The role of polysemy in masked semantic and translation priming

JournalofMemoryandLanguage51(2004)

1–22

JournalofMemoryandLanguage

/locate/jml

Theroleofpolysemyinmaskedsemantic

andtranslationpriming

MatthewFinkbeiner,a,*KennethForster,bJanetNicol,b

andKumikoNakamurab

ab

DepartmentofPsychology,HarvardUniversity,Cambridge,MA02138,USADepartmentofPsychology,UniversityofArizona,Tucson,AZ85721,USA

Received1December2003;revisionreceived5January2004

Availableonline5March2004

Abstract

Awell-knownasymmetryexistsinthebilingualmaskedprimingliteratureinwhichlexicaldecisionisused:namely,maskedprimesinthedominantlanguage(L1)facilitatedecisiontimesontargetsinthelessdominantlanguage(L2),butnotviceversa.Insemanticcategorization,ontheotherhand,primingissymmetrical.InExperiments1–3wecon rmthistaskdi erence, ndingrobustmaskedL2–rmulatinganaccountforthese ndings,webeginwiththeassumptionofarepresentationalasymmetrybetweenL1andL2lexicalsemanticrepresentations,suchthatL1representationsarerichlypopulatedandL2representationsarenot.Accordingtothisrepresentationalaccount,L2–L1primingdoesnotoccurinlexicaldecisionbecauseaninsu cientproportionoftheL1lexicalsemanticrepresentationisactivatedbytheL2prime.Insemanticcategorization,wearguethatthesemanticinformationrecruitedtogenerateadecisionisrestrictedbythetaskcategory,andthatthisrestrictionenhancesthee ectivenessoftheL2prime.InExperiments4–6,theseassumptionsweretestedinawithin-languagesettingbypairingmany-sensewords(e.g.,‘‘head’’)withfew-sensewords(e.g.,‘‘skull’’).Inlexicaldecision,robustprimingwasobtainedinthemany-to-fewdirection(analogoustoL1–L2),but,noprimingwasobtainedinthefew-to-manydirection(analogoustoL2–L1)usingthesamewordpairs.Priminginsemanticcategorization,ontheotherhand,wasobtainedinbothdirections.WeproposetheSenseModelasapossibleaccountofthese ndings.Ó2004ElsevierInc.Allrightsreserved.

Keywords:Maskedtranslationpriming;Semanticpriming;Bilinguallexicalprocessing;Bilinguallexicon;Lexicaldecision;Semanticcategorization

Themaskedtranslationprimingparadigmhaspro-ventobeinstrumentalinthedevelopmentofpsycho-linguisticaccountsofbilinguallexicalrepresentationandprocessing.Oneimportant ndingthatthispara-digmhasrevealed,andonethatanysuccessfulmodelofbilinguallexicalprocessingmustbeabletoexplain,isthetranslationprimingasymmetryinlexicaldecision

Correspondingauthor.Fax:1-617-496-6262.

E-mailaddress:msf@wjh.harvard.edu(M.Finkbeiner).

*

(Isitaword?).Thiswell-establishedasymmetry,whichhasbeenreportedbyseveralresearchersworkingwithdi erentbilingualpopulations,ischaracterizedbythefollowingpatternof ndings:masked(subliminal)primesinthedominantlanguage(L1)facilitaterecog-nitionoftranslation-equivalentwordsinthenon-dominantlanguage(L2)(deGroot&Nas,1991;Gollan,Forster,&Frost,1997;Jiang,1999;Jiang&Forster,2001;Keatly,Spinks,&DeGelder,1994;Wil-liams,1994);however,maskedL2primesdonotreli-ablyfacilitaterecognitionofL1translation-equivalent

0749-596X/$-seefrontmatterÓ2004ElsevierInc.Allrightsreserved.

doi:10.1016/j.jml.2004.01.004

2M.Finkbeineretal./JournalofMemoryandLanguage51(2004)1–22

words(Gollanetal.,1997;Grainger&Frenck-Mestre,1998;Jiang,1999;Jiang&Forster,2001;Keatlyetal.,1994;Sanchez-Casas,Davis,&Garcia-Albea,1992).1Aswewillseebelow,mostaccountsofthetranslationprimingasymmetryhaveappealedtoa‘‘limitingfactor’’explanation,wherebyapropertyspeci ctotheL2lexi-conisthoughttolimitprimingintheL2–L1direction.OneobviouspossibilityisthatwordsinL2arenotprocessede ectivelywhentheyaremasked;however,thesamesubjectsexhibitrobustwithin-L2maskedrepetitionpriminge ects(Finkbeiner,inpress;Gollanetal.,1997;Jiang,1999;Jiang&Forster,2001).Thatis,bilingualsarefastertorespondtoatargetintheirL2(e.g.,‘‘HOUSE’’)whenitisprecededbyamaskedpresentationofthesamewordbutindi erentcase(e.g.,‘‘house’’)thantheyarewhenthetargetispre-cededbyacontrolprime(e.g.,‘‘truck’’).This ndingsuggeststhatalackofL2–L1translationprimingcannotbeduetoaninabilitytoe ectivelyprocessthemaskedprimes.

TheRevisedHierarchicalModel(Kroll&Stewart,1994;Kroll&Tokowicz,2001),adominantmodelinthe eld,accountsforthetranslationprimingasymmetrybysuggestingthat:(1)thelocusofthetranslationpriminge ectisatthelevelofmeaningandthat(2)relativetoL1representations,L2lexicalformsareonlyweaklycon-nectedtomeaning-levelrepresentations.Accordingtothisaccount,primingise ectiveintheL1–L2directionbecausethemaskedL1primeservestoactivateasharedconceptualnode,whichthenpreactivatestheL2trans-lation-equivalentlexicalform.However,primingisnote ectiveintheL2–L1directionbecauseL2primesdonotautomaticallyactivatetheirconceptualrepresenta-tions,resultinginnopreactivationoftheL1translation-equivalentformandthusnopriming.Accordingtothisaccount,theL2form–meaningconnectionstrengthisthelimitingfactorthatpreventsprimingintheL2–L1direction.

Thisaccountisappealing,butitsu ersfromamajorweaknessinthatitpredictsnowithin-L2priming.Ifthelocusofprimingisatthelevelofmeaning,andL2formscannotautomaticallyactivatetheirmeanings,thenhowdoesmaskedL2–L2primingoccur?Toexplaintheex-istenceofwithin-L2priming,asecondlocusofprimingwouldhavetobeposited:namely,itwouldhavetobearguedthatwithin-L2primingoccursatthelevelofform.Insupportofthisclaim,itcouldbepointedoutthatwithin-L2priminge ectshaveprimarilybeendemonstratedwithrepetitionpriming,notsemanticpriming(Gollanetal.,1997;Jiang,1999;althoughsee

1

Thetranslationprimingasymmetryiseliminatedwhenprimeandtargetwordsarecognatesfromsame-scriptlan-guages(e.g.,rico-RICH)(Sanchez-Casasetal.,1992).

Frenck-Mestre&Prince,1997),andconsequentlymaybeattributedtoprimingattheformlevelalone.How-ever,translationpriminge ectswithcognatestimuli(e.g.,rico-RICH)cannotbeattributedtoanoverlapinform-levelpropertiesalone,sincetheprimingismea-suredrelativetoabaselineinvolvingasimilardegreeofformoverlap(e.g.,rict-RICH).Thereforeitappearsthatcognateprimingmustbeattributedtoanoverlapofinformationatthelevelofmeaning.

ArelatedproblemfortheRHMisitsunderspeci -cationwithrespecttothenatureoftheL2form–meaningconnection.ThisispresentedmostclearlyintheawkwardproposalthattheL2form–meaningcon-nectioncansimultaneouslysupportrobustprimingintheL1–L2directionandlimitprimingintheL2–L1direction.Toexplainbothofthese ndings,itisas-sumedthatthemappingbetweenL2formandmeaningissu cientlystronginthemeaning-to-formdirection(therebypermittingL1–L2priming),butthatthesamemappingistooweakintheform-to-meaningdirection(therebylimitingL2–L1priming).

Recently,another ndingusingthemaskedtrans-lationprimingparadigmhasbeenreportedwhichpromisestoelucidatetherelationshipbetweenL1andL2lexicalrepresentations.Thisisthepossibilityofataskdi erenceinmaskedL2–L1repetitionpriming.Todate,therehavebeentworeportedinstancesofsuchataskdi erence.The rstwasreportedbyGraingerandFrenck-Mestre(1998),whoobservedmaskedtransla-tionpriminge ectsintheL2–L1directionfornon-cognateswhensemanticcategorizationwasused,but,inlinewithseveralotherresearchers,notwhenlexicaldecisionwasused.This ndingappearstobequiteinconsistentwiththenotionthatL2wordsareinca-pableofactivatingsemanticrepresentationswhenmasked,andhenceitiscriticaltoexaminethisissuefurther.ThisisespeciallytruesincetheGraingerandFrenck-Mestre(1998) ndingstandsinmarkedcon-trasttothatofSanchez-Casasetal.(1992),whoalsousedthesemanticcategorizationtask,butreportednomaskedL2–L1translationprimingunlessthetransla-tionpairswerecognates.

Thesecondexampleofataske ectwasreportedbyJiangandForster(2001),whoobservedmaskedL2–L1translationprimingwhenan‘‘old–new’’episodicrec-ognitiontaskwasused,but,again,notwhenlexicaldecisionwasused.Theirexplanationofthise ect,whichwerefertoasthe‘‘separatememorysystemsaccount,’’o ersaverydi erentinterpretationofbilinguallexicalprocessing.Intheirstudy,participantswereaskedtomemorizealistofL1wordsinthe rstphaseoftheexperiment.Inthesecondphase,participantsperformedaspeeded‘‘old–new’’task,inwhichL1wordshadtobeclassi edaccordingtowhethertheywereontheoriginallistornot.Unknowntotheparticipants,theL1targetwordwasprecededbyamaskedtranslationprimeinL2.

M.Finkbeineretal./JournalofMemoryandLanguage51(2004)1–223

Usingthisprocedure,JiangandForster(2001)obtainedsigni cantmaskedtranslationpriminge ectsintheL2–L1directionfor‘‘old,’’butnot‘‘new’’ingthesamematerialsandtwo-phasedesign,noprimingwasobtainedwhenthetaskwaschangedtolexicaldecision.JiangandForsterexplainedtheirresultsbyproposingthatL1andL2lexiconsarestoredinseparatememorysystems:theL1lexiconisstoredinlexicalmemory,whiletheL2lexiconisstoredinepisodicmemoryasasetofassociationsbetweenL2wordsandtheirL1counter-parts(representedepisodically).Ifresponsesarecon-trolledbyactivationpatternsinepisodicmemory(astheywouldbeinan‘‘old–new’’task),thenL2–L1primingcaneasilybeexplained.Incontrast,whenre-sponsesarecontrolledbyactivationpatternsinlexicalmemory(astheywouldbeinalexicaldecisionorse-manticcategorizationtask),theL2primeisthoughttobeunabletoactivatethelexicalrepresentationofitsL1counterpart.Thisisduetoacriticalassumptionoftheseparatememorysystemsaccount,whichholdsthatactivationacrossmemorysystemsrequiresawareness.Althoughthisaccountcanexplainthetaskdi erencebetweenlexicaldecisionandepisodicrecognition,itcannotstraightforwardlyexplaintheresultsobtainedwhensemanticcategorizationisused.Sincesemanticcategorizationclearlyrequiresthatthelexicalrepresen-tationoftheL1targetbeactivated,itshouldyieldthesamee ectsasthelexicaldecisiontask,i.e.,noL2–L1priming.Hence,ataskdi erencewheremaskedtrans-lationprimingisobservedinsemanticcategorizationbutnotlexicaldecisionwouldbeachallengetoboththeconnection-strengthexplanationprovidedbytheRHMaswellastheexplanationlaidoutintheseparatememorysystemsaccount.

Thepurposeofthepresentstudyis:(1)tocon rmthetaskdi erenceoriginallyreportedbyGraingerandFrenck-Mestre(1998)betweensemanticcategorizationandlexicaldecisioninmaskedL2–L1translationprimingand(2)toproposeandtestatheoreticalac-countofmaskedtranslationprimingthatatonceisabletoaccommodatethetaskdi erence,themaskedtrans-lationprimingasymmetry,andtherobustwithin-L2maskedpriminge ects.Toanticipateourresults,wefoundmaskedtranslationprimingintheL2–L1direc-tionusingsemanticcategorizationbutnotlexicaldeci-sion.Inthesecondpartofthearticle,weproposeanaccountofthese ndings,whichwetentativelyrefertoasthe‘‘SenseModel.’’TheSenseModelisdi erentfromothermodelsofbilinguallexicalprocessinginthatitproposesthatprimingbetweensemanticallyrelatedwordsdependsontheproportionofsharedsenses.ItseemsreasonabletosupposethatL1wordsareassoci-atedwithmanymoresemanticsensesthantheirL2counterparts.FromthisitfollowsthattheproportionofprimedsensesbelongingtothetargetwordwillbemuchhigherwhenanL2wordisprimedbyitsL1

counterpartthanwhenanL1wordisprimedbyitsL2counterpart.Notethatsimilarrepresentationalasym-metriesexistbetweensemanticallyrelatedwordswithinasinglelanguagetoo(e.g.,‘‘head’’hasmanysenseswhere‘‘skull’’onlyhasoneortwo).Hence,itshouldbethecasethatprimingasymmetriesexistforwithin-lan-guageprimingaswell(e.g.,‘‘head’’shouldprime‘‘skull,’’but‘‘skull’’shouldnotprime‘‘head’’).Inthethirdpartofthearticle,wetestspeci cpredictionsoftheSenseModel,usingwithin-Englishsemanticallyre-latedprime–targetwordpairs.Toanticipateourresultsonceagain,wefoundthatwewereabletorecreatetheprimingasymmetryseenintranslationprimingwhentheselectionofmaterialsadheredtotheassumptionsoftheSenseModelfortranslationequivalents.Thatis,inlexicaldecision,primingoccurredbetweenmany-andfew-sensewordsinthemany-to-few(analogousto‘‘L1–L2’’)direction,butnotinthefew-to-many(or‘‘L2–L1’’)direction.Furthermore,weobservedpriminginbothdirectionsinthesemanticcategorizationtask,therebyrecreatingthe(cross-language)taskdi erencebetweenlexicaldecisionandsemanticcategorization.Takentogether,these ndingsprovidestrongsupportfortheassumptionsoftheSenseModel.Moregener-ally,these ndingspresentdi cultchallengestoothercurrentmodelsofbilinguallexicalrepresentationandprocessing.

