LanguageLearning ISSN0023-8333 Who’s Afraid of the Big Bad Whorf? Crosslinguistic Differences in Temporal Language and Thought DanielCasasanto StanfordUniversity Theideathatlanguageshapesthewaywethink,oftenassociatedwithBenjaminWhorf, has long been decried as not only wrong but also fundamentally wrong-headed. Yet, experimentalevidencehasreopeneddebateabouttheextenttowhichlanguageinflu- encesnonlinguisticcognition,particularlyinthedomainoftime.Inthisarticle,Iwill firstanalyzeaninfluentialargumentagainsttheWhorfianhypothesisandshowthatits anti-Whorfianconclusionisinpartanartifactofconflatingtwodistinctquestions:Dowe thinkinlanguage?andDoeslanguageshapethought?Next,Iwilldiscusscrosslinguis- ticdifferencesinspatialmetaphorsfortimeanddescribeexperimentsthatdemonstrate correspondingdifferencesinnonlinguisticmentalrepresentations.Finally,Iwillsketch asimplelearningmechanismbywhichsomelinguisticrelativityeffectsappeartoarise. Althoughpeoplemaynotthinkinlanguage,speakersofdifferentlanguagesdevelopdis- tinctiveconceptualrepertoiresasaconsequenceofordinaryandpresumablyuniversal neuralandcognitiveprocesses. Areourownconceptsof‘time,’‘space,’and‘matter’giveninsubstantially thesameformbyexperiencetoallmen,oraretheyinpartconditionedby thestructureofparticularlanguages?(Whorf,1939/2000,p.138) Doeslanguageshapethought?Docrosslinguisticdifferencesinthelexiconor grammar have nonlinguistic consequences, such that people who talk differ- entlyendupthinkingdifferentlyincorrespondingways?Ifso,howdoesthis happen:WhatfeaturesoflanguageaffectwhichcognitiveprocessesandWhat arethemechanisms?SinceBenjaminWhorfdirectedscholars’attentiontothis ThisresearchwassupportedinpartbyanNSFGraduateResearchFellowshipandNSFDis- sertationFellowship,byNRSApost-doctoralfellowship#F32MH072502,andbyagrantfromthe SpanishMinistryofEducationandScience#SEJ2006-04732/PSIC,DGI. Correspondence concerning this article should be addressed to Daniel Casasanto, Stanford University,JordanHall,Bldg.420,Stanford,CA94305.Internet:[email protected] LanguageLearning58:Suppl.1,December2008,pp.63–79 63 (cid:2)C 2008LanguageLearningResearchClub,UniversityofMichigan Casasanto Who’sAfraidoftheBigBadWhorf? setofquestionsoverhalfacenturyago,fewtopicshaveincitedsomuchpas- sionatedisagreementamonglinguists,anthropologists,andpsychologistsalike. Inthisarticle,IwillnotreviewthesanguinaryhistoryoftheWhorfiandebate (seeGentner&Goldin-Meadow,2003;Gumperz&Levinson,1996).Rather, Iwillstartbydisentanglingtwoquestionsabouttherelationshipbetweenlan- guage and thought, the conflation of which may contribute to the fervor that theWhorfianhypothesisinspiresinitsopponents.1 Next,Iwillreviewexper- imental evidence that crosslinguistic differences in the lexicalization of time correlatewithprofounddifferencesinthewayspeakersofdifferentlanguages mentallyrepresentduration.Finally,Iwilldescribeamechanismbywhichlan- guagecausesthesedifferencestoarise,whichsuggestsareframingofWhorf’s questionaboutrelationshipsamonglanguage,concepts,andexperience. Thelong-standingmajorityviewontheWhorfianhypothesisissummarized inStevenPinker’s1994bestsellerTheLanguageInstinct.Pinkerfirstposeda questioninspiredbyGeorgeOrwell’s(1949)dystopianfantasy,thenovel1984, inwhichthegovernmentseekstorendersubversivethoughtsunimaginableby makingthemnamelessintheprescribedlanguage,Newspeak: Isthoughtdependentonlanguage?DopeopleliterallythinkinEnglish, Cherokee,Kivunjo,orby[theyear]2050,Newspeak?Orarethoughts couchedinsomesilentmediumofthebrain—alanguageofthoughtor “mentalese”—andmerelyclothedinwordswheneverweneedto communicatethemtoalistener?