ChomskyIdeas and Ideals

Neil Smith

published by the press syndicate of the university of cambridge

The Pitt Building, Trumpington Street, Cambridge, United Kingdom

cambridge university press

The Edinburgh Building, Cambridge, cb2 2ru, UKhttp://www.cup.cam.ac.uk

40 West 20th Street, New York, ny 10011-4211,usa http://www.cup.org

10 Stamford Road, Oakleigh, Melbourne 3166, Australia

© Cambridge University Press 1999

This book is in copyright. Subject to statutory exception and to the provisionsof relevant collective licensing agreements, no reproduction of any part maytake place without the written permission of Cambridge University Press.

First published 1999

Printed in the United Kingdom at the University Press, Cambridge

Typeset in Plantin 10/12pt [vn]

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

Library of Congress Cataloguing in Publication data

Smith, N. V. (Neilson Voyne)Chomsky: ideas and ideals/Neil Smith

p. cm.Includes bibliographical references and index.isbn 0 521 47517 1. – isbn 0 521 47570 8 (pbk.)1. Chomsky, Noam. 2. Linguistics. I. Title

P85.C47S64 1999

410'.92–dc21 98–48326 CIP

ISBN 0 521 47517 1 hardbackISBN 0 521 47570 8 paperback

Contents

Acknowledgments page ix

Introduction 1

Chapter 1 The mirror of the mind 7

Linguistics as a science 8

Modularity 17

Competence and performance 28

Performance, parsing, and pragmatics 35

Evolution and innateness 38

Natural language and the language of thought 45

Chapter 2 The linguistic foundation 49

Introduction 49

Knowledge of language 50

The lexicon 50

Knowledge of language: structure 53

Knowledge of language: structural relations 54

Levels of representation 56

Constituents and rules 57

Deep structure 59

Description versus explanation 62

Government and Binding theory 69

Empty categories 76

The status of transformations 80

Principles and parameters 81

Lexical and functional categories 83

Minimalism 86

A historical progression 90

Evolution 91

Chapter 3 Psychological reality 93

Causality and observability 95

Psychological reality and the nature of evidence 97

Language processing 106

Language acquisition (Plato’s problem) 116

Language pathology 126

Connectionism: the behaviorists strike back 131

vii

Chapter 4 Philosophical realism: commitments and controversies 136

Commitments 136

Controversies 145

Language and the world 146

Language and the community 149

Language and the individual 155

Problems of semantics 162

Innateness 167

UniWcation and reduction 174

Conclusions 176

Chapter 5 Language and freedom 177

Explanation and dissent: the common threads 177

Rationality, modularity, and creativity 180

The anarchist background 186

The critique of (American) foreign policy 190

The critique of domestic policy 195

The critique of media control 199

The technique of dissection 203

Moral absolutes and options for the future 208

The positive program 211

Conclusion 213

Envoi 214

Notes and references 215

Bibliography 241

Index 263

viii Contents

Introduction

A Child of the Enlightenment. Chomsky, 1992b: 158

Chomsky’s achievement

Why is Chomsky important? He has shown that there is really only onehuman language: that the immense complexity of the innumerable lan-guages we hear around us must be variations on a single theme. He hasrevolutionized linguistics, and in so doing has set a cat among the philo-sophical pigeons. He has resurrected the theory of innate ideas, demon-strating that a substantial part of our knowledge is genetically deter-mined; he has reinstated rationalist ideas that go back centuries, butwhich had fallen into disrepute; and he has provided evidence that‘‘unconscious knowledge’’ is what underlies our ability to speak andunderstand. He has overturned the dominant school of behaviorism inpsychology, and has returned the mind to its position of preeminence inthe study of humankind. In short, Chomsky has changed the way wethink of ourselves, gaining a position in the history of ideas on a par withthat of Darwin or Descartes. And he has done this while devoting themajority of his time to dissident politics and activism: documenting thelies of government, exposing the hidden inXuences of big business, devel-oping a model of the social order, and acting as the conscience of theWest.

In this century his peers in inXuence are such disparate Wgures asEinstein, Picasso, and Freud, with each of whom he has something incommon. Like Freud – but with added intellectual rigor – he has changedour conception of the mind; like Einstein, he blends intense scientiWccreativity with radical political activism; like Picasso, he has overturnedand replaced his own established systems with startling frequency. Per-haps his greatest similarity is to Bertrand Russell, whose early work,Principia Mathematica, redeWned the foundations of mathematics, andwho devoted much of his life to political writing and activism. But whileeveryone knows something about mathematics, that most people have

1

even heard of linguistics is largely due to Chomsky. His renown inlinguistics, philosophy, and psychology Wrst ensured that a few peoplewould listen to his political views; subsequently, his political fame, ornotoriety, has attracted attention to his academic work, which hasbrought the study of language into the mainstream of scientiWc research,and simultaneously made it relevant to the rest of the humanities and thenatural sciences.

This book is not a biography. I am concerned with Chomsky’s ideas,rather than the details of his private life. This is not through lack ofinterest. Fascinating snippets of information emerge from his interviews:endearing tales of childhood visits to a baseball match with his school-teacher or insights about his feelings when forced to take boxing at college.However, Chomsky is ‘‘really a hermit by nature’’ and has repeatedlyemphasized that his personal views are irrelevant to his scientiWc ideas;indeed, that ‘‘to the extent that a subject is signiWcant and worth pursuing,it is not personalized.’’ For those who want personal glimpses beyond thefollowing few notes, the book by Barsky and the interviews with Bar-samian and Peck are the best sources (see Bibliography).

Chomsky was born on 7 December 1928. From the age of two, he spentten years in a progressive Deweyite school in Philadelphia, where therewas a congenial emphasis on individual creativity. From there he movedon to a regimented and stiXing high school, about which he claims toremember ‘‘virtually nothing.’’ Thereafter he attended the University ofPennsylvania where he met Zellig Harris, a leading linguist and politicaltheorist, who had a profound inXuence on his life. He graduated in 1949,with an undergraduate thesis about Modern Hebrew, that was laterrevised and extended as his master’s thesis. That same year he marriedCarol Schatz, a fellow student who has made a signiWcant contribution tolanguage and linguistics in her own right. He entered graduate schoollater the same year and in 1951 became one of the Society of Fellows atHarvard, from where he moved to the Massachusetts Institute of Tech-nology (MIT) in 1955. He has been there ever since, although a large partof each year is devoted to traveling around the world giving countlesslectures and interviews.