Experiment1—replicationofGraingerandFrenck-Mes-tre(1998)

AcomparisonoftheSanchez-Casasetal.(1992)studyandtheGraingerandFrenck-Mestre(1998)studyrevealsseveraldi erencesthatcouldhaveproducedthecontradictory ndings.Perhapsthemoststrikingdif-ferenceisinhowthecategorieswerepresentedtotheparticipants.Itisoftenthecaseinsemanticcategoriza-tiontasksthatitemsarepresentedinablockeddesignsuchthatalloftheexemplars(andanequalnumberofnon-exemplars)appeartogether(Bueno&Frenck-Mestre,2002;Forster&Hector,2002;Forster,Mohan,&Hector,2003;Frenck-Mestre&Bueno,1999).Gra-ingerandFrenck-Mestre(1998)usedthisprocedure,whereasintheSanchez-Casasstudy,categorieschangedoneachsuccessiveitem.Aconcernwithacategoryswitchingprocedureisthateachexperimentalitemmaye ectivelyconstituteaseparatetask,andtheconse-quenceofthis‘‘task-switchingparadigm’’maybedi-minishede ects.Asanexampleinsupportofsuchapossibility,Catchpole(1987)foundthatblockingitemsintocategoriesrestoredthefrequencye ectfornon-ex-emplars,ane ectthathadpreviouslygoneundetectedinexperimentsusingdi erentcategoriesoneachtrial(Balota&Chumbley,1984).Ifswitchingcategoriesoneachsuccessivetrialcanservetoeliminateane ectas

4M.Finkbeineretal./JournalofMemoryandLanguage51(2004)1–22

robustasthefrequencye ect,thenitfollowsthatblockingitemsaccordingtosemanticcategoryinse-manticcategorizationtasksmaybenecessarytoobserveanylexicalprocessinge ects,includingtranslationpriminge ects.Accordingly,itemsinthepresentex-perimentwereblockedaccordingtocategory,andpar-ticipantsweregivenpracticeitemswitheachnewcategory.

Justasthereisreasontobeconcernedthatthede-signoftheSanchez-Casasetal.studymayhaveledtoaTypeIIerror,thereisalsoreasontobeconcernedaboutparticularaspectsoftheGraingerandFrenck-Mestrestudy.Forexample,therewerearelativelysmallnum-berofparticipantsintheGraingerandFrenck-Mestrestudy(N¼12;twoofwhomweretheauthors,asop-posedtoanNof21intheSanchez-Casasstudy)andeachitemwaspresentedatotalof24di erenttimestoeachparticipant.Obviously,thisdegreeofrepetitionmayleadtostrategice ects.Similarly,thedominantlanguageoftheEnglish–FrenchbilingualsintheGra-ingerandFrenck-MestrestudymayhavebeentheirL2French(asaresultoflivinginFranceformanyyears),whereastheparticipantsintheSanchez-Casasetal.studywerecertainlynotdominantintheirL2English.Ifthiswerethecase,itmayexplainwhyGraingerandFrenck-MestreobservedprimingintheFrench–English(L2–L1)directionandwhySanchez-Casasetal.didnot(exceptforcognates)intheEnglish–Spanish(againL2–L1)direction.Inthepresentexperiment,thedominantlanguageofallofourJapanese–EnglishbilingualswastheirL1(Japanese).Method

Participants

TwentyJapanese–EnglishbilingualswererecruitedfromtheUniversityofArizonacampuscommunity.AllparticipantswerenativespeakersofJapaneseandwereemployedasgraduatestudentsattheUniversityofArizona.Participantshadreceivedaminimumof6yearsofEnglishinstructionwhileinJapanandatthetimeoftestingallhadbeenlivingintheUnitedStatesforatleast2years.Allparticipantswerepaidfortheirparticipation.

Materials

InordertoensuretranslationequivalencyforeachEnglish–Japaneseprime–targetwordpair,5Japanese–Englishbilinguals(fromthesamepopulationastheparticipantsintheexperiment)wereaskedtotranslatealistof170itemsfromEnglishintoJapanese(L2–L1);anothergroupof5wasaskedtotranslatethesameitemsintheoppositedirection(L1–L2).Onlythosewordpairsthatweretranslatedidenticallyineachdi-rectionbyallparticipantswereselectedascriticalitems.Fifty-twowordpairsmetthiscriterion.The

criticalitemsbelongedto11di erentsemanticcate-gories(partofbuilding,familyrelative,color,unitoftime,animal,profession,insect,scienti cdiscipline,partofthebody,kindofmetal,readingmaterial).Inordertomakeuseofablockeddesign,anadditional58itemswereselectedtoserveas‘‘practice’’itemsontrialsprecedingthecriticalitemsineachcategory.Thisre-sultedinatotalof110‘‘Yes’’items(10percategory),withaminimumof4percategorybeingcritical.Ad-ditionally,110non-exemplartargets(NOitems)werechosen.Thesewerechosensoastoensurethattheycouldnotbeconstruedasbelongingtoanyofthe11categories.Anadditional220Englishwordswerese-lectedtoserveasprimesoncontrolandpracticeex-emplartrials,aswellasonnon-exemplartrials.Thesewereunrelatedtotheirtargets,butwerematchedwiththecriticalprimesforfrequency,concreteness,image-ability,andword-length.

CarewastakentoensurethatnoneoftheJapanesetargetssharedcognatestatuswiththeirEnglishprimes.Alltargetswerepresentedintheirstandardscript,whichwasKanjiforallcategoriesexceptforthoseappearingintheINSECTcategory(thestandardscriptforinsectsisKatakana).

Designandprocedure

Itemswereblockedaccordingtosemanticcategory.Foreachcategory,therewere10exemplarand10non-exemplartrials.Eachtrialconsistedofthefollowingsequence(adaptedfromGraingerandFrenck-Mestre): rst,theparticipantwaspresentedwithaforwardmask(##########)for500ms,followedbyanEnglishprime(translationorcontrol)inlowercaselettersfor50ms,followedbyabackwardmaskfor150ms,andthentheJapanesetargetwordfor500ms.Thebackwardmaskwasinterpolatedbetweentheprimeandthetargetinordertoensurethatparticipantshadsu cienttimetoprocesstheL2prime.IthasbeensuggestedthatL2–L1primingmaynotoccurbecauseofaspeedofprocessingasymmetry(cf.Jiang,1999).Accordingtothisargu-ment,theprocessingofL2primeslagstoofarbehindtheprocessingofL1targetstoproduceanyobservablepriming.Consideringthispossibility,wepresentedabackwardmaskforarelativelylongduration(150msforatotalSOAof200ms).Apotentialconcernwiththisprocedureisthattheprimewillappearto‘‘popout’’ofthebackgroundcreatedbyidenticalforwardandback-wardmasks.Inordertopreventthispossibility,thebackwardmaskdi eredfromtheforwardmaskbothinfonttype(ArialBlack,asopposedtoTimesNewRoman)andsize(18asopposedto11).Thisparticularpresentationsequencewasapparentlye ectiveinpre-ventingparticipantsfrombecomingawareofthepres-enceoftheprimeastheyallexpressedsurpriseduringtheirdebrie ngtolearnthatanEnglishwordhadpre-cededtheJapanesetarget.

M.Finkbeineretal./JournalofMemoryandLanguage51(2004)1–225

Halfofthecriticaltargetsperlistwereprecededbytheirtranslationequivalentsandhalfwereprecededbyacontrolprime.Noothertargetswereprecededbytranslation-equivalentprimes.TwocounterbalancedlistswereconstructedsuchthatifatargetwasprecededbyitstranslationprimeonListA,itwasprecededbyitscontrolprimeonListBandviceversa.Notargetwordorprimewordwasrepeatedwithinlists.Partici-pantswereaskedtoindicatewhetherthetargetbe-longedtothecategorybypressingeitheraYESbuttonoraNObutton.Stimuliwerepresentedrandomlywithincategories,usingtheDMDXpackagedevelopedattheUniversityofArizonabyJ.C.Forster(Forster&Forster,2003),withtheonlyconstraintthatthe‘‘practice’’exemplarsprecededthecriticalitems.Re-sponsetimes(RTs)wererecordedtothenearestmilli-second.

Resultsanddiscussion

Datafromtrialsonwhichanerroroccurredwerediscardedandoutlierswerereplacedwithvaluesequaltocuto sestablished2SDunitsaboveandbelowthemeanforeachparticipant.Meanresponsetimeswere475msinthetranslationprimecondition,and494msinthecontrolprimecondition.AnANOVAshowedthatthis19mspriminge ectwashighlysigni cant(F1ð1;19Þ¼12:84,P¼:002;F2ð1;51Þ¼24:38,P<:001).Themeanerrorratewas1.2%anddidnotdi erbetweenconditions.

The ndingsofExperiment1revealthatourpar-ticipantspossessedsu cientL2lexicalprocessingskillstoexhibitreliablemaskedtranslationpriminge ectsintheL2–L1directioninasemanticcategori-zationtask.Assuch,theseresultscon rmthe ndingsofGraingerandFrenck-Mestre(1998).2AswepointedoutintheIntroduction,itisnotclearhowthesepa-ratememorysystemsaccountcouldaccommodatethese ndings.Asfortheconnection-strengthsaccountofthetranslationprimingasymmetry,these ndingsindicatethat,fortheseparticularbilinguals,L2form–meaningconnectionstrengthsdidnotconstitutealimitingfactorinL2–L1priming.Theconnection-strengthhypothesisshouldthereforepredictthattheseparticularbilingualsshouldalsoexhibitL2–L1prim-ingwhenthetaskischangedtolexicaldecision.WetestthispossibilityinExperiment2,usingthesamematerialsandmaskingprocedure,changingonlythetasktolexicaldecision.

2

This ndingalsosuggeststhatthemoste ectiveprocedureistoblockitemsbycategory,ratherthanchangingcategoryfromtrialtotrial,aswasthecaseintheSanchez-Casasetal.(1992)study.

Experiment2—maskedprimingwithL2primesinlexicaldecisionMethod

Participants

EighteenJapanese–Englishbilingualswererecruitedforthelexicaldecisionexperiment.Fourteenofthese18participantshadparticipatedinthesemanticcategori-zationexperiment6monthsprior.

Materials

ThematerialsfortheL2–L1translationprimingconditionwereidenticaltothoseusedinthesemanticcategorizationtaskexceptforthe110Japanesenon-wordsthathadtobecreated.Thesewerecreatedbydividingacolumnoftwo-characterJapanesewordsintotwocolumnsandrandomlysortingoneofthosecol-umnsandthenrejoiningthetwocolumns.Anyresultingcombinationsthatwerethoughtbythefourthauthor(anativespeakerofJapanese)tobewordswerediscarded.Additionally,80Englishwordswereselectedforthewithin-L2repetitionprimingcondition.Halfofthesewereabstractwordsandhalfwereconcretewords.ConcretenessvaluesweretakenfromtheMRCpsy-cholinguisticdatabase(availableonthewebathttp://www.psy.uwa.edu.au/mrcdatabase/uwa_mrc.htm).Theabstractwordshadameanconcretenessvalueof244.6(onascaleof100–700);theconcretewordshadameanconcretenessvalueof592.6.Thewordswerefurtherdividedonthebasisoffrequency.Halfofthewordswerehigh-frequencyitems(meanfrequencyof279oc-currencespermillion;CELEX)andhalfwerelow-fre-quencyitems(meanfrequencyof5.73occurrencespermillion).Thepurposeofselectingtheseparticularstimuliwastoseehowcloselythelexicalperformanceofthesebilingualspatternedtothatofnativespeakers.Ifthebilingualparticipantsexhibitclearfrequencyandconcretenesse ects,wecanberelativelysurethattheyareprocessingtheseitemsinlargelythesamewayasnativespeakers.

Inaddition,80non-wordtargetswereselectedtoelicit‘‘No’’responses.Non-wordsweregeneratedbytheARCNonwordDatabase(http://www.maccs.mq.edu.au/~nwdb/)suchthatallnon-wordshadortho-graphicallyexistingonsetsandbodies,aswellaslegalbigrams.

Designandprocedure

ThedesignandprocedurewaslargelyidenticaltothatofExperiment1withtheobviousdi erencethatinthistaskparticipantswereaskedtoindicatewhetherthetargets(JapaneseintheL2–L1condition;EnglishintheL2–L2condition)werewordsornot.Itemconstruction,includingprimedurationandSOA,wasidenticaltothatofExperiment1.Twolistswere

6M.Finkbeineretal./JournalofMemoryandLanguage51(2004)1–22

constructedineachprimingconditioninordertocounterbalanceacrosstheprimefactor.Inthetransla-tioncondition,targetsthatwereprecededbytheirmaskedL2translationequivalentonListAwerepre-cededbyamaskedcontrolprimeonListB,andviceversa.Inthewithin-L2repetitionprimingcondition,criticaltargetspairedwitharepetitionprimeonListAwerepairedwithacontrolprimeonListB,andviceversa.Therepetitionprimeconsistedofthesameletterstringasthetarget,butwaspresentedinlowercaseletters.Thecontrolprimewascompletelyunrelatedtothetargetexceptthatitwasmatchedwiththetargetonlength,frequency,andconcreteness.Notargetwordorprimewordwasrepeatedwithinlists.Itemswerepre-sentedrandomlytoeachparticipantandRTswerere-cordedtothenearestmillisecond.Results

ThesametrimmingprocedureusedinExperiment1wasemployedagain.Inthetranslationcondition,meanresponsetimeswere529msforJapanesetargetspre-cededbytheirEnglish(L2)translationequivalents,and525msfortargetsprecededbyEnglishcontrolprimes.

