Concreteness in word processing ERP and behavioral eff - PDF document

Concretenessinwordprocessing:ERPandbehavioraleffectsinalexicaldecisiontaskHoracioA.Barber,LeunJ.Otten,Stavroula-ThaleiaKousta,GabriellaViglioccoDepartmentofCognitivePsychology,UniversityofLaLaguna,SpainInstituteofCognitiveNeuroscience,UniversityCollegeLondon,UK Correspondingauthor.Address:DepartamentodePsicologíaCognitiva,UniversidaddeLaLaguna,CampusdeGuajaras/n,LaLaguna,Tenerife38205,Spain.E-mailaddress:(H.A.Barber). Brain&Language125(2013)47…53 ContentslistsavailableatSciVerseScienceDirectBrain&Languagejournalhomepage:www.elsevier.com/locate/b&l proposalthattheN700re”ectsimageryprocesses.Morerecently,AdorniandProverbio(2012)usedsourcelocalizationanalysestoassociatethelateconcretenesseffectwiththeventrolateralpre-frontalcortex,anareafunctionallyrelatedtotop-downcontrolofsemanticmemory.ThepatternofERPsassociatedwiththefasterrespondingtoconcretethanabstractwordshasbeenaccountedforbyacon-text-extendeddualcodinghypothesis(Holcombetal.,1999).Thisaccountintegratestwoofthedominantcognitivetheoriescon-cerningdifferencesbetweenconcreteandabstractwords:thedual-codingtheory(Paivio,1986)andthecontext-availabilityhypothesis(Schwanen”ugel,1991).Dualcodingarguesthatcon-cretewordsarerepresentedinaverbalaswellasnon-verbalcode.Abstractwords,incontrast,areonlyrepresentedinaverbalcode.Accordingtothecontext-availabilityhypothesis,differencesbe-tweenconcreteandabstractwordswouldinsteadariseinasingleverbal-semanticstorebecauseofstrongeranddenserassociativelinksforconcretethanabstractwords.Whereaseitherthedual-codingorcontext-availabilitytheorycanaccountforthereactiontime(RT)advantageforconcreteoverabstractwords,theycannotalsoaccountfortheERPresults.Thus,theirintegrationisproposedbythecontext-extendeddualcodinghypothesis(Holcombetal.,).Accordingtothisaccount,thelargerN400forconcretewordsrepresentspost-lexicalprocessingeffortsassociatedwithintegratingdenserassociativenetworks,inlinewiththecontext-availabilityhypothesis.Concretewordswouldalsobene“tfromactivationofvisualimageryatalaterstage,asindexedbytheN700differences,inlinewiththedualcodingview.ThefactthatthestimulithatelicittheshorterRTs(concretewords)alsoelicitlargerN400s(ratherthansmallerasonewouldexpectfromthelit-erature)canbeaccountedforbyproposingthatthebehavioraladvantageforconcretewordsarisesbecausegreatersemanticinte-grationdemandsareusedstrategicallyasanindexoflexicalityinlexicaldecisions.However,Kouniosandcolleagues(etal.,2009)foundlargerN400amplitudesforconcretewordswithalowernumberofassociates(i.e.lowersemanticrichness)relativetothosewithahighernumberofassociatesandabstractwords.Similarly,arecentstudythatindependentlymanipulatedcon-cretenessandsemanticrichness(numberoflistedsemanticfea-tures)reportedlargerN400amplitudesforconcretethanabstractswords,butsmallerN400amplitudesforwordsassociatedwithricherthanpoorersemanticconcepts(Amsel&Cree,inpressbutseeRabovsky,Sommer,&Abdel-Rahman,2012fortheoppositeresult).Thesestudiescontrastwithpredictionsfromthecontext-extendeddualcodinghypothesis,thusleavingopenthepossibilitythattheERPcorrelatesoftheconcretenesseffectmaybelinkedtodifferentcognitiveprocesses.Itiscriticaltonotethatcomparisonsbetweenconcreteandab-stractwordsinvolveacontrastbetweenitemsthatcandifferalongotherlexicalandsublexicaldimensionsinadditiontoconcrete-ness.First,thecontext-extendeddualcodinghypothesisas-sumesthatdifferencesinconcretenesscomeaboutbecauseofdifferencesinotherdimensions,namelydifferencesincontextavailability(whichcanbeoperationalisedintermsofratingsofhowmanycontextsonecanthinkofforagivenword,”ugel,1991)anddifferencesinimageabilitybetweenconcreteandabstractwords.Hence,ifthesedimensionsarecontrolled,noRTandERPdifferencesbetweenthewordswouldbeexpected.Sec-ond,itisalsothecasethatmostoftenstudiesdidnotcontrolforadditionallexicalvariables.Forinstance,concreteandabstractwordsmatchedforwrittenfrequencycanstillgreatlydifferinfamiliaritytotheadvantageofconcretewords(e.g.,),andfamiliarityhasbeenshowntomodulatethebehavioralconcretenesseffect(Levy-Drori&Henik,2006).Inthepresentstudywematcheditemsforalargenumberoflexicalandsub-lex-icalvariablesincludingfamiliarityand,critically,bothimageabilityandcontextavailability.Intheliterature,itisinvariablyassumedthatthepsycholinguisticconstructsofconcretenessandimage-abilitytapintothesameunderlyingtheoreticalconstruct.Thus,concretenessandimageabilityratingshavebeenusedinterchange-ablyinmostoftherecentliteratureinthe“eld(e.g.Binder,West-bury,McKiernan,Possing,&Medler,2005;Fliessbach,Weis,Klaver,Elger,&Weber,2006;Giesbrecht,Camblin,&Swaab,2004;Rich-ardson,2003).However,concretenessandimageabilitytapinto,atleastpartially,differentaspectsofsemanticrepresentations.Mostoftheconcretewordsareconsideredimageable,whereasab-stractwordsshowhighervariabilityinimageabilityratings.More-Kousta,Vigliocco,Vinson,Andrews,andDelCampo(2011)analyzedratingsformorethan4000wordsintheMRCPsycholin-guisticDatabaseandshowedthatthefrequencydistributionofconcretenessratingsisbimodal(withtwodistinctmodesforab-stractandconcretewords),capturingthecategoricalontologicaldistinctionbetweenconcrete(spatiotemporally-boundconcepts)andabstractconcepts(non-spatiotemporallyboundconcepts).Onthecontrary,thedistributionofimageabilityratingsisuni-modal,indexingthegradedamountsofsensory(primarilyvisual)informationassociatedtothewords.Koustaetal.(2011)havefurthershownthatwhenallknownlexicalandsublexicaldimensions(includingfamiliarity,imageabil-ityandcontext-availability)arecontrolled,orpartialledoutinregressionanalyses,abstractwordselicitfasterRTsthanconcretewordsinlexicaldecisiontasks,resultingthereforeinannesseffectKoustaetal.,2011).IntheexperimentsandregressionanalysesbyKoustaetal.(2011)itwasfoundthatemotionalva-lencedifferedbetweenconcreteandabstractwords,withabstractwordsbeingmorevalencedthanconcreteones,adifferencethatcouldaccountfortheRTadvantageforabstractwordsreported.However,itisimportanttonotethatERPsandbehavioralcon-cretenesseffectsarenotnecessarilyrelatedtothesamecognitiveprocesses.Infact,Holcombetal.(1999)proposedtheoppositeonthebasisoftheparadoxicallargerN400amplitudesassociatedwithconcretewords.Therefore,itisanopenquestionifastandardN400…N700concretenesseffectcanbefoundaftercontrollingforimageabilityandcontext-availability.Ifso,concretenessERPef-fectswouldbedif“culttoaccountforintermsofthecontext-ex-tendeddualcodinghypothesis(Holcombetal.,1999Alternatively,concretenessERPeffectscouldberelatedtothegreateramountofmultimodalinformationactivatedforconcretethanabstractwords.Whiletheactivationofconcretewordsim-pliestheactivationofsensoryandmotorrepresentationsassoci-atedwiththeirmeanings,abstractwords(especiallywhenpresentedoutofcontext)wouldonlyactivatesuper“cialasocia-tionswithrelatedconcepts.Thedifferencesintypeofactivatedinformationcouldalsomodulatethecontrolandintegrationmech-anismsassociatedwithlexico-semanticretrieval(asproposedbyAdorni&Proverbio,2012).Differencesbetweenconcreteandab-stractwordswouldbeduemoretothetypeofinformationthatisactivatedinasinglelexico-semanticsystemthantothenumberofcontextualasociationsormentalimagestriggeredbythem.Here,wecontrastconcreteandabstractitemsthatdonotdifferalongalargenumberoflexicalandsublexicalfactors,includingimageability,contextavailabilityandalsoemotionalvalencetoestablishwhetherwecan“ndabehavioralAdditionally,andconsideringthatelectrophysiologicalandbehav-ioralmeasurescanre”ectdifferentcognitiveprocesses,weassessERPdifferencesinordertotestthecontext-extendeddualhypoth-esis.ThishypothesiswouldpredictthatweshouldnotobserveERPdifferencesbetweenconcreteandabstractwordsgiventhatitemsarematchedforbothcontext-availabilityandimageability.Thus,ifwewereto“ndERPdifferencesbetweenconcreteandabstractwordswewouldhavetoconcludethatthecontext-extendeddualcodinghypothesisdoesnotprovideageneralaccountforEEGH.A.Barberetal./Brain&Language125(2013)47…53 