint (‘initial derive the candidates). In practice, stops must be aspirated’), then [pɪk] only a few ‘reasonable’ candidates are could be derived from either [pɪk] or considered in analyses (outlandish [phɪk]. Which of these is the under- candidates are assumed to be ruled lying form is simply not an important out by high-ranked faithfulness con- (or relevant) question. This character- straints). The candidates include istic is known as the Richness of [bund], which satisfies *FOV, with no the Base, and has some theoretical voicing in its final obstruent. [bund] benefits. In the German case, how- violates Ident(), but is still the output ever, alternations (such as Bund~ because other candidates violate Bundes [bʊnd]~[bʊndəs] ‘federation’~ higher-ranked constraints. The other ‘federation-genitive’) show that the two candidates do not violate *FOV, underlying must be voiced. as they have no final obstruent OT analyses can often be linked to (having undergone deletion), but they pre- or non-OT work by implement- do violate Max, also ranked above ing as a violable constraint what such Ident(). This ranking is shown by the work sees as a universal principle. unbroken line between Max and This has likely contributed to OT’s Ident(). The broken line between success because it expresses what pho- *FOV and Max shows that their nologists have wanted to be able to mutual ranking is irrelevant here. The say: certain principles often play a role above ranking is specific to languages in the phonology of languages, but with FOD. In English, with no FOD, not always. For example, the Obliga- the constraint *FOV must be ranked tory Contour Principle (OCP) was below Ident(). introduced in the 1970s to forbid It is worth noting that the right can- adjacent identical aspects of structure, didate would still be selected if the such as adjacent high tones or specifi- input were /bʊnd/. In fact, in some cations for place. By the late 1980s, it cases, the precise nature of the input is had been shown that the OCP is not

147 OPTIMALITY THEORY absolute, but is best understood as a analyse simple FOD are subject to tendency in languages, elevated to the debate: there is not necessarily one level of an inviolable principle in straightforward way to analyse any some. This is exactly what OT pre- phenomenon, because there is no clear dicts, and OT thus allows phonolo- theory of Con. Some see this as an gists to formalise tendencies. opportunity for debate, others as a Theoretical development in OT has problem: how can we know which largely involved either work on the constraints exist? A more fundamen- types of constraints allowed, driven tal criticism questions whether we by attempts to find new, better ways to really should formalise all tendencies express linguistic generalisations, or as cognitive constraints on surface work on additions to the basic theo- forms. Should ‘ease of articulation’ retical machinery, often to remedy (formalised in OT as Lazy) be seen as perceived shortcomings in the model, the same kind of thing as constraints such as its inability to account for on the faithfulness of features? opacity. The former includes work on Despite such criticism, OT has a ‘positional faithfulness’, which prefer- firm place in theoretical linguistics, as entially preserves input specifications the framework for most analysis and in strong positions, such as the onset. debate in phonology, and consider- FOD can also fit into a positional able work in other areas of linguistics. faithfulness model: a positional It provides a novel set of principles to IdentOnset(voice) can require that guide analysis and has allowed new the value of [voice] be the same in input answers to old problems. For those and output in onsets, and conflict with who follow its direction, it has a general *Obstruent/ Voice, which focused work on the status of con- requires that no obstruent be specified straints and moved attention away for [voice]. This achieves the same from representations and underlying result, but also opens up new ‘posi- levels. The framework for constraint tional’ analytical options. interaction has also been adopted Opponents of OT often argue that by linguists who reject generative its central tenets are mistaken. Critics assumptions. Its influence is felt have claimed that the status of inputs widely in linguistic theory. is problematic. The Richness of the Base has been criticised on both psy- Primary sources cholinguistic and theoretical grounds Prince, Alan and Paul Smolensky (much previous theoretical success (1993/2004). Optimality Theory: Con- was due to the ability to specify the straint Interaction in Grammar. Rut- characteristics of underlying forms). gers University and University of The fact that OT has largely led its Colorado at Boulder. Oxford: Black- practitioners to abandon work on well. phonological representation (espe- McCarthy, John and Alan Prince (1993). cially at the segmental level) has been Prosodic Morphology: Constraint condemned. The contents of Con Interaction and Satisfaction. University have also proved controversial. As we of Massachusetts, Amherst and Rutgers have seen, the constraints needed to University.

148 ORDINARY LANGUAGE PHILOSOPHY

McCarthy, John and Alan Prince (1995). philosophy or , ‘Faithfulness and reduplicative iden- but these terms should not taken as tity’. In Jill Beckman, Laura Walsh accurately descriptive. It is true that Dickey and Suzanne Urbanczyk (eds), Oxford University served as its princi- University of Massachusetts Occasional pal locus, with key Oxford philoso- Papers in Linguistics 18: 249–384. phers such as Gilbert Ryle, P. F. Strawson and J. L. Austin contribut- Further reading ing to its development. But it is often McCarthy, John (2002). A Thematic pointed out that many of the ideas of Guide to Optimality Theory. Cam- the later Wittgenstein also had an bridge: Cambridge University Press. impact on the movement, and John Wisdom, associated like Wittgenstein Patrick Honeybone with the , is generally considered a sympathiser of OLP. Furthermore it is not true that there was anything approaching a ORDINARY consensus among the philosophers of LANGUAGE Oxford regarding the main ideas of the movement; it is arguable that the PHILOSOPHY group did not even constitute a move- ment or a school in any real sense. At Ordinary language philosophy (OLP) best they shared some common atti- is the name given to a philosophical tudes concerning the nature of lan- movement that developed in England guage and ways of doing philosophy. during the years between the two To use a term attributable to Wittgen- World Wars and enjoyed its heyday in stein, the farthest one may go in char- the late 1940s through the early acterising the group is to say that there 1960s. Its adherents saw ordinary, was a ‘’ among the everyday language as the starting positions assumed by many of its point for their philosophical inquiries. members. Through the central ideas of its lead- The term ‘linguistic philosophy’ is ing figures such as J. L. Austin and more to the point. These philosophers Ludwig Wittgenstein, OLP continues put language at the centre of their to echo in a wide variety of disciplines. inquiry. This was in sharp contrast with the customary practice of regarding it See also: Analytic Philosophy; as merely (and, in the views of some, Implicature; Speech Act Theory; lamentably) a tertium quid (third ele- Truth Theories ment) between René Descartes’ res Key Thinkers: Austin, J. L.; extensa (extended matter or material Derrida, Jacques; Grice, H. P.; reality) and res cogitans (thinking sub- Russell, Bertrand; Ryle, Gilbert; stance or, simply, the mind). In this Searle, John; Strawson, P. F.; sense, linguistic philosophy marked a Wittgenstein, Ludwig major departure from the long tradi- Ordinary language philosophy is tion of analytic philosophy, of which it also variously referred to as Oxford is nevertheless an offshoot. However,

149 ORDINARY LANGUAGE PHILOSOPHY even here, one cannot point to any con- arguing that such an approach would sensus. John Searle, of the University of clarify matters by blocking misuses of California at Berkeley, probably the language. In other words, most of phi- most famous of Austin’s followers, has losophy, on careful inspection, would categorically denied being a linguistic boil down to straightforward linguistic philosopher, claiming to be only a analysis. philosopher of language. J. L. Austin, one of the principal The term ‘ordinary language phi- exponents of ordinary language phi- losophy’ captures an important ele- losophy, used to quip that if one wants ment that distinguishes the movement to embark on a career in philosophy from the work of earlier philosophers, one had better equip oneself with a particularly those inspired by logical good dictionary. His point was that positivism* or what may contrastively our ordinary, everyday language, be referred to as ‘ideal language phi- unlike its ideal or formal counterpart, losophy’. Bertrand Russell, for one, was fashioned into its present form by was openly scornful of ordinary lan- generations of speakers. In that long guage which he thought was full of process, the language has been ambiguity* and vagueness and hence invested with most, if not all, of the inadequate for the philosophers’ pur- distinctions that its speakers felt nec- poses. Unlike many of their con - essary. Not that the ordinary language temporaries and predecessors who cannot be improved or further dis- believed in first positing a logically tinctions introduced into it to suit perfect language and then lamenting fresh needs. Ordinary language may how our ordinary, everyday language not provide us with the last word on pales in comparison with it, ordinary philosophical problems, but it should language philosophers insisted on certainly be the starting point of all examining ordinary, everyday lan- philosophical inquiry. guage at its face value. The philosophical importance of In stark contrast with the apologists conferring pride of place upon ordi- of ideal language, Wittgenstein argued nary language can hardly be overesti- in his Philosophical Investigations mated. In a way it debunked the very (1953) that ordinary language was per- enterprise of philosophy or at least a fectly in order as it stood, and that traditional way of going about doing many of the puzzles that professional philosophy. The image of the proto- philosophers and linguists encountered typical philosopher is best captured by and sought to resolve were actually the the famous statue called ‘The Thinker’ result of an inadequate understanding by the French sculptor Auguste Rodin. of the subtleties of ordinary, everyday Lost in meditation, the philosopher is language. In his essay ‘Systematically completely out of touch with the Misleading Expressions’, Ryle (1932) work-a-day world. Introspection is his made a strong case for a careful analy- preferred modus operandi and soli- sis of ordinary language expressions as tude his self-imposed ambience. This is a way of doing philosophy, or rather, perfectly in tune with philosophy’s as a way of ‘dissolving’ many of the proverbial disdain for language in the problems that crop up in philosophy, everyday sense of the word. Gottfried

150 ORDINARY LANGUAGE PHILOSOPHY

Wilhelm Leibniz, the German philo - the hilt. True to the austere tradition of sopher, is famously said to have sober empiricism, the hallmark of exclaimed that if Almighty were British thought, he preferred to start one day to descend upon the earth, He his philosophical inquiry with the tan- would certainly address us ordinary gibles – to start from the concrete mortals in the language of mathemat- (words) and work his way towards the ics – it was inconceivable for the general or the abstract, rather than the philosopher that an all-perfect God other way around. And he was in no would have recourse to any language hurry to finish that job. He believed other than the all-perfect language of that philosophy demanded a great deal mathematics. of painstaking spadework. He once Traditionally, philosophy is con- said that the sentence ‘Neither a be-all cerned with conceptual analysis. It is nor an end-all be’ could make an excel- not primarily concerned with words. lent motto for philosophy. Words are believed to embody con- Another essay by Austin, ‘A Plea for cepts at best. For the most part, tradi- Excuses’, is widely regarded as a tional philosophers distrusted words, prime example of OLP at work. By judging them misleading representa- carefully teasing out the different uses tions of the concepts behind them. of the word ‘excuse’ (which excul- Recall the Socratic practice, illus- pates the doer of an action), Austin trated in several of Plato’s Dialogues, distinguishes it from ‘justification’ of insisting that his interlocutor pro- (which, by contrast, ‘absolves’, as it vide a definition of, say, piety, instead were, the action itself of any imputa- of pointing to examples of pious per- tion of wrong-doing). This essay also sons. The moral is that true under- highlights what may be seen as yet standing comes from pure conceptual another hallmark of OLP – deflecting analysis which is what a definition is the focus from the naming of an act all about. In Socrates’ view, examples (along with the hypostatisation that it give the false impression that one has invariably involves) to the more mun- got to grips with the concepts behind dane doing of the act or, simply put, them. the action itself. This idea found its Austin’s attitude to this time- full expression in How to Do Things honoured practice among philoso- with Words (1962a), undoubtedly phers was eloquently expressed in a Austin’s most famous work. Before paper entitled ‘Are there any a priori proceeding any further, it is important concepts?’, originally presented in to comment on the title of the book, 1939 before the Aristotelian Society. In originally presented as a series of it Austin surprised his audience by twelve lectures presented at the Uni- arguing that he had no idea of what versity of Harvard in 1955 under the that question meant because he did not title ‘Words and Deeds’ and published know what concepts were to begin posthumously. It has been suggested with. So the question of concepts being that Austin chose that title somewhat a priori or a posteriori simply did not lightheartedly after Dale Carnegie’s arise. On this question and on many bestselling How to Win Friends and others, Austin was an Aristotelian to Influence People. In fact, many an

151 ORDINARY LANGUAGE PHILOSOPHY

Austin reader has a rather difficult expected answer would have it, ‘a time coming to terms with what is best barn’. Austin thus challenges what described as his blithe irreverence for may be described as a mainstay of Pla- the sober, lofty style that one typically tonic realism, namely, the idea that associates with philosophy, and his epistemology can only play second use of a thoroughly colloquial style. fiddle to ontology or, to put it more But it is important to realise that simply, what there is can by no means Austin is making an important philo- be affected by what one comes to sophical point here (Rajagopalan know about it. 2000a). The performative view of language In How to Do Things with Words thus has important things to say in Austin argued that a sentence such as respect of what has, from time imme- ‘The cat is on the mat’, which has long morial, been the philosopher’s Holy been seen as a declarative (or, ‘consta- Grail: the concept of truth. P. F. Straw- tive’ as he calls it) and hence capable son advanced what is referred to as of being judged true or false, is actu- the ‘performative theory of truth’ ally a ‘performative’* which cannot according to which truth is not some- be judged true or false but only happy thing to be approached solely with the or otherwise, depending on the cir- tools of the trade available in philoso- cumstances in which it is uttered or phy in its conventional sense, but ‘performed’. Thus was launched the something invariably mediated by idea that speaking a language is per- language. Truth claims, in other forming a series of acts. This seminal words, take precedence over truth idea has inspired a number of scholars ipsis. To call something true is to per- from a number of diverse disciplines form the speech act of endorsing it or including linguistics, psychology, giving it one’s stamp of approval. anthropology, sociology, and even of philosophy often treat such unlikely fields as economics. OLP as a chapter in the unfinished A discussion of Austin’s legacy book of analytic philosophy but the would be incomplete without a men- point is debatable. Both the later tion of his book Sense and Sensibilia Wittgenstein and Austin produced (1962b) (the echo of Jane Austen’s cel- immense challenges to what was the ebrated novel is unmistakable here). established dogma in philosophy. Per- Austin takes on the fashionable haps the best proof of the radical approach to the analysis of perception nature of their philosophical positions in terms of sense data* and proceeds is the publication in 1959 of a book by to deconstruct the so-called ‘argument Ernest Gellner entitled Words and from illusion’. For instance, he says, Things: An Examination of, and an the right answer to the question as to Attack on, Linguistic Philosophy. what it is that you actually see, posed The book even carried a laudatory apropos of a church camouflaged so preface by Bertrand Russell in which as to look like a barn, is precisely what he lamented that his former pupil that description says, namely, ‘a Wittgenstein seemed to have taken a church camouflaged so as to look like holiday from serious intellectual activ- a barn’ and not, as the traditionally ity, meaning thereby that he considered

152 ORDINARY LANGUAGE PHILOSOPHY

OLP at best a diversion from philoso- he was widely acclaimed as Austin’s phy proper and at worst an aberration. intellectual legatee. The book was But the very fact that the book became welcomed by many as genuinely Aus- a major sensation immediately after it tinian in spirit and an improvement was published speaks volumes for the upon Austin’s own How to Do Things popularity of OLP at that time and the with Words, which had left many threat it represented to mainstream readers perplexed and unsure of just philosophy. No doubt Gellner’s book what Austin had in mind. What many did quite some damage to OLP, though readers of Searle’s book failed to take once the heat of the moment had died serious notice of was something Searle down, many scholars were of the opin- had declared right in the subtitle of his ion that the book threw very little light book: An Essay in the Philosophy of on substantive issues, instead content- Language. The philosophy of lan- ing itself with mud-slinging and vitu- guage casts a much wider net than lin- perative rhetoric. guistic philosophy, which is one way Perhaps the real damage to OLP was of philosophically looking at language done not by die-hard detractors like but not the only one. In fact, in work Gellner, but by those responsible since then, Searle has evinced little for clever appropriations of the her- sympathy for many of the trademark itage of some of its principal figures, positions assumed by linguistic philo - notably Wittgenstein and Austin. We sophers, notably Austin himself have already seen how Bertrand (Rajagopalan 2000b). Russell, a towering figure in British It has been argued that Searle rein- philosophy who personally supervised terpreted Austin and put his ideas Wittgenstein’s earlier work Tratactatus back on the beaten track of traditional Logico-Philosophicus (Wittgenstein analytic philosophy. In the process, 1922/1961), simply brushed aside his many of the potentially subversive ele- later work, Philosophical Investiga- ments in Austin’s thought were either tions (Wittgenstein 1953/1968), dis- downplayed or simply ignored. With- missing it as at best frivolous. This out doubt this made Austin a house- furnished the vital clue to many suc- hold name in disciplines such as ceeding generations to stay with the linguistics, but many critics have com- earlier Wittgenstein and to summarily plained that the Austin appropriated dismiss the later Wittgenstein. by Searle is a far cry from the Austin With Austin, the philosophical who so vigorously and implacably establishment adopted a different defended linguistic philosophy. tactic. After his untimely death in Be that as it may, the later Wittgen- 1960, his mantle was widely consid- stein and Austin have had a consider- ered to have fallen on John Searle who able influence elsewhere in academia, had studied at Oxford in the early and continue to do so. Wittgenstein’s 1950s and who therefore knew him influence on contemporary linguistic personally. In 1969 Searle published thought is undeniable. His insistence his book Speech Acts: An Essay in the on the normative character of lan- Philosophy of Language which was guage has profound implications for an instant success in linguistics, where contemporary emphasis on ethics and

153 ORDINARY LANGUAGE PHILOSOPHY politics in relation to language. Note Judith Butler’s notions of ‘performa- that Wittgenstein’s position is in stark tivity’ and ‘performance’ (Butler contrast with the position assumed in 1997), both inspired by Austin’s ‘per- much of linguistics where the belief formative utterance’. Building on has been that a grammar must ideally Austin’s insights, Butler argues that be descriptive rather than prescrip- gender is not a matter of what one is tive. Another area that has been influ- (constative), but rather what one does enced by Wittgenstein’s thought is (performative). While OLP is often cognitive science. The theory of pro- associated with a specific time and totypes* – a model of graded cate- place, its influence goes much wider gorisation that eschews the classic, than linguistics and the philosophy of Aristotelian distinction between language thanks to the ongoing work essences and accidents – is a direct of scholars in other disciplines who spin-off from Wittgenstein’s notion of find inspiration in its central tenets. ‘family resemblance’. The idea that class membership does not have to be Primary sources a matter of all or nothing, but can be Austin, J. L. (1939). ‘Are there a priori a matter of more or less, has impor- concepts?’. In J. L. Austin (1961), Philo- tant consequences for how different sophical Papers. London: Oxford Uni- societies categorise entities in the versity Press. 32–54. work-a-day world (Rosch and Lloyd Austin, J. L. (1956). ‘A plea for excuses’. 1978; Lakoff 1987). In C. Lyas (ed.) (1971), Philosophy and Austin’s influence on contemporary Linguistics. London: Macmillan Press. thought is less often acknowledged, 79–101. Originally published in the thanks to the widespread tendency to Proceedings of the Aristotelian Society. credit it to Searle instead. Jacques 1956–7. Reprinted in J. L. Austin Derrida has made no secret of the (1961), Philosophical Papers. London: enormous influence of Austin’s think- Oxford University Press. 175–204. ing on his own. He even became Austin, J. L . (1961). Philosophical Papers. involved in a protracted exchange London: Oxford University Press. with Searle over the legacy of Austin. Austin, J. L. (1962a). How to Do Things An unlikely area of inquiry taken by with Words. Oxford: Clarendon Press. storm in the 1980s, thanks to the Austin, J. L. (1962b). Sense and Sensibilia. influence of OLP, was economics. In Oxford: Clarendon Press. 1985 Deirdre (then, Donald) Mac- Gellner, Ernest (1959). Words and Things: Closkey published a book in which An Examination of, and an Attack on, she argued that economists are closer Linguistic Philosophy. London: Victor to poets and story-tellers than they Gollancz. think, and that their claims are Ryle, Gilbert (1932). ‘Systematically mis- mistakenly taken to be constative, leading expressions’. In Richard Rorty whereas they are just as performative (ed.) (1967), The . as any other. Perhaps a most signifi- Chicago: University of Chicago Press. cant testimony to the vitality of lin- Searle, J. R. (1969). Speech Acts: An Essay guistic philosophy in general and in the Philosophy of Language. Cam- Austin’s heritage in particular is bridge: Cambridge University Press.

154 Wittgenstein, Ludwig (1922/1961). Trac- PERFORMATIVE tatus Logico-Philosophicus. London: Routledge and Kegan Paul. Wittgenstein, Ludwig (1953/1968). Philo- A description of utterances, sentences sophical Investigations. Oxford: Basil and speech acts the use of which is Blackwell. meant to create facts in addition to the fact of one of them having been used. Further reading Often, the appearance of ‘hereby’ Butler, Judith (1997). Excitable Speech: A marks the performative nature of the Politics of the Performative. New York: expression, as in ‘Master Little John is Routledge. hereby created Sheriff of Nottingham’ Hacking, Ian (1975). Why Does Language (Robin Hood). Matter to Philosophy? Cambridge: See also: Speech Act Theory Cambridge University Press. Key Thinkers: Austin, J. L.; Searle, Lakoff, George (1987). Women, Fire and John Dangerous Things: What Categories Reveal about the Mind. Chicago: Uni- For a long period, philosophers of lan- versity of Chicago Press. guage and logicians of natural lan- McCloskey, Deirdre (1985). The Applied guage were interested in sentences the Theory of Price. Second edition. New utterance of which in appropriate York: Macmillan. contexts expresses propositions that Rajagopalan, Kanavillil (2000a). ‘Austin’s are either true or false, for example, humorous style of philosophical dis- ‘Shakespeare is the author of Romeo course in light of Schrempp’s interpreta- and Juliet’ or ‘All ravens are white’. It tion of Oring’s “incongruity theory” of was always clear that not all sentences humor’. Humor: An International Jour- are of that type, for example English nal of 13 (3): 287–311. sentences in the or the Rajagopalan, Kanavillil (2000b). ‘On imperative. It was J. L. Austin (1961) Searle [on Austin] on language’. who drew attention to sentences of a Language and Communication 20 (4): seemingly different type that are quite 347–91. commonly used, the ‘performatives’. Rajagopalan, Kanavillil (2004a). ‘John The leading intuition underlying Langshaw Austin’. In Philipp Strazny the distinction between propositional (ed.) (2004), of Linguistics. utterances and performative ones is New York: Fitzroy Dearborn. 98–100. that in the former case the story of Rajagopalan, Kanavillil (2004b). ‘John the utterance, as commonly intended, Searle’. In Philipp Strazny (ed.) (2004), conveyed and understood, is that of Encyclopedia of Linguistics. New York: presenting a given fact about the state Fitzroy Dearborn. 936–8. of affairs, while in the latter case the Rosch, Eleanor and Barbara B. Lloyd (eds) story involves some additional ele- (1978). Cognition and . ment that creates a new fact. Thus sin- Hillsdale, NJ: Lawrence Erlbaum cerely saying ‘The door is open’ is Associates. describing what the speaker takes to be a fact about the door, under Kanavillil Rajagopalan the circumstances of utterance, while

155 PHONEME

sincerely saying ‘I promise to return Bach K. and R. M. Harnish (1992). ‘How the book’ is creating a commitment on performatives really work’. Linguistics the part of the speaker to return the and Philosophy 15: 93–110. book. The commitment is created by Searle, John (1989). ‘How performatives the linguistic institution of promising. work’. Linguistics and Philosophy 12: According to the rules that constitute 535–58. and regulate that institution, uttering a sentence of the form ‘I promise to do Asa Kasher this and that’ counts as undertaking a commitment to do this and that. Promising is a linguistic institution. Appointing is a non-linguistic institu- PHONEME tion, according to the rules of which an utterance of ‘I hereby appoint you A term generally understood in the deputy sheriff’, under appropriate cir- British and American structuralist tra- cumstances, counts as creating you a ditions to refer to a fundamental deputy sheriff. What is common to abstract linguistic unit physically re - both linguistic and non-linguistic insti- alised by multiple context- dependent, tutions is that they determine phonetically similar speech sounds, conditions for creating new facts, such and possessing the capacity to change as a certain person having a certain meaning but bearing no semantic commitment or serving in a certain properties of its own. Several defini- capacity, which are institutional facts. tions of the term exist, and disagree- The question whether a performa- ment persists over when and by whom tive utterance is an utterance of an it was first used. assertion has been debated. Searle See also: Distinctive Features; (1989) answers it in the negative, Emic/Etic; Generative Phonology; while Bach and Harnish (1992) argue Optimality Theory; that such an utterance is primarily Poststructuralism; Type/Token assertoric. Key Thinkers: Bloomfield, According to the debatable Perfor- Leonard; Firth, J. R.; Hockett, mative Hypothesis developed within Charles; Jakobson, Roman; Jones, generative semantics*, the deep struc- Daniel; Martinet, André; Pike, ture of every sentence, including Kenneth; Saussure, Ferdinand de; indicative ones, includes a prefix of Trubetzkoy, N. S.; Whorf, the form ‘I + (performative) verb’. Benjamin Lee Hence, every utterance is made within the framework of some institution. Daniel Jones (1967) characterises phonemes as ‘small families of Primary sources sounds, each family consisting of an Austin, J. L. (1961) ‘Performative Utter- important sound of the language ances’. In J. O. Urmson and G. J. together with other related sounds Warnock (eds), J. L. Austin, Philosoph- which, so to speak, “represent” it in ical Papers. Oxford: Clarendon Press. particular sequences or under particu- 233–52. lar conditions of length or stress or

156 POLITENESS intonation’ (p. 7). These family mem- taken as self-evident by researchers in bers are known as allophones (derived domains outside theoretical phonol- from the Greek for ‘other sound’) fol- ogy, for instance in the study of first lowing Benjamin Lee Whorf’s first use and second language acquisition, of the term in the early 1930s. speech and hearing disorders and A language’s phonemes stand in remediation, speech technology, psy- relationships of contrast with one cholinguistics and psychology more another in that commutation (switch- generally. The phoneme is also argued ing) of one phoneme for another in a to be a structural element of signed given phonological frame through the languages. creation of a ‘minimal pair’ (for exam- ple, substituting /p/ for /b/ in ‘big’ to Primary works produce ‘pig’) brings about a change Jones, Daniel (1957). The History and in the words meaning. Phonemes are Meaning of the Term ‘Phoneme’. therefore said to be in ‘parallel distri- London: International Phonetic Associ- bution’. Their allophones, which are ation. contextually determined, are said to Jones, Daniel (1967). The Phoneme: Its be in ‘complementary distribution’. Nature and Use. Third edition. London: In The History and Meaning of the W. Heffer & Sons. Term ‘Phoneme’ (1957), Jones attrib- Swadesh, Morris (1934). ‘The phonemic utes the simultaneous ‘discovery’ of principle’, Language Vol. 10: 117–29. the phoneme to the late nineteenth and early twentieth-century linguists Further reading Baudouin de Courtenay and Henry Anderson, Stephen R. (1985). Phonology Sweet, although elsewhere (Jones in the Twentieth Century: Theories of 1967) he cites J. R. Firth’s contention Rules and Theories of Representations. that the term ‘phoneme’ as distinct Chicago: University of Chicago Press. from ‘phone’ (that is, an actual speech Goldsmith, John (ed.) (1996). The Hand- sound) was coined in 1879 by Nikolaj book of Phonological Theory. Oxford: Kruszewski, one of Baudouin de Blackwell. Courtenay’s students. Jones points to various kinds of evidence indicating Dominic Watt that all spoken language users have intuitions about phonemic units, and cites the development of alphabetic writing systems as a good example of POLITENESS evidence of the ‘natural sense’ in which native speakers consider The formulation of linguistic utter- sounds to ‘have a kind of sameness’, ances in such a way that they con- even when their phonetic forms differ tribute to good social relationships from each other. between participants in an exchange. Although the phoneme was super- Since their inception in the 1970s, seded in the 1980s by non-linear politeness studies have developed into models of phonological representa- one of the most active research areas tion, its existence still appears to be in pragmatics and sociolinguistics,

157 POLITENESS with several studies focusing on the endearment terms and in-group lan- realisation of politeness in different guage), and negative face satisfied cultures. through avoidance of contact (as sig- nalled through ‘for example, ques- See also: Implicature; Speech Act tioning and hedging). These two Theory; Conversation Analysis aspects of face allow for a four-way Key Thinkers: Grice, H. P.; Searle, classification of speech acts into acts John that inherently threaten the speaker’s The importance of politeness as a or the hearer’s positive or negative factor motivating particular turns of face. Prior to performing an act x that phrase was acknowledged early on by inherently threatens face, known as a linguists such as Charles Bally and face threatening act or FTA, the Otto Jespersen. Later, it was also speaker assesses the weightiness of acknowledged by H. P. Grice (1967) the threat using the formula and John Searle (1975), who associ- Wx=D(S,H)+P(H,S)+Rx. In this for- ated it with departures from rational mula, D(S,H) stands for the distance efficiency and with indirectness, between the speaker and the hearer, a respectively. However, it was not until symmetric measure of familiarity or the 1970s with Robin Lakoff’s work similarity between them; P(H,S) is the on the rules of politeness (1973) and, power of the hearer over the speaker, most notably, the publication of Pene- an asymmetric measure of the amount lope Brown and Stephen Levinson’s of control the hearer can exercise over essay ‘Politeness: Universals in Lan- the speaker; and Rx is the ranking, or guage Usage’ (1978) that the inte - seriousness, of the imposition entailed gration of politeness into existing by x in the culture in question. The theoretical frameworks became a seri- sum of these three sociological values ous theoretical concern. A major con- guides the speaker’s choice among five tribution of Brown and Levinson lay strategies for the realisation of FTAs, in proposing the notion of face, ranging from ‘Bald on record’ (Wx is inspired by Ervine Goffman’s work on negligible), to ‘Don’t do the FTA’ (Wx ritual aspects of everyday exchanges, is very high) (see Figure 4). The bulk as the unifying principle underlying of Brown and Levinson’s essay is ded- the expression of politeness through icated to cataloguing the linguistic language. expressions that may realise the first Brown and Levinson distinguish four strategies (since the last one two aspects of face: positive face results in silence), which they illus- refers to the desire to be liked and trate with examples from English, approved of, while negative face cor- Tamil (a Dravidian language) and responds to the wish for privacy and Tzeltal (a Mayan language). freedom from imposition. The two It is hard to overestimate the impact aspects of face may thus be viewed as of Brown and Levinson’s model to pulling in opposite directions, with field of politeness studies. Integrat- positive face promoting sociability ing politeness with anthropological and contact with others (as signalled, notions such as face, as well as the the- for example, through, the use of oretical frameworks of implicature*

158 POLITENESS

Lesser

1. without redressive action, baldly

on record 2. Positive politeness

with redressive action Do the FTA

4. Off record 3. Negative politeness

5. Don’t do the FTA Estimation of risk of face loss face of risk of Estimation

Greater Figure 4 Strategies for performing FTAs (Brown and Levinson 1987: 60)

and Speech Act Theory*, their work and surrounding social conventions has been instrumental in shaping our (an aspect of ‘discernment’) with con- thinking about how human dyadic comitant over-emphasising of individ- relationships are reflected in, and con- ual rationality (also termed ‘’); stituted through, language. Moreover, and the universal applicability of their their comprehensive analysis of field- claims. work data has inspired a multitude of More recently, and in tandem with articles and books exploring politeness the challenging of Gricean and Sear- phenomena in a variety of languages lean accounts of meaning as not and cultures, often using the data col- flexible enough to account for the lection method of Discourse Comple- ongoing co-construction of meaning tion Tests (DCTs). Nevertheless, these by participants in interaction, a new works have also produced criticisms set of concerns focusing on social and refinements of Brown and Levin- theoretic aspects of politeness and son’s model. Some of the most impor- favouring a more holistic approach to tant ones concern the definition of face politeness phenomena have emerged. and prioritising of negative over posi- Responding to a paradigm shift tive aspects; their focus on face threat- within politeness studies, recent stud- ening acts to the exclusion of face ies distinguish between first-order enhancing/boosting acts; the defini- politeness (Politeness1), correspon- tions and adequacy of the three socio- ding to participants’ own definitions logical variables and their assumed and perceptions of politeness in inter- independence; the association of action, and second-order politeness politeness with degree of indirectness; (Politeness2), corresponding to the the nature and content of politeness technical definition of politeness implicatures; the (inscrutable) role of by researchers, building on, for exam- silence in their model; the little atten- ple, anthropological notions such tion paid to the role of the audience as face. Although researchers do not

159 POLITENESS necessarily agree on which of the two this debate. In other words, the fre- should be the focus of scholarly quent, contextually-conditioned asso- analysis, the distinction between ciation of a particular perlocutionary Politeness1 and Politeness2 has sev- effect with a particular expression in eral important consequences, includ- corpus data, if such association can be ing the possibility of finer gradation established, provides an empirically between behaviour that is merely ade- verifiable baseline as to what consti- quate (now termed ‘politic’), and tutes the ‘unmarked’ case, and an behaviour that goes beyond that (now analyst-independent vantage point termed ‘polite’), with behaviour that from which to locate and analyse var- is inappropriate falling at either end ious cases as ‘marked’. of a continuum that ranges from Politeness is neither inherent in lin- ‘over-polite’ to ‘impolite.’ In this way, guistic forms in isolation from their the question of the scope of a theory context of utterance (it is not a matter of politeness is also raised, with impo- of structure), nor does it reside liteness/rudeness increasingly attract- (wholly) in the speaker’s intention, ing scholarly attention. independently of its recognition by the At the level of methodology, the hearer (it is not a matter of agency). focus has shifted from isolated utter- Rather, politeness may be viewed as ances to longer chunks of discourse, the mutual constitution of face by par- which are often analysed using con- ticipants in an exchange, which is versation-analytic tools. Rather than greatly facilitated by their partaking aiming at providing an inventory of of similar societal norms – what may devices, either lexical or structural, by be described in terms drawing on which politeness is expressed, as Bourdieu as their having developed in earlier studies, emphasis is now ‘homologous habitus’. Approaches to placed on the utterance situation as a politeness have thus increasingly whole, including the addressee’s turned to social-theoretic notions, reception of the speaker’s utterance, such as habitus and communities of prosodic aspects, and any paralinguis- practice, to analyse the contribution tic cues available. The use of recorded of language in constructing, maintain- conversational data is paramount in ing and endangering good social rela- this respect. Further to the qualitative tionships. analysis of these data, the value of their quantitative analysis is also Primary sources increasingly acknowledged. Such Brown, Penelope and Stephen Levinson quantitative analysis presupposes the (1978). ‘Politeness: universals in lan- availability of large conversational guage usage’. In Goody, E. (ed.), Ques- corpora*, in which regularities of tions and Politeness: Strategies in Social usage (or norms) may be investigated. Interaction. Cambridge: Cambridge While the existence of such norms, University Press. 56–324. Reprinted as and hence their value as analytic tools, Brown, Penelope and Stephen Levinson, have been brought into question, (1987) Politeness: Some Universals in quantitative analysis of conversa- Language Usage. Cambridge: Cam- tional corpora may hold the key to bridge University Press.

