A Cognitive Linguistic Analysis - Heidelberg University

ImageSchemaVerbsinJapaneseACognitiveLinguisticAnalysis

InauguraldissertationzurErlangungderDoktorwürdederPhilosophischenFakultätder

UniversitätHeidelberg

vorgelegtvonAronWittfeld

Erstgutachterin:Prof.Dr.JuditÁrokayZweitgutachterin:Prof.Dr.TaniguchiKazumi(UniversitätKyôto)

2017

ii

AcknowledgementsThisthesiswasbornoutofthefruitfulcooperationbetweenKyôtoUniversityand

HeidelbergUniversity,fosteredbytheHeKKSaGOnAlliance.Anditwouldnothavebeen

possiblewithoutthefinancialsupportoftheBaden-WürttembergFoundation,which

providedmewiththeopportunitytoconductresearchattheGraduateSchoolofHuman

andEnvironmentalStudiesatKyôtoUniversityin2013.

IwouldliketoexpressmydeepgratitudetoProf.Dr.Taniguchiforherinvaluable

academicadviceaswellasherconstantsupportandencouragement.Thisprojecthas

immenselybenefitedfromthecordialandintellectuallystimulatingatmosphereatthe

Taniguchikenkyûshitsu.Iwouldliketothankallmy“labmates”–HayashiTomoaki,Itô

Kaoru,KanzawaKatsunori,KimotoYukinori,KomatsubaraTetsuta,KônoWataru,

KurodaIppei,SaitôHayato,SugayaTomokatsu,TaguchiShinya,FujitaAyumi,Ogawa

Haruka,SaitôMotoki,SasakiHideaki,andTamaruAyumi.Thankyouforthelively

debates,foryourhelpandinput,andallthefunwehad!

OntheGermansideofthings,mygratitudegoestomyacademicsupervisorProf.Dr.

Árokayforgivingthisprojectachanceandforherunwaveringsupport.Further,Iwould

liketothanktheparticipantsoftheInternationalPhDColloquiumattheInstitutefor

JapaneseStudiesatHeidelbergUniversityfortheirfeedbackonpartsofthisthesis.

Lastbutnotleast,Ithankmyfamilyandfriendsforputtingupwithmethroughallof

this.

iii

Abstract

ThepresentstudyexaminesthesemanticstructureofaspecificclassofJapaneseverbs

within the framework of Cognitive Linguistics. The verbs in question are highly

polysemousandassumed tobe centeredaroundaparticular spatial or force-dynamic

schema – hence the name image schema verbs. Further, they partake in verb-verb

compounding as grammatical “auxiliaries” (V2s) which functionally resemble the

particlesofEnglishandGermanverbparticleconstructions(VPCs).Overthecourseof

fivecasestudiesitisshownthattherespectiveV2sareinherentlymeaningfulandthat

their senses aremotivated by the same image schematic structures thatmotivate the

senses of the simplex. Thus, simplex and V2 are entangled in a complex network of

familyresemblences.Mechanismsofmeaningextensionsuchasmetaphor,metonymy,

andimageschematransformationareexaminedinsomedetailandoftenfromacross-

linguisticpointofview.Rejectingaprincipleddivisionbetweenlexiconandgrammarin

favorof the symbolic continuumhypothesis, argument structurephenomenaare then

reexamined and reframed as issues of cognitive prominence. In the same spirit, the

traditionaldichotomyof“lexical”vs“grammatical”V-Vcompounds,astapleofJapanese

linguistics, is challenged from a usage-based perspective. Based on the results of the

casestudies,thethesiscloseswithabriefcross-culturalinquiryintoembodiedcognition,

showing that directly embodied source domains tend to have similar metaphorical

scopeinJapaneseandGerman.

Keywords: cognitive semantics, image schemas, polysemy, V-V compounds, lexicon-

grammarcontinuum

iv

Glossabbreviations:Theinterlinearglossusedinthisthesislargelyfollowsthe“LeipzigGlossingRules”(https://www.eva.mpg.de/lingua/resources/glossing-rules.php).

• Morphemesareseparatedbyhyphens“-“(e.g.:neko-ga=cat-NOM).• Forpracticalreasons,somecomplexgrammaticalconstructionsaretreatedas

singlemorphemes(e.g.:-teiru=PROG/RES,-teshimau=IRR).• Whenasingleobject-languageelementcorrespondstomorethanonemeta-

languageelement,dots“.”areused(e.g.:desu=COP.POL).• SincepasttenseisindicatedbyPAST,theglossfornon-pastisomitted

(e.g.:tabe-ta=eat-PAST;but:taberu=eat).ABL=ablativeACC=accusativeALL=allativeATT=attributiveCOM=comitativeCON=conjectureCONJ=conjunctionCOND=conditionalCOP=copulaDAT=dativeDES=desiderativeEMPH=emphaticmarkerEVI=evidentialHUM=humbleHON=honorificIMP=imperativeINF=infinitiveINS=instrumentalINT=intentionalIRR=“irreversible”aspect(-teshimau)LK=nominallinkerLOC=locativeM=maleNEG=negativeNMLZ=nominalizer/nominalizationNOM=nominativePAST=pastPL=pluralPOL=politePOT=potentialPROG=progressive/continuousaspectQ=questionparticleQT=quotativeRES=resultativeaspectTE=“conjunctiveform”(-te)TOP=topicVOL=volitional

v

Contents

0. Scope,Aims,andStructureoftheThesis 1 PARTI:THEORETICALFOUNDATIONS 1. TheFramework:CognitiveLinguistics 4 1.1. TheObjectivistTradition 4 1.2. ExperientialRealism 6 2. ImageSchemas 9 2.1. PhilosophicalandLinguisticFoundations:Johnson(1987)andLakoff(1987) 9 2.2. RefiningandDefiningtheNotion:DivergingOpinions 13 3. Trajector/LandmarkOrganization(FigureandGround) 17 3.1. LinguisticStructureasaReflectionofCognitiveProminence 17 3.2. RelevanceforImageSchematicStructure 18 4. Metaphor 20 4.1. Complex,Primary,andImageMetaphor 20 4.2. MetaphorandMeaningExtension 23 5. Metonymy 25 5.1. MetonymicShiftsandEncyclopedicKnowledge 25 5.2. MetonymyandMeaningExtension 26 6. ANoteontheRelationbetweenMetaphorandMetonymy 28 7. PolysemyandLexicalNetworks 29 7.1. TheCaseforPolysemy 29 7.2. MakingSenseofSenses–SomeProposals 31 7.3. PolysemyasaFuzzyNotion–TheLangacker/TuggyModel 32 7.4. ImplicationsforthePresentStudy 35 8. SpatialExpressionsinCognitiveLinguistics:AnOverviewoftheLiterature 36 PARTII:CASESTUDIES 9. KAKARUandtheCONTACTSchema 39 9.1. TheSensesofKAKARU 40 9.1.1. Sense(Ia):PhysicalSupport 40 9.1.2. Sense(Ib):ImaginedSupport 40 9.1.3. Sense(Ic):PhysicalForce 41 9.1.4. Sense(Id):PsychologicalBurden 41 9.1.5. Sense(Ie):PreconditionforSuccess 42 9.1.6. Sense(If):OntologicalDependence 43 9.1.7. Sense(II):ElicitedEffect 43 9.1.8. Sense(III):Covering 46 9.1.9. Sense(IVa):PhysicalRestraint 48 9.1.10. Sense(IVb):AbstractRestraint 49 9.1.11. Sense(Va):ExternalControl 50 9.1.12. Sense(Vb):AgentiveControl 50 9.1.13. Sense(VIa):PhysicalArrival 51 9.1.14. Sense(VIb):Transmission 52 9.1.15. Sense(VIc):TemporalArrival 52 9.1.16. Sense(VII):ResourceRequirement 52

vi

9.1.17. Sense(VIIIa):PhysicalLink 53 9.1.18. Sense(VIIIb):RelevanceLink 55 9.1.19. Summary 55 9.2. TheSensesofV-KAKARU 56 9.2.1. TheGermanPrepositionanandtheCONTACTSchema 57 9.2.2. CONTACTandDirectedness:TheDOTOWARDSSenseofan-VandV-

KAKARU58

9.2.3. CONTACTandInchoativity:TheSTARTSenseofan-VandV-KAKARU 67 10. DERUandtheEXITSchema 75 10.1. TheSensesofDERU 75 10.1.1. Sense(I):SpatialExit 75 10.1.2. Sense(II):Activity 79 10.1.3. Sense(III):Incubation 80 10.1.4. Sense(IV):Transfer 81 10.1.5. Sense(V):Access 82 10.1.6. Sense(VI):Excess 84 10.1.7. RelationsBetweenSensesandCategorialFringeCases 85 10.1.8. OntheChoiceofkaravswo 87 10.2. TheSensesofV-DERU 92 10.2.1. SpatialV-DERU 92 10.2.2. Activity 92 10.2.3. Incubation 93 10.2.4. Transfer 93 10.2.5. Access 93 10.2.6. InchoativeV-dasu 94 11. KiruandtheSPLITSchema 98 11.1. TheSensesofkiru 98 11.1.1. Sense(Ia):PhysicalDiscontinuity-LMisaSolidExtentofMatter 98 11.1.2. Sense(Ib):UnintentionalSelf-injury-NoSegmentation 98 11.1.3. Sense(Ic):Opening-LMisaCONTAINER 99 11.1.4. Sense(Id):TraversalofNon-solid,UnboundedLM 100 11.1.5. Sense(Ie):Disconnection-LMisanAssemblyofFunctionalParts 101 11.1.6. Sense(If):Disconnection-LMisanAbstractRelation 101 11.1.7. Sense(Ig):TemporalDiscontinuity-LMisanActivity 102 11.1.8. Sense(Ih):Reduction-LMisanAbstractScalarExtent 103 11.1.9. Sense(II):FocusonObsoletePortionofLM 104 11.1.10. Sense(III):FocusonPointofSegmentation 105 11.1.11. Sense(IV):FocusonManner 106 11.2. TheSensesofV-kiru 106 11.2.1. PreviousSuggestions 106 11.2.2. Discussion 107 11.2.2.1. SomeRemarksonLimitvsAccomplishment 109 11.2.2.2. ‘Odd’Cases 111 11.2.3. RevisitingV-kiru:ACategorizationBasedonSchematicTopology 112 11.2.3.1. Sense(I):TheV1ProfilesaWayofPhysicalSegmentation 112 11.2.3.2. Sense(IIa):TheV1ProfilesanInherentlyGoal-orientedProcess

