ORIGINALRESEARCHARTICLEpublished:10February2015doi:10.3389/fnhum.2015. - PDF document

ORIGINALRESEARCHARTICLEpublished:10February2015doi:10.3389/fnhum.2015.00028 Metaphorsarephysicalandabstract:ERPstometaphoricallymodiednounsresembleERPstoabstractBálintForgácsMeganD.BardolphBenD.AmselKatherineA.DeLongMartaKutasKutasCognitiveElectrophysiologyLab,DepartmentofCognitiveScience,UniversityofCalifornia,SanDiego,CA,USA FrontiersinHumanNeurosciencewww.frontiersin.orgFebruary2015|Volume9|Article28 Forgácsetal.Metaphorsarephysicalandabstracttheleftmiddlesuperiortemporalsulcussimilarlytoabstractparaphrases(“Thepublicunderstoodtheidea”),bothanabstractandphysicalcomponentareimplicated.Lessstricttheoriesofembodiedlanguageprocessing(e.g.,BinderandDesai,2011)suggestthatonlynovelexpressionsactivatesensorimotorregions,andfamiliarexpressionsand/orfamiliarcontextsrelyonmoreabstractrepresentations.Thisproposalresonateswithlanguageprocessingmodelssuchasthegradedsaliencehypothesis(Giora,1997,2003)orthecoarsesemanticcodingtheory(Beeman,1998;Jung-Beeman,2005)thatpredictdifferentprocessesfornovelexpressions(i.e.,bytherighthemisphere),regardlessofgurativeness,andforconventionalexpressions(i.e.,bythelefthemisphere)asaresultoftheirsalientmeaningand/orhighdegreeofassociation.Thesepotentialdifferencesbetweencomprehensionofconventionalandnovelmetaphorsfalloutsidethescopeofthecurrentinquiry,sincewefocussolelyonrelativelynovelexpressions.Ouraiminthisreportistobetterunderstandtherolethatphysical(orconcrete,aswewillalsorefertothem)propertiesofindividualwords(adjectives)and/orconcepts(expressedbyadjective-nounpairs)playinnovelmetaphorcomprehensioninreal-time.Tothatend,weemployanonlinemethodologythatpermitsmoment-by-momentexaminationofthemetamorphosisfromconcrete,literallanguageintometaphorical,emergentconcepts.Specically,werecordedevent-relatedbrainpotentials(ERPs)—amethodenablingnotjustquantitative,butalsoqualitative,comparisonsoftheneuralprocessingrelatedtolinguisticphenomenaastheyunfoldintime.ERPstudiesofmetaphorprocessingareoftencenteredontheN400ERPcomponent.ThevisualN400(KutasandHillyard,1980)isanegative-goingcentroparietallymaximalERPcomponentpeakingapproximately400msafterstimulusonset,sensitivetotheeaseordifcultyofsemanticmemoryaccess(forareview,seeKutasandFedermeier,2011).AlthoughtheN400issensitivetoawidevarietyoffactorsthatvaryatthesingle-wordlevel(e.g.,concreteness,frequency,orthographicneighborhoodsize,repetition),theeffectsoftop-downcontextualinformationgenerallyoutweighthoseofbottom-upinformation.SeveralERPstudieshavereportedlargerN400stowordsappearinginmetaphorical(e.g.,“powerisastrongintoxicant”)vs.literal(“whiskeyisastrongintoxicant”)expressions(e.g.,Pynteetal.,1996;CoulsonandVanPetten,2002,2007;Tartteretal.,2002;Arzouanetal.,2007;Laietal.,2009;LaiandCurran,2013);overall,however,resultshavebeeninconsistent.Conventionalmetaphors(e.g.,“brokenheart”)haveaxedgurativemeaning,andmightbestoredaslexicalunits(Jackendoff,1997).Perhapsasaconsequencetheyhavebeenfoundtobeprocessedfasterandmoreaccuratelythannovelmetaphors(e.g.,“rustymoves”),forwhichgurativemeaningneedstobecomputedon-line(e.g.,FaustandMashal,2007;Forgácsetal.,2014).Investigationsofnovelmetaphors,however,differconsiderablyintheirdetails.Pynteetal.(1996),forexample,modiedthetopicsofconventionalmetaphors(“ghters”in“Thoseghtersarelions”),notthevehicles(“lions”)thatcarrythegurativemeaning;inTartteretal.(2002),sentencenalwordswerenotidenticalacrossconditions,leadingtodifferencesinfrequencyandclozeprobability;(CoulsonandVanPetten,2002,2007)controlledfortheclozeprobabilityofsentencenalwords,butnotforthenoveltyandcomplexityoftheexpressionsthemselves(forfurtherconcernswiththeirstimuliseeLaietal.,2009).Laietal.(2009)carriedoutawell-controlledstudy,usingamixtureofnoun-,adjective-andverb-basedmetaphorsinsentencesofvaryingcomplexity.TheyshowedthatwhileconventionalandnovelmetaphorsbothelicitedlargeramplitudenegativitiesrelativetoliteralsentencesearlyintheN400timewindow(320–440ms),processingofconventionalmetaphorsconvergedwiththatoftheliteralsentences,whereasnovelmetaphorscontinuedtobetreatedmorelikeanomaloussentences.Theyattributedthesustainednegativity(between440–560ms)elicitedbynovelmetaphorstosemanticintegrationprocesses.Figurativelanguagealsohasbeenstudiedusingsemanticallylinkedwordpairsthatconstituterelativelyminimallinguisticcontexts.Arzouanetal.(2007)comparedliteral,conventionalmetaphoric,novelmetaphoric,andunrelatedtwo-wordexpressionsbymanipulatingtherstwordwhilematchingthesecondwordonseveralpsycholinguisticmeasures.TheyfoundthattheN400tothesecondwordmonotonicallyincreasedfromliteral,toconventionalmetaphorical,tonovelmetaphorical,tounrelatedpairs.Theyalsofounddifferencesinscalptopographyandtimingthatsuggestqualitativedifferencesbetweentheprocessingofconventionalandnovelmetaphoricalexpressions;specicallytheysuggestedthatalatenegativewave(between550–880ms)reectssecondarysemanticintegration,specictonovelmetaphors.However,whennovelmetaphorsarecomparedtoconventionalliteralexpressionsortosentences,noveltyandgurativenessareconfounded;hencethesourceoftheeffectisnotclear.Comparingnovelmetaphorstoconventionalmetaphorsisnotanoptimalsolution;rstlybecauseitis,inessence,amanipulationoflanguageconventionality;andsecondly,andperhapsmoreimportantly,theremightbedifferentprocessesinvolvedincomprehendingthetwo(cf.,BowdleandGentner,2005;Forgácsetal.,2012).Together,thesestudiesdemonstrate,nevertheless,thatmetaphoricityinuencesrealtimelanguageprocessingwithinthesametimewindow(i.e.,200–900mspoststimulusonset)asmanyothersemanticfactors(KutasandFedermeier,2011).TheN400timewindowoftheERPislikewisesensitivetotheconcrete-abstractdimensionofwords,whichmightplayakeyroleinthecreationandcomprehensionofmetaphors,whichofteninvolvemappingbetweenanabstract(target)andamoreconcrete(source)concept.Aftercontrollingforpotentialconfoundingfactorsbetweenconcreteandabstractwords,recentworkshowsbehavioralprocessingadvantagesforabstractwords(Koustaetal.,2011;Barberetal.,2013).ERPstudiesalsoindicaterapiddifferentialprocessingofconcreteandabstractwordsbyabout300mspoststimulusonset(e.g.,KouniosandHolcomb,1994;Holcombetal.,1999;WestandHolcomb,2000;LeeandFedermeier,2008;Barberetal.,2013;forasummaryseeKutasetal.,2006).ERPconcretenesseffectsaretypicallycharacterizedasgreaternegativitytoconcretewords