160 Lakoff, Robin (1973). ‘The logic of polite- POLITICAL ness; or minding your p’s and qs’. In Papers from the Ninth Regional CORRECTNESS Meeting of the Chicago Linguistic Soci- ety. Chicago: Chicago Linguistic Soci- In its most general sense the term ety. 292–305. ‘political correctness’ refers to an indi- Leech, Geoffrey (1983). Principles of vidual’s or group’s conscious avoid- Pragmatics. London: Longman. ance of linguistic terms associated principally with race, ethnicity, Further reading nationality, gender, religious belief Bousfield, Derek and Miriam Locher (eds) system and sexual orientation that are (2007). Impoliteness in Language. perceived to be pejorative or at any Berlin: Mouton de Gruyter. rate to have the potential to cause Eelen, Gino (2001). A Critique of Polite- offence to others. A more specialised ness Theories. Manchester: St Jerome. sense refers to an uncritical adherence Grice, Herbert Paul (1967). ‘Logic and to and advocacy of a particular polit- conversation’. William James Lectures, ical credo, such as Marxism. Harvard University typescript. In P. Cole See also: Conversation Analysis; and J. Morgan (eds) (1975), Syntax and (Critical) Discourse Analysis; Semantics. Vol. III: Speech Acts. New Feminism; Metaphor; Names; York: Academic Press. 41–58. Reprinted Politeness in Grice, H. P. (1989), Studies in the Way Key Thinkers: Cameron, Deborah; of Words. Cambridge, MA: Harvard Milroy, Lesley; Tannen, Deborah; University Press. 22–40. Whorf, Benjamin Lee Kerbrat-Orecchioni, Catherine (2005). ‘La Politesse dans le discours en interac- The term ‘political correctness’ origi- tion’. In C. Kerbrat-Orecchioni. Le Dis- nated on American campuses among cours en Interaction. Paris: Colin. New Left activists in the 1960s and 187–284. 1970s, when it was used primarily in Searle, John (1975). ‘Indirect speech acts’. connection with feminist criticisms of In P. Cole and J. Morgan (eds.), syntax perceived sexism in language. Nowa- and Semantics. Vol. III: Speech Acts. days the term is used mainly by critics New York: Academic Press. 59–82. of proposed linguistic and social Terkourafi, Marina (2005). ‘Beyond the reforms and is therefore almost exclu- micro-level in politeness research’, Jour- sively pejorative. Opponents of politi- nal of Politeness Research Vol. 1:2, cal correctness claim that is is overly 237–62. concerned with trivial linguistic pre- Watts, Richard (2003). Politeness. Cam- scriptivism. Those on the other side of bridge: Cambridge University Press. the debate draw on a form of ‘weak Watts, Richard, Sachiko Ide and Konrad Whorfianism’ to claim that language Ehlich (eds) (2005). Politeness in Lan- influences perception, therefore that guage. Second edition. Berlin: Mouton linguistic change is a necessary pre- de Gruyter. cursor to changing social attitudes. While practically all language users Marina Terkourafi observe the conventions prohibiting

161 POLITICAL CORRECTNESS the public use of taboo forms, the forms of legislation seeking to pro- degree of sanction attached to individ- scribe the public use of ‘offensive’ lan- ual words or phrases in the social guage. Since 2000 extensions to domains mentioned above is highly existing linguistic legislation have fluid, such that terms considered been made in the United Kingdom, for acceptable at a particular place or time example, ostensibly in the interests of may quickly become stigmatised if protecting religious minorities from they are considered insufficiently persecution in the aftermath of the politically correct or ‘PC’ by one or events of 11 September 2001. The more politically or economically influ- Racial and Religious Hatred Act 2006 ential groups. Awareness of the often specifies that it will be an offence to serious consequences of deliberately incite (or ‘stir up’) hatred against a or accidentally flouting the conven- person on the grounds of his or her tions of politically correct language religion, by prohibiting ‘threatening have strongly influenced linguistic words’ and ‘the display of any written habits at almost all levels of society in material which is threatening’ if the English-speaking countries, perhaps intention by so doing is to stir up reli- most particularly in the United States, gious hatred. where a high degree of sensitivity to Resistance to the introduction of certain terms perceived to betray racist such measures takes several forms. attitudes has resulted in the sacking of Key among these are the holding up of several high-profile public figures. For PC language to public ridicule, for example, the use of the word ‘coon’ by example James Finn Garner’s series of American radio talk-show host Dave ‘politically correct stories’, and the Lenihan in an item about US Secretary appropriation (and thereby subver- of State Condoleezza Rice broadcast sion) of ‘non-PC’ terms such as in 2006 was apparently accidental (he ‘queen’ and ‘nigger’ by the gay and the claims to have intended to say ‘coup’), black communities, respectively. while the previous year Las Vegas TV weatherman Rob Blair’s alleged use of Primary works the same form, supposedly substituted Cameron, Deborah (1995). Verbal for ‘King’ in the phrase ‘Martin Luther Hygiene. London: Routledge. King Junior Day’, could plausibly Dunant, Sarah (1995). The War of the have resulted from an anticipatory Words: The Political Correctness speech . Both Lenihan and Blair Debate. London: Virago. were nonetheless dismissed from their posts almost immediately. In 1999 Further reading vociferous objections to the contextu- Fairclough, Norman (2003). ‘Political cor- ally valid use in a private meeting of rectness: the politics of culture and lan- the word ‘niggardly’ by David guage’. Discourse and Society Vol. 14:1, Howard, an aide to the mayor 17–28. of Washington DC, led to Howard’s Holborow, Marnie (1999). The Politics of resignation. English. London: Sage. Political correctness is tightly bound to , and to other Dominic Watt

162 PORT-ROYAL LOGIC modifications from the scholastic studies. Propositions* are understood to be complexes formed by acts of will Colloquial name for La Logique ou (judgement) that combine a subject l’Art de Penser (Logic or the Art of idea with a predicate idea, with the Thinking) written by Antoine Arnauld verb carrying affirmative force; a neg- and Pierre Nicole but published ative judgement is the converse, where anonymously in 1662 (the fifth edi- the predicate is separated from the tion was published in 1683). The subject. The Logic falls prey here to Logic elaborated a traditional syllo- not being able to distinguish a propo- gistic logic wedded to a Cartesian sition from an attitude towards it. The epistemology and metaphysics. The reasoning, however, brings into relief work also contains many insights in the ‘unity problem’ that was to bedevil natural language syntax that were not Bertrand Russell. surpassed until the work of Noam The Logic also distinguishes Chomsky in the mid-twentieth cen- between explicating and determining tury. subordinate propositions (relative clauses). For example, ‘The invisible See also: Analytic/Synthetic; God created the visible universe’ can Empiricism/Rationalism; Logic; be analysed as ‘The God who is invis- Propositional Attitudes; ible created the universe which is visi- Transformational-Generative ble’, where the first relative clause Grammar explicates its subject, and the second Key Thinkers: Aristotle; Arnauld, further determines its subject. This Antoine; Chomsky, Noam; development was perhaps the first Descartes, René; Frege, Gottlob; formal account of the analytic/syn- Russell, Bertrand thetic* distinction. With its companion, Grammaire The Logic fell into neglect after the Générale et Raisonnée de Port-Royal rise of modern function-argument (published by Arnauld and Claude logic and the prevailing behav- Lancelot in 1660), the Logic seeks iourism* empiricism* in the first part to show that the new Cartesian of the twentieth century. In the mid- philosophy of ideas provides a gen- 1960s, however, Chomsky argued eral account of judgement and rea- that the Port-Royal approach was the soning and a sound basis for a high point of a Cartesian first ‘cogni- demarcation of ‘good’ and ‘bad’ tive revolution’, in its search for arguments. It further provides many underlying mental structures behind detailed analyses of how linguistic the surface of language. Although ‘surface forms’ systematically mis- Chomsky’s historical remarks have match the complex combination of been broadly misconstrued, they did ideas they express. stimulate fresh interest in Port-Royal Following Aristotle, the Logic and both the Logic and Grammar are categorises judgements in terms of now recognised as the forerunners of quantity and quality, and adopts much that is characteristic in modern the standard syllogistic forms with linguistic analysis.

163 POSSIBLE WORD SEMANTICS

Primary sources Key Thinkers: Carnap, Rudolf; Arnauld, A. and P. Nicole (1996). Logic or Kripke, Saul; Leibniz, Gottfried the Art of Thinking. Trans. and ed. J. V. Wilhelm; Lewis, David; Montague, Buroker. Cambridge: Cambridge Uni- Richard versity Press. From a common-sense point of view Arnauld, A. and C. Lancelot (1975). Gen- the idea of possible worlds may seem eral and Rational Grammar The Port- at odds with normal intuitions since Royal Grammar. Trans. and ed. J. one may think the world is what it is; Rieux and B. E. Rollin. The Hague: what it might have been instead is just Mouton. a matter of speculation. But Leibniz saw the world as a manifestation of Further reading God’s will and in God’s mind there are Chomsky, N. (1966). Cartesian Linguis- an infinite number of possible worlds, tics. New York: Harper and Row. from which God has chosen this one. Since God is both benevolent and John Collins omnipotent He has chosen the best of all the possible worlds. When taking up the idea in a modern context, the POSSIBLE WORLD first kind of concern is what is under- stood by the concept ‘possible world’ SEMANTICS if it is not the notion proposed by Leibniz. It may most easily be cap- A theory of formal and natural lan- tured in the perspective of the choices guage semantics that assumes truth- we make as human beings. Consider conditions to be the meaning of the sentence: linguistic expressions and sentences, and claims that the truth value* of (1) John bought a red car yesterday particular sentences relative either to Whoever John is, he chose to buy a car the actual world or to one or more of and he chose the colour red. Recon- the other possible worlds is what structing the situation in retrospect we makes any unique sentence true or know that he (i) might have chosen not false, thereby establishing its mean- to buy a car and, if anyhow he did buy ing. The idea goes back to Gottfried a car, he (ii) might have chosen another Wilhelm Leibniz and has been elabo- colour. Had he chosen otherwise, the rated by Rudolf Carnap, but the main world would also have been otherwise figures of the modern version, espe- in the respect that John might have no cially associated with modal logic, car or a car with another colour. We were Saul Kripke and David Lewis. can also, in retrospect, construe the Possible world semantics was, to some situation in which John made his extent, utilised by Richard Montague. choices and imagine the scenarios that See also: Connotation/Denotation; were going through his mind at the car Correspondence Theory; Definite dealer: (i) car versus . no car, and (ii) Descriptions; Logic; Modality; red versus other colours. Each sce- Names; Sense/Reference nario represents a minor difference in

164 POSSIBLE WORLD SEMANTICS the way the world may look depend- world – what then is the meaning of a ing on the choices made by John; in the false sentence? Can a lack of corre- history of the universe it may seem spondence be the meaning of a sen- insignificant what John chose to do, tence? Possible world semantics may but in the context of what linguistic not have a clear answer to this. expressions – and especially sen- Another problem is the question tences – mean, it may be crucial. about the relation between words and Taking (1) as an ordinary life utter- the world, in that we may ask: what ance, it is expected to have an inherent world? It seems evident that we can meaning more or less evident to the talk directly about things in our envi- interlocutors engaged in a conversa- ronment because we can observe them tion, and in modern philosophical while talking. In general we also feel semantics this is supposed to be estab- fairly sure that our environment is rel- lished by its truth value, that is, atively stable: for instance, that when whether it is true or false according to we get up in the morning, our home its truth conditions. This in turn is the will look the way it looked when we way the world looks like in the cir- went to bed. We may realise that the cumstances surrounding John’s buying building may have burned down over a car or not. night, but only as a hypothetical pos- So far one might accept the claim sibility. Philosophers have spent much that the meaning of sentences has to time debating the fact that we think do with the relation between the we know, and can talk about, things words uttered and the world, but we do not observe while talking, and there are two problems here. One is basically we cannot be totally sure that it may not be evident how the that we do know. The fact that we meaning of sentences is identical with nevertheless maintain that we actually the question of whether they are true do know is based on our experience or false. Intuitively one may think that and the belief that our experience is in order to decide the truth value (rel- reliable. Therefore one of our linguis- ative to a specific state of affairs) of a tic skills – affiliated with our mental sentence, one has to understand it, faculty and our memory – is the abil- and to understand it is to know its ity of displacement, namely that we meaning. But, it is claimed, to know can talk about things that are not its meaning is to know its truth value present, or even things we have never based on its truth conditions, which it experienced anything about, like follows is the same thing as its mean- Napoleon, the far side of the moon or ing. There seems to be a kind of circu- Jack the Ripper. This, too, is one of the larity predicament as an immanent issues for possible world semantics. In part of truth-value semantics, and this the special and technical framework is one of the problems possible world of formal logic, possible world seman- semantics tries to solve. A further tics will also comprise a kind of modal complication is the following ques- logic, the modality* of which is so- tion: if meaning is truth conditions – called alethic modality, which deals that is, there is a kind of correspon- with propositions expressing what is dence between a sentence and the possible versus what is necessary.

165 POSSIBLE WORLD SEMANTICS

The modern notion of possible and, accordingly, contingent proposi- world semantics has its roots in tions are true at some possible worlds Carnap’s Meaning and Necessity and false at some other possible (1947) in which, as opposed to his ear- worlds. Obviously things might have lier work, he engaged in semantic gone otherwise, and (4) questions, adjusting Gottlob Frege’s basic notions of Sinn (‘sense’) and (4) The present queen of Denmark was Bedeutung (‘meaning’) in order to crowned queen in 1971. make them work in his system of modal logic. In the general modern might have been true if only her father version of possible world semantics – the king had not lived the last few in which the term ‘possible world’ had days of his life. Therefore (4) is a its technical meaning coined by proposition that is possibly true – that Kripke in a paper in 1959 – it works is, in possible world semantics it is as a way of clarifying the meaning of true in some (at least one) possible either normal declarative sentences or worlds. In contrast, (5) is an impossi- sentences containing the words ‘possi- ble proposition: ble’ and ‘necessary’ (or some of the (5) The present queen of Denmark is a morphological derivations of them) or man. containing epistemic modals (verbs in the past tense expressing counterfac- This is true in no possible world and is tuals or unsubstantiated information). therefore false by necessity, while (6) Consider the sentence below: (6) The present queen of Denmark is a (2) The present queen of Denmark was woman. crowned queen in 1972. is a proposition that is true by neces- This proposition is true in the actual sity – that is, it is true at all possible world because her father the king died worlds. When using the word ‘possi- that year. This is a normal declarative bly’, one is able to set up an explicit sentence and its truth value is a matter hypothetical scenario: of contingency, that is, it may be either true or false, and one has to check the (7) The queen of Denmark is possibly evidence in order to decide what the meeting the American president this facts are. One also has to do so when moment. listening to (3) in order to conclude that it is a false contingent proposition: and the same job can be done with an epistemic modal: (3) The present queen of Denmark has an older brother. (8) The queen of Denmark might be meeting the American president this A possible world semantics will tackle moment. these details by saying that (2) is true at the actual world, as mentioned, Both (7) and (8) may be true in some while (3) is false at the actual world, possible worlds and false in others,

166 POSSIBLE WORLD SEMANTICS but the trouble with such sentences is (12) ‘if the “Big Bang” initiated the the fact that their truth value is not universe [antecedent] then it will given by the meanings (reference of expand forever [consequent]’ the subject and extension of the predicate) of their parts, as is the and adding the alethic expression ‘it is case with the previously mentioned possible that’ as a prefix. This can be subject-predicate sentences. Hence formalised, first, as (7) and (8) cannot be used as truth- functional conjuncts in the same way. (13) p → q This other kind of logic may be expressed in a formal and symbolised where p is the antecedent and q the language (so-called alethic formal consequent in (12) and, second, as logic) using special logical operators. Thus in (9) (14) ◊( p → q) = ‘it is possible that (12)’.

(9) ᮀp = ‘it is necessary that p’ In possible world semantics this would be equivalent to ‘ “if the ‘Big the ᮀ is an operator signifying Bang’ initiated the universe then it will the necessity of the content of the expand forever” is true at some (at proposition ‘p’. If p is supposed to be least one) possible worlds’. On this (6) ‘The present queen of Denmark is basis one is able to combine more a woman’ then p is true – that is, (6) is complex formulae of deduction. true at all possible worlds. However, if Whether modal alethic logic is the p is, for instance, (5) ‘The present ideal (for) modal logic or just some queen of Denmark is a man’ and subtype of it is a debated issue among somebody claims that philosophers, but some common ground seems to be that there are (10) ᭛p = ‘it is possible that p’ certain relations between possible worlds. Sometimes this is referred to using ᭛ as a possibility operator, then as accessibility or alternativeness (10) is false by necessity because the between worlds (see McCawley 1981: sentence p is true in no possible world. 276), figuratively how you can ‘get’ By combinations of these operators from, say, w1 to w2, irrespective of and connectors from traditional logic, the stipulated ontological status of the one is able to construct larger formal worlds. That means that the tradi- systems (languages) with compound tional mathematical principles of propositions. Using the letters w, v, u reflexivity, symmetry and transivity as variables for possible worlds, one (McCawley 1982: 276) apply to the may state the propositions: formulae produced by systems of formal modal (alethic) logic. A further (11) ‘the ‘Big Bang’ initiated the uni- extension of such systems is the pro- verse’ and ‘the universe will duction of predicate logic systems for expand forever’ modal logic (see McCawley 1982: 285–96). Lewis (1973) takes up the putting them together in question of relations between possible

167 POSSIBLE WORLD SEMANTICS world in the context of counterfactu- designators. He developed an idea als, that is what is not true at the now called the causal theory of refer- actual world, and he offers a clarify- ence, and he has profoundly changed ing account of these relations in that the common view on the basic logical he proposes not an abstract distance – notions of a priori and a posteriori rel- illustrated by closeness in numerical ative to analytical and synthetic measure – but a three-place predicate truths. But in the context of modal notation expressing the relative close- logic his major contribution is the ness between three possible worlds solutions he offers to the problems (McCawley 1982: 312). concerning modal logic, possible Lewis also defends the idea that world semantics and the metaphysical possible worlds do in fact exist as consequences of a number of princi- entities in some way or other. This is ples appealed to by logicians. This has called absolute realism (Schurz 2006: to do with, for instance, the above- 443) and is a subtype of realism, a mentioned realism as one kind of metaphysical claim that holds that assertion about whether possible universals (what can be used as pred- things exist or not; another view is icates in propositions about more , which says that everything than one entity) are real. An alterna- that exists is what is actual – that is, tive view is reductive realism which what we can say there is a fairly clear says that possible worlds do exist but consensus about in the real world. can be reduced to entities that are Now, the problem is that if one is an more familiar, and this view is the actualist and accepts the formal con- most common among scholars in the sequences of modal logic in what is field. A third position is anti-realism, called a simplest quantified version, and from the meaning of this label one also has to accept the unattractive one may infer that it denies both outcome that possible things do exist. the realistic stance (that universals Kripke’s solution to the problem exist) and, accordingly, that possi- requires a rather technical account ble worlds exist and that possible- but it has to do with the way he han- world-sentences are meaningful. In dles the semantics (specifically the this respect Kripke conceived of a domains) of individual terms in modal possible world as ‘a counterfactual propositions, a theoretical suggestion course of history’ (Kripke 1981: 6), called ‘Kripke Models’. An unpreten- and, in a broader perspective, one tious interpretation might say that he may set up a quite impressive number claimed that possible things can be of combinations of different kinds of said to exist but only, so to speak, in existence of possible worlds, maybe their own worlds, from which it fol- ending with the mundane view that lows that propositions about them they are just useful fictions used in put forward in the actual world end logical reasoning. up being false. In this way the prob- Kripke made a number of contribu- lem with their truth value, that they tions to philosophical logic, notably seem to be true also in the actual to the theory of the semantics of world (which has some unpleasant names* – which he called rigid metaphysical consequences), has been

168 POSSIBLE WORLD SEMANTICS

dissolved. Kripke’s solution is not Kripke, S. (1981). Naming and Necessity. undisputed but his intellectual Oxford: Blackwell. achievements in the field are widely Lewis, D. K. (1973). Counterfactuals. acknowledged. Oxford: Blackwell. Possible world semantics presents McCawley, J. (1981). Everything That the linguist and the philosopher with Linguists Have Always Wanted to a number of suggestions that may Know About Logic, But Were Ashamed solve some traditional problems. For to Ask. Oxford: Blackwell. 273–359. instance, the apparent problem of cir- Menzel, C. (1990). ‘Actualism, ontologi- cularity in a truth-conditional seman- cal commitment, and possible world tics for linguistic expressions may be semantics’. 85: 355–89. solved by rephrasing the question Schurz, G. (2006). ‘Alethic modal logics from asking about relations between and semantics’. In D. Jaquette (ed.), A meanings of sentences in this world Companion to Philosophical Logic. to asking about relations between Oxford: Blackwell. 442–77. different worlds, called possible worlds, as meanings of sentences. Further reading This may also solve the question Benthem, J. van et al. (2006). ‘Part IX. about the meaning of a false sentence Modal logic and semantics’. In D. in that a false sentence may be true in Jaquette (ed.), A Companion to Philo- some possible world, which thereby sophical Logic. Oxford: Blackwell. presents an intuitively acceptable 389–509. positive correspondence between the Dancy, J. (1985). Introduction to Contem- sentence and that particular world. poraryEpistemology.Oxford:Blackwell. However, this leaves open the ques- Groenendijk J. and M. Stokhof (2002). tion of whether this will make us ‘Type-shifting rules and the semantics of understand what a false sentence ’. In P. Portner and B. H. means in our world. It also does not Partee (eds), Formal Semantics. Oxford: dissolve the relation between sen- Blackwell. tence, meaning and possible world; it Kratzer, A. (2002). ‘The notational cate- does not solve the correspondence gory of modality’. In P. Portner and B. problem. It would not be unfair to H. Partee (eds), Formal Semantics. say that possible world semantics Oxford: Blackwell. chooses to ignore such problems con- Martin, R. M. (1987). The Meaning of centrating on constructing formal Language. Cambridge, MA: MIT Press. systems that can handle traditional Robertson, T. (1998). ‘Possibilities and the technical problems in other formal arguments for origin essentialism’. systems. Mind 107. 428: 729–49. Saeed, J. I. (2003). Semantics. Second edi- Primary sources tion. Oxford: Blackwell. Carnap, R. (1947). Meaning and Neces- Stalnaker, R. C. (2002). ‘Assertion’. In P. sity. Chicago: Chicago University Press. Portner and B. H. Partee (eds), Formal Kripke, S. (1959). ‘A completeness theo- Semantics. Oxford: Blackwell. rem in modal logic’. Journal of Sym- bolic Logic 24: 1–14. Hans Götzsche

169 POSTSTRUCTURALISM entities themselves are not positively defined, but identified at best as mere place-holders. This has the conse- Poststructuralism is an outgrowth of quence that all structures are by defi- structuralism* It is a method of analy- nition hermetically closed unto sis which had a tremendous impact in themselves. It is only on this condition the last decades of the twentieth cen- that a structure can be regarded as tury on work done in the fields of lin- composed of pure negativities. To put guistics, anthropology, psychology, it differently, the ‘structurality’ of a literary criticism and elsewhere. given structure is conditional upon the Arguably, it remains a potent force to entire space having been taken up by reckon with in the early twenty-first binary relations: the forceful removal century, though it has been at the of any entity will automatically trigger centre of intense and often acrimo- a rearrangement of the remaining nious debates. entities. This in turn means that all structures are fully integrated, each See also: Signs and Semiotics; with respect to itself, and autonomous Structuralism; Deconstruction with respect to other structures. Key Thinkers: Bourdieu, Pierre; Structuralism was a huge success as Derrida, Jacques; Lacan, Jacques; an idea in the early decades of the Saussure, Ferdinand de twentieth century and it held sway Like so many other terms that take the over several academic disciplines well prefix ‘post-’, poststructuralism is a into the second half of that century. It complex notion. Part of the difficulty revolutionised linguistics and set it in getting to grips with the term has to apart from philology and historical do with just how one is supposed to linguistics, which dominated the disci- view its relation to structuralism. In pline in the nineteenth century. In fact one sense, the term does imply that it the neat separation between syn- regards structuralism as a thing of the chrony and diachrony that Ferdinand past (as in ‘post-World War Two de Saussure advocated was but a period’). But equally it also captures straightforward consequence of the the idea that it is an offshoot of struc- requirement that a structure, in order turalism or, alternatively, it is a move- to function the way it was required to, ment that draws its strength from the had to be closed unto itself. But it soon legacy of structuralism (as in ‘post- became clear that this was at odds graduate student’). with the further requirement that a In other words, to understand what structure also had to be resilient, so as poststructuralism is about, one needs to permit ‘the rearrangement of its to have some idea of what constitutes internal units’ in order to accommo- structuralism. Structuralism is a mode date eventual structural changes. of inquiry which looks at the phe- From a structuralist point of view, all nomena under its scrutiny as made up changes were, so to speak, sudden and solely of the relations among the enti- cataclysmic and, while keeping the ties in question, rather than those enti- overall structure intact, would result ties themselves. In other words, the in an internal rearrangement of the

170 POSTSTRUCTURALISM network of relations within the struc- to incorporate into structuralism ture. History, in this world-view, pro- insights from Marxism and Lacanian gressed by fits and starts, rather than psychoanalysis. The speaking subject in a smooth and gradual continuum. was unceremoniously dislodged from In the philosophy of science, Thomas the high pedestal where he/she had Kuhn famously advocated a view of remained safely and majestically the progress of science which had all ensconced. The subject was decentred the trappings of structuralism in this and demoted to the status of a mari- sense. onette. Marxist structuralists like As already noted, the twin require- Louis Althusser advocated this second ments of the closure and the resilience alternative, on the strength of the con- of a structure are mutually incompat- viction that the ‘mature Marx’ had ible. In a ground-breaking essay called made a crucial break with his own ear- ‘Structure, Sign, and Play in the Dis- lier Romantic , whereby course of the Human Sciences’, what would be left at the centre is an Jacques Derrida (1966) drew atten- illusory subject, seeming to be in con- tion to this irreconcilable incompati- trol of him/herself and his/her mean- bility and argued that the only way ings, while actually being subjected or structuralism could be salvaged is by subjugated by the structure. admitting that a structure can have Derrida contended that both of the only an imaginary centre. ‘The centre options signalled by structuralism cannot be the centre’ concluded the result from a misunderstanding of the French philosopher in what is often function of the absent centre. The cited as one of his most enigmatic structure’s absent centre functions statements. Now, this imaginary rather like the black hole at the centre centre does indeed appear to be either of a galaxy (a structure like any controlling or, contrariwise, being other). It is here that Derrida recog- controlled by the structure. nises the effects of history within the The first alternative results in the structure, traditionally thought to positing of an all-powerful subject, offer no room for origin or history Cartesian in its lineage, whose inten- (whereof the separation of synchrony tions underpin meanings and prevent from diachrony). What this means is them from going astray or floating that the subject can be neither Carte- around freely. Humanism, alongside sian (all-knowing, intending) nor a its offshoots existentialism and phe- being at the mercy of forces beyond nomenology, provides the necessary his/her control. The structuralist sub- philosophical backdrop against which ject, in other words, must give way to such a subject could be envisaged. a poststructuralist agent who must act Intentionality* (‘aboutness’) would be in history (not independently of it) the hallmark of the subject of lan- while, no doubt, constrained by the guage. He/she would constitute the structure around him/her. fulcrum around which language turns. This then can be pointed out as the In the late 1960s and early 1970s, hallmark of poststructuralism: the however, there was a concerted emergence of the historical subject move among French intellectuals within the very entrails of a structure.

171 POSTSTRUCTURALISM

This new subject is one who has been notion of ‘habitus’. In Bourdieu’s empowered to act on his/her own and view, social agents develop, over a is endowed with agency. Once the period of time, a ‘feel for the game’ presence of an agent who is in a posi- which is a kind of eminently practical tion to subvert , and bodily knowledge. thwarting it from within, is recog- Poststructuralism is closely tied to nised, it is but a short step to reject the and often confused with postmod- existence of all pre-ordained, founda- ernism. While the two indeed share tionalist, essentialist and totalising many common concerns, it is impor- conceptual schemes. tant to bear in mind that the latter is played an impor- concerned to critique, in a way the tant role in the development of post- former is not, the Enlightenment proj- structuralism. Having been a die-hard ect and the sheer arrogance of those structuralist himself, Foucault grew who claimed to be fighting the battle increasingly discontented with two of of Reason against Irrationality. structuralism’s key assumptions. On Poststructuralism, alongside post- the one hand, while recognising the modernism, opens up an entirely pervasiveness of structures in many of new set of possibilities for thinking human activities, Foucault became about ethical issues. With the meta- sceptical of the idea that structures narratives of foundationalist, essen- invariably control and regulate the tialist and totalising discourse human condition. On the other completely discredited, poststruc- hand, he also came to recognise the turalism leaves us no option but that inevitable ‘situatedness’ of our gaze of regarding the path of ethics as an and the impossibility of ever attaining exercise in tight-rope walking, with a transcendental standpoint from no safety net in case a false step is which to contemplate things. taken. In other words, it foregrounds Freeing the subject of language the question of personal responsibil- from the shackles of structurally ity. In the United Kingdom, a special- imposed subjugation is key to a poli- ist group called ‘Post-Structuralism tics of identity. But then the escape and Radical Politics’ has been actively from the prison-house of structure engaged in promoting discussion on should not herald a return to liberal the political implications of poststruc- individualism whereby the individ- turalism in all walks of life. ual determines his/her own through the exercise of a series of Primary sources rational choices. This is the central Belsey, Catherine (2002). Poststructural- thrust of and moving impulse behind ism: A Very Short Introduction. the work of Pierre Bourdieu. In his Oxford: Oxford University Press. book Outline of a Theory of Practice Bourdieu, Pierre (1977). Outline of a (1977) Bourdieu made a proposal for Theory of Practice. London: Cambridge avoiding both the Scylla of the total University Press. subjugation of the subject of language Derrida, Jacques (1966). ‘Structure, sign, and the Charybdis of navel-gazing and play in the discourse of the human individualism, by putting forward the sciences’. In Jacques Derrida (2001),

172 PRESUPPOSITION

Writing and Difference. London: Rout- See also: Connotation/Denotation; ledge. Definite Descriptions; Implicature; Sarup, Madan (1993). An Introductory Names; Possible World Semantics; Guide to Poststructuralism and Post- Sense/Reference; Speech Act modernism. Second edition. Athens, Theory; Truth Value GA: University of Georgia Press. Key Thinkers: Frege, Gottlob; Sturrock, John (1979). Structuralism and Grice, H. P.; Russell, Bertrand; Since. London: Oxford University Press. Strawson, P. F.; Montague, West, David. (1996). An Introduction to Richard Continental Philosophy. Cambridge: Polity Press. The notion of presupposition as a the- Williams, James (2006). Understanding oretical concept in linguistics and the Poststructuralism. Canada: McGill- philosophy of language tries to solve Queen’s University Press. the problem that in discourse utter- ances may be articulated and fully Further reading understood by the interlocutors, while Valentine, Jeremy and Alan Finlayson a detached and uninitiated observer (eds) (2002). Poststructuralism and Pol- may find it hard to understand the itics: An Introduction. Edinburgh: Edin- meaning of what is being said. Imag- burgh University Press. ine I go to the supermarket together with my wife and, seeing among the Kanavillil Rajagopalan amount of goods a certain kind of soup in small plastic bags, I say ‘Look, there is this special Italian soup in small plastic bags’, and my wife PRESUPPOSITION responds ‘My Italian language course has no more classes’. The background A type of extra and assumed meaning knowledge assumed by both of us is attached to the basic meaning of utter- that I do not cook and therefore I just ances or sentences. In a broad sense warm up some prefabricated meal for presupposition can be defined as what us when she comes home from her must by necessity be assumed to be the Italian classes. The line of reasoning is case in order to interpret a short that when she has no more Italian ordered sequence of linguistic expres- classes, I will not have to do this any sions in a meaningful way. The idea more and therefore we will not have was introduced by Gotlob Frege who to buy more plastic bag soup. Accord- said that (in German: ing to H. P. Grice, this instance of Voraussetzung ‘(pre)condition’) are background knowledge and tacit line particular conditions that have to be of reasoning may be called a kind satisfied for single linguistic expres- of (conversational) implicature*, or, sions to have a denotation (Beaver in later developments, presumptive 1996), but twentieth-century linguis- meaning or preferred interpretation tics and philosophy has focused more (Levinson 2000), but it might also be on the presuppositional meaning asso- considered a kind of presupposition. ciated with sentences and utterances. Which option is chosen depends on

173 PRESUPPOSITION whether one sees presupposition as a heated discussion and controversy semantic or as a pragmatic function among philosophers and linguists, and furthermore on whether one’s and while later on Strawson modified concern is natural languages or formal his views, many twentieth-and twenty- languages. Frege’s main objective was first-century scholars prefer not to to nurture the technical languages of commit themselves to any of the science in the form of formal logic* standpoints. and mathematics, and the specific lin- However, the philosophical impli- guistic functions of these languages cations of this issue are by no means that now are regarded as presupposi- trivial since, on the one hand, they tional functions were the main interest deal with some aspects of the basic of the central figures Bertrand Russell problem of how linguistic expressions and P. F. Strawson. can be used to refer to things in the Working with his famous, and to world and, on the other hand, how we some scholars notorious, example ‘The can express knowledge of the world in present king of France is bald’ (Russell so-called existential sentences like 1905: 483), Russell claims that the ‘there is a [one and only one] queen of proposition does not fail to have a Denmark’, as opposed to so-called truth value* (a point of view held by identity sentences like ‘a queen is a Frege) but has to be broken down into female monarch’ and predications like a number of component propositions ‘the queen of Denmark is intelligent’. each of which has to be true for the One further concern is how we refer compound proposition to be true. to fictional and abstract entities and, Since one of the propositions (‘there in the present context, how we use exists a king of France’) is false, then such kinds of reference in presupposi- the compound proposition is false, and tions. These are, in the end, epistemo- therefore meaningful according to the logical questions, and some of the view that truth conditions are the formal properties of presuppositions meaning of sentences. This view was have some bearing on deeper philo- challenged by Strawson (1950) who sophical contemplations. One such said that the proposition singled out by common property shared by presum- Russell – together with the second ably all theories of presuppositions is component proposition, ‘there is no the way presuppositions work under more than one king of France’ – is not negation. Consider the following asserted but presupposed since the def- sentence: inite description* ‘the [present] king of France’ fails to refer (it did in 1950 and (1) The queen of Denmark is intelligent. has done since then). In this respect Strawson agrees with Frege. Therefore Accepting the convention that (1) can the proposition containing the definite be represented by the symbol q, then description cannot be used to make an it follows that the truth value of q as assertion and accordingly it is mean- well as its negation ~q depends on ingful but has no truth value. The posi- the truth value of the presupposition tions held by Russell and Strawson ‘there is a queen of Denmark’, which respectively have given rise to much can be represented by the symbol p.

