JOURNALOFMEMORYANDLANGUAGE31,485506(1992)
WhatDoidiomsRealtyMean? RAYMONDW.GIBBS,JR. Universifyof
California,SantaCruz Thedeadmetaphorviewof idiomaticitysuggests
thatidiomswereoncemetaphoricbut
havelosttheirmetaphoricityovertimeandnowareequivalenttosimpleliteralphrasessuch
thatblowyourstack=togetveryangry,crackthewhip=toexertauthority,and
spillthebeans=torevealasecret.Thepurposeofthepresentstudieswastodemon-
stratethatidiomsarenotdeadmetaphorsbuthavemorecomplexmeaningsthataremoti-
vatedbyconceptualmetaphorslinkingidiomphraseswiththeirfigurativeinterpretations.
Sixexperimentsarereportedthatexaminethedifferencebetweenidiomsandtheirliteral
paraphrases.Afuststudyexaminedpeoplesintuitionsaboutdifferentknowledgedomains
toillustratethatthemeaningsofidiomsareconsistentwiththesource-to-targetdomain
mappingsoftheconceptualmetaphorsthatmotivatethesephrasesfigurativemeanings.
ThedatafromExperiments2 and3 indicatethatpeopleviewidiomsas
havingmorecomplex
meaningsthandotheirroughly,equivalentliteralparaphrases.Experiments4through6
showthatidiomsaremostappropriatetouseandeasiesttocomprehendwhentheyare
encounteredindiscoursesituationsthatareconsistentwiththeentailmentsof
theconcep-
tualmetaphorsthatmotivatethesephrasesidiomaticmeanings.Thefindingsfromthese
studiessuggestthatidiomsarenotdeadmetaphorswithsimplefigurativeinterpretations.
Instead,idiomshavecomplexmeaningsthataremotivatedbyindependentlyexistingcon-
ceptualmetaphorsthatarepartlyconstitutiveof everydaythought.D ISZ
Academicpress,w. Oneofthemostpersistentideas inboth folkand
linguisticaccounts ofidiomsis thatthese phrases aredeadmetaphors.
Itiscommonlyassumed thatidiomswere
metaphoricalintheirorigins,buthavelost theirmetaphoricityovertime
and nowexist as frozen,semantic units oras deadmet-
aphors.Althoughmetaphors are lively,cre- ative,and resistant to
literalparaphrase,id- ioms aredead,hackneyed expressions that
areequivalentinmeaningtosimpleliteral
phrases.Tomanyscholars,classifyingan utteranceorphraseas
idiomaticis tan- tamounttoatheoreticalexplanationinit- self,since
itassumes thatidiomsaredead metaphorsandreallybelongtothewaste-
basket offormulasand phrases that are sep-
ThisresearchwassupportedbyGrantMH42980
fromtheNationalInstituteofMentalHealthandbya
FacultyResearchGrantfromtheUniversityofCali-
fornia,SantaCruz.Correspondenceandreprintre-
questsmaybesenttoRaymondW.Gibbs,Jr.,Pro-
graminExperimentalPsychology,ClarkKerrHall,
UniversityofCalifornia,SantaCruz,CA95064. arate fromthe
generativecomponentofthe grammar(Gibbs,1990, inpress).
Mostidiomdictionariesgivesimpledeli-
nitionsforidioms(Boatner,Gates, &Mak- kai,1975;
Cowie,Mackin,& McCaig,1983; Long&Summer,1979).Forinstance,
crack thewhipis definedas tobeincon- trol,spillthebeansmeanstomake
knownasecret,gotopiecesmeansto becomedistressed,andblowyourstack
means tobecomeveryangry.Ofcourse, dictionarydefinitionsdonot
necessarily re- flectthe complexityin peoplesmental rep-
resentationsforwordsandphrases.Yet
manysemantictheoriesassumethatthe meanings ofidiomsare best
representedby simpledefinitionsbecauseidiomsare
mostlydeadmetaphors(Cruse,1986; Palmer,1981).