FrontiersinHumanNeurosciencewww.frontiersin.orgFebruary2015|Volume9|Article28|2 Forgácsetal.MetaphorsarephysicalandabstractrelativetoabstractwordsstartingwithintheN400timewindow.Thesedifferencessometimesextendintoalatertimewindow(500–800ms),wheretheytypicallyexhibitmoreanterior,andmorerightlateralized,scalptopographies(e.g.,WestandHolcomb,2000).ThesepotentiallyseparableelectrophysiologicalconstituentsoftheconcretenesseffectareconsistentwithPaivio(2007)dualcodingtheory,underwhichtherearetwosemanticsystems:alinguisticonethatencodesbothabstractandconcretewords,andanon-verbalimagisticsystemthatencodesonlyconcretewords.Onthistheory,concretewordsenjoyaprocessingadvantagebecausetheyactivatedualrepresentationsandtapintoneuralresourcesinboththelinguisticandimagisticsystems.Concretenesseffectshavebeenobservedinweaklyconstrainingsentencecontexts(e.g.,WestandHolcomb,2000),aswellasinsinglewordcontexts.Swaabetal.(2002)presentedabstractandconcretewordsfollowing(un)relatedprimewords,andobservedcanonicalN400primingeffects(largeramplitudeN400stowordsprecededbyunrelatedrelativetorelatedwordprimes)forbothconcreteandabstractwords.However,theyalsofoundtopographicdifferencesbetweentheabstractandconcretewords,andanenhancedfrontalnegativityforconcretewords,regardlessofprimerelatedness,consistentwithstructuraland/orqualitativeprocessingdifferences.Itappearsthatincontrasttosententialcontextsthateliminatetopographicdifferences(Holcombetal.,1999),singlewordcontextsdonotsufcetooverridequalitativeERPconcretenesseffects.Nonetheless,itseemsthatcontextalsomayhavesomebearingontheelicitationofconcretenesseffects.WhereasminimalcontextERPstudieshavetypicallymanipulatedconcretenessbypresentingdifferentsetsofconcreteandabstractwordsthatthuscoulddifferonanynumberofotherfactors,Huangetal.(2010)cleverlyreliedondifferentadjective-samenouncombinationstomanipulatewhetheragivennounwasmodiedinaconcreteorabstractfashion.TheyconductedadividedvisualeldERPstudyinwhichpolysemousnouns(e.g.,“book”)werepresentedintheleftandrightvisualelds(LVF,RVF),modiedeitherbyabstractadjectives(“interesting”)orconcreteadjectives(“thick”).FollowingashestotheRVF(lefthemisphere),concretelymodiednouns(“thickbook”)evokedreducedN400s(300–500ms)relativetoabstractlymodiednouns(“interestingbook”);thisisthereverseofthecanonicalERPconcretenesseffect.Theauthorssuggestedthatconcreteadjectives(whichthemselvesevokedthecanonicalconcretenesseffect)establishedamoreconstrainingcontextthanabstractadjectives,andtheresultingincreasedexpectancyledtoreducedN400s.FollowingLVF(righthemisphere)presentation,concrete(vs.abstract)expressionsevokedasustainednegativityoverfrontalelectrodesitesonlyinalater500–900mstimewindow,consistentwithpreviouslyreportedqualitativeprocessingdissociations,andthereforewithsomeversionsofthedual-codingtheory.Basedontheresultsofthetwo-wordstudiesofSwaabetal.(2002)andHuangetal.(2010),thecanonical(context-driven)N400expectancyeffectobservedinpublishedmetaphorstudiesmightbeindependentofthelexicalconcretenesseffectseeninthesamewindow,asthetwoeffectsseemtogoinoppositedirections.Tosumup,metaphoricalexpressionsveryoftenrelyonphysicalexpressionsdenotingconcretesourcedomainstodescribeabstracttargetdomains.Whereasgurativemeaningclearlygoesbeyondthesumofitsparts(i.e.,thephysicalsensesofconstituentwords),itislesscleartowhatextent(andwhen)thephysicalsensesofconstituentconcretewordsimpactimmediateprocessingofmetaphoricalexpressions.ElectrophysiologicalstudiesofmetaphorprocessinggenerallyshowsmalleramplitudeN400stoliteralrelativetometaphoricalexpressions.Incontrast,electrophysiologicalstudieswithcentrallypresentedsinglewordsorexpressionstypicallyreportagreaternegativitywithintheN400timewindow(andsometimesbeyond)tomoreconcreterelativetomoreabstractwords.Againstthisbackgroundliterature,wesetouttoassesswhethermetaphoricalexpressionscreatedbycombiningphysicaladjectivesthatdonotliterallymodifynouns(e.g.,“stickymeeting”)areprocessedmorelikeconcreteorabstractadjective-nounexpressions.WeadoptedthewordpairparadigmofHuangetal.(2010)inwhichdifferentadjectivesarecombinedwiththesamenountoruleoutanypotentiallexicaldifferencesbetweentargetstimuli.Giventhatfamiliaritycanmediatebetweenconcretenessandcontexteffects(Levy-DroriandHenik,2006),welimitedourexplorationtonovelmetaphoricaladjective-nounwordpairs,therebyrulingoutconventionalmetaphorsthatmightbestoredinthelexicon,andthusinvokedifferentprocessing.Wecomparedandcontrastedthefollowingconditions,forwhichindividualstimulusitemswereformedbycombiningthreedifferentadjectiveswiththesamenoun:(1)AbstractLiteral(AL)expressionswhichwerecomprisedofanabstractadjective+noun(e.g.,“conditionalschedule”);(2)ConcreteLiteral(CL)expressionswhichwerecomprisedofaconcreteadjective+noun(e.g.,“printedschedule”);and(3)Metaphorical(MET)expressionswhichwerecomprisedofadifferentconcreteadjective+noun(e.g.,“thinschedule”)thatwerelikelytobeinterpretedmetaphoricallyastheycouldnotsensiblybeinterpretedliterally(SeeTable1foradditionalrepresentativestimuli).Table1|Examplestimuli. MetaphoricalConcreteAbstract(MET)literal(CL)literal(AL)adjectiveadjectiveadjectiveNoun FluffyNasalIneffectiveSpeechStickyLoudConstructiveMeetingStaleScaryComprehensiveMovieVelvetyHotProtectedLakeMagneticSlimyIntelligentBrainBuzzingLivelyDiligentReceptionistGuttedLushMysticalForestFragileSlopedUnknownPathWoundedSaltyRadioactiveEarthDrippingSopranoSymbolicToneRustyPainfulImprovisedMovesSparklingLuxuriousIllegalParty FrontiersinHumanNeurosciencewww.frontiersin.orgFebruary2015|Volume9|Article28|3 