(“Limit”Sense)112

11.2.3.3. Sense(IIb):TheV1ProfilesaNon-goal-orientedProcess;TheLMoftheV1FunctionsasaTelicModifier(“Accomplishment”Sense)

113

11.2.3.4. OtherSenses 114 11.3. RelatedConstructions 116 11.3.1. V-kiri/V-takiri(da) 116 11.3.2. NumeralClassifier+kiri 117 11.3.3. kiri-ganai 117

vii

12. AGARUandtheUPSchema 119 12.1. TheSensesofAGARU 119 12.1.1. Sense(Ia):SpatialAscensionofaZero-dimensionalTR 119 12.1.2. Sense(Ib):AbstractAscensionofZero-dimensionalTR(SocialAscension) 120 12.1.3. Sense(IIa):SpatialExtensionalongtheVerticalAxis 120 12.1.4. Sense(IIb):AbstractExtensionalongtheVerticalAxis 121 12.1.5. Sense(III):SubtractiveCompletion 122 12.1.6. Sense(IV):Access 124 12.2. TheSensesofV-AGARU 126 12.2.1. SpatialAscensionofaZero-dimensionalTR 126 12.2.2. AbstractAscensionofaZero-dimensionalTR(SocialAscension) 128 12.2.3. GOAL-orientedSpatialMovement(BleachedVerticality) 128 12.2.4. GOAL-orientedNon-spatialMovement(BleachedVerticality) 130 12.2.5. SpatialExtensionalongtheVerticalAxis 131 12.2.6. AbstractExtensionalongtheVerticalAxis 132 12.2.7. MultidimensionalSpatialExtension 132 12.2.8. VerticalEncroachment 133 12.2.9. Completion1:SubtractiveCompletion 133 12.2.10. Completion2:AchievementofSufficientState 135 12.2.11. SomeNotesonReflexiveTRs 136 13. TÔRUandthePATHTRAVERSALSchema 139 13.1. TheSensesofTÔRU 139 13.1.1. Sense(Ia):LMisaVolumeinPhysicalSpace 139 13.1.2. Sense(Ib):LMisa“Floor”SurfaceinPhysicalSpace 141 13.1.3. Sense(Ic):LMisa“Wall”SurfaceinPhysicalSpace 142 13.1.4. Sense(Id)LMisaMassofUnspecifiedDimensionalityinPhysicalSpace 144 13.1.5. Sense(II):LMisaTemporalExpanse 144 13.1.6. Sense(III)LMisaNon-spatial,Non-temporalExpanse 145 13.1.7. Sense(IV):LMisanInstrument(wotôshite) 146 13.2. TheSensesofV-TÔRU 149 13.2.1. LMisanX-dimensionalExpanseinPhysicalSpace 149 13.2.2. LMisaTemporalExpanse(ExtendedProcess) 150 13.3. RelatedConstructions:N-notôri(N-dôri)/V-tôri 152 CasesStudies:Summary 153 PARTIII:BEYONDTHENETWORK 14. CompositionalDisparity 155 14.1. ButWhataboutSyntax? 155 14.2. V-VCompoundsinTeramura(1969),Nagashima(1976),andYamamoto(1984) 155 14.3. CompositionalDisparityasanUmbrellaTerm 158 14.4. SalienceandAbstractEntities:SomeCompoundswithGrammaticalV2s 159 14.5. GrammaticalV2sand“FakeTransitivity” 164 14.6. OtherSourcesofCompositionalDisparity 165 14.6.1. “Subordination”,ConceptualAutonomy,andDiscourseContext 165 14.6.2. ActiveZones 170 14.6.3. TowardsanActiveZoneAnalysisofSomeGrammaticalV2s 171 14.6.4. PartialMetaphoricalMappings 174 14.7. LexiconvsSyntax?TowardsaUnified,Schema-basedAccount 175 15. EmbodimentandtheScopeofMetaphorinGermanandJapanese 181 15.1. (I)TheWeightScale:HeavyandLight 181 15.1.1. (Ia)EFFORTFULACTIVITYISHANDLINGHEAVYOBJECTS 181 15.1.2. (Ib)ABSTRACTBURDENSAREPHYSICALWEIGHTS 182 15.1.3. (Ic)INTENSITYISWEIGHT 184 15.1.4. (Id)IMPORTANCEISWEIGHT 185 15.1.5. (Ie)RESPECT/DIGNITYISWEIGHT 186

viii

15.2. (II)EdgeProperties:SharpandDull 187 15.2.1. (IIa)SynaestheticMappings 187 15.2.2. (IIb)UNPLEASANTINTENSITYISSHARPNESS 188 15.2.3. (IIc)PRECISIONISSHARPNESS 189 15.2.4. (IId)INTELLIGENCEISSHARPNESS 189 15.2.5. Excursion:OverlapofMetaphorsinaSingleExpression 190 15.3. (III)SurfaceProperties:SmoothandRough 191 15.3.1. (IIIa)ABSTRACTREFINEMENTISSMOOTH,LACKOFABSTRACT

REFINEMENTISROUGH191

15.3.2. (IIIb)GOODDEVELOPMENTISSMOOTH,BADDEVELOPMENTISROUGH 192 15.3.3. (IIIc)DEGREEOFDETAILISGRANULARITYOFSURFACESTRUCTURE 194 15.4. ObservationsandExplanations 195 ConcludingRemarksandProspects 197 References 199

0.Scope,Aims,andStructureoftheThesis

Themain objective of this study is to give an account of the semantic structure of a

specific class of Japanese verbs, i.e. image schema verbs, in accordance with general

cognitiveprinciples.

Theverbsunderconsiderationherefulfilltwocriteria.

(i) Theyarehighlypolysemouswithabasicspatialorforce-dynamicmeaning.

(ii) Theypartakeinverb-verbcompoundingasgrammatical“auxiliaries”.

Toillustrate,considerthefollowingexamples:

(1) Kabe-ni e-ga kakat-teiru.

Wall-DAT picture-NOM KAKARU-RES ‘Apictureishangingonthewall.’

(2) Inochi-wo kake-ta gyanburu

Life-ACC KAKERU-PAST gamble ‘Agamblewithone’slifeatstake’

(3) Tarô-ga teki-no keiryaku-ni kakat-ta

Tarô-NOM enemy-LK scheme-DAT KAKARU-PAST

‘Tarôfellvictimtotheenemy’sscheme.’

(4) Kuruma-wo kauØ-ni-wa okane-ga kakaru.

Car-ACC buyNMLZ-DAT-TOP money-NOM KAKARU

‘Oneneedsmoneytobuyacar.’

(5) Hanako-ga Tarô-ni warai-kake-ta.

Hanako-NOM Tarô-DAT smile-KAKERU-PAST ‘HanakosmiledatTarô.’

(6) Jirô-ga hon-wo yomi-kake-ta tokoro-e denwa-ga nat-ta.

Tarô-NOM book-ACC read-KAKERU-PAST moment-ALL phonecall-NOM ring-PAST

‘AsTarôbegantoreadthebook,thephonerang.’

(7) Tarô-ga jiko-ni at-te, shini-kake-tei-ta.

Tarô-NOM accident-DAT meet-TE die-KAKERU-RES-PAST

‘Tarôgotintoanaccidentandwasonthevergeofdying.’

Myworkinghypothesis,inthisexemplarycase,isthatthevarioussensesofkakaru(and

its transitive variant kakeru) are structured around the abstract schema CONTACT.

Severalotherimageschemaverbsaregivenbelow:

verb centralschema

iru/ireru(enter/putsth.in) CONTAINMENT:ENTRY

deru/dasu(moveout/putsth.out) CONTAINMENT:EXIT

agaru/ageru(rise/raise) VERTICALITY:UP

tsukiru/tsukusu(runout/useup) DEPLETION

kiru(cut) SPLIT

1

tôru/tôsu(gothrough/letpass) PATHTRAVERSAL

au(meetwith/match) MATCHING

etc.etc.

Aspertheabovecriteria,Idonottaketheclassofimageschemaverbstohaveclearly

demarcatedboundaries.Sincethenotionofimageschemaasdiscussedintheliterature

is itself somewhat fuzzy, it follows that imageschemaverbs, too,arebest treatedasa

prototype-centered,opencategory(seechapter2).

Whyaretheseverbsofanyinterestatall?Thereareseveralreasons.First,thereisa

vastbodyofresearchonthetopicofverb-verbcompounds(fukugôdôshi).Andwhilea

huge part of it is dedicated to the analysis of verbs like the above, their existence as

simplexverbsrarely receivesmore thanapassingglance. Inotherwords, fewstudies

draw a connection between themeaning of the simplex verb and themeaning of the

corresponding auxiliary (henceforth called V2). The neglection of the simplex in the

literatureeithervaguelypresupposesitssemanticsignificanceortacitlydeniesit.Inthis

context,thepresentstudyseekstofillatheoreticalvoid:Thesemanticsofthesimplex

andthesemanticsoftheV2aretreatedastwosidesofthesamecoin.Throughaseries

ofcasestudiesIhopetoshowthattheV2isinherentlymeaningfulandthatitssenses,

togetherwiththesensesofthesimplex,formacomplexnetworkoffamilyresemblances.

Secondly, image schema verbs serve to illustrate the inextricable relation between

lexicon and syntax. The present study assumes that both are poles on a continuum

rather than discrete components and aims to show how lexical semantics, in tandem

with salience, directly affects syntactic phenomena such as argument selection in the

caseofverbalcompounds.

Finally, image schema verbs in many ways resemble the particles of verb particle

constructions(VPCs)inotherlanguages.ThroughoutthecasestudiesofthisthesisIwill

discuss Japanese image schema verbs in contrast and comparison with VPCs from

GermanandEnglishsuchasan-VorVup.Therefore,thestudyassumesacross-linguistic

perspectiveandwillhopefullybeanassettofutureinvestigationsintoVPCsandsimilar

constructionswithintheCognitiveLinguisticsframework.

The thesis consists of three parts. Part 1 lays out the theoretical foundations by

introducingtheframeworkofCognitiveLinguisticsanditsmajorguidingassumptions.

Basicconceptsthatareparticularlyrelevanttothepresentpurposearesingledoutand

discussed insomedetail.Thestudy is thensituated in thecontextofpastandpresent

researchonspatialexpressions.