Apart from his major inXuence on linguistics, philosophy, and psychol-ogy, Chomsky has had a minor but not insigniWcant eVect on a range ofdisciplines from anthropology to mathematics, from education to literarycriticism. To understand this pervasive inXuence requires a grasp of thedeWning characteristics of Chomsky’s scientiWc program of GenerativeGrammar, and some insight into the appeal of his social and politicalthought. What follows is an attempt to explain Chomsky’s work byanalyzing and putting into context the key contributions he has made to

2 Introduction

the study of language and the study of mind. This involves dealing withissues, some of them technical and profound, in linguistics, psychology,and philosophy. His work in all these areas has been systematicallyinnovative and systematically controversial. Misunderstanding of hisviews is widespread in all three communities of scholars, and part of myaim is to explain why it is that he has been both adulated and viliWed. Insome instances the task is straightforward: the preconceptions that causethe misapprehensions are reasonably superWcial and clear. In others it isharder to see why the hostility is so uncomprehending.

The book is intended to be accessible to everyone. Accordingly, I havechosen not to clutter the text with footnotes, but detailed references,sources, and suggestions for further reading are collected together at theend of the book. All quotations are identiWed there and it should bepossible to locate any source in a few moments. References are in allinstances to Chomsky’s work, unless explicit indication to the contrary isgiven. Much of Chomsky’s work is extremely technical and I haveattempted to simplify his ideas in the interest of comprehensibility.Nonetheless, I have occasionally included a brief technicality in order tomake it clear to my professional colleagues what it is I am simplifying. Inevery case, it is worth emphasizing that the linguistic examples I cite willneed mulling over, if their implications are to be fully grasped.

Chapter 1 begins by putting language and the study of language in awider context as part of the scientiWc investigation of human nature. Thisinvolves a discussion of the structure of mind, with evidence drawn fromstudies of both normal and pathological cases of the dissociation ofhuman faculties, and with language as the ‘‘mirror of the mind.’’ Thisopening chapter is followed by a detailed and partly historical expositionof Chomsky’s linguistic theorizing, which constitutes the bedrock onwhich the rest is built. The aim of this section is to give the reader someunderstanding of current theory by showing how we got where we are. Anaccount is given of the ideas for which Chomsky is best known (deep andsurface structure, for instance) and why they are no longer part of hiscurrent Minimalist framework; but most importantly, I try to give a Xavorof the kind of argument that Chomsky has used in his work over the lastWfty years. The next two chapters are devoted to the psychological andphilosophical implications of Chomsky’s work. Chapter 3 looks at thevexed question of what is meant by psychological reality, and providesevidence for it from language processing, from the child’s acquisition of aWrst language, and from language breakdown in pathology. At the core ofthis chapter is a discussion of Chomsky’s solution to ‘‘Plato’s problem,’’the puzzle of how children can acquire their Wrst language on the basis ofso little evidence. Chapter 4 turns to the philosophical aspects of

3Introduction

Chomsky’s ideas, outlining his intellectual commitments to realism,mentalism and naturalism, and explaining the controversies which havesparked so much debate in the philosophical community. The Wnalchapter is devoted to a discussion of his political ideas and how these Wt inintellectually with his ‘‘academic’’ work. Despite Chomsky’s own dis-avowal of any very close connection, it can be argued that there arefundamental ideas of rationality, creativity, and modularity which drawthe disparate strands of his output together. The book ends with anannotated bibliography.

The task of summarizing Chomsky is daunting, and I am conscious ofLeonardo da Vinci’s aphorism that ‘‘abbreviators do injury to know-ledge.’’ Chomsky’s output is vast: he has published about seventy-Wvebooks, hundreds of articles, and written tens of thousands of letters. Hismastery of a huge literature is awe-inspiring: in current aVairs throughoutthe world, in politics, history, linguistics, philosophy, psychology, mathe-matics . . . there are few areas where he has no knowledge. To achievethis mastery of many Welds demands ‘‘fanaticism’’ plus in his words, theability and dedication to ‘‘work like a maniac.’’ It also takes immensecourage, ceaseless energy and the sacriWce of any leisure. He wrote: ‘‘Ittakes a big ego to withstand the fact that you’re saying something diVerentfrom everyone else.’’ He views his own contribution as ‘‘pre-Galilean,’’though Berlinski is probably right to consider him ‘‘As big as Galileo.’’ Atthe end of the sixteenth century Galileo founded the experimentalmethod which underpins the whole of modern science; at the end of thetwentieth century Chomsky is generally viewed as the originator of thecognitive revolution which is beginning to extend that method to thestudy of the mind.

Not everyone shares this positive evaluation of him. The philosopherRichard Montague reportedly called him one of the ‘‘two great frauds oftwentieth century science’’ (the other was Einstein, so at least he was ingood company); he has been viliWed as an ‘‘opportunist, . . . applauder ofcorruption, and apologist for government indiVerence to protests againstwar and colonialism’’; he has been called the ‘‘great American crackpot’’and ‘‘outside the pale of intellectual responsibility.’’ He has been repeat-edly jailed for his political activism and has frequently been the victim ofdeath threats. Such polarization of opinion demands explanation, andone of the reasons for writing this book is to provide the foundations forsuch an explanation. Chomsky says: ‘‘You have a responsibility to explainwhy what you are doing is worth doing.’’ For me, his work is illuminating,but I think it is under-appreciated and worth broadcasting more widely,so I have tried to distill the essence into a few brief chapters.

4 Introduction

On heroes and inXuences

Most people need heroes to act as role models, whose exploits they canemulate or, more mundanely, simply use as a basis for deWning the kind ofactivity it is appropriate, morally defensible, and at least partly feasible tofollow. This is not the mindless homage of hero-worship. Close scrutinyusually leads to the discovery that one’s heroes – like everyone else in theworld – have feet of clay, which can be an encouragement if it puts themon the same mundane plane as oneself. I am happy to admit thatChomsky is a hero for me. It does not follow that I always agree with him,though if I didn’t agree with him on many issues, I almost certainlywouldn’t have written this book: I do not identify with those who idolizepolitical leaders because of their strength of leadership, irrespective of thedirection in which they lead.