(p.56) Inresponse,Pinkerwrote: [T]hefamousSapir-Whorfhypothesisoflinguisticdeterminism,stating thatpeople’sthoughtsaredeterminedbythecategoriesmadeavailableby theirlanguage,anditsweakerversion,linguisticrelativity,thatdifferences amonglanguagescausedifferencesinthethoughtsoftheirspeakers[...] iswrong,allwrong.(p.57) Pinker denounced not only radical linguistic determinism but also linguistic relativity,whichismyconcernhere,andWhorf’sconcerninthe1939passage quotedabove.Tojustifyhisanti-Whorfianposition,Pinkerasserted: Theideathatthoughtisthesameaslanguageisanexampleofwhatcan becalledaconventionalabsurdity[.](p.57) Here, Pinker illustrates a confusion that is rampant in the literature on rela- tionships between language and thought. The idea that “thought is the same aslanguage”(asOrwellsuggested)shouldnotbeconflatedwiththeideathat LanguageLearning58:Suppl.1,December2008,pp.63–79 64 Casasanto Who’sAfraidoftheBigBadWhorf? “differencesamonglanguagescausedifferencesinthethoughtsoftheirspeak- ers”(asWhorfsuggested).OrwellandWhorfraisedtwodistinctquestions:Do wethinkinlanguage?andDoeslanguageshapethought?Itispossible(and,I willargue,true)thatlanguagecanshapethewaypeoplethinkeveniftheydo notthinkinlanguage. ConfusionoftheWhorfianquestionwithwhatIwillcallthe“Orwellian” questionpervadesPinker’sdiscussion: Theideathatlanguageshapesthinkingseemedplausiblewhenscientists wereinthedarkabouthowthinkingworks,orevenhowtostudyit.Now thatscientistsknowhowtothinkaboutthinking,thereislessofa temptationtoequate[thinking]withlanguage.(pp.58–59,italicsadded) NotonlydoesPinkerzigzagbetweenOrwell’sideaandWhorf’s,healsoimplies arelationshipbetweenthemthatismisleadingand,infact,logicallyfallacious. ThereaderisledtobelievethattheabsurdityoftheOrwellianproposalshould betakenasoneofthestrongestargumentsagainsttheWhorfianhypothesis.To unpackthisfallacy,letusturntheOrwellianandWhorfianquestionsintotwo statements,OandW[(1a)and(b)]: (1a) O:Wethinkinlanguage. (1b) W:Languageshapesthought. Thereisaclearrelationshipbetweenthesestatements.Ifpeoplethinkinlan- guage,thenitmustbethecasethatlanguageshapesthought.Inotherwords, if Orwell was right, then Whorf must necessarily be right, too. This can be expressedintheproposition(2):ifO,thenW. (2) O→W Ifweagreetothetruthofthisproposition,thenwemustalsoagreetothetruth of its contrapositive (3): If language does not shape thought, then we do not thinkinlanguage.Inotherwords,provingthatWhorfwaswrongwouldentail thatOrwellwaswrong,aswell. (3) ∼W→∼O However,thisisnotwhatPinkerargued.Rather,hesuggestedthatpeopledonot thinkinlanguage;therefore,languagedoesnotshapethought.Inotherwords, because Orwell was wrong, we should believe that Whorf must be wrong. However,thisargumentassumesthattheinverse(4)ofourpropositionistrue, alogicalfallacyknownas“denyingtheantecedent.” 65 LanguageLearning58:Suppl.1,December2008,pp.63–79 Casasanto Who’sAfraidoftheBigBadWhorf? (4) ∼O→∼W It may sound reasonable at first to say that “if people think in language, then language shapes thought, but because people do not think in language, then language does not shape thought”—provided you do not think about it too carefully.Infact,thisisequivalenttosaying,“ifJohnhasadaughterwhoplays theviolin,thenhemustbeahuman,butbecauseJohndoesnothaveadaughter whoplaystheviolin,thenheisnothuman.” AlthoughtheOrwellianquestionandtheWhorfianquestionareimportantly related,theyarenotrelatedinthewaythatPinkersuggested(andheisnotalone in making this logical error). Evidence in favor of Orwell’s idea would also support Whorf’s, but evidence against the idea that people think in language