2

Part2isconcernedwiththeanalysisofsemanticstructure.InaseriesofcasestudiesI

investigate five image schema verbs: kakaru, deru, kiru, agaru, and tôru. These verbs

werechosenbecauseoftheirhighprototypicality, i.e.theyarebothhighlypolysemous

andprominentasgrammaticalV2s.Thefollowingquestionspresentthemselves:What

isthesemanticstructureofthesimplex?Canthemeaningofthesimplexaccountforthe

meaning of the V2? What mechanisms of meaning extension are involved? Can we

maintain the hypothesis that each of these verbs’ semantics is centered around a

particularimageschema?

Based on this analysis, part 3 discusses further theoretical issues. The chapter on

“compositionaldisparity”isconcernedwithtwomainquestions:Givenanon-algebraic

approach to grammar, how can one account for the compositional properties of

Japanese verb-verb compounds in the framework of Cognitive Linguistics?Moreover,

howdoweappraochtheissueof“lexicalvssyntacticcompounds”fromanon-modular

point of view? The second chapter of part 3 shifts the focus away from questions of

compositionalityandargumentstructuretowardsanimportanttopicattheperipheryof

ourmainenterprise:Theuniversalcharacterofembodiedexperience. Inasmall-scale

comparative study of German and Japanese I explore the degree of variance in

metaphoricalscope,startingoutfromacommonsetofembodiedsourcedomains.

3

PARTI:THEORETICALFOUNDATIONS

1.TheFramework:CognitiveLinguistics

Beforeweturnourattentiontothecasestudies,itisessentialtoclarifywhatconception

oflanguageandlinguisticstheyarebasedonandwhythatconceptionispreferabletoits

alternatives.CognitiveLinguisticsisnotacompacttheoryoflanguagebutratheraloose

framework, a relatively new paradigm of linguistic inquiry, consisting of diverse

theoreticalapproachessharingacommonperspective.InthefollowingIwillsketchout

whatIconsidertobethephilosophicalfoundationofthecognitivelinguisticenterprise–

the position known as experiential realism. First, however, we must consider the

objectivisttraditioninoppostiontowhichitemerged.

1.1.TheObjectivistTradition

ManynowclassicworksinthefieldofCognitiveLinguisticsincludeatleastonepassage

or chapter akin to a “manifesto” inwhich the respective authors distance themselves

fromatraditioninthephilosophyofmindandlanguageoftendescribedasobjectivism

(e.g.LakoffandJohnson2003;Lakoff1990b;Johnson1990;Langacker1990;Sweetser

1991). The termobjectivismwas coined by Lakoff and Johnson in their seminal 1980

bookMetaphorsWeLiveByandisprobablymostconciselyexplainedinLakoff(1990b).

Ashepointsout,objectivism isnota theoryofmindor languagebutratherasetofa

prioriassumptionsdeeplyentrenchedinthehistoryofWesternphilosophy–sodeeply

thatmanyof themdateback to antiquity andhavebeen taken for granted ever since

(Lakoff 1990b: xii). I believe that objectivist linguistics is best broken down into two

maintenetsfromwhichitsotherassumptionsthenfollow(basedonLakoff1990b):

The correspondence model of meaning: Linguistic expressions have meaning only in

virtueof theirdirect correspondence to the things, relations, and statesof affairs that

makeupobjectivelygivenreality.Thatis,meaningisarelationbetweenwordsandthe

world without any kind of human mediation. Lexemes correspond to pre-existing

categoriesand theirmeaning isadequately representedby featurebundles. Sentences

correspondtoobjectivestate-of-affairsandtheirmeaningisadequatelygiveninterms

oftruthconditions.

4

The computational model of the mind: The human mind essentially functions like a

machine operating on a set of algebraic rules. Linguistic activity is the application of

combinatorialrulestoalistoflexemesinordertoassemblewell-formedsentences.

Thesetwotenetshaveseveralimportantimplications:

Meaning is disembodied and culture-independent: If meaning is a relation between

wordsandobjectivereality,itfollowsthatthehumanperspectivehasnoparttoplayin

categorystructure.Thebiologicalnichewehavecometooccupy,thephysiologyofour

bodies, the socio-cultural contextwe live in,while obviously shaping our view of the

world,hasnobearingonsemantics.

Meaningisallabouttruthandreference:Categorystructureisneatinthesensethatit

essentiallyboilsdowntoa“checklist”ofnecessaryandsufficientfeatureswhichhelpus

identify categorymembers. In otherwords, themeaning of a lexeme is a guide to its

reference.Forexample,abachelorisanunmarriedadultman.Everythingelseweknow

aboutorassociatewithbachelors isnotpartof“semanticsproper”.Sentencesexpress

propositionswhichareeithertrueorfalse.Tounderstandthemeaningofasentenceis

tobeabletogiveitstruthconditions.Simplyput:ThesentenceSnowiswhiteistrueif

and only if snow is white. Again, all non-truth conditional aspects of a sentence (e.g.

speech act meaning, grammatical voice, politeness, etc.) are not part of semantics.

Consequently,thereisasharpdividebetweenthemeaningofanexpression(semantics)

anditsuse(pragmatics).

Language is autonomous and compartmentalized: According to the computational

model of the mind, language is an autonomous faculty, i.e. our linguistic ability is

independentfromtherestofcognition(e.g.attention,figure-groundorganization,etc.).

Within the language faculty semantics, syntax, and phonology exist as distinct

components, eachwith their own set of rules and constraints. These components are

complementedby the lexicon,a listof lexicalentriesuponwhich theyoperate. In this

model “pragmatics” is an umbrella term for everything concerning the actual use of

language(asopposedtotheinternalworkingsofthelanguagefaculty).1

1Itis,ofcourse,impossiblewithintheconfinesofthisthesistosurveyovertwothousandyearsofWesternphilosophyandgiveahalf-waysatisfiyinghistoricalportrayalofobjectivism.Nonethelesssome

cornerstonesshouldbementioned.Theideathatanobjectiveeternalrealitydivorcedfromhuman

experienceisaccessiblebydisembodiedthoughtcanbetracedbacktoPlato’stheoryofforms.The

5

1.2.ExperientialRealism

Given the above positions,we can see how the objectivist paradigmmarginalizes the

roleof theconceptualizer.Meaningmirrors thestructureofanobjectiverealityand is

grasped by logical thought. Thus, the conceptualizer has no active role in shaping

semantic structure. This is problematic because, if it were true, many linguistic

phenomenawouldbecomeeitherinexplicableorirrelevant.Forillustration,considera

passagefromSweetser(1991).Whileadmittingthatquestionsofconceptualizationmay

beoflittlerelevanceindeterminingthetruthvalueofsentencessuchasSnowiswhite,

shegoesontomakethefollowingpoint:

Butsupposethat,insteadofwhite,ItakeLatincandidusasmysampleword.Candidusmeant,amongotherthings,“white”and“bright”;butitalsomeant“open,honest”–asinitsEnglishdescendent,candid.Butitseemsunlikelythatthereisanyobjectivecorrelationintherealworldbetweenwhitethingsandhonest

things,oranylargerobjectivelychosencategorywhichincludesjusttheseandnoothers.The“realworld”,

ifwemeanonewhich isoutsideofhumancognitiveorganization, isnotsoconstructedas togroup the

white with the honest. Rather, it is our cognitive structuring of the world which can create such an

identification.Andiflanguageusesawordofourcognitivecategory,thenlanguagecannotbedescribedin

termsofWord andWorld: unless, byWorld,wemeanour experiential picture of theworld. (Sweetser

1991:4-5)

Whatwecantakefromthispassageisthatcategoriesinnaturallanguageusuallyexhibit

polysemy, i.e. one lexical item often has several related senses (as illustrated by

candidus).Givinganaccountofmeaningextension inordertoexplainphenomena like

polysemyanddiachronicchange isobviouslypartof linguistics.However,asSweetser

points out above, such an account can only be given in terms of “human cognitive

organization”–whichhasnobearingonsemanticstructureaccordingtoobjectivism.

“dictionarymodel”(orchecklistmodel)oflexicalsemantics(e.g.Katz1972)originatedwithArtistotle’s

decompositionalaccountofcategoriesintermsofnecessaryandsufficientconditions.Mind-bodydualism

(i.e.therigidseparationofimmatierialmindandmaterialbody)isacentralthemeinDescartes’

epistemologyandphilosophyofmind.

However,itisnotuntiltheemergenceoftheFregeanphilosophyoflanguageattheendofthe19th

centurythatwecanspeakofanobjectivistlinguistics.Themajortenetsofobjectivistsemanticswere

formulatedinFrege’s(1892)seminalpaperÜberSinnundBedeutung(OnSenseandReference),inwhichhelaysthegroundworkforatruth-conditionaltheoryofmeaning.Wittgenstein’spicturetheoryof

meaning(i.e.theviewthatsentencemeaningmirrorsstates-of-affairs)andhisfamousclaimthat“[t]o

understandapropositionmeanstoknowwhatisthecaseifitistrue”(Wittgenstein1922:4.024)are

hardlyconceivablewithoutFregeanphilosophy.Allsubsequentformalapproachestosemanticssuchas

Tarsky([1944]2004),Montague(1973),andDavidson(1967)arebuiltuponWittgenstein’sidentification

ofsentencemeaningwithtruth-conditions.Unsurprisingly,approachestolanguagebasedonlogicand

algebrafavoracomputationalmodelofthemindsuchastheoneassumedbythevariousiterationsof

ChomskyianGenerativeGrammar(e.g.Chomsky1957,1965).

ForamoredetaileddepictionofseveralobjectivistpositionsseeJohnson(1990).ParttwoofLakoff

(1990b)launchesacomprehensiveattackonthephilosophicalfoundationsofobjectivism.

6

AnotherexampleisthefollowingsentencepairfromLangacker(1990:13):

(1a) BillysentawalrustoJoyce.

(1b) BillysentJoyceawalrus.

According to Langacker, these sentences offer an alternate construal of the same

conceptualcontent.Inshort,to in(1a)emphasisesthepathtakenbythewalrus,while

thedirectjuxtapositionofJoyceandawalrusin(1b)emphasisesthepossessiverelation

(Langacker 1990: 13-14). In other words, by choosing either (1a) or (1b) the

conceptualizerdecidesonaspecificwayofpackagingandpresentingthesamecontent.