174 PRESUPPOSITION

Thus, it seems to be a matter of fact The denial of (4b) automatically that p is the necessary condition of makes (4a) false while the denial of both q and ~q. In this respect a pre- (4a), as in (4c), has no influence on the supposition differs from an entailment truth value of (4b) since the woman in that the same does not seem to may have died from other causes. This hold for entailments. Consider the is opposed to (3a) and (3c) in that if sentence: (3b) is denied: ‘John has not been smoking’, then both (3a) and (3c) (2) Jack the Ripper killed the woman. become false by necessity, because one cannot stop doing something one has It follows from the meaning of the not been doing. This is the basis of sentence that the woman must be what is sometimes called presupposi- dead, and this has to do with the tion failure. Imagine somebody is meaning of the word ‘kill’. The basic being asked the question: meaning of ‘kill’ is that somebody being ‘killed’ must be dead. This is not (5) Have you stopped beating your the case with, for instance, the word dog? ‘shoot’. The sentence ‘the policeman shot the murderer’ may refer to an Questions like these are much debated incident where the murderer survived. in because the respon- Thus entailments are closely con- dent is being trapped: he can neither nected with the lexical meanings of say ‘yes’ or ‘no’, because in both cases the words used (verbs like ‘kill’ and he will confess to the fact that he has ‘shoot’) while presupposition is a rela- actually beaten his dog in the past, tion between sentences and between even though he may not have done so, sentences and the world. This can be and then the only relevant way to illustrated by sets of sentences such as respond is to question the question the following: itself, viz. its presupposition(s). In this case many would feel that the presup- (3a) John has stopped smoking position is disguised by the implicit (3b) John has been smoking meaning of the words. It may seem (3c) John has not stopped smoking more evident that somebody is dead if one has been killed (4a and 4b), and it If John has stopped smoking (3a), may seem even trivial to say of the then he must have been smoking in the sentence ‘the king of France is bald’ past, but the same goes for the case in that the existential sentence ‘there is a which he has not stopped smoking (one and only one) king of France’ is a (3c), and therefore (3b) is a presuppo- presupposition of that sentence, sition of both (3a) and (3c). Not so because if somebody mentions a ‘king with entailments: of France’ then there must be a king of France. (4a) Jack the Ripper killed the woman. Nevertheless this was the subject (4b) The woman is dead. matter of the controversy between (4c) Jack the Ripper did not kill the Russell and Strawson and of the fol- woman. lowing debates. Russell’s interest had

175 PRESUPPOSITION to do with the semantics of what he above, held that propositions picking called ‘definite descriptions’ like ‘the them up as subjects were false. Straw- (red) car’ which he conceived of as son, then, said that Russell had got it what he called ‘names*’ (that is, not all wrong in that he ignored the dis- only proper names in our understand- tinction between ‘a sentence’, ‘the use ing), a stance based on the argument of a sentence’ and ‘an utterance of a mentioned above, viz. that a phrase sentence’. In Strawson’s view, sen- like ‘the (red) car’ formally should be tences have no truth value, only the dissolved into the proposition ‘there use of a sentence can produce a exists a red car’. Accordingly Russell proposition which is either true or claimed that ‘names’, which he false. As for the definite description defined as linguistic expressions being ‘the king of France’ Strawson argued able to be used as logical subjects that ‘To say “The king of France is in propositions, included definite wise” is, in some sense of “imply”, to descriptions, and therefore these imply that there is a king of France’ could be used – as referring expres- (1950 (1971: 12)), and this is the kind sions in propositions – to refer to of implication that gave rise to the things in the world, for instance a red notion of presupposition; the term car. This solved the problem that in was coined by Strawson in his 1952 ordinary predicate logic there are no publication. The outcome was the definite descriptions, only individual position that certain sentences, like or quantitative terms like ‘Aristotle’, ‘the king of France is bald’, are seen as ‘he’ or ‘all dogs’, and accordingly Rus- having unfulfilled presuppositions sell was able to justify predicate sen- and therefore they have no truth tences like ‘the (red) car is rusty’. value, or they have the truth value This has further implications in zero. One way to go around the prob- modern logic but it also presented lem, if one does not feel at ease with Russell with the problem that definite sentences that have the form of pred- descriptions may be empty, at vari- icate propositions but nevertheless ance with our common-sense view seem to lack truth value and seman- that the use of the definite article tics, is to adopt a ‘possible world’ implies that there is some well-known semantics*. In this kind of semantics, entity out there. But as is also well propositions can be seen as functions known, definite descriptions may from the set of all possible worlds to sometimes have no referents – for the truth values ‘true’ or ‘false’ and in instance in fictional and abstract con- this kind of logic, presuppositions texts – that is, no entities that satisfy can be conceived of as fulfilled in their description, such as is the case only a restricted set of worlds. Hence with the definite noun phrase ‘the the propositional functions can be king of France’. From this it follows defined as having value only in the set that propositions containing them of these worlds, that is where the pre- cannot specify truth conditions and suppositions of a proposition are ful- accordingly they have no semantics. filled, a set which is sometimes called According to Frege, they were mean- the domain of that proposition (All- ingless, while Russell, as mentioned wood et al. 1977: 150).

176 PRESUPPOSITION

Apart from these technical and epis- ‘assume’, and it may seem a little odd temological questions, some issues to say that sentences are able to assume concerning presuppositions are on the something, because a common-sense borderline between such basically understanding might claim that only philosophical problems and the daily people can assume. The same question use of language. Some of them deal may arise when considering the theo- with the formulaic or grammati- retical idea that a sentence like ‘the calised expressions associated with king of France is bald’ presupposes the specific speech acts: sentence ‘there exists a king of France’, because it would be counter-intuitive (6) What did Jack the Ripper do? to think otherwise. This has led to the suggestion that there are, at least, two In (6) the wh-expression connected types of presupposition: one is a with the fixed (and a cer- semantic notion dealt with in modern tain prosodic contour) signifies the logic where scholars try to solve the speech act of a question, and it pre- technical problems of formal lan- supposes that Jack the Ripper did guages, and another is a pragmatic something (cf. Lyons 1977: 597). This notion dealing with what people pre- kind of presupposition is not indiffer- suppose in their discourse interactions. ent to negation because the sentence: To contemplate this question is like opening a Pandora’s box in linguistics (7) What did Jack the Ripper not do? and philosophy. As Levinson puts it, ‘there is more literature on presupposi- presupposes that there was something tion than on almost any other topic in that Jack the Ripper did not do. pragmatics (excepting perhaps speech Another kind of presupposition (cf. acts)’ (1983: 167). Lyons 1977: 599) is found in sen- In order to approach the subject we tences expressing so-called proposi- may take the Italian-soup-in-small- tional attitudes*, for example, plastic-bags scenario described above sentences containing expressions like as an example. A main topic here will ‘think’, ‘believe’, ‘realise’: be what are in general called presup- position triggers. A presupposition (8) The police realised that Jack the trigger is one or more words, or Ripper had killed the woman ‘aspects of surface structure in gen- eral’ (Levinson 1983: 179) that gen- In (8) it is a presupposition of the com- erate a relation of presupposition pound sentence that Jack the Ripper between what is actually expressed actually killed the woman and as for and what must be assumed. At face the ‘historical’ Jack the Ripper, of value only the word ‘Italian’ seems to whom little is known, a sentence like be common to the two utterances (8) does not seem justified. thereby suggesting there is a connec- As pointed out by Lyons (1977: tion between them. But this similarity 600), a pre-theoretical notion of pre- may elicit a rather large number of supposition as a verb ‘presuppose’ possible scenarios and it is in fact means almost the same as the word deceptive. My wife and I may have

177 PRESUPPOSITION any kind of fast food after her Italian be interpreted as a ‘No, we should buy classes. If, instead of fantasising, we no more soup’. But this is, in a narrow look at the details of what is being sense, not something presupposed. It said, my utterance was ‘Look, there is may belong to the ‘normal’ state of this special Italian soup in small plas- affairs in certain cultures or subcul- tic bags’, and this may be interpreted tures, but none of the words or other as a man-and-wife convention for the surface aspects can, directly, trigger speech act of a question: ‘Should we such presuppositions. Only knowledge buy some?’ Taken as a question, the about the family and a number of cir- presupposition is ‘either we buy some cumstantial facts may yield the infor- or we don’t buy some’. In this case it mation that this has nothing to do with is the unspoken but intended question a scenario where, for instance, the stu- that triggers the presuppositional rela- dents of my wife’s language class eat tion in that the specific clause surface Italian soup every time they meet. structure of questions – the syntactic However, in a broad sense these inversion and the of ques- inferred assumptions may be called tions produced by native speakers of presuppositions. Some scholars have English – will come to my wife’s mind defended radical theories of pragmatic when I paraphrase my intended ques- presupposition in which the key con- tion as the declarative sentence I actu- cepts are ‘appropriateness’ or ‘felicity’ ally utter. If this reading seems fair (Levinson 1983: 204) and ‘mutual enough, the logical response from my knowledge’ or ‘common ground’, wife should be an answer implying mean ing that a presupposition can be either that we should or we should not ‘appropriately used if it is assumed in buy the soup-in-small-plastic-bags, the context that the propositions indi- and it actually may be so. The sen- cated by the presupposition-triggers tence ‘My Italian language course has are true’ (Levinson 193: 205). The con- no more classes’ may be paraphrased textual prerequisite is furthermore that as ‘My Italian course has stopped’, this is known by the participants, and and the verb ‘stop’ is a so-called consequently what is not mutually change-of-state verb triggering the known by them will not come up as necessary presupposition(s) ‘I have presuppositions because they are inap- attended an Italian language course in propriate. In this framework the chain the past and now I don’t attend it any of inferences assumed in my conversa- more because there are no more tional transactions with my wife on classes’. ‘Italian soup’ may well be called pre- In order to establish a connection suppositions, but one problem is just between my question ‘Do we buy or the criterion about mutual knowledge. not buy?’ and the ‘Italian classes’, one When examining actual discourse it is may speculate what my wife’s absence fairly hard to make justified claims or late home-coming has to do with about what people know – as opposed fast food and one only has to infer that to claims about what they say. in this family fast food is on the menu At the core of theories on presup- when the housewife is away or late, positions there are other difficulties, and that my wife’s response may boldly among them the problem of whether

178 PRIVATE LANGUAGE presuppositions are compositional Strawson, P. F. (1950). ‘On referring’. relative to the complex clauses they Mind 59: 320–44. are presuppositions of (Levinson Strawson, P. F. (1952). Introduction to 1983: 191). There is not a consensus Logical Theory. London: Methuen. among scholars in the field about what constitutes a standard notion of Further reading presupposition in linguistics and the Allwood, J, L.-G. Andersson and Ö. Dahl philosophy of language. (1977). Logic in Linguistics. Cam- bridge: Cambridge University Press. Primary sources Levinson, S. C. (1983). Pragmatics. Cam- Asher, N. and A. Lascarides (1998). ‘The bridge: Cambridge University Press. semantics and pragmatics of presuppo- Levinson, S. C. (2000). Presumptive sitions’. Journal of Semantics 15, 3: Meanings. The Theory of Generalized 239–300. Conversational Implicature. Cam- Beaver, D. I. (1996). ‘Presupposition’. In J. bridge, MA: MIT Press. van Benthem and A. ter Meulen (eds), Lyons, J. (1977). Semantics. Cambridge: Handbook of Logic and Language. Cambridge University Press. Amsterdam: Elsevier Science B.V. Saeed, J. I. (2003) Semantics. Second edi- Frege, G. (1892). ‘Sinn und Bedeutung’. In tion. Oxford: Blackwell Publishing. Zeitschrift für Philosophie und philosophische Kritik, NF 100:25–50. Hans Götzsche English translation, ‘On sense and meaning’. In Peter Geach and Max Blach (eds) (1980), Translations from the Philosophical Writings of Gottlob PRIVATE Frege. Oxford: Blackwell. 56–78. First LANGUAGE edition 1952. Journal of Semantics Vol. 9 no. 3 1992. A private language is a putative lan- Special issue on presupposition. guage the meaning of whose expres- Journal of Semantics Vol. 9 no. 4 1992. sions are essentially grounded in the Special issue on presupposition. subjective or private states of the users McCawley, J. (1981). Everything That of the language. The notion was first Linguists Have Always Wanted to employed by Ludwig Wittgenstein Know About Logic, But Were Ashamed (Philosophical Investigations) in order to Ask. Oxford: Basil Blackwell. to refute both classical and contempo- Portner, P. and B. H. Partee (eds) (2002). rary conceptions of linguistic mean- Formal Semantics. Oxford: Blackwell ing. The notion remains at the centre Publishing. of Wittgenstein scholarship and has a Reyle, U. (1993). ‘Existence presupposi- highly controversial use in philosophy tions and background knowledge’. of language and mind more generally. Journal of Semantics 10, 2: 113–22. Russell, B. (1905). ‘On denoting’. Mind See also: Deconstruction; 14: 479–99. Empiricism/Rationalism; Language Russell, B. (1957). ‘Mr Strawson on refer- Games; Logical Positivism; ring’. Mind 66: 385–89. Presupposition

179 PRIVATE LANGUAGE

Key Thinkers: Ayer, A. J.; Carnap, in approach and associated with John Rudolf; Chomsky, Noam; Fodor, McDowell, takes the argument to be a Jerry; Frege, Gottlob; Kant, principled declining of a justification Immanuel; Kripke, Saul; Russell, of our normative activity, for the Bertrand; Wittgenstein, Ludwig reason that any non-normative basis, such as consensus, would never suffice In the first half of the twentieth cen- to establish the kind of correctness tury, the received understanding of a involved in meaning. classical empiricist theory of language Wittgenstein’s argument has also was that words are outward signs of surfaced in cognitive science. Jerry private states (ideas or images). Such a Fodor and Noam Chomsky have position found its contemporary argued that meaning, linguistic and advocates in Bertrand Russell, Rudolf non-linguistic, has an empirical basis Carnap and A. J. Ayer. Arguably fol- in an unconscious system of rules that lowing Immanuel Kant and Gottlob determine interpretation as opposed Frege, Ludwig Wittgenstein argued to normatively governing it. They that such an essentially private lan- defend this approach against the pri- guage was impossible and, therefore, vate language argument by pointing the empiricist conception of meaning out that the rules at issue are not is confused. Although the argument essentially private, but simply not has been interpreted in many different available to first person conscious ways, its basic structure is as follows. access; ultimately, whether there are (1) Rule following is constitutive of such rules or not is an empirical issue competent language use. (2) Rule fol- that cannot be a priori determined. lowing is a normative activity – that is, there are essential conditions on Primary sources whether one is following the rules cor- Wittgenstein, Ludwig (1953). Philosophi- rectly or not. (3) If linguistic meaning cal Investigations. Oxford: Blackwell. is private, then the agent must be fol- Kripke, Saul (1984). Wittgenstein on lowing private rules. (4) If the rules Rules and Private Language. Oxford: are private, then there is no difference Blackwell. between following the rules correctly McDowell, John (1984). ‘Wittgenstein or not; whatever seems right will be on following a rule’. Synthese 58 (3): right. (5) Therefore, there can be no 325–64. such private language. Fodor, Jerry (1975). The Language of In general terms there are two Thought. Cambridge, MA: Harvard schools of thought on the argument. University Press. One school, most prominently repre- Chomsky, Noam (1986). Knowledge of sented by Saul Kripke, views the Language. Westport: Praeger. argument as an attempt to show Further reading that linguistic meaning is essentially Miller, Alexander, and Crispin Wright public, in that only a public check (eds) (2002). Rule-Following and by other speakers could establish Meaning. London: Acumen. whether one is following the rules or not. Another school, more diagnostic John Collins

180 PROPOSITIONAL (2) a. Jane believes that Fido is barking. ATTITUDES b. Fido exists. c. There is something that Jane A propositional attitude is what some- believes is barking. one is described as having by a propo- sitional attitude report: for example, Like other ordinary relational claims, by an utterance of ‘Jane believes that (1a) has existential import that (2a) Fido is barking’ or ‘Henry wishes his and other attitude reports seem to cat would have kittens’. The subtle lack. In particular, (1a) entails both semantics and pragmatics of proposi- (1b) and (1c) while (2a) apparently tional attitude reports, as well as the entails neither (2b) nor (2c). After all, role of propositional attitudes them- Jane could have the belief reported in selves in language use, have made (2a) even if Fido is a figment of her them one of the more controversial imagination, contrary to (2b); and in topics in linguistics. that case, contrary to (2c), nothing would have the property of being See also: Compositionality; believed by Jane to be barking, for if Definite Descriptions; there were, that thing would be Fido, Descriptivism; Intentionality; and there is no Fido. Propositions; Sense/Reference Another difference relates to the Key Thinkers: Frege, Gottlob; law of identity. This law, compelling Kripke, Saul; Quine, W. V. O.; in the context of ordinary relations, is Russell, Bertrand less plausible in the context of attitude A standard route into the topic of reports. Thus, suppose Fido is Rex. propositional attitudes begins with This identity and (1a) could not both the observation that propositional be true without (1d) also being true. In attitude reports differ from ordinary contrast, the identity and (2a) do not relational claims. On the face of it, entail (2d). After all, Jane may not for Jane to believe that Fido is realise that Fido and Rex are one and barking is for her to bear a two-place the same dog. relation to the proposition that Fido is barking – just as for her to (1) stroke Fido is for her to bear a two- a. Jane strokes Fido. place stroking relation to the dog d. Jane strokes Rex. Fido. This surface similarity is (2) deceptive. a. Jane believes that Fido is barking. One difference can be illustrated by d. Jane believes that Rex is barking. comparing (1) with (2): These two distinctive features of (1) propositional attitude reports – a. Jane strokes Fido. apparent lack of existential import b. Fido exists. and apparent failure – c. There is something that Jane give rise to a range of semantic puz- stroked. zles. Any solution to these depends on

181 PROPOSITIONAL ATTITUDES a proper appreciation of the nature of conditional properties of their simpler a fundamental notion in the philoso- components. For example, the truth phy of language and mind: proposi- condition stated in (4a) is a derivable tions*, that is the things that attitude consequence of (4b), (4c), and some reports report us as believing, desir- background semantic principles. ing, intending, and so on. After pre- senting two such puzzles, one relating (4) to each distinctive feature we will a. ‘Fido is a dog’ is true iff Fido is a think about how different theories of dog. propositions require differing treat- b. ‘Fido’ refers to x iff x = Fido. ments of the second of these puzzles. c. ‘is a dog’ is true of x iff x is a dog. The lack of existential import threat- ens to sabotage solutions to the already But what about (1a)? Its truth condi- notorious problem of negative existen- tion, stated in (5a), ought to be deriv- tials. This is the problem of under- able from (5b) plus background standing how we can ever correctly semantic axioms for the other parts of deny something’s existence when, in the sentence. order to do so, we would need to refer to it, and hence need it to exist. Various (5) purported solutions are available, but a a. ‘Jane believes that Fido is barking’ plausible condition on their acceptabil- is true iff Jane believes that Fido is ity is that they make sense of the legiti- barking. macy of inferences with the form in (3): b. ‘Fido’ refers to x iff x is Fido.

(3) a does not exist, therefore not . . . a The problem is that any compositional ... derivation of (5a) using (5b) seems to license a parallel derivation of the ille- For example, if Saladin never existed, gitimate (6a) from (6b). Given that it follows that Saladin could not have Fido is Rex, (6b) follows from (5b), and fought Richard the Lionheart. But the the derivation of (6a) from (6b) could inference appears to fail when (3) is then mimic that of (5a) from (5b). Yet ‘Fido does not exist, therefore Jane (6a) seems wrong: Jane’s belief that does not believe that Fido is barking’. Fido is barking is not enough to make Any theory of negative existentials the sentence ‘Jane believes that Rex is therefore has an additional burden: to barking’ true, even if Rex is Fido. explain exceptions to (3) thrown up by propositional attitude reports. (6) The second feature, apparent sub- a. ‘Jane believes that Rex is barking’ is stitution failure, threatens the compo- true iff Jane believes that Fido is sitionality thesis, a cornerstone of barking. most approaches to formal semantics. b. ‘Rex’ refers to x iff x is Fido. In one form, this thesis holds that the referential or truth conditional prop- Notice that this problem does not erties of complex expressions are a arise for ‘Rex is a dog’, which is true function of the referential or truth if Fido is a dog and Rex is Fido.

182 PROPOSITIONAL ATTITUDES

These puzzles may seem like mere Since the elements of propositions are technical challenges, but their actual entities, the referents of the endurance suggests otherwise. What words used, substituting co-referring follows is a summary of two promi- expressions will alter neither the iden- nent theories of propositions, stressing tity of the proposition expressed nor, how each confronts the puzzle of therefore, the truth value of the sen- apparent substitution failure. Both the- tence. Russellians must explain the ories treat propositions as structured appearance of substitution failure in complexes of elements, but they dis- propositional attitude reports with- agree on the nature of these elements. out granting that this appearance is Followers of Bertrand Russell genuine. assume that propositions are struc- Before 1972 the standard way was tured complexes of real-world Russell’s own. Ordinary names* like entities. Sentences map onto the ‘Fido’ and ‘Rex’ are not in fact refer- propositions they express via a map- ring expressions at all, he said. They ping that takes individual words onto are really disguised definite descrip- the real-world entities they refer to. tions*, which have an internal struc- The proposition expressed by a typi- ture of their own but no referents. The cal utterance of (1a), for example, contrast between (2a) and (2d) does would be (1e), the first element of not, therefore, constitute a case of which is a two-place relation, and the changing truth value by substituting second element of which is an ordered co-referring expressions. pair consisting of Jane then Fido. In more detail, suppose ‘Fido’ in Jane’s is a disguised version of (1) ‘The dog I bought last year’ and ‘Rex’ e. > is a disguised version of ‘The dog I ran over two years ago’. In that case, (2a) Thus (1e), and hence (1a) itself, will be and (2d) would express the distinct true just so long as the ordered pair propositions (2g) and (2h) respec- satisfies the two-place relation. More tively (where ᭙ can be read as roughly generally, the truth value* of proposi- equivalent to ‘for every’, ᭚ as ‘exists’ tions is determined compositionally and → as ‘entails’; see the entry on from its components. ‘Logic’ for a fuller account of these So (2a) expresses the proposition logical constants). (2e), which breaks down further into (2f). The simple proposition Jane pur- (2) portedly believes is a component of g. > sented in (2f) as ‘ ‘. h. > e. < believes, > No actual dog is a component of f. < believes, >> either proposition, so ‘Fido’ or ‘Rex’

183 PROPOSITIONAL ATTITUDES lack a shared referent, or indeed any But Fregeans have a ready explana- referent. tion for exceptions such as (2a) and Russell’s defence of Russellianism (2d): when a word occurs within the fell out of favour after Saul Kripke’s scope of a propositional attitude verb, 1972 criticisms of descriptivism about Fregeans say it refers to its customary names. More recent Russellians (for sense, not its customary referent. example, Salmon 1986) have instead Since ‘Fido’ and ‘Rex’ have distinct tried to explain the appearance of sub- customary senses, the propositions stitution failure as a pragmatic effect. expressed by (2a) and (2d) have dif- While (1a) and (1d) express the same ferent components even after con - Russellian proposition at the semantic version into Russellian propositions, level, the small difference between and hence potentially different truth them carries pragmatic significance values. easily mistaken for a semantic differ- The most pressing question ence. Fregeans must answer is: what is a The second view of propositions, sense? If a sense is just a disguised Fregeanism, is similar to Russellian- definite description, anti-descriptivist ism save that the elements in a propo- objections to Russell carry over. Some sition are not objects but perspectives take the sense of a word to be the on, or ways of thinking about, or (to concept it expresses. But concepts are use Gottlob Frege’s own ) often defined as components of pro - ‘modes of presentation’ of, objects. positions, and if ‘the sense of “Fido” ’ The mapping from sentences onto the is unpacked merely as ‘that compo- propositions they express is therefore nent of a proposition which is a function from a word onto, not an expressed by “Fido” ’, Frege’s solution object, but a specific mode of presen- seems hollow. More substantial and tation of an object. Frege calls this controversial theories of concepts are function the word’s sense*. Two available, but at the very least words, such as ‘Fido’ and ‘Rex’ (or Fregeans must address profound ‘’ and ‘Eric Blair’), can questions in the philosophy of mind have a common referent but distinct before they can claim to have solved senses. For this reason, substituting the substitutivity puzzle. one of the words for the other within To sum up, propositional attitude a sentence will change the proposition reports appear to behave differently expressed. It will not (normally) from sentences devoid of psychologi- change the truth value, however. The cal attitude verbs. These differences truth value of Fregean propositions is pose a challenge to semanticists. The determined in the same way as Rus- compositionality thesis, in particular, sellian ones, by referents rather than is difficult to defend in this context modes of presentation; and Fregean without taking potentially controver- propositions can be converted into sial stands on the nature of proposi- Russellian propositions by replacing tions, a nature whose investigation each perspective on an object with the belongs as much to the philosophy of object itself. Substitution will there- mind as it does to the philosophy of fore not (normally) affect truth value. language.

184 PROPOSITIONS

Primary sources In the analysis of sentence meaning Frege, G. (1892/1997). ‘On sinn and the application of the concept of Bedeutung’. In M. Beaney (ed.), The propositions has proved to be very Frege Reader. Oxford: Blackwell. fruitful. In formal semantics, truth- Kripke, S. (1972/1980). Naming and conditional semantics and speech act Necessity. Cambridge, MA: Harvard theory sentences are analysed with University Press. reference to the underlying proposi- Russell, B. (1919). Introduction to Math- tions that they contain. It is central to ematical Philosophy. London: George these theories that in natural lan- Allen and Unwin. guages propositions are not sentences. Salmon, N. (1986). Frege’s Puzzle. Cam- ‘I am hungry’ uttered by different bridge, MA: MIT Press. speakers is obviously an instance of the same sentence but as the reference Further reading changes with every speaker we have as Kripke, S. (1979). ‘A puzzle about belief’. many different propositions as we In A. Margalit (ed.), Meaning and Use. have speakers. Dordrecht: Reidel. 239–83. In formal semantics mathematical and logical techniques are used to Alex Barber describe the semantic structure of nat- ural languages. Applying proposi- tional calculus, we can deduce how logical operators such as negation, PROPOSITIONS conjunction, disjunction and implica- tions build up complex sentences out of other sentences. Propositions are A term adopted from philosophy and central to truth-conditional seman- formal logic*, referring to the lan- tics based on Alfred Tarski (1956) guage-independent common denomi- and to Montague Grammar, where nator of the meaning of all the Richard Montague (Thomason 1974) sentences that express the truth of a applies semantic principles developed certain state of affairs, independent of for artificial languages to natural their illocutionary form. Propositions languages. are the central elements used in the In his development of J. L. Austin’s investigation of sentence meaning in (1962) speech act theory, John Searle formal semantics and in speech act (1969) divided locutionary acts into theory*. utterance acts (uttering noises and See also: Artificial Intelligence; words) and propositional acts, by Logic; Logical Form; which the speaker is referring and Model-Theoretic Semantics; predicating. In uttering the sentences Possible World Semantics; Speech (1) ‘Sam smokes habitually’. (2) ‘Does Act Theory; Truth Theories; Truth Sam smoke habitually?’ (3) ‘Sam, Value smoke habitually!’ (4) ‘Would that Key Thinkers: Austin, J. L.; Sam smoked habitually’ (Searle 1969: Montague, Richard; Searle, John; 22), the speaker performs an asser- Tarski, Alfred tion, asks a question, gives an order

185 or expresses a wish. But at the same PROTOTYPE time she/he refers to a certain person called Sam and expresses the predica- tion ‘smokes habitually’ with respect explains how people to that person. In other words, the understand the meaning of a word by reference and the predication are the reference to the best example of the same in all four sentences, and thus object indicated by the word. This the same proposition is made, regard- means that, for example, in identifying less of their respective illocutionary an object as belonging to the category acts. This is in opposition to mathe- of furniture, the human mind is geared matical languages, where every to thinking in terms of ‘chair’, that is, proposition is expressed in the form the prototype member of that cate- of a statement. gory, rather than with the more mar- Logical analysis based on proposi- ginal ones such as ‘lamp’ or ‘stove’. tions has had a major influence on See also: Ambiguity/Vagueness; semantic studies of natural languages. Cognitivism; Intuition; Language Although the theories are controver- Games; Language of Thought sial, they can lead to more advanced Key Thinkers: Aristotle; Fodor, studies in semantic linguistics, partic- Jerry; Whorf, Benjamin Lee; ularly with respect to sentence mean- Wittgenstein, Ludwig ing, and in artificial intelligence. Every conceivable thing is in principle Primary sources distinct from every other. In order for Searle, John R. (1969). Speech Acts: An them to make sense, people group Essay in the Philosophy of Language. together exemplars of the same kind Cambridge: Cambridge University into categories. But how precisely the Press. mind works in this respect is not Thomason, Richmond H. (ed.) (1974). always clear. In the Aristotelian tradi- Formal Philosophy: Selected Papers of tion, categorisation is explained as the Richard Montague. New Haven: Yale conjunction of necessary and suffi- University Press. cient features. Hence a bird has to be a winged creature as its necessary fea- Further reading ture and possess a beak as its sufficient Austin, John L. (1962). How to Do Things feature. All members of a category are With Words. Oxford: Oxford Univer- thought to have an equal status. Once sity Press. established, a category will separate Kneale, William (1972). ‘Propositions and entities that belong to it from those truth in natural languages’ Mind 81: that do not. No ambiguity or varia- 225–43. tion is allowed. Tarski, Alfred (1956). Logics, Semantics In the 1970s Eleanor Rosch con- Metamathematics: Papers from 1923 to ducted a number of cognitive psycho- 1938. Trans. J. H. Woodger. Oxford: logical experiments on how people Clarendon Press. operate with categories and how they learn about things through them. Not Jürg Strässler all members of a category turned out

186 PROTOTYPE to be equally representative. Some are a higher, superordinate level in more central to the category; others contrast to lower, subordinate level are marginal and even dubious. For terms such as ‘robin’, ‘canary’, example, the people in Rosch’s study ‘parrot’, and so on. Between these two categorised different kinds of birds levels stands the basic level term, in according to a graded scale of recog- this case, ‘bird’. In creating categories, nisability ranging from the best exem- people tend to opt for the basic level plar, such as ‘robin’, to the least likely terms. Thus when referring to an members like ‘ostrich’. Between these avian creature, one normally says, opposite ends there is a range of ‘Look at this bird!’ rather than men- what people will consider as a bird. tioning a specific bird (robin, and so Very close to ‘robin’, but already less on) or using the superordinate term typical, were birds like ‘dove’, ‘spar- ‘animal’. Basic level terms are the row’, ‘canary’. Then slightly away most natural form of categorisation, from the centre was another group and for that reason they are said to with ‘parrot’, ‘owl’, ‘pheasant’, and have the highest degree of cue-validity. ‘toucan’ in it. Somewhat more distant After being shown a word that rep- are ‘peacock’ and ‘duck’, then ‘pen- resents a higher level category, people guin’, until finally, with some hesita- tend to find it easier to agree that two tion, the ‘ostrich’ is considered as words representing central members belonging to the category. Of all the of a category are the same. Thus when birds in Rosch’s study, ‘robin’ excelled primed with the word ‘furniture’, as the most representative member. It people generally take less time in was the prototype of the category. deciding that ‘chair-chair’ are the Similar experiments with other cate- same words than ‘stove-stove’. People gories such as vegetables and furniture are fairly consistent in identifying the give the same result. These categories best examples of a category. Again, exhibit a graded scale of membership, the speed with which people judge each with a prototype of its own, in this whether an object belongs to such case, ‘pea’ and ‘chair’. This suggests and such category also determines that people group things together and the place the object will occupy on the exclude others according to their simi- graded scale within the category. The larity to a prototype. Hence the notion shorter the response time, the more of prototype gives a better account of central an exemplar will be. People how the mind sets up categories and also seem to rely on the properties of uses them than the Aristotelian frame- the central members in dealing with work. The finding has a strong impact other exemplars. This means that not only in cognitive psychology but attention is normally given first to the also in linguistic research, especially in central members. It is therefore easier the works of R.W. Langacker in cogni- to elicit them than the marginal mem- tive grammar and in George Lakoff’s bers. semantic studies. All these are called prototype An important notion that links pro- effects. They play an important role in totype with linguistic meaning is the the way people mentally organise the ‘basic level term’. The term ‘animal’ is outside world. However, these effects

187 PROTOTYPE are not the same as the cognitive about domestic cats and dogs as representation itself. The difficulty typical pets, allowing for snakes and comes from the role of prototype caimans as less common pets. But ‘pet Though intuitively a robin is the pro- fish’ creates some problems. It does totypical bird, it cannot be used as a not matter whether the fish in ques- unique criterion for being a bird. tion, usually a goldfish, is prototypical Properties of some other birds are also fish or not, and it is not relevant either taken into consideration. The ques- where in the scale of pets the goldfish tion is which properties are to be stands. In fact the combination included and which ones are to be between ‘pet’ and ‘fish’ here is no ignored – in other words, how far one longer understood in terms of the can extend the category. While an prototypes of each category. ostrich may be quite far removed from Prototypes are a mental construc- a typical bird, it is still a bird, not a tion that helps us to understand the birdlike creature the way a bat is. A world. This construction or model is colour is no longer considered to be built upon an immense storage of red as it moves towards being orange concepts that can be retrieved to or purple. But where are the bound- combine with one another in order to aries of these categories? Such cases form a model to represent reality in show that there are no clear-cut the mind. One such powerful model boundaries – in other words, cate- is family resemblance, a notion that gories are inherently fuzzy. goes back to Wittgenstein. This is Furthermore, many words have dif- exemplified by the use of the word ferent meanings, evoking different ‘game’. Although every game has sets of membership. The word ‘jail- some similarity with other games, bird’, for example, refers to a criminal there is no single feature that links who is locked up behind bars like a them all. Instead, there is a compli- canary and who may be persuaded to cated network of similarities overlap- ‘sing’ (confess) in order to get better ping and criss-crossing. There is treatment. This shows that the word something that makes all members of ‘bird’ is associated not with a single a category resemble one another even category, namely the avian category, if it cannot be said that they share a but also with other categories, in this common set of properties. What hap- case prisoner. is the basis pens is that each member is linked for metaphors*. The expression ‘he is to the others by an intermediate a fox’ to describe a cunning and sly member with which each is in close person is an example where the meta- contact and thus shares some proper- phorical use of the word ‘fox’ gives it ties. Thus, schematically, it is possible a new sense. But obviously slyness is that in a category ABC the members not part of the ordinary sense of the A and C have nothing in common word ‘fox’. and yet both are part of a family In addition to the problems above, a simply because each is in contact with combination of categories can pro- member B. duce a more complex concept without As synthesised by John R. Taylor, reference to prototypes. People talk the notion of prototype is very influen -

188 PSYCHOANALYSIS tial in forming grammatical cate- Löbner, Sebastian (2002). Understanding gories. For example, what is proto- Semantics. London: Arnold. Chapter 9. typical in transitive verbs is the presence of two different participants, Agustinus Gianto namely the agent and the goal. If the agent and the goal are the same, then the verb is transitive-reflexive and is said to be less prototypical of the class PSYCHOANALYSIS of transitive verbs. The notion of pro- totypes also operates in other areas of A method of analysing discourse in linguistics but it is in this cognitive order to discover its hidden meanings, arena that the idea has had its most originally developed as a cure for significant influence. various mental disorders, but later extended to become a method of Primary sources analysing literature, film and other Lakoff, George (1987). Women, Fire, and cultural phenomena. Psychoanalysis Dangerous Things. What Categories was developed as a technique by Sig- Reveal About the Mind. Chicago: mund Freud from the 1890s to the Chicago University Press. 1930s; his disciple Langacker, R.W. (1987, 1991). Founda- imported structural linguistics into the tions of Cognitive Grammar. Vols 1–II. theory, and extended it into a general Stanford: Stanford University Press. philosophy of everyday life that has Rosch, Eleanor (1975). ‘Cognitive repre- been influential in the continental tra- sentations of semantic categories’. Jour- dition. nal of Experimental Psychology: General Vol. 104: 192–233. See also: Signs and Semiotics; Rosch, Eleanor (1978). ‘Principles of Cat- Structuralism; Poststructuralism egorization’. In E. Rosch and B. B. Key Thinkers: Hegel, G. W. F.; Lloyd (eds), Cognition and Categoriza- Jakobson, Roman; Kristeva, Julia; tion. Hillsdale; NJ: Lawrence Erlbaum. Lacan, Jacques; Plato; Saussure, 27–48. Ferdinand de Rosch, Eleanor and C. B. Mervis (1975). ‘Family resemblances: studies in the Psychoanalysis was founded as a dis- internal nature of categories’. Cognitive cipline by in the Psychology Vol. 7: 573–605. 1890s, but it was through the work of Taylor, John R. (1995). Linguistic Catego- Lacan that it took a ‘linguistic’ turn. rization. Prototypes in Linguistic Freud, who was contemporary with Theory. Second edition. Oxford: Frans Boaz, Émile Durkheim and Oxford University Press. Ferdinand de Saussure, conceived of psychoanalysis as a scientific disci- Further reading pline and his successors typically Aitchison, Jean (2003). Words in the viewed it as one of the ‘human sci- Mind. An Introduction to the Mental ences’ in the manner of anthropology, Lexicon. Third edtion. London: Black- sociology or linguistics. The range of well. Chapters 4–6. conditions which psychoanalysis is

189 PSYCHOANALYSIS used to treat has shrunk since Freud’s and a baseball. The interpreter puts time, from encompassing psychosis, the two images together to arrive at aphasia, schizophrenia and what the single word ‘screwball’, which as a Freud and his contemporaries called signifier of Daffy’s daffiness partakes ‘hysteria’, to being confined to the rel- of the meaning of neither ‘screw’ nor atively minor condition of neurosis ‘ball’. In order to arrive at this inter- and associated personality disorders. pretation, the interpreter must per- Yet psychoanalysis continues to have form a two-stage ‘decoding’: first he currency as a tool of cultural critique. or she must translate each picture into Freud’s claim to importance in lin- a word – a movement from symbol guistics comes primarily through to signification – and then translate Chapter VI of The Interpretation of the individual words into a combina- Dreams (1899), ‘The Dream-Work’. tion that is homophonically, but According to Freud, the ‘dream- not semantically, related to them – a work’ (the work in which the uncon- movement from signification to scious mind of the dreamer engages in meaning. The dreamer, meanwhile, order to create a dream) consists of has already performed the same four stages: condensation, displace- sequence of operations as an encoding ment, figurability and secondary revi- process, this constituting the dream- sion. Condensation is the process work. whereby a detailed and seemingly In later works Freud extends the long dream can take place within a province of psychoanalysis as an inter- very short amount of real time. pretive technique from dreams to Displacement is the process of replac- other, more explicitly linguistic phe- ing material the dreamer wishes to nomena, such as jokes and flashes of repress by other elements which wit, and ultimately to the discourse of apparently have a different meaning. the more generally. It is Figurability is the process of turning this fact, that psychoanalysis takes dream material into pictures – most as its working material the discourse dreams are pictorial, or are a combi- of the patient, that inspires Lacan’s nation of pictures and language. And ‘linguisticisation’ of Freud, famously finally secondary revision is the claiming that ‘the unconscious is process of turning the dream into a structured as a language’ (Lacan 1972: ‘day-dream’, of organising the mate- 188). According to Lacan, Freud’s rial so that it may be remembered as ‘condensation’ corresponds to Roman a coherent sequence. Jakobson’s concept of metaphor*, The first three of these processes, ‘displacement’ corresponds to Jakob- claims Freud, organise the dream into son’s metonymy, while ‘figurability’ a series of rebuses. A rebus is a kind of (or ‘transposition’) corresponds to ‘the picture-puzzle, whereby pictures sym- sliding of the signified under the signi- bolise words in an indirect manner. fier’ (Lacan 2006: 425). For example, in a 1940s Warner Bros To take the third of these phenom- Merrie Melodies cartoon, Daffy Duck ena first, Lacan develops Saussure’s is shown with a flag attached to his theory of the sign, as represented in tail, on which are depicted a screw the Figure 5 (Saussure 1992/1983).