Forgácsetal.MetaphorsarephysicalandabstractAtissuewaswhetherprocessingofMETexpressionswouldbedriven(1)bytheconcrete(physical)natureoftheadjective(e.g.,“thin”)—inwhichcaseERPstotheMETnounswouldmimicthosetotheCLnouns;(2)bythenon-literalabstractnatureofthenounphrases—inwhichcaseERPstotheMETnounswouldmimicthosetotheALnouns;or(3)bythenon-literal,metaphoricalnatureofthenounphraseinterpretation(e.g.,“thinschedule”)—inwhichcaseERPstotheMETnounswoulddifferfromthosetoboththeCLandALnouns,elicitingthelargestN400and/orlatenegativityasinmostERPstudiesofnovelmetaphors.WeconsiderseveralpotentialoutcomesintheN400window(300–500ms)oftheadjectiveaswellasthenoun.WewillrstinspecttheERPselicitedbytheadjectivestoobtainalexicalconcretenesseffectbaseline.WeexpecttoseelargerN400stoconcreteadjectives(easilyexperiencedwiththesenses)inboththeCLandMETconditionscomparedtotheabstractadjectives(noteasilyexperiencedwiththesenses)intheALcondition.Iftheconcretenessoftheadjectivedrivestheprocessingandinterpretationofthenounphrase,thenweexpecttoseeanN400concretenesseffectatthenounsuchthatCLexpression,butalsotheMETexpressions,exhibitlargerN400sthanALexpressions(CL=MET�AL).Conversely,ifitistheabstractnessoftheemergentconcepttowhichthenounphraserefersratherthantheabstractnessoftheadjectivepersethatdrivesprocessingandinterpretation(suchthattheMETnounisprocessedasifitfollowedanabstractadjective)thentheMETandALnounswouldelicitequivalentlyreducedN400amplitudes(CL�MET=AL).If,however,thesystemconcurrentlydistinguishesbetweenemergentconcreteness,andbetweenabstractconceptsthatareliteralvs.thosethataremetaphorical,thentheN400tometaphorsmaybeevenlargerthantheN400forabstractexpressions(MET�AL)duetoincreasedprocessingdemandsofunderstandinganovelexpressionformedbyanadjectivereferringtoaphysicaltraitthatanouncannotliterallypossess.ThisoutcomewouldconvergewiththeERPmetaphorliterature,andwiththedifferentialsensitivityoftheN400toindependentfactorsofconcretenessandeaseofprocessing/expectancy.MATERIALSANDMETHODSSTIMULITocreatethetwo-wordexpressionsusedintheERPstudy,eachof212nounswascombinedwith3differentadjectivestoform636novelwordpairs.Thenounswerepolysemousinthatmetaphorical(MET),concreteliteral(CL)orabstractliteral(AL)expressionscouldresultfrommodicationbythedifferentadjectives.ExamplesofthestimulicanbeseeninTable1.TheALwordpairsconsistedofabstractadjectivesmodifyingnounstoformexpressionsreferringtoabstractconcepts.IntheCLandMETconditions,adjectiveswereconcrete,butintheMETconditionadjectivesmodiednounsinanon-literalmanner:43%oftheadjectivesweresharedacrossthesetwoconditions.CLexpressionsreferredtoentitieseasilyexperiencedbythesenses,whereasALandMETexpressionsreferredtoentitiesnoteasilyexperiencedwiththesenses.Wordpairsweredesignedtobemeaningfulbutnovel,withnoveltycontrolledforbycorpusmeasures.Allwordpairsappeared4timesorlessintheBNCandtheprobabilityofanounfollowinganadjectivewaslessthan0.01.Semanticrelatednessbetweenconstituentsofexpressionswaslow,asmeasuredbyLatentSemanticAnalysis(LSA)(M=0.11,SD=0.11).Toensurethatstimuliwereconsistentwiththedenitionsabove,allwordpairswereratedinanonlinenormingstudyby90UCSDstudentsnotparticipatingintheERPstudy.Wordpairswereratedalongthreedimensions(concreteness,literalness,andmeaningfulness)onsevenpointLikert-scales:(1:notatall—7:completely).Thethreetasksassignedrandomlytoindividualwordpairswere:(1)“Howeasyisittoexperiencewiththesenses?”;(2)“Howliteralisit?”;and(3)“Howmeaningfulisit?”Wechosealiteralnessratinginordertoavoidtheexplanationordenitionof“metaphorical”and/or“gurative”,suspectingthatitmightbeeasiertodeterminewhethersomethingismeantliterallythanguratively.Eachparticipantsaweverywordpairbutratedindividualexpressionsalongonlyonedimension.Acrossparticipants,allwordpairswereratedforalldimensions.PairsinwhichtheCLexpressionwasratedmoreabstractthantheALexpression,orforwhichtheMETwasratedmoreliteralthantheCLorALexpressions,wereexcluded.Ofthe212normeditems,thetworatedleastmeaningfulwerediscarded.Oftheremaining210items,thehalf(105)ratedmostmeaningful(andmostliteralandmostmetaphoricalforthecorrespondingconditions)wereusedasstimuli,withtherestassignedtobellers.ItemnormingstatisticsaresummarizedinTable2.Usingthesametargetnounsineachconditionensuredthatnounlexicalfactorswereidenticallymatched(i.e.,nodifferencesintermsoffrequency,lengthorotherpsycholinguisticmeasure).Aone-wayANOVArevealedsignicantdifferencesbetweenconditionswithrespecttoconcreteness,F.2,312/=162.3,p0.001,2P=0.51,literalness,F.2,312/=387.9,p0.001,2P=0.71,andmeaningfulness,F.2,312/=114,p0.001,2P=0.42.Levene'stestofequalityoferrorvarianceswassignicantforconcreteness,F.2,312/=7.84,p0.001,andmeaningfulness,F.2,312/=5.04,p0.01,andtherewasastrongtrendforliteralness,F.2,312/=3,p=0.051.Therefore,theTamhaneposthoctestwasusedforpairwisecomparisons.Allconditionsweresignicantlydifferentfromeachotherinconcreteness(p0.001),andliteralness(p0.05),whileinmeaningfulnessALandCLexpressionswerenotsignicantlydifferent,withonlyMETdifferingfromtheotherconditions(p0.001,althoughallconditionswerestillabove4,themiddleofthescaleused).Participantsreadeachadjective-nounpairfollowedbyaprobewordthatwaseitherrelatedorunrelatedtothetwo-wordphrase.ExamplesofstimuliandrelatedprobewordsareshowninTable3.Table2|Means(standarddeviations)ofstimulusproperties. METCLAL Concreteness3.46(0.73)4.89(0.55)3.85(0.46)Literalness3.03(0.58)5.37(0.72)5.13(0.7)Meaningfulness4.14(0.74)5.31(0.64)5.33(0.56)LSA0.08(0.09)0.14(0.13)0.11(0.09) FrontiersinHumanNeurosciencewww.frontiersin.orgFebruary2015|Volume9|Article28|4 Forgácsetal.MetaphorsarephysicalandabstractTable3|Examplestimuliandrelatedprobewords. ConditionAdjectiveNounProbe METFluffySpeechExaggerationCLLoudMeetingDisputeCLScaryMovieThrillMETVelvetyLakeBeautyALUnknownPathHikingCLSaltyEarthCoastCLLuxuriousPartyTuxedoMETRustyMovesSports ERPPARTICIPANTSForty-twoUCSDvolunteers(18females)participatedforcoursecreditorwerecompensatedat7h.Participantswereright-handed,nativeEnglishspeakerswithnormalorcorrected-to-normalvision,rangingfrom18–29yearsold(M=21).Ofthe42participants,7wereexcludedfromfurtheranalysisduetoexcessiveeyeblinkormovementartifacts,whichleftaremaining35participantswhosedatawecontinuedtoexamine.PROCEDURETheexperimentwasconductedaccordingtohumansubjectprotocolsapprovedbytheUniversityofCalifornia,SanDiegoInstitutionalReviewBoard.Allparticipantsprovidedtheirinformedconsentinwritingbeforeparticipatingintheexperiment.ERPswererecordedinasinglesessioninasound-attenuated,electricallyshieldedchamber.ParticipantssatonemeterinfrontofaCRTmonitorandreadadjective-nounpairsfollowedbyprobewords.Participantsusedtwohand-heldbuttonstoindicatewhethertheprobeword(e.g.,“leader”)wasrelatedtotheadjective-nounpair(e.g.,“respectedperson”).Importantly,thistechniqueencouragedparticipantstocomprehendthenovelmetaphoricalexpressionsinagurativeratherthanaliteralsense.Responsehandwascounterbalancedacrossparticipantsandlists.StimuliwerecentrallypresentedinwhiteArial26pointfontonablackbackgroundonaCRTmonitor.Participantscompleted6blocksof35itemseachwithshortbreaksbetweenthem.Eachtrialstartedwithablankscreen(1000ms),followedbyaxationcross“+”(1000ms).Theadjectiveappearedcentrallyfor200ms,followedbya300msblankscreen,followedbythenounfor200ms,followedbya1500msblankscreen,andnallyaprobewordappearedfor200ms.After800msfollowingtheprobeonset,aquestionmark“?”wasdisplayeduntilparticipantsrespondedwithabuttonpress.Asmallreddotwaspresentedcentrallyandslightlybelowthetextthroughoutthetrials,exceptduringthequestionmarkandtherst1000msblankscreen;participantswereinstructednottoblinkwhenitwaspresent.Participantssawall105targetnounsonce,andeachwaspairedwithasingleadjectiveonce,resultingin35itemsfromeachcondition,alongwith105llerexpressions.Itemswerearrangedin5differentliststoavoidordereffects.Eachofthe5listswasseparatedinto3sublistssothateachnounwaspairedwithall3adjectivesacrossparticipants.EEGRECORDINGTheelectroencephalogram(EEG)wasrecordedfrom26electrodesarrangedgeodesicallyinanElectro-cap,eachreferencedonlinetoanelectrodeovertheleftmastoid.Blinksandeyemovementsweremonitoredfromelectrodesplacedontheoutercanthiandundereacheye,alsoreferencedtotheleftmastoid.Electrodeimpedanceswerekeptbelow5K2.TheEEGwasampliedwithGrassamplierswithabandpassof0.01–100Hzandwascontinuouslydigitizedatasamplingrateof250samples/second.DATAANALYSISTrialscontaminatedbyeyemovements,excessivemuscleactivity,oramplierblockingwererejectedoff-linebeforeaveraging:thesetrials(8.3%forMET,10.6%forCL,and10.1%forAL)wereexcludedfromfurtheranalysis.Datawerere-referencedoff-linetothealgebraicmeanoftheleftandrightmastoidsandaveragedforeachexperimentalcondition,time-lockedtoadjectiveonsets.ERPswerecomputedforepochsextendingfrom500mspre-to1500mspost-adjectiveonset,usingapre-stimulusbaselineof500ms.Sincewewereinterestedintheprocessingofthetwowordadjective-nounexpression,webaselinecorrectedonlypriortotheadjective,practicallytreatingthetwo-wordcombinationasoneexperimentalunit.ANOVAswereusedtoanalyzemeanamplitudeERPsover6medialcentralelectrodes(MiCe,MiPa,RMCe,LMCe,LMFr,RMFr)whereconcretenesseffectsintheN400timewindowarecommonlyobserved:thesewerethesameelectrodesitesoverwhichadjectiveconcretenesseffectswereassessedtodetermineinclusioninstatisticalanalyses.Basedontheliterature,weanalyzedconcretenesseffectsinthefollowingtimewindows:(1)anadjectiveN400timewindow(300–500mspost-adjectiveonset);and(2)anounN400timewindow(300–500mspost-nounonset).RESULTSBEHAVIORALRESULTSOverallresponseaccuracy(M=88%,SD=6%)forbuttonpressesindicatingwhetherornottheprobewasrelatedtothewordpairsuggestedthatthewordpairswerereadforcomprehension.ERPRESULTSAverageERPsforall35participantsareshowninFigure1.AdjectiveN400(300–500mspost-adjectiveonset)AnANOVAwith3levelsofwordtypeand6levelsofelectrodelocationrevealedamaineffectofwordtype,F.2,68/=10.65,p0.001,withCLandMETadjectivesshowinggreaterN400meanamplitude(�1.78mVand�2mV,respectively)thanALadjectives(�0.67mV).PlannedpairwisecomparisonsindicatedthatthemeanamplitudesforCLandALadjectivesandforMETandALadjectivesweresignicantlydifferent(p0.001).OurALandCLconditionswerebasedonadjective-nounpairconcretenessratings.However,toensurethattheselabelsalsomatchedtheconcretenessoftheadjectivesalone,weobtainedconcretenessratingsforadjectivesintheCLandALconditionsfromBrysbaertetal.(2013).For46itemsnotfoundinthedatabase,concretenessratingswerecollectedfrom7UCSDundergradstudentswhodidnotparticipateintheERPstudy. FrontiersinHumanNeurosciencewww.frontiersin.orgFebruary2015|Volume9|Article28|5 Forgácsetal.Metaphorsarephysicalandabstract FIGURE1|TheleftandrightpanelsshowsixcentralscalpelectrodesfromwhichERPdatawererecordedandstatisticallyanalyzed(locationsindicatedwithanXontheschematicarrayof26scalpelectrodes).InthelefthalfofthegurearegrandaverageERPs(N=35)forthe3experimentalconditions.OntherighthalfofthegurearegrandaverageERPs(N=35)againfortheMETcondition,nowcontrastedwiththelowerhalfoftheexpressionconcretenessratings(fromtheALcondition)andthetophalfoftheexpressionconcretenessratings(fromtheCLcondition).Targetnounonsetoccurs500msfollowingadjectiveonset. Adjectivesweresortedintohighandlowconcretenessconditionsusingamediansplit.AnANOVAwith2levelsofwordtype(highandlowconcretenessadjectives)and6levelsofelectrodelocationrevealedamaineffectofwordtype,F.1,34/=15.59,p0.001,withhighconcretenessadjectiveselicitingagreaternegativity(meanamplitude=�1.73mV)thanlowconcretenessadjectives(�0.62mV).Astheseresultswerenearlyidenticaltotheresultsforourlabeledconditions,weassumethatthedifferencebetweenALandCLadjectivesindeedreectslexicalconcreteness.NounN400(800–1000mspost-adjectiveonset)Aconcretenesseffectintheexpecteddirectionisvisible,withnounsintheCLconditionelicitingalargerN400thanALnouns.METnounsappeartobepatterningwithCLnouns,alsoelicitingalargerN400relativetoALnouns.However,anANOVAwith3levelsofwordtypeand6levelsofelectrodelocationshowednomaineffectofwordtype,F.2,68/=1.24,p=0.3.Inordertoincreasethesensitivityoftheconcretenessmanipulation,wesortedthedatabasedonpairedconcretenessratingsintothemostconcreteandleastconcreteitemswithinconditions.Themostconcrete(tophalfofCL)andleastconcrete(bottomhalfofAL)itemswerecomparedtoMETitemsinorderobtainaclearerpatternofconcretenesseffects—iftheywereindeedpresentinthedata.AnANOVAwith3levelsofwordtype(MET,CL-high,andAL-low)and6levelsofelectrodelocationrevealedamaineffectofwordtype,F.2,68/=4.38,p0.05,withCL-highandMETnounsshowinggreatermeanN400amplitude(�1.39mVand�1.17mV,respectively)thanAL-lownouns(meanamplitude=�0.29mV)(Figure1).PairwisecomparisonsindicatedthatthedifferencebetweenCL-highandAL-lownounsandMETandAL-lownounswasstatisticallysignicant(p0.05forbothcomparisons).ThustheN400patternatthenounresemblesthatattheadjective.LikethemediansplitoftheCLandALconditions,wesplitMETitemsbasedonparticipantratingsofpairconcreteness.Thesetwoconditions,MET-highandMET-low,wereanalyzedinordertobetterunderstandhowmetaphoricalitemsmaybeprocessedonthebasisoftheirratedconcreteness.AnANOVAwith2levelsofwordtype(MET-highandMET-low)and6levelsofelectroderevealedasignicantmaineffectofwordtype,F.1,34/=5.98,p0.05:theMET-lowgroupwasassociatedwithalargerN400meanamplitude(�1.62mV)thantheMET-highgroup(�0.53mV).WenextcomparedthesetwoMETgroupstothehighandlowCLERPs(Figure2).First,thehigh-concretenessMETgroupwascomparedtothemostabstractandmostCLconditionsdescribedabove.AnANOVAwith3levelsofwordtype(MET-high,CL-high,andAL-low)and6levelsofelectrodelocationrevealedamaineffectofwordtype,F.2,68/=3.18,p0.05,withMET-highnounsshowingareducedN400meanamplitude(�0.53mV)comparedtoCL-highnouns(�1.39mV).PairwisecomparisonsshowedthatthedifferencebetweenMET-highandCL-highnounswasborderlinesignicant(p=0.08)andtherewasnostatisticaldifferencebetweenMET-highandAL-lownouns(p=0.58). FrontiersinHumanNeurosciencewww.frontiersin.orgFebruary2015|Volume9|Article28|6 