Each version conveys a different manner of experiencing the world. Importantly,

construalreflectsgeneralprinciplesofhumancognition: inthiscase,differentwaysof

distributing attention across a given scene (see e.g. Talmy 2003a: 76ff.). From an

objectivistperspectivenoneofthismatters.Sincebothversionsaretruth-conditionally

equivalent (they are true in the same set of possible worlds), the preposition to is

consideredtobesemanticallyvacuous.

Finally,considerthestory-tellingfunctionoflanguage.InthefollowingpassageLakoff

andJohnson(2003)illustratethecreationofacoherentnarrative:

[...]facedwiththeenergycrisis,PresidentCarterdeclared“themoralequivalentofwar.”TheWAR

metaphorgeneratedanetworkofentailments.Therewasan“enemy”,a“threattonationalsecurity”,

whichrequired“settingtargets”,“reorganizingpriorities”,“establishinganewchainofcommand”,

“plottingnewstrategy”,“gatheringintelligence”,“marshalingforces”“imposingsanctions”,“callingfor

sacrifices”,andonandon.Themetaphorwasnotmerelyawayofviewingreality;itconstitutedalicense

forpolicychangeandpoliticalandeconomicaction.(LakoffandJohnson2003:156)

Ihavecitedthispassage,becauseitunderscoreshowhumanunderstandingworks.We

understandCarter’snarrative(regardlessofwhetherweacceptorrejectit),becausewe

understand the metaphorical correspondences it is built on. Yet, from an objectivst

perspective these correspondences cannot be part of a theory of meaning which

assumes a principled distinction between literal and figurative speech, as well as

betweensemanticsandpragmatics.Since“semanticsproper”isonlyconcernedwithso-

called literalmeaningandtruthconditions, themeaningofCarter’sviewamountstoa

set of false and/or nonsensical propositions instead of a coherent whole. An

experientialist account, on the other hand, recognizes communication as the primary

functionoflanguage.Themetaphoricalcorrespondences,ultimatelygroundedinhuman

experience,areapreconditionforunderstandingCarter’snarrative.Andallphenomena

7

involved in understanding linguistic communication should be part of a theory of

semantics.2

Muchmorecouldbesaidaboutthedifferencesbetweenanobjectivistandacognitive

linguisticapproachtolanguage,butIthinktheaboveexamplescapturethespiritofthe

experientialist enterprise quite well. In summary, then, experiential realism is the

positionthatthoughtandmeaningarisefromembodiedexperienceandareimaginative

in nature. Reality is not objectively given but only accessible via our species-specific

sense-perceptual capabilities which in tandem with general cognitive principles give

rise to imagistic structures (e.g. psychological gestalts) that go far beyond thekindof

propositionalentitiespositedbyformalsemanticists(Lakoff1990b:xv).Foratheoryof

languagethispositionhastwomajorimplications:

Cognitive holism: As the above examples from Sweetser, Langacker, and Lakoff and

Johnson illustrate, linguistic phenomena reflect general principles of human cognition

(in the above cases: categorisation, attention, and conceptualmetaphor, respectively).

“Eveniftheblueprintsforlanguagearewiredgeneticallyintothehumanorganism,their

elaboration into a fully specified linguistic system during acquisition, and their

implementationineverydaylanguageuse,areclearlydependentonexperientialfactors

and inextricably bound up with psychological phenomena that are not specifically

linguisticincharacter”(Langacker1987:13,emphasismine).Thus,CognitiveLinguistics

canbedescribedasembracingaholisticviewoftherelationshipbetweenlanguageand

cognition.

Centralityofmeaning:Sincemuchofwhatisknownaboutthemindfromthecongnitive

sciencesisincompatiblewithamodular,algebraicviewoflanguage(Lakoff1990a:42)

cognitive linguists havebeen free to discard the central commitments of such a view.

First and foremost, rejecting the paradigm of empty symbol manipulation – a

consequenceof thestrictseparationofsyntaxandsemantics–hasreopened thedoor

for ameaning-based linguistics. In otherwords, there is noapriori need for positing

semanticallyvacuousstructures.Noristhereanyconvergingevidencesuggestingtheir

2SeealsoLangacker(1990:2):“Meaningisequatedwithconceptualization.Linguisticsemanticsmustthereforeattemptthestructuralanalysisandexplicitdescriptionofabstractentitieslikethoughtsand

concepts.Thetermconceptualizationisinterpretedquitebroadly:itencompassesnovelconceptionsas

wellasfixedconcepts;sensory,kinesthetic,andemotiveexperience;recognitionoftheimmediatecontext

(social,physical,andlinguistic);andsoon.”

8

existence. Therefore, Cognitive Linguistics assumes that “[l]exicon, morphology, and

syntax form a continuum of symbolic units, divided only arbitrarily into separate

components;itisultimatelyaspointlesstoanalyzegrammaticalunitswithoutreference

totheirsemanticvalueastowriteadictionarywhichomitsthemeaningsof its lexical

items”(Langacker1990:29).

InthefollowingsectionsIwillintroduceseveralkeyconceptsthatarecentraltomy

argumentationinthesubsequentchapters.

2.ImageSchemas

2.1.PhilosophicalandLinguisticFoundations:Johnson(1987)andLakoff(1987)

Image schemas are non-propositional imagistic patterns which arise from bodily

experience and structuremuchof our conceptualization and reasoning.As such, their

importancefortheprincipleofembodiedcongnitioncanhardlybeoverstated.Theterm

image schema was introduced by Mark Johnson ([1987] 1990) and George Lakoff

([1987] 1990b) in two separate book-length studies and from slightly different

perspectives.WhileJohnsonstressesthephilosphicalunderpinningsofimageschemas,

Lakoffpresents linguistic evidence in the formof a case study.Letusbriefly consider

bothperspectivesinturn.

Johnson’s overarching goal is to present an alternative to the Cartesianmind-body

dualism that – tacitly or explicitly – has dominated a significant part of Western

philosophyandcontinuestobeinfluentialtothepresentday.Allphilosophicaltheories

whichareCartesianinspiritfaceacommonchallenge:Theymustsomehowexplainhow

themindisconnectedtothebodyandhowthetwocouldpossiblyworktogether.Asa

prominentattempttoanswerthisquestion,JohnsonpointstoKant’stheoryofmindin

theCritiqueofPureReason.Inshort,Kantassumesadivisionoflaborbetweensensory-

motorperception(sensibility)anddisembodiedconceptformation(understanding):

[...]thereisanoverlyrigiddichotomybetweentheconceptualandthebodily.Conceptsareproductsofourunderstanding,whichisformal,spontaneous,andrule-governed;sensationsarebodily,giventhroughour

sensibility, which is material, passive, and lacking in any active principle of combination or synthesis.

(Johnson1990:xxviii)

9

In order to bridge the gap between sensibility and understanding, Kant postulates an

intermediary faculty, imagination, tasked with unifiying the slew of perceptual

impressionsintoacoherentwhole:

Forexample,inmyperceptionofadog,Kantthoughtthatimaginationorderedvarioussenseimpressions

(e.g., thefeeloffur, fourlegs,atrunk, longteeth,etc.) intoasingleperceptualexperience(e.g.,aunified

imageof a furry creature), such that I can then recognize it (conceptualize it) as adog. (Johnson1990:

xxviii)

Johnson goes on to argue that Kant’s account of imagination is ultimately self-

contradictory. Imagination seems to belong in equal parts to the realms of bodily

experienceandconceptual thought–despite the fact thatKant’sphilosophical system

precludesthisverypossibility.“Somehowimaginationissupposedtohaveafootinboth

worlds(inthe‘formal’andthe‘material’),andyetitisnotclearhowitcanhavethisdual

nature”(1990:166).Johnson’ssolutiontothisproblemistodoawaywiththeCartesian

legacywhichgaverisetoitinthefirstplace:“Oncewenolongerdemandadisembodied

(or non-physical) rationality, then there is no particular reason to exclude embodied

imagination from the bounds of reason” (1990: 168). It is for this embodied view of

reason that the notion of image schema is integral, since image schemas are both

groundedinbodilyexperienceandemployedinabstractthought:

TheviewIamproposing is this: inorder forus tohavemeaningful,connectedexperiences thatwecan

comprehend and reason about, there must be pattern and order to our actions, perceptions, and

conceptions. A schema is a recurrent pattern, shape, and regularity in, or of, these ongoing orderingactivities. These patterns emerge as meaningful structures for us chiefly at the level of our bodilymovementsthroughspace,ourmanipulationofobjects,andourperceptualinteractions.(Johnson1990:

29)

JohnsonillustratesthisbypointingoutthepervasivenessoftheINandOUTschemas:

Considerjustasmallfractionoftheorientationalfeatsyouperformconstantlyandunconsciouslyinyour

dailyactivities.Consider,forexample,onlyafewofthemanyin-outorientationsthatmightoccurinthefirstfewminutesofanordinaryday.Youwakeoutofadeepsleepandpeeroutfrombeneaththecoversintoyourroom.Yougraduallyemergeoutofyourstupor,pullyourselfout fromunderthecovers,climbintoyourrobe,stretchoutyourlimbs,andwalkinadazeoutofthebedroomandintothebathroom.Youlookinthemirrorandseeyourfacestaringoutatyou.Youreachintothemedicinecabinet,takeoutthetoothpaste,squeezeoutsometoothpaste,putthetoothbrushintoyourmouth,brushyourteethinahurry,andrinseoutyourmouth.Atbreakfastyouperformahostoffurtherin-outmoves–pouringoutthecoffee,settingoutthedishes,puttingthetoastinthetoaster,spreadingoutthejamonthetoast,andonandon.Once you aremore awake youmight even get lost in the newspaper,might enter into a conversation,whichleadstoyourspeakingoutonsometopic.(Johnson1990:30-31)

The key point is that, from early infancy and on a daily basis, we are subjected to

myriads of sensory-motor experiences involving CONTAINERS (e.g. grasping objects,

10

eating, being located in various bounded spaces) which eventually lead to the

emergenceofapreconceptualdynamicpatternentrenchedat thenon-conscious level.

Thispattern,theimageschemaCONTAINER,willinturngivemeaningtoallourfuture

encounters with containers. Crucially, as we shall see below, image schemas are a

preconditionforabstractthought,e.g.whenabstractstates(walkinadaze,enterintoa

conversation)aremadesenseofintermsofphysicallocations.

So far,wehavesketchedout Johnson’sphilosophicalmotivations forpositing image

schemasandgivenapreliminarycharacterizationofthenotion.Togetabetterideaof

what image schemas are and how they structure our thought, a slight change of

perspectivemightbehelpful.IthereforesuggestthatwetakealookatGeorgeLakoff’s

workoncategorystructureinordertoobserveimageschemas“inaction”.