Thisslightinhibitorye ectof4mswasnotsigni cant(allFs<1).Themeanerrorratewas5.6%onnon-wordtrialsand1.3%onexperimentaltrials.Errorratesonexperimentaltrialsdidnotdi erbetweenconditions.Cruciallyforthepurposesofthisstudy,anANOVArevealedasigni cantinteractionbetweenthepriminge ectsofExperiments1and2(F1ð1;36Þ¼4:62,P¼:03;F2ð1;51Þ¼6:70,P¼:012).Thisinteraction,depictedinFig.1,providesconvincingsupportforataskdi erence,withreliablepriminginsemanticcate-gorizationbutnotlexicaldecision.

Inthewithin-L2repetitionprimingcondition,twoofthelow-frequencyabstractitemshadtobeexcludedfromanalysisbecauseofexceptionallyhigh-errorrates.Inordertobalancethedesign,twoitemswererandomlyselectedandremovedfromeachoftheotherwordtypes.ThemeanlexicaldecisiontimesandpercenterrorratesarepresentedinTable1.Foreachcomparisonofin-terest,twoanalysesofvariancewereperformed,onetreatingsubjectsasarandomfactor(F1),theothertreatingitemsasarandomfactor(F2).ThefactorswereWordType(abstractvs.concrete),Frequency(highvs.low),andPrimeType(identityvs.unrelated).ThefactorofPrimeTypewasarepeatedmeasuresfactorinbothanalyses,butthefactorsofFrequencyandWordTypewererepeatedinthesubjectanalysisbutnotintheitemanalysis.

AsTable1shows,therewereclearmaine ectsofwordtype,frequency,andprimetype.Participantsexhibitedaclearconcretenesse ect,respondingonaverage89msfastertoconcretewordsthantoab-stractwords.Thisdi erencewasfoundtobehighlyreliable,F1ð1;17Þ¼15:48,P<:01;F2ð1;68Þ¼12:67,P<:01.Likewise,participantsexhibitedaclearfre-quencye ect.Meanresponsetimeswere263msfasterforhigh-frequencyitemsthantheywereforlow-fre-quencyitems.Again,thisdi erencewasfoundtobehighlyreliable:F1ð1;17Þ¼88:70,P<:01;F2ð1;68Þ¼174:19,P<:01.Crucially,participantsexhibitedaclearrepetitionpriminge ect(83ms),whichwasfoundtobereliableinboththesubjectanditemsanalyses,F1ð1;17Þ¼18:95,P<:01;F2ð1;68Þ¼25:67,P<:01.Noneoftheinteractionsbetweenfactorsreachedsigni cance.These ndingsrevealrobust

Table1

Meanlexicaldecisiontimes(ms),percenterrorrates,andpriminge ectsasafunctionofwordtypeinwithin-L2repetitionpriming(Experiment2)

WordType

Abstract

Low-frequency

ControlPrimePriminge ect

990(29)913(33)77

High-frequency659(3)590(2)69

Low-frequency860(22)737(16)123

Concrete

High-frequency630(3)567(2)63

M.Finkbeineretal./JournalofMemoryandLanguage51(2004)1–227

lexicalprocessingskillsand,hence,indicatethatthelackofL2–L1priminginlexicaldecisionwasnotduetoparticipantsÕinabilitytoprocessL2primesundertheseparticularmaskingconditions.Discussion

Takentogether,the ndingsofExperiments1and2demonstrateclearlythatL2–L1primingistaskdepen-dent.Despiteobservingrobustwithin-L2maskedrepe-titionpriming(con rmingthe ndingsofGollanetal.,1997;Jiang,1999),andclearL2–L1maskedtranslationpriminginsemanticcategorization(con rmingthe ndingreportedbyGrainger&Frenck-Mestre,1998),therewasamarkedabsenceofL2–L1translationpriminginthelexicaldecisiontask(cf.Gollanetal.,1997;Grainger&Frenck-Mestre,1998;Jiang,1999;Keatlyetal.,1994).Modelsofbilinguallexicalpro-cessingsuchastheRHMhavedi cultyaccountingforthistaskdependencybecauseL2form–meaningcon-nectionstrengths,proposedtobethelimitingfactorsinL2–L1priming,cannotatthesametimelimitpriminginonetaskbutnotanother.

Onemaywanttoargue,though,thatthetaskdif-ferenceisduesimplytothefactthatconsiderationofmeaningiscriticalforsemanticcategorizationbutnotforlexicaldecision.Thatis,participantsessentiallyig-noremeaningwhentheymakelexicaldecisions,butnotwhentheymakesemanticcategorydecisions.Theproblemwiththisargumentisthatitfailstoexplaintheexistenceofstrongsemanticandassociativeprim-inge ectsinlexicaldecision(seeNeely,1991forareview),whichsuggeststhatsemanticandconceptualpropertiesarehighlyrelevanttothedecisionprocess.Alternatively,itcouldbearguedthatthesemanticcategorizationtaskitselfampli esthestrengthoftheform–meaningconnections,sothattheweakL2form–meaningconnectionsarenowstrongenoughtopro-ducepriming.Ifthiswerethecase,thenweshouldobserveprimingforbothexemplars(aswasthecaseinExperiment1)andnon-exemplars.Analternativepossibilityisthatinthesemanticcategorizationtask,theexistenceofthecategoryitselfmightactasacontext,whichmayservetoemphasizethecategory-relevantsemanticpropertiesofboththeprimeandtarget,therebyincreasingthesemanticoverlapbetweenthem.Ifthiswerethecase,thenprimingwouldbeexpectedonlyforexemplars,sincethenon-exemplarswouldnothaveanycategory-relevantsemanticfea-tures,andhencewouldbeuna ectedbythecategorycontext.Thus,itiscriticaltodeterminewhetherprimingisobtainedfornon-exemplarsaswellasex-emplars:asimpleampli cationmodelpredictsprimingforboth,butacontextselectionmodelpredictsprim-ingonlyforexemplars.WetestthispossibilityinExperiment3.

Experiment3—semanticcategorizationwithnon-exemp-lars

Experiment3wasidenticaltothatofExperiment1,theonlyexceptionbeingthatthecriticalitemsfromExperiment1werenowpresentedasnon-exemplars.Thiswasachievedbyrandomlyselecting48oftheori-ginal52itemsanddistributingthemevenlybetween6newcategories(militarytitle,partofspeech,weapon,naturalearthformation,weatherphenomenon,andgirl’s rstname).Theseparticularcategorieswerechosenfortworeasons:(1)noneofourcriticalitemscouldbeconstruedashavingexemplarstatusinanyofthecate-goriesand(2)noneoftheexemplarsinthesecategorieswerecognateswiththeirEnglishtranslationequivalents.Eachcategoryconsistedof30items,15exemplarsand15non-exemplars.InordertomaintainconsistencywithExperiment1,thecriticalitems(N¼8percategory)wereinterleavedwithanequalnumberofexemplarsandpresentedrandomlywithincategories,withtheonlyconstraintbeingthatthecriticalitems(andanequalnumberofexemplars)werepresentedsubsequenttotheotheritemsineachcategory.Theparticipants(N¼20)wererecruitedfromthesamestudentpopulationofin-dividualstestedinExperiments1and2.Again,eachparticipantwaspaidfortheirparticipation.Results

ThesametrimmingprocedureusedinExperiments1and2wasemployedagain.Meanresponsetimesforthecriticalitemsasnon-exemplarswere605mswhenpre-cededbyEnglish(L2)translation-equivalentprimes,and609mswhenprecededbyEnglishcontrolprimes.Thisslightpriminge ectof4mswasnotfoundtobesig-ni cant(allFs<1).Errorrates(M¼3:5%)didnotdi ersigni cantlybetweenconditions.Importantly,therewasaninteractionbetweenthepriminge ectsofExperiment1,whenthecriticalitemswereexemplars,andthoseofthepresentexperiment,whenthecriticalitemswerenon-exemplars(F1ð1;36Þ¼6:99,P¼:01;F2ð1;92Þ¼3:53,P¼:07).Discussion

Thelackofareliablepriminge ectinthepresentexperimentfornon-exemplarssuggeststhatthere-quirementsofthesemanticcategorizationtaskalonedonotsimplystrengthentheform–meaningconnectionsnecessarytoproduceL2–L1priming.Furthermore,thenatureoftheinteractionbetweenthepriminge ectsobservedinExperiment1andthelackofpriminginthepresentexperimentsuggeststhatreliableprimingintheL2–L1directionisobservedinsemanticcategorizationonlywhenparticipantsgeneratedecisionsforexemplars.Thisresultisconsistentwiththeproposalthatthe

8M.Finkbeineretal./JournalofMemoryandLanguage51(2004)1–22

categoryactsasakindofcontextwhichemphasizesthecategory-relevantpropertiesoftheprimeandtarget.Suchane ectwouldbeexpectedonlyforexemplarsofthecategory.Inwhatfollows,weprovideapossibleexplanationforthispatternof ndings.

TheSenseModel

Webeginwithanissuethathasnotreceivedade-quateattentionbeforeinthecross-languageprimingliterature:mostwordsarepolysemousandtherangeofsensesthatawordcanhavetendstobelanguagespeci c.3Forexample,theEnglishword‘‘black’’andtheJapaneseword‘‘kuroi’’aretranslationequivalents,despitehavinglittlebesidestheonesenseCOLORincommon.AccordingtoWordNet1.6(Fellbaum,1998),theword‘‘black’’has21di erentsenses.Forexample,‘‘black’’isusedtorefertoatypeofhumor,aswellasacalamitousdayonWallStreet(blackMonday).InJapanese,‘‘kuroi’’canbeusedtorefertothosewhoareevil-minded(blackbelly),aswellasthosewhoarewelltannedorguiltyofacrime.Hence,thelanguage-speci cshadesofmeaning,orsenses,forthesetwowordsextendwellbeyondthesinglesensedeterminingtranslationequivalency.Therelevanceofthisfortranslationprimingisthattheamountofprimingmaydependnotonlyontheoverlapinthesemanticsensesactivatedbytheprimeandthetarget,but,crucially,ontheratioofprimedtounprimedsensesassociatedwiththetarget.

Inthinkingabouthowtoformulateanaccountofthetranslationprimingasymmetrythatmakescentralthefactofpolysemy,wehavebegunwithasetofassump-tionsthat,thoughfundamentallydi erentfromthoseonwhichtheRevisedHierarchicalModelandtheSeparateMemorySystemsModelarebased,resonatefairlyclo-selywiththoseoftheDistributedConceptualFeatureModel(deGroot,1992).IndeGrootÕsdistributedcon-ceptualfeaturemodel(DCFM),lexicalnodesareasso-ciatedwithadistributedsetofconceptualfeatures,asopposedtolocalistconceptualrepresentations.Ac-cordingtotheDCFM,therearevaryingdegreesofoverlapofconceptualfeaturesbetweenL1andL2meanings,dependingonwhattypeofwordisrepre-sented.Forexample,deGroot(1992,1993)hasarguedthatthereismorefeaturaloverlapbetweentranslationequivalentsforconcretewordsthanabstractwordsbe-causeconcretereferentstendtohavethesameshape,size,andfunctioncross-linguistically.TheDCFMis

3Here,wedistinguishpolysemyfromambiguity.Apolyse-mousword(e.g.,‘‘call’’)hasseveralshadesofmeaningorusages(i.e.,senses)thatareallconsideredtobepartofthesameword(e.g.,‘‘telephonecall’’and‘‘tocallout’’).Anambiguouswordlike‘‘bark,’’though,canbeconsideredtobetwodistinctwords(‘‘tobark’’and‘‘treebark’’)thatjusthappentosharethesameorthographicandphonologicalform.

intuitivelyappealingonseveralfronts,notleastofwhichisitsstraightforwardaccountofhowtranslationequivalentscanhavelanguagespeci cmeanings.Infact,deGroot(1993)hasreferredtotheDCFMasa‘‘mixed’’modelbecauseitispossibleforthemeaningrepresen-tationstorangefromhavingcompletefeaturaloverlap(essentiallythearchitectureofacompoundbilingual)tohavingnofeaturaloverlap(i.e.,thearchitectureofacoordinatebilingual)dependinguponthetypeofwordinquestion.

AcriticalproblemwiththeDCFM,ascurrentlyspeci ed,though,isitsdi cultyinaccountingforasymmetriesintranslationperformance.Thatis,be-causethismodelattributestranslationperformancetotheconceptualfeaturescommontobothL1andL2,andbecausetheamountoffeaturaloverlapbetweentwotranslationequivalentsisconstantregardlessoftrans-lationdirection,anye ect(suchaspriming)foundinonedirectionshouldoccurequallystronglyintheotherdirection.Assuch,theDCFMhasaparticulardi cultyinaccountingforthemaskedtranslationprimingasymmetry.Intheirrevisionoftheconceptualfeaturemodel,KrollanddeGroot(1997)haveaddressedthisshortcomingbyproposingweakconnectionsbetweenL2lexicalnodesandconceptualfeatures.Asthe ndingsofExperiments1–3indicate,though,theconnection-strengthapproachmaynotbethecorrectone.Instead,wehavechosentopursueadi erentmodi cationtotheDCFM,oneinwhichlexicalsemanticrepresentationsareassumedtobeboundedandcomprisedofdistinct‘‘bundles’’offeaturescorrespondingtodistinctusages,whichwerefertoassemanticsenses.Thesemodi ca-tionstotheDCFMpresentuswiththehypothesisthatwhatiscriticalinobservingtranslationprimingisthedegreetowhichthecompletelexicalsemanticrepresen-tation(asopposedtojustthefeaturesincommonbe-tweentranslationequivalents)hasbeenactivatedbytheprime.