190 PSYCHOANALYSIS

Figure 5

Here, the picture of the tree represents which determines how psychic phe- the concept (‘signified’) of a tree in the nomena are organised, and hence how speaker’s mind, while the word arbor subjects behave: according to this represents the ‘sound image’ (‘signi- example, men and women line up fier’) of that concept. The horizontal before the doors according to the sig- line shows that signifier and signified nifiers that label them. In other words, are discrete manifestations, while the sex is determined socially, rather than oval shows that nevertheless together biologically. Moreover, the bar sepa- they form a psychological unity. The rating signifier from signified, which choice of a tree (arbre in French) for was a relatively minor part of Saus- illustrative purposes is a little on sure’s theory, assumes a major role in Saussure’s part, since the relation Lacan’s: it represents nothing other between signifier and signified is held than the bar of repression. Thus even to be arbitrary (arbitraire) – and like- in everyday life, signifiers do not sig- wise the signified is separated from the nify signifieds directly, but rather, they signifier by a barre (barre being an signify, according to Lacan, other sig- anagram of arbre). nifiers in the signifying chain that con- Lacan (2006: 416), meanwhile, stitutes the subject’s discourse. It is the replaces Saussure’s diagram with the task of analysis to uncover the true another (see Figure 6). signifieds of these signifiers, which This both inverts Saussure’s dia- have been displaced. gram and reduplicates it. The signifier This theory of linguistic displace- is shown to have priority for the sub- ment leads to Lacan’s appropriation of ject over the signified, by being placed Jakobson’s theory of metaphor and above it. Meanwhile, the two doors metonymy. In a series of papers on are identical, so that only the signifiers aphasia, Jakobson defines metaphor as distinguish them. Accordingly, the sig- the selection or substitution of terms nifier for Lacan has a material reality, one for another, while metonymy is

191 PSYCHOANALYSIS

Figure 6 defined as combination of one term the desire to be desired’. This latter with another. Lacan, meanwhile, version introduces the Other into the conceives of Jakobson’s distinction equation, so that an alternative for- between metaphor and metonymy in mulation is ‘desire is the desire of the terms of Hegelian dialectics: his exam- Other’, in which the word ‘of’ can ple, ‘Your freedom or your life!’, is bor- likewise be read as either ‘for’ or ‘by’. rowed from Hegel’s Phenomenology of Lacan’s theory has always been Mind. Superficially it looks like a controversial: as a linguist, Lacan is Jakobsonian metaphorical equation, indebted to his predecessors Jakobson freedom=life, where one term can and Saussure, and hence his theory is be substituted for another. However, open to the same criticisms as is struc- if the slave ‘chooses freedom, he loses turalist linguistics more generally – for both [freedom and life] immediately – example, that it fails to account for if he chooses life, he has life deprived creativity* in language. However, psy- of freedom’ (Lacan 1979: 212). choanalysis still has currency as a According to Lacan, insofar as the sub- method of cultural analysis, particu- ject exists as a speaking being, he is larly within literary and film studies, subjected to this demand by being as and as exemplified through the work such. of Slavoj Žižek. While metaphor is thus equated with demand, metonymy is equated Primary sources with desire. The formula for the Freud, Sigmund (1976). ‘The Dream- metonymy of desire is, says Lacan, Work’. In The Interpretation of ‘desire is the desire of a desire’. This Dreams. Trans. James Strachey. Har- can be read either as ‘desire is the mondsworth: Penguin. 381–651. desire to desire’ (which, following Hegel, G. W. F. (1967). The Phenomenol- Plato, means that by definition desire ogy of Mind. Trans. J. B. Baillie. New can never be fulfilled), or as ‘desire is York and Evanston: Harper.

192 RELEVANCE THEORY

Jakobson, Roman (1971). Selected Writ- the study of human communication ings II: Word and Language. The Hague which is firmly grounded in a general and Paris: Mouton. view of human cognitive design. With Lacan, Jacques (1972). ‘Of structure as an H. P. Grice, relevance theorists assume inmixing of an otherness prerequisite to that human communication is charac- any subject whatever’. In Richard teristically intention-based, and so Macksey and Eugenio Donato (eds), they see verbal comprehension as The Structuralist Controversy: The involving not just the decoding of Languages of Criticism and the Sciences speech signals, but also the recogni- of Man. Baltimore and London: Johns tion of the speaker’s communicative Hopkins University Press. 186–200. intentions. Lacan, Jacques (1979). The Four Funda- See also: Implicature; Logical mental Concepts of Psychoanalysis. Form; Presupposition Trans. Alan Sheridan. Harmondsworth: Key Thinkers: Chomsky, Noam; Penguin. Fodor, Jerry; Grice, H. P. Lacan, Jacques (2006). Écrits: The First Complete Edition in English. Trans. Relevance is usually defined as a poten- Bruce Fink with Héloïse Fink and Rus- tial property of inputs (such as sell Grigg. New York and London: assumptions, thoughts, utterances) to Routledge. cognitive processes. Sperber and Saussure, Ferdinand de (1922/1983). Wilson (1995) advance two principles Course in General Linguistics. Second of relevance embodying two central edition. Trans. Roy Harris. London: claims about human cognition and Duckworth. communication: a first or ‘cognitive’ principle of relevance, and a second or Further reading ‘communicative’ principle of rele- Forrester, John (1980). Language and the vance. The first of this principles states Origins of Psychoanalysis. Basingstoke: that: ‘Human cognition tends to be Macmillan. geared to the maximisation of rele- MacCabe, Colin (ed.) (1981). The Talking vance’ (1995: 260). The second is the Cure: Essays in Psychoanalysis and statement that ‘[e]very act of ostensive Language. Basingstoke: Macmillan. communication communicates a pre- sumption of its own optimal relevance’ Karl Simms (1995: 260). The communicative prin- ciple of relevance does not have the same status as Grice’s Cooperative Principle and conversational maxims, RELEVANCE for it is simply a generalisation about ostensive-inferential communication. THEORY This generalisation about communica- tion applies without exception: it is not A cognitive theory of pragmatics orig- something that speakers ‘follow’ or inally developed in the 1980s by Dan can ‘opt out’ of, for example. Sperber and Deirdre Wilson. Rele- Relevance theory claims that lin- vance theory offers a new approach to guistic communication (and, in fact,

193 RELEVANCE THEORY all human communication) is rele- On the basis of these two compet- vance driven. Relevance is defined ing factors, the relevance of an input within this framework as a trade-off to an individual may be comparatively of two competing factors: cognitive or assessed as follows: (1) other things contextual effects and processing being equal, the greater the positive effort. A positive cognitive effect is ‘a cognitive effects achieved by process- worthwhile difference to the individ- ing an input, the greater the relevance ual’s representation of the world’ of that input; (2) other things being (Wilson and Sperber 2004: 608), such equal, the greater the effort required as a true belief or conclusion. Accord- in processing an input, the lower the ing to relevance theory, an input is rel- relevance of that input. The following evant to an individual just in case example, adapted from Wilson and positive cognitive effects result from Sperber (2004: 609), can be used to the processing of that input. illustrate the comparative relevance of Processing new information in a alternative inputs to an individual. context may yield three main types of Suppose that Peter, a friend of ours, positive cognitive effect. First, it asks us who we phoned last night. Let may yield a contextual implication us assume, furthermore, that we deducible from the combination of phoned Kim and Sandy last night. new and existing assumptions, but Each of the following three alternative from neither of these alone. Second, it utterances would constitute a true and may provide evidence that strengthens relevant answer to Peter’s question: an already existing assumption. ‘We phoned Kim and Sandy’, ‘We Third, it may contradict and eliminate phoned Kim’, and ‘We phoned Kim information already held. For exam- and Sandy or 2 ϩ 2 ϭ 5’. However, ple, given the contextual assumption these answers would not be relevant ‘If the lights are on, then Mary is to the same degree: the first would be home’, an utterance of the sentence comparatively more relevant than the ‘The lights are on’ as we approach our other two. Notice that ‘We phoned house may yield the contextual impli- Kim and Sandy’ entails ‘We phoned cation ‘Mary is home’. Contextual Kim’, and so it is a more relevant implications are the central type of answer because it yields the positive positive cognitive effect. cognitive effects of the second utter- Processing effort, on the other ance and more. The first utterance hand, is the effort of perception, would be a more relevant answer than memory and inference that must be ‘We phoned Kim and Sandy or 2 ϩ 2 expended in computing cognitive ϭ 5’ because, although these two effects. For instance, an utterance of a utterances are semantically equiva- wordy and syntactically complex sen- lent, the latter is obviously more costly tence would take more effort to to process. On the whole, when a sim- process than an utterance of a less ilar amount of effort is expended in wordy and simpler version of that sen- processing alternative inputs, the tence. Likewise, an indirect answer to more relevant of these inputs is the a question would require more pro- one that yields more positive cogni- cessing effort than a direct one. tive effects. Conversely, when similar

194 RELEVANCE THEORY

positive cognitive effects are derivable one that satisfies their expectations of from the processing of alternative optimal relevance. Notice that the inputs, the one which is less costly to term ‘interpretive hypotheses’ as used process is the more relevant. in the definition above includes not How much relevance are individu- just the proposition the speaker als entitled to expect? According to intended to communicate, but also the relevance theory, maximal relevance is contextual assumptions, implicatures an unreasonably high expectation in and attitudes intended by the speaker. communication because, for example, The relevance-theoretic compre- our interlocutors might be unwilling hension procedure does not guarantee or unable to produce information that that communication will always be would yield the most positive cogni- successful, of course. Misunderstand- tive effects for the least processing ings do occur from time to time. effort (Higashimori and Wilson Rather, this heuristic offers an account 1996). In light of this, Sperber and of how addressees select the interpre- Wilson (1995) have argued that while tive hypothesis they are entitled to cognition tends to be geared to the assume is the one overtly intended by maximisation of relevance, acts of their interlocutors. ostensive communication simply Relevance theory rejects the tradi- create an expectation of optimal rele- tional assumption that every pragmat- vance. In other words, for any osten- ically determined aspect of utterance sive stimulus (for example, a verbal interpretation other than reference utterance) addressees are only entitled assignment and disambiguation must to expect a degree of relevance that is be an implicature*. Central to this sufficient to warrant their effort in framework is the claim that the processing it, and which is also the explicit side of communication should highest degree of relevance that their also fall under the scope of a theory of interlocutors are able to achieve given pragmatics (Bach 1994; Carston their abilities, goals and preferences. 2002, 2004b; Sperber and Wilson The presumption of optimal rele- 1993, 1995). Thus, according to vance suggests the following general relevance theory, there are two comprehension procedure: ‘Check types of communicated ‘assumptions’ interpretive hypotheses in order of (conceptual representations of the accessibility, that is, follow a path of actual world): explicitly communi- least effort, until an interpretation cated assumptions, or ‘’, which satisfies the expectation of rele- and implicitly communicated ones vance is found; then stop’ (Carston (implicatures). When is an assump- 2002: 45). Every utterance gives rise tion communicated by an utterance to a number of possible interpretive ‘explicit’? Sperber and Wilson (1995: hypotheses that are compatible with 182) suggest that an is an the linguistic meaning of the sentence inferential development of the propo- uttered. According to this general sitional template or ‘logical form’* criterion, addressees follow a path encoded by an utterance. In other of least effort in considering such words, an explicature involves a com- hypotheses, stopping once they reach bination of linguistically decoded

195 RELEVANCE THEORY material and pragmatic enrichment. intention that a specific implication By contrast, implicatures are commu- should be recovered. As Sperber and nicated assumptions resulting from Wilson (1995: 197) put it: ‘Some impli- pragmatic inference alone. Let us catures are made so strongly manifest consider an example of a relevance- that the hearer can scarcely avoid theoretic explicature involving the recovering them. Others are made less restriction of the domain of a quan- strongly manifest’. Consider, for exam- tificational expression. An utterance ple, the utterance ‘I don’t like action of the sentence ‘Every cat has white thrillers’ as an answer to the question paws’ does not make a patently false ‘Have you seen the latest Harrison universal claim about cats, but will Ford film?’. It is not difficult to see that typically be used to express a more such a response can be used to implic- restricted true proposition – for exam- itly communicate the assumptions that ple, a proposition about cats in our ‘The latest Harrison Ford film is an local shelter. Thus, an explicature of action thriller’ and, in addition, that ‘I this utterance would be ‘Every cat in haven’t seen the latest Harrison Ford our local shelter has white paws’. As film’. According to relevance theory, pointed out above, this conceptual these assumptions are strong implica- representation is a pragmatic develop- tures because their recovery is essential ment of the propositional schema to understand the speaker’s intended corresponding to the conventional meaning. In other words, the expecta- meaning of the sentence uttered. tion of optimal relevance that an utter- The class of explicatures of a given ance of ‘I don’t like action thrillers’ utterance within the relevance- gives rise to requires that the addressee theoretic framework includes not just take these highly salient assumptions the proposition expressed by that (or very similar ones) as being implic- utterance, but also a range of so-called itly communicated. The response ‘higher-level’ explicatures, which are above may also implicitly communi- obtained by embedding that proposi- cate the assumption that the speaker tion under an appropriate pro - has a general dislike for commercial positional attitude* or speech act blockbusters, for example. However, description (Sperber and Wilson this assumption is a weak implicature 1993). For example, an utterance of because it need not be supplied by the the sentence ‘The boss is coming’ may addressee in order to satisfy his expec- be developed inferentially into the tation of optimal relevance. Indeed, higher-level explicature ‘The speaker there are many other similar implica- believes that the boss is coming’, or tures which may have been derived on even ‘The speaker is warning that the the basis of the speaker’s response, for boss is coming’. example that the speaker is not really a An interesting distinction is made in film enthusiast, or the speaker frowns relevance theory between strong and upon the use of violence. In general, weak implicatures (and communica- the more obvious the speaker’s com- tion, more generally). The strength of municative intentions, the stronger the an implicature crucially depends on communication. Conversely, the wider the manifest strength of the speaker’s the range of interpretive possibilities

196 RELEVANCE THEORY allowed by the speaker, the weaker the spective, using expressions which communication. The relevance-theo- encode procedures for the identifica- retic approach to the implicit-explicit tion of intended cognitive effects distinction has led to a significant would obviously reduce the processing reassessment of the interface between cost involved in achieving those semantics and pragmatics: the exis- effects, a result that is in consonance tence of pragmatic aspects of pro - with the communicative principle of positional content which do not relevance. For example, Blakemore correspond to items present in the (2002) links the use of the sentential syntactic representation, as argued by connective ‘but’ with the cognitive relevance theory, strongly suggests effect of contradiction and elimina- that context-sensitivity at this level is tion. Hence, according to Blakemore, widespread. the use of ‘but’ activates an inferential An important development within process resulting in the contradiction relevance theory was the recognition, and elimination of an assumption due to the seminal work of Diane which the speaker has reason to Blakemore (1987, 2002), that linguis- believe is accessible to the hearer. For tic meaning can encode constraints on example, an utterance of the sentence the inferential phase of utterance com- ‘Kim is rich, but unhappy’ activates an prehension. Thus, linguistic meaning inferential process whereby the hearer may affect the inferential processes contradicts and eliminates the accessi- that characterise utterance compre - ble assumption that wealth leads to hension in two different ways: while a happiness. majority of linguistic expressions Research in relevance theory has encode constituents of conceptual rep- also made an important contribution resentations, there are also expressions to the study of figurative language. which encode inferential procedures, Verbal irony, for example, is analysed which we could think of as instructions in this framework as an ‘echoic use’ of to increase the of a particular language in which the speaker dissoci- type of inferential process. ates himself tacitly from an attributed Blakemore justified the distinction utterance or thought (Wilson 2006; between conceptual and procedural Wilson and Sperber 1992). For exam- encoding in both cognitive and com- ple, uttering ‘You really are good at municative terms. Since, as it is this!’ after a friend has failed to score assumed in relevance theory, the inter- an easy goal in a quick counter-attack pretation of utterances involves carry- can be construed as an instance of ing out computations over conceptual verbal irony because we are tacitly representations, it is reasonable to dissociating ourselves from a thought expect from a cognitive point of view or utterance with a similar content that languages encode information (such as a reassurance that our friend about the inferential procedures in is a skilful footballer) which may which such conceptual representations have been attributed to us had the cir- enter (and not just the constituents cumstances of the game been differ- of conceptual representations them- ent. This analysis is a departure from selves). From a communicative per- more traditional Gricean accounts,

197 RELEVANCE THEORY according to which verbal irony con- versus implicating) in the study of stitutes an overt violation of the Qual- meaning in natural language. More- ity maxim, thus giving rise to a related over, it has helped to enhance our true implicature which contradicts understanding of the semantics- the literal meaning of the sentence pragmatics interface by arguing contro- uttered. The relevance-theoretic versially that the contribution of approach to verbal irony is in har- pragmatics to the propositional content mony with the framework’s assump- of utterances goes far beyond disam- tion that an expectation of relevance, biguation and reference assignment. rather than one of truthfulness, is a standard of verbal communication Primary sources (Wilson and Sperber 2002). Blakemore, Diane (1987). Semantic Con- A recent line of research in relevance straints on Relevance. Oxford: Black- theory explores in detail the idea that well. the interpretation of words is also Blakemore, Diane (2002). Relevance and highly context-sensitive, and that lexi- Linguistic Meaning: The Semantics and cal context-sensitivity is not restricted Pragmatics of Discourse Markers. Cam- to indexical expressions. This line of bridge: Cambridge University Press. investigation puts forward the hypoth- Carston, Robyn (1988). ‘Implicature, esis that lexical-pragmatic processes explicature and truth-theoretic seman- such as narrowings, broadenings, tics’. In Ruth Kempson (ed.), Mental approximations and metaphorical Representations: The Interface between extensions are the result of a single Language and Reality. Cambridge: pragmatic procedure which fine-tunes Cambridge University Press. 155–81. the conventional meaning of words in Carston, Robyn (2002). Thoughts and communication (Carston and Powell Utterances: The Pragmatics of Explicit 2006; Wilson 2003). Communication. Oxford: Blackwell In sum, research carried out within Publishing. the framework of relevance theory has Carston, Robyn (2004a). ‘Relevance yielded interesting and insightful results theory and the saying/implicating dis- (see, for example, Carston 1988, 2002, tinction’. In G. Ward and L. Horn (eds), 2004a, 2004b; Carston and Powell The Handbook of Pragmatics. Oxford: 2006; Sperber and Wilson 1987, 1993, Blackwell. 633–56. 1995). First, it has provided a cognitive Carston, Robyn (2004b). ‘Explicature and alternative to Gricean and neo-Gricean semantics’. In S. Davis and B. S. Gillon theories of pragmatics. It has also intro- (eds), Semantics: A Reader. Oxford: duced (or thrown new light into) a Oxford University Press. 817–45. number of important theoretical con- Carston, Robyn and George Powell cepts (explicature, echoic uses of lan- (2006). ‘Relevance theory – new direc- guage, strength of communicated tions and developments’. In E. LePore assumptions) and distinctions (decod- and B. Smith (eds), Oxford Handbook ing versus inference, explicature versus of Philosophy of Language. Oxford: implicature, conceptual versus proce- Oxford University Press. 341–60. dural meaning, interpretive versus Higashimori, Isao and Deirdre Wilson descriptive uses of language, saying (1996). ‘Questions on relevance’. UCL

198 SENSE DATA

Working Papers in Linguistics 8: F. Keil (eds), MIT Encyclopaedia of the 111–24. Cognitive Sciences. Cambridge, MA: Sperber, Dan and Deirdre Wilson (1987). MIT Press. 719–22. ‘Précis of relevance: communication and cognition’. Behavioral and Brain Iván García Álvarez Sciences 10: 697–754. Sperber, Dan and Deirdre Wilson (1993). ‘Linguistic form and relevance’. Lingua 90: 1–25. SENSE DATA Sperber, Dan and Deirdre Wilson (1995). Relevance: Communication and Cogni- Mental images of physical objects that tion. Oxford: Blackwell. First edition we perceive directly, with observed 1986. properties representing those of the Wilson, Deirdre (2003). ‘Relevance theory physical object itself. Popular among and lexical pragmatics’. Italian Journal philosophers through the early twen- of Linguistics/Rivista di Linguistica 15: tieth century, the sense data theory has 273–91. been criticised for its inability to Wilson, Deirdre (2006). ‘The pragmatics solidly define and recognise sense data of verbal irony: echo or pretence?’. as such. Lingua 116: 1722–43. See also: Empiricism/Rationalism; Wilson, Deirdre and Dan Sperber (1992). Ideational Theories; ‘On verbal irony’. Lingua 87: 53–76. Indeterminacy; Mentalism; Wilson, Deirdre and Dan Sperber (2002). Sense/Reference ‘Truthfulness and relevance’. Mind 111: Key Thinkers: Austin, J. L.; Ayer, 583–632. A. J.; Berkeley, George; Descartes, Wilson, Deirdre and Dan Sperber (2004). René; Hume, David; Kant, ‘Relevance theory’. In G. Ward and L. Immanuel; Locke, John; Moore, Horn (eds), The Handbook of Pragmat- G. E.; Russell, Bertrand ics. Oxford: Blackwell. 607–32. Although philosophers as far back as Further reading John Locke, George Berkeley and Bach, Kent (1994). ‘Conversational implic- David Hume wrote about human per- iture’. Mind and Language 9: 124–62. ception and understanding, the term Blakemore, Diane (1995). ‘Relevance ‘sense data’ (or its singular form ‘sense theory’. In J. Verschueren, J. Östman datum’) is more modern, having first and J. Blommaert (eds), Handbook of appeared in the early twentieth cen- Pragmatics. Amsterdam: Philadelphia. tury in the works of such thinkers as 443–52. G. E. Moore and Bertrand Russell. In Blakemore, Diane (2002). Understanding the , the Utterances: An Introduction to Prag- sense data theory argues that proper- matics. Oxford: Blackwell. ties of mental images directly correlate Huang, Yan (2007). Pragmatics. Oxford: to those of the physical objects they Oxford University Press. represent. Growing criticism to the Wilson, Deirdre (1999). ‘Relevance and contrary – most notably by J. L. relevance theory’. In R. Wilson and Austin, Frank Jackson and Wilfrid

199 SENSE/REFERENCE

Sellars – maintains that we form inter- data receive increasing support in lin- pretations, not mere mental images, guistic circles, where words in lan- that lack the same properties of those guage trigger direct awareness of physical objects. internal interpretations, not property- In support of the sense data theory, bearing sense data. Hume and Russell use an example of how a table appears differently to us Primary sources as we physically move alongside it or Austin, J. L. (1962). Sense and Sensibilia. away from it in order to describe how Oxford: Clarendon Press. our sensory experience adjusts to per- Berkeley, George (1998). Three Dialogues spectival variation, or changes in our between Hylas and Philonous. Ed. J. physical relationship to external Dancy. New York: Oxford University objects. Because the size and shape of Press. the physical table do not change, our Hume, David (2000). An Enquiry Con- perception is our awareness of the cerning Human Understanding: A Crit- table’s corresponding mental image. ical Edition. Ed. T. Beauchamp. Other arguments supporting sense Oxford: Clarendon Press. data have to do with perception of Russell, Bertrand (1964). The Problems of mental images not directly related to Philosophy. New York: Oxford Univer- physical objects. This includes optical sity Press. illusions, hallucinations, double vision and even time delays between Further reading the existence of a physical object and Firth, Roderick (1965). ‘Sense data and our perception of it. the percept theory’. In R. Swartz (ed.), More recently, criticisms of the sense Perceiving, Sensing, and Knowing. New data theory have directly refuted its York: Anchor Books. 204–70. supporting arguments. For example, Chrucky, Andrew (1992). ‘The alleged sense data theorists would argue that if fallacy of the sense datum inference’. thing X were made to look exactly like Eastern Pennsylvania Philosophical thing Y, what we are directly aware of Association. Bloomsburg University. in our perception is thing Y. This Huemer, Michael (2004). ‘Sense data’. mental image of thing Y is evidence of Stanford Encyclopedia of Philosophy sense data. Opposing arguments, how- (http://plato.stanford.edu/). ever, state that this illusion does not change the fact that we are still looking Jennifer A. Baldwin at and perceiving properties of a real (or material) thing X. Originating from discussions on human knowledge and understanding, SENSE/REFERENCE the sense data theory is an important component of the philosophy of per- The distinction between sense and ref- ception. As the debate continues over erence (German ‘Sinn’ and ‘Bedeu- whether mental phenomena behind tung’) goes back to the German perception exist as mental images or as logician Gottlob Frege. According to pure interpretation, critics of sense his understanding,

200 SENSE/REFERENCE are two different aspects of meaning of contrast, identified meaning with linguistic expressions, reference being the reference of the expression. In con- the entity to which an expression nection with this denotational concep- refers and sense being the way in tion of meaning, he worked out a which that entity is presented. For theory around two main puzzles about example, the expressions ‘Tom’ and language. ‘the prisoner’ may well refer to the The first puzzle is in dealing with same person, but they differ in sense. identity statements such as ‘the Morn- This highly controversial distinction ing Star is the Evening Star’. Of course had, and still has, a strong influence on this sentence conveys the information semantics and philosophy of language. that the objects called the Morning Star and the Evening Star (this pair is See also: Analytic Philosophy; used by Frege) are physically identical Compositionality; (both are the planet Venus). If we sub- Connotation/Denotation; Definite stitute the expression ‘Evening Star’ Descriptions; Names; with the co-referring expression Propositional Attitudes; Truth ‘Morning Star’ in the example above, Value it results in ‘the Morning Star is the Key Thinkers: Carnap, Rudolf; Morning Star’. In contrast to the first Dummett, Michael; Frege, statement, this is a tautology. It gives Gottlob; Kripke, Saul; Leibniz, only trivial information and does not Gottfried Wilhelm; Mill, J. S.; tell us anything new because the state- Russell, Bertrand; Searle, John; ment holds a priori. It is not possible Strawson, P. F. to comprehend and above all to Gottlob Frege, the father of modern explain this difference in the meaning mathematical logic, and also consid- of the two sentences solely by the ref- ered one of the founders of analytic erence of a sign, because this is the philosophy, developed the distinction same for both expressions. between sense and reference mainly in This observation leads Frege to con- his essay ‘On Sense and Reference’ clude that there is a second aspect to (‘Über Sinn und Bedeutung’, 1892). the meaning of a sign besides the ref- This paper is considered one of the erence, namely the sense (‘Sinn’). The most important historical sources of sense of a sign corresponds to the modern semantics. In it Frege sug- mode of presentation (‘Art des gested that in order to grasp the mean- Gegebenseins’). In the example the ing of an expression, one must expressions ‘Morning Star’ and comprehend not only the object it ‘Evening Star’ present two different refers to but also the way the object is ways of characterising the planet presented, that is to say the sense of Venus, the brightest shining visible the referring expression. celestial body in either the evening or Frege’s theory was opposed to the the morning. The difference between then common understanding, dating them in the example given above lies back to John Locke, that identified in a difference in the mode of presen- the meaning of a word with the inter- tation of the signs since their refer- nal idea connected with it. Frege, in ence, the planet Venus, is the same.

201 SENSE/REFERENCE

The direct reference approach – But a referent can correspond to sev- assuming that the meaning of an eral senses (not necessarily a unique expression is its referent – was already sense), for example in different lan- claimed by Mill but the two problems guages. With ‘sign’ Frege refers to described above (identity statements proper names, under which he and empty names) led Frege and later widely subsumes all expressions P. F. Strawson and John Searle to claim which denote or could denote objects, that proper names* have sense, and because a grammatically correct additionally but not necessarily, refer- meaningful expression having a sense ence. This view is also referred to as might lack a reference. For instance, mediated reference theory. Philoso- the expression ‘the least rapidly con- phers like Saul Kripke, John McDow- vergent series’ has no reference ell and Gareth Evans have argued because for each convergent series a against this position, convinced that less rapidly convergent series can be Frege’s treatment of so-called empty found. Other examples of empty names was inconsistent. Kripke names are figures from mythology like attacked Frege’s and Bertrand Rus- ‘Odysseus’, as already seen above. sell’s position in his lectures ‘Naming This was another reason for Frege to and Necessity’ asserting that proper distinguish a second aspect of mean- names do not have senses at all. ing. Frege holds the view that this Kripke said that the reference of a inconsistency marks a deficiency of proper name is determined by the natural language and even of symbolic associated description and he language of mathematical analysis. explained that contrary to the des - For a logically perfect language criptive properties a proper name (‘Begriffsschrift’) he demands that refers to the same object across all every correctly formed expression has possible worlds (counterfactual a reference. worlds included). Sense and reference need to be dis- The terms used by Frege are ‘Sinn’ tinguished from the associated idea and ‘Bedeutung’. This terminology is (‘Vorstellung’), the internal image of confusing. ‘Bedeutung’ is generally an object, which is subjective. It varies translated as ‘meaning’. But what is from person to person and even for commonly understood as meaning, one person it is not always the same. language internal and independent In Frege’s theory of meaning the sense from facts, rather corresponds to lies between the reference of a proper Frege’s other term ‘Sinn’ or sense. Other name (the object itself) and the sub- authors adapt the connotation/denota- jective idea. It is neither subjective nor tion or intension/extension distinction. is it the object itself. Frege developed The essay is translated as ‘Sense and the theory of sense and reference into Reference’ by Max Black (1948), but as a philosophy of language. Thus, ‘Sense and Meaning’ in Frege’s Posthu- departing from proper names he gen- mous Writings (1979). They are herein eralises the twofold semantic function after called ‘sense’ and ‘reference’. to entire declarative sentences. The According to Frege’s theory, a sign meaning of a sentence (or complex has a sense, and this has a reference. expression) is determined by the

202 SENSE/REFERENCE meanings of its constituents. This is This theory of sense and reference can the principle of compositionality* only be maintained if we assume that controversially attributed to Frege. in these contexts the expressions do Hence it should be possible to substi- not have their customary reference. In tute constituents of the sentence with his essay ‘Sense and Reference’, Frege co-referring constituents without analyses various types of subordinate affecting the reference of the entire clauses in detail. sentence. If we substitute the proper name Leibniz had formulated this regu- Charlotte Brontë with her pseudonym larity as his law of substitution and in Currer Bell in the following true sen- Frege’s theory the objective content of tence ‘Dorothy believes that Charlotte a declarative sentence is called a Brontë is the author of Jane Eyre’, the thought (‘Gedanke’) and is under- meaning of the entire sentence is stood as the sense of a sentence. Sub- changed because the resulting sen- stitution might change the sense or tence is not necessarily correct: thought as in the example given ‘Dorothy believes that Currer Bell is below. the author of Jane Eyre’. Dorothy might not be aware of the fact that ‘The morning star is a body illuminated Charlotte Brontë at that time pub- by the sun.’ lished under the assumed masculine ‘The evening star is a body illuminated name Currer Bell; therefore she might by the sun.’ believe that ‘Charlotte Brontë is the author of Jane Eyre’ and at the same In contrast what remains the same is time believe that it can not be the case the truth value of the sentence, which that ‘Currer Bell is the author of Jane Frege considers the reference. The Eyre’, despite the fact that they are idea of truth values comes from the one and the same person. discipline of logic or logical semantics. The issue described above is also It refers to a function which maps sen- known as Frege’s second puzzle about tences on to the truth values true or language. How can it be that the prin- false. Corresponding to what is said ciple of substitution fails in the con- about proper names, Frege claims that text of propositional attitude* reports sentences need to have sense but not (or of indirect )? The answer necessarily reference. This can be is that in these cases we talk about the shown in the sentence ‘Miss Marple words themselves. Expressions corre- investigated the assassination in the spondingly do not have their custom- Orient Express’. This sentence is ary reference, but they have their meaningful but, as it contains the indirect reference, coinciding with name ‘Miss Marple’, a fictional char- what is customarily their sense. acter whose reference is doubtful, it Frege introduces the term ‘indirect’ cannot be said to have a reference, (German ‘ungerade’) reference/sense that is a truth value*. contrary to the ‘customary’ (‘gewöhn- The following paragraph highlights lich’) reference/sense. This leads to the several subordinate clauses as exemp- position that under the given circum- tions from the substitution principle. stances a clause’s reference is not a