BuildingmainlyonpreviousworkbyBrugman(1981),Lakoff(1990b)askshowthe

varioussensesofEnglishoverarerelatedtooneanother.Oneof themostremarkable

discoveriesof the study iswhathecalls transformationallinks. Consider the following

sentences(adaptedfromLakoff1990b):

(2a) SamwalkedovertheHill. (pathfocus)

(2b) SamlivesovertheHill. (end-pointfocus)

(3a) Theguardswerepostedalloverthehill. (multiplex)

(3b) Theboardisoverthehole. (mass)

(4a) Thebirdflewovertheyard. (moving0Dentity)

(4b) Thepowerlinestretchesovertheyard. (1Dstaticentity)

In (2a)wementally trace the trajectoryofamovingentity (Sam). Ifwe then focuson

thatentity’srestinglocationweendupwith(2b).Thehillin(3a)iscoveredbymultiple

individualentities(amultiplexity).However,ifwementally“zoomout”onthescene(or

squintoureyes)theguardswillappearasanundifferentiatedwhole(amass)similarto

theboardin(3b).In(4a)wetracethebird’spathaswewouldtraceamovingdot.Ifwe

thenmentallyconnectallindividuallocationssubsequentlyoccupiedbythebird,weend

upwithastaticone-dimensionalentity,justlikethepowerlinein(4b).3

As these examples show, over can be applied to a variety of spatial configurations.

However, these configurations do not constitute an arbitrary collection. Instead, the

3AsLakoff’suseofdouble-pointedarrowsindicates(cf.1990b:442f.),allofthesetransformationsarereversible.Theend-pointin(2b)presupposesapath,whichcanbementallyreconstructed.Orimagine

lookingatsandthroughamagnifiyingglass,sothattheindividualgrainsbecomediscernable(massto

multiplex).Similarly,wecanmentallytracethepowerlinein(4b)frombeginningtoendinthesame

fashionwewouldfollowazero-dimensionalobject(“Thepowerlinerunsovertheyard.”).

11

varioususesofover in(2)-(4)aremotivatedbyschemassuchasSOURCE-PATH-GOAL,

MULITPLEX,MASS,etc.aswellasthementaloperationsweperformonthem.Crucially,

these representations are neither propositional nor inherently linguistic in nature. As

Lakoff puts it, “image schemas are a reflection of our sensory and general spatial

experience” (1990b: 443). Furthermore, they enable abstract thought by serving as

input for metaphorical source domains. For example, the UP-DOWN axis can lend

imagisticstructuretothedomainofcontrol,thusmotivatingexpressionssuchasShehas

astrangepoweroverme(Lakoff1990b:435f.).4

Like Johnson, Lakoff concludes that image schemas “structure our perceptions and

that theirstructure ismadeuseof inreason”(1990b:440).Andcertainly,subsequent

applicationsof image schema theory to topics asdiverseasEnglishmodals (Sweetser

1991),caseinGerman(Smith1992),mathematicalreasoning(LakoffandNúñez2000),

andliterarytheory(LakoffandTurner1989;Turner1991)providestrongevidencein

favor of this hypothesis. However, at the same time the notion of image schema, as

introducedbyJohnsonandLakoff,remainsrathervagueinseveralrespectsandthishas

ledtosomecontroversyregardingitsexactcharacterization.Letustakeinventory,then,

ofwhatwe know about image schemas at this point. So far,we have established the

following:

• Imageschemasareentrenchedsensory-motorpatternsthatemergefrom

continuedbodilyinteractionwithourenvironment.

• Asembodiedrepresentationstheyplayanimportantroleinstructuringboth

perceptionandthought.

• TheyareimagisticinthesensethetermimageisusedinGestaltpsychology,i.e.imageschemasarenotmerelyvisualbutmultimodalrepresentations.For

example,schemaslikeBALANCE,BLOCKAGEorCONTACTrelyheavilyonnon-

visualsense-data.

• Theyareschematicinthesensethattheyaremaximallyunderspecified.Inother

words,eventhemostgeneraldepiction(adrawing/diagram)of,say,acontainer

wouldstillhavetospecifytheboundaries,shape,size,degreeofopacityetc.of

thecontainertosomedegreeandthereforefallshortintermsofschematicity.

(Furthermore,adiagramnecessarilyfailstocapturethedynamicandmultimodal

aspectsofanimageschema.)

Yet,aswewillseeinthefollowingsection,theexactdefinitionofthetermimageschema

remainsahotlycontestedtopicamongresearchers.

4Iwilldiscussmetaphoricalmappingsinmoredetaillateron(seechapter4).

12

2.2.RefiningandDefiningtheNotion:DivergingOpinions

Aquickreviewoftheliteraturerevealsthatdifferentresearchersholddifferentopinions

as to what counts as an image schema. Themost liberal use of the term is probably

foundinTurner(1991):

Thefollowingarealldifferent imageschemas:Acirclewithamarkedpointat itscenter;acirclewitha

markedpointsomewhereelseinteriortotheboundary;acirclewithamarkedpointontheboundary;and

acirclewithamarkedpointexteriortotheboundary.[...]Theimageschemaassociatedwithcupneednothaveanexactdegreeofcurvature;butifweflattenoutthesidestothepointthattheimageschemabegins

toapproachanimageschemaassociatedwithplate,forexample,thenwehavenotpreservedtheoriginalimageschema.(Turner1991:177)

Following Johnson’s (1990) understanding of images as multimodal gestalts, Turner

(1991: 57) describes image schemas as “extremely skeletal images that we use in

cognitive operations”. However, as the above passage shows, this characterization is

apparently broad enough to include complex geometrical constellations as well as

relativelyrichvisualimages(cup,plate)intothecategory.

Anentirelydifferent approach is takenbyClausnerandCroft (1999), forwhom the

criteriumofpervasivenessiscentral.Theysuggestthatimageschemascanbeequated

withthosebackgroundknowledgestructures(i.e.domains)thatarepresupposedbythe

largestnumberofconcepts:

Forexample,thedomainAPPLEisconcrete,thatis,itisrelativelynon-schematic.Inthescopeofhuman

experienceitispresupposedbyrelativelyfewotherdomains(e.g.,CIDER).Ontheotherhand,almostall

domainsmake some reference to SCALES; for example, anydomain involvinggradableproperties.Also

IDENTITY and SIMILARITY can be found in nearly every concept profile. The domains of TIME and

CHANGE(thatis,thePROCESSimageschema)canbefoundinthematrixofanyeventorprocessconcept.

An enormous number of domains involving physical objects ormotion include SPACE in their domain

matrix.

These factssuggestanaturaldefinitionof imageschematicity:domainswhichareimageschematicarethose found in the largest number of domain matrices (for the concepts used in human experience).(ClausnerandCroft1999:21-22)

While this characterization excludes rich images such as cup and plate, it allows

extremely general concepts such as the basic domains of TIME and SPACE into the

category.Infact,basedonthepervasivenesscriterium,thesebasicdomainswouldbefar

betterexamplesofimageschemasthantheCONTAINERschema,whichisonlygranted

peripheralmembershipbyClausner andCroft (1999:22). Evidently, this clasheswith

the understanding of image schematicity in Johnson (1990: 126), who includes

CONTAINER (but neither SPACE nor TIME) in his list of “the more important image

schemata.”

13

Grady, on the other hand, stresses the importance of perceptual grounding in his

definitionofimageschemasas“mentalrepresentationsoffundamentalunitsofsensory

experience” (2005:44).Heargues that several schemas suchasCYCLEorSCALE (see

Johnson 1990: 126) fail to qualify as image schemas because they are not inherently

perceptual: “While schemas like CYCLE and SCALE may be strongly associated with

perceptual content such as CIRCLE, PATH, etc., the schemas are also recognizable as

free-standingconceptsintheirownright,referringtobasic(nonsensory)dimensionsof

phenomenological experience, independent of the sensory associations”(Grady 2005:

41).Butisthere,inprinciple,anyreasontoexcludeourexperienceoftemporalpassage

from the realmofperception?Surely, themechanisms that enableus tokeep trackof

timearebiologicallynolessrealthanourothersenses.Andcanwetrulyconceiveofthe

SCALEschemainawaythatis“independentofsensoryassociations”?Itshouldbekept

inmindthatscalarityandgradablequality(e.g.amount,intensity,etc.)arenotthesame

thing – scalarity is the superimposition of gradable quality onto the SOURCE-PATH-

GOAL schema. As these examples show, it is extremely difficult to make a clear-cut

binarydistinctionbetweenwhatisinherentlyperceptualandwhatismerelyassociated

withperceptualcontent.Rather,thecategoryofperceptualrepresentationitselfseemsto

exhibitprototype-effectsandgradedmembership.

InarecentpaperMandlerandPagánCánovas(2014)attempttodefinethenotionof

image schema fromadevelopmental perspective.They argue that the term shouldbe

reserved for those “[r]epresentationsof simple spatial events” (2014:17) that infants

relyonmostheavilyinordertomakesenseoftheirsurroundingsuptotheageofsixto

seven months. They note, for example, that infants are aware of occlusion and

containment events from the ageof twoandahalfmonths (2014:6) and acquire the

conceptofagoal-directedmotioneventataboutfivemonths(2014:8).Thesuggestion

isthatwedifferentiatebetweenthesesimpleevents(e.g.PATHTOTHING,THINGINTO

CONTAINER), the building blocks they aremade up of (e.g. PATH, CONTAINER), and

morecomplexrepresentations thatemergebyaddingnon-spatialelements (e.g. force,

time, emotion) to spatial events (2014: 17). As a consequence, many image schemas

from Johnson’s list (1990: 126) such as PATH, LINK, THING or CONTAINER are

14

“demoted” to spatial primitives while others, such as FORCE, are considered part of

morecomplexschematicblendsduetotheirnon-spatialnature.5

All of this shows that there is very little consensus on how to define the notion of

image schema. In particular, we see that different researchers emphasize different

aspects: For Clausner and Croft pervasiveness/schematicity is central. According to