Forgácsetal.Metaphorsarephysicalandabstract FIGURE2|TheupperpanelsshowERPdataatarepresentativemidlinecentralelectrode(MiCe,akaCz).TheadjectiveN400(300–500ms)andnounN400(800–1000ms)timewindowsareindicatedwithshading.Ontheleft,theERPsassociatedwiththelowconcreteness-ratedMETexpressionsarecontrastedwiththelowerhalfoftheliteralexpressionconcretenessratings(fromAL)andthehigherhalfoftheliteralexpressionconcretenessratings(fromCL);ontherightisthesamecomparisonfortheERPsassociatedwiththehighconcreteness-ratedMETexpressions.InthelowerpanelareN400meanamplitudes(fromthedataintheupperpanels)averagedacrossthe6centralchannelsindicatedinFigure1.ErrorbarsindicateSEM. Second,thelow-concretenessMETgroupwascomparedtothemostabstractandmostCLconditionsdescribedabove.AnANOVAwith3levelsofwordtype(MET-low,CL-high,andAL-low)and6levelsofelectrodelocationrevealedamaineffectofwordtype,F.2,28/=4.10,p0.05,withMET-lownounsshowingincreasedN400meanamplitude(�1.62mV)comparedtoAL-lownouns(�0.29mV).PairwisecomparisonsshowedthatthedifferencebetweenMET-lowandAL-lownounswasstatisticallysignicant(p0.05)andtherewasnostatisticaldifferencebetweenMET-lowandCL-highnouns(p=0.66).Toensurethattheobserveddifferencesatthenounarenotmerelyspilloverfromtheadjectives,weconductedthreeANOVAsasaboveintheadjectiveN400timewindow(300–500mspost-adjectiveonset).ComparingMET(�2.01mV),CL-high(�2.35mV),andAL-low(�0.8mV)revealedamaineffectofwordtype,F.2,68/=8.73,p0.001,withasignicantdifferencebetweenAL-lowandbothCL-high(p0.001)andMET(p0.01).TheANOVAincludingMET-high(�1.71mV),CL-highandAL-lowalsoshowedamaineffectofwordtype,F.2,68/=6.98,p0.01,whereonlyMET-highwasdifferentcomparedtoAL-low(p0.05).TheANOVAwithMET-low(�2.32mV),CL-high,andAL-lowlikewiserevealedamaineffectofwordtype,F.2,68/=7.64,p0.01,whereonlyMET-lowwasdifferentfromAL-low(p0.01).Insum,whilethepatternofN400effectsatthenounmimickedthatattheadjectiveinthe(highvs.low)literalconditions,thiswasnotthecaseforthehighvs.lowMETconditions,whichreversedtheirdirectionfromadjectivetonoun.DISCUSSIONInthecurrentstudyweexaminedthereal-timeprocessingofnovelmetaphorical(“stickymeeting”),AL(“constructivemeeting”),andCL(“loudmeeting”)two-word(adjective-noun)expressions.Wereplicatedthewell-knownN400lexicalconcretenesseffectontheinitialadjectivesofthetwowordexpressions.AreliableconcretenesseffectalsoemergedforthenounsoftheliteralexpressionswhenthemostCLexpressionswerecomparedwiththemostALexpressions—despitetheabsenceofanydifferenceinratedlexicalnounconcreteness.WealsofoundthatonaveragetheN400tothemetaphoricalexpressionspatternedwiththattothemostCLexpressionsratherthanwiththattothemostALexpressions,contrarytowhatweexpected.Upondividingthemetaphoricalexpressionsintomoreconcretevs.moreabstractsubgroupsbasedonpairconcretenessratingswefoundthat,paradoxically,themoreabstractsubgroupofmetaphorswereassociatedwithalargerN400thannotonlythemostALexpressionsbutalsothemoreconcretemetaphorexpressions. FrontiersinHumanNeurosciencewww.frontiersin.orgFebruary2015|Volume9|Article28|7 Forgácsetal.MetaphorsarephysicalandabstractTheN400concretenesseffectontheprenominaladjectivesresemblesthatreportedforsinglewords(Huangetal.,2010;Rabovskyetal.,2012;AmselandCree,2013;Barberetal.,2013).Thisndinghasbeenhypothesizedtoreectactivationofarichernetworkofsemanticrepresentations(orgreateractivationwithinagivennetwork)duringtheprocessingofconcretevs.lessconcretewords.Withintheliteralexpressionssplitbypairwiseconcreteness,targetnounselicitedaprolongednegativitystartingintheN400timewindowthatvariedinamplitudewiththeratedconcretenessoftheexpression.Huangetal.(2010)hadshowedthatmodifyinganouninamoreconcretevs.moreabstractmannercanleadtoaconcretenesseffect(e.g.,for“book”in“interestingbook”vs.“thickbook”).LikeHuangetal.(2010),wendthatforliteral(non-metaphorical)expressions,theconcretenessofanadjectiveseemstodeterminetheconcretenesseffectonasubsequentnoun,atleastattheextremes(thedirectionoftheireffectcannotbedirectlycomparedwithoursastheyemployedvisualhalfeldpresentationandadifferenttask,amongotherdifferences).Thismanifestationoftheconcretenesseffectatthenounisparticularlystrikinggiventhatthenounsthemselvesdonotdifferonthisverymeasure(ofconcreteness).UnlikesomeERPstudies(WestandHolcomb,2000;Huangetal.,2010),wendnoevidencethatourconcretenesseffectsreectimagery-relatedprocessesoverandabovetheprocessesthatroutinelyinuenceN400amplitude.Specically,ourconcretenesseffectsatneithertheadjectivesnorthenounsofliteralexpressionsexhibitedmorefrontaland/orrighthemisphericdistributionsthanthecanonicalN400distributiontowrittenwords.Aswealreadynoted,ourN400concretenesseffectattheadjectiveisconsistentwithaproposedrichnessoftheactivatedconceptualrepresentationsinalexico-semanticsystem(e.g.,Holcombetal.,1999;Barberetal.,2013).Followingthesamelogic,ourconcretenesseffectfortheconcretelyvs.abstractlymodiednounscouldreecttherichnessoftheemergenthigher-levelconceptualrepresentation—atleastfortheliteralexpressions.Forexample,reading“thickbook”inordertodetermineitsrelationtoanupcomingprobewordcouldactivatearichernetworkoffeaturesoftheconcreteconceptBOOKthan“interestingbook”.Thispossibilitydoesnotnecessarilyimplicatesensory/motoractivationsfortheinterpretationoftheCLexpressions,asitcouldjustaswellreectgreateractivationwithinanamodalsemanticsystem(PlautandShallice,1993).Ourresultsfortheliteralexpressionsextendtheresultsofsinglewordstudies(e.g.,Barberetal.,2013)andPaivio(2007)dualcodingtheoryinsofarastheydemonstratethatconcretenessneednotbeastrictlylexicalproperty(i.e.,peggedtosinglewordmeanings),butanemergentpropertyofhigher-levelconceptsaswell.Forthemetaphoricalexpressions,however,ourN400datapatterndivergesfromthatofourliteralexpressions,aswellasfromHuangetal.