203 SENSE/REFERENCE truth value but a thought. Thus we the antecedent clause nor the conse- can only substitute the subordinate quent clause qualifies to express an clause with another with the same entire thought. It is their combination thought (that is, reference) in order to which expresses one single complete keep the truth value of the complex thought. While several types of subor- sentence but not necessarily substitute dinate clauses are in a way incom- another with same truth value. plete, others – for example, concessive The situation is comparable in sen- clauses – are complete in this respect. tences with ‘it seems that’ and ‘com- The remaining cases of failure of mand’, ‘ask’, ‘forbid’, or ‘doubt the substitution test are explained whether’, to name a few. In all these with additional subsidiary thoughts cases the words in the subordinate not explicitly expressed. One example clause have an indirect reference and given by Frege himself is a causal sub- this determines that the reference of ordinate clause: ‘Because ice is less the subordinate clause is not a truth dense than water it floats on water’. value but a thought, a command, a Frege argues that the compound request, a question and so on. Accord- expresses more than one thought per ing to Frege, the subordinate clause clause namely three thoughts alto- following these expressions may also gether: be understood as proper name of the thought it represents in the compound 1. Ice is less dense that water. sentence. 2. If anything is less dense that water, The discussion about sense and ref- it floats on water. erence of propositional attitude 3. Ice floats on water. reports, initialised by Frege, is still alive. One topic is the question, for The clause ‘because ice is less dense example, of how nested or than water’ does not only express the propositional attitudes – leading to a first but also part of the second hierarchy of senses – fit into Frege’s thought. This overlapping is the theory. Apart from cases of indirect reason why it is not possible to reference of words, there are other exchange the subordinate clause given cases where sense and reference do above by another one with the same not correspond to thought and truth truth value without doing harm to the value respectively. In the sentence truth of the entire sentence. It thus ‘whoever invented the ingenious Miss does not serve to disqualify the theory Marple was a fanciful person’, the of sense and reference of sentences in grammatical subject ‘whoever’ has no this view. independent sense, but only in con- Frege’s ideas had a significant nection with the main clause. This impact on the development of modern accounts for the fact that the sense of semantic theories. The distinction the subordinate clause is not a com- between sense and reference continues plete thought. Besides, the reference is to be the subject of research in philos- not a truth value but the person ophy. Besides the topics already men- Agatha Christie. A further example is tioned, another issue still discussed in conditional sentences. Usually neither the literature is the questions raised by

204 SIGNS AND SEMIOTICS sentences which contain expressions Harman (eds), Semantics of Natural depending on context, that is deictical Language. Dordrecht: Reidel. terms such as ‘today’. As Frege con- Russell, Bertrand (1905). ‘On denoting’. siders a thought to be complete, the Mind: 479–99. question of what is the sense of these Quine, W. V. O. (1980). ‘Reference and expressions arises among Frege’s modality’. In From a Logical Point of interpreters and is broadened to a gen- View: Nine Logico-Philosophical eral quest for an adequate theory of Essays. Cambridge, MA: Harvard Uni- deixis. versity Press. 139–59. First edition 1953. Primary sources Black, Max and Peter Geach (eds) (1980). Further reading Translations from the Philosophical Beaney, Michael and Erich H. Reck (eds) Writings of Gottlob Frege. Oxford: (2005). Gottlob Frege: Critical Assess- Blackwell. First edition 1952. ments of Leading Philosophers. Vol. IV: Church, Alonzo (1956). Introduction to Frege’s Philosophy of Thought and Lan- Mathematical Logic I. Princeton: guage. London: Routledge. Princeton University Press. Revised and Burge, Tyler (2005). Truth, Thought, enlarged version. Reason: Essays on Frege. Oxford: Frege, Gottlob (1879). ‘Begriffsschrift, Oxford University Press. eine der arithmetischen nachgebildete Carl, Wolfgang (1994). Frege’s Theory of Formelsprache des reinen Denkens’. Sense and Reference. Its Origin and Halle: Louis Nebert. Trans. J. Van Hei- Scope. Cambridge: Cambridge Univer- jenoort as ‘Concept script, a formal lan- sity Press. guage of pure thought modelled upon Dummett, Michael (1973). Frege: Philos- that of arithmetic’. In J. van Heijenoort ophy of Language. London: Duck- (ed.) (1967), From Frege to Gödel: A worth. Source Book in Mathematical Logic, Ricketts, Thomas (ed.) (forthcoming). The 1879–1931. Cambridge, MA: Harvard Cambridge Companion to Frege. Cam- University Press. bridge: Cambridge University Press. Frege, Gottlob (1892). Über Sinn und Sluga, Hans D. (1993). The Philosophy of Bedeutung’. Zeitschrift für Philosophie Frege. Four vols. New York: Garland. und philosophische Kritik Vol. 100: 25– 50. Eva Herrmann-Kaliner Frege, Gottlob (1892/1948). ‘Sense and reference’. Trans. Max Black. The Philosophical Review Vol. 57, no. 3 (May 1948), 209–30. SIGNS AND Frege, Gottlob (1979). Posthumous Writ- ings. Ed. Hans Hermes, Friedrich Kam- SEMIOTICS bartel, Friedrich Kaulbach. Trans. Peter Long, Roger White. Chicago: University The term Semiotics (Semiologie) in the of Chicago Press. sense that it is understood in the Kripke, Saul (1972). ‘Naming and neces- twenty-first century was first used by sity’. In Donald Davidson, Gilbert Ferdinand de Saussure in his 1908

205 SIGNS AND SEMIOTICS

lessons on general linguistics. A sign is Ferdinand de Saussure defined what any entity representing another entity: he called Semiologie as the science of smoke as a sign of fire, or a stop signal signs. He conceived of semiotics as alerting drivers to come to a halt at a part of social psychology and there- crossroad. To the whole set of signs in fore of general psychology. Semiotics human and non-human communica- is part of psychology because the sign tion belong different subsets depend- is a twofold psychic entity: such an ing on their quality. They can thus be entity is composed of two sides, the encompassed as visual signs, auditory concept and the acoustic image, both signs, verbal signs, cloth signs, and the of which reside in the ‘same psychic like. As far as linguistics is concerned, site’. semiotics is the science that studies lin- Saussure was introduced to a sci- guistic signs, which is tantamount to ence of signs by his professor Michel saying that linguistics is part of a more Jules Alfred Bréal, whose course in general science of semiotics which in semantics he attended in 1881. In his turn is a branch of general psychology. lessons, later collected in his famous Essai de Sémantique (1897), Bréal See also: Langue/Parole; claimed that ‘les mots sont des signes: Structuralism ils n’ont pas plus d’existence que les Key Thinkers: Bopp, Franz; gestes du telegraphe aérien ou que les Hjelmslev, Louis; Morris, Charles; points et les traits du télégraphe Peirce, C. S.; Saussure, Ferdinand de Morse’ (‘Words are signs: they have The very first mention of the label no further existence than the move- ‘semiotics’, spelled ‘semeiotics’, dates ment of the railway semaphore signal back to 1670 when the English physi- or the dots and dashes of Morse code’) cian Henry Stubbes used it to denote (Bréal 1897: 835), and departed from the branch of medicine relating to the the eighteenth-century semiologic tra- interpretation of symptoms. Twenty dition in order to show the inconsis- years later John Locke used the label tency of the organicistic conception of to discuss his tripartite subdivision of the language. Bréal was influenced by science: the studies of and and the notion of the third branch may be called Semei- sign they utilised in their language otike, or the doctrine of signs; the most investigation, and by the philosophi- usual whereof being words, it is aptly cal perspective of Etienne Bonnot de enough termed also Logike, logic: the Condillac and the importance he gave business whereof is to consider the to psychology in his philosophical nature of signs, the mind makes use of studies. Bréal’s semantics is thus for the understanding of things, or con- based on the psychological laws of veying its knowledge to others. (Locke the human mind, and traces of John 1963: 174) Locke’s and Condillac’s ideas can be found in his work as well as insights But we have to wait until the end of from the studies on psychology and the nineteenth century to see the label memory sciences like those offered by applied to language. Hippolite Taine, Théodule-Armand

206 SIGNS AND SEMIOTICS

Ribot and Paul Pierre Broca. To the Whitney’s claim of the arbitrary thirty-year shaping of his theory of nature of sign, Saussure contended sign, Bréal included a strong psycho- that the meaning of a sign is not con- logical value that he drew from the tained within it, but arises in its inter- studies on the human mind which pretation. This means that meanings were so pervasively spreading in the do not exist per se but are established second half of the nineteenth century. by the language users in relation to He thus attempted to search the cog- the context of use. In other words, nitive and intellectual features of the the role of the interpreter must be human mind which regulate the accounted for, either implicitly as in nature of language. Saussure, or explicitly as in Peirce. This aspect of Bréal’s semantics is Saussure’s apparatus was followed evident in the work of Saussure, who by Hjelmslev (1943) who substituted aimed to find a bridge between the the terms expression and content to psychic essence of the concept and the refer to the signifier and signified concrete reality of the word. He came respectively. He also referred to up with the idea of ‘sign’ as an indi- planes of expression and content, each visible pair consisting of abstract con- having substance and form. Thus cept and concrete realisation. Until there are four categories which may one of his 1911 lessons, when he first facilitate analytical distinctions: sub- used the two terms signifiant and sig- stance of expression, form of expres- nifié, Saussure generally spoke of sign sion, substance of content, and form in the same way as was done by his of content. contemporaries, that is as the phono- In the United States semiotics devel- logical counterpart of an entity. Later oped in the second half of the nine- he developed a more refined theory teenth century within the field of where the sign is conceived of as a philosophy where Charles Peirce dyadic entity formed by the indivisible claimed that is ‘an action, or combination of a signifier (signifiant, influence, which is, or involves, a the acoustic image) and a signified dynamical operation of three subjects, (signifié, the mental representation of such as sign, its object, and its reality). The signifier and the signified interpretant, this tri-relative influence are intimately linked by an associative not being in any way resolvable into link whereby each triggers the other. action between pairs’ (Hartshorne et Hence they stand in a static dyadic al. 1958: 5.484). Whereas in the Saus- relationship which goes under the surean theoretical apparatus, the sign label of signification. With reference is a bipartite entity, Peirce conceived to language, a linguistic sign is not a of it as a dynamic triadic relation: link between a thing and a name, but between a concept and a sound pat- a sign or representamen is something tern. This means that language is which stands to somebody for some- mainly symbolic, since the relations thing in some respect or capacity. It between the sound sequences and addresses somebody, that is, creates in their meanings are conventional, or the mind of that person an equivalent arbitrary, and have to be learnt. On sign, or perhaps a more developed sign.

207 SIGNS AND SEMIOTICS

The sign which it creates I call the inter- iourism* and investigated the under- pretant of the first sign. The sign stands standing of the unitary process of for something, its object. (Hartshorne semiosis. He proposed to focus on the et al. 1958: 2.228) relation that the sign can establish with the other entities in the semiotic Semiosis is therefore an endless pro- process. Thus the sign-object relation cess, something that Eco (1976) points to the dimension of semantics; defined as ‘unlimited semiosis’. the sign-sign relation refers to syntax; In his investigation of the triadic and the sign-interpreter relation refers nature of sign (Triadism), Peirce to pragmatics. Such a tripartite divi- claimed that ‘all thought whatsoever sion became normalised in linguistics. is a sign, and is mostly of the nature Semiotics is a broad discipline which of language’ (Hartshorne et al. 1958: deals with any type of signification 5.421). In his triadism, Peirce and communication. It encompasses analysed the sign in itself, in relation branches like social semiotics, visual with the object, and in relation with semiotics, zoosemiotics, music semiol- the interpretant, thus developing a ogy, , and lit- huge number of categories. As far as erary semiotics, to mention but a few. language is concerned, the most rele- vant triad is the one which originates Primary sources from the relation that a sign estab- Bréal, Michel (1897). Essai de séman- lishes with itself. The sign-to-sign rela- tique. Science des significations. Paris: tion produces three modes, an icon, Hachette. an index, and a symbol. The icon is a Eco, Umberto (1976). A Theory of Semi- mode in which the signifier is per- otics. Bloomington: Indiana University ceived as imitating the signified, as is Press/London: Macmillan. the case with onomatopoeia. An index Hartshorne, Charles, Paul Weiss and is a mode in which the signifier is arbi- Arthur W. Burks (eds) (1958). Collected trary but connected in some way, Papers of . Eight physically or causally, to the signified, volumes. Cambridge, MA: Harvard for example in the case of demonstra- University Press. tive . A symbol is a mode in Hjelmslev, Louis ([1943]1953). Prolegom- which the signifier is fundamentally ena to a Theory of Language. Balti- arbitrary. These three modes give rise more: Indiana University Publications to three basic principles which are in Anthropology and Linguistics. fundamental to investigate linguistic Morris, Charles W. (1938/1970). Founda- phenomena: the principle of indexi- tions of the Theory of Signs. Chicago: cality, of iconicity, and of symbolicity. Chicago University Press. Peirce’s scientific contribution went Saussure, Ferdinand de (1916/1983). almost unknown during his life- Course in General Linguistics. Trans. time, and his semiotic approach was Roy Harris. London: Duckworth. spread by Charles Morris who, draw- ing largely on the Peircean theoretical Further reading framework, approached semiotics Chandler, Daniel (2001). Semiotics: The through the lenses of Mead’s behav- Basics. London: Routledge.

208 SITUATIONAL SEMANTICS

Eco, Umberto (1984). Semiotics and the called ‘situations’ and can be indi - Philosophy of Language. Bloomington: viduated by cognitive agents. Thus, Indiana University Press. people perceive situations, cause them Locke, John (1689/1963). ‘Of the division to be brought about, and have all sorts of the sciences’. Book IV, chapter XXI. of attitudes toward them. One fact An Essay Concerning Human Under- remains: we are at all times in situa- standing. Ed. Peter N. Nidditch. tions (cf. Norbert Hornstein: ‘Situa- Oxford: Clarendon. tions people the world. They are dated and located.’). Annalisa Baicchi While the Barwise-Perry volume (1983) is exceptional in its pro - grammatic employment of situations (applied, among others, to naked- SITUATIONAL infinitive perception and belief reports), historically there was always SEMANTICS some interest in situations. Two note- worthy – albeit cryptic – passages in An information-based approach to Zettel (Wittgenstein 1981: 2, 13) natural language semantics. Formu- show that Wittgenstein thought that lated by Jon Barwise and John Perry in situations a person is embedded in are their influential book Situations and of key value in making their behav- Attitudes (1983), it is built upon the iour intelligible. Authorities of prag- notion of a ‘situation’ – a limited part matics like J. L. Austin, H. P. Grice of the real world that a cognitive agent and Peter Strawson could be regarded can individuate and has access to. A as friendly to a situational approach, situation represents a lump of infor- for they try to come to terms with mation in terms of a collection of the notion of ‘context’. And for some, facts. It is through the actualist ontol- situations are generalised versions ogy of situations that the meaning of of ‘events’ as conceived by Donald natural language utterances can be Davidson and others. elucidated. A situation is a rich object consist- ing of individuals enjoying various See also: Logic; Possible World properties and standing in a variety of Semantics relations. It is a ‘small’ world. Inci- Key Thinkers: Austin, J. L.; dentally, there is a crucial difference Davidson, Donald; Frege, Gottlob; between situation-theoretic and Grice, H. P.; Lewis, David; mathematical relations. The latter are Montague, Richard; Strawson, P. set-theoretic constructs whereas the F.; Tarski, Alfred; Wittgenstein, former are relations of the kind recog- Ludwig nisable by cognitive agents. A situa- Situational semantics (‘situation tion may extend quite far in space and semantics’ in the sequel) starts with time. An agent can watch a film about the hypothesis that what is called ‘the a past assassination, scrutinise the world’ is an inconceivably large total- latest videos from the Jupiter mission, ity. Limited parts of the world are or chat with someone who relates

209 SITUATIONAL SEMANTICS their adventures in the Pampas of intricacy was cited by Gadamer Argentina. (1975: 268–9) who saw that the very One of the features of situation idea of a situation necessitates that an semantics is its information-based dis- agent is not located outside of it and position. Let us define something’s hence may be unable to have objective being P (a property) or something’s epistemic access to it. having R (a relation) to something else Human beings and lower organ- as a ‘state of affairs’ (Armstrong isms display a fundamental ability to 1997). In , ‘infons’ discern similarities between situa- are posited as discrete items supplying tions. This is accomplished via regu- such bits of information. An infon is larities, that is individuals, relations, shown as an (n + 2)-tuple , where R is an n-place relation uation to another. Thus, I believe that (properties being 1-place relations); snow makes driving difficult, that a1, . . ., an are objects appropriate for doctors are available for medical the respective argument places of R; assistance, that parents care about and p is polarity. If p=yes (respectively, their offspring, that I will receive a no) then a1, . . ., an stand (respectively, present on Father’s Day. do not stand) in the relation R. Barwise and Perry note that agents Abstract situations are proposed to ‘must constantly adapt to the course of be counterparts of real situations in events in which they find themselves’ order to make the latter amenable to (1983: 10). This takes formal manipulation. Given a situa- place as an upshot of attunement to tion s, the set {i | s =| i}, where i stands similarities between situations (‘uni- for an infon, is the corresponding formities’). Thus, a useful uniformity abstract situation. Notice that this set in my life has to do with the milkman. collects all facts (infons that are made Every morning (a different situation), true by s). Alternatively, s is said to he brings the milk at about 8 o’clock ‘support’ (make it the case that) i – and leaves it on our doorsteps. By just denoted as s =| i above – just in case i being attuned to this uniformity, I con- is true of s. tribute to my well-being. Violation of Devlin (1991: 31) has studied what a uniformity is possible; there is no situations might amount to and how milk service on holidays. we can ‘individuate’ them. A scheme Representation of uniformities of individuation – a way of carving the yields ‘types’. Suppose Bob was eating world into uniformities – is an essen- cookies yesterday and is eating cook- tial facet of the situational approach. ies now. Both of these situations share This way we can single out – say, via the same constituent sequence . These events, occur- treat situations as entities that can ring at different times, have the same later be referred to. When agents indi- type. In the same vein, consider two viduate a situation, they cannot be ‘parametric’ infons and , of all that the situation comprises: sit- where ĝ and ĥ are placeholders for uations are vague objects. Another individuals. Their meaning can be

210 SITUATIONAL SEMANTICS

rendered as ‘Someone embraces Carol’ For instance, an utterance of ‘I am and ‘Someone embraces someone’, smiling’ defines a meaning relation. respectively. Anchoring parameters Given d, c, and e, this relation holds of an infon yields (parameter-free) just in case there is a location l and a infons. For example, given , if F(ĝ) = David (F is an and, in e, s is smiling at l. In interpret- anchoring) then we obtain . context, there is a flow of information, Networks of abstract links between partly from the linguistic form situation types provide information encoded in f and partly from contex- flow (Dretske 1981). Thus, the state- tual factors provided by the utterance ment ‘smoke means fire’ expresses the situation u. These are combined to law-like relation that links situations form a set of constraints on the where there is smoke to situations described situation e. where there is a blaze. If a is the type Ideas from situation semantics of smoky situations and b is the type have been applied to a number of of fire situations, then having been issues in logic*, language, cognition attuned to the constraint a » b (read ‘a and information. To take three involves b’) an agent can pick up the comprehensive projects, Barwise and information that there is a fire in a Etchemendy (1987) analyse self- particular site by observing that there reference and paradox, Gawron is smoke. and Peters (1990) deal with pronom- According to situation semantics, inal , and Cooper (1996) meanings of expressions reside in sys- focuses on generalised quantifiers. tematic relations between different Unlike the classical approaches to types of situations. They can be iden- meaning (including Fregean senses, tified with relations on discourse situ- Tarskian truth, Montague grammar), ations d, connections c, the utterance there is an ordinary feel to situation situation u itself, and the described sit- semantics; it does not impose human- uation e. Some public facts about u – made assumptions in our conceptual such as its speaker and time of utter- scheme (in contra-distinction to ance – are determined by d. The ties of Lewisian possible worlds, for exam- the mental states of the speaker and ple). It is an archetype of what a nat- the hearer with the world constitute c. uralised theory of semantics should A discourse situation d involves the look like. expression uttered, its speaker, spa- tiotemporal location of the utterance, Primary sources and the addressee. Each of these Barwise, Jon (1989). The Situation in defines a linguistic role (role of the Logic. Stanford, CA: CSLI Publications. speaker, of the addressee, and so on). Barwise, Jon and John Etchemendy The utterance situation u constrains (1987). The Liar. New York: Oxford the world in a certain way, depending University Press. on how the roles for discourse situa- Barwise, Jon and John Perry (1983). Situ- tions, connections and described situ- ations and Attitudes. Cambridge, MA: ation are to be filled. MIT Press.

211 SPEECH ACT THEORY

Devlin, Keith (1991). Logic and Informa- Gadamer, Hans-Georg (1975). Truth and tion. New York: Cambridge University Method. Trans. and ed. Garrett Barden Press. and John Cumming. New York: Devlin, Keith (2006). ‘Situation theory Seabury Press. and situation semantics’. In Dov Gawron, Jean Mark and Stanley Peters Gabbay and John Woods (eds), Hand- (1990). Anaphora and Quantification book of the , vol. 7. in Situation Semantics. Stanford, CA: Amsterdam: Elsevier. 601–64. CSLI Publications. Seligman, Jerry and Larry Moss (1997). Grice, H. P. (1989). Studies in the Way of ‘Situation theory’. In van Benthem, Words. Cambridge, MA: Harvard Uni- Johan and Alice ter Meulen (eds), versity Press. Handbook of Logic and Language. Strawson, P. F. (1997). Entity and Identity. Amsterdam: Elsevier. 239–309. Oxford: Oxford University Press. Wittgenstein, Ludwig (1981). Zettel. Ed. Further reading G. E. M. Anscombe and G. H. von Aczel, Peter, David Israel, Yasuhiro Kata- Wright. Trans. G. E. M. Anscombe. giri, and Stanley Peters (eds) (1993). Sit- Oxford: Basil Blackwell. uation Theory and Its Applications, vol. 3. Stanford, CA: CSLI Publications. Varol Akman Armstrong, D. M. (1997). A World of States of Affairs. Cambridge: Cam- bridge University Press. Austin, J. L. (1979). Philosophical Papers. SPEECH ACT Ed. J. O. Urmson and G. J. Warnock. Oxford: Oxford University Press. THEORY Barwise, Jon, Jean Mark Gawron, Gordon Plotkin, and Syun Tutiya (eds) (1991). Speech act theory accounts for an act Situation Theory and Its Applications, that a speaker performs when pro- vol. 2. Stanford, CA: CSLI Publications. nouncing an utterance, which thus Cooper, Robin (1996). ‘The role of situa- serves a function in communication. tions in generalized quantifiers’. In Since speech acts are the tools that Shalom Lappin (ed.), The Handbook allow us to interact in real-life situa- of Contemporary Semantic Theory. tions, uttering a speech act requires Cambridge, MA: Blackwell Publishers. knowledge not only of the language 65–86. but also of its appropriate use within Cooper, Robin, Kuniaki Mukai, and John a given culture. Perry (eds) (1990). Situation Theory and Its Applications, vol. 1. Stanford, See also: Logical Positivism; CA: CSLI Publications. Ordinary Language Philosophy; Davidson, Donald (1980). Essays on Performative Actions and Events. Oxford: Clarendon Key Thinkers: Aristotle; Austin, J. Press. L; Ayer, A. J.; Grice, H. P.; Husserl, Dretske, Fred (1981). Knowledge and the Edmund; Kant, Immanuel; Ryle, Flow of Information. Cambridge, MA: Gilbert; Searle, John; Wittgenstein, MIT Press. Ludwig

212 SPEECH ACT THEORY

Speech act theory was first developed Although he explicitly deems the by J. L. Austin whose seminal Oxford nature of sentences to be uninteresting Lectures in 1952–4 marked an impor- in his inquiry on apophantic logos, tant development in the philosophy of Aristotle represents the first account language and linguistics. Austin’s pro- of language as action. posal can be viewed as a reaction to Aristotle’s standpoint influenced the extreme claims of logical posi- the study of language for centuries tivists, who argued that the meaning and paved the way for a tradition of of a sentence is reducible to its research on verifiability, but several verifiability, that is to an analysis German and British philosophers which verifies if utterances are true or anticipated a view of language as a false. Austin contended that most of tool to change a state of affairs. The our utterances do more than simply issues of language and conversation making statements: questions and were addressed by Immanuel Kant orders are not used to state something, who anticipated some concepts like and many declarative sentences do not ‘context’ and ‘subjective idealisation’, lend themselves to being analysed in the rules that articulate conversation, terms of their falsifiability. Instead, and the para-linguistic used they are instruments that allow speak- in the accomplishment of speech acts. ers to change the state of affairs. This But it was only at the end of the nine- is tantamount to saying that we use teenth century that a more elaborate language mainly as a tool to do things, treatment of language as action was and we do so by means of performing initiated. hundreds of ordinary verbal actions of The first, although non-systematic, different types in daily life, such as study of the action-like character of make telephone calls, baptise children, language was conducted by Thomas or fire an employee. Reid, who described different acts The fact that not all sentences are a that can be performed through lan- matter of truth verifiability was first guage, and grouped them into two advanced by Aristotle who, in his De categories: ‘solitary acts’ like judge- Interpretatione, argued that: ments, intentions, deliberations and desiring, which can go unexpressed; as there are in the mind thoughts which and ‘social operations’ like command- do not involve truth or falsity, and also ing, promising or warning, which, by those which must be either true or false, their very social nature, must be so it is in speech. [. . .] A sentence is a sig- expressed. Reid’s contribution to the nificant portion of speech [. . .] Yet every inception of a speech act theory can be sentence is not a proposition; only such fully understood if viewed from the are propositions as have in them either wider perspective of the philosophical truth or falsity. [. . .] Let us therefore dis- developments of his time. miss all other types of sentence but the Franz Brentano’s distinction proposition, for this last concerns our between physical and psychological present inquiry, whereas the investiga- phenomena is particularly relevant in tion of the others belongs rather to the this respect because it reintroduced to study of rhetoric or of poetry. (1–4) philosophy the scholastic concept of

213 SPEECH ACT THEORY

‘intentionality’, which allows for a Marty offered the first account of uses distinction between mental acts and of language meant to direct others’ the external world. As far as speech behaviour, like giving an order, act theory is concerned, suffice it here requesting, or giving encouragement. to say that Brentano argued that every Marty stated that sentences may hint mental, psychological act has a con- at the speaker’s psychic processes and tent and is directed at an object (the argued that ‘deliberate speaking is a intentional object), which means special kind of acting, whose proper that mental phenomena contain an goal is to call forth certain psychic object intentionally within themselves phenomena in other beings’ (1908: and are thus definable as objectifying 284). Stemming from Brentano’s tri- acts. The Brentanian approach to partite subdivision of mental phenom- intentionality* allows for a distinc- ena into presentation, judgements, tion between linguistic expressions and phenomena of love and hate, describing psychological phenomena Marty discriminated linguistic forms and linguistic expressions describing into names, statements and emotives non-psychological phenomena. Fur- (utterances arousing an interest), thermore, Brentano claimed that which is a model that closely resembles speaking is itself an activity through Karl Bühler’s Sprachtheorie.Itispre- which we can initiate psychic phe- cisely to Bühler that we owe the nomena. Edmund Husserl picked up coinage of the label ‘speech act theory’. the importance of what Brentano’s He offered the first thorough study of psychological investigation could the functions of language – Darstel- bring to logic*, in particular the con- lung (representation), Kindgabe (inti- trast between emotional acts and mation or expression), and Auslösung objectifying acts. Husserl tackled the (arousal or appeal) – thus endowing issue of human mental activities non-representational sentences with (‘acts’) and how they constitute the their own status. ‘object’ of knowledge through experi- A more complete treatment we find ence. In his Logical Investigations in the work of Adolf Reinach, who (1900/1) he developed a theory of offered the first systematic theory of meaning based on ‘intentionality’ speech acts. Reinach received a doc- which, for him, meant that conscious- torate in philosophy from the Univer- ness entails ‘directedness’ towards an sity of Munich; his dissertation was object. It is on the notion of ‘objecti- on the concept of cause in penal law. fying acts’, that is acts of representa- It was within the context of legal lan- tion, that Husserl shaped his theory of guage that Reinach argued in favour linguistic meaning, thus emphasising of the relevance of speech acts which the referential use of language. Collat- he referred to, presumably independ- erally he treated the non-representa- ently of Reid’s work, as ‘social tional uses of language, that is acts acts, that is acts of the mind that are like asking questions, commanding or performed in the very act of speak- requesting. ing’. Reinach (1913) provided a Following Brentano and moving detailed taxonomy of social acts as within the field of psychology, Anton performative* utterances and their

214 SPEECH ACT THEORY modification, and stated very clearly Oxford linguist philosophers like that the utterance (Äusserung) of a Gilbert Ryle and J. L. Austin, who social act is different from the inner instead were greatly influenced by experience of emotions like anger or Ludwig Wittgenstein. He claimed that shame and from statements (Konsta- a language consists of a wide multi- tierungen) about experiences. It is pre- plicity of structures and usages that cisely the recourse to the physical logical positivists had neglected to medium, the Äusserung, that trans- analyse but which encompass the forms the philosophical category of majority of what human beings say in action into a social act. Drawing on their construction of meaning. previous literature, Reinach separated Following Wittgenstein’s insights actions from internal experiences. into language and putting himself Then he discriminated between exter- against the positivist background, nal actions like kissing or killing and Gilbert Ryle rejected the Cartesian linguistic actions, and within this class mind-body dualism in The Concept of he distinguished between social acts, Mind (1949), and revived the central- which are performed in every act of ity of the standard uses of language, speaking, and actions, where signs are thus contributing to the development used but no speech act is performed of ‘ordinary language philosophy’* in such as in ‘solitary asserting’ and emo- Oxford. tive uses of language. The final dis- Taking the same veil and influenced tinction refers to the linguistic actions by Husserl, Austin rejected the performed in uttering performative account that only sentences that are formulae and the linguistic and non- meant to describe a state of affairs are linguistic actions whose performance worth studying, and he observed that has an effect on the state of affairs and verifiable sentences are only a small even changes it. part of the large amount of utterances While Reinach’s ideas were spread- produced by language users. Not all ing through the Munich scholars, at utterances express propositions: Oxford A. J. Ayer, considered the many perform actions as, for exam- philosophical successor of Bertrand ple, greetings or orders, which resist Russell, deemed philosophically inter- a truth-conditional analysis. Indeed, esting only those sentences that can be most of the sentences uttered by subject to the truth-condition analy- speakers are used in such a way as to sis. In line with the logical positivism* perform more fundamental things in of the Vienna Circle, Ayer developed verbal interactions, such as naming a the verification principle in Language, ship, marrying a couple, or making a Truth and Logic (1936) where he request. In daily life we perform many stated that a sentence is meaningful ordinary verbal actions, and utter- only if it has verifiable import. Sen- ances are used in speech events to tences expressing judgements, evalua- accomplish all that is achieved tion and the like were not to be objects through language. Austin’s speech act of scientific inquiry. This stance, theory was first delineated in the notes which is now known as the ‘descrip- he prepared for some lectures interest- tive fallacy’, led him into conflict with ingly entitled Words and Deeds which

215 SPEECH ACT THEORY he delivered at Oxford University act to be successful, it must fulfil some from 1952 to 1954. Such notes con- appropriateness conditions, or ‘felic- stituted the basis on which he devel- ity’ conditions: locution is successful if oped his Harvard lectures in 1955, words and sounds are correctly pro- posthumously published in 1962. In duced; illocution is appropriate if it the first phase of development of his meets the conditions for its realisation; theory, Austin retained the Aris- perlocution may be effective when it totelian distinction between apophan- produces consequences desired by tic and non-apophantic logos, and the producer. The notion of illocution- introduced the terms of constative ary force embodies the philosophical utterances and performative utter- notion of intentionality, which can ances, where the former describe or be expressed by performing a speech constate a state of affairs and the act through three modalities: (1) latter perform actions. Austin later directly or indirectly through the per- realised that a clear distinction formance of another speech act (‘Pass between the two types of utterances is me the salt’ versus ‘Can you pass me unsustainable. If, for example, we say the salt?’); literally or non-literally ‘There is a rat under your chair’, we depending on the way words are used do more than assert a state of affairs: (‘Stick it in your head’); (3) explicitly we warn someone about a possible or inexplicitly when meaning is spelled danger. Assertions can thus be used to out fully or incompletely (‘I’ll be back perform such acts as to warn, to apol- later, Mary’s ready’). Indirectness and ogise, and many more. Austin then nonliterality are disambiguated by abandoned the dichotomy and con- way of a conversational implicature*, tended that to say something equals to whereas explicitation is achieved perform something. through expansion or completion of According to Austin, when we say what one says. something, we perform three acts John Searle, one of Austin’s stu- simultaneously: a , an dents, contributed widely to develop- , and a perlocutionary ing speech act theory, which he act. At the locutionary level, a speaker addressed from the viewpoint of produces sounds (phonetic act) which intentionality. Specifically he con- are well ordered with respect to the ceived of linguistic intentionality as phonological system and grammar of derived from mental intentionality. In a particular language (phatic act), and his Speech Acts (1969) Searle claimed carry some sense with respect to the that Austin’s ‘felicity conditions’ are semantic and pragmatic rules of that constitutive rules of speech acts to the language (rhetic act). At the illocution- extent that to perform a speech act ary level, he is expressing his intention means to meet the conventional rules by virtue of conventions shared in his which constitute a specific speech act. speech community. At the perlocution- Moving from this approach and ary level, he performs a third act which analysing the act of promising, Searle includes the consequences of his proposed a classification of speech speaking, and he has only limited con- acts into four categories: (1) proposi- trol over them. In order for the speech tional content (what the speech act is

216 SPEECH ACT THEORY about); (2) preparatory condition, what they call ‘illocutionary scenar- which states the prerequisites for the ios’. They are formed by a before, a speech act; (3) sincerity condition (the core, and an after component. If a speaker has to sincerely intend to keep person wants someone to bring him a promise); and (4) essential condition his pen, he can utter a direct speech (the speaker’s intention that the utter- act like ‘Bring me my pen’, which ance counts as an act and as such is to exploits the core component, or he be recognised by the hearer). One of can make his request indirectly Searle’s major contributions to the exploiting either the before compo- theory refers to indirectness, that is nent (‘Can you bring me my pen?’) the mismatch between an utterance where the modal verb ‘can’ points to and an illocutionary force. the hearer’s ability to perform the The interpretation of indirect action, or the after component (‘You speech acts has drawn a great deal of will bring me my pen, won’t you?’) attention. Drawing on H. P. Grice’s where the auxiliary ‘will’ instantiates pragmatics, most scholars assume the after component of the request that some inferential work on the part scenario. Panther (2005) makes the of the hearer is required in order to point that provide natu- identify the speaker’s communicative ral ‘inference schemas’ which are intention and the core question is how constantly used by speakers in such inference can be computed. meaning construction and interpreta- Searle (1975) assumes that the hearer tion. Scenarios may be accessed recognises both a direct-literal force, metonymically by invoking relevant which he understands as the second- parts of them. Indirect requests like ary force, and an indirect-nonliteral ‘Can you open the door?’, ‘Will you force, which is the primary force. Sim- close the window?’, ‘Do you have hot ilarly Dan Gordon and George Lakoff chocolate?’ exploit all pre-conditions (1975) argue that inference rules that for the performance of a request, that they label ‘conversational postulates’ is, the ability and willingness of the reduce the amount of inferential com- hearer, and his possession of the puting necessary to disambiguate an required object. Such pre-conditions indirect speech act. Jerrold Sadock are used to stand for the whole (1974) departs from the inferential speech act category. By means of the hypothesis and proposes ‘the explicit mention of one of the com- model’ by claiming that a speech act ponents of the scenario, it is possible like ‘Can you pass me the salt?’ is for the speaker to afford access to the promptly interpreted as a request and hearer to the whole illocutionary cat- needs no inference. egory of ‘requesting’ in such a way Speech act theory is now receiving that the utterance is effortlessly inter- great attention and valid theoretical preted as a request. With a view to proposals from cognitive linguists. improving Panther’s proposal, Fran- Klaus Panther and Linda Thornburg cisco Ruiz de Mendoza (2007) con- (1998) claim that our knowledge of tends that illocutionary meaning is illocutionary meaning may be sys- directly tied to the notion of Idealised tematically organised in the form of Cognitive Models (ICMs), which are

217 SPEECH ACT THEORY principle-governed cognitive struc- Panther, K. U. and L. Thornburg (1998). tures. Illocutionary scenarios repre- ‘A cognitive approach to inferencing in sent the way in which language users conversation’. Journal of Pragmatics construct interactional meaning rep- 30: 755–69. resentations abstracted away from a Panther K. U. (2005). ‘The role of concep- number of stereotypical illocutionary tual metonymy in meaning construction’. situations. In an indirect request like In F. Ruiz de Mendoza and S.Peña (eds), ‘I fancy going out for dinner’ the Cognitive Linguistics. Internal Dynam- hearer understands the implicated ics and Interdisciplinary Interaction. meaning by relying on high-level sit- Berlin: Mouton de Gruyter. 353–86. uational ICMs – that is, on the Reinach, A. (1913). ‘Die apriorischen generic knowledge that expressing a Grundlagen des bürgerlichen Rechtes’. wish indirectly corresponds to asking In Jahrbuch für Philosophie und for its fulfillment. Thus, it is exactly phänomenologische Forschung 1: the quick and easy retrieval from our 685–847. long-term memory of a stored illocu- Ruiz de Mendoza, F. (2007). ‘High level tionary scenario that allows us to cognitive models: in search of a unified identify the nature of indirectness. framework for inferential and gram- Speech act theory is a thought- matical behavior’. In Krzysztof Kosecki provoking issue which has attracted (ed.), Perspectives on Metonymy. the interest of philosophers of lan- Frankfurt: Peter Lang. 1130. guage and linguists from diverse theo- Ryle G. (1949). The Concept of Mind. retical . Manifold aspects London: Hutchinson. of the theory are being debated such Sadock J. (1974). Toward a Linguistic as the classification of speech acts, the Theory of Speech Acts. New York: Aca- relationship between speech acts and demic Press. culture, and the acquisition of speech Searle J. R. (1969). Speech Acts. Cam- acts by children, which proves how bridge: Cambridge University Press. this area of language research still Searle J. R. (1975). ‘Indirect speech acts’. provides room for developments and In P. Cole and J. L. Morgan (eds), new insights. Syntax and Semantics 3: Speech Acts. New York: Academic Press. 59–82. Primary sources Wittgenstein L. (1953). Philosophical Aristotle (1941). De Interpretatione. New Investigations. Oxford: Blackwell. York: Random House. 38–61. Austin, J. L. (1962). How to Do Things Further reading with Words. Oxford: Oxford University Ayer, A. J. (1936). Language, Truth and Press. Logic. London: Gollancz. Gordon D. and G. Lakoff (1975). ‘Con- Brentano, F. (1874). Psychologie vom versational postulates’. In P. Cole and J. empirischen Standpunkt. Leipzig: L. Morgan (eds), Syntax and Semantics, Duncke and Humbolt. Speech Acts. New York: Academic Marty, A. (1908). Untersuchungen zur Press. 83–106. Grundlegung der allgemeinen Gram- Husserl, E. (1900/1). Logische Unter- matik und Sprachphilosophie. Halle: suchungen. Halle: Nyemeier. Nyemeier.