Grady,thecriteriumofperceptualgroundingtakesprecedence.AndMandlerandPagán

Cánovassuggest that imageschemasarebestunderstoodas simplestorieswhichare,

firstandforemost,dynamicandspatial.Itisworthnotingthatalloftheseaspectsplaya

role in Johnson’s and Lakoff’s early characterizations of image schemas. However, in

their work no particular aspect seems to utterly outrank any other aspect. When

Johnson revisits the notion in a later paper, he gives a relatively inclusive and broad

characterization of an image schema as “a dynamic recurring pattern of organism-

environment interactions” that “will often reveal itself in the contours of our basic

sensory-motor experience” (2005: 19). Instead of postulating a set of definitional

criteria, Johnson suggests thatmost image schemaswill “show themselves” through a

methodof“informalphenomenologicalanalysis”(2005:21).Hegivesseveralexamples:

Askyourselfwhatthemostfundamentalstructuresofyourperception,objectmanipulation,and bodily

movement are, given thathumanbodies share several quite specific sensory-motor capacities keyed to

thesizeandconstitutionofourbodiesandthecommoncharacteristicsofthedifferentenvironmentswe

inhabit.Certainobviouspatterns immediately jumpoutatyou.Forexample,given therelativebilateral

symmetryofourbodies,wehaveanintimateacquaintancewithright-leftsymmetry.[...]Becauseofour

particularembodiment,weprojectRIGHTandLEFT,FRONTandBACK,NEARandFAR, throughout the

horizonofourperceptualinteractions.[...]Becauseofourongoingbodilyencounterwithphysicalforces

thatpushandpullus,weexperiencetheimageschematicstructuresofCOMPULSION,ATTRACTION,and

BLOCKAGEOFMOVEMENT....[...]Becausewemustcontinuallymonitorourownchangingbodilystates,

weareexquisitelyattunedtochanges indegree, intensity,andqualityof feelings,which is thebasis for

oursenseofscalesofintensityofaquality(theSCALARITYschema).Becausewemustconstantlyinteract

withcontainersofallshapesandsizes,wenaturallylearnthe“logic”ofcontainment(fortheCONTAINER

schema).(Johnson2005:20-21)

Onemight be inclined to dismiss this sort of survey as too subjective and vague, and

object that itdodgesthereal issue.But Ibelievethat Johnsonhasgoodreasonfornot

attempting to give a defintion of image schematicity. To be sure, image schema has

become a technical term and a key notion in Cognitive Linguistics and it is therefore

crucial to make sure that everyone is talking about the same thing. Nevertheless,

expectingadefinition tosolve the issueseemsoddtobeginwith–especiallywithina

theoretical framework that has time and time again critisized how definitional

5“[...]sofarasweknow,prelinguisticimageschemasarestrictlySPATIAL.[...]Neitherforcenoranyothernon-imageableinformationisavailabletotheconceptualsystemwhenimageschemasbegintobeformed”

(MandlerandPagánCánovas2014:17-18).

15

approachestypicallymisrepresentcategorystructure.Afterall,wewouldnotexpecta

setofdefinitionalcriteriatosettleonceandforalltheseeminglymundanequestionof

what a table is. Carpenters might eventually agree on a definition as the result of a

lengthyinternationalconference,butthenewnotionwouldbeartificialandsomewhat

arbitrary.Ithereforeproposethatweshouldavoidanimpoverishedexpertdefinitionof

imageschematicityinlieuoftheviewthatthecategoryimageschemaexhibitsprototype

effectsandgradedmembership(Imadethesamesuggestionaboveregardingthenotion

perceptualrepresentation).Fromthisperspective,allofthepreviouslydiscussedcriteria

contribute to category structure. That is, a prototypical image schema is cognitively

pervasive,groundedinperception,dynamicandspatial.Itisimportanttonote,however,

that not all image schemas rank equally in respect to each of these parameters. For

example,aFORCEschemalikeCOMPULSIONishighlydynamicbutdoesnotnecessarily

evoke thespatialdomainasstronglyas theCONTAINERschemadoes.Conversely, the

CONTAINER schema is less dynamic: While the schema is inextricably linked to our

experiencesofthingsmovingINandOUT,astaticCONTAINERisstillacoherentmental

representation.SchemaslikeOBJECTandPATHaremorepervasive(i.e.presupposedby

alargernumberofcognitivedomains)thanLEFTandRIGHT.TheITERATIONschema,

while grounded in perceptual experience, has no perceptible instances. (I.e., the

perceptibleeventsthatareiteratedareinstancesoftheeventschema,notinstancesof

theITERATIONschema.)NowcomparethistotheLINKschemaanditsinstances,which

arethemselveslinks.

Theconclusionisratherstraightforward.Thebestexamplesofthecategoryarethose

imageschemasthatrankhighlywithrespecttoalloftheparameters.Attheotherendof

the spectrumwe have fringemembers that rank low in average orwith respect to a

particular parameter. From this perspective it is probably impossible to exhaustively

enumerateallimageschemasinexistence.Butthisdoesnotimplythat“anythinggoes”,

either.Forinstance,wecanconfidentlyexcluderelativelyrichvisualimagessuchascup

andplatefromtheimageschemacategoryfortheirlackofpervasiveness.Forthesame

reason,imageschemascannotbeinfinitelycomplex.Infact,CONTAINER–withitssub-

schemasINTERIOR,EXTERIOR,andBOUNDARY–seemstobeamongthemorecomplex

ones.6

6Ihaveintentionallyleftoutdiscussionofonekeyaspectwithintheimageschemadebate,namelytheneuro-psychologicalstatusofimageschemas(seee.g.Gibbs2005;Rohrer2005).Thereareseveral

reasonsforthis:First,researchintothistopicisstillintheearlydevelopingstagesandhighlyspeculative.

16

3.Trajector/LandmarkOrganization(FigureandGround)

3.1LinguisticStructureasaReflectionofCognitiveProminence

Imagine(visualize)thescenesdescribedbythefollowingsentences:

(5) Thedogisrunningacrossthefield.

(6) Thebookisonthetable.

(7) Thelittlestarcirclesthebigstar.

Ineachofthesescenesoneparticipantstandsoutasparticularlysalientagainsttherest

of the environment: The dog, the book, and the little star. This asymmetrical

segmentationofthevisualfieldintoaprominentfigureandalessprominentgroundisa

basicprincipleofhumanperceptionandacornerstoneofGestaltpsychology(seeRubin

[1915]1958;Bahnsen1928;Koffka1935).

Cognitive linguists (e.g. Talmy 1975; Langacker 1987, 1991) have subsequently

adapted the notion of figure/ground organization into their frameworks on the

assumption that linguistic structure reflects conceptual organization. Again, consider

(5)-(7): The prominent subject role and sentence initial position are reserved for the

entity corresponding to the figure of the encoded scene,while the ground appears as

directorprepositionalobject.AccordingtoTalmy,figureandgroundhavethefollowing

associatedcharacteristics:

figure ground

• moremovable • morepermanentlylocated

• smaller • larger

• geometricallysimpler(often

pointlike)initstreatment

• geometricallymorecomplexinits

treatment

• morerecentlyonthescene/in

awareness

• morefamiliar/expected

• ofgreaterconcern/relevance • oflesserconcern/relevance

• lessimmediatelyperceivable • moreimmediatelyperceivable

• moresalient,onceperceived • morebackgrounded,oncefigureis

perceived

• moredependent • moreindependent

(adaptedfromTalmy2003a:315-16)

Secondly,discussionwouldrequiretheintroductionofnumerousexperiments,whichinturnpresuppose

conceptsfromneuroscience.Aproperrepresentationofthedebatewouldthereforegobeyondthescope

ofthisintroduction.Andlastly,whileanchoringthenotionofimageschemaintheempiricalsciencesis

ultimatelyamatteroffundamentalimportance,noknowledgeofthetopicisrequiredforunderstanding

anyoftheargumentspresentedinthisthesis.

17

BasedonTalmy’swork,Langackerhasdemonstratedthat figure/groundasymmetry

playsafundamentalroleintheanalysisofgrammaticalstructure.7Hereferstothemost

prominentparticipantofaprofiledrelationasthetrajector(TR)andtothesecondmost

prominentparticipantasthelandmark(LM).Bywayofillustration,consider(8):

(8) [thelamp]TRabove[thetable]LM

AccordingtoLangacker,theprepositionaboveencodesaspatialrelationbetweentwo

schematicentities.Theseentitiesareelaborated(instantiated)bythelampandthetable.

Itisimportanttonote,however,thatabovedoesnotconstruethetwoasequallysalient.

I.e., the thing profiled by the phrase the lampabove the table is a lamp, not a table.8

Hence,thelampisthemostsalientparticipantintherelationandcorrespondstotheTR

ofabove,while thetableassecondmostprominentparticipantcorrespondstheLMof

above(Langacker1990:25).Insummary,thetermsTRandLMarefunctionallyroughly

equivalent to the notions figure and ground. For Langacker, TR/LM alignment is an

instanceoffigure/groundasymmetrypertainingtotheleveloflinguisticstructure.But

since the terms TR/LM have become prevalent in Cognitive Linguistics (e.g. Lindner

1981; Lakoff 1990b), I will use them exclusively, evenwhere non-linguistic cognitive

structureisconcerned.

3.2.RelevanceforImageSchematicStructure

Recall fromthepreviousdiscussionthat imageschemasareimagisticandgroundedin

perception.Assuch,theprincipleofTR/LMorganizationwillnaturallyplayarolewhen

wetalkaboutagivenschema.TaketheCONTACTschema,forinstance.CONTACT,even

when considered in the most abstract, presupposes at least two schematic entities

betweenwhichtherelationobtains.ConfrontedwithascenewhereAandBarerelated

via CONTACT,wewill recognize one entity as the TR and the other entity as the LM.

WhilethebareimageschemaCONTACTbyitselfisneutralintermsofTR/LMalignment,

any scene which instantiates the schema will necessarily have to be construed in a

certainway,i.e.forceustoimposeTR/LMorganizationuponit.Inotherwords,wewill

7OneofthemostsignificantconsequencesofthisanalysisisLangacker’scharacterizationofthegrammaticalnotionssubjectanddirectobjectasclausaltrajectorandlandmark,respectively(1987:324).8IntheterminologyofCognitiveGrammarlampistheprofiledeterminantofthephrase.

18

(16a) Thepaperplaneflewintothehouse.

(16b) Janethrewthepaperplaneintothehouse.

The preposition into profiles a spatial relation between a CONTAINER (the LM) and

someother entity (theTR)whichmoves from itsEXTERIOR to its INTERIOR. Inboth

sentences the TR and LM of into are elaborated by the paper plane and the house,

respectively. However, in (16b) the paper plane also corresponds to the LM of threw

(whiletheTRofthrewiselaboratedbyJane).Furthermore,inJapanese,themotionpath

expressedbytheprepositional“satellite“intoiscodedbytheverbitself9:

(17a) Saru-ga ori-ni hait-ta.