For Chomsky ‘‘Nobody is a hero,’’ and he usually avoids answeringquestions about whom he admires, though the list of those who haveinXuenced him and whom he respects is lengthy. It includes anarchistthinkers like Mikhail Bakunin, Peter Kropotkin, and Rudolf Rocker; theleft Marxist Anton Pannekoek; a long series of philosophers: Descartes,Humboldt, and Rousseau; John Dewey and Charles Sanders Peirce;more recently Nelson Goodman and W. V. Quine; linguists like ZelligHarris and Otto Jespersen; and libertarians like A. J. Muste and BertrandRussell (‘‘one of the very few people that I actually admire’’). At a greaterremove, it would doubtless include Galileo, Kant, and Newton. Some ofthe inXuences are less obvious than others: Ahad Ha-’am, a culturalZionist at the turn of the century, whose work was later considered notonly to be anti-Zionist, but to show ‘‘an excess of rationalism,’’ was anearly inXuence on both Chomsky and his parents. His father, WilliamChomsky, not only inXuenced him politically, but also exposed him earlyin life to classical Semitic philology: his book Hebrew: The Eternal Lan-guage (dedicated to Noam and his brother) appeared in the same year,1957, as his son’s Syntactic Structures, the accepted beginning of theChomskyan revolution.

Despite his ability to overthrow the ediWces he has himself created,there is a timelessness about his moral commitments and the intellectualfoundations of his work, that clearly date to his childhood. His views arenever adopted unthinkingly, and none of the inXuences is accepteduncritically. In linguistics as in politics what is striking is Chomsky’sability to see to the heart of issues; to extract that which is defensible andconstructive and to dismiss that which is dishonest, immoral or irrational.In both domains he defends the insights of those whose general positionhe has no time for and criticizes the perceived failings of his intellectual

5Introduction

allies. Moreover, he does it with grace and humor. Intellectually, he isperhaps closest in spirit, as well as achievement, to Darwin, who wrote tohis friend and mentor Henslow: ‘‘I believe there exists, & I feel within me,an instinct for truth, or knowledge or discovery, of something [the] samenature as the instinct of virtue, & that our having such an instinct is reasonenough for scientiWc researches without any practical results ever ensuingfrom them.’’

6 Introduction

1 The mirror of the mind

One reason for studying language – and for me personally the mostcompelling reason – is that it is tempting to regard language, in thetraditional phrase, as ‘‘a mirror of mind.’’ Chomsky, 1975a: 4

Frogs are not like us. They are better at catching Xies but not, it seems, atexplaining how they do it. The frog mind is narrowly specialized tocontrol tasks such as locating small black specks, escaping predators andWnding mates, but not for reXecting on the ethics of eating insects or theissue of equal rights for toads.

This view of the limited intellectual capabilities of amphibians is un-likely to be controversial. If I extended it to apes the reaction might bediVerent, and it would clearly be false of humans. How do we know?Because humans can tell us so and the others cannot. Although having alanguage is not a prerequisite for having a mind, language is overwhel-mingly our best evidence for the nature of mind. Language is deWnitionalof what it is to be human, and the study of language is a way in to the studyof the human, but not the frog, mind.

Despite the complexity and variety of animal communication systems,no other creature has language like ours. Although chimpanzees andbonobos can be taught to manipulate an impressive array of signs and usethem to communicate with us or with each other, human language, inparticular the syntax of human language, is sui generis. As far as we know,even the singing of whales and the color communication of cuttle-Wshhave nothing like syntax. In one respect this uniqueness is trivial: theinherent interest of our abilities would not be diminished just because itturned out that our close genetic relatives had even more in common withus than we had previously suspected. But if we want to understand whatwe are – how we are unique – our linguistic ability is central, andChomsky’s work in generative grammar provides the most important andradical insights in this domain. He has achieved this by studying languagewith the rigor and the methodology of the hard sciences in combination

7

with the philosophical insight of the Cartesian tradition in a way that hadpreviously never been attempted.

In this chapter I look Wrst at the implications of the assumption thatlinguistics should be part of the natural sciences, and then at the positionof language in relation to the rest of cognition. This involves investigatinga range of human abilities, their interrelations and dissociations, thecontrast between knowledge of language and the use of that knowledge,and taking a Wrst glance at questions of innateness and the relation oflanguage to thought.

Linguistics as a science

Linguistics had long been deWned as the scientiWc study of language, butthe science was restricted to taxonomy and a naıve methodology.Hockett, one of the leading Wgures of the American structuralism thatChomsky’s revolution replaced, opens one of his early papers with thedeWnitional claim that ‘‘linguistics is a classiWcatory science.’’ One ofChomsky’s achievements has been to make plausible the claim thatlinguistics is scientiWc in the more interesting sense that it can provide notonly explicit descriptions but also explanations for the classiWcation.There are several strands to such a claim. The Wrst is that linguisticsprovides a general theory explaining why languages are the way they are:each language is a particular example of a universal faculty of mind,whose basic properties are innate. The second is that the theory shouldspawn testable hypotheses: like a physicist or a biologist, the linguistmanipulates the environment experimentally to see what happens and,crucially, he or she may be wrong. The experiments are usually not ashigh-tech as those in the hard sciences, but they allow for testing: if youranalysis entails that English speakers should Wnd John speaks XuentlyEnglish as acceptable as John speaks English Xuently, then it is wrong andmust be replaced by a better one. A corollary of this emphasis on seekingtestable explanations is that the central concern is evidence rather thandata. Every linguist (a term which is ambiguous between theorist oflanguage and polyglot) has suVered the question ‘‘So how many lan-guages do you speak?’’ It is often hard to convince people that the answerdoesn’t really matter. Having a little knowledge of half a dozen languagesis less useful than knowing one language with native proWciency. You maybe reasonably Xuent in French, for instance, without being quite surewhether the French equivalent of the unacceptable English sentenceabove is acceptable or not: ‘‘Jean parle couramment l’anglais.’’ If you’renot sure, your knowledge is of little more use than an unreliable balance.Even if I assure you that it is acceptable, and that this reXects a systematic

8 The mirror of the mind

diVerence between the two languages, this is still just another fact until Ican use it as evidence for some particular theoretical assumption, atwhich point it may acquire vital importance for deciding between con-Xicting theories.