218 STRUCTURALISM

Reid, T. (1894). The Works of Thomas Jakobson, Roman; Sapir, Edward; Reid. Edinburgh: Maclachlan and Saussure, Ferdinand de; Whorf, Stewart. Benjamin Lee

Annalisa Baicchi The term structure is derived from Latin structura (from struere, ‘to build’) and just as human beings build houses, so structuralism contends that STRUCTURALISM human existence – the physiological and mental set-up of individuals and their social life – is also built from A theoretical and methodological structures in a way that more or less approach in linguistics and other governs what people are able to think human (including social) sciences that and do. In addition to the limitations attempts to gain insights into its sub- of the laws of physics and rules of ject matter by assuming that every- social behaviour, structuralism also thing to do with human beings is built maintains that less overt structures of more or less autonomous systems as restrict psychological and behavioural relations of oppositions. These oppo- alternatives by controlling individu- sitions may be of different types but in als’ preferences. In linguistics struc- general are binary relations. The Swiss turalism is affiliated with the so-called linguist Ferdinand de Saussure is often Saussurean ideas about language and said to have initiated a structuralist other sign-systems (dealt with in movement, school or intellectual ‘semiology’, later ‘semiotics’) and it world view, rather than developing may be most easily understood in the a coherent theory, and in linguistics conceptual framework that is attrib- a distinction is traditionally made uted to Saussure. between (Saussurean) European struc- As opposed to the use of language turalism and American structuralism, (French parole), there is a system of the main figure of which is the linguist languages and language (langue) that Leonard Bloomfield. Structuralism in is a set of inherent relations that build a broad sense has mainly been applied a structure. In order to arrive at an in anthropology, especially by the exhaustive and consistent description French anthropologist Claude Lévi- of this system, one has to assert that Strauss and other French thinkers, and the description is historically specific: in literary studies. characteristic of an abstraction from See also: Distinctive Features; language use at a certain time and Glossematics; Langue/Parole; place. In other words, it produces Poststructuralism; Sense/Reference; what was labelled a synchronic Signs and Semiotics; description of the system. This is Transformational-Generative opposed to the view prevalent in the Grammar nineteenth century that the history Key Thinkers: Barthes, Roland; and the genealogy of languages was Boas, Franz; Bloomfield, Leonard; the only (legitimate) theme in the lan- Derrida, Jacques; Hjelmslev, Louis; guage sciences; the results yielded in

219 STRUCTURALISM this tradition was labelled by struc- system. Phonology is the study of turalists as diachronic description. It phonemes*: that is, abstract entities is a general experience that words are identified as generalisations of speech put together in chains, and a basic sounds and building a mental system notion in structural linguistics is that of oppositions that makes the individ- of syntagm. But since this is a matter ual language user able to decode the of language use, the corresponding sound chains as words in a language. notion of paradigm is often more The enterprise of setting up such interesting for structuralists. A para- phonological systems is often a fairly digm is not a discernible and evident straightforward project, but the entity like a syntagm because we can analysis of word meanings is a more only identify a paradigm by abstrac- challenging task. Some approaches in tion and experiment: by playing with semantics can be regarded as con- the words of a sentence by substitut- ceived within the framework of struc- ing them. In that way we learn that turalism, for instance the notions of paradigms can be said to be sets of semantic fields and semantic compo- words that can replace each other on nents describing the features creating certain positions in chains of words, the basis of semantic oppositions. words that accordingly must be dif- The methods of structuralism make ferent in one respect and similar in no consistent conceptual framework. some other respect. This potential of Phonology investigates the sound words is said by structuralists to be a systems of a language or a dialect in matter of how their properties are order to find the distinctive sound fea- structural and can be described as tures that separate words. But some such. linguists take different methodologi- The basic opposition, then, is that cal approaches. For example, Louis between the form of a word, the signi- Hjelmslev, the inventor of glossemat- fier (signifiant), and its meaning, the ics*, started out by dividing a text into signified (signifié). According to Saus- two and he continued like this until he sure, this relation is arbitrary, a claim ended up with the phonemes. One of that has caused some controversy. But the basic problems of structuralism is one might say that the relation is, in what answers are given to the question principle, arbitrary while it is, evi- about the nature of meaning. There is dently, not historically arbitrary since obviously a difference between word we use words in – almost – the same meaning and the meaning of life, even way as we have experienced them though appeal to word meanings as when acquiring our mother tongue. oppositional relations may not be From this point of departure, both unequivocally convincing and even form and meaning can be subject to though the question of the meaning of structural scrutiny. The extremes of general, abstract and fictional expres- such scrutiny are, on the one hand, sions may not be straightforwardly phonology and, on the other, semantic answered. This vagueness in what orderings of the sense and meanings meaning is may be the background of of words and their mutual relations the diffusion of the concept of struc- within the whole of the language turalism into other academic fields

220 STRUCTURALISM such as anthropology and literature, It may seem ironical that Jakobson, and its popularity in semiotics. a linguist and passionate reader of Although it is often stated as fact, Russian poetry, with a background the attribution of the idea of struc- affiliation with Russian , turalism in linguistics to Saussure may was to become the main inspiring not be totally justified. The source of force of European structuralism, the what we know about his thoughts Prague School (which was, with is Cours de Linguistique Générale the Geneva School, the centre of (Course in General Linguistics) from European structuralism), and of 1916, a work that is based on the lec- American structuralism. Passing ture notes of some of his students. Fur- through Europe and ending up in the thermore, Saussure may be considered United States, he and other Russians only part of an emerging movement in contributed significantly to what has linguistics that could not reconcile become known as the Prague School itself to nineteenth-century positivism. of structuralism. Fleeing from World Finally, the term structuralism had War Two, he visited Denmark and been used by psychologists in the nine- influenced the glossematic version of teenth century but it was not used by structuralism, and in the United States linguists before the end of the 1920s; he inspired anthropology and other Roman Jakobson probably was the sciences, at the same time more or less first to offer a definition of the theoret- directly influencing Chomsky’s ideas ical concept. The expressions ‘struc- about linguistic universals. ture’ and ‘structural’ in discussing European structuralism was not linguistic phenomena were also used in only a countermovement with respect the first half of twentieth century. to positivism but after World War One Nevertheless Saussure is usually it was also an opportunity to find an considered the founding father of lin- alternative to the dominance of the guistic structuralism, and his thoughts nineteenth-century German neogram- as they are presented in Cours have marians and their ‘sound laws’. Most had a considerable impact on general of the characteristics of structuralism structuralist thinking; structuralism mentioned above apply to European reached its peak as an -ism between structuralism, and a number of other 1930 and 1960. The broad movement features can be mentioned. Structural- of structuralism was not a unified ism deals with systems (a word that endeavour but rather a patchwork of can, in this context, be taken as a syn- different groupings with different onym for structure; structuralists who goals, different basic assumptions and talk about ‘structured systems’ use a different kinds of subject matter. In pleonastic term), and knowledge retrospect, European structuralism about these systems (the entities of can be said to encompass some more which are considered at least as real as important groups of people and some observable entities) are arrived at more peripheral groupings and indi- through abstraction and analysis. The viduals, while one person in particular systems are also regarded as social. is a travelling herald of the structural- Even though Saussure’s basic assump- ist message: Roman Jakobson. tion about the linguistic sign was that

221 STRUCTURALISM it is a entité psychique (Saussure 1916: conceiving of them as ‘grammars’ 99), or psychological entity, in general intrinsic to the human mind. Whereas structuralists maintain that they talk Lévi-Strauss studied the built-in mean- about social phenomena, albeit stud- ing of tribal behaviour, other French ied through analysis of what people structuralists like and say. Indeed this may be a necessary Jacques Derrida have applied struc- theoretical prerequisite if one wishes turalist thought to literature; the to avoid philosophical inquiry into the French intellectual Jacques Lacan did problem of what is private, what is the same to psychoanalysis and the public, and what role language plays French philosopher Louis Althusser in making the private public. founded a ‘Marxist structuralism’. In European structuralism has had Europe, therefore, structuralism as a wide-ranging consequences for the broad notion is nowadays mostly asso- sometimes not particularly explicit ciated with French structuralism, and basic assumptions of all disciplines of while the structural mind in twenty- modern European linguistics. This is first century Europe is most likely not especially true in phonology, where a member of a club named ‘structural- any textbook on and ism’, structuralist beliefs are almost speech sounds takes up the phoneme in many theoretical, system of the language in question. often implicit, basic assumptions. This involves ascertaining the distinc- In contrast to the European struc- tive, contrastive features of minimal turalistic vein – which was linguistic pairs (of words) in order to identify in its outset and all the time concerned single phonemes, and, following this at some level with linguistic meaning – procedure, the objective of the process American structuralism tended to is to find all the phonemes of a single ignore meaning and focus on linguis- language. This modus operandi is tic form, while also in general main- now almost a standard method in taining so-called linguistic relativity*. phonology, whether phonologists per- Both attitudes can be explained in ceive themselves as structuralists or terms of the background of American not. linguistics. The United States was Phonology is a fairly technical disci- built on a mixture of immigrants and pline in linguistics, but structuralist their descendants whose cultures had ideas have not confined themselves to long philological traditions, while intricate linguistic phenomena: they also comprising the ethnic Native have also been adopted by scholars in American cultures, the anthropologi- fields outside linguistics. In anthropol- cal and linguistic documentation of ogy, one of the main figures is Claude which was an immense task. It may Lévi-Strauss, who met Jakobson in therefore seem natural that describing New York, and who attempted to each language on its own terms would show that the myths and rituals of be a reasonable goal, and this view, in tribal cultures work as regulating kin- combination with the low priority ship systems and other social institu- given to semantics, logically gave rise tions. He also analysed them in line to the fairly extreme idea that the way with structural linguistic analysis people think depends on the structure

222 STRUCTURALISM of their language, the so-called Sapir- that goes back to Boas and the (basi- Whorf hypothesis. cally pragmatic) idea that human Franz Boas is regarded as the beings communicate in sentences, not founder of American structuralism. by using words in isolation. The other An anthropologist untrained in lin- thing is the fact that, contrary to gen- guistics, he contributed notably to eral opinion, Bloomfield did not the description of the phonological totally abandon linguistic semantics and grammatical structures of from his linguistic approach. For Native American languages. Boas instance, he appeals to a special kind incremented the prestige of language of meaning: studies as a part of anthropological studies, which traditionally included Since our study ordinarily concerns and cultural and physical only the distinctive features of form and anthropology. But he was also an meaning, I shall henceforth usually advocate of relativism, a view that he omit the qualification linguistic or dis- passed on to his students and follow- tinctive, and speak simply of forms and ers, and, in line with this, one of his meanings, ignoring the existence of main concerns was to promote the non-distinctive features. A form is often basic methods of fieldwork. If lan- said to express its meaning. (Bloomfield guages are more or less self-contained 1935: 141) entities, it sounds reasonable to call for careful and detailed investigation This is also by necessity consistent into each particular language in order with his hierarchy of categories which to offer exhaustive descriptions. And would otherwise have been impracti- this may be one of the only reasons for cal because the notions ‘meaningful’ the predicate structuralism to this and ‘meaningless’ are necessary theo- stage of American structuralism: what retical concepts for the distinctions was accounted for through linguistic between the classes of linguistic units, fieldwork were the internal structures for instance in the expression the of particular languages. ‘smallest meaningful unit . . .’. The same characteristic may be But some of Bloomfield’s followers attributed to the most prominent in American structuralism took it fur- American structuralist, Leonard ther away from semantics and linguis- Bloomfield. In his principal theoretical tic meaning, one of them being Zellig work, Language (1935), the words Harris, whose principal work actually ‘structure’ and ‘structural’ are not fre- has the word ‘structural’ in its title. quently found. The book offers more And these two themes, the meanings of a methodological account of cate- of the words ‘structuralism, structural gories and their hierarchies than an and structure’, and the question of all-encompassing and consistent whether semantics can be disposed of, theory, but there are two features that are being transferred to the modern are worth mentioning. In opposition world’s most famous linguist, Noam to European structuralism, Bloomfield Chomsky. It may seem ironical that and American structuralism empha- Chomsky’s programmatic work, Syn- sise sentences as linguistic units, a bias tactic Structures (1957), dealt with

223 STRUCTURALISM formal (mathematical in a broad phenomena they study. Saussure and sense) descriptions of phrase struc- his followers see language as words, tures, accordingly using the term primarily their phoneme structure and ‘structure’ while explicitly abandon- their semantics, while Chomsky sees ing the idea of meaning having any language as phrase structures. role to play in this formal approach to One of the distinctions set up by linguistics. The position is repeated in some structuralists in linguistics is New Horizons in the Study of Lan- that between form and substance, guage and Mind: meaning that, for instance, the physi- cal nature of speech sounds studied in As soon as the first attempts were made phonetics is interesting but secondary to provide actual descriptions of lan- to the important effort of finding and guages forty years ago, it was discov- modelling the phonemes – as mental ered that the intricacy of structure is far and material entities – as the form of beyond anything that had been imag- language systems. This illustrates a ined, that traditional descriptions of fundamental problem with structural- form and meaning merely skimmed the ism, that some of its theoretical surface while structuralist ones were notions are extremely general, almost almost irrelevant. (2000: 122) universal. Likewise, there is a certain kind of vagueness in the terms ‘form’ The outcome of the structuralist proj- and ‘structure’. From their classical ect will find its place (cf. Chomsky origin they are the heirs of words that 2000: 5) in this the latest theoretical did not have different meanings. paradigm, labelled generative gram- ‘Form’ and ‘structure’ meant almost mar, in which the term ‘structure’ is a the same, and in some contexts they basic and axiomatic one, and in which still do. In formal logic, for instance, the unearthed structures are assumed the raison d’être of arguments is that to be mental, but have nothing to do they have a certain structure, while in with meaning. On the one hand, other contexts, for instance in archi- Chomsky seems to carry further some tecture, it may be reasonable to distin- of tenets of American structuralism guish between the form (shape and and, on the other, he dissociates him- surface) of a building and its inner self from its basic theoretical ideas structure. So basically, the proposal of and methods. In a way he is back at structuralism seems to be no more the starting point of Saussure (maybe than the idea that things in the world not as a historical figure but as the are ordered in ways that make them icon of European structuralism), and more than silhouettes in a nebulous in another way he is not. He says, like landscape. As such, structuralism has Saussure, that he studies mental (what pervaded modern Western civilisa- Saussure calls psychological) phenom- tion. ena, and, like Saussure, he does so as abstractions from mental products: Primary sources words. In this respect they put Benveniste, Emile (1939). ‘Nature du forward comparable ideas. Where signe linguistique’. Acta Linguistica they part is the question of which Vol. I, fasc. 1: 23–9.

224 SYSTEMIC-FUNCTIONAL GRAMMAR

Bloomfield, Leonard (US 1933, UK 1935). Newmeyer, Frederick J. (1998). Language Language. London: George Allen and Form and Language Function. Cam- Unwin. bridge, MA: MIT Press. Boas, Franz (ed.) (1911, 1922, 1934). Handbook of American Indian Lan- Hans Götzsche guages, Parts 1, 2, 3. Washington, DC: Smithsonian Institution, Bureau of American Ethnology. Bröndal, Viggo (1939). ‘Linguistique SYSTEMIC- structurale’. Acta Linguistica Vol. I, FUNCTIONAL fasc. 1: 2–10. Chomsky, Noam (1957). Syntactic Struc- GRAMMAR tures. The Hague. Paris: Mouton. Chomsky, Noam (2000). New Horizons An approach to in the Study of Language and Mind. which aims to provide a comprehensive Cambridge: Cambridge University account of how language is used in con- Press. text for communication. The approach Harris, Zellig (1951). Methods in Struc- views language as a resource that is fun- tural Linguistics. Chicago: University of damentally shaped by the uses that Chicago Press. people make of it; it therefore aims to Jakobson, Roman (2002). Selected Writ- explain the forms of language in terms ings Vol. I–VIII. Third edition. of the meanings that they express, and Berlin/New York: Mouton de Gruyter, to develop a grammar which is 1962–88. designed to ‘make it possible to say sen- Lévi-Strauss, Claude (1958). Anthropolo- sible and useful things about any text, gie structurale. Paris: Plon. spoken or written’ (Halliday 1994: xv). Saussure, Ferdinand de (1916). Cours de See also: Corpora; (Critical) Linguistique Generale. Paris: Payot. Discourse Analysis; Zwirner, Eberhard (1939). ‘Phonologie Integrationism; Langue/Parole; und Phonetik’. Acta Linguistica Vol. I, Metaphor; Modality; fasc. 1: 29–47. Transformational-Generative Grammar Further reading Key Thinkers: Halliday, M. A. K.; de Beaugrande, Robert (1991). Linguistic Bernstein, Basil; Firth, J. R.; Theory: The Discourse of Fundamental Hjelmslev, Louis; Malinowsky, Works. New York: Longman. Bronislaw; Sinclair, John; Whorf, Huck, Geoffrey J. (1995). Ideology and Benjamin Lee Linguistic Theory. Noam Chomsky and the Deep Structure Debates. London Systemic-functional grammar (SFG) and New York: Routledge. originated with M. A. K. Halliday, Koerner, E. F. K. and R. E. Asher (eds) building especially on the ideas of his (1955). Concise Histrory of the Lan- tutor J. R. Firth, in publications from guage Sciences. From the Sumerians to the 1960s on, with major contributions the Cognitivists. Oxford: Elsevier Sci- by other scholars such as Ruqaiya ence, Pergamon. Hasan and, in more recent years, Jim

225 SYSTEMIC-FUNCTIONAL GRAMMAR

Martin and Christian Matthiessen be represented as a system, labelled among many others. From early in its determination (names of systems are development, SFG has had two main traditionally written in small capitals distinguishing features, which are in SFG). The entry condition (the lin- reflected in the name. First, whereas guistic context in which the choices many approaches focus on the syntag- apply) is ‘ group’; the first matic, ‘horizontal’ dimension of how choice (at least in English) is between constituents may be combined with ‘specific’ (‘the [cat]’) and ‘non-specific’ other constituents in a well-formed (‘a [cat]’). Each option taken opens up structure, SFG prioritises the choices a further set of choices until a formal that are open to the speaker at any par- realisation is reached: for example, ticular point in an utterance – the para- selecting ‘specific: personal: interac- digmatic, ‘vertical’ dimension. The tant: addressee’ leads to the deictic grammatical structures are then seen as (determiner) form ‘your [cat]’, whereas the outcome of choices from those selecting ‘specific: demonstrative: available (the technical term in SFG is selective: near ϩ plural’ leads to ‘these that structures ‘realise’ choices). Sets of [cats]’. As this last instance shows, choices between options can most eco- some sets of choices in the system may nomically be shown in the form of sys- be simultaneous: that is, rather than tems: for example, ‘if A is the case, only choosing one of two or more there is a choice between B and C; if B mutually exclusive options, the is chosen, there is then a choice speaker chooses from two sub-systems between D, E and F; but if C is chosen, at the same level. Thus, taking the there is then a choice between G and ‘selective’ option means choosing both H’. Systems embody the Saussurean between ‘near’ and ‘far’ and between concept of valeur: a linguistic form has ‘plural’ and ‘non-plural’. Part of the meaning by virtue of the other possible system is shown in Figure 7 (three dots forms that could have been chosen indicates where more delicate choices instead. Hence this is a ‘systemic’ have been omitted). Simultaneous grammar. Second, the model is ori- choices are enclosed by a curly bracket; ented primarily towards meaning and the formal realisations (in this case rather than form: that is, its aim is to specific words rather than general describe how wordings are used in structures) are signalled by downward- expressing meanings. What a linguistic slanting arrows. A complete version of form consists of is seen as less impor- the system can be found as Figure 6-2 tant than the function that it performs in Halliday and Matthiessen (2006: in the clause: hence, this is a ‘func- 313). tional’ grammar. The following para- Systems do not operate in isolation: graphs expand on these distinguishing they interact with each other. For features in turn. example, the system of polarity The systemic nature of the grammar (positive/negative) interacts with a can be illustrated with a relatively number of other systems, including simple example. The choices at nomi- determination: here, a combination of nal group level between different kinds ‘negative’ with ‘non-specific: total’ of determiner (such as ‘the’ and ‘a’) can gives the deictic ‘no’ as in ‘no [cats]’.

226 SYSTEMIC-FUNCTIONAL GRAMMAR

As relatively simple systems build up inal group is being determined, or into system networks in this way, the specified, in terms of possession or of complexity increases, but that reflects location in relation to the speaker. faithfully the complexity of the mean- Even traditional labels such as subject ing choices that are realised in any are reinterpreted in SFG as not being utterance. purely syntactic (‘controlling’ agree- The other key feature of the gram- ment with the verb) but as having mar is its functional nature. In con- meaning: the subject expresses the trast to form-oriented approaches entity which is represented as respon- which concentrate more or less exclu- sible for the validity of the proposition sively on how clauses are constructed, expressed in the clause (that is, the with meaning excluded and dealt with proposition is represented as true for, separately in a semantics component, or applicable to, this entity). SFG is designed to map the relation- This connects with a second impli- ships between forms and meanings in cation: whereas many approaches are a consistent way. This has a number of parsimonious in that they place great important implications. First, the most value on achieving maximally eco- important labels are functional, or nomical descriptions of grammar, SFG semantic (telling us what the elements is ‘extravagant’ in Halliday’s term. In mean), rather than formal (telling us order to capture the multivariate rela- what they consist of). This can be seen tions between meaning and structure, in the terms used in Figure 7: they indi- the grammatical model has to allow cate what meaning choice is being for different perspectives on the made: for example, whether the nom- clause, providing a much richer and

speaker Deictic: ‘my’ interactant speaker-plus Deictic: ‘our’ personal addressee (possession) Deictic: ‘your’

non-interactant … specific

non-selective Deictic: ‘the’

demonstrative non-plural nominal Deictic: (proximity) group ‘this’ plural Deictic: selective ‘these’ near Deictic: ‘that’ far Deictic: ‘those’ non-specific …

Figure 7 Part of the system of DETERMINATION in English

227 SYSTEMIC-FUNCTIONAL GRAMMAR more informative set of descriptions. to support this claim). Thus, mood This then leads on to a third important (declarative, interrogative, imperative) implication: since the communicative interacts with polarity in that, for function of utterances can only be example, the position of the negative fully understood in relation to their particle ‘n’t’ varies in relation to the meaning in context, the grammar has subject with different mood choices; to be designed in such a way that the but transitivity roles (see below) are analyst can ‘shunt’ between specific unaffected by whether the clause is choices at clause level or below and the declarative or interrogative. The three context within which the utterance is groupings of systems make up what used. For example, an interrogative Halliday calls three metafunctions, such as ‘Why aren’t you leaving?’ may each of which realise different broad function as a question, but in certain types of meaning. contexts it may instead function as a The interpersonal metafunction command. The ‘context’ here would comprises those systems which func- include who the interactants are, what tion to enact social relations between their relationship is, how power is dis- addressers and addressees, to express tributed in the culture (who has the the speaker’s viewpoint on events and right to give commands to whom in things in the world, and to influence particular circumstances), and so on. the addressee’s behaviour or views. In many approaches, the use of the Apart from mood and polarity, some interrogative as a command would be of the main systems are modality*, treated as in some way an extension of mood tag, and, at the semantic level, its ‘literal’ decontextualised use as a speech function (whether the clause question and dealt with separately as a functions as statement, question, com- matter of pragmatics; but, in SFG, the mand or offer). This is the area of the analyst would take all these factors grammar in which differences in mean- into account in arriving at a full gram- ing such as the following are captured: matical characterisation of the utter- ance, exploring the reasons why a The flight is confirmed. (declarative, particular meaning expressed in a par- statement) ticular way at a particular point in an Is the flight confirmed? (interroga- interaction is likely to have appeared tive, question) appropriate to the speaker in a partic- Confirm the flight. (imperative, com- ular situation. mand) The two key bases of the model, Would you confirm the flight? system and function, come together in (modalised interrogative, command) the concept of metafunctions. Halli- The flight mustn’t be confirmed. (neg- day argues that the system networks ative modalised declarative, command) that can be identified fall into three main groupings, with interaction As the examples show, SFG highlights between systems in any one group, but the crucial distinction between mood little or no interaction between the sys- and speech function: for example, not tems across groupings (Matthiessen all interrogatives function as ques- 2006 provides corpus-based evidence tions. The choice of forms may seem

228 SYSTEMIC-FUNCTIONAL GRAMMAR to be imposed naturally by the con- 2006 expand on this relation be - text. For example, if you want infor- tween grammar and cognition): mate- mation you use an interrogative rial (processes of action), mental question; but in fact speakers in dif- (processes of sensing, which con- ferent contexts consistently exploit strue our interior worlds), relational choices. There is, for instance, an (processes of being and having), verbal important difference between express- (processes of conveying messages, by ing a command through the impera- saying, and so on.), behavioural (char- tive or the modalised interrogative acteristically human physiological wordings shown above; and there is processes), and existential (processes also a choice between the interroga- of existing). Each of these types can be tive question ‘Is the flight confirmed’ further subdivided into more delicate and the declarative question ‘The options: for example, one subdivision flight is confirmed?’. These choices of material processes, the largest depend on, and simultaneously reflect group with the greatest number of the speaker’s view of, the context of cross-cutting options, is between ‘cre- utterance. A basic tenet of SFG is that ative’ (a process which results in a any difference in wordings is the result new entity, such as ‘build’) and trans- of meaningful choice. formative (a process which involves a The second metafunction is the pre-existing entity, such as ‘cut’). The experiential. This is language seen participants in each process type are from the perspective of how it is used given different labels which reflect to talk about events, states and entities their relation to the process. There is in the world, to construe the speaker’s no space to give a full account, but view of the world. This kind of mean- Table 2 gives a flavour of transitivity ing is traditionally taken – even in analysis. many linguistic approaches – as ‘real’ It is worth highlighting that transi- meaning. In SFG, however, it is only tivity may be ‘blind’ to certain parts one kind of meaning: in simple terms, of the clause. The clause in Table 3 why something is said (the interper- has the same transitivity configuration sonal angle) and how it is said (the tex- as the first example in Table 2. tual – see below) is as important as The unlabelled constituents are dealt what is said (the experiential). An with in another part of the grammar (in experiential analysis of a clause the example above, they are interper- focuses on the process (realised by the sonal in nature). Transitivity labels are verb), the participants in that process semantic, but it is important to note (typically, but not always, realised by that each corresponds to different nominal groups), and the circum- grammatical possibilities. Halliday stance(s) in which the process happens (1994: xx) stresses that ‘all the cate- (typically realised by adverbial groups gories employed must be clearly or prepositional phrases). Processes “there” in the grammar of the lan- can be divided into six main types, guage’. There is only space to mention which reflect the cognitive categories two of these grammatical reflexes as that we use to make sense of the events examples: material processes cannot around us (Halliday and Matthiessen project (that is have a ‘that’ clause as

229 SYSTEMIC-FUNCTIONAL GRAMMAR

The man drove the car fast Actor Process: material, Goal Circumstance transformative

I recognised the driver Senser Process: mental, cognition Phenomenon The man was a fast driver Carrier Process: relational, Attribute attributive

Table 2

Might the man have driven the car too fast perhaps ? - Actor - Process: Circumstance material Goal

Table 3 their complement), whereas mental and is theme), which signal that there is verbal processes can; mental and behav- some kind of specific contextual pres- ioural processes must normally have sure at work, often associated with a human sensers/behavers, whereas mate- change of textual frame. For example, rial processes have no such restriction. in the following extract from a The third metafunction is the tex- guide to a historical monument, the tual. This is the part of the grammar choice of themes (underlined) is which is concerned with how the designed to help the reader to follow meanings in the clause are organised the text organisation. The marked to fit in with the co-text of the sur- themes (in italics) signal that the text rounding messages and with the is moving on to deal with a different wider context of the utterance. The part of the church, whereas the textual system that has been most unmarked themes signal that the text fully explored is that of theme: the is, for the moment, continuing to theme of a clause is the initial experi- focus on the same part of the church ential constituent, which has a special introduced in the preceding marked role in signalling how the current theme. clause relates to clauses around it. Very broadly, there is a choice On the northern side of the church between unmarked themes (where was a porch . . . Further east, on the theme and subject are the same), west side of the north transept was a which typically signal continuity of room . . . The east wall of the room was some kind in the topic, and marked divided into three parts . . . Its use is not themes (where something other than known for certain, but it may have been subject, such as an adverbial adjunct, a vestry and sacristy . . .

230 SYSTEMIC-FUNCTIONAL GRAMMAR

Rise in radiation exposure leads to warning Actor Process: material Circumstance

Table 4

One factor which complicates the used in that way in some texts); it already complex three-dimensional allows new kinds of meanings to be picture above is grammatical meta- made. In the headline, ‘radiation’ is an phor, which has been identified in instance of metaphor, but it is SFG as a crucial mechanism by which extremely difficult to unpack it: it has language can expand the potential become an accepted term for referring meanings that can be expressed. in a condensed form to a complex phe- Grammatical metaphor* involves nomenon. Grammatical metaphor is the use of wordings which do not often ignored in other grammatical map in a straightforward way onto approaches; but, since SFG is a gram- the meanings that they express. This mar of discourse, it has a central place can be seen in the following news - in the model. paper headline: ‘Rise in radiation SFG is the most fully-developed exposure leads to warning’. A transi- alternative to what was, for much of tivity analysis of this clause is in Table the twentieth century, the main lin- 4. guistic paradigm, which aimed to split This captures part of the meaning as the problem of describing language it is expressed; but it is clearly possible into separate areas such as syntax, to unpack the clause into more ‘con- semantics, sociolinguistics, and so on. gruent’ wordings which reflect more SFG rejects that kind of ‘divide-and- naturally the events in the world that conquer’ solution, on the grounds that are being represented. One possi- language is a resource for social com- ble wording (some of the ‘missing’ munication and can only be properly information is supplied from the news- understood if that whole picture is paper article itself) would be ‘Because taken into account at all stages of [= ‘leads to’] doctors are exposing [= investigation. Its orientation to lan- ‘exposure’] patients to radiation [= guage in use means that it has ‘radiation’] more often [= ‘rise’], been widely adopted in discourse researchers are warning [= ‘warning’] analysis* and corpus linguistics and that . . .’. This is a much less econom- in a range of other areas, most ical way of representing the meaning, notably education, natural language but it is also more natural, in that it is generation and language acquisition easier to process in speech and easier studies. for a non-specialist to understand. Grammatical metaphor of this kind is Primary sources particularly associated with formal Halliday, M. A. K. (1978). Language as technical writing. It is not just an Social Semiotic: The Social Interpreta- impressive-sounding way of saying tion of Language and Meaning. simple things (though it is certainly London: Arnold.