doesnotcountagainstthepossibilitythatlanguageshapesthought.Clarifying thisrelationshipisimportant:Ifpeoplebelievethatlinguisticrelativityentails (orimplies)theOrwelliannotionthatpeoplethinkinlanguage,thennowonder theyareafraidofthebigbadWhorf! OrwellandWhorf:DivideandConquer? Having distinguished the Orwellian and Whorfian questions, it is possible to evaluatethemseparately.Pinker’s1994discussion,whichhelaterdubbedhis “obituary” for the Whorfian hypothesis (Pinker, 2007, p. 124), reviews stan- dardargumentsagainstboththeideathatwethinkinlanguageandtheideathat language shapes thought. Although arguments against the one idea are often misrepresented as arguments against the other, many of these objections are compellingwhenproperlyconstrued.Pinkertakesashotgunapproachtodis- creditingtheOrwellianequationoflanguagewiththought,startingwithappeals tointuition:Wehaveallhadthefeelingthatweknowwhatwewanttosaybut wedonotknowhowtosayit,thereforetheremustbesomedifferencebetween whatwesayandwhatwewanttosay.Otherargumentsstresstheinadequacy of language as a medium for thought, pointing to problems such as ambigu- ity,deixis,andcoreference,whichseemtorequireextralinguisticresourcesto resolve. For example, when we read the ambiguous headline “Hershey Bars Protest”(Pinker,1994,p.119),theinformationweneedtodecidewhetherthis storydiscussesanoppressivechocolatemanufacturerorsomerebelliouscandy bars does not appear to reside in language, per se. Still other arguments call uponresultsfrompsychologicalexperiments,suchasRogerShepardandLynn Cooper’s(1982)studiessuggestingthatpeoplerotateobjectsintheirmind’seye usingimagistic,picturelikerepresentations.Mostcompellingly,Pinkerpointed LanguageLearning58:Suppl.1,December2008,pp.63–79 66 Casasanto Who’sAfraidoftheBigBadWhorf? out that much of what can reasonably be construed as “thinking” happens in themindsofbabies,deafisolates,andaphasics,wholackthefulluseoflan- guage, and even in the minds of animals that lack language altogether. The fact that animals can learn, remember, navigate, communicate, reason about causes, infer goals, build shelters, use tools, cooperate, deceive one another, andconstructsocialhierarchiessuggeststhatquitealotofthinkingcanhappen withouthumanlanguage. These anti-Orwellian arguments do not necessarily support Pinker’s con- clusionthatallthinkingoccursinaFodorian“mentalese”(Fodor,1975).They do not rule out the possibility that certain kinds of thinking are mediated by language (see Carruthers, 1996; Spelke & Tsivkin, 2001). They do suggest, however,thatevenifsomethinkingtakesplaceinthemediumofnaturallan- guage,notallthinkingdoes,andthattheOrwellianproposalisproblematic,in principle. Standard arguments against the Whorfian hypothesis, including Pinker’s, are of a different sort. Rather than attacking linguistic relativity in principle, they criticize the data and methods that have been used to support Whorfian claims.DotheEskimosreallythinkaboutsnowmorepreciselythanspeakers ofaStandardAverageEuropean(SAE)languagebecausetheycancategorize itmorepreciselyinlanguage(Whorf,1940/2000)?AreHopispeakersunable to conceptualize time as we do because of their impoverished temporal vo- cabulary(Whorf,1939/2000)?Thereisnogoodevidencethatthisisthecase. TwothingswerewrongwithsomeofWhorf’smostnotoriousclaims.Tostart, many of Whorf’s linguistic observations did not stand up to scrutiny by later scholars.Arguably,speakersofEskimolanguagesdonotuseanymorewords for snow than your average snowboarding enthusiast (Pullum, 1991), and the Hopispeakers’wayoftalkingabouttimemaynotbeasdifferentfromtheSAE