Thenumberofbundledfeatures,orsenses,presentinalexicalsemanticrepresentationcanrangefromonetoseveraldozen.Theexactnumberdependsonhowmanyusagesthatparticularwordhasand,importanttothediscussionofthebilinguallexicon,howknowledgeabletheindividualisofthoseusages.Becausetheshadesofmeaningandrangeofusagesthataparticularwordmayhavevaryfromlanguagetolanguage,itisassumedthatseveralofawordÕssenses(i.e.,thewayinwhichcon-ceptualfeaturesareboundtogether)willbelanguagespeci c,withanobviousexceptionforthesemanticsense(s)determiningtranslationequivalency.These-manticsensedeterminingtranslationequivalencyissharedbetweenL1andL2lexicalentries,andisthusconsideredtobelargely,ifnotcompletely,identicalbetweenL1andL2(seeFig.2).

Byassumingthateachsenseofawordconstitutesadistinctmentalrepresentationwithinalexicalsemantic

M.Finkbeineretal./JournalofMemoryandLanguage51(2004)1–229

representation,wearepresentedwiththepossibilityofarepresentationalasymmetrybetweenrelatedwords.In-stancesofthisasymmetrycanbefoundbetweenwordswithinalanguageaswellasbetweentranslationequiv-alentsacrosslanguages.TakingawordpairwithinEn-glishasanexample,arepresentationalasymmetryissaidtoexistbetweentwolexicalrepresentationswhen(one)theyshareatleastonesemanticsenseand(two)whenonewordhasmanydi erentsenses(e.g.,‘‘head’’hasmultipleusages,including:partofananimal’sbody;headofstate;fullheadofsteam;headinginadirection,etc.)andtheotherhasveryfew(e.g.,‘‘skull’’onlyhasonesense:partofananimal’sbody).Withinlanguages,itisrelativelyuncommonto ndwordpairsthatmeetthesecriteria.Wesuggestthattranslationequivalents,ontheotherhand,frequentlymeetthecriteriaofarepresentationalasymmetry.Forexample,translationequivalents,byvirtueofbeing‘‘equivalent,’’presumablyshareatleastonesemanticsense.Furthermore,becausebilinguals,generallyspeaking,aremorepro cientintheirL1thanintheirL2and,thus,aremorefamiliarwiththerangeofusagesthatL1wordsmayhave,thenumberofsensesassociatedwithL1wordsisthoughttoexceedthenumberofsemanticsensesassociatedwithL2words.

Thetranslationprimingasymmetryinlexicaldeci-sionisseentobethenaturalconsequenceofthisrep-resentationalasymmetry.Recentworkinvestigatingthein uenceofsemanticsensesonlexicaldecisionpro-cesseshasfoundthatresponsetimesaresensitivetothenumberofsensesassociatedwithaparticulartarget(Rodd,Gaskell,&Marslen-Wilson,2002;seealsoGeneraldiscussion).Giventhis nding,itisreasonabletoconsiderthepossibilitythatsemanticprimingre ectsthee cacyoftheprimewordtopreactivatethese-manticsensesassociatedwiththetargetword(cf.Cree,McRae,&McNorgan,1999,forasimilar‘‘featureoverlap’’accountofsemanticpriming).Accordingtothisaccountofsemanticpriming,theamountofprimingwouldincreaseastheproportionofprimedtounprimedsensesinthelexicalsemanticrepresentationofthetargetwordincreases.IntheL1–L2direction,weproposethattheproportionofL2sensesprimedbytheL1primeisgoingtobeveryhigh,ifnotcomplete.Thisisbecausemost,ifnotall,ofthesensesassociatedwiththeL2formarealsoassociatedwiththeL1equivalentform(muchlikethesenseof‘‘skull’’isalsoassociatedwith‘‘head’’).Priminginthisdirection,then,shouldbeveryreliable,anditis(deGroot&Nas,1991;Gollanetal.,1997;Jiang,1999;Keatlyetal.,1994;Williams,1994).However,inthereversedirection(L2–L1),theproportionofL1sensesprimedbytheL2primeisgoingtobeverylow(e.g.,1:8).Thisisbecauseitisfrequentlythecasethattherearemanysensesassoci-atedwiththeL1formthatarenotsimilarlyassociatedwiththeL2prime(again,muchliketherelationshipbetween‘‘head’’and‘‘skull,’’wheremanysensesof‘‘head’’arenotassociatedwith‘‘skull’’).Accordingly,priminginthisdirectionshouldbeveryweak,anditis(Gollanetal.,1997;Grainger&Frenck-Mestre,

1998;

10M.Finkbeineretal./JournalofMemoryandLanguage51(2004)1–22

Jiang,1999;Jiang&Forster,2001;Keatlyetal.,1994;Sanchez-Casasetal.,1992;aswellasExperiment2ofthepresentstudy).

IfitiscorrecttoassumethatL2–L1primingdoesnotoccur(orisveryweak)inlexicaldecisionbecauseaninsu cientnumberofL1sensesarepreactivatedbytheL2prime,thentheremustbesomethingaboutthesemanticcategorizationtaskthatservestorestrictthenumberofsensesrecruitedwhenadecisionisgener-ated.Aswehavealreadyindicated,itisplausibletoarguethatthecategoryactsasacontextthatservestofocushowdecisionsaremadeinresponsetothetarget,sothatonlythesemanticsensethatisrelevanttothecategoryistakenintoaccount(e.g.,inordertodecidethat‘‘black’’isacolor,onedoesnotneedtotakeintoconsiderationthemultiplesensesthat‘‘black’’mayhave).Essentially,wearguethatthesemanticcatego-rizationtaskturnsmany-sensetargetsintoone-sensetargetsby‘‘ ltering’’outcategory-irrelevantsensesfromthedecisionmakingprocess.Assumingthattheprime–targetpairshavebeenproperlyselected,thissuggeststhattheL2primewillbemoree ectiveinprimingtheL1targetbecausetheonlysensethatisrelevantforthedecisionwillbetheverysensethathasbeenprimedbytheL2prime.

TheSenseModeldi ersfromothermodelsofbi-linguallexicalrepresentationandprocessinginthatitsexplanationofthetranslationprimingasymmetryisnotrestrictedtoapeculiarityofthebilinguallexicon(e.g.,weakconnectionstrengths).Thatis,theSenseModelpredictsthesamepatternofpriminge ects(andtaskdi erences)forwordpairswithinalanguagewhoserelationshipmeetstherepresentationalasym-metrycriteriaspeci edabove.Infact,therearereasonstothinkthatthestrongesttestoftheSenseModelwouldbetousewithin-languagewordpairsinamaskedprimingexperimentwithlexicaldecisionandsemanticcategorization.IftheassumptionsoftheSenseModelarecorrect,thenwordpairslike‘‘head–skull,’’whiche ectivelyinstantiatetheassumedrepre-sentationalasymmetrybetweentranslationequivalents,shouldproducethesameprimingasymmetryinlexicaldecisionbutnotsemanticcategorization.Thesepre-dictionsoftheSenseModelaretestedinthefollowingexperiments.

Experiment4—within-languageprimingasymmetryinlexicaldecision

Inthepresentexperiment,wetestthepredictionsoftheSenseModelinamaskedprimingtaskwithlexicaldecision.IftheassumptionsoftheSenseModelarecorrect,Englishmaterialsthate ectivelyrecreatetherepresentationalasymmetrythoughttoexistbe-tweentranslationequivalentsshouldproducepriming

inthemany-to-onedirection(e.g.,‘‘head–SKULL’’),butnotintheone-to-manydirection(‘‘skull–HEAD’’).Participants

Forty-fourundergraduatesattheUniversityofAri-zonaparticipatedforcoursecredit.Twenty-twooftheparticipantsweretestedinthemany-to-one(thoughttobeanalogoustoL1–L2)direction(Experiment4a)and22weretestedintheone-to-many(thoughttobeanal-ogoustoL2–L1)direction(Experiment4b).Allpartic-ipantswerenativespeakersofEnglish.Materials

Severaldi erentselectioncriteriawereemployed.First,itwasimperativetoselectwordpairsthatwerethoughttoshareacommonsense.Thiswasdonebyselectingsynonyms,orclosesynonyms.Second,itwasimportantthatthematerialssatis edthenumericaldiscrepancycriterionoftherepresentationalasymme-trybetweenthewords.Thiswasdonebyensuringthatonewordinthepairhadonlyonesense(e.g.,‘‘oar’’hasonlyonesense)andthattheotherwordinthepairhadseveral(e.g.,‘‘paddle’’has10senses).SensecountsweretakenfromWordNet1.6(Fellbaum,1998).One-sensewordswerepairedwithwordshavingelevensensesonaverage(seeAppendixAforacompletelistofwordpairsandsensecounts).Fortywordpairsintotalwerecreated.Anequalnumberofnon-wordswasgeneratedbytheARCNonwordDatabase(Rastle,Harrington,&Coltheart,2002)andinterleavedwiththeexperimentalitems.Additionally,40wordsthatwerematchedinlengthandfrequencywiththeex-perimentalprimeswereselectedtoserveascontrolprimes.

Designandprocedure

Experiments4aand4bwereidenticalindesignandprocedure.ParticipantsinbothexperimentswereaskedtodecidewhetherthetargetitemwasanEn-glishwordornot.InExperiment4a,theprimeshadmanysenses,whilethetargetshadjustone.ThiswasreversedinExperiment4b.Eachtrialconsistedofthefollowingsequenceofevents,eachonefollowingimmediatelyaftertheother:forwardmask(########)for500ms,primeinlowercaselettersfor41ms,andtargetinuppercaselettersfor500ms.TwolistswereconstructedforbothExperiments4aand4binordertocounterbalanceacrosstheprimefactor.TargetsthatwereprecededbytheirrelatedprimeonListAwereprecededbyacontrolprimeonListBandviceversa.ItemswerepresentedrandomlytoeachparticipantandRTswererecordedtothenearestmillisecond.

M.Finkbeineretal./JournalofMemoryandLanguage51(2004)1–2211

Results

ThesametrimmingprocedureusedinExperiments1–3wasemployedhere.InExperiment4a,themany-to-one(or‘‘L1–L2’’)condition,meanresponsetimeswere550msfortargets(e.g.,OAR)precededbyarelatedprime(e.g.,paddle)and573msfortargetsprecededbyacontrolprime.AnANOVAshowedthatthisfacilitatione ectof23mswasreliable(F1ð1;20Þ¼12:41,P¼:002;F2ð1;38Þ¼14:26,P<:001).Themeanerrorratewas5.9%andwasnotdi erentbetweenexperimentalandcontrolconditions(allFs<1).

InExperiment4b,theone-to-manycondition(or‘‘L2–L1’’direction),thepatternofresultswasmarkedlydi erentdespiteusingthesamewordpairs,albeitintheoppositedirection.Meanresponsetimeswere538msfortargets(e.g.,PADDLE)precededbyarelatedprime(e.g.,oar)and528msfortargetsprecededbyacontrolprime.AnANOVAshowedthatthisinhibitorye ectof10mswasnotreliable(F1ð1;20Þ¼1:71,P¼:21;F2ð1;38Þ¼3:24,P<:079).Themeanerrorratewas5.7%andwasnotdi erentbetweenconditions(allFs<1).Importantly,therewasasigni cantinteractionbetweentheresultsofExperiments4aand4b(F1ð1;40Þ¼10:49,P¼:002;F2ð1;72Þ¼15:70,P<:001).

Onepossibleexplanationforthisinteractionthatweconsiderhastodowiththelargedi erenceinsurfacefrequencybetweentheprimesandthetargets.Many-sensewords,byvirtueofhavingseveraldi erentusages,haveamuchhighersurfacefrequencythanfew-sensewords.Becausesurfacefrequencyandnumberofsensesaresohighlycorrelated,wewereunabletocontrolap-propriatelyforthisinouritemselection,andthecon-cernisthathigh-frequencywordsmaybebettersemanticprimesthanlow-frequencywords.Althoughitiswellknownthatfrequencyrarelyinteractswithmaskedrepetitionpriminge ects(cf.Experiment2),muchlessisknownaboutthepossibilityoffrequencyinteractingwithmaskedsemanticpriminge ects.Inordertoaddressthispossibility,wecalculatedthepriminge ectforeachitemfromExperiments4aand4bandenteredthemintoasimultaneousmultipleregres-sionanalysiswiththesensecountsofthetargetwordsandthesurfacefrequenciesoftheprimewordsincludedaspredictors.Thisanalysisrevealedthatalthoughthesensecountsofthetargetwordsaccountedforasig-ni cantproportionofthepriminge ect(t¼À2:97,P¼:003),thesurfacefrequenciesoftheprimewordsdidnot(t¼À0:27,P¼:788).Discussion

These ndingscon rmthepredictionsoftheSenseModelandhenceprovidesupportfortheassumptionthatarepresentationalasymmetrybetweenL1andL2

equivalentsmaybethesourceofthetranslationprimingasymmetry.Maskedprimingwasobservedinthelexicaldecisiontaskbetweenmany-sensewords(e.g.,‘‘paddle’’)andone-sensewords(e.g.,‘‘oar’’)inthemany-to-onedirection.Thise ectisanalogoustothemaskedtrans-lationpriminge ectsfoundwithbilingualsintheL1–L2direction(deGroot&Nas,1991;Gollanetal.,1997;Jiang,1999).Usingthesamewordpairs,butnowintheone-to-manydirection,noprimingwasobserved.This ndingisanalogoustothelackofmaskedtranslationprimingintheL2–L1direction(Gollanetal.,1997;Grainger&Frenck-Mestre,1998;Jiang,1999;Sanchez-Casasetal.,1992;andExperiment2above).