231 Halliday, M. A. K. (1985, second edition TRANSFORMATIONAL- 1994, third edition, with Christian Matthiessen, 2004). An Introduction GENERATIVE to Functional Grammar. London: GRAMMAR Arnold. Halliday, M. A. K and Ruqaiya Hasan (1976). Cohesion in English. (English Any formal description of language Language Series 9). London: Longman. consisting of an algorithm which gen- Halliday, M. A. K. and Christian M. I. erates sentence structures and of a set M. Matthiessen (2006). Construing of transformations which modify Experience Through Meaning: A Lan- them systematically. In particular the guage-based Approach to Cognition. theory developed by Noam Chomsky, Second edition. London and New York: whose goal it is to account for the Continuum. implicit knowledge of language inher- Halliday, M. A. K. (2002–). Collected ent in the human mind by means of a Works of M. A. K. Halliday. Ed. formalised system of rules. Transfor- Jonathan Webster. Ten vols. London mational-generative grammar claims and New York: Continuum. to produce all and only the possible Martin, J. R. (1992). English Text: System sentences of a language determined by and Structure. Amsterdam and Philadel- intuition* and the evaluation by com- phia: John Benjamins. petent native speakers and to account Matthiessen, Christian M. I. M. (1995). for language acquisition and speakers’ Lexicogrammatical Cartography: Eng- knowledge of language. lish Systems. Tokyo: International Lan- See also: Acceptability/ guage Sciences Publishers. Grammaticality; Behaviourism; Empiricism/Rationalism; Further reading Innateness; Language of Thought; Butler, Christopher S. (2003). Structure Logical Form; Minimalism; and Function: A Guide to Three Major Universal Grammar Structural-Functional Theories. Two Key Thinkers: Chomsky, Noam; vols. Amsterdam: John Benjamins. Fodor, Jerry; Katz, J. J.; Montague, Eggins, Suzanne (2004). An Introduction Richard to Systemic . Second edition. London: Pinter. Transformational-generative gram- Matthiessen, Christian M. I. M. (2006). mar has been one of the most influen- ‘Frequency profiles of some basic gram- tial linguistic theories since the matical systems: an interim report’. In publication of Chomsky’s (1957) Syn- Geoff Thompson and Susan Hunston tactic Structures. It is a systematic, (eds), System and Corpus: Exploring objective, scientific formalisation of Connections. London: Equinox. 103–42 grammar, based on the belief that the Thompson, Geoff (2004). Introducing structure of language is determined Functional Grammar. Second edition. by the structure of the human mind, London: Arnold. that all languages share some common, universal characteristics Geoff Thompson and that the species-specific creativ-

232 TRANSFORMATIONAL-GENERATIVE GRAMMAR ity* of human language – that is, the sentences of the language under analy- capacity of all native speakers to sis’ (1957: 11). Furthermore such a produce and understand an infinite theory ‘is to provide a general method number of sentences that they have for selecting a grammar for each lan- never heard before – must be guage’. In order to achieve this aim, accounted for. Furthermore, such a Chomsky relied heavily on features theory should also reflect processes of that are important for the success of language acquisition and language theories in physics, and thus arguably use. Transformational-generative established linguistics as a science. grammar has been constantly devel- After rejecting linear finite state oped into more powerful theories, grammars and after showing the mainly by Chomsky himself, based on limitations of phrase-structure (PS) suggestions and criticism from many descriptions, a formalised version scholars, perhaps the best known of of immediate constituent analysis, whom are , J. J. Katz, and Chomsky developed a grammar with Richard Montague. Chomsky’s views a tripartite structure. It consisted of have always been controversial, but a set of phrase-structure rules, a set they cannot be ignored. of morphophonemic rules that con- In Syntactic Structures, the first of vert sequences of morphemes into Chomsky’s continually developing sequences of phonemes, and an inter- models, Chomsky declared that mediate level of transformations mod- syntax is a completely autonomous ifying the output strings of elements part of language, independent of generated by the PS-rules into strings semantics and the phonological that can serve as the input to the mor- system. Furthermore he proclaimed phophonemic rules. With a generative that the syntax of a language can be component generating the underlying formalised in a mathematically pre- structures and a transformational cise way by means of different rules component modifying them into sur- operating on different levels. His pro- face structures, this approach was the gramme was revolutionary in two birth of what was to be known as ways. First, it was an attempt to for- transformational-generative gram- malise at least some of the features of mar. language; and, second, Chomsky dis- The PS-rules given in Chomsky sociated himself from the prevailing (1957: 111), which generate the idea that a grammatical description of underlying structures, are the follow- a language can only be derived from ing (slightly modified): (1) ⌺ → NP ϩ the observation and analysis of actu- VP, (2) VP → Verb ϩ NP, (3) NP → → ϩ ally occurring data. For Chomsky {NPsing / NPpl}, (4) NPsing T N, (5) → ϩ → ‘[s]yntax is the study of the principles NPpl T N + S, (6) T the, (7) N and processes by which sentences are → {man, ball, etc.}, (8) Verb → Aux + constructed in particular languages. V, (9) V → {hit, take, walk, read, etc.}, Syntactic investigation of a given lan- (10) Aux → C(M) (have + en) (be + guage has as its goal the construction ing), (11) → {will, can, may, shall, of a grammar that can be viewed as a must}. ⌺ stands for sentence, NP for device of some sort for producing the noun phrase, VP for , Ɵ

233 TRANSFORMATIONAL-GENERATIVE GRAMMAR for zero, T for the definite article, N face structure. In contrast to PS-rules, for noun, V for verb, S for the plural transformations are not applied to , Aux for auxiliaries, M for single elements but to phrase markers, modals, and C is the element that will the structural description of the later be interpreted by a transforma- underlying structure generated by the tional rule as concord; en stands for PS-rules. In other words, a transfor- the past participle morpheme and ing mational rule takes one structure as its accounts for the progressive form. input and modifies it into another Wavy brackets indicate a set of possi- structure. bilities from which one must be Formally, transformations consist chosen, and round brackets stand for of two components: the structural optionality. Thus rule (10) states that description, which serves as the con - C must be chosen and that we may dition for the transformation to be choose zero or more of the parenthe- applicable, and the structural change, sised elements in the given order, which describes the output. As an resulting in eight different possibilities example of an optional transforma- to rewrite Aux. tion, consider the structural analysis Phrase-structure rules are re-write NP – Aux – V – NP. In a formal rules which formalise immediate con- approach these four elements are stituent structures of sentences on an numbered linearly from left to right as abstract level. All the elements to the X1 – X2 – X3 – X4. The structural right of the arrow are the proper con- change now gives the precise instruc- stituents of the single element to the tion X1 – X2 – X3 – X4 ⇒ X4 – X2 + left. In other words, rule (1) formalises be + en – X3 – by + X1 , that is, in a the fact that in the sentence “The man less formal way, to re-write the input hit the ball” ‘the man’ and ‘hit the string NP1 – Aux – V – NP2 as NP2 – ball’ are proper constituents of the Aux + be + en – V – by + NP1. It can sentence, whereas ‘the man hit’ is not. easily be seen that this transformation The output of the PS-rules are kernel turns the underlying structure of an strings, that is the underlying struc- active sentence such as ‘The man hit tures of kernel sentences. In order to the ball’ into the passive ‘The ball was arrive at the surface structure of the hit by the man’. kernel sentences we have to apply all It must be noted that in the theory the obligatory transformations which of Syntactic Structures, the PS-rules, regulate the morphological processes, together with the obligatory transfor- for example, concord. mations, only generate simple, active, Transformations were first intro- affirmative, declarative sentences, so- duced in linguistics by called kernel sentences, and that all (1952) as a means to account for other sentences are generated by the relationship between linguistic applying optional transformations. expressions at surface structure. In Though it would not be impossible to Chomsky’s terminology, transforma- generate all the interrogative, nega- tions are re-write rules applied on the tive, passive sentences, and so on by a output of the PS-rules, the underlying set of PS-rules alone, the obvious rela- structure, in order to arrive at the sur- tionship between, for example, pas-

234 TRANSFORMATIONAL-GENERATIVE GRAMMAR sive and active sentences would be were responsible for the recursiveness lost. All transformations are based on of language, another solution to han- deletion or insertion of elements dling this vital feature of a comprehen- resulting in substitutions or permuta- sive theory had to be found. So, instead tions and are either obligatory or of taking two sentence structures and optional. Whereas the obligatory conjoining them, it was proposed that transformations account for the by integrating a rule of the type NP + necessary morphological processes N + (S) the task could be shifted into needed to arrive at the actual sen- the PS-rules and the generalised trans- tences, optional transformations such formations could be discarded. Simi- as negation, passivisation, question larly, instead of having a negative or formation and so on change the mean- interrogative transformation, inserting ing by introducing new semantic the necessary element into the respec- information. Note that sentences such tive affirmative or declarative kernel as (a) ‘The man didn’t hit the ball’ (b) string, it was suggested that an abstract ‘The ball was hit by the man’ (c) ‘Did marker (neg) or (Qu), or both of them, the man hit the ball?’ and so on, are is inserted in the underlying structure not derived from (d) ‘The man hit the as an optional element, yielding the ball’ but that the kernel sentence (d) is rule S (Qu) (neg) + NP + VP. If these generated by means of obligatory markers are chosen in the generative transformations and sentences (a) to process, the respective obligatory (c) by optional transformations from transformations are triggered off by the same underlying kernel string. these markers to produce negative, Furthermore Chomsky distin- interrogative or negative-interrogative guishes between singular and gener- sentences. alised transformations. Singular J. J. Katz and Paul M. Postal sug- transformations operate on individual gested in their seminal book An strings of elements, whereas gener- Integrated Theory of Linguistic alised transformations combine two Descriptions (1964) that the underly- different strings into one, resulting in ing structure already contained all the embedding and conjunction and thus necessary elements for the semantic accounting for the possibility of recur- interpretation of sentences, and that sive constructions and the infinite transformations only account for the capacity of the grammar. necessary adjustments in order to The model proposed in Syntactic arrive at the surface structure and thus Structures had some major draw- are obligatory. Furthermore as the backs: the growing complexity of the meaning of the sentences is stated by transformational component, particu- means of PS-rules, these transforma- larly with respect to the generalised tions must also be meaning-preserving. transformations, the covert relation- Based on these suggestions, Chom- ship between, for instance, affirmative sky proposed in Aspects of the Theory and negative sentences, and above all of Syntax (1965) a much stronger the fact that it was a purely syntactic transformational-generative model theory. than his original one. The base com- As the generalised transformations ponent of the Aspects Theory still

235 TRANSFORMATIONAL-GENERATIVE GRAMMAR

consisted of a set of PS-rules and a lex- lar to Chomsky’s phrase markers as icon to create the deep structure. But in semantic representations from which contrast to the 1957 model, the base the surface structures are generated by not only contained syntactic informa- transformations. However, by trying tion but also all the semantic elements, to include more and more phenomena and could thus serve as the input for into their theory, they had to widen the semantic component, on the one the range of rules so drastically that hand, and the transformational and the model lost its explanatory power. phonological component, on the other, Furthermore, such a theory cannot to account for semantic interpretation, account for universal features of that is meaning and phonological form language and is a move away from respectively. The syntactic part still explanation back to descriptivism*. consisted of a generative and a trans- Apart from generative semantics formational part, but their functions there were a very large number of were different. It was now the base post-Chomskyan models, the most component that accounted for recur- important of which were Charles Fill- siveness and for all the semantically more’s (1968) , operat- relevant options, and therefore there ing with logical cases such as agent, was no need for optional transfor- instrument, experiencer, victim, place mations anymore. In the 1960s and and so on, as constituents in the base 1970s this theory was so important component; Grammars, that it became known as the Standard which considered the verb as the only Theory, and there was hardly any syn- governing element from which all tactic research done that was not based other expressions are dependent; and on this model by either accepting and Montague Grammar, as an attempt to developing it or by rejecting and formalise the semantic structure of replacing it by other theories. natural languages. Major criticism of the Standard Insights from these approaches led Theory came from within generative to a continuous development of the grammar itself. Some of Chomsky’s Standard Theory to what were to be students felt that the scope of gram- known as the Extended Standard mar was too narrow and should be Theory and the Revised Extended extended into other areas of language, Standard Theory. The major change in particularly into semantics. They the Extended Standard Theory was called their approach generative that semantic interpretation could not semantics*. Observing that the gram- be based on the deep structure alone, maticality of a sentence is not inde- but that it is determined by the deep pendent of the beliefs of the speaker or structure as well as by the surface struc- that it has a lot to do with the lexemes ture. However, the deep structure actually chosen, they suggested that keeps its important syntactic role. In syntactic and semantic processes the Revised Extended Standard Theory could not be separated and thus dis- we have a strict delimitation of the dif- carded the notion of deep structure. ferent grammatical components, that They claimed that all aspects of mean- is syntax, semantics, as well as phonol- ing could be captured in a form simi- ogy, and pragmatics, the

236 TRANSFORMATIONAL-GENERATIVE GRAMMAR introduction of marked elements and assignment of abstract Case and its mor- above all the reduction of the number phological realization. Control theory of transformations to a single, heavily determines the potential for reference of constrained move-␣ rule. This devel- the abstract pronominal element PRO. opment led to the Government and Binding Theory (GB-Theory) advo- It is the aim of GB-theory to find the cated in Chomsky (1981). principles and parameters common In GB-Theory, sentence descriptions to all languages so that the syntax of a are simultaneously created on the four particular language can be explained interdependent levels of syntax (or D- along these lines. Evidence and structure), S-structure, phonetic form counter-evidence from specific lan- and logical form, each of which is con- guages has led to continual refinement cerned with specific aspects of the of the theory, so that as much varia- description of the sentence under con- tion in human language as possible sideration. It is based on the principles can be accounted for. and parameters theory, which states In The Minimalist Program (1995), that there is a finite set of fundamental the latest step in the continuous principles common to all natural lan- development of transformational- guages and a finite set of binary generative grammar, Chomsky pro- parameters that determine the range vided a radically new approach to the of permissible variability in language, implementation of his underlying language acquisition and language ideas. The well-established concepts understanding. A major difference of D-structure and S-structure have from the Aspects model is the fact that been discarded as well as government, logical form is now derived from the the central element in GB-theory. Even S-structure and not from the D- the ubiquitous phrase-structure rules structure. have been eliminated from the theory The principles that constrain the to a large degree. The only conceptu- range of possible sentence structures ally necessary categories left are the are contained in closely related sub- lexicon and the two levels of phonetic systems as formulated in Chomsky form and logical form* and it is the (1981: 5–6): role of a grammar to map them onto each other. It is suggested that all the Binding theory poses conditions other categories be dealt with in exter- on certain processes and related items. nally specified systems outside the The central notion of government actual grammar. Using conceptions of theory is the relation between the head economy, assuming that humans use of a construction and categories as economic a system as possible, it is dependent on it. ␪-theory is concerned suggested that an optimal solution of with the assignment of thematic relating phonological form and logi- roles such as agent-of-action, etc. cal form can be found. (henceforth: ␪-roles). Binding theory is As suggested by the title of Chom- concerned with relations of anaphors, sky’s book, minimalism* has not yet pronouns, names and variables to possi- been fully developed into a compre- ble antecedents. Case theory deals with hensive theory, but is a program,

237 TRUTH THEORIES according to which a theory of the Fodor, Jerry A. (1964). The Structure of language faculty could be developed. Language. Readings in the Philosophy Although the minimalist program is of Language. Englewood Cliffs: Pren- well established in linguistics, it is tice Hall. severely criticised by developmental Jackendoff, Ray S. (2007). Language, psychologists, cognitive scientists and Consciousness, Culture: Essays on neuroscientists, particularly because Mental Structure (Jean Nicod Lectures). of its claims with respect to language Cambridge, MA: MIT Press. acquisition and innateness*. Lyons, John (1970). Noam Chomsky. Transformational-generative gram- Glasgow: Fontana/Collins. mar has influenced the study of syn- Otero, Carlos (1994). Noam Chomsky: tax more than any other theory, but it Critical Assessments. Four vols. also had an impact on other fields London: Routledge. within linguistics, such as phonetics Smith, Neil (1999). Chomsky Ideas and and phonology, sociolinguistics, prag- Ideals. Cambridge: Cambridge Univer- matics, discourse analysis and polite- sity Press. ness theory. Furthermore Chomsky’s innovative approach to the study of Further reading language and his methodology have Fillmore, Charles J. (1968). ‘The case for influenced anthropology, philosophy, case’. In E. Bach and R. T. Harms, Uni- psychology and sociology, as well as versals in Linguistic Theory. New York: computer science, neuroscience, liter- Holt, Rinehart and Winston. ary criticism, music theory, and other Harris, Zellig S. (1952). ‘Discourse analy- fields. They have provided new sis’. Language 28: 1–30. insights for second-language teaching Katz, Jerrold J. and Paul M. Postal (1964). and learning and speech pathology, An Integrated Theory of Linguistic and above all, they have changed our Descriptions. (Research Monographs, conception of the mind. 26). Cambridge, MA: MIT Press. Lakoff, George; and John R. Ross (1976). Primary sources ‘Is deep structure necessary?’. In James Bach, Emmon W. (1974). Syntactic D. McCawley (ed.), Syntax and Seman- Theory. New York: University Press of tics 7: 159–64. America. Salkie, Raphael (1990). The Chomsky Chomsky, Noam (1957). Syntactic Struc- Update. Linguistics and Politics. tures. The Hague: Mouton. London: Routledge. Chomsky, Noam (1965). Aspects of the Theory of Syntax. Cambridge, MA: Jürg Strässler MIT Press. Chomsky, Noam (1975). Reflections on Language. New York: Pantheon Books. Chomsky, Noam (1981). Lectures on TRUTH THEORIES Government and Binding. Dordrecht: Foris Publications. Truth theories are philosophical Chomsky, Noam (1995). The Minimalist attempts to pin down the concept of Program. Cambridge, MA: MIT Press. truth. Truth is arguably one of the

238 TRUTH THEORIES most central issues in philosophy the name should avoid. Consider the and truth theories vary a great deal sentence, ‘I am lying’. If one asks from one another in terms of what whether this sentence is true or false, they claim as well as what they take one is soon confronted with a para- for granted, that is, their metaphysi- dox. If it is true, then the speaker is cal baggage. Philosophers have admitting that he/she is lying or that always been at pains to define truth what he/she is saying is false. On the and suggest ways of ascertaining other hand, if it is false, then we have when something is true and when it is the speaker’s word that he/she is not not. lying or that what he/she is saying is true. So, either way, the question as to See also: Analytic/Synthetic; the truth or falsity of the sentence ‘I Correspondence Theory; Definite am lying’ leads one to an intellectual Descriptions; Logic; quagmire. There are many variants of Presupposition; Propositions; the liar paradox that produce similar Truth Value results. Among these are ‘Epimenides Key Thinkers: Aristotle; Ayer, A. the Cretan says all Cretans are liars’ J.; Davidson, Donald; Frege, (is Epimenides lying or telling the Gottlob; Kant, Immanuel; Peirce, truth?) or ‘The barber of Seville is one C. S.; Quine, W. V. O.; Ramsey, F. who, while continuing to live in P.; Russell, Bertrand; Strawson, P. Seville, decides to shave every man in F.; Tarski Alfred; Wittgenstein, Seville who does not shave himself’ Ludwig (does the barber shave himself?). It is clear that, irrespective of the defini- Ever since Pontius Pilate asked the tion of truth one works with, one has question ‘What is truth?’ and, as the to make sure that the paradox pro- saying goes, preferred not to wait for duced in each of these cases must an answer, the issue has been posed somehow be skirted. and intensely probed time and time The solutions proposed by phil - again by philosophers. But truth has osophers for the liar paradox and its proved to be evasive. Over the cen- variants differ considerably. Some rec- turies, philosophers have proposed ommend that we regard such sen- competing theories of truth, but this is tences as meaningless on the grounds an area of investigation where new that one part of a sentence cannot sen- problems keep cropping up faster than sibly refer to the sentence as a whole they are able to tackle them. In philos- and therefore predicate of it that it is ophy, the predicate ‘true’ is always in true or false. Others concede that such contrast with ‘false’, unlike popular sentences do make sense (how else do usage where it is often opposed to ‘fic- we know that they are odd?) but are titious’, ‘fake’ or ‘spurious’. neither true nor false. Others decree Philosophers have long contended that such sentences are ill-formed with a mind-boggler referred to gener- because they violate a ban on self- ically as the ‘liar paradox’. The liar reference. Still others decree that all paradox illustrates an early conun- paradoxical sentences, being outright drum which any theory of truth worth , are false and argue

239 TRUTH THEORIES further that any attempt to derive the cal term proposition*, which they dis- truth of part of a paradoxical sentence tinguish from a sentence (a syntactic by logical means is doomed to fail. entity) and a statement (a unit of prag- Bertrand Russell identified a version matics). Another property of truth is of the ‘liar paradox’ in by that a sentence, once true, will always asking us to imagine ‘a set of sets that be true no matter what the specific cir- are not members of themselves’ and cumstances of its utterance, making wondering, apropos of this larger set, due allowances for such elements as if it is a member of itself. What Russell ‘tomorrow’ that, as it were, date the successfully showed by his discussion sentence. In other words, truth is not of this was the importance only a universal but also an eternal of weeding out the threat of contra- property. Furthermore, it is an inalien- diction from a mathematically sound able property. With reference to the theory of truth, or for that matter, any specific example above, it must be theory whatsoever. pointed out that what is being consid- Anyone setting out to propose a ered truth-apt is neither the sentence theory of truth has to sort out a per se nor the proposition expressed number of initial questions. To begin by it, but a dated instantiation of it, with, it is important to know what it that is to say, a token of that sentence is of which we may sensibly predicate as produced on a certain date and at a truth or, to put it technically, what precise time. This caveat is absolutely sorts of entities can be said to be truth- necessary because, unless we are fur- apt. For instance, the sentence ‘It will ther told when/on which day the sen- rain tomorrow’ is bound to be either tence was uttered, it would be true or false – we will soon discover impossible to verify its truth, for the whether it is one or the other – simple reason that this can only be whereas ‘Will it rain tomorrow?’ or ‘If done retrospectively – a day after it is only it rained tomorrow/Would that it uttered. rained tomorrow’ is not considered to The philosophers who broadly sub- be a candidate for truth ascription. scribe to the view of truth adumbrated On the other hand, ‘Il pleuvra in the foregoing paragraphs are work- demain’, ‘Vai chover amanhã’, ‘kal ing within what is referred to as the pani barsega’, which are the French, correspondence theory* of truth. This Portuguese and translation- is a way of conceiving of truth that equivalents of ‘It will rain tomorrow’ dates back to at least as far back as respectively, are just as true or false Aristotle, who famously claimed in his under the same set of circumstances as Metaphysics (1011b25): ‘to say of the English sentence. This means that what is that it is not, or of what is not what is true or false is not the sentence that it is, is false, while to say of what ‘It will rain tomorrow’ (which, of is that it is, and of what is not that it course, belongs only to the English is not, is true’. Underlying the corre- language). In other words, truth has spondence theory is the assumption to do with what a sentence says or that truth is a matter of correspon- what we may call its meaning. dence between what one says on the Many philosophers use the techni- one hand and what there is out there

240 TRUTH THEORIES in the world (say, a fact or a state of Frege claimed in a paper published affairs) on the other. It is perhaps the toward the end of the nineteenth cen- most satisfactory theory from a purely tury that the truth of the sentence ‘the intuitive point of view and is certainly present king of France is bald’ was the oldest. It is implicit in such routine conditional upon there being a king of practices as a police investigation into France. In other words, the sentence a possible crime in which a witness (or proposition, if you will; note that account is checked and double- Frege had in mind an undated version) checked by comparing it to every would be neither true nor false if put single detail of the crime scene, or the forward at a time when France was no police coming up with an artist’s longer a monarchy. To control the sketch of the physiognomy of the sus- damage such a claim represented to pect on the basis of details furnished binary logic, Bertrand Russell pre- by persons who claim they witnessed sented his famous Theory of Descrip- the crime. tions, under which what appears to be The correspondence theory is a reference to the putative king of underwritten by metaphysical real- France turns out to be a claim to the ism, which is a claim for the existence effect there is a king of France at the of things independently of what any- time the sentence is produced and body might say or believe about them. which, thanks to its declarative mood, It also posits, as in Ludwig Wittgen- can be adjudged true or false, depend- stein’s celebrated analysis in Tractatus ing on the facts of the matter. Logico-Philosophicus, a world con- The Frege-Russell dispute over the sisting entirely of facts which can be need or otherwise to posit presupposi- truthfully and exhaustively mirrored tion* as a condition for a sentence to by language. Furthermore, every be either true or false impacted on proposition is either true or false and work done in linguistic semantics in none can be both simultaneously or the 1970s and the first half of the neither. Now, there can be no doubt 1980s. The initial enthusiasm in that correspondence theory and its favour of the Fregean approach soon underlying metaphysical realism, gave way to the Russellian alternative whether or not explicitly spelt out or of reinterpreting the putative presup- thought through, informs most of positions as straightforward entail- what laypersons believe about these ments, thanks to the work of scholars issues. Incidentally, when it comes to such as Deirdre Wilson and Ruth the layperson, philosophers are wont Kempson. to distinguish epistemic truth (what There are exceptions to the corre- one knows) and doxastic truth (what spondence theory’s claim that truth is one believes) from aletheutic or a matter of correspondence between a alethic truth (truth ipsis). linguistic object and a language exter- But the time-honoured claim of nal reality. These are so-called ‘ana- classical, two-valued logic* to the lytic’ sentences, first identified by effect that between truth and falsity Immanuel Kant, following in the foot- there could be no third value was steps of John Locke and Gottfried thrown into disarray when Gottlob Wilhelm Leibniz. For Kant, analytic

241 TRUTH THEORIES statements are those in the declarative herself, may be seen as putting to mood, where the predicate is con- practical use the central principle of tained in the concept of the subject, as coherence theory–the assumption in ‘All bachelors are unmarried’. By behind the practice being that, if the contrast, synthetic statements such as respondent is telling all the truth, only ‘The dog barked at the stranger’ can the truth, and nothing but the truth, be adjudged true or false, only on the he/she will not give an answer that basis of whether or not they faith- contradicts any other during the same fully report something that actually cross-examination. occurred in the real world. All syn- In the nineteenth century, Charles thetic statements are thus claimed to Sanders Peirce, William James and fall between the limiting cases of tau- defended a theory of tology and contradiction (true and truth that, despite differences among false, respectively, regardless of what these thinkers on details, claimed that happens to be the case in the world). truth is that which effectively works. There are many other theories of This is called the pragmatic (or prag- truth that, in one way or another, matist) theory of truth. Peirce added challenge the correspondence theory. the further requirement that truth is Coherence theory is one such. Coher- ‘the opinion which is fated to be ulti- ence theorists hold that truth is a mately agreed by all who investigate’, matter of internal consistency within a thus emphasising the public character whole system of propositions. In other of truth. William James replaced that words, truth is to be predicated pri- with an emphasis on the process of marily of the system as a whole and belief-formation by the individual only secondarily and derivatively of rather than the product and pointing individual propositions, based on to the utility of holding something as whether or not they cohere with one true as the criterion for judging it to be another. Under such a generous defi- true. John Dewey, agreeing with nition, a system comprised entirely of Peirce that truth is the conclusion to false propositions (that is, false under be arrived down the end of the road, the correspondence theory) may nev- went on to define it in terms of his ertheless receive the certificate of favourite concept of warranted assert- truth, thanks to the fact they cohere ibility. One might say that pragma- with (do not contradict) one another tism, with its steadfast opposition to and thanks also to the fact that the the Cartesian dogma, realism, Conti- coherence theory does not require val- nental rationalism and its emphasis on idation by any factor external to the instrumentalism, is opposed to corre- system. The popular saying ‘Truth spondence theory. But it has also been knows no contradiction’ may be seen argued that incorporates as drawing on the key insight of coher- – or at the very least is not averse to – ence theory. So too the practice of certain elements of both correspon- cross-examination of a suspect by the dence and coherence theories. police or of a defendant by a prosecu- There are other versions of coher- tor, where the attempt is to catch the ence theory that concede that, respondent contradicting himself/ although individual propositions have

242 TRUTH THEORIES nothing do with a language-external alinguistic observation concerning the reality or the way we experience it, the truth of that sentence in the object lan- system as a whole does, at least at its guage. In other words, the truth of ‘p’ fringes. For instance, W. V. O. Quine is claimed to be guaranteed by the truth held that, in talking about the world, of p. This means that one can predicate we typically move from talking in cer- truth only in a language other than the tain terms to talking about them, that one in which the sentence for truth is to say, from the material mode to ascription is presented or,alternatively, the formal mode, in what he famously ‘p’andp belong, as it were, to two dif- referred to as a ‘semantic ascent’. At ferent languages. Evidently, as Tarski the other extreme of the spectrum is himself was the first to admit, his defi- the American neopragmatist philoso- nition of truth does not work for natu- pher Richard Rorty who flatly denies ral languages, where self-referential that there is any relation whatsoever sentences of the form ‘This sentence is between language on the one hand not true’ are permitted, in spite of the and the external world on the other, so difficulty in assigning to it a meaning. that all talk of correspondence is In a formal language, such trouble- philosophical balderdash. some cases can be blocked by fiat. That truth itself might be definable Tarski’s theory of truth is a version – only by means of a sort of semantic albeit a very sophisticated one – of the ascent was what led the Polish logi- correspondence theory. cian Alfred Tarski to develop his In 1927, Frank Ramsey published a semantic theory of truth – a theory of paper in which he defended the idea, truth in formal or artificial (not natu- originally entertained by Frege, that to ral) languages. He captured the say of a sentence that it is true is not insight in the formula: ‘ “p” is true if to say anything else about it. That is to and only if p’ (Tarski 1933). The key say, someone who says ‘p’ is thereby to interpreting the formula lies in already saying, as it were, ‘p is true’. admitting a neat separation between In other words, the only difference object language (the language under between the statements ‘p’ and ‘p is investigation) and (the true’ is that the latter makes explicit language employed to say things what the former says implicitly. Thus about the object language). Tarski’s was born so-called ‘’ or central idea was that truth in a given ‘deflationary’ or ‘disquotational’ (object) language can only be theory of truth. A. J. Ayer took explicated by means of a sentence in Ramsey’s thesis to its obvious conclu- the corresponding metalanguage or, sion, namely that truth is an empty equivalently, truth is invariably a met- predicate. This has been called the alinguistic predicate. ‘disappearance theory’ or the ‘no Thus, in the formula above, ‘p’isa truth theory of truth’. P. F. Strawson sentence in the object language in the used Ramsey’s thesis to come to an sense that it is about this sentence that equally interesting conclusion, by one predicates truth or whatever. The arguing that to say of a sentence that occurrence of p (the one without the it is true is to endorse it or give it a inverted commas) is part of the met- stamp of approval. Strawson’s thesis

243 TRUTH VALUE has received the name of the ‘perfor- Russell, Bertrand (1905). ‘On denoting’. mative* theory of truth’ because it Mind 14: 479–44. draws on J. L. Austin’s speech act Strawson, Peter. F. (1949). ‘Truth’. Analy- theory* which initially posited an sis: 83–97. important distinction between saying Tarski, Alfred (1933). ‘The concept of truth and doing. Quine, to whom the dis- in formalized languages’. Reprinted in quotational theory of truth is gener- Alfred Tarski (1956), Logic, Semantics, ally attributed, interpreted Tarski’s Metamathematius. Trans. J. H. Woodger. semantic theory in the light of Oxfurd: Clarendon Press. 152–78. Ramsey’s insight and argued that Tarski’s was, at bottom, a deflationary Further reading theory to the extent that, in Tarski’s Alston, William P. (1996). A Realistic approach, the predicate ‘is true’ only Conception of Truth. Ithaca, NY: Cor- made sense in a metalanguage, not in nell University Press. the object language. Davidson, Donald (1984). Inquiries into Truth theories, dedicated to answer- Truth and Interpretation. Oxford: ing the deceptively simple-looking Oxford University Press. question of what it is to say that a Dummett, Michael (1978). Truth and sentence, statement or proposition is Other Enigmas. Cambridge, MA: Har- ‘true’, have dominated the work of vard University Press. some of the most eminent thinkers, and Horwich, Paul (1990). Truth. London: have touched on many of the major Basil Blackwell. issues in the philosophy of language. Kirkham, Richard (1992). Theories of Truth: A Critical Introduction. Cam- Primary sources bridge, MA: MIT Press. Aristotle (350 BCE). Metaphysics. Trans. Rescher, Nicholas (1973). The Coherence W. D. Ross. Available at http://clas- Theory of Truth. Oxford: Oxford Uni- sics.mit.edu/Aristotle/metaphysics.html versity Press. (accessed 4 October 2007). Kanavillil Rajagopalan Ayer, Alfred J. (1952). Language, Truth and Logic. New York: Dover Publications. Frege, Gottlob (1892). ‘On sense and mean- ing’. In Peter Geach and Max Black (eds) TRUTH VALUE (1980). Translations from the Philo soph- ical Writings of Gottlob Frege. Oxford: At their simplest, truth values are often Blackwell. First edition, 1952. 56–78. thought to be governed by the princi- Quine, Willard V. O. (1970). Philosophy ple of bivalence, in which any declara- of Logic. Englewood Cliffs, NJ: Prentice tive sentence’s truth value can be either Hall. true (when what the sentence states is Ramsey, Frank P. (1927). ‘Facts and the case) or false (when what the sen- propositions’. Aristotelian Society Sup- tence states is not the case). There is plementary Volume 7: 153–70. debate in logic and the philosophy of Rorty, Richard (1979). Philosophy and language about the appropriateness of the Mirror of Nature. Princeton: Prince- the principle of bivalence and hence, ton University Press. about the nature of truth value.

244 TRUTH VALUE

See also: Definite Descriptions; sition that a given statement is not Logic; Logical Form; Logical true, we can conclude merely that it is Positivism; Propositions either false or indeterminate. Using Key Thinkers: Aristotle; Dummett, bivalent semantics it could only be Michael; Frege, Gottlob; Strawson, false. P. F. Non-bivalent semantics has been motivated by ‘empty terms’, lacking Though problems about truth and fal- any truth value at all (Strawson sity are discussed in ancient logic, for 1950); ‘semantic presupposition’* in example by Aristotle, the phrase which a logically resolvable sentence ‘truth value’ as we use it in modern presupposes something impossible philosophy comes from Gottlob (Strawson 1952; see Allwood, Ander- Frege. He defined the truth value of a sson and Dahl 1977: 149–55); sentence as its being true or false, and ‘semantic paradoxes’, such as the sen- claimed that there are no other truth tence ‘This statement is untrue’ (see values in declarative sentences, which Read 1995: chapter 6; Haack 1978: normally refer to ‘the True’ or ‘the chapters 8, 11); ‘future contingents’, False’. Though Frege’s further view in which a statement refers to events that truth values are objects is dis- that have not yet occurred (see Haack putable, the main controversies about 1978: chapter 11); and mathematical truth value concern what it is that has statements, which in some approaches truth value (see Kneale and Kneale are always false because they refer 1962; Haack 1978: chapter 6) and to non-existent entities (see Key whether every declarative sentence is Thinkers: Dummett, Michael; Read either true or false. 1995: chapter 8). While truth value A sentence must be declarative, in bivalent semantics may appear rather than, for example, interroga- straightforward, wider investigation tive or imperative, if it is to be a can- by philosophers of language and by didate for having a truth value. Thus, logicians suggests that truth value is ‘London is in England’ and ‘Torture is both a complex and a contestable wrong’ are candidates; ‘Where is idea. London?’ and ‘Don’t go there’ are not. Declarative sentences are usually Primary sources called ‘statements’. By the principle of Frege, Gottlob (1892). ‘On sense and ref- bivalence, each statement is either true erence’. In Max Black and Peter Geach or false, but not both. Bivalent seman- (eds) (1960), Translations from the tics, then, recognises only two truth Philosophical Writings of Gottlob values; the semantics of classical Frege. Second edition. Oxford: Black- logic* is bivalent. well. 56–78. In contrast, in a non-bivalent approach to semantics a statement’s Further reading falsehood does not follow from its Allwood, Jens, Lars-Gunnar Andersson being not true. For example, suppose and Östen Dahl (1977). Logic in Lin- our truth values are: true, false, and guistics. Cambridge: Cambridge Uni- indeterminate. Then from the suppo- versity Press.

245 TYPE/TOKEN

Haack, Susan (1978). Philosophy of mental apparatus. The contrast be - Logics. Cambridge: Cambridge Univer- tween type and token maps quite sity Press. closely onto the contrast between uni- Kneale, William and Martha Kneale versals and particulars or onto that (1962). The Development of Logic. between properties and their instances. Oxford: Clarendon. Chapters 8 and 10, Most uses of the words ‘type’ or especially 576–93. ‘token’ by linguists or philosophers Read, Stephen (1995). Thinking about operate broadly along these lines. Logic. Oxford: Oxford University Most appeals to the type/token dis- Press. tinction in a linguistic context belong Strawson, P. F. (1950). ‘On referring’. to one or other of two kinds. For the Mind 59: 320–44. first, consider how many letters there Strawson, P. F. (1952). An Introduction to are in the word ‘London’. Counting Logical Theory. London: Methuen. by letter type, there are four: ‘L’, ‘O’, ‘N’, and ‘D’. Counting by letter token, Stephen McLeod there are six: one ‘L’, two ‘O’s, two ‘N’s, and a ‘D’. This contrast also applies at the level of whole expres- sions. There are, for example, nine TYPE/TOKEN word tokens in (1) but only seven word types: At its most intuitive, the distinction (1) ‘The blue cat ran circles around the between types and tokens is a distinc- green cat.’ tion between two different ways of counting linguistic entities. One This manifestation of the type-token reason for paying attention to the dis- distinction clearly generalises to many tinction is the ease with which failure other . Think, for to do so causes confusion in discus- example, of how many phonemes* sions of language. there are in ‘Mississippi’. See also: Language of Thought The second manifestation of the Key Thinkers: Peirce, C. S. distinction crops up when authors mark the contrast between sentences Suppose you are asked how many cars and utterances of sentences by there are in a car park. You could describing the latter as tokens of the count by type or by token. For exam- former. As with the first use, this also ple, if there are five Fords, one Fiat and helps to eliminate ambiguity*. Sup- three Toyotas, that gives three types of pose Jane and John both utter the sen- car but nine tokens of car. The termi- tence ‘London is in England’. Have nology of ‘type’ and ‘token’ was coined they performed the same utterance? by C. S. Peirce (1966: 4.537), but the The answer is affirmative if we mean distinction it marks has been known the same utterance type, where the under different labels for far longer. type is defined as the sentence pro- Indeed, the distinction is arguably a duced, but negative if we mean the primitive component of everyone’s same utterance token.