speakers’asWhorfmadeitseem(Malotki,1983).Iftherearenotactuallyany relevantdifferencesbetweenlanguages,thenthereisnoreasontopositdiffer- ences in the thoughts of their speakers. Furthermore, the project of inferring cognitive differences solely from linguistic differences is hopelessly circular. Patterns in language can serve as a source of hypotheses about cognitive dif- ferences between members of different language communities, but some sort ofextralinguisticdataareneededtotestthesehypotheses:Otherwise,theonly evidencethatpeoplewhotalkdifferentlyalsothinkdifferentlyisthattheytalk differently! ExperimentalworksinceWhorf’stimehassufferedseveraladditionalprob- lems.Pinkernotedthatsomeapparentbehavioraldifferencesbetweenlanguage groupshaveturnedouttobeartifactsofclumsytranslation.Famously,Alfred 67 LanguageLearning58:Suppl.1,December2008,pp.63–79 Casasanto Who’sAfraidoftheBigBadWhorf? Bloom (1981) alleged that Chinese speakers are less capable of reasoning counterfactually than English speakers because Chinese lacks subjunctives, whichserveascounterfactualmarkersinEnglish(e.g.,Ifitweretorain,then Iwould takeanumbrella).Inhisoriginalexperiment,BloomcreatedEnglish and Chinese versions of a story that required counterfactual reasoning, and he found that Chinese speakers in China failed abysmally to understand the story’scounterfactualstructure,whereasEnglishspeakersintheUnitedStates hadnotroubleatall.TerryAu,anativeChinese-speakingpsychologist,exam- inedthestoriesthatBloomhadusedandfoundtheChineseversiontobe“not veryidiomatic”(Au,1983,p.161).WhenAureplicatedBloom’sexperiment using new stimuli, the crosslinguistic differences in counterfactual reasoning disappeared(Au). Yet,awkwardtranslationsmaybetheleastoftheproblemsexemplifiedby Bloom’sstudy(andfoundinmanysubsequentstudies,aswell).Nomatterhow carefullystimulusmaterialsaretranslated,anyexperimentaldesignthatrelies oncomparingperformanceacrosstranslationsconfoundsdifferencesbetween itemswithdifferencesbetweenconditions(andifmonolingualsareused,also confounds item differences with group differences), making the results hard tointerpret.Furthermore,thisstudypredictedasingledifferencebetweenlan- guage groups: One group should perform better than the other. Even if the predicted difference had been reliable (e.g., if English speakers had shown bettercounterfactualreasoningthanChinesespeakersacrossstudies),itwould behardtoknowwhy:thegroupthatisbetteratcounterfactualreasoningcould alsobebetteratmanyotherkindsofreasoningthatdonotcorrespondtoany crosslinguistic differences. Unless predictions are cast in terms of statistical interactions, researchers risk interpreting spurious relationships between pat- ternsinlanguageandpatternsinperformanceonexperimentaltasks.Finally, the Whorfian hypothesis posits a causal relationship between language and thought.Bloomsoughttotestthehypothesisthatlanguagecausesdifferences incounterfactualreasoningusinganexperimentaldesignthatwasonlycapable ofdemonstratingcorrelation—aproblemthatisnotintractable,butwhichstill plagueswould-beWhorfianresearchaquarterofacenturylater(seeCasasanto, 2005a;Gordon,2004). The Orwellian idea that people think (mostly or entirely) in the medium ofnaturallanguage,andthereforethatlanguagecanbeequatedwiththought, is unsupported empirically and is also problematic in principle, given what is known about language and about thought.2 By contrast, the Whorfian idea that linguistic differences can cause speakers of different languages to think differently faces no such in-principle challenges. When standard arguments LanguageLearning58:Suppl.1,December2008,pp.63–79 