Linguistics before Chomsky (and in many cases even now) was preoc-cupied, like Linnaean botany or Victorian entomology, with achievingcomplete coverage of the respective Welds. Examples are legion, fromHjelmslev’s Prolegomena, which begins with the claim that linguistictheory must permit descriptions which are ‘‘exhaustive,’’ to current ver-sions of Construction Grammar, which criticizes the generative paradigmbecause ‘‘it doesn’t allow the grammarian to account for absolutelyeverything in its terms.’’ It is essential to collect enough data to guaranteerepresentative coverage – missing out marsupials in a taxonomy of mam-mals would be a serious omission – but trying to achieve exhaustivecoverage is a wild-goose chase, and such criticisms are misconceived. Theset of facts is potentially inWnite, but facts which can be used as evidencefor some particular hypothesis are much harder to come by. Considerword order.

DiVerent languages have diVerent word orders: in some, like English,sentences are typically of the form Subject Verb Object (SVO), so we sayFrogs eat Xies; in others, like Japanese, they are of the form Subject ObjectVerb (SOV), so the equivalent sentence would have the order Frogs Xieseat; in yet others, like Arabic, they are of the form Verb Subject Object(VSO), with the order Eat frogs Xies. Assuming that it makes sense to talkof diVerent languages having diVerent characteristic word orders, it wassuggested some years ago that all the world’s languages fell necessarilyinto one of these three types (SVO, SOV, and VSO). The suggestion wasplausible because these are the three orders where the subject precedesthe object which, given our own language background, feels logical. Totest this claim it’s no use just collecting more examples of languages likethe ones mentioned: it’s easy to Wnd hundreds more languages thatconform to the generalization. What is needed is a list of the world’slanguages suYciently exhaustive to tell us whether there are any excep-tions: languages with the word orders VOS, OVS, or OSV. As it happens,the suggestion was wrong: all these types do occur (although the last twoin particular are extremely rare), so all the six logically possible orders areattested. It follows that, as far as this particular observation is concerned,there is nothing more to be said. Whatever language one looks at next, itwill fall into one of the six types listed, because there are no other logicalpossibilities, so every language will exemplify one of the possibilities wealready know about. Even the signed languages of the deaf manifest thesame kind of word order diVerences as spoken languages. Accordingly, if

9Linguistics as a science

word order were the only consideration of interest, there would be nopoint in trekking oV to the Highlands of New Guinea to search foranother example of something we already have. Of course we still haveinnumerable interesting questions: Why are some of these orders so rare?What other properties, if any, correlate with the word order manifested bya particular language? What happens when we consider indirect objectsand adverbs, and other possible additions? It may well be that evidenceabout these issues will come precisely from as yet unknown languages,but to investigate these constructively we need more, and more complex,hypotheses. Our knowledge of language and languages is by now suf-Wciently complex that we are more likely to gain insight by looking ingreater depth at well-studied languages than by looking superWcially atrelatively unknown ones. I spent a fascinating year learning and studyingthe Nupe language of Nigeria, and have used the language ever since tocheck out various claims about language, but many of the things I want tocheck are beyond my Nupe abilities and I have to have recourse to mynative intuitions in English or to the native intuitions of speakers of Nupeto settle the issue.

At this point you might rightly object that saying English is SVO is toosimplistic, because many sentences diverge from this favored pattern. InWhat do frogs eat? or Flies are what frogs eat, the object appears at thebeginning of the sentence, hence before the subject. Such orders occursystematically in English and cannot just be ignored, even if other devi-ations are characteristic only of poetry or archaic forms of the languageand can perhaps be safely left out of consideration. For instance, in thesaying What the eye doesn’t see, the heart doesn’t grieve, the heart is the objectof grieve, so the expression means that what you don’t see doesn’t ‘‘grieveyour heart.’’ There is a sense in which such sayings are part of English,but to infer from this that English word order allows the object either toprecede or to follow the verb would be grossly misleading, predicting thatFrogs Xies eat is on a par with Frogs eat Xies; which it patently is not.Indeed, to bring the saying into conformity with their form of English,many people have changed it to What the eye doesn’t see, the heart doesn’tgrieve over, thereby making the heart unambiguously the subject of grieve.This observation highlights an important and basic assumption of Chom-skyan theory: the notion of language that is being investigated is thelanguage of an individual, not the language of a community or a countryor an era. This special notion is accordingly referred to as ‘‘I-language’’(for ‘‘individual’’), and linguistics is viewed as part of cognitive psychol-ogy, an investigation of what an individual, any individual, knows invirtue of being a speaker of a language. It follows that if we are to describeaccurately what our knowledge of English (or any other language) con-sists in, and if we are to explain why our knowledge takes the form it does

10 The mirror of the mind

and how we come by it, we need to separate out our idiosyncraticfamiliarity with poetic and archaic expressions and concentrate on thecore knowledge reXected in our normal usage, however hard it may be todeWne precisely what that means.

There is a danger associated with the search for depth and explanation:looking for that narrow range of data which bear on a particulartheoretical problem, one may overlook data which would be even morerelevant if only one could recognize the fact. Choosing to ignore theexample of grieve because it is archaic may deprive one of a source ofuseful evidence. In this situation one relies on a combination of factors tosave one from egregious error: a knowledge of the literature outside one’simmediate domain of interest, the correctives of colleagues and thecriticisms of rivals, and serendipity. Amassing new data from as wide arange of languages as possible is a valuable enterprise, indeed an in-valuable enterprise, provided that it is remembered that all data need tobe analyzed and that there are no data without some kind of theory: thatis, the facts need to be described within some framework that makes themuseful to other linguists. Knowing that tankpolozı ewa edzuzı is the Nupefor ‘‘toads catch spiders’’ is of little use to you unless you know what thewords mean, so that you can tell which is subject, which verb and whichobject. Even the notions ‘‘subject,’’ ‘‘verb’’ and ‘‘object,’’ which I havebeen taking for granted, on the assumption that an example or two wouldmake it clear what I meant, are problematic. Some linguists use them,some do not; and those who do use them need to account for the fact thatthe interpretation of such categories is not consistent across all sentences:there is only a partial match between the grammar and the meaning, asshould be apparent from a moment’s reXection on the diVerent inter-pretations given to the subject John in John broke a leg and John broke anegg.

