Encyclopedia of CognitiveScienceVolume 2Lynn NadelUniversity of ArizonaVolume 1Volume 2Volume 3Volume 4Academic Achievement - Environmental PsychologyEpilepsy - Mental Imagery, Philosophical Issues aboutMental Models - Signal Detection TheorySimilarity - Zombiesnature publishing groupLondon, New York and Tokyo
Nature Publishing Group O 2003 Macmillan Publishers LtdAll rights reserved. No reproduction, copy or transmission of this publication may be made without written permission.No part of this publication may be reproduced, stored in or introduced into a retrieval system, or transmitted, in any form,or by any means (electronic, mechanical, photocopying, recording or otherwise) without the prior written permission of thepublisher unless in accordance with the provisions of the Copyright Designs and Patents Act 1988, or under the terms ofany licence permitting limited copying issued by the Copyright Licensing Agency, 90 Tottenham Court Road,London W I P 9HE, UK.Any person who does any unauthorized act in relation to this publication may be liable to criminal prosecution and civil claims fordamages.Great care has been taken to maintain the accuracy of the information contained in this work. However, neither Nature PublishingGroup, the editors nor the authors can be held responsible for any consequences arising from use of the information contained herein.Published byNature Publishing Group, 2003The Macmillan Building, 4 Crinan Street, London, N1 9XW, UKAssociated companies and representatives throughout the worldwww.nature.comISBN: 0-333-792610Distributed on behalf of the Nature Publishing Group in the United States and Canada byGrove's Dictionaries, Inc.345 Park Avenue South,New York,NY 10010-1707,USABritish Library Cataloguing in Publication DataEncyclopedia of cognitive science1. Cognitive science - EncyclopediasI. Nadel, Lynn153' .03Library of Congress Cataloging in Publication DataA catalog for this record is available from the Library of CongressTypeset by Kolam Information Services Pvt. Ltd., Pondicherry, IndiaPrinted and bound by The Bath Press, England
Linguistic RelativitvChomsky N (1961) Some methodological remarks ongenerative grammar. Word 17: 219-239.Fillmore CJ, Kempler D and Wang WS-Y (eds) (1979)Individual Differences in Language Ability and LanguageBehavior. New York, NY: Academic Press.Gerken LA and Bever TG (1986) Linguistic intuitionsare the result of interactions between perceptualprocesses and linguistic universals. Cognitive Science 10:457-476.Greenbaum S (1988) Good English and the Grammarian.London, UK: Longman.917Levelt WJM (1974) Formal Grammars in Linguistics andPsycholinguistics, 3 vols. The Hague, Netherlands:Mouton.McNair L, Singer K, Dobrin LM and AuCoin MM (eds)(1996) CLS 32: Papersfrom the Parasession on Theoy andData in Linguistics. Chicago, IL: Chicago LinguisticSociety.Newmeyer FJ (1983) Grammatical Theoy, its Limits and itsPossibilities. Chicago, IL: University of Chicago Press.Perry TA (ed.) (1979) Evidence and Argumentation inLinguistics. Berlin, Germany: de Gruyter.Linguistic RelativityIntermediate articleLera Boroditsky, Massachusetts Institute of Technology, Cambridge, Massachusetts, USACONTENTSDoes language shape thought?SpaceTimeLanguages differ dramatically from one another interms of how they describe the world. Does havingdifferent ways of describing the world lead speakersof different languages also to have different ways ofthinking about the world?DOES LANGUAGE SHAPE 'THOUGHT?Humans communicate with one another using anamazing array of languages, and each languagediffers from the next in innumerable ways (fromobvious differences in pronunciation and vocabulary to more subtle differences in grammar). Forexample, to say that 'the elephant ate the peanuts'in English, we must include tense - the fact that theevent happened in the past. In Mandarin and Indonesian, indicating when the event occurred wouldbe optional and couldn't be included in the verb. InRussian, the verb would need to include tense andalso whether the peanut-eater was male or female(though only in the past tense), and whether saidpeanut-eater ate all of the peanuts or just a portionof them. In Turkish, on the other hand, one wouldspecify (as a suffix on the verb) whether the eatingof the peanuts was witnessed or if it was hearsay. Itappears that speakers of different languages haveto attend to and encode strikingly different aspectsof the world in order to use their language properly(Sapir, 1921; Slobin, 1996). Do these quirks ofShapes and substancesObjectsConclusionlanguages affect the way their speakers thinkabout the world? Do English, Mandarin, Russian,and speakersk i endh up attending to, partitioning, and remembering their experiencesdifferently simply because they speak differentlanguages?The idea that thought is shaped by language ismost commonly associated with the writings ofBenjamin Lee Whorf (Whorf, 1956). Whorf, impressed by linguistic diversity, proposed that thecategories and distinctions of each language enshrine a way of perceiving, analyzing, and actingin the world. In so far as languages differ, theirspeakers too should differ in how they perceiveand act in objectively similar situations. This strongWhorfian view - that thought and action are entirely determined by language - has long beenabandoned in the field. However, definitivelyanswering less deterministic versions of the 'doeslanguage shape thought' question has proven to bea very difficult task. Some studies have claimedevidence to the affirmative (e.g. Boroditsky, 2001;Bowerman, 1996; Davidoff et al., 1999; Gentner andImai, 1997; Levinson, 1996; Lucy, 1992; Dehaeneet al., 1999), while others report evidence to thecontrary (e.g. Heider, 1972; Malt et al., 1999; Liand Gleitrnan, 2002).In recent years, research on linguistic relativityhas enjoyed a considerable resurgence, and much