Monkey-NOM cage-DAT enter-PAST ‘Themonkeyenteredthecage.’

(17b) Tarô-ga saru-wo ori-ni ire-ta.

Tarô-NOM monkey-ACC cage-DAT putinto-PAST

‘Tarôputthemonkeyintothecage.’

Note that (17) expresses roughly the same schematic content as (16): Y moves into

CONTAINER(aversion)andXcausesYtomoveintoCONTAINER(bversion).Althoughin

(17b)wecannotsaythatsaruelaboratestheTRofsomeovertrelationalelement(like

intointheEnglishsentences),wecansaythat–asfarasmentalimageryisconcerned–

saru corresponds to the TR (the figure) of what we might call the “entry relation“.

Therefore,whenIspeakof“theTR“or“theLM“throughoutthisthesisIwillbereferring

to the level of image schematic structure. For example, in regards to (17b) – and

focussingontheENTRYschema–Iwouldrefertosarusimplyas“theTR“eventhough

theTRofireru(i.e.themostprominentparticipantoftheprocess)iselaboratedbyTarô.

Conversely,whenreferringtothelevelofgrammaticalstructure,Iwillbeasexplicitas

possible, using terms such as “clausal TR“, etc. As mentioned above, this use of

terminologyismainlyaconcessiontotheverb-framednatureofJapanese.

4.Metaphor

4.1.Complex,Primary,andImageMetaphor

Conceptualmetaphor,asfirstdescribedbyLakoffandJohnson([1980]2003),isabasic

principleofhumancognitionwherebyoneconceptualdomain(thetarget)isstructured

9SeeTalmy’s(1991)distinctionbetweensatellite-framedvsverb-framedlanguages.

20

and understood in terms of another conceptual domain (the source). To illustrate,

consider the classic example LOVE ISA JOURNEY,which underlies utterances such as the

following:

Lookhowfarwe’vecome.It’sbeenalong,bumpyroad.Wecan’tturnbacknow.We‘reatacrossroads.Wemayhavetogoourseparateways.Therelationshipisn’tgoinganywhere.We‘respinningourwheels.Ourrelationshipisoffthetrack.Themarriageisontherocks.Wemayhavetobailoutofthisrelationship.

(fromLakoff2006:189)

Accordingtometaphortheory,wecanthinkandtalkaboutthetargetdomainLOVEin

termsofthesourcedomainJOURNEY,becauseasetofmappingsobtainsbetweenthem

(Lakoff2006:190ff.):

JOURNEY LOVE

travellers --> lovers

vehicle --> relationship

impedimentstotravel --> difficultiesencountered

crossroads --> choices

destination --> commonlifegoals

etc.

Thus, themnemonic shorthand LOVE IS A JOURNEY is neither a proposition nor a single

metaphorical expression. Instead it is the name for a cognitive operation thatmaps a

complex body of knowledge about our experienceswith journeys onto the domain of

LOVEviaasetofsystematiccorrespondences.

Note, however, that there is no obvious connection between love relationships and

journeysintherealworldthatcompelsustodrawaconnectionbetweenthesedomains.

Loveand journeysareexperiencesofquiteadifferentnature thatdonot typicallyco-

occur. How, then, is the metaphor motivated? Why does JOURNEY “fit” as a source

domainforLOVE?Whycanwemaptravellersontolovers,destinationsontolifegoals,

andsoon?

Inorder toanswer thequestionofmotivatedness, it is important to realize that the

metaphorLOVEISAJOURNEYcanonlyexistinvirtueofalargersystemofmoreschematic

21

metaphors collectively known as the EVENT STRUCTURE METAPHOR. Lakoff (2006: 204)

characterizesthissystemasfollows:

• Statesarelocations(boundedregionsinspace).

• Changesaremovements(intooroutofboundedregions).

• Causesareforces.

• Actionsareself-propelledmovements.

• Purposesaredestinations.

• Meansarepaths(todestinations).

• Difficultiesareimpedimentstomotion.

• Expectedprogressisatravelschedule;ascheduleisavirtualtraveler,who

reachesprearrangeddestinationsatprearrangedtimes.

• Externaleventsarelarge,movingobjects.

• Longterm,purposefulactivitiesarejourneys.

It immediately springs to attention that LOVE IS A JOURNEY is an instance of long term,

purposeful activities are journeys. But there is more to it. All of the above mappings

constitutegeneric-levelmetaphorsintheirownright.Morespecifically,themajorityof

themarewhatGrady(e.g.1997a,1997b,1999)callsprimaryorcorrelationmetaphors.

Primarymetaphors are different from complex conceptual metaphors like LOVE IS A

JOURNEY in twomajor respects. First, they only involve a singlemappingbetween two

experientially equally basic domains. As Grady (1997a: 26) puts it, primary source

domains have image content: “[...]content which is tied to physical perception or

sensation. The feeling of an itch; the perception of shape, weight, and distance; the

detection of movement – all of these experiences involve the (apparently) direct

perceptionoffeaturesofourbodiesorourenvironments.”Primarytargetdomains,on

theotherhand,haveresponsecontent:“Theyarenotdirectperceptionsoftheworld,but

responsesto[...]ourperceptionsoftheworld”(1997a:26).Inotherwords,whileimage

and response content differ in kind, they do not differ in degree of abstractness or

complexity. Secondly, primary metaphors are directly motivated by the experiential

correlation of source and target domain. To illustrate this point, consider STATES ARE

LOCATIONS:

IfIaminaveryhotplace[...]Iwillfindmyselfinastateofdiscomfort.Moregenerally,itmaybethecase

that we form metaphorical associations between certain sensations and the perception of being in

particularplaces–i.e.thecorrelationinourexperiencebetweentheplacesandthestatesleadstobinding

betweentheconcepts.(Grady1997a:106).

Returning to LOVE ISA JOURNEY, we now see that themetaphor is decomposable into

several primary metaphors such as STATES ARE LOCATIONS, PURPOSES ARE DESTINATIONS,

22

DIFFICULTIESAREIMPEDIMENTSTOMOTION, and so forth. I.e., the states, commongoals, and

difficulties of a love relationship can be understood in terms of the locations,

destinations,andterrainobstaclesofajourneybecausethesourceandtargetdomains

oftherespectiveprimarymetaphors–whichinturnmakeupthecomplexmetaphor–

correlateinexperience.10

Inadditiontocomplexmetaphorsandtheprimarymetaphorsthatconstitutethemwe

need to briefly consider a third kind of projection beforewemove on. The following

expressionsareexamplesofimagemetaphors:

(18) hourglass-waist(basedonLakoffandTurner1989:90)

(19) submarinesandwich(fromBenczes2006:108)

(20) barcodehairstyle(fromBenczes2006:110)

Allof theaboveare“[o]ne-shot imagemappings”(LakoffandTurner1989:91)which

project the shapeproperties of the sourceonto the target concept.Note, for example,

thatasubmarinesandwichderivesitsnameexclusivelyfromtheschematicsilhouetteof

a submarine. No other aspects of the SUBMARINE domain are mapped onto the

SANDWICHdomain(asopposedtothenumerouscorrespondencesinLIVEISAJOURNEY).

Nor is there any salient experiential correlation between submarines and sandwiches

(asopposedtoPURPOSESAREDESTINATIONS).

4.2.MetaphorandMeaningExtension

In the case studies of this thesis we will be interested in metaphor mainly as a

mechanismofmeaningextension.Letusthereforeconsiderseveralexamplesfromthe

literaturethatillustratehowmetaphorfunctionsasadrivingforcebehindpolysemy.

Inherseminal1981study,Lindneranalysesthesystematicrelationshipsbetweenthe

sensesofverbparticleconstructionswithoutandup.Thefollowingpassagedealswith

therelationbetweenspatialupandandwhatwemaycalltheactivitysenseofup:

10Itisworthpointingoutthatthenotionsofimagecontentandimageschemaareco-extensionalforalargenumberofprimarysourcedomains(Grady1997a:179).Observe,forexample,thatthemappings

whichconstitutetheEVENTSTRUCTUREMETAPHORhavethefollowingimageschemasamongtheirsource

domains:MOTION,FORCE,PATH,GOAL,BLOCKAGE,OBJECT.Lakoff’s(2006:199)invarianceprinciple

holdsthat“[m]etaphoricalmappingspreservethecognitivetopology(thatis,theimage-schema

structure)ofthesourcedomain,inawayconsistentwiththeinherentstructureofthetarget

domain.“Fromtheperspectiveofprimarymetaphortheorythispreservationofimageschematictopology

isanemergentaspectoftheexperientialcorrelationbetweenprimarysourceandtarget.Forexample,

destinationsaremappedontopurposesbecausedestination-reachingandpurpose-fulfillmentnaturally

co-occur.

23

[...]whensomethingisinhand,itisavailableforuseoraction.ThuswhenIpickuportakeupmysword,I

amreadytofight.Moreabstractlywemaypickupaconversationwhereweleftoff(‘continueactingonit’)

ortakeupsailing(‘incorporateintotherangeofouractivities’).(Lindner1981:161)

With the gift of hindsight (Lindner’s study falls into the earliest days of metaphor

theory),wecanattributethissemanticextensiontotheprimarymetaphorACTIVITYISUP.

Infact,whatthepassagecitedabovedescribesisexactlytheexperientialcorrelationthe

metaphorisbasedon:elevationtohand-levelandreadinessforuse/action.

Several extensions of over into the non-physical domain are discussed by Lakoff

(1990b).Recalltheexamplementionedbrieflyin2.1.:

(21) Shehasastrangepoweroverme.

AsLakoff(1990b:435)pointsout,thissenseisbasedonthemetaphor(s)CONTROLIS

UP/LACK OF CONTROL IS DOWN. Again, the experiential correlation grounding this

primarymetaphorshouldbeevident:controlisassociatedwithanelevatedpositionand

anoptimaloverview.