Like physics, but unlike logic or literary criticism, linguistics is anempirical science. That is, on a Chomskyan interpretation, which takesthe speaker’s mentally represented grammar to be the correct focus forinvestigation, it makes sense to claim that one analysis is right and anotherwrong. Every time a linguist describes a sentence or postulates a prin-ciple, he or she is making innumerable empirically testable predictions.Those linguists who claimed that subjects precede objects in all languageswere simply wrong: interestingly wrong, because the refutation of theirclaim has led to greater understanding of the nature of language, butwrong. By contrast, a literary critic who claims that ‘‘a song is a form oflinguistic disobedience,’’ or a logician who says that ‘‘nothing is both an Xand a non-X’’ are not formulating hypotheses to be checked out andtested by their colleagues. The observations may be useful, insightful,even inspired, but they are not empirical.

11Linguistics as a science

The nature of idealization

If science aims to explain a few things rather than to describe everything,some things (such as poetic survivals) have to be left out. When Galileodevised the law of uniform acceleration for falling bodies, either bydropping weights from the leaning tower of Pisa or rolling balls down aninclined plane, he ignored the eVects of wind resistance and friction. Infact, we often don’t even know if he carried out the experiments hedescribed: they were thought experiments that relied for their validity asmuch on logical argumentation as precise observation. This was notsloppy experimental practice or ignorance of the eVect of the air on fallingfeathers, rather it was a sensible idealization. The eVect of wind resistanceor friction is irrelevant to the generalization Galileo was seeking to estab-lish. All of science is characterized by the need to exclude from con-sideration those factors which, while undeniably real, are not pertinent tothe issue under investigation. We know that heavenly bodies are notmathematical points, but they can be treated as such for the purposes ofgravitational theory. We know that Boyle’s law applies to ‘‘ideal’’ gases,and that the gases we observe appear less well-behaved, but we do nottake this observation to impugn Boyle’s discovery or to invalidate theidealization. In general, the role of scientiWc experimentation is to get uscloser to the truth, to the ideal, by eliminating irrelevant extraneousconsiderations. In other words idealization reveals what is real, but isusually hidden from view by a mass of detail. All scientists accept thereality of the inverse-square law, whether it is being used to explain theintensity of the light reaching us from a star, of the sound reaching usfrom a jet engine, or the attractive force of a magnet, even though themessiness of experiments means that their measurements never mirror itexactly, giving ‘‘a distortion of reality’’ in Chomsky’s phrase.

One of the idealizing claims in linguistics that has caused the greatestmisunderstanding is Chomsky’s much quoted passage on the Wrst page ofhis classic book Aspects of the Theory of Syntax that ‘‘Linguistic theory isconcerned primarily with an ideal speaker-listener, in a completely homo-geneous speech-community . . . [who] is unaVected by such grammati-cally irrelevant conditions as memory limitations, distractions, shifts ofattention and interest, and errors (random or characteristic) in applyinghis knowledge of the language in actual performance.’’ There are twoissues: Wrst, is such idealization defensible? Second, if it is, which idealiz-ations are likely to be fruitful and not beg important questions? Theanswer to the Wrst question is self-evident: all science involves idealiz-ation, which ‘‘frees the explorer from irrelevant demands,’’ and it isirrational to attempt to avoid it. When we contemplate real speaker-

12 The mirror of the mind

listeners (or gases) in their full complexity, we usually fail to see theprinciples which underlie their behavior, simply because of interferingfactors of one kind or another. To see those (real) principles in full clarityit is necessary that some things be fruitfully ignored. A major stumblingblock for many working in the humanities is that what is fruitfully ignoredin the linguistic study of language may be precisely what they are interes-ted in: the language of poetry, for instance. While acknowledging thatmany aspects of language remain outside the domain of scientiWc inquiry,Chomskyan linguistics has demonstrated that it is fruitful to incorporatesome aspects into natural science. Even here a legitimate focus of ar-gument and disagreement arises with regard to the second question: whatidealizations are to be made? In Aspects Chomsky claimed that suchphenomena as limitations of memory, tiredness, and minor variationsbetween speakers are irrelevant to our understanding of the notion‘‘knowledge of language.’’ English grammars do not in general include astatement saying: ‘‘Insert ‘um’ here if you are tired or uncertain,’’ eventhough languages diVer in terms of the hesitation noises they allow, andeven though the places where such phenomena can occur are not entirelyrandom, but can provide interesting clues about language production. Ifyou say um while speaking French, for instance, you give yourself away asa foreigner (the French say something like [oe]); and um in English ismuch more likely to occur after of, rather than before it, in phrases such asin the middle of Texas. Such examples of idealization are relatively unprob-lematic, so let’s consider a somewhat more vexed case by anticipating thediscussion in chapter 3 of the child’s acquisition of its Wrst language.