918Linguistic Relativitynew evidence regarding the effects of language onthought has become available. This chapter reviewsseveral lines of evidence regarding the effects oflanguage on people's representations of space,time, substances, and objects.SPACELanguages differ considerably in how they describe spatial relations. Many such differenceshave been noted among English, Dutch, Finnish,Korean, and Spanish, among others (Bowerman,1996). For example, English distinguishes betweenputting things into containers ('the apple in thebowl', 'the letter in the envelope') and puttingthings onto surfaces ('the apple on the table', 'themagnet on the refrigerator door'). Cross-cutting thiscontainment/support distinction, Korean distinguishes between tight and loose fit or attachment.For example, putting an apple in a bowl requires adifferent relational term (nehta) from putting aletter in an envelope (kitta), because the first is anexample of loose containment and the secondan example of tight fit. Further, putting a letterin an envelope and putting a magnet on the refrigerator are both described by kitta because bothinvolve close fit.To test whether these cross-linguistic differencesare reflected in the way English and Koreanspeakers represent spatial relations, McDonoughet al. (2000) showed scenes involving tight or loosefit to Korean- and English-speaking adults. Afterthey had seen a few examples of either tight fit orloose fit, the subjects were shown an example oftight fit on one screen, and an example of loose fiton another. While Korean-speaking adults lookedlonger at the kind of spatial relation they had justbeen familiarized with, English speakers did notdistinguish between the tight- and loose-fit scenes,looking equally long at the familiar and novelscenes. Further, when given several examples oftight fit and one example of loose fit (or viceversa), Korean adults could easily pick out the oddpicture, but English speakers could not. Finally,McDonough et al. found that unlike adult Englishspeakers, prelinguistic infants (being raised in bothEnglish-speaking and Korean-speaking households) distinguished between tight and loose fit inthe looking-time test described above. This patternof findings suggests that infants may come ready toattend to any number of spatial distinctions. However, as people learn and use language, the spatialdistinctions reinforced by their particular languageare the ones that remain salient in their representational repertoire.Dramatic cross-linguistic differences have alsobeen noted in the way languages describe spatiallocations (Levinson, 1996). Whereas most languages (e.g. English, Dutch) rely heavily on relativespatial terms to describe the relative locationsof objects (e.g. left/right, front/back), Tzeltal (aMayan language) relies primarily on absolute reference (a system similar to the English north/southdirection system). Spatial locations that are northare said to be downhill, and those south are said tobe uphill. This absolute uphill/downhill system isthe dominant way to describe spatial relations between objects in Tzeltal; no relational equivalents tothe English terms front/back or left/right are available (Levinson, 1996).To test whether this difference between the twolanguages has cognitive consequences, Levinson(1996) tested Dutch and Tzeltal speakers in anumber of spatial tasks. In one study, participantswere seated at a table and an arrow lay in front ofthem pointing either to the right (north) or to theleft (south). They were then rotated 180 degrees to asecond table which had two arrows (one pointingto the left (north) and one to the right (south)), andwere asked to identify the arrow 'like the one theysaw before'. Dutch speakers overwhelmingly chosethe 'relative' solution. If the stimulus arrow pointedto the right (and north), Dutch speakers chose thearrow that still pointed to the right (though it nowpointed south instead of the original north). Tzeltalspeakers did exactly the opposite, overwhelminglychoosing the 'absolute' solution. If the stimulusarrow pointed to the right (and north), Tzeltalspeakers chose the arrow that still pointed north(though it now pointed left instead of right). Thus,Tzeltal speakers' heavy reliance on absolute reference in spatial description appears to have affectedtheir interpretation of (and performance on) a nonlinguistic orientation task.Further studies of this task showed that Englishspeakers (English is the same as Dutch in this respect) do not always favor relative responses; certain contextual factors can be used to induceEnglish speakers to produce both absolute andrelative responses on these tasks (Li and Gleitman,2002). This is not surprising since English speakersuse both absolute and relative forms in their language. It remains to be seen whether the samecontextual factors can induce Tzeltal speakers toproduce relative responses despite an apparentlack of relative terms in Tzeltal.In summary, the evidence available so far suggests that reference frames and distinctions madeavailable by one's language may indeed imposeimportant constraints on one's spatial thinking.