Mycontention,contraryto thedeadmetaphorview,isthatidioms
havecomplexfigurativeinterpretations
thatarenotarbitrarilydeterminedbutare
motivatedbyindependentlyexistingcon- ceptualmetaphors
thatprovidethe founda- 485 0749-596X/92$5.00
CopyrightQ1992byAcademicPress,Inc.
AUtightsofreproductioninanyformreserved. 486 RAYMONDW.GIBBS,JR.
tionformuchofoureverydaythoughtand reasoning.
Therehasrecentlybeenagreatdealof
researchincognitivelinguisticsandpsy-
cholinguisticsthatquestionsthedeadmet-
aphorviewofidiomaticity(Fillmore,Kay,
&OConner,1988;Gibbs,1990;Gibbs&
Nayak,1991,1989;Gibbs&OBrien,1990;
Lakoff,1987;Langacker,1987;Nayak&
Gibbs,1990).Thisworkgenerallysuggests
thatmanyidiomsarenotsimple,dead
metaphors,butactuallyretainagooddeal
oftheirmetaphoricity.Forexample,the
figurativemeaningsofblowyourstackand
flipyourlidarespecificallymotivatedby
twoindependentlyexistingconceptual mappingsinlong-termmemory-MINDIS
ACONTAINERandANGERIS HEATEDFLUIDINACONTAINER.
Theseconceptualmetaphorsallowspeak-
erstorefertoideasaboutgettingangry
throughparticularinstantiationsofthe mappingfromasourcedomain(e.g.,
heatedfluid)ontoatargetdomain(e.g., anger).Speakersmakesenseofidi-
oms,suchasblowyourstackandflipyour
lid,preciselybecausetheirmeaningscanbe
motivatedbytheconceptualmappingsthat
linktheindividualwordsinidiomstotheir figurativemeanings.
Variousexperimentalstudieshaveinves-
tigatedthepsycholinguisticconsequences
ofthisconceptualviewofidiomaticity.
Someresearchshowedthatpeoplestacit
knowledgeofdifferentconceptualmeta- phors(e.g.,theMINDISACONTAINER,
ANGERISHEATEDFLUIDINACON- TAINER)constraintheirmentalimagesfor
idiomaticphrases(Gibbs&OBrien,1990).
Otherstudiesdemonstratedthatpeoples
metaphoricalunderstandingofemotion
concepts,suchasanger,joy,sadness,and
fear,facilitatesthecontext-sensitiveuse
andunderstandingofidiomsindiscourse
(Nayak&Gibbs,1990).Myaiminthe
presentstudieswastoshowthatidiomsare
notequivalentinmeaningtotheirliteral
paraphrases.Numerousstudiesreportthat
peopleprocessidiomsmorequicklythan
theydocorrespondingliteralphrases
(Gibbs,1980,1986;Gibbs&Gonzales,
1985;Gibbs,Nayak,&Cutting,1989;Or-
tony,Schallert,Reynolds,&Antos,1978;
Swinney&Cutler,1979).However,no
workhasspecificallyexaminedhowand
whyidiomsactuallydifferinmeaningfrom theirliteralparaphrases.