correlatesofconcretenesseffects.Concreteandabstractwords,however,candifferwithregardstotheamountofmultimodalinformationthattheyactivatewhenpresentedoutofcontext(greaterforconcretethanabstractwords).ERPeffectscouldthere-foreindexgenuinedifferencesinthelevelofmeaningactivation.Thus,wemightexpecttoobserveagreaterN400forconcretethanabstractwordsofwhethertheabstractwordsleadtoslowerorfasterRTsthanconcretewords.Moreover,N700effectsaftercontrollingforimagebilitycannoteasilybeexplainedonlybymentalimageryprocesses.Thiseffectwouldinsteadbemoreconsistentwithexecutivecontrolmechanismsthatoperateduringmeaningactivationandworkingmemoryoperations.2.Method2.1.ParticipantsEighteenright-handednativeEnglishspeakers(9women;meanofage24years,range18…33)participatedvoluntarilyinexchangeforpayment(£7.50/h).HandednesswasassessedusinganabridgedversionoftheEdinburghInventory(Old“eld,1971).Allparticipantsgaveinformedconsentandreportednoneurologicorpsychiatrichistory.Threeparticipantswereexcludedfromtheanalysesduetoahighnumberofartefactsintheirdata(morethan50%oftherelevantobservations).TheexperimentalprocedureswereapprovedbythejointUniversityCollegeLondonandUniver-sityCollegeLondonHospitalsethicscommittee.2.2.MaterialsSixtyabstract(e.g.)and60concrete(e.g.)wordswerese-lectedfromanumberofdatabases(seeTable1).Concretenessval-uesweretakenfromtheMRCPsycholinguisticDatabase(),whichusesnormsfromGilhoolyandLogie(1980).Con-cretenessratingswereobtainedona7-pointscale;wordsreferringtoobjects,materials,orpersonsweretoreceiveahighconcrete-nessrating,andwordsreferringtoabstractconceptsthatcouldnotbeexperiencedbythesensesweretoreceivealowconcrete-nessrating.Wordslistsdifferedsigni“cantlyinconcretenessbutwerematchedpairwiseforcontextavailability,imageabilityandanumberofother(sub)lexicalvariables.Forgrammaticalclass,wemadesurethatforeachpair,thewordssharedthesamepartofspeechasde“nedintheMRCdatabase(Coltheart,1981).Forcontextavailability,weobtainedratingsbyasking47nativeEng-lishspeakerstoratewordsona7-pointLikertscaleaccordingtohoweasyitistocomeupwithaparticularcontextorcircumstanceinwhichawordmightappear.TheinstructionstoparticipantswereidenticaltothoseusedbySchwanen”ugelandStowe.Weobtainednormsfor650words,witheachwordratedby22or25speakers.Forimageability,weusednormstakenfromWilson(1988)Stadthagen-GonzalezandDavis(2006).Thesenormsindicatetheeasewithwhichwordsinducementalimages.Highimageryratingsaregiventowordsthatquicklyandeasilypro-vokementalimages,andlowratingstowordsthatprovokementalimageswithdif“cultyornotatall.Finally,concreteandabstractwordsdidnotdifferintermsofemotionalvalence(=.11).ValencevaluesweretakenfromBradleyandLang(1999)andKoustaetal.Weusedthesamedatabasestocreate120pseudowords.Sixtyconcreteand60abstractwordswereselectedandmatchedpair-wisewiththeexperimentalitemsforconcreteness,length,andmeanpositionalbigramfrequency.Asingleletterwasthenalteredfromeachmatchedword.Allbut11pseudowordshadoneortho-graphicneighbor(i.e.,theintendedconcreteorabstractword).Nowordsorpseudowordswererepeatedintheexperiment.2.3.ProcedureStimuliwerepresentedateyelevelonamonitoratadistanceofapproximately70cmfromtheparticipant.StringsappearedinablackArialfont(size24)inagrayrectangle(18397pixels)againstablackbackground.Thesequenceofeventsoneachtrialwasasfollows:a“xationcrosshairwasdisplayedfor700ms,fol-lowedbyablankscreenfor300ms,astringfor300ms,aquestionmarkthatremainedonthescreenfor2000msoruntilaresponsewasgiven,andthen“nallyanother“xationcrosshairforaninter-valthatvariedrandomlybetween1250and1750ms.ThephraseTIMEISUPfollowedthequestionmarkfor1000msifnore-sponsewasgiven.Theexperimentstartedwith10practicetrials,followedbyfourlexicaldecisionblockslastingbetween3and5mineach.Stimuliwerepresentedinarandomorder,andthese-quencewasrandomizedanewforeveryparticipant.ToindicatewhetherornotastringwasanEnglishword,participantswereaskedtopositiontheirleftandrightindex“ngersontworesponsebuttonsmarkedyes/noResponsehandswerecounterbalancedacrossparticipants.Bothspeedandaccuracywerestressed.2.4.EEGrecordingsandERPanalysesEEGwasrecordedfrom29scalpsiteswithAg/AgClelectrodes,embeddedinanelasticatedcapaccordingtoanequidistantelec-trodemontage(montage10;www.easycap.de/easycap/e/elec-trodes/13_M10.htm).Twofurtherelectrodeswereplacedontheleftandrightmastoids.Amidfrontalelectrode(equivalenttoFzinthe10…20system)wasusedastheonlinereference;thedatawerealgebraicallyre-referencedtolinkedmastoidsof”ine,rein-statingtheonlinereferencesite.Electrodesplacedaboveandbe-lowtherighteyeandontheoutercanthirecordedverticalandhorizontaleyemovements.Allsignalswereampli“edandband-pass“lteredbetween0.01…70Hzanddigitizedatarateof200Hz.Thedatawerelow-pass“lteredof”inewitha25Hzcut-off(96dBroll-off,zerophaseshift“lter). Table1Meanvaluesforlexicalandsublexicalcharacteristicsofstimuli.AbstractConcrete341547494495Contextavailability556558Emotionalvalence5.585.11Ageofacquisition391419Numberofsynsets4.705.46494492Logfrequency8.598.81Orthographicneighborhooddensity3.333.54Numberofletters5.955.98Numberofsyllables1.831.83Numberofmorphemes1.161.08Meanpositionalbigramfrequency16571635MRCpsycholinguisticdatabase(Coltheart,1981Koustaetal.,2011WordNet(Fellbaum,1998Theenglishlexiconproject(Balotaetal.,2007BradleyandLang(1999),andKoustaetal.