First language acquisition takes place within a particular window ofopportunity known as the critical period (see below and chapter 3), whichlasts for a few years and ends (at the latest) at puberty. Given what weknow about children’s development, it sometimes comes as a shock toread that Wrst language acquisition is idealized to ‘‘instantaneity.’’ Howcan a process which extends over several years be sensibly treated asthough it took no time at all? The paradox is only apparent not real.Although there is a striking uniformity across children learning their Wrstlanguage in respect of the stages of development they go through, they dononetheless diVer from each other. For instance, in the course of mas-tering the system of negation, one child may form negative sentences byusing an initial no, while another may use a Wnal no, giving the contrastbetween no like cabbage and like cabbage no. Despite this developmentaldiVerence, both children will end up with the same system of negation inwhich they use the correct adult form: I don’t like cabbage. As far as we cantell, the early diVerence in the children’s systems has no eVect at all on thegrammar they end up with. If the focus of our interest is on what’s known

13The nature of idealization

as ‘‘the logical problem of language acquisition’’ – children’s transitionfrom apparently having no knowledge of a language to being able to speakand understand it like adults – then we have support for the idealization toinstantaneity, which says that the diVerent stages children go through inthe language acquisition process are of no import to their ultimate psy-chological state. Of course, it may turn out that this surprising claim isfalse. It might be that suYciently sophisticated tests of grammaticalityjudgment, or investigations of neural Wring, or subsequent historicalchanges in the language concerned, showed that the two children’s gram-mars were crucially diVerent, and diVerent in ways that could explainother mysterious facts about knowledge of language. It’s possible, butthere is (as yet) no evidence, and the idealization is accordingly justiWed: itleads us to an understanding of one aspect of the real system we arestudying.

There is of course no guarantee that the idealizations made are themost fruitful ones or even that they are not pernicious. Chomsky talks of‘‘the dangers of illegitimate idealization that isolates some inquiry fromfactors that crucially aVect its subject matter, a problem familiar in thesciences.’’ Many sociolinguists feel that Chomsky’s idealization to thehomogeneity of the speech community comes in the ‘‘illegitimate’’ cat-egory. A recent example is provided by the pained claim that ‘‘theidealization program in practice means that at least for now we should notbe studying any community where we perceive a considerable degree of‘mixing’ or ‘impurity.’’’ That is, of course, essentially all communities,with the implication being that nothing should be studied. This wouldobviously be ridiculous, but fortunately it reveals a deep misunderstan-ding. We should study whatever promises to cast light on the hypothesiswe are currently entertaining, and variational data may well cast suchlight when properly interpreted. The force of Chomsky’s idealization isthat the variation is not necessary to an understanding of the humanlanguage faculty. In particular, it is not the case, contrary to what thewriter goes on to say, that we have only a ‘‘limited range of data’’: we aredrowning in data; what we need is clearly articulated hypotheses forwhich some subset of these data can constitute evidence. If the hypothesisone is interested in itself crucially involves variation, then obviously theidealization to homogeneity is pernicious, but Chomsky is preoccupiedwith the general properties of the language faculty: what we, as speakersof any language, know and how we come by that knowledge. No onedenies that there is variation, but to claim that we need to ‘‘look for waysof documenting it in order to understand language as part of the humancondition’’ is doubly misleading: Wrst, because it suggests that suchdocumentation will provide understanding; second, because no theory of

14 The mirror of the mind

‘‘the human condition’’ which would throw up testable hypotheses issuggested. Worse, there is an implicit suggestion that it would be impos-sible for a child to learn language unless it was confronted with thecontradictory information of dialect mixture, false starts and the like.This seems inherently implausible and no evidence for such a claim isever presented. Despite these strictures, I shall suggest in chapter 4 thatvariation does in fact have philosophical implications which are relevantto Chomsky’s program.

While idealization is necessary, it must be emphasized that idealizingaway from speech errors, for instance, still allows one to use performancemistakes such as slips of the tongue as evidence for the nature of theknowledge that is the focus of enquiry. All our understanding of linguisticknowledge, like our understanding of quantum physics or molecularbiology, has to be supported by evidence, and where that evidence comesfrom is limited only by our imagination and ingenuity. On the assumptionthat our knowledge of language in part determines how we say what we dosay: that ‘‘the rules of grammar enter into the processing mechanisms,’’malfunctions of the system can provide evidence for the nature of therules of grammar. Vicki Fromkin illustrates this possibility on the basis ofthe regular formation of the past tense in English. The usual rule is thatyou add -ed to the verb, producing talked from talk, kissed from kiss, and soon, though for a considerable number of irregular verbs, there are com-plications: come gives came rather than comed, and we say left rather thanleaved. She then cites a host of slips of the tongue involving over-regulariz-ation such as the last I knowed about it, he haved to have it, if he swimmed,indicating that normal adult speakers of the language don’t just access astore of learned items, but that they actually use a rule of the kind thatlinguists, for independent reasons, posit as part of the grammar. Theexample is elementary, but it puts into perspective the objection that theChomskyan framework ignores a crucial range of data. In fact, a majorand innovative characteristic of Chomsky’s linguistics is its exploitation ofthe previously neglected fact that we are able to recognize immediatelythat some sentences are ungrammatical: we have what one might call‘‘negative knowledge.’’ Hamlet can tell Ophelia that I loved you not, butwe know we have to say I didn’t love you; Othello can ask DesdemonaWent he hence now?, which we understand easily enough, though we knowwe have to rephrase it in current English as ‘‘Did he go?’’ We can sayequally well both I asked the way to the school and I inquired the way to theschool, but whereas I asked the number of people in the class is Wne, I inquiredthe number of people in the class is odd. We should be as surprised by the factthat we have these intuitions as by the fact that apples fall down not up, orthat the sea has waves. Newton was not the Wrst to notice apples falling,

15The nature of idealization

but his insight that why apples fall is in need of an explanation ledultimately to his theory of gravity. Chomsky was not the Wrst to notice theelementary facts I have cited here, but his insight that our intuitions cantell us something profound about the human mind is of comparableimportance.