246 UNIVERSAL GRAMMAR

This second usage may seem at thought* as having been uttered, and odds with the first. To obtain two ‘tokened’ is a useful alternative. To tokens of the word ‘Alfred’ in the token a sentence of the language of second usage of the type/token dis- thought is for that sentence to occur tinction would require two utterances explicitly in one’s brain. Philosophers of the word. But in the first usage there of mind also make a related distinction are two tokens of this word in the sen- between types and tokens of actions tence ‘Alfred Brendel was named after (where only the latter are specific Alfred the Great’, even if that sentence events) and types and tokens of mental is not uttered even once. state (such as the belief that London is The relation between the first and in England, which anyone might have, second usages can be understood if we versus Jane’s belief that London is in return to our non-linguistic example. England). The vehicles in a car park can be typed The type/token distinction admits, according to manufacturer, to colour, then, of several different manifesta- to engine size, or to age, and so on. tions in linguistics. Using the label to Similarly, the utterances produced at a mark the contrast between a sentence particular party can be typed in differ- and a particular utterance of it may be ent ways: by the identity of the utterer, slightly out of kilter with other uses for example, or by loudness, pitch, but it has become dominant and is at time of occurrence,and so on. But the least comprehensible in context. most useful way of typing utterances in linguistics is usually by the expres- Primary sources sion produced. So dominant is this Peirce, Charles S. (1966). Collected Papers way of comparing and contrasting of Charles Sanders Peirce. Cambridge, utterances that it has become normal MA: Harvard University Press. to talk as though sentences themselves are the type, while utterances are Further reading tokens of them. Though this use Wetzel, Linda (2006). Types and Tokens: is, arguably, erroneous, it is simple An Essay on Universals. Cambridge, enough to understand what an author MA: MIT Press. has in mind in writing this way. The practice has even given rise to talk of a Alex Barber sentence’s being ‘tokened’, by which is meant its being uttered. A related practice is to describe the of expressions such as ‘I’, ‘now’, and UNIVERSAL so on as ‘token reflexivity’, since indexicals depend for their referential GRAMMAR properties on features of the context of their utterance. In Noam Chomsky’s theory of human A further factor driving this second natural language, the genetically use of ‘token’ derives from the philos- determined initial state of the lin - ophy of mind. It makes little sense to guistic competence in the individ- talk of sentences of the language of ual human mind/brain, which is the

247 UNIVERSAL GRAMMAR

starting point of the linguistic devel- the linguistic competence of a child, opment of the individual into a beyond the lexical dimension. mature user of a natural language. Amazement is often a plea for expla- nation. What enables the human mind/ See also: Acceptability/ brain to move from a seemingly poor Grammaticality; Continuity; initial state to an obviously complex Empiricism/Rationalism; mature state? This is the fundamental Generative Phonology; Generative question of language acquisition, Semantics; Innateness; Mentalism; dubbed by Chomsky ‘Plato’s problem’. Transformational-Generative Within the spectrum of answers given Grammar to that question, there are two extreme Key Thinkers: Chomsky, Noam; and commonly rejected positions and Kant, Immanuel. then two intermediate ones, both of which are related to rich philosophical The linguistic development of every traditions that have been in rivalry for human being is amazing. When a baby many years. One of these intermedi- is born, it does not speak, nor does it ate answers is the idea of universal understand the language of its par- grammar. ents. But some years later, every child All the different answers are posi- understands much of what is being tions with respect to the role played in said in the natural language used at language acquisition by the human home and uses parts of it in its spon- mind/brain and by human experience. taneous speech. After some additional One extreme position portrays the years, the boy or girl is a fluent user of whole process of language acquisition the same natural language and seems as resting solely on human experience; to share his or her linguistic compe- in the initial state of the process the tence with adult, mature users of it. baby’s mind/brain is a blank tablet. Such a development is amazing on The opposite extreme position por- several accounts. First, it apparently trays the whole process as resting requires no learning, in the sense used solely on innate human knowledge. when we describe the child as having Both of these positions are obviously learnt to use a spoon or a pencil prop- untenable. The former cannot explain erly and regularly. Second, it appar- the fact that different children, whose ently requires no effort–neither on the experiences, in terms of the utterances part of the baby or the child, nor on made in their vicinity, are quite differ- the part of the parents. What is ent from each other, turn out to speak required is that the latter use the lan- the same natural language. The later guage in the presence of the former. cannot explain the fact that there are Third, it takes place independently of many different natural languages. other facets of human development, Thus, conceptions of language acqui- such as spatial orientation or face sition that have the form ‘either nature recognition. Fourth, and most impor- or nurture’ are wrong. Accordingly, tantly, the product of development is the intermediate positions see both very complex, as is easily realised by nature and nurture as offering essen- anyone who tries to describe in full tial contributions to the language

248 UNIVERSAL GRAMMAR acquisition process. However, they inductive learning functions from the involve different views of the contri- utterances babies and children hear. bution of the human mind/brain to the An example of such an abstract process, and consequently, also of the property would be the very idea of contribution made by human experi- structure. A mature speaker’s linguis- ence to the process. tic competence induces a distinction A position that reflects the philo- between what counts as a proper sen- sophical tradition of empiricism takes tence in one’s natural language and the mind/brain of the baby when born what does not count as such. A gen- to include a general learning device, eral learning device has to capture this one that processes what is observed distinction, simply by processing during experience and gradually devel- utterances of sentences or other lin- ops all human cognitive competencies, guistic expressions. Such a general including language, but also face device would have to reach the con- recognition and calculation, for exam- clusion that the distinction does not ple. The mind/brain of the baby knows involve numerical calculation but nothing about language but acquires it rather structural processing. There is by applying the general learning device no reason to assume that general to the linguistic experience. learning devices will have the ability It is important to understand that to reach such a conclusion and also do such a position is actually a research it quickly. This is especially so in the programme. To hold such a position context of the limited linguistic data- within the framework of scientific base of a small child. research is to be committed to an The major alternative to the idea of investigative effort to propose a a ‘general learning device’ in lan- theory that specifies the structure and guage acquisition is that of universal operation of the general learning grammar. Here the position is that device and to put it to empirical test in the mind/brain of a new-born baby a variety of areas of human cognitive includes an innate concept of natu- competency, including language. ral language. That is a system of The major argument that has been properties shared by all human lan- raised against that position is called guages, historically actual, or psycho- ‘the poverty-of-the-stimulus argu- logically possible, which constitutes ment’. It has been argued by Chomsky the essence of human language. Thus, and many of his followers that the for example, the fact that the syntax natural language acquired by the child of a human natural language is struc- has highly abstract properties that tural rather than numerical is a fact cannot be the product of any general about language that is included in learning device, especially since the the universal grammar. It is not input received consists only of utter- learnt by a general learning device ances made in its presence. For exam- from experience, but innately given ple, each sentence of a natural to human beings as a feature of the language has a syntactic structure. mind/brain. Those structures have abstract prop- It should be noted that universal erties that cannot be formulated by grammar is not a grammar in the

249 UNIVERSAL GRAMMAR

ordinary sense. It is a system of con- though ancient, is particularly related ditions that every natural language to seventeenth-century philosophy of fulfills precisely because every natural language and cognition. Its generic language is an end-product of an nature is related to Kant’s conception acquisition process that begins with of the mind. Much of generative the universal grammar. Universal linguistics should be viewed as taking grammar is uniquely linguistic. It is place within the research programme what the human mind/brain knows, of universal grammar. by genetic , about natural language. This is one facet of the Primary sources modularity of language: its cognitive Chomsky, Noam (1966). Cartesian Lin- independence of the rest of the guistics: A Chapter in the History of mind/brain. Universal grammar is Rationalist Thought. New York and generic, in the sense of serving as the London: Harper and Row. initial state of different acquisition Piattelli-Palmarini, Massimo (ed.) (1980). processes that lead to different natu- Language and Learning: The Debate ral languages. between Jean Piaget and Noam Chom- The unique linguistic nature of the sky. Cambridge, MA: Harvard Univer- universal grammar and its generic sity Press. property are manifest in a theory of Chomsky, Noam (1986). Knowledge of principles and parameters that has Language: Its Nature, Origin, and Use. been developed within the research New York, Westport, CT and London: programme of universal grammar. Praeger. Universal grammar includes a class of Chomsky, Noam (1982). The Generative parameters that have to be set, on Enterprise. A discussion with Riny grounds of linguistic experience. Huybregts and Henk van Riemsdijk. When the parameters are set, much of Dordrecht: Foris; The Generative Enter- the nature of the natural language to prise Revisited (2004). Discussions with be acquired has been determined. An Riny Huybregts, Henk van Riemsdijk, example is the head parameter: the Naoki Fukui and Mihoko Zushi, with a head of a noun phrase is a noun, the new Foreword by Noam Chomsky. head of a verb phrase is a verb, and so Berlin and New York: Mouton de on. In a language, all heads appear on Gruyter. the same side of the phrase. Some nat- Chomsky, Noam (1988). Language and ural languages are head-first ones: Problems of Knowledge: The Managua English is an example. Some are head- Lectures. Cambridge, MA and London: last ones: Japanese is an example. In MIT Press. order to set the head parameter, the Chomsky, Noam (2002). On Nature and child does not have to hear more than Language. Ed. Adriana Belletti and a few examples of appropriate expres- Luigi Rizzi. Cambridge: Cambridge sions or sentences. University Press. The universal grammar research programme emerged from the philo- Further reading sophical tradition of rationalism. The Cook, Vivian and Mark Newson (1996). related conception of innateness, Chomsky’s Universal Grammar: An

250 USE/MENTION

Introduction. Second edition. Oxford: (1) Paris is a bustling metropolis. Blackwell. (2) ‘Paris’ begins with the sixteenth McGilvray, James (ed.) (2005). The Cam- letter of the English alphabet. bridge Companion to Chomsky. Cam- bridge: Cambridge University Press. Here are two implausible claims:

Asa Kasher (3) ‘Paris’ is a bustling metropolis. (4) Paris begins with the sixteenth letter of the English alphabet. USE/MENTION (1) and (3), while differing in accu- racy, sound the same when spoken aloud. The same goes for (2) and (4). To utter the word ‘Paris’, for example, This is why failure to be clear on while referring to the French city, is to whether one is talking about what a use that word. To utter the word while word refers to or about the word itself referring to the word itself is to men- can cause confusion. tion the word. The distinction matters In writing various conventions allow to those investigating language for easy disambiguation between use because they are required to talk and mention. Mentionings are often sometimes about a word and some- enclosed within single inverted times about what that word refers to, commas; under another convention, giving rise to potential confusion over they are italicised. In speech, context which they are doing. usually suffices to make clear which interpretation is intended. If that is not See also: Propositional Attitudes; possible, speakers sometimes make a Sense/Reference deliberate effort to highlight the dis- Key Thinkers: Frege, Gottlob; tinction by explaining that they mean Leibniz, Gottfried Wilhelm; Quine, ‘the word “Paris”’ or ‘Paris the city’. W. V. O. For an example of where confusion In spoken language the distinction might occur, consider the contrast between the use of a word and the between (5a) and (5b): mention of a word is often difficult to work out. In most languages there is (5) (a) What does ‘citizenship’ mean? no convention for indicating through (b) What does citizenship mean? pronunciation which is being employed. Speakers and hearers rely An utterance of (5a) would typically on the context in which a word is used be inviting a simple definition of the in order to convey and understand word, such as might be given in a dic- what is meant. Yet the distinction has tionary. (5b), though it sounds the a significant effect on meaning to the same, is more plausibly asking for extent that it can make the difference someone’s view on the broader impli- between sense and nonsense. For cations of citizenship as a status. example, here are two plausible Several factors make the use/mention claims: contrast difficult to define precisely.

251 USE/MENTION

One is that the very label for the con- (7) The final letter in ‘Paris’ is left trast is misleading: to mention a word unvoiced by the people who live often is to use it, albeit to refer to itself. there. Another is that one can mention a word (8) The people of Paris do not pro- without actually using it, as with (6): nounce it that way. (9) ‘Giorgione’ was so-called because (6) The five letter word that refers to of his size. the capital of the largest French- speaking European country has two The use/mention distinction can be syllables. difficult for language users to negoti- ate, yet it is important in discussions Since such cases do not give rise to of language precisely because it exists confusion, they are not what people at the interface between expression would have in mind as typical men- and meaning. The fact that for speak- tionings, but mentionings they never- ers and hearers the distinction is theless are. Yet another complication often imperceptible without addi- is that non-mentioning uses of a word tional communicative actions, includ- have little in common beyond being ing gestures and intonation, has made non-mentionings. They are certainly use/mention a central problem in not all cases of using the word to refer pragmatics as well as in the philoso- to a standard referent. Think of ‘the’ phy of language more generally. or ‘Santa Claus’, for example, or the use of words to refer to a non- Primary sources standard referent in malapropisms. Frege, Gottlob (1892). ‘On sense and ref- Anaphora – in which one word or erence’. In P. Geach and M. Black (eds) expression is referentially dependent (1980), Translations for the Philosoph- on another – introduces yet more ical Writings of Gottlob Frege. Third complexity. In (7), the final word is edition. Oxford: Blackwell. anaphoric to the fifth word, even Quine, W. V. (1956). ‘Quantifiers and though the fifth word refers to itself propositional attitudes’. Journal of Phi- while the final word refers to a city. losophy 53, 177–87. With the ‘it’ in (8) the situation is reversed. Following W.V.O. Quine’s Further reading work on propositional attitudes*, Davidson, Donald (1979). ‘Quotation’. In example (9) is still more complex. Donald Davidson (2001), Inquiries into The ‘so’ in ‘so-called’ refers to the Truth and Interpretation. Oxford: name ‘Giorgione’ (which means ‘tall Oxford University Press. George’) while ‘his’ refers to the Ren- Devitt, Michael and Kim Sterelny (1987). aissance artist himself, making it Language and Reality. Oxford: Black- unclear whether the name to which well. 34–5. both are tied anaphorically should be in quotation marks or not. Alex Barber

252 Index

Entries are indicated in bold. acceptability/grammaticality, 1–2, Birdsong, D., 105 105 Blakemore, Diane, 197 acquisition (of language), 20, 49, 50, blending, 23 51–3, 71–3, 96, 111, 130, 157, Bloomfield, Leonard, 18, 58, 60, 231, 232–3, 237, 238, 249–50 115, 130, 219, 223 adequacy, 2–3, 137 Boas, Franz, 58, 115, 189, 223 adjacency pairs, 31, 34 Bourdieu, Pierre, 160, 172 alethic, 165, 167, 241 Bréal, Michel Jules Alfred, 206 Althusser, Louis, 171, 222 Brentano, Franz, 5, 100–1, 213–14 ambiguity/vagueness, 3–4, 98, 150 Broca, Paul Pierre, 207 analytic philosophy, 4–11, 126, 152, Brown, Gillian, 32, 60–1 153 Brown, Penelope, 158–60 analytic/synthetic, 9, 11–14, 129, Brugmann, Karl, 57 163, 168, 241–2 Bühler, Karl, 214 animals (communication of), 19, Butler, Judith, 154 27–8, 96 Bybee, Joan, 50–1 anthropology, 35, 59, 66, 97, 152, 170, 189, 219, 221, 223, 238 Cameron, Deborah, 75–6 Aquinas, Thomas, 42 Carnap, Rudolf, 8, 10, 12, 19, arbitrariness, 27, 28, 114, 143, 144, 128–9, 164, 166, 180 191, 207, 208, 220 Carston, Robyn, 195, 198 Aristotle, 25, 28, 42, 47, 101, 107, Chase, Stuart, 118 122, 163, 213, 240, 245 Chomsky, Noam, 1, 2–3, 19–20, 27, Arnauld, Antoine, 163 30, 43–4, 49–50, 52, 58, 60, 65, Artificial Intelligence (AI), 14–17, 97, 71–3, 77, 78, 96–7, 103–6, 107, 131, 142, 186 110–11, 127, 130–1, 134–8, 163, Austin, J. L., 8, 79, 111, 129, 149, 180, 221, 223–4, 232–8, 247–51 150–6, 185, 199, 209, 213–16, cognitivism, 20, 21–4, 96 244 cohesion, 60, 61 Ayer, A. J., 68, 129, 180, 215, 243 compositionality, 24–5, 83, 142, 184, 203 Babbage, Charles, 15 computational linguistics, 15 Bach, K., 128, 141, 156, 195 concordance, 36, 41 Bakhtin, Mikhail, 115 Condillac, Etienne Bonnot de, 206 Bally, Charles, 113, 158 conditioning, 18–19, 130 Barthes, Roland, 115, 222 connotation/denotation, 25–6, 202 Barwise, Jon, 209, 210, 211 continuity, 26–8 Beaugrande, Robert de, 60–1 conventional meaning, 28–9, 145, behaviourism, 17–20, 21, 93, 130 196, 198 Berkeley, George, 70, 199 conversation analysis, 29–36, 59

253 INDEX

Cook, Guy, 31 existentialism, 171 Copenhagen Linguistic Circle, 79 extension, 26, 167, 202 corpora, 23, 36–42, 106, 160 corpus linguistics, 39, 231 Fairclough, Norman, 63 correspondence theory, 42–3, 240–3 feminism, 74–7, 161 Coulthard, Malcolm, 60, 61 Firth, J. R., 36, 157, 225 creativity, 16, 27, 43–4, 130, 192 Fishman, Pamela, 74 (critical) discourse analysis, 58–65, Fodor, Jerry, 90, 110–12, 180, 233 97, 99, 238 Foucault, Michel, 172 Francis, Nelson, 37 Davidson, Donald, 24, 209 Frege, Gottlob, 6, 9, 24, 54, 68, deconstruction, 44–7 123–6, 141, 143, 166, 173–4, 176, deduction/induction, 47–53 180, 184, 201–5, 211, 241, 243, definite descriptions, 6, 54–7, 127, 245 176, 184 Full Interpretation (principle of), deixis, 205, 226, 227 136 Dekker, P., 109 Dennett, Daniel, 103, 112 Gadamer, Hans-Georg, 210 denotation see game theory, 107–8 connotation/denotation Garnfikel, Harold, 29 Derrida, Jacques, 44–7, 154, 171–2, generative grammar see 222 transformational generative Descartes, René, 5, 15, 44, 49, 70, grammar 101, 130, 149 generative phonology, 65, 77–8 descriptivism, 57–8, 83, 184, 236 generative semantics, 14, 21, 78–9, Devlin, Keith, 210 156, 236 Dewey, John, 242 genetics, 28, 96–7 discourse completion tests, 159 genre, 33, 34, 36, 59, 62 discourse markers, 34 glossematics, 79–81 dispreferred response see Goffman, Ervine, 158 preferred/dispreferred responses Gopnik, Alison, 52 distinctive features, 25, 65–6, 77 Gordon, Dan, 217 Donnellan, Keith, 56 government and binding theory, 135, Dressler, Wolfgang, 60–1 237–8 Dreyfus, Hubert, 16 grammaticality see Dummett, Michael, 9–10 acceptability/grammaticality Durkheim, Émile, 114, 189 Green, Georgia, 79 Grice, H. P., 28–9, 79, 84, 85, Eco, Umberto, 208 86–92, 108–9, 111, 144–5, 158, edge feature, 137 159, 173, 193, 197–8, 209, 217 emic/etic, 66–7 empiricism/rationalism, 9, 18, 43, Halle, Morris, 65, 77 49–50, 52, 67–74, 130, 151, 242, Halliday, M. A. K., 58, 60–1, 249, 250 225–31 English for specific purposes (ESP), Harnish, R. M., 156 61–2 Harris, Roy, 97–8 epistemology, 43, 67, 152, 163 Harris, Zellig, 60, 223, 234 ethnography, 59, 60, 99 Hasan, Ruquaiya, 60–1, 225 etic see emic/etic Hegel, G. W. F., 192

254 INDEX

Heidegger, Martin, 44 Kripke, Saul, 92–5, 143, 164, 166, Herder, Johann Gottfried, 117 168–9, 180, 184, 202 Hintikka, Jakko, 107 Kroeber, Alfred, 117 Hjelmslev, Louis, 79–80, 207, 220 Kruszewski, Nikolaj, 157 Hockett, Charles, 18, 27, 58 Kucera, Henry, 37 holism, 13, 81–4, 93, 129 Kuhn, Thomas, 171 Horn, Laurence, 90–1 Hornstein, Norbert, 209 Labov, William, 58, 60 humanism, 171 Lacan, Jacques, 171, 189–92, 222 Humbolt, Wilhelm von, 44, 116–17 Lakoff, George B., 21, 49, 79, Hume, David, 48–9, 68, 70, 132–3, 154, 187, 217 199–200 Lakoff, Robin, 74, 79, 158 Husserl, Edmund, 100–1, 214, 215 Langacker, Ronald, 21, 79, 187 language games, 106–9 iconicity, 21, 23, 208 language of thought (LOT), 81, idealism, 5–6, 8 110–12 ideational theories, 84–5, 98 langue/parole, 113–16, 219 implicature, 86–92, 145, 158–9, 173, Leibniz, Gottfried Wilhelm, 70, 151, 195–6, 198, 216 164, 203, 241 indeterminacy, 9, 92–5, 98–9, 100 Lévi-Strauss, Claude, 219, 222 induction see deduction/induction Levinson, Stephen, 91, 158–60, 173, innateness, 71, 96–7, 111, 238, 250 177–8, 179 insertion sequence, 31 Lewis, David, 29, 108, 164, 167–8, integrationism, 97–100 211 intentionality, 100–3, 110, 112, 171, lexicon, 21, 28, 33, 123, 124, 237 214, 216 linguistic relativity, 116–19, 222 international phonetic alphabet linguistic variable, 24, 119–20 (IPA), 67 literary criticism, 45, 170 intuition, 1, 61, 68, 85, 103–6, 157, literature, 47, 189, 221, 222 232 Locke, John, 70–1, 84–5, 199, 201, 206, 241 Jackson, Frank, 199 logic, 6–10, 55, 107, 120–7, 139, Jakobson, Roman, 65, 77, 115, 163, 167–9, 175–7, 203, 244–5 190–2, 221–2 logical atomism, 7 James, William, 242 logical form, 7, 55, 122, 127–8, 142, Jefferson, Gail, 30 195, 237 Jesperson, Otto, 158 logical positivism, 8, 12, 68, 73, Johnson, Mark, 131–4 81–2, 128–30, 215 Jones, Daniel, 156–7 Lorenzen, Paul, 107 Lovelace, Ada, 15 Kamp, Hans, 15 Kant, Immanuel, 5, 11–12, 42–3, 87, McCarthy, John, 146 180, 213, 241, 250 McCawley, James, 78, 79, 167–8 Karmiloff-Smith, Annette, 53 MacCorquodale, Kenneth, 20 Kasher, Asa, 90 Martin, James, 62, 225–6 Katz, J. J., 233, 235 Marx, Karl, 171 Kempson, Ruth, 241 Marxism, 115, 161, 171, 222 Knowles, Murray, 131, 134 mathematics, 2, 9, 55, 68, 70, 129, Kövecses, Zoltán, 131, 133–4 151, 174

255 INDEX

Matthiessen, Christian, 226, 228, 229 Peirce, Charles Sanders, 5–6, 49, Mead, Margaret, 208 116, 144, 207–8, 242, 246 Mendoza, San Francisco Ruiz de, 217 performatives, 152, 154, 155–6, mentalese, 110–11 214–16 mentalism, 21, 130–1 Perry, John, 209–10 mention see use/mention phase, 137 merge operation, 135 phenomenology, 101 Merin, Arthur, 108 phoneme, 65, 156–7, 220, 222, 224, metaphor, 16, 21–3, 43, 49, 109, 233, 246 131–4, 188, 190, 191–2, 198, 231 phonology, 23, 38, 58, 65–6, 77–8, metonymy, 132, 190–2, 217 113, 115, 147, 148, 220, 222, Mill, J. S., 26, 68, 143, 202 238 minimalism, 134–8, 237–8 Piaget, Jean, 53 Mitchell, T. F., 60 Pike, Kenneth, 58, 66–7, 115 modality, 62, 138–9, 165, 228 Plato, 5, 7, 28, 46, 96, 143, 151, 192 model theoretic semantics, 139–42 politeness, 90, 157–61 Montague, Richard, 24, 139–40, political correctness, 161–2 164, 185–6, 211, 233, 236 Popper, Karl, 43, 49, 52 Moon, Rosamund, 131, 134 Port Royal Logic, 110, 163–4 Moore, G. E., 6, 7, 10, 42, 88, 199 possible world semantics, 23, 164–9, morphology, 13, 38, 58, 139 176, 202, 211 Morris, Charles, 208 Postal, Paul, 78–9, 235 postmodernism, 47 names, 25–6, 54, 142–4, 168, 176, poststructuralism, 47, 75, 170–3 183–4, 202–3, 214 pragmatics, 3, 22, 28–9, 59–60, 78, narrative, 33, 63 79, 91, 109, 113, 157, 177, 181, natural language processing (NLP), 193, 195, 197–9, 208, 217, 228, 15, 39–41 236, 252 negative evidence, 96, 104 Prague Linguistics Circle (Prague Neurath, Otto, 128 School), 65, 115, 221 Nicole, Pierre, 163 preferred/dispreferred responses, 31 Nietzsche, Friedrich, 44 prescriptivism, 1, 57, 154 nonnatural meaning, 29, 144–5 presupposition, 55, 89, 173–9, 241, noun phrase, 25, 136–7, 176, 233–5, 245 250 Prince, Alan, 146 private language, 85, 111, 179–80, Ockham, William of, 26, 42 222 openings and closings, 30 propositional attitudes, 82–3, 110, optimality theory, 65, 77–8, 109, 112, 177, 181–5, 204, 252 145–9 propositions, 4, 48, 55, 81, 95, 108, ordinary language philosophy (OLP), 122, 155, 163, 165, 167–8, 174, 8, 55, 149–55, 215 176, 182–3, 184, 185–6, 215, 242 orthography, 35, 38 prototype, 21, 79, 186–9 psychoanalysis, 171, 189–93, 222 Pânini, 120 psychology, 13, 18, 19, 20, 21, 30, Panther, Klause, 217 48–9, 66, 71, 72, 81, 85, 96, 104, parameters, 50, 108, 211, 237, 250 112, 130–1, 145, 152, 157, 170, parole see langue/parole 187, 206, 214, 238 Pascal, Blaise, 15 Putnam, Hilary, 85

256 INDEX

Quine, W. V. O., 8–9, 11–12, 13–14, science, 9, 12–14, 18, 20, 43, 47, 52, 19, 25, 92–3, 95, 129, 243–4, 73, 79, 81, 97, 133, 171, 174, 252 206, 221, 233 Searle, John, 15–16, 79, 101–3, Ramsey, Frank, 243–4 111–12, 143–5, 150, 153–4, Rask, Rasmus, 57 155–6, 158, 185, 202, 216–17 rationalism see Sechehaye, Albert, 113 empiricism/rationalism Sellars, Wilfred, 199–200 realism, 6, 152, 168, 241, 242 semantics, 13–14, 21, 23, 26, 58, 78, recursion, 2, 96 95, 97, 107, 109, 113, 126, 131, reference see sense/reference 138, 139–42, 164–9, 176, 181, register, 1, 59, 62 182, 185, 197–8, 201, 203, 206–8, Reid, Thomas, 213, 214 209–12, 220, 222, 224, 227, 231, Reinach, Adolf, 214–15 233, 236, 241, 245 reinforcement, 19, 20, 63 sense data, 199–200, 152 relative clause, 2, 163 sense/reference, 200–5 relevance theory (RT), 91, 193–9 sequence pairs, 31 repair, 30, 32, 33, 34 signs and semiotics, 28, 98–9, response see stimulus/response 114–15, 116, 142–4, 201, 205–9, revised extended standard theory, 215 135, 236 Sinclair, John, 60 Ribot, Théodule-Armand, 206–7 situational semantics, 209–12 Rorty, Richard, 243 Skinner, B. F., 19–20, 130 Rosch, Eleanor, 22, 79, 154, 186–7 Smolensky, Paul, 146 Ross, John (Haj), 22, 78, 79 sociolinguistics, 24, 58, 60–1, 76, 78, Russell, Bertrand, 6–7, 9, 10, 15, 42, 97, 99, 120, 157, 231, 238 49, 54–6, 123, 126, 127, 143, sociology, 35, 45, 59, 114, 152, 158, 150, 152–3, 163, 174, 175–6, 189, 238 180, 183–4, 199–200, 215, Socrates, 13, 28, 46, 151 240–1 speech act theory, 59, 60, 86–9, 91, Ryle, Gilbert, 8, 19, 149, 150, 215 97, 101–2, 109, 112, 129, 152–3, 155, 158–9, 177–8, 185, 196, Sacks, Harvey, 29–32 212–19, 244 Sadock, Jerrold, 217 Sperber, Dan, 90, 91, 193–9 St Anselm, 26 Spinoza, Benedictus de, 70 Sampson, Geoffrey, 52 stimulus/response, 18–20, 43–4, Sapir-Whorf hypothesis, 52, 116, 49–50, 52, 70–1, 96, 195, 249 117, 223 Strawson, P. F., 8, 55–6, 145, 149, Sapir, Edward, 58, 66, 115, 116, 152, 174–6, 202, 209, 243, 245 117, 118, 130 structuralism, 113, 170–1, 219–25 Saussure, Ferdinand de, 26, 28, 46, Stubbs, Michael, 60–1 58, 59, 97, 98, 113–16, 144, 170, Swales, John, 62 189–92, 205–7, 219, 221–2, 224, synecdoche, 132 226 syntax, 2, 13, 38, 58–9, 79, 97, 111, Schegloff, Emanuel, 30–1, 33 113, 123, 126, 127, 141–2, 146, Schiffer, Stephen, 56, 145 163, 208, 231, 233, 236–7, 249 Schiffrin, Deborah, 33–4, 63 synthetic see analytic/synthetic Schlick, Moritz, 128 systemic-functional grammar (SFG), Schütze, C. T., 2, 105–6 62, 98, 225–32

257 INDEX tagging, 37–9 Uldall, Hans Jørgen, 79 tagmemics, 66 universal grammar (UG), 3, 44, Tain, Hippolite, 206 49–50, 77, 96, 247–51 Talmy, Len, 21 use/mention, 251–2 Tannen, Deborah, 75 Tarski, Alfred, 42–3, 139–40, 185, vagueness see ambiguity/vagueness 211, 243–4 van Dijk, Teun, 63 tense, 50–1, 62, 119, 136–7, 142, Van Rooy, R., 109 166 verification, 8, 68, 73, 129, 215 Thornburg, Linda, 217 Vienna Circle, 8, 9, 128–9, 215 Thorndike, Edward, 130 Titchener, Edward Bradford, 130 Waismann, Friedrich, 128 token see type/token Watson, John B., 18, 130 Tomasello, Michael, 51–2 Weizenbaum, Joseph, 15 Toolan, Michael, 99–100 Whitney, William Dwight, 206–7 transcription, 34–5, 38, 61 Whorf, Benjamin Lee, 116–19, 130, transformational-generative 157, 161 grammar (TGG), 1, 2–3, 21, Wilson, Deirdre, 91, 193–8, 241 58, 77, 78–9, 111, 135, Wisdom, John, 149 232–8 Wittgenstein, Ludwig, 7–8, 10, 19, translation, 9, 15, 39–40, 81–2, 42, 85, 93, 107, 111, 149–50, 92–3, 118, 143 152–4, 179–80, 188, 209, 215, Trubetzkoy, N. S., 65, 77 241 truth theories, 43, 238–44 Wodak, Ruth, 63 truth values, 56, 120–1, 125–6, World War I, 149, 221 140–1, 164–8, 174–6, 183–4, World War II, 149, 170 203–4, 244–6 Turing, Alan, 15 Yule, George, 32, 60–1 turn taking, 31–3 type/token, 49, 51, 110, 240, Zadeh, Lotfi, 79 246–7 Žižek, Slavoj, 192

258

ERRATUM

The publisher and editors of Key Ideas in Linguistics and the Philosophy of Language wish to apologise for the of fact which they mistakenly introduced into the entry on Innateness in this volume, errors which were neither originated nor seen by the author. The corrected and author-approved entry is printed here.

Innateness The claim that some aspects of linguistic competence are genetically specified rather than learnt through experience. This claim has been driving research in generative linguistics and language acquisition since the late 1950s.

See also: Acceptability/Grammaticality; Continuity; Mentalism; Transformational- Generative Grammar; Universal Grammar Key Thinkers: Chomsky, Noam; Descartes, René; Plato

Noam Chomsky proposed that humans possess domain- and species-specific knowledge of the structure of possible languages, which enables children to acquire language with the speed, efficiency and uniformity that they do. This view can be traced back to Platonic philosophy and Cartesian cognitivism. Opponents of this view claim that language acquisition is innately constrained but only by the same mechanisms that underlie general cognitive ability. In other words, although it is uncontroversial that linguistic development is innately constrained, exactly what is innate is still a matter of debate. The argument that has most forcefully been used in support of the position that we are born with innate knowledge of linguistic constraints and principles is the poverty of the stimulus argument (Chomsky 1980: 34): 1. Language is a complex system that could only be acquired through experience if negative evidence was available (that is, information about what sequences are grammatically illicit). 2. Children only ever have access to positive evidence (information about grammatical sentences). 3. Despite this, children successfully acquire language. If one accepts these premises, one has to conclude that humans are genetically hard-wired for language: children are born equipped with Universal Grammar (Chomsky 1981), containing information about universal linguistic principles which enable children to form specific hypotheses about the structure of the language they are learning. The proposed existence of a critical period for language (i.e. typical language acquisition is not possible after a certain age) and studies of deaf children who spontaneously develop sign language have been used as further evidence in support of linguistic nativism. Opponents of linguistic nativism claim that the richness of empirical data actually available to children is vastly underestimated; that general cognitive ability can explain how language is learnt from experience; and that the poverty of the stimulus argument holds only given the generative of what language is. The most articulated alternative to the nativist perspective has been provided within connectionist psychology (Elman et al. 1996), where artificial neural networks are trained to reproduce complex linguistic behaviour based solely on past experience with relevant data. Research on innateness has spurred comparative research with animals in an effort to identify exactly what aspect of linguistic knowledge could be species- and domain-specific. Parallel research in genetics also attempts to specify the genetic basis of language. In 2001, Lai and colleagues suggested that a mutation in the FOXP2 gene is causally involved in language disorders. Despite widespread enthusiasm in the popular press about the ‘language gene’, the exact role of this gene in relation to language development is far from clear. Although the innateness question is still unresolved, interdisciplinary research that straddles the gap between linguistics, anthropology, artificial intelligence, genetics and neuroscience has opened up novel, exciting ways in which the question can be empirically addressed.

Primary sources Chomsky, N. (1980). Rules and Representations. Oxford: Blackwell. Chomsky, N. (1981). Lectures on Government and Binding: The Pisa Lectures. Holland: Foris Publications.

Further Reading Elman, J. L., E. A. Bates, M. H. Johnson, A. Karmiloff-Smith, D. Parisi and K. Plunkett (1996). Rethinking Innateness: A connectionist perspective on development. 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. Lai, C. S. L., S. E. Fisher, J. A. Hurst, F. Vargha-Khadem and A. P. Monaco (2001). ‘A forkhead-domain gene is mutated in severe speech and language’. Nature 413: 519–23.

Stavroula-Thaleia Kousta