68 Casasanto Who’sAfraidoftheBigBadWhorf? are properly interpreted, they do not sound a death knell for the Whorfian hypothesis,butratheracallformorerigorousresearch. TimeinLanguageandThought Since Pinker’s (1994) “obituary,” Whorfian research has experienced a re- naissance. Experimental evidence has reopened debate about the extent to which language influences nonlinguistic cognition in domains such as space (Levinson, 1996; Li & Gleitman, 2002; Majid, Bowerman, Kita, Haun, & Levinson,2004),color(Gilbert,Regier,Kay,&Ivry,2006;Kay&Kempton, 1984; Robertson, Davies, & Davidoff, 2000; Witthoft, et al., 2003), number (Casasanto, 2005a; Gordon, 2004; Gelman & Gallistel, 2004; Miller, Major, Shu,&Zhang,2000;Pica,Lemer,Izard,&Dehaene,2004;Spelke&Tsivkin, 2001),andtime(Boroditsky,2001;Casasantoetal.,2004;Chen,2007;January &Kako,2007;Nu´n˜ez&Sweetser,2006).Oneobstacletoresolvingthiscon- troversyhasbeendevisingtrulynonlinguisticteststoevaluatehowspeakersof differentlanguagesperceiveorremembertheirexperiences,particularlyinthe moreabstractconceptualdomainssuchastime. Across languages, people use the same words to talk about time that they use to talk about space (Alverson, 1994; Clark, 1973; Gruber, 1965; Haspelmath,1997;Jackendoff,1983;Lakoff&Johnson,1980;Traugott,1978). Forexample,Englishspeakersmighttalkaboutalongvacationoralongline andmovingameetingforward ormovingatruckforward.Evidencefrombe- havioralexperimentssuggeststhatpeoplenotonlytalkabouttimeusingspatial language, they also think about time using mental representations of space (Boroditsky, 2000, 2001; Boroditsky & Ramscar, 2002; Casasanto, 2005b, in press; Casasanto & Boroditsky, 2003, 2008; Casasanto et al., 2004; Cohen, 1967; Gentner, 2001; Nu´n˜ez & Sweetser, 2006; Piaget, 1927/1969; Torralbo, Santiago,&Lupia´n˜ez,2006;Tversky,Kugelmass,&Winter,1991).Although using spatial metaphors for time may be universal (Alverson, 1994; cf. Silva Sinha,Sinha,Zinken,&Sampaio,2008),theparticularmappingsfromspace totimevaryacrosslanguages.Forinstance,dependingonthelanguage,speak- ers might talk about the future as if it lies ahead of us (in English), behind us (in Aymara), or below us (in Mandarin Chinese). Behavioral studies sug- gest that speakers of languages that use different spatiotemporal metaphors mayindeedthinkabouttimedifferently(Boroditsky,2001;Nu´n˜ez&Sweetser, 2006). There is, however, an important limitation shared by these crosslinguistic studiescomparingEnglishspeakers’mentalrepresentationsoftimewiththose 69 LanguageLearning58:Suppl.1,December2008,pp.63–79 Casasanto Who’sAfraidoftheBigBadWhorf? ofAymaraandMandarinspeakers:Subjectsweretestedontasksthatrequired them to produce or understand language. Nu´n˜ez and Sweetser’s participants were producing co-speech gestures, and Boroditsky’s were judging sentences containingspatialortemporallanguage.Perhapsthesestudiesshowedrelation- shipsbetweenspatialandtemporalthinkingthatwereconsistentwithlinguistic metaphors only because participants were required to process space or time in language (i.e., because they were “thinking for speaking”; Slobin, 1996). Wouldthesamerelationshipsbetweenmentalrepresentationsofspaceandtime befoundifparticipantsweretestedonnonlinguistictasks? To address this question, my collaborators and I developed simple psy- chophysicaltaskstoinvestigatewhethermetaphorsinlanguageinfluenceeven our basic, perceptuomotor representations of time. Do speakers of languages thatusedifferentspatiotemporalmetaphorsthinkabouttimedifferently—even when they are not using language? First, we analyzed