Considerthephrasesblowyourstack,
jlipyourlid,andhittheceiling.Whywould speakersuse
theseexpressionstomeanto getveryangry?Myhypothesisisthat
theseidiomsarenotequivalenttoasimple,
literalparaphrasesuchastogetveryan-
grybecausetheyhavecomplexsemantic
configurationsthataremotivatedbytwo conceptualmetaphors-MINDISACON-
TAINERandANGERISHEATED FLUIDINACONTAINER.TheMINDIS
ACONTAINERmetaphorispartofthe moregeneralCONDUITmetaphor(Reddy,
1979),andtheANGERISHEATmetaphor comesfromthecommonfolktheorythat
thephysiologicaleffectsofangerarein-
creasedbodyheat,increasedinternalpres-
sure,andagitation(Kovecses,1986;La-
koff,1987).Thus,peoplesmetaphorical
mappingofknowledgefromasourcedo-
main(e.g.,heatedfluidinacontainer)onto
atargetdomain(e.g.,theangeremotion)
helpsthemconceptualizeinmoreconcrete
termswhatisunderstoodaboutthetarget
domain.Thismetaphoricalmappingpre-
servesthestructuralcharacteristicsorthe
cognitivetopologyofthesourcedomain
(Lakoff,1990).Forexample,ourunder-
standingthattoomuchheatedfluidcan causeasealedcontainertoexplodeis
mappedontothetargetdomainofanger suchthatweconceptualizetheexplo-
sionofsomeonesangerasbeingper- formedunintentionallyandwithgreatin-
tensity.Variousspecificentailmentsresult
fromthesegeneralmetaphoricalmappings,
entailmentsthatprovidespecificinsight
intothecauses,intentionality,manner,and
consequencesoftheactivitiesdescribedby IDIOMATICMEANING487
stacksblowing,lidsflipping,andceilings beinghit.
Themetaphoricalwaysinwhichwepar-
tiallyconceptualizeexperiences,suchas
anger,providepartofthemotivationfor
whylinguisticexpressionssuchashitthe
ceiling,jlipyourlid,orblowyourstack
makesenseinhavingthefigurativemean-
ingstheydo(Gibbs,1990;Gibbs&Nayak,
1991).Ontheotherhand,literalpara- phrasesof theseidiomssuchas
togetvery angrydonotconveythesameinferences
aboutthecauses,intentionality,andman-
nerinwhichsomeoneexperiencesandex-
presseshisorheranger.Mostliteralpara-
phrasesofidiomsarenotmotivatedbythe
samesetofrichconceptualmetaphorsas
areidioms(Gibbs&OBrien,1990).This
differencebetweenidiomsandtheirliteral
paraphrasesisnotsimplyduetothefact thatidiomsarea
typeoffigurativelanguage andthusmorelikelytobemotivatedby
conceptualmetaphorsthanareliteralex-
pressions.Indeed,manyliteralexpressions
makesensetouspreciselybecausethey tooaremotivatedbyconceptualmeta-
phors.Forexample,theexpressions:He
attackedeveryweakpointinmyargument;
Zdemolishedhisargument;andHiscriti-
cismswererightontargetappeartomost speakersas
beingfairlyliteral.Yeteachex- pressionismotivatedbythesameconcep-
tualmetaphorwherebyargumentsareun-
derstoodintermsofwars.Myclaimthat
literalparaphrasesofidiomsarenotcleady
motivatedbyconceptualmetaphoristhere-
forenotacommentonliterallanguageper
se.Rather,myhypothesisisthatsimplelit-
eralparaphrasesofidioms,suchastoget veryangryortorevealthesecretare
notbythemselvesmotivatedbysinglecon-
ceptualmetaphorsandthereforedonot
possessthekindofcomplexinterpretations as
doidiomaticphrases.Thepresentexper-
imentsaimedtodiscoverwhetherpeople viewedidiomsas
havingdifferentmeanings fromtheirliteralparaphrases.Thesestud-
iesaresignificantnotonlyforpsycholin-
guistictheoriesoffigurativelanguageuse,
butalsobecausetheyprovideadditional
evidenceonthemetaphoricalfoundationof everydaythought. EXPERIMENT1
Thefirststudyattemptedtoprovidedata
showinghowthefigurativemeaningsofid-
iomsaremotivatedbyconceptualmeta- phor.Thepreviousworkonmentalimag-
eryforidiomsindicatedthatpeoplehave
verysimilarintuitionsabouttheactions
thataredescribedbyidiomaticexpressions
(Gibbs&OBrien,1990).Consideranger
idiomssuchasblowyourstack,flipyour
lid,andhittheceiling.Participantsinthe
earlierstudiesstronglyagreedaboutthe
causes,intentionality,andmannerinwhich
stacksareblown,lidsareflipped,andceil-
ingshitwhentheyformmentalimagesfor
theseangeridioms.Thisconsistencyin
peoplesintuitionsabouttheirmentalim-
agesforidiomswasattributedtothecon-
strainingpresenceofspecificconceptual
metaphorsthatmotivatedthefigurative
meaningsoftheseidioms.Fortheangerid-
iomsstudied,theconceptualmetaphorAN- GERISHEATEDFLUIDINACON-
TAINERprovidespartofthelinkbetween
anidiomanditsfigurativemeaningandalso
constrainstheinferencespeoplemake aboutwhattheseidiomsmean.