Theseresultsareimportantinatleasttwodi erentways.First,theyprovidejustthekindofevidencenee-dedtocon rmthecentralassumptionoftheSenseModel.Second,theyrepresentarelativelyunusualcaseofsemanticpriminginthatthisisamaskedsemanticpriminge ectinlexicaldecision;mostpreviousstudiesreportingsemanticpriminge ectshaveusedmuchlongerprimedurationsorsemanticjudgmenttasks.4

Experiment5—symmetricalpriminginsemanticcatego-rization

InExperiment4,thecentralassumptionoftheSenseModelwascon rmed,namelythattheprimingasym-metryfrequentlyobservedintranslationprimingistheconsequenceofarepresentationalasymmetrybetweenL1andL2translationequivalents.Wenowturnourinvestigationtothetaskdi erenceinL2–L1translationpriming.Here,thequestionishowthesemanticcate-gorizationtaskeliminatestheprimingasymmetrythatisclearlypresentinlexicaldecision.Wehavesuggestedthatinlexicaldecision,thereisnotaskrelevantcategorytorestricthowparticipantsgeneratedecisions,e ec-tivelydilutingthee cacyoftheL2prime.Insemanticcategorization,though,thepresenceofrelevantcategoryinformationmayservetofocushowdecisionsaremadeinresponsetothetarget.Theconsequenceofthis‘‘fo-cusing’’e ectisthatonlythesemanticsensethatisrelevanttothecategoryistakenintoaccount,therebyallowingtheL2primetobee ective.Thatis,byhy-pothesis,semanticcategorizatione ectivelyturnsmany-sensewordsintoone-sensewordsforthepurposesofthetask.Ifthishypothesisiscorrect,thenprimingshouldbesymmetricalbetweenmany-andfew-sensewordsin

4

Subsequentreplicationsofthisexperimenthaveuncoveredacuriousfact:many-to-onepriminge ectsareonlyobservedwhenatleast50%ofthepositiveitemsareofthe‘‘paddle–OAR’’type.Currently,wehavenoexplanationforthisresult,otherthantopointoutthatthismaybesimilartotherelatednessproportione ectsreportedbyBodnerandMasson(2001)inmaskedpriming.

12M.Finkbeineretal./JournalofMemoryandLanguage51(2004)1–22

semanticcategorization,despitebeingclearlyasymmet-ricalinlexicaldecision(Experiment4).Wetestthishypothesisinthepresentexperiment.

Themaskingprocedureadoptedinthisexperimentdi ersfromthepreviouscross-languagesemanticcate-gorizationexperiments(Experiments1and3)inthatthedurationoftheinterpolatedmaskisreducedfrom150to13.7ms.ThiswasnecessarybecausetheprimeswerenowpresentedinL1ratherthanL2,andwitha150mspost-primemask,theprimemayhavebeenvisible(i.e.,primevisibilitydependsonlanguagecompetence).Thereasonforincludinganinterpolatedmaskatallwasthatareviewoftheliteraturerevealedthatthisisthetypicalpresentationsequenceemployedinmaskedsemanticprimingexperimentsusingsemanticcategorizationwithnativespeakers(Bueno&Frenck-Mestre,2002;Frenck-Mestre&Bueno,1999).Italsomakesthesewithin-languageexperimentsmorecomparabletotheearlierbetween-languageexperiments.However,becauseofconcernsaboutthevisibilityoftheprimespresentedinL1,wealsoincludedatestofprimeawarenessthatwasgivenaftertheexperimentproper.Participants

FortyundergraduateparticipantsattheUniversityofArizonaparticipatedforcoursecredit.Twentyoftheparticipantsweretestedinthemany-to-one(‘‘L1–L2’’)direction(Experiment5a)and20weretestedintheone-to-many(‘‘L2–L1’’)direction(Experiment5b).Materials

ItemswereselectedinlargelythesamewaythatitemsforExperiment1hadbeenselected.Forty-twowordstakenfrom8di erentsemanticcategorieswereselectedtoserveasmany-sense(or‘‘L1’’)targetsintheone-to-manyprimingdirection.Thesewerechosenaccordingtotwodi erentcriteria:one,theyhadtobeamongthe rst10exemplarsofthecategory(Battig&Montague,1969;Uyeda&Mandler,1980)and,two,theyhadtohaveatleast5semanticsensesassociatedwiththem(WordNet1.6).Thesemanticcategorieswereasfollows:articleofclothing,articleoffurniture,bird,carpenter’stool,color,four-footedanimal,kitchenutensil,bodypart.Addition-ally,42wordswereselectedtoserveasone-sense(‘‘L2’’)primes.Selectionoftheseprimewordswasdoneac-cordingtothesamecriteriausedinExperiment4:thatis,itemshadonlyonesense(WordNet)andthatsensehadtobesharedwiththetarget.Anexampleofaprime–targetpairis‘‘skull–HEAD.’’Here,‘‘skull’’onlyhasonesense,whichissharedwith‘‘head.’’‘‘Head,’’ontheotherhand,hasmorethan20senses,onlyoneofwhichconstitutesthemeaningof‘‘skull.’’Inadditiontothese42prime–targetpairs,additionalexemplarswereselectedforatotalof10exemplarspercategory.These

lleritemswerealsofromthe rst10exemplarsforeachcategory,butthenumberofsensesassociatedwiththemwasnotcontrolledfor.Eightynon-exemplarswerealsoselected.Thesewerematchedwiththeexemplarsonlengthandfrequency,butcarewastakentoensurethatnoneoftheseitemscouldbeconstruedashavingex-emplarstatusinanyofthecategoriesintheexperiment.Finally,160additionalwordswereselectedtoserveasprimewordsforthe llerexemplars,thenon-exemplars,andforthecriticaltargetsinthecontrolcondition.Thesewerematchedwiththeexperimentalprimesonlengthandfrequency,butagaincouldnotbeconstruedashavingexemplarstatusinanyofthecategoriespresentintheexperiment.Designandprocedure

ThedesignandprocedurewasidenticalforExperi-ments5aand5b.MuchlikeExperiment1,itemswereblockedaccordingtosemanticcategory.Foreachcate-gory,therewere10exemplarand10non-exemplartri-als.Itemswerepresentedinadi erentrandomorderforeachparticipantwiththeconstraintthatthecriticalitemsfollowedthe lleritemsineachcategory.Criticaltargets(e.g.,‘‘HEAD’’)werematchedwitharelatedprime(e.g.,‘‘skull’’)aswellasacontrolprime(e.g.,‘‘shop’’).TwolistswereconstructedforbothExperi-ments5aand5binordertocounterbalanceacrosstheprimefactor.TargetsthatwereprecededbytheirrelatedprimeonListAwereprecededbyacontrolprimeonListBandviceversa.AsinExperiments1and3,eachtrialconsistedofthefollowingsequenceofevents: rst,aforwardmask(##########)waspresentedfor500ms,followedbytheprimeword(relatedorcontrol)inlowercaselettersfor41ms,followedbyabackwardmaskforonerefreshcycle(13.7ms),andthenthetargetwordinuppercaselettersfor500ms.Becauseofcon-cernsaboutthevisibilityoftheprimespresentedinL1,wealsoincludedatestofprimeawarenessthatwasgi-venaftertheexperimentproper.Thistestrequiredparticipantstoindicatewhetherornottheprimewordcontainedtheletter‘‘e.’’Participantswere rstshownseveralpracticetrialsatamuchslowerpresentationratesothattheycouldseetheprimeword.Afteritwasclearthattheyunderstoodtherequirementsofthisnewtask,theywerepresentedwith42trialsinadi erentrandomorderforeachparticipant.Halfofthesetrialshadprimewordsthatcontainedtheletter‘‘e,’’andhalfdidnot.Thetargetswerecompletelyunrelatedwords,andnoneofthetargetscontainedtheletter‘‘e.’’Thirty-twoofthetrialswerepresentedusingthesametemporalcharac-teristicsusedintheexperimentproper;onthe10re-mainingtrials(5withtheletter‘‘e’’and5without),theprimewordwaspresentedfor10refreshcycles(137.3ms).Onthesetrials,theprimewordwasvisible,thoughonlyverybrie y.Thepurposeofpresenting

M.Finkbeineretal./JournalofMemoryandLanguage51(2004)1–2213

trialswithvisibleprimeswastoencourageparticipantstocontinuetryingtodothetask.Anacceptableaccu-racyrateonthesetrialsallowsustosafelyassumethatparticipantswereperformingthetasktothebestoftheirabilities.

Resultsanddiscussion

Experiment5a—one-to-manyinsemanticcategorizationThesametrimmingproceduresusedinthepreviousexperimentswereagainemployedhere.InExperiment5a,meanresponsetimesintheone-to-many(or‘‘L2–L1’’)directionwere473msfortargets(e.g.,HEAD)precededbyrelatedone-senseprimes(e.g.,skull)and486msforthesametargetswhenprecededbyacontrolprime(e.g.,shop).AnANOVAshowedthatthisfacili-tatione ectof13mswasreliable(F1ð1;18Þ¼8:15,P¼:01;F2ð1;40Þ¼7:02,P¼:01).Furthermore,justastherewasaninteractionbetweentheresultsofthese-manticcategorizationandlexicaldecisiontasksintranslationpriming(Experiments1and2),sowasthereaninteractionbetweentheresultsofthepresentexper-imentandthoseofExperiment4b(F1ð1;36Þ¼6:54,P¼:01;F2ð1;76Þ¼5:47,P¼:02).Theseresultscon- rmthetaskdi erencebetweensemanticcategorizationandlexicaldecisioninone-to-manypriming.Insodo-ing,these ndingsprovidefurthersupportforthehy-pothesisthatsemanticcategorizationservestorestricttheamountofsemanticinformationrecruitedtogen-erateadecision,therebyenhancingthee ectivenessofthefew-senseprimeword.

Themeanerrorrateinthisexperimentwas5.4%:3.3%onthetrialswithrelatedprimesand7.5%ontrialswithcontrolprimes.Thisdi erencewassigni cantinboththesubjectanditemsanalyses(F1ð1;18Þ¼7:06,P¼:01;F2ð1;40Þ¼7:25,P¼:01),indicatingthatpar-ticipantsbene tedfromarelatedprimeintermsofbothreactiontimesandaccuracy.Theerrorratesconstitutetheonlydi erencebetweenthisexperimentandExper-iment1.InExperiment1,bilingualsdidnotexhibitanydi erenceinerrorratesbetweenconditions.Here,ontheotherhand,thereisthesuggestionthatinformationpresentedintheprimestimuluswasnotonlyactivatingthetargetrepresentation,butalsoenteringintothede-cisionmakingprocess.Althoughitissomewhatunclearhowthiscouldoccur(participantswereunawareoftheprime,seebelow),thepatternoferrorsspeakstothepossibilityofadecisioncon ict.Thatis,participantsmayhavebegungeneratingadecisiononthebasisoftheprimeÕscategorymembership,whichwouldfacilitatedecisionsontheexperimentaltrials,butnotonthecontroltrials.Thisisapossibilitythatweconsiderfur-therinExperiment6.

WithrespecttoparticipantsÕperformanceontheprimeawarenesstask,themaskingprocedureappearstohavebeene ectiveinpreventingawarenessoftheprime.

Ontrialsinwhichthepresentationsequencewasthesameasinthesemanticcategorizationtask,participantscorrectlydetectedprimewordswiththeletter‘‘e’’51.5%ofthetime.Theirperformancedidnotdi erfromchance(F<1).Ontrialsinwhichtheprimewaspre-sentedfor10refreshcycles,participantscorrectlyiden-ti edprimewordswiththeletter‘‘e’’87.8%ofthetime.Takentogether,these ndingsindicate:(1)thatpartici-pantswereperformingthe‘‘e-detection’’tasktothebestoftheirabilitiesand(2)thatonthetrialsinwhichprimedurationmatchedtheprimedurationusedinthese-manticcategorizationtask(41ms),participantsweregenerallynotabletodetecttheprime.Assuch,these ndingsruleoutawarenessasapossiblesourceofthepriminge ectobservedinExperiment5a.

Experiment5b—many-to-oneinsemanticcategorizationAlthoughExperiment5acon rmsthetaskdi erenceintheone-to-manydirection,and,hence,providesstrongsupportfortheassumptionsoftheSenseModel,ouraccountofthetaskdi erencepredictssymmetricalpriminginsemanticcategorization.Forthesakeofcompleteness,then,weconductedExperiment5btocon rmthatmany-to-oneprimingoccursbothinlexicaldecisionandsemanticcategorization.Inthisexperi-ment,meanresponsetimeswere532msfortargets(e.g.,SKULL)precededbyrelatedmany-senseprimes(e.g.,head)and552msforthesametargetswhenprecededbyacontrolprime.This20mse ectwasfoundtobesig-ni cantinboththesubjectanditemsanalyses(F1ð1;18Þ¼7:28,P¼:01;F2ð1;40Þ¼7:67,P¼:008).Importantly,thise ectdoesnotdi erfromthe23mspriminge ectfoundinthemany-to-onedirectioninlexicaldecision(Experiment4a).Again,theseresultscon rmthepredictionsoftheSenseModel.AccordingtotheSenseModel,thereshouldbenotaskdi erencewhentargetshaveveryfewsenses.ThisisbecausetheproportionofthetargetÕssensesactivatedbytheprimewillbeveryhigh,whichwillfacilitatedecisionsinbothlexicaldecisionandsemanticcategorization.Inotherwords,the‘‘focusinge ect’’ofsemanticcategorizationdoesnotprovideanybene twhenthetargethasonlyonesensetobeginwith.