(2010).WhenwecomparethemetaphornounN400stothenounN400softhemostabstractandmostCLexpressions,ourdata(atrstglance)suggestthattheconcretenesseffectatthenounisdrivenbythelexicalconcretenessoftheadjective,asseemstobethecaseintheliteralexpressionsandinHuangetal.(2010).Totheextentthatconcretenesseffectsatthenounaremerelyanextension(spillover)oftheERPconcretenesseffectattheadjective,thispatternshouldremainunchangedforallmetaphoricalexpressions.However,whenwedivideourmetaphoricalexpressionsbypairedconcreteness,themoreconcretemetaphorsappearedtobeprocessed(i.e.,looked)morelikeALexpressions,andthemoreabstractmetaphorslookedmorelikeCLexpressions.Inotherwords,theelicitednegativityisreversedwithinthemetaphoricalexpressions,withexpressionsratedasmoreabstractelicitinglargernounN400sthanthoseratedasmoreconcrete.IfthenegativityformetaphorsobservedintheN400timewindowwereaconcretenesseffectproper,highconcretenessmetaphorsshouldhaveelicitedagreaternegativitythanlowconcretenessmetaphors.Yetthemoreconcreteametaphorwasrated,thesmallerthenegativityitelicited.Ataminimum,thispatternofresultsdemonstratesthattheprocessingofthenounsinthemetaphoricalexpressionscannotbedrivenstrictlybyeitherlexicalconcretenessorhigher-levelemergentconcreteness.Ofcourse,concretenessisonlyoneofmanyfactorsknowntoinuencetheERP,andinparticulartheN400.LessliteralandmorenovelexpressionshavebeenshowntoelicitlargerN400s.TargetwordsinnovelmetaphorsusuallyelicitlargerN400amplitudesthantargetwordsinliteralexpressions(CoulsonandVanPetten,2002,2007;Arzouanetal.,2007;Laietal.,2009),andrelativetoconventionalmetaphorstheyelicitlargernegativitiesslightlylateraswell,post-N400(Arzouanetal.,2007;Laietal.,2009).Contrathismonotonicrelationship,ourhighconcretenessmetaphoricalexpressionconditiondidnotelicitlargerN400sthanourALcondition,andstrikingly,wasreducedincomparisonwithCLexpressions(seeFigure2).Eventhoughtheirconcretenesspositivelycorrelatedwithliteralness(r.105/=0.59,p0.001)andthemoreabstractmetaphoricalexpressionsdidelicitalargerN400thanthemoreALexpressions,theydidnotdifferfromthemoreCLexpressions.IftheincreasednegativityformetaphorsincomparisonwithmoreALexpressions(Figure1)wereduetometaphoricityperse,itshouldhavemanifestforallmetaphors,butitdidnot.Onereasonwhyourresultsdivergedinpartfromotherinvestigationsofmetaphoricallanguagemaybethatourexpressionswerematchedonnoveltyacrossconditions,whereasintheaforementionedstudiesonlythenovelmetaphorswereunfamiliar.Asaresult,allthreeofourexperimentalconditionsmayhaveinvokedsomeadditionalconstructiveorintegrativeprocessing(linkedinpreviousreportstothepost-N400,sustainednegativity).Onthispossibility,ourndingofequivalentN400sformoreALexpressionsandmoreconcretemetaphoricalexpressions(despitealexicalconcretenessdifferenceattheadjectives)suggeststhatreadersneednotnecessarilyconstructtheliteral(i.e.,physical)interpretationofanovelmetaphoricalexpressionbeforeunderstandingitsgurativemeaning.Thisinterpretationisconsistentwithparallelmodelsofmetaphorcomprehension(Glucksberg,2003):theabstract,gurativemeaningofmetaphorsmightbereadilyanddirectlyavailable,asalsoinferredbyBlaskoandConnine(1993).Ourndingsargueagainstothermodelsofserialprocessingofmetaphors FrontiersinHumanNeurosciencewww.frontiersin.orgFebruary2015|Volume9|Article28|8 Forgácsetal.Metaphorsarephysicalandabstractaswell.Forexample,Giora(1997,2003)proposedthatthecomprehensionofnovelmetaphorsrequirestherejectionofasalient,literalmeaningbeforearrivingatanon-salientmetaphoricalmeaning.Ifweassumethatserialprocessingwouldresultinnon-identicalERPresponses,thelackofdifferencesbetweenmoreconcretemetaphorsandmoreALwordpairsdoesnotsupporttheserialprocessingassumption(unlessthelatterhavebothasalientandanon-salientliteralmeaning).Moreover,ourresultsindicatethatgurativemeaningneednotbedirectlyderivativeofthephysicalaspectsofverbalexpressions,butrathermayattimesemergeabstractlyatleastbythetimewindowoftheN400,awell-establishedmarkerofsemanticanalysis.Consequently,ourdatamightposeachallengetostrongviewsofembodiedcognition(e.g.,LakoffandJohnson,1999).Onastrongembodimentview,sensorimotorsourcedomains(e.g.,physicalsensationofwarmth)areactivatedinparallelwithmoreabstracttargetdomains,soastoprovidestructureandsemanticcontentforunderstandingmetaphoricalexpressions(asin“warmsmile”).GalleseandLakoff(2005)proposethat“graspinganidea”involvessomeofthesamemotoractivationsas“graspingabanana”.Inotherwords,duringconceptualintegrationbothconceptualdomainsshouldbeactiveatthesametime;ifso,weexpectedtoseethisreectedinacanonicalN400concretenesseffectatthenoun.Wedidnot.Amongotherpossibleinterpretations,theapparentabsenceofaprocessingdifferencebetweenmoreconcretemetaphoricalexpressionsandmoreALexpressionsatthenouncouldbetakentomeanthatbythetimesomemetaphoricalmeaningsareconstructed,physicalaspectsofthewordsmightnolongerbeplayingatangibleroleincomprehension.Acumulativeconclusionthusfaristhatneitherconcretenessnormetaphoricityperse,canfullyaccountfortheprocessingdifferencesamongournovelliteralandmetaphoricalexpressions,atleastintheN400timewindow.Wecanspeculateaboutwhatadditionalfactormaybeinuencingourresults.Amongmetaphoricalexpressions,ratedpairconcretenesscorrelateswithmeaningfulness,r.105/=0.53,p0.001(aphenomenonobservedalsobyForgácsetal.,2014).Thus,thegreaterN400elicitedbythelessconcretemetaphorscouldreectthetypicalinverserelationshipbetweencontext-drivenexpectancyandN400amplitude,ratherthanprocessesspeciceithertolexicalconcreteness,ortogurativemeaning.Perhapsthemetaphoricalexpressionsratedmoreconcreteandmoremeaningfulweremorelikelytoincreasesemanticexpectanciesfortheupcomingnoun.OurndingthatmoremeaningfulandmoreconcretemetaphoricalexpressionsseemtobeprocessedlikemoreALexpressionstsnicelywithanewlyemergingpictureofmetaphorcomprehension.Onthisview,thereisnoempiricalreasontoassumethatprocessingofmetaphorsinvokesspecialprocessesthatarenotalsorequiredforcomprehendingliterallanguage.Indeed,despitelongheldassumptionsaboutthespecialroleoftherighthemisphereingurativelanguage,recentresultssuggestthatitdoesnotplayaprivilegedroleinmetaphorcomprehensionafterall(Rappetal.,2004,2007;CoulsonandVanPetten,2007;Bohrnetal.,2012;Forgácsetal.,2012,2014).Likewise,thereisnosupportfortheproposalthatgurativemeaningofnovelmetaphoricalexpressionsproceedsonlyafterattemptsat(salient)literalmeaningfail(Forgácsetal.,2014).Forgács(2014)hasdevelopedanoveltheoreticalframeworkformetaphorcomprehension—AbstractConceptualSubstitution(ACS).Accordingtothisview,foraninitialmetaphoricalinterpretationitmightsufcetosubstitutethevehicleterm(“uffy”in“uffyspeech”)withoneofitsabstract,non-physicalproperties,priortoanysystematicmapping,orstructuralalignment,etc.ThistakeonmetaphorinterpretationiscloselyrelatedtothatofSperberandWilson(2008),andthelexicalpragmaticaccountofWilsonandCarston(2007),Carston(2010).Theyproposethatmetaphorsarepartofacontinuumoflooselanguageuse(togetherwithhyperboleandapproximation,forexample),whichareunderstoodviathegenerationofadhocconcepts.Forexample,intheexpression“uffyspeech”theconceptFLUFFYistransformedintoFLUFFY*,whichisconceptuallybothbroaderandnarrower(i.e.,moregeneralandmorespecicatthesametime),inwaysleftasyetunspecied,thantheoriginal,encoded,lexicalconcept.Forgácsspeciesthisbroadening/narrowingintermsoftheabstract-concretedimension:FLUFFY*couldbroadenthelexicalconceptFLUFFYbyactivatingmoreofitsabstractproperties(e.g.,superuous,cushy,etc.),butnarrowthelexicalconceptbysuppressingallofitsconcrete/physicalproperties(e.g.,physicallyprotrudinguff,textile,texture,etc.).ThisapproachissimilartoGlucksberg(2003)categoryassertionview,butdoesnotrelyonthecreationofsuperordinateadhoccategoriesoronthegenerationofadhocconcepts.Instead,itmightsufcetoconceptuallysubstitutethemostrelevant(i.e.,contextuallymostactivated)abstractpropertyforthevehicleterm(“uffy”),creating“superuous,cushyspeech”.Thisisnotmerelyaparaphrase,however,becauseexpressingsuperuouswith“uffy”bringsalongwithitseveralcognitiveconsequences,suchasdeniability,negotiability,etc.,muchlikeindirectspeech(cf.Pinkeretal.,2008).Thelackofaconcretenesseffectatleastforthemoremeaningful,moreconcretemetaphoricalexpressionsvs.themoreALexpressionsisconsistentwiththisabstractsubstitutionviewinthatthesystemseemstosubstituteabstractbutnotconcretepropertiesforthevehicleterminournovelmetaphoricalexpressions.Tosumup,ourresultssuggestthattheconcretenesseffectdoesnotmerelyreecttheconcretenessofindividualwords,butmayalsobesensitivetotheconcretenessofhigher-levelconceptualinformation.AtleastintheN400timewindow,andseeminglyonlyformoremeaningful,moreconcreteadjectivalmetaphors,ourndingssuggestthatmetaphoricallanguagemaybeprocessedandpresumablyunderstoodinanabstractmanner,despitetheconcretenatureofitsconstituentparts.Inconclusion,itappearsthatcomprehendingcertainmetaphoricalexpressionscreatedfromphysicalconceptsandwordscanbeasreadilygrasped,andasrapidlydigestedasALexpressions,althoughnotstrictlydrivenbyconcreteness.AUTHORCONTRIBUTIONSBálintForgácsconceivedresearch;BálintForgács,MeganD.Bardolph,BenD.Amsel,KatherineA.DeLong,andMartaKutasdesignedresearch;BálintForgács,MeganD.Bardolph FrontiersinHumanNeurosciencewww.frontiersin.orgFebruary2015|Volume9|Article28|9 Forgácsetal.Metaphorsarephysicalandabstractperformedresearch;BálintForgács,MeganD.Bardolph,BenD.Amsel,KatherineA.DeLong,andMartaKutasanalyzeddata;BálintForgács,MeganD.Bardolph,BenD.Amsel,KatherineA.DeLong,andMartaKutaswrotethepaper.ACKNOWLEDGMENTSWewouldliketothanktheinvaluablehelpofTomUrbach,PhD,inexperimentaldesignandstimulusnorming,JamieAlexandre,PhD,inprogramming,GabrielDoyle,PhD,incorpuslinguisticmeasures,andLindsayCrissmanandAlexKuoindatacollection.BálintForgácswassupportedbyaFulbrightfellowship,andbyanEmergence(s)programgrantfromtheCityofParistoJuditGervain.TheresearchwasfundedbygrantNICHD22614toMartaKutas.REFERENCESAmsel,B.D.,andCree,G.S.(2013).Semanticrichness,concretenessandobjectdomain:anelectrophysiologicalstudy.Can.J.Exp.Psychol.67,117–129.doi:10.1037/a0029807Arzouan,Y.,Goldstein,A.,andFaust,M.(2007).Brainwavesarestethoscopes:ERPcorrelatesofnovelmetaphorcomprehension.BrainRes.1160,69–81.doi:10.1016/j.brainres.2007.05.034Barber,H.A.,Otten,L.J.,Kousta,S.T.,andVigliocco,G.(2013).Concretenessinwordprocessing:ERPandbehavioraleffectsinalexicaldecisiontask.BrainLang.125,47–53.doi:10.1016/j.bandl.2013.01.005Beeman,M.J.(1998).“Coarsesemanticcodinganddiscoursecomprehension,”inRightHemisphereLanguageComprehension:PerspectivesfromCognitiveNeuroscience,edsM.BeemanandC.Chiarello(Mahwah,NJ:Erlbaum),255–284.Binder,J.R.,andDesai,R.H.(2011).Theneurobiologyofsemanticmemory.TrendsCogn.Sci.15,527–536.doi:10.1016/j.tics.2011.10.001Blasko,D.,andConnine,C.M.(1993).Effectsoffamiliarityandaptnessonmetaphorprocessing.J.Exp.Psychol.Learn.Mem.Cogn.19,295–308.doi:10.1037//0278-7393.19.2.295Bohrn,I.C.,Altmann,U.,andJacobs,A.M.(2012).Lookingatthebrainsbehindgurativelanguage–Aquantitativemeta-analysisofneuroimagingstudiesonmetaphor,idiomandironyprocessing.Neuropsychologia50,2669–2683.doi:10.1016/j.neuropsychologia.2012.07.021Bowdle,B.,andGentner,D.(2005).Thecareerofmetaphor.Psychol.Rev.112,193–216.doi:10.1037/0033-295x.112.1.193Bréal,M.(1900).Semantics.(Mrs.H.Cust,trans.).NewYork:HenryHolt.Brysbaert,M.,Warriner,A.B.,andKuperman,V.(2013).Concretenessratingsfor40thousandgenerallyknownEnglishwordlemmas.Behav.Res.Methods46,904–911.doi:10.3758/s13428-013-0403-5Carston,R.(2010).Lexicalpragmatics,adhocconceptsandmetaphor:fromarelevancetheoryperspective.Ital.J.Linguist.22,153–180.Coulson,S.,andVanPetten,C.(2002).Conceptualintegrationandmetaphor:anevent-relatedpotentialstudy.Mem.Cognit.30,958–968.doi:10.3758/bf03195780Coulson,S.,andVanPetten,C.(2007).Aspecialrolefortherighthemisphereinmetaphorcomprehension?ERPevidencefromhemieldpresentation.BrainRes.1146,128–145.doi:10.1016/j.brainres.2007.03.008Desai,R.H.,Binder,J.R.,Conant,L.L.,Mano,Q.R.,andSeidenberg,M.S.(2011).Theneuralcareerofsensory-motormetaphors.J.Cogn.Neurosci.23,2376–2386.doi:10.1162/jocn.2010.21596Faust,M.,andMashal,N.(2007).Theroleoftherightcerebralhemisphereinprocessingnovelmetaphoricexpressionstakenfrompoetry:adividedvisualeldstudy.Neuropsychologia45,860–870.doi:10.1016/j.neuropsychologia.2006.08.010Forgács,B.(2014).FiguresofLanguageinCognitiveScienceintheLightofFigurativeLanguageProcessingintheBrain.Hungary:BudapestUniversityofTechnologyandEconomics.Availableonlineat:http://www.omikk.bme.hu/collections/phd/Termeszettudomanyi_Kar/2014/Forgacs_Balint/ertekezes.pdfForgács,B.,Bohrn,I.,Baudewig,J.,Hofmann,M.J.,Pléh,C.,andJacobs,A.M.(2012).Neuralcorrelatesofcombinatorialsemanticprocessingofliteralandgurativenounnouncompoundwords.Neuroimage63,1432–1442.doi:10.1016/j.neuroimage.2012.07.029Forgács,B.,Lukács,A.,andPléh,C.