Linguistic Relativity'TIMELanguages also differ from one another on theirdescriptions of time. While all languages usespatial terms to talk about time ('looking forwardto a brighter tomorrow', 'proposing theories aheadof our time', 'falling behind schedule'), differentlanguages use different spatial terms. For example,in English, we predominantly use front/back termsto talk about time. We can talk about the good timesahead of us, or the hardships behind us. We canmove meetings forward, push deadlines back, andeat dessert before we're finished with our vegetables. On the whole, the terms used to order eventsare the same as those used to describe asymmetrichorizontal spatial relations (e.g. 'he took threesteps forward' or 'the path is behind the store'). InMandarin, front/back spatial metaphors for timeare also common (Scott, 1989). Mandarin speakersuse the spatial morphemes qian (front) and hbu(back) to talk about time. What makes Mandarininteresting for present purposes is that Mandarinspeakers also systematically use vertical metaphorsto talk about time (Scott, 1989). The spatial morphemes shling (up) and xili (down) are frequentlyused to talk about the order of events, roughlytranslated into English as last and next. Earlierevents are said to be shling or 'up', and later eventsare said to be xi2 or 'down'. In summary, bothMandarin and English speakers use horizontalterms to talk about time. In addition, Mandarinspeakers commonly use the vertical terms shlingand xi;.So, do the English and Mandarin ways oftalking about time lead to differences in howpeople think about time? Specifically, are Mandarin speakers more likely to construct vertical timelines to think about time, while English speakersare more likely to construct horizontal timelines?A collection of studies showed that Mandarinspeakers tend to think about time vertically evenwhen thinking for English (Boroditsky, 2001). Forexample, Mandarin speakers were faster to confirm that March comes earlier than April if theyhad just seen a vertical array of objects than ifthey had just seen a horizontal array. The reversewas true for English speakers. Another studyshowed that the extent to which Mandarin-Englishbilinguals think about time vertically is relatedto how old they were when they first beganto learn English. In another experiment nativeEnglish speakers were taught to talk about timeusing vertical spatial terms in a way similar toMandarin. On a subsequent test, this group of English speakers showed the same bias to think about919time vertically as was observed with Mandarinspeakers.This last result suggests two things: (1) languageis a powerful tool in shaping thought, and (2) one'snative language plays a role in shaping habitualthought (how we tend to think about time, forexample) but does not completely determinethought in the strong Whorfian sense (since onecan always learn a new way of talking, and withit, a new way of thinking).SHAPES AND SUBSTANCESLanguages also differ in the extent to which theymake a grammatical distinction between objectsand substances. For example, in English, objectslike candles and chairs have distinct singular andplural forms (e.g. one candle versus two candles),but substances like mud and wax do not. Further,objects and substances are distinguished in Englishin counting. While one can say 'one candle, twocandles, three candles' and so on, counting substances is a bit trickier. Instead of saying 'onemud, two muds', English speakers must specifythe unit of measurement such as 'one mound ofmud' or 'one cup of mud' (words like 'mound'and 'cup' here are called 'unitizers' because theyspecify the unit of measurement).Unlike English, some languages do not havea grammatical boundary between objects andsubstances. In Yucatec Mayan, for example, allnouns act almost as if they refer to substances. Allnouns require a unitizer when counting (usuallyspecifying shape or form, for example 'one longthin unit'), and don't necessarily need to take distinct plural and singular forms (Lucy and Gaskins,2001). This means that 'two candles' in English ismore like 'two long thin units of wax' in Yucatec.Does talking about objects as if they were substances in their language lead Yucatec Mayans toattend more to the materials and substances thatcomprise the objects? Several studies suggest thatthis is indeed the case (e.g. Lucy and Gaskins,2001). English speakers and Yucatec Mayans wereshown an example object (e.g. a plastic comb with ahandle) and asked to choose which of two otherobjects was more similar to this example. The twochoices varied from the example either in shape(a plastic comb with no handle), or in material (awooden comb with a handle). English speakerspreferred the shape match, saying that the twocombs with a handle were more similar (eventhough they were made of different materials).Yucatec Mayans, on the other hand, preferred thematerial match, saying that the two plastic combs