ThemeanerrorrateinExperiment5bwas8.5%.Again,therewasadi erencebetweenexperimentalandcontroltrials.Ontrialsinwhichthetargetwasprecededbyarelatedprime,theerrorratewas6.02%;theerrorrateonthecontroltrialswas10.93%.Thisdi erencewasveryclosetoreachingsigni canceinthesubjectanalysisanddidreachsigni canceintheitemsanalysis(F1ð1;18Þ¼3:82,P¼:07;F2ð1;40Þ¼5:04,P¼:03).Again,thisdi erencesuggeststhatparticipantswereabletobene tfromtheprimebothintermsofreactiontimesandaccuracy.Onepossibleinterpretationis,assuggestedabove,participantswereinitiatingaresponsebasedonthecategorymembershipoftheprime.Ifthis

14M.Finkbeineretal./JournalofMemoryandLanguage51(2004)1–22

wasthecase,thenthefacilitatione ectwouldbeacongruencye ect,notapriminge ect.Again,thisisapossibilitythatweaddressinExperiment6.

Turningnowtotheprimeawarenesstask,thepar-ticipantsinExperiment5bdidnotdoanybetterthanthoseinExperiment5a.Themeanaccuracyrateontrialswitha41msprimeduration(asinthesemanticcategorizationtask)was53.2%.Thisdidnotdi ersig-ni cantlyfromchance(F¼2:5;p¼:11).Participantsdidmuchbetterontrialswheretheprimewaspresentedfor10refreshcycles(137.3ms),correctlyidentifyingtheprimewordscontainingtheletter‘‘e’’83%ofthetime.Again,thispatternof ndingsindicatesveryclearlythat:(1)participantsweretryingtodothistaskand(2)thatdespitetheire orts,theywerenotabletoperceivetheprimewhenthepresentationsequencewasidenticalinitstemporalcharacteristicstothatusedinthesemanticcategorizationtask.Followingfromthis,itisunlikelythattheprimingexhibitedbythesesameparticipantsinthesemanticcategorizationtaskcouldhavebeenduetotheirawarenessoftheprimeÕspresence.

Experiment5hascontributedinanimportantwaytoourunderstandingofthetaskdi erencebetweense-manticcategorizationandlexicaldecisionintranslationpriming.Itisclearfromthisexperimentthatsemanticprimingoccursequallyinboththeone-to-manyandthemany-to-onedirection,despitetherepresentationalasymmetrybetweenprimeandtarget.Thisisquitedif-ferentfromtheresultsseeninlexicaldecision.Inlexicaldecision,primingonlyoccursinthemany-to-onedi-rection.Althoughthispatternof ndings tsverywellwiththepredictionsoftheSenseModel,thereisapossiblealternativeexplanationforthetaskdi erence,whichweaddressnow.Thisisthepossibilityofadeci-sioncon ict.

ThispossibilitywaspointedoutbyDavis,Kim,andSanchez-Casas(2003)intheircritiqueoftheoriginal ndingreportedbyGraingerandFrenck-Mestre(1998).TheyarguedthatthecontrolconditionusedintheGraingerandFrenck-Mestre(1998)experiment(thesameasinourExperiments1and5)mayhavebeeninappropriate.Forthecriticalitems,theunrelatedprimewasanon-exemplar,whilethetargetwasanexemplar.Thismayhavegeneratedadecisioncon ictforunre-lateditems,butnotforrelateditems(sincebothprimeandtargetwereexemplars),whichwouldproduceacongruencee ectintheplaceofatruepriminge ect.Previousattemptstodiscoversuchcongruencee ects,however,havenotbeensuccessful.Inarecentexperi-ment,Damian(2001)foundthatdecisioncon ictsinasizejudgmenttaskaroseonlywhenprimeshadprevi-ouslyappearedastargets.Damianinterpretedthose ndingstomeanthatdecisioncon icts,whentheyar-ose,weretheresultofautomatizedstimulus–responsemappings,notunconsciouscategorizationoftheprime.Becausenoneoftheprimesinthepresentsetof

experimentseverappearedastargets,thereisnoreasontobelievethatourparticipantseverdevelopedsuchautomatizedstimulus–responsemappings.Inaseparateinvestigationofthesameissue,Forsteretal.(2003),usingthecategoryANIMAL,foundnodi erencebe-tweencongruentpairs(e.g.,‘‘shark–ROBIN’’)andin-congruentpairs(e.g.,‘‘badge–ROBIN’’),suggestingthatcongruencee ectscouldsafelybeignored.However,thiswastrueonlyforexemplars.Fornon-exemplars,acongruencee ect(15ms)wasdetected.SimilarresultsforexemplarswereobtainedbyBuenoandFrenck-Mestre(2002),whoreportedsimilarpriminge ectsfor‘‘dolphin–WHALE’’whenthecontrolconditionusednon-exemplarsasprimes(e.g.,‘‘helmet–WHALE’’)andwhenthecontrolconditionusedcongruent,butse-manticallydissimilar,exemplarsasprimes(e.g.,‘‘spar-row–WHALE’’).Sinceneitheroftheseexperimentsreportevidenceofacongruencee ectforexemplars,adecisioncon ictaccountforthepriminge ectsseeninExperiments1and5seemsveryunlikely.However,re-centevidence(Forster,inpress)indicatesthatverysubstantialcongruencee ectscanbegeneratedforbothexemplarsandnon-exemplarswhenasmallcategoryisused(e.g.,farmanimal),butnotwhenalargecategoryisused(e.g.,animal).SincesomeofthecategoriesusedinExperiments1and5mightbedescribedas‘‘small,’’itseemedwisetoeliminatethispossibleexplanationofthepriminge ect.Accordingly,weaddressthispossibilityinExperiment6byconductingareplicationofExperi-ment5.Theonlydi erencebetweenthetwoexperimentswasthatinExperiment6bothexperimentalandcontrolprimessharedexemplarstatuswiththetarget.

Experiment6—semanticcategorizationwithexemplarsascontrolprimesParticipants

Twenty-fourundergraduatestudentsfromtheUni-versityofArizonaparticipatedforcoursecredit.Materials

ThematerialswereidenticaltothoseusedinExper-iment5exceptforthecontrolprimesonthe42criticaltrials.Thesewerereplacedsothatbothprimeandtargetwereexemplarsofthesamecategory.Inselectingthenewcontrolitems,carewastakentoensurethatthecontrolprimesweresemanticallydistantfromthetar-gets.Forexample,intheBODYPARTcategory,thetarget‘‘HAND’’waspairedwiththeexperimentalprime‘‘wrist’’andthecontrolprime‘‘kidney.’’Ifparticipantsaregeneratingaresponsebasedonthecategorymem-bershipoftheprimeword,thenthereshouldbenodi erenceinresponsetimesbetweenexperimentaland

M.Finkbeineretal./JournalofMemoryandLanguage51(2004)1–2215

controltrials.If,though,theprimestimulusisprimingthetargetbyactivatingasemanticsenseincommonbetweentheprimeandtarget,participantsshouldbesigni cantlyfasterontheexperimentaltrials.Designandprocedure

TheexperimentaldesignandprocedurewasidenticaltothatofExperiment5,exceptthatweonlytestedinonedirection.Becausethesourceofthetaskdi erencebetweenlexicaldecisionandsemanticcategorizationintheone-to-manydirectionisthecriticalquestionthatthisexperimentaddresses,itwasonlynecessarytotestinthisdirection.Asigni cantpriminge ectintheone-to-manydirectionwhenusingexemplarsascontrolprimeswouldallowustoruleoutthepossibilityofadecisioncon ict.

Resultsanddiscussion

Thesametrimmingproceduresusedinthepreviousexperimentswereusedagainhere.Contrarytothepredictionsofthedecisioncon ictaccount,semanticprimingintheone-to-manydirectionwasnota ectedbythepresenceofcongruent,butsemanticallydissim-ilar,exemplarsasprimes.Meanresponsetimefortar-gets(e.g.,‘‘HAND’’)precededbyexperimentalprimes(e.g.,‘‘wrist’’)was463ms;andmeanresponsetimeforthesametargetsprecededbycontrolprimes(e.g.,‘‘kidney’’)was475ms.Thispriminge ectwasfoundtobereliableinboththesubjectanditemsanalyses(F1ð1;22Þ¼5:79,P¼:02;F2ð1;40Þ¼6:25,P¼:01).These ndingscon rmthosethathavebeenreportedonatleasttwooccasionsnow(Bueno&Frenck-Mes-tre,2002;Forsteretal.,2003).Thatis,maskedpriminge ectsforexemplarsinsemanticcategorizationaresimilarinmagnituderegardlessofwhetherthecontrolconditionincludescategorycongruentitemsornot.Assuch,wecanconcludethatthetaskdi erencebetweenlexicaldecisionandsemanticcategorizationintheone-to-manydirection(andbyextension,intheL2–L1direction)isnotattributabletoadecisioncon ictinsemanticcategorization.Rather,these ndingssuggestthatthetaskdi erenceisduetothefactthatthese-manticcategorizationtaskservestofocushowdeci-sionsaremade,sothatonlythesemanticsensethatisrelevanttothecategoryistakenintoaccount.Taking‘‘hand’’asanexample,onedoesnotneedtotakeintoconsiderationthemultiplesensesthat‘‘hand’’mayhaveinordertodecidewhetheritisabodypartornot.Essentially,wearguethatthesemanticcategorizationtaskturnsmany-sensetargetsintoone-sensetargetsby‘‘ ltering’’outcategory-irrelevantsensesfromthede-cisionmakingprocess,andthatthisenhancesthee -cacyoftheone-senseprime.ThisissuewillbeconsideredagainintheGeneraldiscussion.

Generaldiscussion

Theresearchhereestablishesseveral ndingsthat,whentakentogether,presentanimportantchallengetomodelsofbilinguallexicalrepresentationandprocess-ing.These ndingsare:(1)asymmetricaltranslationpriminginlexicaldecision,(2)symmetricaltranslationpriminginsemanticcategorization,(3)robustwithin-L2maskedpriminge ects,and(4)thesamepatternof ndingsforwithin-languagewordpairsthathavethesamerepresentationalasymmetryassumedbetweentranslationequivalents(seeTable2).Aswehaveseen,modelsofbilinguallexicalprocessingthatproposelim-itingfactorssuchasweakL2form–meaningconnectionstrengths(e.g.,TheRevisedHierarchicalModel),orthattheL2lexiconmaybestoredinaseparatememorymodulefromtheL1lexicon(asintheseparatememorysystemsaccount),cannotaccountforthese ndings.Inwhatfollows,wediscussinmoredetailhowtheSenseModelprovidesaparsimoniousexplanationforeachofthese ndingsinturn.

MaskedtranslationprimingasymmetryinlexicaldecisionTheaccountoftheprimingasymmetrygivenbytheSenseModelfollowsdirectlyfromtheassumptionsthatitmakeswithrespecttohowlexicalsemanticinforma-tionisrepresented.AswementionedinthediscussionofExperiment3,theSenseModel,muchlikethedistrib-utedconceptualfeaturemodel(DCFM)(deGroot,1992;deGroot,Dannenburg,&vanHell,1994;Kroll&deGroot,1997),holdsthatlexical(form-level)repre-sentationsmapontodistributedlexicalsemanticrepre-sentations,asopposedtolocalistconceptualrepresentations.UnliketheDCFM,though,acentralassumptionoftheSenseModelisthatsemanticfeaturesarebundledintosemanticsenseswithindistinctlexicalsemanticrepresentations.Followingfromthis,theSenseModelassumesthatmaskedtranslation(andsemantic)primingisattributednotonlytotheoverlappingse-manticfeaturesbetweenprimeandtarget,butalsototheratioofprimedtounprimedsenses.Whenthese-manticfeaturesincommonbetweenprimeandtargetconstitutealargeproportionofthetargetÕslexicalse-manticrepresentation,primingshouldbeobserved.Ontheotherhand,whenthesemanticfeaturesincommonbetweenprimeandtargetconstituteonlyasmallpro-portionofthetargetÕslexicalsemanticrepresentation,theamountofprimingmaybetoosmalltoa ectdeci-sionmakinginanysigni cantway.

Whyshouldweassumethatlexicaldecisionissen-sitivetotheproportionofsemanticsensesactivatedforagiventarget?SomesupportforthisassumptioncomesfromrecentworkbyRoddetal.(2002).Roddetal.(2002)reportthatthecontroversialclaimthatthereisanambiguityadvantage(i.e.,thatwordswithmany

16M.Finkbeineretal./JournalofMemoryandLanguage51(2004)1–22

Table2

Comparisonofpriminge ectsanderrorrates(inparenthesis)forall-Englishprime–targetpairsbydirection(many-to-onevs.one-to-many)andbytask(lexicaldecisionvs.semanticcategorization)inExperiments4–6

PrimeconditionPrime

LexicaldecisiontaskSemanticcategorizationtaskSemanticcategorizationtaskwithexemplarsascontrolprimes

Many-to-one(Experiment4a)One-to-many(Experiment4b)One-to-many(Experiment5a)Many-to-one(Experiment5b)One-to-many(Experiment6)

550538473532463

(5.70)(5.1)(3.3)(6.0)(3.3)

Control573528486552475

(5.75)(6.2)(7.5)(10.9)(5.5)

23)10132012Priminge ect

meaningsarerespondedtofasterthanunambiguouswords)isactuallyasenseadvantage.Theyfoundthatlexicaldecisiontimeswereactuallyslowerforambigu-ouswords(wordswithdistinctmeanings,like‘‘bank’’)thanforunambiguouswords(thoughnotsigni cantlyslower),whereasdecisiontimesforwordswithmanysenses(e.g.,‘‘hammer’’)weresigni cantlyfasterthanforwordswithfewsenses(e.g.,‘‘cake’’).Allwordswerematchedforfrequency,length,familiarityandcon-creteness.Similar ndingswereobtainedbyFinkbeiner(2002)usingalargerandrandomlygeneratedsetofitems.5ThisinvolvedusingWordNet(Fellbaum,1998)togeneratelistsoffew-sensewords(nomorethantwosenses)andmany-sensewords(15ormoresenses).Eightyitemsappearinginafrequencybandof20–80permillionwerethenrandomlyselectedfromeachlistofpossiblewordsforatotalof160experimentalitems(itemshadameanfrequencyof45.8andameanlengthof5.1lettersoneachlist).Usingtheseitemsinalexicaldecisiontask,areliabledi erencebetweenresponsetimesformany-sensewords(533ms)andfew-sensewords(564ms)wasfound(F1ð1;14Þ¼14:44,P¼:001;F2ð1;78Þ¼36:54,P<:001).Takentogether,these ndingssuggestthatlexicaldecisiontimesaresensitivetothenumberofsemanticsensesassociatedwithaparticulartarget.Itlogicallyfollowsfromthisthatinordertoshowpriming,alargeproportionofatargetwordÕssemanticsensesmustbepreactivatedbytheprime.