Common sense

The similarity of linguistics to the other sciences becomes salient as soonas one tries even to demarcate the Weld of study. Our common-senseunderstanding of the world gives a useful hint as to the domains that wemight expect to be illumined by scientiWc research, but there should be noexpectation that the categories of common sense should carry over intactto a scientiWc theory. The case of language is no diVerent in this respectfrom the case of physics or biology. We do not deride physical theoriesbecause physicists’ ideas about sunsets deviate from common sense, orbiological theories because they do not provide an account of the generalproperties of pets. Likewise we should not attack linguistic theoriesbecause they have no room for traditional notions of ‘‘language,’’ asrevealed in such usage as Chinese is the oldest language, or The Englishlanguage has been spoken in Britain for 1000 years, or She uses the mostappalling language. These remarks may be true, and they are clearly aboutlanguage, but they are not observations of linguistics, because they followfrom no linguistic theory. More importantly, it is almost certainly the casethat there is no theory of language in the sense in which it is being used inthese examples. Similarly, The book fell oV the table is a statement about thephysical world, but it is not a statement couched in the language ofphysics, and there are no physical laws which would enable one todiVerentiate such a remark from The book fell on the Xoor: an observationthat could truly describe exactly the same situation. Here too, it is almostcertainly the case that there is no physical theory of books or of falling:these are just not domains which lend themselves to theory construction,however salient they may be to us as humans. The other side of this coin isthat linguists, again like physicists, need to use concepts and vocabularythat are alien to non-linguists or non-physicists. The technical terminol-ogy of generative grammar can be as startlingly opaque as the vocabularyof quantum physics; only the mathematical underpinning is less forbid-ding. In both domains science and common sense frequently come apart,but they share the need to break problems down into a manageable size.

16 The mirror of the mind

Modularity

Humans are complex. From this it follows by virtue of the meaning of‘‘complex’’ that we have internal structure, and the mind is one part ofthat structure. But the mind itself is not an undiVerentiated general-purpose machine: it is compartmentalized in such a way that diVerenttasks are subserved by diVerent mechanisms. The mind is ‘‘modular.’’Sight and smell, taste and touch, language and memory, are all distinctfrom each other, from our moral and social judgment, and from ourexpertise in music or mathematics. In many cases, we even have a goodidea of where in the brain these diVerent functions of the mind arelocalized, though the question of localization is independent of the fact oftheir modular status. Even if we did not know which bit of the brain wasresponsible for interpreting the messages from our eyes and ears, wewould still have no hesitation in separating vision from hearing.

It has been recognized for two thousand years that the language facultycan be selectively impaired as the result of damage to diVerent parts of thebrain, but it is only since the middle of the nineteenth century that wehave had systematic evidence for the ‘‘lateralization’’ of language: that is,that our linguistic ability is largely the preserve of the left hemisphere ofthe brain, while visuo-spatial abilities are largely the responsibility of theright hemisphere. In each case it is necessary to generalize: both languageand vision are so rich and diverse that it would be naıve to expect a singlesimple relation between mind and brain. Parts of the visual systemresponsible for the perception of color, shape, and motion involve inter-action among many parts of the brain, and face processing appears toshow some sharing of responsibilities between the hemispheres, with theright side concentrating on holistic patterns and the left on individualfeatures. Similarly, diVerent linguistic functions are typically localizeddiVerently, with the left hemisphere taking the major responsibility forsyntax, but semantic and pragmatic processes also invoking the righthemisphere. Irrespective of the detail of their neural localization, it isuncontroversial that diVerent parts of the brain are ‘‘domain-speciWc’’ inthat they are dedicated to sight, to smell, and to the other senses. It wouldmake little sense to postulate abstract structures that are neutral asbetween sight and taste, for instance, as the incoming stimuli are radicallydiVerent in kind, and the representations the mind constructs out of themare likewise idiosyncratic: there is little in common between the percep-tion of purple and the smell of soap.

In Jerry Fodor’s inXuential work (inspired in part by Chomsky’s),human cognition is treated in terms of a basic dichotomy between the‘‘central system’’ and a number of ‘‘input systems.’’ The senses – taste,

17Modularity

sight, smell, hearing and touch – are dedicated input systems each ofwhich constitutes one module (or cluster of modules) of the mind, andfeed into the central system, which is responsible for rational thought,problem solving and what philosophers call the ‘‘Wxation of belief.’’ (Suchbeliefs need not be profound. On the basis of the smell of burning, thesight of smoke and one’s memory of similar past events, one comes tobelieve that the toast is burning.) Fodorian modules have a number ofwell-deWned characteristics: they are specialized for particular domains,they operate fast and mandatorily (one has no choice but to hear soundsone is exposed to), they are ‘‘hard-wired’’ with a particular part of thebrain dedicated to each, their structure and function are largely innatelydetermined, and they are ‘‘informationally encapsulated’’; that is, theyoperate without interference from central control. Fodor makes twofurther controversial claims not shared by Chomsky: that language is aninput system analogous to those devoted to the senses, and that thecentral system is essentially unstructured and putatively uninvestigable.The reason for this pessimism about the central system is that you cansolve problems by drawing on your knowledge from every conceivabledomain: the celebrated physicist Richard Feynman got his initial in-spiration for solving a problem in quantum theory from seeing a cafeteriaplate wobble as someone tossed it spinning in the air. If you can make aconnection between ‘‘the axial wobble of a cafeteria plate and the abstractquantum-mechanical notion of spin,’’ it seems likely that your mind canmake connections of every conceivable kind. We presumably don’t havebits of our brain specialized for dealing with spinning plates, still less forquantum mechanics, so this suggests a construct of awesome complexity.Nevertheless, the pessimistic conclusion is unnecessary; a more approp-riate one is that the central system, while internally structured, must allowessentially unlimited interaction among its components in a way thatFodorian modules deWnitely do not.

The most important criterion for modularity in Fodor’s sense is infor-mational encapsulation, which says that the internal workings of anymodule are oblivious to the workings of the central system. The classicexample is provided by the Muller-Lyer optical illusion, in which twolines, Xanked by inward or outward pointing arrow-heads are displayed:

The visual system perceives the lower line as longer than the upper line.Even if you take a ruler and convince yourself that the two lines are indeedof identical length, your eyes still interpret them as being diVerent. That

18 The mirror of the mind

is, the working of the visual system is impervious to the explicit knowledgeprovided by the central system. Examples of this kind are legion: forinstance, the visual system seems to have two distinct modes, one acces-sible to consciousness and one not. When we move around in the world,with the unerring accuracy characteristic of normal unthinking activity,our movements are subserved by a system inaccessible to consciouscontrol, indeed consciousness makes things worse. In an interestingexperiment, subjects were presented with two discs in an array where avisual illusion made judgment of their relative size problematic, and wereasked to grasp the left-hand disc if it was larger than the right hand one.Subjects were consistently wrong in their conscious visual judgment ofhow large the disc was, but measurements of their grasping hand as itreached for the disk demonstrated that unconsciously they must know thecorrect size, because the conWguration of their hands reXected the actualsize of the disc with considerable accuracy.