Thepresentexperimentextendedthese
earlierstudiestoshowthatpeoplesunder-
standingofidiomaticmeaningreflectsthe
particularentailmentsoftheirunderlying
conceptualmetaphors.Eachconceptual
metaphormapsknowledgefromaspecific sourcedomain(e.g.,HEATEDFLUIDIN
ACONTAINER)ontoadissimilartarget domain(e.g.,ANGER).Myhypothesis
wasthatthemetaphoricalmappingsthat
motivateidiomaticmeaningspreservesthe
structuralcharacteristicsofthesourcedo-
main.Forexample,peoplesunderstanding
ofthecauses,intentionality,andmannerof
physicalevents,suchasheatingfluidin
containers(i.e.,sourcedomains),shouldbe 488RAYMONDW.GIBBS,JR.
similartotheirunderstandingsofthe
causes,intentionality,andmannerofthe
angertowhichidiomssuchasblowyour
stack,jlipyourlid,andhittheceilingrefer. ThepurposeofExperiment1
wastosim- plyassesspeoplesunderstandingofthe
causes,intentionality,andmannerofthe
actionsindifferentsourcedomains(e.g.,
heatedfluidinacontainer,thebehaviorof
brittleobjectsincontainers,andsoon).
Theseeventscorrespondedtoparticular
sourcedomainsinvariousconceptualmet- aphors(e.g.,ANGERISHEATEDFLUID
INACONTAINER,IDEASAREPHYS- ICALENTITIESINCONTAINERS,
THEMINDISABRITTLEOBJECT, CONTROLISPOSSESSIONOFSOME
OBJECT)thathavebeenseeninprevious researchas
motivatingthefigurativemean- ingsofidiomssuchas blowyourstack,spill
thebeans,loseyourgrip,andlaydownthe
law(Gibbs&OBrien,1990).Participants
readbriefscenariosdescribingspecific
sourcedomains.Thesescenariosmakeno
referencetoanythingaboutidiomsortothe
targetdomainstowhichidiomsrefer(e.g.,
anger,therevelationofsecrets,goingin-
sane,etc).Afterreadingeachscenario,for
example,aboutfluidinsideasealedcon-
tainer,theparticipantsansweredspecific
questionsregardingthecause,intentional-
ity,andmannerofvariouspossibleevents, suchas
whatmightcausethefluidtoescape fromasealedcontainer.Ifidiomsarepar-
tiallymotivatedbyconceptualmetaphors,
thenpeoplesintuitionsaboutthecausa-
tion,intentionality,andmannerofaction
forthesemetaphorssourcedomains shouldbeverysimilartowhatpeoplegen-
erallyperceiveasbeingthefigurative
meaningsoftheseidioms.Theresultsof
thisstudy,therefore,providethebasisfor
makingspecificpredictionsaboutwhatidi-
omsmean,basedonanindependentassess-
mentofpeoplesintuitionsabouttheindi-
vidualsourcedomainsintheconceptual
metaphorsthatmotivatethefigurative meaningsofidioms. Methods
Subjects.Thirty-eightundergraduatestu-
dentsfromtheUniversityofCalifornia,
SantaCruzservedasparticipantsinthis
study.Theyreceivedcoursecreditfortheir
service.Alltheparticipantswerenative Englishspeakers.