Insummary,themaskedtranslationprimingasym-metryisarguedtobethestraightforwardconsequenceofarepresentationalasymmetrybetweenL1andL2lexico-semanticrepresentations.L1wordsarearguedtohaverelativelymoresensesassociatedwiththemthanL2words,andtranslation-equivalentwordspresumablyonlysharearestrictednumberofsenses.Followingfromthis,itisarguedthattheproportionofL1sensesprimedbyanL2equivalentisgoingtobelowerthanthepro-

portionofL2sensesprimedbyanL1equivalent,whichleadstotheoftenobservedtranslationprimingasym-metryinlexicaldecision.

Taskdi erencesinmaskedtranslationpriming

Wereturnnowtothequestionofhowsemanticcategorizationmighteliminatetheprimingasymmetry.IfitiscorrecttoassumethatL2–L1primingdoesnotoccur(orisveryweak)becauseaninsu cientnumberofL1sensesarepreactivatedbytheL2prime,thentheremustbesomethingaboutthesemanticcategorizationtaskthatservestorestrictthenumberofsensesrecruitedwhenadecisionisgenerated.Essentially,wearguethatthesemanticcategorizationtaskturnsmany-sensetar-getsintoone-sensetargetsby‘‘ ltering’’outcategory-irrelevantsensesfromthedecisionmakingprocess.Thatis,insemanticcategorization,thesensethattriggersadecisionwillalwaysbethesensethatisrelevanttothecategory,andthisallowstheL2primetobemoreef-fective.Assumingthatthematerialshavebeendesignedappropriately,thecategory-relevantsensewillbethesensethatboththeL2primeandtheL1targetshare,meaningthatthesensethattriggersadecisionwillal-readyhavebeenactivatedbytheL2prime.

Thisaccountraisesseveralinterestingquestions.First,howisthecategoryabletoselectouttherelevantsense?Itisinterestingtonotethatasimilarphenomenonapparentlyoccursinthecontextofneighborhoode ects(Forster&Hector,2002).Theseinvestigatorsfoundthatinasemanticcategorizationtask(e.g.,‘‘Isitanani-mal?’’),non-wordsthatwereone-letterdi erentfrommanywords(e.g.,walley)wereclassi edjustasrapidlyasnon-wordswithfewneighbors(e.g.,braln),unlessoneofthoseneighborshappenedtobeananimalname(e.g.,turple),inwhichcaseasubstantialinhibitorye ectwasobserved.Evidently,aspellingcheckwastriggeredforturplebecauseofitssimilaritytoaword(whichhap-penedtobeanexemplar),butnotforanon-wordsuchascishop,despiteitssimilaritytoaword.Althoughthisishardlyasurprisingresult,itdoesraisethequestionofhowthedecisionsystemknowswhethertoignorethepresenceofawordneighbor.Indeed,non-wordsthathad

Forster(2000)hasarguedthatinthiskindofresearch,itisimperativethatitemsareselectedrandomlytoavoidinadver-tentexperimenterbias.

5

M.Finkbeineretal./JournalofMemoryandLanguage51(2004)1–2217

alargenumberofneighbors,oneofwhichwasananimal(e.g.,goan)tooknolongerthannon-wordswhichalsohadananimalneighbor,butveryfewotherneighbors(e.g.,cadel).Thepuzzlehereistoexplainhowtheexis-tenceofananimalneighborcouldbedetectedwithoutapplyingsomekindofsemantictesttoeachneighbor.Somehowthecategory-irrelevantneighborswere lteredout,leavingjustthecategory-relevantneighbor.Weproposethatasimilarmechanismoperatesinsense-se-lection;category-irrelevantsensesareautomatically l-teredout,leavingjustthecategory-relevantsense.

Althoughweprefertheexplanationprovidedbythe lterhypothesis,thereareotherinterpretationsthatneedtobeconsidered.AccordingtotheSenseModel,thelocusofthepriminge ectisatthelevelofdecisionmaking.Whenthesemanticsenseofthetargetresponsiblefortriggeringasemanticjudgmenthasalreadybeenactivatedbytheprime,lesstimeisneededtogenerateadecision,leadingtoasavings(orpriming)e ect.Accordingtothisaccount,then,thecriticalstageintheprocessoccurswellaftertheform-levelpropertiesofthetargethavebeensuccessfullyretrieved.Butthisisclearlynottheonlypointintheoverallprocessatwhichasavingscouldoccur.Forexample,primingmaybeattributedtofacilitationintheinitialstageofwordformrecognition.Alternatively,itmaybethatthecriticalstageintheprimingprocessin-volvesadegreeofsimilarityinthecategorizationpro-cessesofboththeprimeandtarget.Belowweconsiderthesealternativesinmoredetail.

The rstalternativeexplanationsuggeststhatwhatisprimedinthesemanticcategorizationtaskisnotase-manticsense,butrather,theform-levelpropertiesofthetarget.Accordingtothisaccount,primingoccurswhentheprocessesinvolvedininterpretingtheprimealterthestateofthelexicalentryforthetarget,sothatthetar-getÕswordformisrecognizedmorerapidly.Inorderforthistooccur,atop-downprocessisnecessary,whereactivation owsfromthesemanticlevelbackdowntotheformlevel(e.g.,Besner&Smith,1992).Thisaccountessentiallyassumesthatprimingisaperceptuale ect,i.e.,thatasemanticallyrelatedprimefacilitatesrecog-nitionofthetargetÕsform.However,ifthiswerethecase,thenitwouldbedi culttoexplainwhythereisnoL2–L1priminginlexicaldecision,butthereisinse-manticcategorization.Fasterperceptionofthetargetshouldleadtofasterresponses,whateverthetask.

Anotherpossibleexplanationisthatwhatgetsprimedisthecategorizationprocessitself.Forexample,itmightbethatacategorizationprocessisinitiatedfortheprimeaswellasthetarget.Whenthetargetissemanticallyverysimilartotheprime,itmaymeanthatlesse ortisinvolvedinreachingadecisionconcerningthetarget.So,inthecaseofanitemsuchas‘‘whale–DOLPHIN,’’itcouldbethathavingjustcomputedthemeaningof‘‘whale,’’andclas-sifyingitasananimal,itismucheasiertodecidethat‘‘dolphin’’mustalsobeananimal,becausetheprocesses

involvedinclassifyingthesewordsareverysimilar.Thisanalysisisdi erentfromthesuggestionthatprimingisthesimpleconsequenceofasequenceoftwoYES–YESre-sponses(i.e.,acongruencee ect),onetotheprimeandonetothetarget,and,hence,couldaccountforthe nd-ingspresentedinExperiment6,wherewefoundrobustprimingfor‘‘wrist–HAND’’(thecategorybeingBodyPart)evenwhenthecontrolprimewouldalsohaveledtoaYESresponse(‘‘kidney–HAND’’).Accordingtothisanalysis,primingisobservedinthe‘‘wrist–HAND’’(orL2–L1translation)conditionbecausenearidentityofreferenceofthewordspermitsfastersemanticcategori-zation.Thiswouldnotbetrueforthewords‘‘kidney–HAND’’inthecontrolcondition.Inessence,whenboththeprimeandtargetsharemanyofthesameconceptualfeatures,classifyingthetargetisgoingtobefaster.Theappropriateanalogyhereiswithasyllogismoftheform‘‘IfXisamemberofacategory,andYisverysimilartoX,thenitislikelythatYwillalsobeamember.’’Arguably,thesimilarityofreferenceisnotrelevantinlexicaldecisionbecausewhatiscriticalinthattaskiswhetherthestimulusisacorrectlyspelledword.This,then,capturesanim-portantaspectofthetaskdi erencethatweobservedbetweensemanticcategorizationandlexicaldecision.Butthereisalsogoodreasontoquestionthisanalysis.Ifthesimilarityinthecategorizationprocessbetweenprimeandtargetisthesourceofatimesavingsforexemplars,itshouldalsoleadtoasavingsfornon-exemplars.6But,aswesawinExperiment3,thiswasnotthecase.Translationprimingonlyoccurredforexemplars.Also,thisanalysismaybetoostronginitssuggestionthatthesimilarityofreferenceisnotrelevantinlexicaldecisionbecauseweknowthatmaskedL1–L2translationprimingoccursinlexicaldecision(seealsoExperiment4a).

Asnotedearlier,thereisanothertaskthatyieldsL2–L1priming,namelyepisodicrecognition(Jiang&For-ster,2001).Inthisstudy,thereweretwophasestotheexperiment.Inthe rstphase,participantswereaskedtostudyalistofL1words.Inthesecondphase,ingthisspeeded‘‘old–new’’task,JiangandForster(2001)reportedsigni cantmaskedtranslationpriminge ectsintheL2–L1directionforold,butnotnew,targets.Howcanthise ectbereconciledwiththeSenseModelaccounto eredhere?Inanold–newtask,theresponseisclearlydeterminedbywhethertheL1targetstimulusmatchestheepisodicre-cordoftheearlierpresentationofthesameword.Sincesemanticfactorswillberelevantintheencodingofthe

6

Thisdependsonthesyllogism.Suchapredictionisvalidifitisoftheform‘‘IfXisnotamemberofcategory,thenitislikelythatYisalsonotamember,ifXandYareverysimilarobjects.’’

18M.Finkbeineretal./JournalofMemoryandLanguage51(2004)1–22

stimulusduringthestudyphase,itisrelevanttoaskwhethertheepisodicrecordincludesallofthesensesassociatedwiththeto-be-remembered(TBR)word,orjustthedominantsense.Previousworkinvestigatinghowwordsareencodedintoepisodicmemoryhasindicatedsupportforanencodingspeci cityprinciple(Tulving&Thomson,1973),whichassumesthatonlyasinglesenseoftheTBRwordisencodedduringthestudyphase(thisbeingdeterminedbycontext).InanuncuedlearningconditionsuchastheoneusedbyJiangandForster(2001),itisreasonabletoassumethatparticipantswouldencodeonlythedominantsenseoftheTBRwordduringthestudyphase.SinceitisfrequentlythecasethatL1andL2translationequivalentssharethedominantsenseoftheL1word,thismeansthatthemaskedL2primecouldhavefacilitateddecisionlatenciesontheL1target.Inthetestphaseoftheexperiment,decisionspresumablyarenotmadeonthelexicalstatusofthetargets,but,rather,onwhetherornotthetargetcancuearecollectionofaTBRword.If,asTulvingandThomson(1973)haveargued,theTBRwordislimitedinitsmeaningtoaparticularsense,whichwehavearguedisfrequentlytheverysensethattheL2primeiscapableofactivating,thenweshouldnotbesurprisedtoobservepriminginthistask.Hence,theSenseModelisabletoprovideanex-planationfortheepisodicprimingresultsreportedbyJiangandForster(2001).

Therearetwofurtherissuesthatdeserveattentionbeforemovingontothe nalsection.The rstoftheseisthesubsetmodelproposedbyDufourandKroll(1995),whichispotentiallyatoddswiththeassumptionsoftheSenseModel.Theseresearchersfoundthatless uentbilingualswereabletocategorizeL2targetsfasterwhenthecategory(e.g.,COLOR)wasgiveninL2comparedtowhenitwasgiveninL1.DufourandKroll(1995)arguedthatduringtheearlieststagesofL2acquisition,L2wordsareassociateddirectlywithasmallbutwell-de nedsetofconceptualrepresentations,essentiallyasubsetofthecategoryknowledgeinL1.Followingfromthis,theysuggestedthatwhenbothcuesandtargetsweregiveninL2,onlyexemplarsbelongingtothissubsetofconceptualrepresentationswereactivated.Consequently,therewaslessinterferencefromrelatedconceptsrelativetowhenL1cueswereusedand,hence,facilitation.Accordingtotheirmodel,L1formsareas-sociatedwithawiderangeofcompetingconcepts,allofwhichbecomeactivewhenevertheL1formisencoun-tered.Theconsequenceofthiswidespreadactivation,accordingtotheirargument,isthatit‘‘...mayactuallyinhibittheretrievalofconceptsthatare(otherwise)ac-cessiblefromL2,becausealargenumberofconceptsthatareunknowninL2willalsobeactivated’’(p.176).The ndingsreportedinthepresentarticledirectlycontradictthisconclusion.Forexample,therobustL2–L1priminge ectsobservedinExperiment1demon-stratethatcategorycuespresentedinL1donotpreventsuccessfulprocessingoftheL2prime.Furthermore,byassumingthatL1cuesinterferewithL2processing,itisnotclearhowthesubsetmodelcouldaccountfortherobustpriminge ectsfrequentlyreportedintheL1–L2direction(seeIntroduction),whichindicatethatactiva-tioncausedbytheL1primefacilitatesL2processing.Howthencantheinterferencee ectsreportedbyDufourandKroll(1995)bereconciledwiththefacili-tatione ectsreportedhere?Wefeelthatthedi erencebetweenthetwo ndingsmaybeaccountedforbya‘‘switchcost’’presentintheDufourandKrollstudybutnotinours(cf.Meuter&Allport,1999).Inbothstudies,participantswerepresentedwithacategoryexemplar(e.g.,ANIMAL),which,importantly,mayalsoserveasalanguagecue.IntheDufourandKrollstudy,partic-ipantsmayhavebeenslowestoveralltocategorizeL2targetswhencategorycueswereinL1becausetheyhadtosuppresstheirjust-activatedL1inordertocategorizetheL2target.ThiswouldnothavebeennecessarywhenbothcategorycueandtargetweregiveninL2.Inourstudy,no‘‘switchcost’’wasobservedbecause(a)bothcategorycueandtargetwereinL1and(b)becauseparticipantswereunawareoftheL2prime.Crucially,though,the ndingsofthepresentstudy,whichcon rmthosereportedearlierbyGraingerandFrenck-Mestre(1998),makeitclearthatcategorycuespresentedinL1donotinterferewithprocessingoftheL2prime.