previously unexplored crosslinguistic differences in metaphors for duration in English and Greek. Next,wetestedwhethertheselinguisticdifferencescorrelatewithdifferences in English and Greek speakers’ performance on low-level, nonlinguistic du- ration estimation tasks. Finally, we conducted a training study to evaluate a causal role for language in shaping time representations (Casasanto, 2005b; Casasantoetal.,2004). DurationinOneorThreeDimensions Literature on how time can be expressed verbally in terms of space (and, by hypothesis,conceptualizedintermsofspace)hasfocusedprincipallyonlinear spatialmetaphors.However,istimenecessarilyverbalized(andconceptualized) in terms of unidimensional space? Some theorists have suggested so (Clark, 1973;Gentner,2001),andalthoughthismaybetrueregardingtemporalsuc- cession,linguisticmetaphorssuggestanalternativespatializationforduration. Englishspeakersnotonlydescribetimeasaline,buttheyalsotalkabouthav- ingoceansoftime,savingtimeinabottle,andlikenthedaysoftheirlivesto sandsthroughthehourglass.Quantitiesoftimearedescribedasamountsofa substancethatoccupiesthree-dimensionalspace(i.e.,volume). Languagesdifferintheextenttowhichtheydescribedurationintermsof distance as opposed to amount of substance. In English, it is natural to talk about a long time or a long meeting, borrowing the structure and vocabulary of a linear spatial expression like a long rope. In Greek, the words makris and kontos are the literal equivalents of the English spatial terms long and short. They can be used in spatial contexts much the way long and short are used in English (e.g., ena makry skoini means “a long rope”). In temporal LanguageLearning58:Suppl.1,December2008,pp.63–79 70 Casasanto Who’sAfraidoftheBigBadWhorf? contexts,however,makrisandkontosaredispreferredinmanyinstanceswhere long and short would be used naturally in English. It would be unnatural to translate“alongmeeting”literallyas“miamakriasynantisi.”Ratherthanusing distance terms, Greek speakers typically indicate that an event lasted a long time using megalos, which in spatial contexts means physically large (e.g., a bigbuilding),orusingpoli,whichinspatialcontextsmeansmuch(e.g.,much water). Toquantifytherelativeprevalenceof“distance”and“amount”metaphors for duration across languages, the frequencies of the English phrases long timeandmuchtimewerecomparedwiththeGreekphrasesthatnativespeak- ersreportedmostnaturallyexpressedthesameideas,makrykronikodiastima “longtimedistance”andpoliora“muchtime.”Eachexpressionwasentered as a search term in a very large multilingual text corpus (www.google.com). Search results showed that the distance metaphor was dramatically more fre- quentthantheamountmetaphorinEnglish,whereastheoppositepatternwas found in Greek. Results were corroborated in a questionnaire study showing that English speakers tend to use distance metaphors to describe the dura- tion of events (e.g., long meeting, long party), whereas Greek speakers tend to use amount metaphors to describe the durations of the same events (e.g., synantisi pou diekese poli “meeting that lasts much,” parti pou kratise poli “partythatlastsmuch”).Althoughbothlanguagesusesomedistanceandsome amountmetaphorsforduration,therelativestrengthsofthesemetaphorsvaries across languages. These preliminary studies by no means captured all of the complexities of how duration is metaphorized in terms of space within or betweenlanguages,butfindingscorroboratednativeEnglishandGreekspeak- ers’intuitionsandprovidedquantitativelinguisticmeasuresonwhichtobase predictionsaboutbehaviorinnonlinguistictasks. Do