920Linxuistic Relativitywere more similar (even though they differed inshape). These findings suggest that aspects ofgrammar can in fact shape the way speakers of alanguage conceptualize the shapes and materials ofobjects.OBJECTSFinally, languages also differ in how names ofobjects are grouped into grammatical categories.One such common feature of languages is grammatical gender. Unlike English, many languageshave a grammatical gender system whereby allnouns (e.g. penguins, pockets, and toasters) areassigned a gender. Many languages only have masculine and feminine genders, but some also assignneuter, vegetative, and other more obscure genders.When speaking a language with grammaticalgender, speakers are required to mark objects asgendered through definite articles and genderedpronouns, and often need to modify adjectives oreven verbs to agree in gender with the nouns. Doestalking about inanimate objects as if they weremasculine or feminine actually lead people tothink of inanimate objects as having a gender?A recent set of studies suggests that the grammatical genders assigned to objects by a languagedo indeed influence people's mental representations of objects (Boroditsky et al., in press). Forexample, Spanish and German speakers wereasked to rate similarities between pictures ofpeople (males or females) and pictures of objects(the names of which had opposite genders in Spanish and German). Both groups rated grammaticallyfeminine objects to be more similar to females andgrammatically masculine objects more similar tomales. This was true even though all objects hadopposite genders in Spanish and German, the testwas completely nonlinguistic (conducted entirelyin pictures with instructions given in English), andeven when subjects performed the task during averbal suppression manipulation (which wouldinterfere with their ability to subvocally name theobjects in any language). Other studies demonstrated that Spanish and German speakers alsoascribe more feminine or more masculine properties to objects depending on their grammaticalgender. For example, asked to describe a 'key' (aword masculine in German and feminine in Spanish), German speakers were more likely to usewords like 'hard, heavy, jagged, metal, serrated,and useful', while Spanish speakers were morelikely to say 'golden, intricate, little, lovely, shiny,and tiny'. To describe a 'bridge', on the other hand,(a word feminine in German and masculine inSpanish), German speakers said 'beautiful, elegant,fragile, peaceful, pretty, and slender', while Spanish speakers said 'big, dangerous, long, strong,sturdy, and towering'. These findings once againindicate that people's thinking about objects is influenced by the grammatical genders their nativelanguage assigns to the objects' names. It appearsthat even a small fluke of grammar (the seeminglyarbitrary assignment of a noun to be masculine orfeminine) can have an effect on how people thinkabout things in the world.CONCLUSIONLanguages appear to influence many aspects ofhuman cognition: evidence regarding space, time,objects, and substances has been reviewed in thisarticle, but further studies have also found effectsof language on people's understanding of numbers,colors, shapes, events, and other minds. Considering the many ways in which languages differ, thefindings reviewed here suggest that the privatemental lives of people who speak different languages may differ much more than previouslythought.Beyond showing that speakers of different languages think differently, these results suggest thatlinguistic processes are pervasive in most fundamental domains of thought. That is, it appears thatwhat we normally call 'thinking' is in fact a complexset of collaborations between linguistic and nonlinguistic representations and processes. Further research into linguistic relativity may help uncoverthe exact nature of the interactions between thesemany processes in the service of complex cognitivefunction, as well as help us to estab
Encyclopedia of cognitive science 1. Cognitive science - Encyclopedias I. Nadel, Lynn 1 53' .03 Library of Congress Cataloging in Publication Data A catalog for this record is available from the Library of Congress Typeset by Kolam Information Services Pvt. Ltd., Pondicherry, India Printed and bound by The Bath Press, England