Onefurtherissuethatrequiressomediscussioncon-cernsthefactthatwehaveusedaninterpolatedmaskbetweentheprimeandtargetineachofthesemanticcategorizationexperiments(ashaveotherinvestigators,e.g.,Bueno&Frenck-Mestre,2002;Frenck-Mestre&Bueno,1999;Grainger&Frenck-Mestre,1998),whereasthisisnotnormallydoneinlexicaldecisionexperiments(e.g.,Experiments4aandb).Thisraisesthepossibilitythatitistheinterpolatedmaskthatproducessymmetryofpriming,notthetask.However,thisisnotthecase.InExperiment2,thetaskwaslexicaldecisionwithL2–L1translationpriming,andaninterpolatedmaskwasused(asinExperiment1),yetnoprimingwasobtained.Also,Jiang(1999)attemptedtoobtainL2–L1priminginalexicaldecisiontaskbyinterpolatingamask,butdidnotsucceed.

Nevertheless,itisworthconsideringwhysuchaprocedureisusedinthesemanticcategorizationtask(cf.Bueno&Frenck-Mestre,2002;Frenck-Mestre&Bueno,1999;Grainger&Frenck-Mestre,1998;andtheexperimentsreportedhere).Themostobviouspossi-bilityisthatreliablepriminginthistaskrequirestheinsertionofamask.SomesupportforthispossibilityisprovidedbythefactthatapilotversionofExperiment5accidentallyomittedthemask,andfailedtoproduceanypriming.Clearly,furtherworkisrequiredtoes-tablishwhetherthemaskisreallyrequired,butitisneverthelessofinteresttoconsiderwhyitmightberelevant.Theoriginalpurposeofincludingamaskwas

togivetheprocessingofL2primesmoreofaheadstartovertheL1targetsinalexicaldecisiontask(e.g.,Jiang,1999).Thefactthatthisprocedurefailedtoproduceprimingsuggestedthatthemaskwasirrelevant,butwecannowsaythatthisfailurewasprobablyduetothenatureofthelexicaldecisiontask.Thus,itmightbethecasethatinasemanticcategorizationtask,extratimefortheprocessingoftheprimeisrequiredwhentheoverlapbetweentheprimeandtargetispurelysemanticorconceptualinnature.

Within-L2maskedrepetitionpriming

The nalpointthatweconsiderherehastodowithwithin-L2maskedrepetitionpriming,whichhasbeenobservedtobequiterobust(e.g.,Gollanetal.,1997;Ji-ang,1999;andExperiment2inthepresentstudy).OneofthestrengthsoftheSenseModelliesinitsabilitytopro-videasingleaccountofthemaskedtranslationprimingasymmetryandwithin-L2repetitionpriming.Accordingtothisaccount,therearenoweakconnectionsandsothemaskedL2stimulusservestoactivateitscorrespondingL2lexicalrepresentation,whichservestoactivateitslexicalsemanticrepresentation.Aswehavealreadyar-gued,whenthetargetsareinL1,theproportionofthetargetÕslexicalsemanticrepresentationactivatedbytheL2primemaybeinsu cienttoproduceanoticeablepriminge ect.However,whenthetargetsarerepetitionsoftheL2prime,thetarget,byvirtueofsharingthesameorthographicandlexicalsemanticrepresentations,willalreadyhavebeenpreactivatedbytheprime.Thisshouldleadtoveryrobustpriminge ects—anditdoes.

ternof ndingswasinterpretedtosuggestthatpreviousaccountsoftheasymmetry,whichhaveproposedthatweakL2form–meaningconnectionsconstitutealimitingfactorintranslationpriming,cannotbecorrect.Alter-natively,wehaveproposedtheSenseModel.ThismodelattributesthetranslationprimingasymmetryinlexicaldecisiontothenaturalconsequenceofarepresentationalasymmetrybetweenL1andL2lexicalsemanticrepre-sentationsandclaimsthatsemanticcategorizationin-troducesa lteringprocess,whichservestoeliminatetherepresentationalasymmetry.Wetestedtheseclaimsdirectlybyrecreatingtheassumedrepresentationalasymmetrywithwithin-languagewordpairs(e.g.,‘‘skull–HEAD’’),inwhichonehadmanysenses(HEAD)andonehadonlyoneortwosenses(skull).Usingthesewordpairs,wereproducedthesamepatternofprimingfoundbe-tweentranslationequivalents.Thatis,many-sensewordsprimedfew-sensewordsinlexicaldecisionandsemanticcategorization,butthesamewordpairsinthefew-sensetomany-sensedirectionproducedprimingonlyinsemanticcategorization,notlexicaldecision.Theseresults,paral-lelingdirectlythoseobservedintranslationpriming,providestrongsupportfortheassumptionsoftheSenseModel.

Acknowledgments

Theauthorsgratefullyacknowledgesupportforthisresearchfromanumberofsources,includingtheFacultySmallGrantsProgram(O ceoftheVicePresidentforResearchandGraduateStudies,inconjunctionwiththeUniversityofArizonaFoundation),GrantNIDCDDC-01409(aspartoftheNationalCenterforNeurogenicCommunicationDisorders,UniversityofArizona),andtheCognitiveScienceProgram,UniversityofArizona.WewouldalsoliketothankJonathanGrainger,JudithKroll,TamarGollan,MerrillGarrett,andAlbertCostaaswellastwoanonymousreviewersforhelpfulcom-mentsandsuggestionsonanearlierdraftofthispaper.

Conclusion

Theresultsreportedinthepresentpaperprovideevi-denceofataskdi erenceinmaskedtranslationprimingintheL2–L1direction,withprimingbeingobservedinse-manticcategorizationbutnotlexicaldecision.Thispat-

AppendixAExperimentalitemsusedinExperiment1a

Partofabuildingroof

ArelativePartofthehumanbodyAkindofmetal

20M.Finkbeineretal./JournalofMemoryandLanguage51(2004)1–22

AppendixA(continued)

Colorblack

Unitoftime

Readingmaterial

Ascience

AppendixB.ExperimentalitemsusedinExperiment4AppendixB(continued)

Onesense

ManysenseSTONEDOCKARMFOOTLONGDOGCATWINDBAG

STREAMFLYFASTMean

#ofsensesformany-sensewords

10118121579161510201511.25

Onesenseoxygenmoistmoviesofasurgeonoartombviolinclergyelderlywealthycoyotelanternhugelawnwheatlungcottagedairypovertysaucepan stpistolprincebeerfabriclawyertiny

ManysenseAIRWETFILMCHAIRDOCTORPADDLEGRAVEBASS

MINISTERAGEDRICHFOXLIGHTLARGEGRASSGRAINHEARTHOUSEMILKPOORPOTHANDGUNKINGDRINK

MATERIALJUDGESMALL

#ofsensesformany-sensewords

2087771089610129301210111013791115861112717

pebblemarinawristanklelengthypuppykittengalesuitcasecreekmosquitorapidly

AppendixC.ExperimentalitemsusedinExperiment5

Many-sensewordsdressslacksskirtcoatjacketchairtable

#ofsenses

191276767

Few-sensewordsgowntrouserskiltparka

windbreakerreclinerdesk

#ofsenses

4111111

M.Finkbeineretal./JournalofMemoryandLanguage51(2004)1–2221

AppendixC(continued)

Many-sense#ofsenses

Few-sense#ofsenses

wordswordscouch4sofa1bed9futon1dresser5cupboard1dove8pigeon1eagle5osprey1crow9blackbird2canary5sparrow1saw25lathe1hammer10mallet3drill9awl1 le9chisel1wrench7pliers1red9crimson2blue19lavender2green15teal2purple8violet3black20ebony2dog7puppy2cat9kitten2horse7zebra1cow4bison1wolf6coyote1pig9boar2bear15grizzly1knife4cleaver1spoon5ladle2cup11goblet1pan7skillet1plate15platter1head40skull1legs9thigh2hand15wrist1foot12ankle1eye6retina1nose

13

nostril

1

References

Balota,D.A.,&Chumbley,J.I.(1984).Arelexicaldecisionsa

goodmeasureoflexicalaccess.Theroleofwordfrequencyintheneglecteddecisionstage.JournalofExperimentalPsychology:HumanPerceptionandPerformance,10,340–357.

Battig,W.F.,&Montague,W.E.(1969).Categorynormsfor

verbalitemsin56categories:AreplicationandextensionoftheConnecticutcategorynorms.JournalofExperimentalPsychology,80,1–46.

Besner,D.,&Smith,M.C.(1992).Modelsofvisualword

recognition:Whenobscuringthestimulusyieldsaclearerview.JournalofExperimentalPsychology:Learning,Mem-ory,andCognition,18,468–482.

Bodner,G.,&Masson,M.(2001).Primevaliditya ects

maskedrepetitionpriming:Evidenceforanepisodicresourceaccountofpriming.JournalofMemoryandLanguage,45,616–647.

Bueno,S.,&Frenck-Mestre,C.(2002).Rapidactivationofthelexicon:Afurtherinvestigationwithbehavioralandcom-putationalresults.BrainandLanguage,SpecialIssue:MentallexiconII,81(1–3),120–130.

Catchpole,J.(1987).Thelocusofthewordfrequencye ect.Unpublishedhonorsthesis,DepartmentofPsychology,MonashUniversity.

Cree,G.,McRae,K.,&McNorgan,C.(1999).Anattractormodeloflexicalconceptualprocessing:Simulatingsemanticpriming.CognitiveScience,23(3),371–414.

Damian,M.F.(2001).Congruitye ectsevokedbysubliminallypresentedprimes:Automaticityratherthansemanticpro-cessing.JournalofExperimentalPsychology:HumanPer-ceptionandPerformance,27,154–165.

Davis,C.W.,Kim,J.,&Sanchez-Casas,R.M.(2003).Maskedprimingacrosslanguages:Aninsightintobilinguallexicalprocessing.InS.Kinoshita&S.Lupker(Eds.),Maskedpriming:Stateoftheart.

deGroot,A.M.B.(1992).Determinantsofwordtranslation.JournalofExperimentalPsychology:Learning,Memory,andCognition,18,1001–1018.

deGroot,A.M.B.(1993).Word-typee ectsinbilingualprocessingtasks:Supportforamixed-representationalsystem.InR.Schreuder&B.Weltens(Eds.),Thebilinguallexicon(pp.27–51).Amsterdam:JohnBenjamins.

deGroot,A.M.B.,&Nas,G.L.J.(1991).Lexicalrepresentationofcognatesandnoncognatesincompoundbilinguals.JournalofMemoryandLanguage,30,90–123.

deGroot,A.M.B.,Dannenburg,L.,&vanHell,J.G.(1994).Forwardandbackwardwordtranslationbybilinguals.JournalofMemoryandLanguage,33,600–629.

Dufour,R.,&Kroll,J.F.(1995).Matchingwordstoconceptsintwolanguages:Atestoftheconceptmediationmodelofbilingualrepresentation.Memory&Cognition,23,166–180.

Fellbaum,C.(Ed.).(1998).WordNet:anelectroniclexicaldatabase.Cambridge,MA:MITPress.

Finkbeiner,M.(2002).Bilinguallexicalmemory:TowardsapsycholinguisticmodelofadultL2lexicalacquisition,representation,andprocessing.(Doctoraldissertation,Uni-versityofArizona,2002).DissertationAbstractsInterna-tional,63(12),6114B.(UMINo.AAT3073282).

Finkbeiner,M.(inpress).Task-dependentL2–L1translationpriming:Aninvestigationoftheseparatememorysystemsaccount.InProceedingsofthe4thinternationalsymposiumonbilingualism.Somerville,MA:CascadillaPress.

Forster,K.I.(2000).Thepotentialforexperimenterbiase ectsinwordrecognitionexperiments.Memory&Cognition,28,1109–1115.

Forster,K.I.,&Forster,J.C.(2003).DMDX:Awindowsdisplayprogramwithmillisecondaccuracy.BehaviorRe-searchMethods,Instruments,&Computers,35(1),116–124.

Forster,K.I.,&Hector,J.(2002).Cascadedvsnon-cascadedmodelsoflexicalandsemanticprocessing:Theturplee ect.Memory&Cognition,30,1106–1116.

Forster,K.I.,Mohan,K.,&Hector,J.(2003).Themechanicsofmaskedpriming.InS.Kinoshita&S.Lupker(Eds.),Maskedpriming:Stateoftheart.PsychologyPress.

本文来源：https://www.bwwdw.com/article/8hi4.html