A linguistic example of such encapsulation comes from research onmemory by Dan Schacter and his colleagues. Subjects were Wrst read a listof words and were subsequently presented with a written list containingsome of the words they had heard and some they had not. Judgments insuch experiments are notoriously unreliable. If you have heard sharp,point, prick and pain, you are likely to ‘‘remember’’ hearing needle, even ifyou have not in fact heard needle at all. But your brain really does know.Schacter used a PET scan to monitor the Xow of blood in diVerentregions of the brain while subjects were deciding whether or not they hadheard the words before. This showed equivalent activity near the hippo-campus both for words that had been heard and for those that had not.But for those words that had been heard previously, there was additionalactivity in the left temporoparietal region, an area responsible for proces-sing auditory information. The hippocampus is generally associated withmemory, the major preserve of the central system, and the left temporallobe is the major locus of the language faculty. It seems then that theinternal workings of the language module are inaccessible to the centralsystem in precisely the way modularity predicts.

Chomsky’s work over several decades has provided a wealth of evi-dence that the language faculty does indeed constitute a separate module,akin in many respects to any other organ of the body. Moreover, he hasprovided more, and more rigorous, evidence about the precise internalstructure of this module than has been provided for any other domain(except perhaps vision). However, it is important to diVerentiate hisposition from Fodor’s with respect to the two ancillary claims mentionedabove. First, language is not merely an input system, in the way that visionor taste are, but is crucially also an output system geared to the expression

19Modularity

and communication of thought. Moreover, this output system is obvious-ly correlated with the input system: no one can speak only one languageand understand only another, diVerent one. What is common to the twosystems is necessarily part of a cognitive, central system; hence much oflanguage is ‘‘central.’’ Second, the ‘‘central system’’ (to the extent that itmakes sense to postulate a single entity of this kind) is itself intricatelystructured: it is multiply modular in that a number of human abilitieswhich go beyond the limits of purely sensory perception are apparentlyindependent of each other. For instance, we have separate mechanismsunderlying our ability to recognize faces, to provide moral judgments,and to evaluate social relations.

There are other areas in which Chomskyan and Fodorian modularitydiverge. Chomsky suggests that the language faculty interprets everythingthe external world throws at it, not just well-formed sentences andphrases, but even the squeaking of a door. This rather startling suggestionimplies a view of language which is incompatible with the domain-speciWcity of (Fodorian) modularity. As Chomsky observes, the princip-les of language and the principles of vision seem to be ‘‘entirely distinct.’’On Fodor’s account this distinctness means that the systems are sensitiveto diVerent kinds of input. On Chomsky’s account, each system may haveaccess to a wider range of inputs, but what it can do with that input isdetermined by the diVerences in the principles that characterize thatmodule. As the language system is in part parasitic on the auditory orvisual system (for spoken or signed language respectively) it is diYcult todecide at what stage the sensitivity to a particular form of stimulus makesthe transition from auditory/visual representations to linguistic represen-tations, or indeed whether the linguistic system plays any role in integ-rating information into the central system. As numerous writers haveobserved, coughs can be used with communicative intent, much as lan-guage can, so Chomsky’s position is part of a long tradition. It is alsocoherent, and there is no doubt that one processes the squeak of a door orthe swaying of a tree branch. However, I know of no evidence that suchprocessing involves the linguistic system as such, unless the central sys-tem’s operating language (the language of thought) is the same as thenatural language one knows, in which case such processing does involvethe linguistic system, even if only trivially so.

One might object that this multiplication of autonomous modules isunnecessary, and that life would be simpler if we were provided with asingle general-purpose processor. It might seem simpler, but we knowthat it would also be wrong. For example, evidence that humans comeinto the world supplied with a face-recognition module is now overwhel-ming. Thus infants a few hours old respond preferentially to faces and

20 The mirror of the mind

‘‘learn’’ to recognize them with markedly greater speed, generality, andfacility than they recognize animals or plants. Likewise, there is veryconsiderable evidence for a ‘‘theory-of-mind’’ module, whose operationis prerequisite to the understanding of other people, and whose absence(e.g. in autism) gives rise to a range of pathological behaviors. Forinstance, autistic children (and indeed normal children at an early age)are incapable of recognizing the fact that another person’s representationof the world may diVer from their own, and so systematically misinterpretsituations involving false belief. The classic demonstration of this is givenby the Sally-Anne task. In one version of this test, the subject and anotherobserver watch while the experimenter hides a toy. The observer is thensent out of the room and, in full view of the subject, the toy is moved fromits Wrst position and hidden in a new position. After ensuring that thesubject was aware of where the toy was Wrst hidden and of when theobserver was last present, he or she is then asked where the observer willlook for the toy on returning to the room. From about the age of four,normal people indicate the Wrst hiding place. Children under the age offour, and autistic subjects, usually indicate the second hiding place, wherethe toy actually is. That is, they are unable to entertain the idea thatsomeone else could have a representation of the world which deviatesfrom reality, they cannot understand ‘‘false belief.’’ The standard ex-planation for this phenomenon is in terms of the absence or malfunc-tioning of the theory-of-mind module (more accurately a theory of otherminds), a component of the mind that enables you to take someone else’spoint of view irrespective of whether that point of view conforms toreality.

It is an interesting empirical issue to determine the relations, mentaland physical, among the various components of the human organism. Isthe theory-of-mind module autonomous or dependent on the languagefaculty (or vice versa)? Is the language faculty autonomous or dependenton intelligence, or age, or eye color, or something else? The range ofpossibilities is vast. Fortunately the range of evidence is also vast. Thebest comes from dissociations, especially double dissociations.

Double dissociation

Respiration and digestion, locomotion and reproduction are clearly in-dependent of each other, and are served by autonomous if interactingorgans. This independence is seen most clearly when things go wrong inillness or disease. Breathing problems do not necessarily hamper one’sdigestion; marathon runners may be infertile: the functions ‘‘dissociate.’’As with the workings of the body, so with the workings of the mind:

21Double dissociation