these differences in metaphor frequency between languages lead En- glish and Greek speakers to think about time differently, even when they are notusinglanguage?Tofindout,weaskedEnglishandGreekspeakerstoesti- matethedurationofbriefeventsthatcontainedeitherdistractinginformation about linear distance (distance interference) or distracting information about amount (amount interference). In the distance interference condition, partici- pantsviewedaseriesoflines“growing”acrossthescreen,forvariousdistances and durations. After each line disappeared, they were asked to reproduce its duration by clicking the mouse twice (in the same spot) to indicate the time thatelapsedfromtheinstantthatthelineappearedtotheinstantthatitreached itsmaximumlength.Thedistancethatthelinegrewwasirrelevanttothetask and was varied orthogonally to the line’s duration. As such, distance served 71 LanguageLearning58:Suppl.1,December2008,pp.63–79 Casasanto Who’sAfraidoftheBigBadWhorf? as distractor: a piece of information that was unrelated to the task but could potentially interfere with task performance. In the amount interference con- dition, subjects viewed a schematic drawing of a container filling gradually with liquid and were asked to reproduce the duration of the “filling” event. Analogously to the distance interference condition, the amount of fill varied orthogonallywiththedurationoftheeventand,assuch,servedasadistractor forthesubjects’taskofestimatingduration. InpreviousstudiesusingasimilardistanceinterferencetaskwithEnglish speakers(Casasanto&Boroditsky,2003,2008),English-speakingparticipants wereunabletoignorethedistancethatalinegrewwhenestimatingitsduration, eventhoughdistancewasirrelevanttothetimeestimationtask.Theresultwas apatternofcross-dimensionalinterference:Spatialinformationinterferedwith participants’ temporal judgments in a particular way. Although, on average, all stimuli lasted the same duration, participants judged lines that traveled a shorterdistancetolastashortertimeandlinesthattraveledalongerdistance tolastalongertime. Is this conflation of distance and duration universal to humans, or does it dependinpartontheconflationofdistanceanddurationinlanguage?Ifpat- ternsinlanguagearepartlyresponsibleforthespace-timeconfusionobserved in English speakers, then irrelevant distance and amount information should interfere with English and Greek participants’ duration estimates differently. Englishspeakersshouldshowmoreinterferencefromdistancethanamounton their time estimates. Greek speakers should show the opposite pattern, being more distracted by amount than by distance interference. Results supported these predictions: The pattern of cross-dimensional interference observed in EnglishandGreekspeakersonthispairofnonlinguistictimeestimationtasks followedthepatternofspatiotemporalmetaphorsfoundinEnglishandGreek. Englishspeakerswerestronglyaffectedbythedistancethatalinetraveledbut only weakly affected by the fullness of a container, whereas Greek speakers showed the opposite pattern of cross-dimensional interference. The structure ofpeople’slow-level,nonlinguistictimerepresentationsisnotuniversal:These simplepsychophysicaltasksindicatethatatabasiclevel,thewaywementally representtimecovarieswiththewaywetalkaboutitinournativelanguages. However,doesexperienceusingdifferentmetaphorscausespeakersofdifferent languagestothinkdifferently? HowLanguageShapesTime Previoussuggestionsforhowlanguagecouldinfluencerepresentationsoftime have failed because they have attempted to give an Orwellian answer to a LanguageLearning58:Suppl.1,December2008,pp.63–79 72