Stimulianddesign.Fourdifferentcon-
ceptualmetaphorsthatmotivatedthefigu-
rativemeaningsofidiomsreferringtoan-
ger,insanity,exertingcontrol,andrevela- tionwereemployedas
theprimarystimuli. Theseconceptualmetaphorswereprevi-
ouslyanalyzedasmotivatingthemeanings
ofdifferentidiomsinGibbsandOBrien (19!30)andincludedANGERISHEATED
FLUIDINACONTAINER(foranger), theMINDISABRITTLEOBJECT(for
insanity),CONTROLISAPOSSESSION (forexertingcontrol),andIDEASARE
ENTITIESINACONTAINER(forreve- lation).
Ashortscenariowaswrittentodepict thebasicelementsineachofthefour
sourcedomains.Forexample,thescenario
forthesourcedomainofheatedfluidina containerstatedImaginethatyouare
lookingatacontainerthatisshapedlikea
cylinder.Thetopofthecontainerissealed.
Thecontaineriscompletelyfilledwith
somesortoffluid.Followingeachsce-
nariowerethreequestionsthatqueriedpar-
ticipantsaboutvariouseventsrelevantto
thesesourcedomains.Onequestionas-
sessedpeoplesintuitionsaboutthecausa-
tionofsomeevent(e.g.,Describesome-
thingthatwouldcausethefluidtocome spontaneouslyoutofthecontainer).A
secondquestionassessedpeoplesintui-
tionsabouttheintentional@ofthatevent (e.g.,
Imaginethatsomethingcausedthe fluidtocomeoutofthecontainer.Doyou
thinkthatthefluidcomesoutonpurposeor
doesthefluidjustsomehowgetoutbyac-
cident?).Afinalquestionassessedpeo-
plesintuitionsaboutthemannerinwhich
theeventisperformed(e.g.,Imagine IDIOMATICMEANING 489 again
thatthefluidcomes outofthesealed container.Doyou thinkthe
fluidcomes out inagentlemannerordoesitexplode out?).Table1
presentsthescenarios and thethreeactionquestionsforeachofthe
foursource domains. Procedure.Eachparticipantwaspre-
sentedwithatestbookletthatcontained the
experimentalinstructionsalong withall the
stimulimaterials.Theparticipantswere
toldthatthepurposeofthestudywasto
examinetheirintuitionsaboutsimpleob- jectsand eventsinthe
realworld.Thepar- ticipants read the firstscenario and then an-
sweredthethreequestionsthatfollowed.
Thiswasthendoneforthesecond,third,
and,finally,fourthsourcedomains.Nei-
thertheexperimentinstructionsortheex- perimentersaid
anythingaboutthisstudy relatingtolinguisticsor,morespecifically,
tothemeanings ofidioms.Theexperiment tookapproximately15
mintocomplete. ResultsandDiscussion The participantsresponses to
each ques- tion wereanalyzed in the followingmanner. First,each
persons response toeach ques- tionwas analyzed forits
generalcharacter- istics. Forexample,whena participantre-
portedthathisorherresponse tothecau- TABLE1 SCENARIOS
ANDACTIONQUESTIONS USEDINEXPERIMENT1 Sourcedomain
Fluidinc1container-Imaginethatyouarelookingata
containerthatisshapedlikeacylinder.Thetopof
thecontainerissealed.Thecontaineriscompletelyfilledwithsomesortof
fluid.
Describesomethingthatwouldmakethefluidexplodeoutofthesealedcontainer.(Causation)
Imaginethatthefluidisheatedtoahightemperatureandthatthefluidcomesoutofthecontainer.Do
youthinkthatthefluidcomesoutonpurpose(thatisintentionally)duetoitsownwillordoesthefluidjust
somehowgetoutunintentionallyorbyaccident?(Intentional&y)
Imagineagainthatonceheatedtoaveryhightemperaturethatthefluidcameoutofthesealedcontainer.