(2011),scalefrom0to10(5=neutral).Thebristolnorms(Stadthagen-Gonzalez&Davis,2006 Arelatedfactorthathasbeenshowntoaffectlexicaldecisiontimesisawordscontextualdistinctiveness(McDonald&Shillcock,2001).AlthoughCDvaluesarenotavailableforallourstimuli,theanalysisofarepresentativesubsetofourwordlists(45(/60)abstractand43(/60)concretewords)didnotshowstatisticallysigni“cantdifferences(=.31).H.A.Barberetal./Brain&Language125(2013)47…53 Epochsof1280mslength,startingat100msbeforeitemonset,wereextractedfromthecontinuousrecordSeparateERPswerecalculatedforeachexperimentalcondition,subject,andelectrodesite.Onlytrialswithcorrectlexicaldecisionswereincluded.Acor-rectionprocedurewasappliedtominimizeblinkartefacts(Mark,Gilchrist,&Roberts,1997).Trialscontainingartefactsotherthanblinkswerediscarded(8.6%ofthetrials,evenlydistributedacrossconditions).ERPswerequanti“edbycomputingmeanamplitudesacrossthe250…450and700…1000mslatencyregions,relativetothe100mspre-stimulusbaseline.TheseintervalswerechoseninaccordancewiththepreviousERPliteratureonconcrete-nesseffects(e.g.Kanske&Kotz,2007;Kellenbachetal.,2002;West&Holcomb,2000),andadjustedaftervisualinspectionofthegrandaverages.Theanalysesincorporatedamplitudesat20scalpsites(excludingelectrodesfromthemidlineandthecentralline)topartitiontheelectrodegridaccordingtotheleft/rightandanterior/posteriorpositionofelectrodes.Thedatawereanalyzedwithrepeatedmeasuresanalysesofvariance(ANOVAs),incorpo-ratingtheGreenhouse-Geissercorrectionforviolationsofspheric-ity(Keselman&Rogan,1980).TheANOVAsincludedfactorsofCONCRETENESS(concrete/abstractword),ROSTRALITY(anterior/posterior),LATERALITY(left/right),andELECTRODE(4sites).Forthe250…450msinterval,anadditionalANOVAwasdirectedatef-fectsoflexicality.Thisanalysiscomparedthedataelicitedbypseudowordsandwords,collapsingacrossconcreteandabstractwords.ThisANOVAincorporatedthefactorofLEXICALITY(words/pseudowords)ratherthanCONCRETENESS.3.Results3.1.BehavioraldataAnANOVAcomparinglexicaldecisionlatenciesacrossconcretewords,abstractwords,andpseudowordsindicatedasigni“cantdifference((2,28)=18.037,.01).Pairwisecomparisonsre-vealedthatlexicaldecisionswerefasterforabstractwords(559ms;sd=73)thanconcretewords(577ms;sd=72)(1,14)=7.872,=.014)andfasterforbothabstractandconcretewordsrelativetopseudowords(664ms;sd=123)(1,14)=19.616and17.007,.01).Therewasatrendinthesamedirectioninaccuracyrates((2,28)=3.102,=.061:con-cretewords=95.8(sd=2.5);abstractwords=95.0(sd=2.7);pseudowords=93.8(sd=3.7)).3.2.ElectrophysiologicaldataFig.1showsthegrandaverageERPwaveformsforwords(col-lapsedacrosswordtype)andpseudowordsatafrontalelectrodesite.VisualinspectionsuggestsdifferencesinthetimewindowoftheN400component,startingataround250msandlastingfor200ms.Asexpectedinataskthatengagessemanticinformation,pseudowordselicitedmorenegativeamplitudesthanwordsinthisinterval.Thedifferenceisbroadlydistributedacrossthescalpwithamaximumoverfrontocentralsites(seesplinemapinFig.1).AnANOVAonthemeanamplitudesinthe250…450msintervalre-vealedasigni“canteffectofLEXICALITY[(1,14)=39.23,.001,=.737].LEXICALITYdidnotinteractwithROSTRALITYorLATERALITY((1,14)=2.18,&#x-237;&#x.600;.20;1,respectively).ThewaveformsseparatingconcreteandabstractwordsareshowninFig.2.IntheN400timewindow,concretewordselicitedmorenegativeamplitudesthanabstractwords.Thisdifferenceexhibitedawidespreadscalpdistributionwithamaximumoverfrontalscalpsites(seesplinemapinFig.2).AnANOVAonthemeanamplitudesinthe250…450msintervalshowedasigni“canteffectofCONCRETENESS((1,14)=4.94,.05,partial=.261),withtheinteractionwithROSTRALITYjustfallingbelowsigni“-cance((1,14)=3.98,=.066).Differencesasafunctionofconcretenessarealsoevidentatla-terlatencies.Startingataround700msandcontinuinguntiltheendoftheanalysisepoch,concretewordsagainelicitedamorenegativede”ectionthanabstractwords.Thisde”ectionislargestoverfrontalscalpsites.Inevidence,anANOVAontheamplitudesinthe700…1000msintervaldemonstratedasigni“cantinteractionbetweenCONCRETENESSandROSTRALITY((1,14)=7.63,.05,=.353).TheinteractionwiththeLATERALITYfactorwasnotsigni“cant(1).4.DiscussionWeexaminedthevisualrecognitionprocessesassociatedwithconcreteandabstractwordsmatchedonalargenumberoflexicalandsublexicalvariables,includingimageabilityandcontextavail-ability.Replicatingrecent“ndingswithsimilarlymatchedstimuli,lexicaldecisionresponseswerefasterforabstractwords(etal.,2011).Thiscontrastswithmostoftheexistingliterature,whichtypicallydescribesfasterresponsetimesforconcretewordsKanske&Kotz,2007;Kounios&Holcomb,1994;Schwanen-”ugel,1991).Crucially,despitetheabstractnessbehavioraleffect,wefoundthesamepatternofERPdifferencesthathasbeen Fig.1.Lexicalityeffect.GrandaverageERPwaveformsforwords(collapsedacrosswordtype)andpseudowordsatamidfrontalelectrodesite.Inthisandthenext“gure,verticallinesmarktheonsetoftheletterstrings(=0).Positivevaluesareplottedupward.Thetwo-dimensionalvoltagesplinemapontherightillustratesthedistributionacrossthescalpofthelexicalityeffectinthe250…450mstimewindow.Theeffectwascomputedbysubtractingthevaluesassociatedwiththeunweightedaverageofconcreteandabstractwordsfromthoseassociatedwithpseudowords.Mapsinthisandthenext“gurearescaledsymmetricallydependingonthemaximumnegativevalueofeacheffect.H.A.Barberetal./Brain&Language125(2013)47…53 reportedinpreviousstudies.Namely,ERPsforconcretewordsweremorenegative-goingthanERPsforabstractwordsbetween250and450msafterwordonset.ERPconcretenesseffectswiththispolarityandlatencyhavebeeninterpretedasmodulationsoftheN400component(Kounios&Holcomb,1994).Differencesbe-tweenconcreteandabstractwordswerealsoobservedinalatertimewindow,between700and1000ms.Thislattereffect,whichwaslargestoverfrontalsites,conformstothepreviouslyreportedN700concretenesseffect(e.g.West&Holcomb,2000TheN400concretenesseffecttookplaceinthesametimewin-dowasthelexicalityeffect,aneffectthathasbeenshowntomod-ulatealsotheN400component,withlargeramplitudesforpseudowordscomparedtowords(Barber,Vergara,&Carreiras,).ThemodulationoftheN400componentbythelexicalstatusofthestimulicanbetakenasevidenceofwordmeaningactivationwhenperformingthelexicaldecisiontask.Theconcretenesseffecthadafrontocentraldistributioninlinewithpreviousstudiesthatcomparedconcreteandabstractwords(e.g.Holcombetal.,1999;Kanske&Kotz,2007;West&Holcomb,2000).Lexicalityandcon-cretenesseffectsshowedsimilarbutnotidenticalscalpdistribu-tions.Concretenesseffectsshowedamorefrontaldistributionthanlexicality,thoughthedistributionofthelatterwasstillmorefrontalthanthestandardcentroparietalN400effectforwordspre-sentedincontext.ThesevariationsinthedistributionoftheeffectsareconsistentwithageneralviewoftheN400componentasthesumofseveralneuralgenerators,whoseactivationlevelsdependonboththeavailablecontext-relatedinformation(andtheresult-ingintegrationprocesses)andthelexico-semanticfeaturesofthestimulithatmustberecoveredfromlong-termmemory.Thus,whereastightlymatchedstimuliresultedinfasterre-sponsesforabstractwords,ERPstoconcretewordsstillshowedthepreviouslyreportedlargerN400andN700de”ections.These“ndingsallowtwomainconclusions.First,theresultsdonotsup-portthepredictionsoftheprevailingfunctionalinterpretationofERPcorrelatesofconcreteandabstractwordprocessing,namelythecontext-extendeddualcodinghypothesis.Second,concrete-ness/abstractnesseffectsinlexicaldecisionresponsetimesandelectricalbrainactivityaredissociable.Distinctprocessesmustthereforeunderlieeachofthem.Previousresearchhasarguedthatconcretewordshaverichersemanticrepresentations,withstrongeranddenserlinkstoassoci-atedsemanticinformationthanabstractwords(Holcombetal.,1999;West&Holcomb,2000).Insuchaview,thelargerN400forconcretewordsresultsfromthegreatereffortthatisrequiredtointegratethemoreextensivesemanticinformationintoahigh-er-levelrepresentation.However,inourexperiment,concreteandabstractwordswerematchedonanumberofsemanticvariablesthatmeasuresemanticrichness,criticallyincludingcontextavail-ability(seefootnote1inrelationtocontextualdistinctiveness).ThischallengesaninterpretationoftheN400concretenesseffectintermsoftheamountofsemanticinformationassociatedwiththewords.Importantly,wordsintheexperimentalsodidnotdifferwithregardstoemotionalvalence,consideredtobeacriticaldeterminantofabstractnessbehavioraleffectsinKoustaetal.Thetypesofknowledgeunderlyingabstractandconcretecon-ceptscanbesigni“cantlydifferent.Whileconcreteconceptsrelyonmodality-speci“cfeatures,abstractconceptsrelyonemo-tionalassociationsaswellasavarietyofothersituationalandlinguisticinformation(Andrews,Vigliocco,&Vinson,2009staetal.,2011BarsalouandWiemer-Hastings(2005)havepro-posedthat,althoughsituations(i.e.contexts)arenecessaryforacompleterepresentationofbothabstractandconcreteconcepts,theyareessentialforabstractconcepts.Whenwordsarepre-sentedoutofcontextandundertimepressure(asinthepresentexperiment),abstractmeaningswillthereforereceiveminimalprocessing.Instead,abstractwordswillactivateanumberofsuper“cialassociationswithotherwords,whichcannotneces-sarilybeintegratedinauni“edconcept.TakentheN400compo-nentasanindexofmeaningactivation(Molinaro,Conrad,Barber,&Carreiras,2010),itsamplitudevariationsinourexper-imentmayre”ectdifferencesinthelevelofmeaningactivationwhenwordsarepresentedinisolation;whereasconcretewordswouldactivateandintegratemultimodal(sensory-motor)fea-turesfromdistributedcorticalnetworks,processingofabstractwordswouldresultinashalloweractivationprocess.Consistentwiththispossibility,neuroimagingstudiessuggestthatconcretewordprocessingengagesalargenumberofnetworkslinkedwiththespeci“csensory-motorpropertiesoftheitem(seere-viewsinBinder,Desai,Graves,&Conant,2009;Cappa,2008;Martin,2007).Incontrast,studiesconvergeinshowingthatab-stractwordsengageperisylvianandprefrontalareastoagreaterdegreethanconcretewords(Binderetal.,2009;Goldberg,Per-fetti,Fiez,&Schneider,2007 Fig.2.Concretenesseffect.GrandaverageERPwaveformsforconcreteandabstractwordsatarightfrontalelectrodesite.Shownontherightarethevoltagesplinemapsillustratingthescalpdistributionsoftheconcretenesseffects(amplitudedifferencesbetweenconcreteandabstractwords)inthe250…450msand700…1000msintervals.H.A.Barberetal./Brain&Language125(2013)47…53 Concreteandabstractwordswerenotonlymatchedforcontextavailability,butalsoforimageability.Nonetheless,wefoundalar-gerN700forconcretewords,justasinpreviousresearch(cf.combetal.,1999;Kanske&Kotz,2007;West&Holcomb,2000WestandHolcomb(2000)relatedtheN700concretenesseffecttoenhancedimagery-relatedprocessingforconcretewords,con-sistentwiththedualcodingtheory(Paivio,1986).AlthoughandHolcomb(2000)didnotspecifytheexactnatureofsuchimag-eryprocesses,theysuggestedthatitcouldberelatedtovisualimagery(Kosslyn,Thompson,&Alpert,1997).Thefactthatvisualimageabilitywasmatchedacrosswordtypesinourexperimentmakessuchaninterpretationunlikely.Here,thegreaterN700elic-itedbyconcretewordsmayre”ectdifferencesintheamountofsensory-motorfeaturesthatneedtobeintegratedandmanipu-latedtobuildandmaintainamentalrepresentation.Inapreviousstudy,wereportedasimilarsustainedfrontaleffectthatdifferen-tiatednounsandverbswithdifferentamountsofassociatedmotorandsensoryfeatures(Barber,Kousta,Otten,&Vigliocco,2010ThereforealthoughtheN700effectcannotjustbereducedtovisualimagery(whichwasmatchedinthepresentstudy),itcouldstillberelatedtointernalrepresentations,andtheeffectmayre”ecttheamountandcharacteristicsofthemultimodalinformationin-volvedinamentalrepresentation.ThisviewisalsocompatiblewiththeproposalthattheN700effectcouldbelinkedwithactivityinventrolateralprefrontalcortex,whichcouldcontrolandmodu-latetheretrievalofrepresentationsfromthelong-termmemorysystem(Adorni&Proverbio,2012Interestinglyourresultsshowadissociationbetweenelectro-physiologicalandbehavioralmeasures.InspiteoftheclassicalN400-N700concretenesseffect,responselatencieswereshorterforabstractthanforconcretewords.Althoughthereisextensiveevidenceofsemanticeffectsinlexicaldecisions(Balotaetal.,),wordversusnon-worddecisionscanalsotakeadvantageofotherlinguisticcues(e.g.orthography,phonologyorgenerallev-elsofactivation).Inthepresentstudy,stimuliwerematchedforallthoseothervariablesthatnormallyfavorconcretewords,suchasfamiliarity,ageofacquisition,imageabilityandcontextavailabil-ity.Underthesecircumstances,theproposedlargernumberofsuper“cialwordassociationstriggeredbyabstractwords(&Wiemer-Hastings,2005)canbeexploitedasaquickindexforcon“rminglexicality.Therefore,wesuggestthatERPconcretenesseffectscanbetakenasanindexofmeaningactivationprocesses,modulatedbythedegreeofmultimodalityofthesemanticinfor-mationbeingintegrated(greaterforconcretethanabstractwords).Thebehavioralabstractnesseffectmay,instead,re”ectdecisionandresponse-selectionmechanismsthatareextremelysensitivetocontrol(Neely,1991).Regardlessofthevalidityofthisinterpre-tation,thecurrentdataclearlypointtoadissociatonbetweenbehavioralandelectrophysiologicaldifferencesbetweenconcreteandabstractwords.Insummary,wehaveshownthattheadequatecontrolofcon-foundinglexicalandsublexicalvariablesiscriticalforstudiesonwordconcretenesseffectsusingfactorialdesigns.However,evenafterthecontrolofseveralcriticalvariablesweshowthattheERPconcretenesseffectcanbefound.Theseresultsarenoteasilyexplainedintheframeworkofthecontext-extendeddualcodinghypothesis.Alternatively,weproposethatdifferencesbetweentheprocessingofconcreteandabstractswordsemergefromtheamountofsensory-motorinformationthatisactivatedandinte-gratedwithinasinglelexico-semanticsysteminvolvingextensivecorticalnetworksinthebrain.AcknowledgmentsSupportforthisworkwasprovidedbyEuropeanUnion(FP6-2004-NESTPATH)Grant028714andbyESRCGrantRES-062-23-2012toGabriellaVigliocco,andtheGrantPSI2010-19767oftheSpanishMinistryofSciencetoHoracioA.Barber.AppendixA.SupplementarymaterialSupplementarydataassociatedwiththisarticlecanbefound,intheonlineversion,athttp://dx.doi.org/10.1016/j.bandl.2013.Adorni,R.,&Proverbio,A.M.(2012).Theneuralmanifestationofthewordconcretenesseffect:Anelectricalneuroimagingstudy.Neuropsychologia,50Amsel,B.D.,&Cree,G.S.(inpress).Semanticrichness,concreteness,andobjectdomain:Anelectrophysiologicalstudy.CanadianJournalofExperimentalAndrews,M.,Vigliocco,G.,&Vinson,D.(2009).Integratingexperientialanddistributionaldatatolearnsemanticrepresentations.PsychologicalReview,(3),463…498.Balota,D.A.,Yap,M.J.,Cortese,M.J.,Hutchison,K.I.,Kessler,B.,Loftis,B.,etal.(2007).TheEnglishlexiconproject.BehaviorResearchMethods,39,445…459.Barber,H.A.,Kousta,S.T.,Otten,L.J.,&Vigliocco,G.(2010).Event-relatedpotentialstoevent-relatedwords:Grammaticalclassandsemanticattributesintherepresentationofknowledge.BrainResearch,21(1332),65…74.Barber,H.A.,&Kutas,M.(2007).Interplaybetweencomputationalmodelsandcognitiveelectrophysiologyinvisualwordrecognition.BrainResearchReviews,,98…123.Barber,H.A.,Vergara,M.,&Carreiras,M.(2004).Syllable-frequencyeffectsinvisualwordrecognition:EvidencefromERPs.Neuroreport,15(3),545…548.Barsalou,L.W.,&Wiemer-Hastings,K.(2005).Situatingabstractconcepts.InD.Pecher&R.Zwaan(Eds.),Groundingcognition:Theroleofperceptionandactioninmemory,language,andthought(pp.129…163).NewYork:CambridgeUniversityBinder,J.R.,Desai,R.H.,Graves,W.W.,&Conant,L.L.(2009).Whereisthesemanticsystem?Acriticalreviewandmeta-analysisof120functionalneuroimagingCerebralCortex,19,2767…2796.Binder,J.R.,Westbury,C.F.,McKiernan,K.A.,Possing,E.T.,&Medler,D.A.(2005).Distinctbrainsystemsforprocessingconcreteandabstractwords.JournalofCognitiveNeuroscience,17,905…917.Bradley,M.M.,&Lang,P.J.(1999).AffectivenormsforEnglishwords(ANEW):Stimuli,instructionmanualandaffectiveratings.TechnicalreportC-1,Gainesville,FL.TheCenterforResearchinPsychophysiology,UniversityofCappa,S.F.(2008).Imagingstudiesofsemanticmemory.CurrentOpinioninNeurology,21(6),669…675.Coltheart,M.(1981).MRCpsycholinguisticdatabase.QuarterlyJournalofExperimentalPsychology,33a,497…505.Fellbaum,C.(1998).WordNet:Anelectroniclexicaldatabase.Cambridge,MA:MITFliessbach,K.,Weis,S.,Klaver,P.,Elger,C.E.,&Weber,B.(2006).Theeffectofwordconcretenessonrecognitionmemory.NeuroImage,32,1413…1421.Giesbrecht,B.,Camblin,C.C.,&Swaab,T.Y.(2004).Separableeffectsofsemanticprimingandimageabilityonwordprocessinginhumancortex.CerebralCortex,,521…529.Gilhooly,K.J.,&Logie,R.H.(1980).Ageofacquisition,imagery,concreteness,familiarityandambiguitymeasuresfor1944words.BehaviourResearchMethodsandInstrumentation,12,395…427.Goldberg,R.F.,Perfetti,C.A.,Fiez,J.A.,&Schneider,W.(2007).SelectiveretrievalofabstractsemanticknowledgeinLeftPrefrontalCortex.TheJournalofNeuroscience,27(14),3790…3798.Holcomb,P.J.,Kounios,J.,Anderson,J.E.,&West,W.C.