(2014).Lateralizedprocessingofnovelmetaphors:disentanglinggurativenessandnovelty.Neuropsychologia56,101–109.doi:10.1016/j.neuropsychologia.2014.01.003Gallese,V.,andLakoff,G.(2005).Thebrain'sconcepts:theroleofthesensory-motorsysteminconceptualknowledge.Cogn.Neuropsychol.22,455–479.doi:10.1080/02643290442000310Giora,R.(1997).Understandinggurativeandliterallanguage:thegradedsaliencehypothesis.Cogn.Linguist.8,183–206.doi:10.1515/cogl.1997.8.3.183Giora,R.(2003).OnourMind:Salience,ContextandFigurativeLanguage.NewYork:OxfordUniversityPress.Glucksberg,S.(2003).Thepsycholinguisticsofmetaphor.TrendsCogn.Sci.7,92–96.doi:10.1016/s1364-6613(02)00040-2Holcomb,P.J.,Kounios,J.,Anderson,J.E.,andWest,W.C.(1999).Dual-coding,context-availabilityandconcretenesseffectsinsentencecomprehension:anelectrophysiologicalinvestigation.J.Exp.Psychol.Learn.Mem.Cogn.25,721–742.doi:10.1037//0278-7393.25.3.721Huang,H.W.,Lee,C.L.,andFedermeier,K.D.(2010).Imaginethat!ERPsprovideevidencefordistincthemisphericcontributionstotheprocessingofconcreteandabstractconcepts.Neuroimage49,1116–1123.doi:10.1016/j.neuroimage.2009.07.031Jackendoff,R.(1997).TheArchitectureoftheLanguageFaculty.Cambridge,MA:MITPress.Jung-Beeman,M.(2005).Bilateralbrainprocessesforcomprehendingnaturallanguage.TrendsCogn.Sci.9,512–518.doi:10.1016/j.tics.2005.09.009Kounios,J.,andHolcomb,P.J.(1994).Concretenesseffectsinsemanticprocessing:ERPevidencesupportingdual-codingtheory.J.Exp.Psychol.Learn.Mem.Cogn.20,804–823.doi:10.1037//0278-7393.20.4.804Kousta,S.T.,Vigliocco,G.,Vinson,D.P.,Andrews,M.,andDelCampo,E.(2011).Therepresentationofabstractwords:whyemotionmatters.J.Exp.Psychol.Gen.140,14–34.doi:10.1037/a0021446Kutas,M.,andFedermeier,K.D.(2011).Thirtyyearsandcounting:ndingmeaningintheN400componentoftheevent-relatedbrainpotential(ERP).Annu.Rev.Psychol.62,621–647.doi:10.1146/annurev.psych.093008.131123Kutas,M.,andHillyard,S.(1980).Readingsenselesssentences:brainpotentialsreectsemanticincongruity.Science207,203–205.doi:10.1126/science.7350657Kutas,M.,VanPetten,C.,andKluender,R.(2006).“PsycholinguisticselectriedII:1994–2005,”inHandbookofPsycholinguistics,2ndEdn.edsM.TraxlerandM.A.Gernsbacher(NewYork:Elsevier),659–724.Lai,V.T.,andCurran,T.(2013).ERPevidenceforconceptualmappingsandcomparisonprocessesduringthecomprehensionofconventionalandnovelmetaphors.BrainLang.127,484–496.doi:10.1016/j.bandl.2013.09.010Lai,V.T.,Curran,T.,andMenn,L.(2009).Comprehendingconventionalandnovelmetaphors:anERPstudy.BrainRes.1284,145–155.doi:10.1016/j.brainres.2009.05.088Lakoff,G.,andJohnson,M.(1980).MetaphorsWeLiveBy.Chicago:UniversityofChicagoPress.Lakoff,G.,andJohnson,M.(1999).PhilosophyintheFlesh:TheEmbodiedMindandItsChallengetoWesternThought.NewYork:BasicBooks.Lee,C.L.,andFedermeier,K.D.(2008).Towatch,toseeandtodiffer:anevent-relatedpotentialstudyofconcretenesseffectsasafunctionofwordclassandlexicalambiguity.BrainLang.104,145–158.doi:10.1016/j.bandl.2007.06.002Levy-Drori,S.,andHenik,A.(2006).Concretenessandcontextavailabilityinlexicaldecisiontasks.Am.J.Psychol.119,45–65.doi:10.2307/20445318Paivio,A.(2007).MindandItsEvolution:ADualCodingTheoreticalApproach.Mahwah,NJ:LawrenceErlbaumAssociatesPublishers.Pinker,S.,Nowak,M.A.,andLee,J.J.(2008).Thelogicofindirectspeech.Proc.Natl.Acad.Sci.USA105,833–838.doi:10.1073/pnas.0707192105Plaut,D.C.,andShallice,T.(1993).Deepdyslexia:acasestudyofconnectionistneuropsychology.Cogn.Neuropsychol.10,377–500.doi:10.1080/02643299308253469Pynte,J.,Besson,M.,Robichon,F.H.,andPoli,J.(1996).Thetime-courseofmetaphorcomprehension:anevent-relatedpotentialstudy.BrainLang.55,293–316.doi:10.1006/brln.1996.0107 FrontiersinHumanNeurosciencewww.frontiersin.orgFebruary2015|Volume9|Article28|10 Forgácsetal.MetaphorsarephysicalandabstractRabovsky,M.,Sommer,W.,andAbdel-Rahman,R.(2012).Thetimecourseofsemanticrichnesseffectsinvisualwordrecognition.Front.Hum.Neurosci.6:11.doi:10.3389/fnhum.2012.00011Rapp,A.M.,Leube,D.T.,Erb,M.,Grodd,W.,andKircher,T.T.J.(2004).Neuralcorrelatesofmetaphorprocessing.BrainRes.Cogn.BrainRes.20,395–402.doi:10.1016/j.cogbrainres.2004.03.017Rapp,A.M.,Leube,D.T.,Erb,M.,Grodd,W.,andKircher,T.T.J.(2007).Lateralityinmetaphorprocessing:lackofevidencefromfunctionalmagneticresonanceimagingfortherighthemispheretheory.BrainLang.100,142–149.doi:10.1016/j.bandl.2006.04.004Sperber,D.,andWilson,D.(2008).“Adeationaryaccountofmetaphors,”inTheCambridgeHandbookofMetaphorandThought,edR.W.Gibbs(NewYork,NY:CambridgeUniversityPress),84–105.Swaab,T.Y.,Baynes,K.,andKnight,R.T.(2002).Separableeffectsofprimingandimageabilityonwordprocessing:anERPstudy.BrainRes.Cogn.BrainRes.15,99–103.doi:10.1016/s0926-6410(02)00219-7Tartter,V.C.,Gomes,H.,Dubrovsky,B.,Molholm,S.,andStewart,R.V.(2002).Novelmetaphorsappearanomalousatleastmomentarily:evidencefromN400.BrainLang.80,488–509.doi:10.1006/brln.2001.2610West,W.C.,andHolcomb,P.J.(2000).Imaginal,semanticandsurface-levelprocessingofconcreteandabstractwords:anelectrophysiologicalinvestigation.J.Cogn.Neurosci.12,1024–1037.doi:10.1162/08989290051137558Wilson,D.,andCarston,R.(2007).“Aunitaryapproachtolexicalpragmatics:relevance,inferenceandadhocconcepts,”inPragmatics,edN.Burton-Roberts(Basingstoke:PalgraveMacMillan),230–259.ConictofInterestStatement:Theauthorsdeclarethattheresearchwasconductedintheabsenceofanycommercialornancialrelationshipsthatcouldbeconstruedasapotentialconictofinterest.Received:15April2014;accepted:12January2015;publishedonline:10February2015.Citation:ForgácsB,BardolphMD,AmselBD,DeLongKAandKutasM(2015)Metaphorsarephysicalandabstract:ERPstometaphoricallymodiednounsresembleERPstoabstractlanguage.Front.Hum.Neurosci.9:28.doi:10.3389/fnhum.2015.00028ThisarticlewassubmittedtothejournalFrontiersinHumanNeuroscience.Copyright©2015Forgács,Bardolph,Amsel,DeLongandKutas.Thisisanopen-accessarticledistributedunderthetermsoftheCreativeCommonsAttributionLicense(CCBY).Theuse,distributionandreproductioninotherforumsispermitted,providedtheoriginalauthor(s)orlicensorarecreditedandthattheoriginalpublicationinthisjournaliscited,inaccordancewithacceptedacademicpractice.Nouse,distributionorreproductionispermittedwhichdoesnotcomplywiththeseterms. FrontiersinHumanNeurosciencewww.frontiersin.orgFebruary2015|Volume9|Article28|11