Doyouthinkthatthefluidcomesoutofthesealedcontainerina
gentlemannerordoesitexplodeout? (Manner)
Fragileobjectincontniner-Imaginethatyouhaveacontainerandinsideofitisaverybrittle,fragile
object.
Describehowthefragileobjectsinsidethecontainermightbreakorfallapart.(Causation)
Imaginenowthatsomethinghappenstothecontainerandthatthiscausesthefragileobjectinsideto
break.Doyouthinktheobjectfallsapartintentionallythroughitsownwillordoesitfallapartbyaccident?
(Intentional&y)
Imagineagainthatsomethinghappenstothecontainerandthatthefragileobjectbreaksorfallsapart.Do
youthinktheobjectfallsapartgracefullyandslowlyordoesitfallapartallatonce?(Manner)
Smallobjectsincontainer-Imaginethatyouarelookingatanothercontainer.Thecontaineris
fullof many smallpiecesofsomething.
Describesomethingthatwouldmakethesmallpiecesofmaterialcomeoutofthecontainer.(Causation)
I
fthesmallpiecesdidsomehowgetoutofcontainer,doyouthinkthiswouldhappenonpurposethrough
thewillofthepiecesorwouldthishappenaccidentally?(Intentional&y)
Imagineagainthatsomethinghappenstothecontainerandthatthesmallpiecesofthematerialgetoutof
thecontainer.Dothesepiecesgetoutslowlyordotheysomehowgetoutofthecontainerquitequickly,
perhapsallatonce?(Manner)
Takingcontrolofsomeobject-Considerthesituationwhereyoutakesomeactioninordertotakesome
objectoreventunderyourcontrol.
Whatmakesyouexertcontrolontheobjectorsituation?(Causation)
Istheactionyoudototakesomethingunderyourcontroldoneonpurposethroughyourowndesireor
doesit justoccurbyaccident?(Intentional&y)
Istheactionyoudototakesomethingunder,yourcontroldonegentlyandslowlyorisitperformedwith
someauthorityandforce?(Manner) 490RAYMONDW.GIBBS,JR.
sationquestionforthefluidinacontainer
scenariowasifyouheatthefluidtoa high
temperatureitmightexplodeout,thiswas scoredas
referringtoheatorinternalpres- sureas
themaincauseofthefluidescaping. Twoindependentjudgesexaminedthere-
sponseprotocolsandinitiallyreached96%
agreementastohowtheindividualpartici-
pantsresponsestoquestionsshouldbe
scored.Subsequentdiscussionamongthe
judgesproducedcompleteagreement.In
thesecondstageoftheanalysis,thediffer-
entgeneralcharacteristicsforpeoplesre-
sponsestoeachquestionacrosspartici-
pantsweretallied.Fromthis,thesingle
mostfrequentanswertoeachquestionfor eachsourcedomainwas
determined.Table 2 presentstheproportionoftotalresponses
acrossalltheparticipantsthatconformed
tothemostfrequentlynotedresponsefor eachquestionineachsourcedomain.
TheresponseslistedinTable2 generally
showthatpeoplewerequiteconsistentin
theirintuitionsregardingthecausation,in-
tentionality,andmannerofeventsforthe
fourdifferentsourcedomainsstudied.On
average89%oftheparticipantsresponses
wereinagreementcollapsedacrossthe
threetypesofquestionsandthefoursource
domains.Thisresultreflects,forexample,
thefindingthatpeoplecommonlyagree
thattheprobablecauseoffluidescaping
outofasealedcontainerissomeinternal
pressurecausedbytheincreaseintheheat
ofthefluidinsidethecontainer;thatthis
explosionisunintentionalbecausecontain-
ersandfluidhavenointentionalagency;
andthattheexplosionoccursinaviolent manner.
Ananalysisoftheseintersubjectspro-
portionsindicatednosignificantvariability
acrossthedifferentquestions.Thepartici-
pantswereinhigheragreementintheirre-
sponsestothequestionsforthefragileob- jectsina
containerdomainthantothesmall objectsdomain,z=2.05,p