(1999).Dual-coding,contextvailability,andconcretenesseffectsinsentencecomprehension:Anelectrophysiologicalinvestigation.JournalofExperimentalPsychology:Learning,Memory,andCognition,25,721…742.Huang,H.,Lee,C.,&Federmeier,K.D.(2010).Imaginethat!ERPsprovideevidencefordistincthemisphericcontributionstotheprocessingofconcreteandabstractconcepts.NeuroImage,49,1116…1123.Kanske,P.,&Kotz,S.A.(2007).Concretenessinemotionalwords:ERPevidencefromahemi“eldstudy.BrainResearch,1148,138…148.Kellenbach,M.L.,Wijers,A.A.,Hovius,M.,Mulder,J.,&Mulder,G.(2002).Neuraldifferentiationoflexico-syntacticcategoriesorsemanticfeatures?Event-relatedpotentialevidenceforboth.JournalofCognitiveNeuroscience,14Keselman,H.J.,&Rogan,J.C.(1980).RepeatedmeasuresFtestsandpsychophysiologicalresearch:Controllingthenumberoffalsepositives.Psychophysiology,17,499…503.Kosslyn,S.M.,Thompson,W.L.,&Alpert,N.M.(1997).Neuralsystemssharedbyvisualimageryandvisualperception:Apositronemissiontomographystudy.Neuroimage,6,320…334.Kounios,J.,Green,D.L.,Payne,L.,Fleck,J.I.,Grondin,R.,&McRae,K.(2009).Semanticrichnessandtheactivationofconceptsinsemanticmemory:Evidencefromevent-relatedpotentials.BrainResearch,28(1282),95…102.H.A.Barberetal./Brain&Language125(2013)47…53 Kounios,J.,&Holcomb,P.J.(1994).Concretenesseffectsinsemanticprocessing:ERPevidencesupportingdual-encodingtheory.JournalofExperimentalPsychology:Learning,Memory,andCognition,20,804…823.Kousta,S.T.,Vigliocco,G.,Vinson,D.P.,Andrews,M.,&DelCampo,E.(2011).Therepresentationofabstractwords:Whyemotionmatters.JournalofExperimentalPsychology:General,140(1),14…34.Kutas,M.,&Federmeier,K.D.(2011).Thirtyyearsandcounting:FindingmeaningintheN400componentoftheevent-relatedbrainpotential(ERP).AnnualReviewofPsychology,62,621…647.Levy-Drori,S.,&Henik,A.(2006).Concretenessandcontextavailabilityinlexicaldecisiontasks.AmericanJournalofPsychology,119(1),45…65.Martin,A.(2007).Therepresentationofobjectconceptsinthebrain.AnnualReviewofPsychology,58,25…45.McDonald,S.A.,&Shillcock,R.C.(2001).Rethinkingthewordfrequencyeffect:Theneglectedroleofdistributionalinformationinlexicalprocessing.LanguageandSpeech,44,295…323.Molinaro,N.,Conrad,M.,Barber,H.A.,&Carreiras,M.(2010).OnthefunctionalnatureoftheN400:Contrastingeffectsrelatedtovisualwordrecognitionandcontextualsemanticintegration.CognitiveNeuroscience,1,1…7.Neely,J.H.(1991).Semanticprimingeffectsinvisualwordrecognition:Aselectivereviewofcurrent“ndingsandtheories.InD.Besner&G.W.Humphreys(Eds.),Basicprocessesinreading(pp.264…336).Hillsdale,NJ:LawrenceErlbaumNittono,H.,Suehiro,M.,&Hori,T.(2002).WordimageabilityandN400inanincidentalmemoryparadigm.InternationalJournalofPsychophysiology,44Old“eld,R.C.(1971).Theassessmentandanalysisofhandedness:TheEdinburghNeuropsychologia,9,97…113.Paivio,A.(1986).Mentalrepresentations:Adualcodingapproach.Oxford:OxfordUniversityPress.Rabovsky,M.,Sommer,W.,&Abdel-Rahman,R.(2012).Thetimecourseofsemanticrichnesseffectsinvisualwordrecognition.FrontiersinHumanNeuroscience,6Richardson,J.(2003).Dualcodingversusrelationalprocessinginmemoryforconcreteandabstractwords.EuropeanJournalofCognitivePsychology,15Rugg,M.D.,Mark,R.E.,Gilchrist,J.,&Roberts,R.C.(1997).ERPrepetitioneffectsinindirectanddirecttasks:Effectsofageandinter-itemlag.Psychophysiology,34Schwanen”ugel,P.J.,&Stowe,R.W.(1989).Contextavailabilityandtheprocessingofabstractandconcretewordsinsentences.ReadingResearchQuarterly,24Schwanen”ugel,P.(1991).Whyareabstractconceptshardtounderstand?InP.Schwanen”ugel(Ed.),Thepsychologyofwordmeanings(pp.223…250).Hillsdale,NJ:Erlbaum.Stadthagen-Gonzalez,H.,&Davis,C.J.(2006).TheBristolnormsforageofacquisition,imageability,andfamiliarity.BehaviorResearchMethods,38Welcome,S.E.,Paivio,A.,McRae,K.,&Joanisse,M.F.(2011).Anelectrophysiologicalstudyoftaskdemandsonconcretenesseffects:Evidencefordualcodingtheory.ExperimentalBrainResearch,212,347…358.West,W.C.,&Holcomb,P.J.(2000).Imaginal,semantic,andsurface-levelprocessingofconcreteandabstractwords:AnelectrophysiologicalJournalofCognitiveNeuroscience,12,1024…1037.Wilson,M.(1988).MRCpsycholinguisticdatabase:Machine-usabledictionary,version2.00.BehaviorResearchMethods,Instruments,andComputers,20,6…10.H.A.Barberetal./Brain&Language125(2013)47…53