Finally,consideraslightlymorecomplexexampleconcerninggrammaticalratherthan

lexical polysemy. Sweetser (1991) has famously claimed that the epistemic sense of

modals suchasmust andmay aremetaphoricallyderived fromtheir root (ordeontic)

sense.Her argument is basedon a systemof correspondencesknownas themind-as-

body metaphor for which she provides extensive diachronic and cross-linguistic

evidence – mainly in the form of mappings from the sense perceptual to the mental

domain(Sweetser1991:38):

senseperceptual/externaldomain mental/internaldomain

VISION --> KNOWLEDGE

HEARING --> INTERNALRECIPTIVITY

FEEL --> EMOTION

TASTE --> PERSONALPREFERENCE

AsSweetser(1991:45)putsit,“[t]heinternalselfispervasivelyunderstoodintermsof

thebodilyexternalself,andishencedescribedbymeansofvocabularydrawn[...]from

thephysicaldomain.”Inlightofthissystem,whichwemightsumupunderthegeneric

level shorthand INTERNALISEXTERNAL, themodalsmay andmust canbeparaphrased in

thefollowingfashion(adaptedfromSweetser1991:61):

(22a) Johnmaygo.(root)

24

“Johnisnotbarredbymyorsomeotherauthorityfromgoing.”

(22b) Johnmaybethere.(epistemic)

“Iamnotbarredbymypremisesfromtheconclusionthatheisthere.”

(23a) Youmustcomehomebyten.(Momsaidso.)(root)

“Thedirectforce(ofMom’sauthority)compelsyoutocomehomebyten.”

(23b) Youmusthavebeenhomelastnight.(epistemic)

“Theavailable(direct)evidencecompelsmetotheconclusionthatyouwerehome.”

That is, the epistemic senses are derived from the root senses bymapping the image

schemas FORCE, BLOCKAGE, and ENABLEMENT from the external sociophysical onto

theinternalpsychologicaldomain.

5.Metonymy

5.1.MetonymicShiftsandEncyclopedicKnowledge

Metonymy isanothermappingprocesscentral tohumancognition.Here,oneconcept

serves as a point of access for another concept from within the same experiential

domain or domain matrix (e.g. Langacker 1993: 30; Kövecses and Radden 1998: 39;

Barcelona2000:37).Thefollowingexamplesillustratethis:

(24) TheGiantsneedastrongerarminrightfield. (BODYPARTFORPERSON)

(25) I’llhaveaLöwenbräu. (PRODUCERFORPRODUCT)

(26) Thesaxhastheflutoday. (OBJECTFORUSER)

(27) Watergatechangedourpolitics. (PLACEFOREVENT)

(fromLakoff&Johnson2003:38-39)

Thenotionofadomain,as introducedbyLangacker(1987:147ff.), isessential to the

encyclopedicviewofsemanticsassumedbycognitive linguists.Onthisview, linguistic

itemsarepointsofaccesstopotentiallyopen-endedbackgroundknowledgestructures

(as opposed to being definable by feature bundles). For example, thewordknuckle is

understoodrelativetotheconceptualdomainFINGER.Ifagivenconceptisunderstood

relative tomore thanonedomain (which is thenorm),wespeakof itsdomainmatrix

(Langacker 1987: 147). Thus, additional domains in the domain matrix of KNUCKLE

include HUMAN BODY PARTS, ANATOMY, BAR FIGHTS, and so forth. Note that some

domainsarestrictlypresupposedbyaconcept(FINGERforKNUCKLE)whileothersare

moreperipheral (e.g.BARFIGHTS).Nevertheless, all arepartof thedomainmatrixas

theycontributetoourencyclopedicknowledge.

25

Wecannoweasilyseehowtheabovesentencesinvolveadomaininternalshiftfrom

oneentitytoanother.(24)isunderstoodagainsttheexperientialdomainofBASEBALL,

which includes theplayers, their functions and thebodypartsused in the game. (25)

evokesaRESTAURANTdomain–aconceptualstructurethatincludesdrinksasphysical

objectsaswellasknowledgeabouttheirproducers,branding,andsoon.Thedomainof

MUSICEVENTSincludesmusiciansandtheirinstrumentsamongitsparts.ThePOLITICS

domainincludesknowledgeofpoliticaleventsaswellastheplacestheyoccurat.

AspointedoutbyLangacker(1993)andKövecsesandRadden(1998),metonymical

shiftsarenotarbitrarybutgovernedbyprinciplesofrelativesalience.Accordingtothe

classificationofKövecsesandRadden (1998:63ff.), relative salience isdeterminedby

four major factors: human experience (e.g. HUMAN OVER NON-HUMAN, CONCRETE OVER

ABSTRACT), perceptual selectivity (e.g.MORE OVER LESS, GOOD GESTALT OVER POOR GESTALT),

culturalpreferences(e.g.STEREOTYPICALOVERNONSTEREOTYPICAL,IDEALOVERNON-IDEAL),and

communicative principles (CLEAR OVER LESS CLEAR, RELEVANT OVER IRRELEVANT).

Furthermore,they(1998:71ff.)observethatseveralconflictingprinciplesmayapplyat

thesametime.Forexample,in(26)theprincipleHUMANOVERNON-HUMANisatoddswith

RELEVANTOVERIRRELEVANT. In such cases, oneormoreprinciples canbeoverridden for

“social,communicativeoraestheticreasons”(1998:71).Giventhecontextof (26), the

musician’s functionofplaying the sax is considered themost important aspectby far,

whilehisorherotherhumanqualitiesarehardlyrelevant.Hence, thePRINCIPLEHUMAN

OVERNON-HUMANisreversedinfavorofRELEVANTOVERIRRELEVANT.

5.2.MetonymyandMeaningExtension

Aswithmetaphor,wewillbeconcernedwithmetonymyprimarilyasamechanismof

semanticextension.Toillustrate,considertheentryfortheverbpaint11intheMerriam-

WebsterDictionary:12

transitiveverb

1 a (1) toapplycolor,pigment,orpaintto

(2) tocolorwithacosmetic

b (1) toapplywithamovementresemblingthatusedinpainting

(2) totreatwithaliquidbybrushingorswabbing

11TheverbpaintisalsousedasanexamplebyTuggy(1993)inhisdiscussionofambiguity,polysemyandvagueness(see7.3.),althoughnotspecificallyinthecontextofmetonymy.12"Paint."Merriam-Webster.com.,retrieved18Aug.2016.

26

2 a (1) toproduceinlinesandcolorsonasurfacebyapplyingpigments

(2) todepictbysuchlinesandcolors

b todecorate,adorn,orvariegatebyapplyinglinesandcolors

c toproduceorevokeasifbypainting

3 totouchuporcoveroverbyorasifbypainting

4 todepictashavingspecifiedorimpliedcharacteristics

intransitiveverb

1’ topracticetheartofpainting

2’ tousecosmetics

Recall fromabovethenotionofadomain.WhatelementswouldthedomainofPAINT

include?Surely,itincludestheapplicationofasubstanceontoasurface.Thisisdonefor

some purpose which can be artistic or utilitarian. There is a set of characteristic

movements involved.Further,painting canhaveavarietyof effects, i.e. the result can

have an expressive, evocative and/or representational function. All of this (andmuch

more)ispartofourencyclopedicknowledgeaboutpainting.

Now, allowing for the possibility that the entry cited is not an entirely accurate

representation of semantic structure, we still get a rough idea of what some of the

sensesofpaintareandhowtheydifferfromoneanother.Themainpointisthis:Manyof

thesensescharacterizedabovecanbedistinguished in termsof therelativeweightof

the various elements constituting the domain(-matrix) of PAINT.13This is a salience-

basedphenomenonwhich is essentiallymetonymic in character (see also Croft 1993:

348).Consider,forexample,aninstanceof1b(2):

(28) Thedoctorpaintedthewoundwithiodine.

Here, theuseofpaint is licensedbyshiftingthebulkofsaliencetothesubdomainof

ASSOCIATED CHARACTERISTIC MOVEMENT and away from other aspects such as

DEPICTION,ARTISTICEXPRESSION,etc.Comparethistothedistributionofsaliencein

thefollowingsentence,aninstanceof1’:

(29) Thebeautyoflifeinspireshertopaint.

Inthiscase,paintislicensedbyforegroundingtheaestheticandexpressiveaspectsof

paintingabsentin(28)whileignoringallutilitarianaspects.

13SeealsoLangacker’snotionofcentrality:“Themultitudeofspecificationsthatfigureinourencyclopedicconceptionofanentityclearlyformagradationintermsoftheircentrality.Somearesocentralthatthey

canhardlybeomittedfromeventhesketchiestcharacterization,whereasothersaresoperipheralthat

theyholdlittlesignificanceevenforthemostexhaustivedescription”(1987:159).Wecanthereforesay

thatmanysensesofpaintdifferintermsofthecentralityofdomain-internalelements.

27

6.ANoteontheRelationbetweenMetaphorandMetonymy

Aparticularlychallengingaspectofthetheoryofconceptualmetaphorandmetonymyis

theinteractionanddistinctionbetweenthetwo(seee.g.Goossens1990,Radden2000).

Insomeexpressionstherelationshipisquiteintricate:

(30) Helaiddownthepenandtookupthesword.

Here,asindicatedbylaydownandtakeup,theMETAPHORUPISACTIVE/DOWNISINACTIVEis

atwork (Lindner1981:161). In this regard, (30) isnodifferent fromasentence like

(31):

(31) Shetookupknittingasahobby.

However, (30) also features an encapsulatedmetonymy, i.e. pen and sword stand for

scholarship and warfare, respectively (INSTRUMENT FOR ACTION). Goossens (1990) has

coined the termmetaphtonymy for the interplaybetweenmetaphorandmetonymy in

linguisticexpressions.

Now,note thatmeaningextension, too,may involvemetaphorandmetonymyat the

sametime.Considerthefollowingsentence:

(32) AwitnessatJianGhomeshi’strialpaintedhimasaviolentegomaniac.14

Ontheonehand,thisuseofpaint(sense4listedintheMerriamWebsterentry)isclearly

metaphorical, since the domain of VERBAL DESCRIPTION is understood in terms of

PAINTING.Ontheotherhand,thedescriptiveandevocativefunctionofpainting(EFFECT

FORACTION) is part of our encyclopedic knowledge about the PAINTINGdomain – and

thus metonymic. In other words, the meaning extension process seems to have a

bipartitestructure,wherebythemetonymylicensesthemetaphor:

(i) Metonymy:EFFECTOFPAINTING(description,evocation)FORPAINTING

(ii) Metaphor:DESCRIBINGISPAINTING

14http://www.vice.com/read/lucy-decouteres-testimony-paints-jian-ghomeshi-as-a-violent-egomaniac,retrieved23Aug.2016.

28

Infact,thisexampleisnodifferentfromtheprimarymetaphorsdiscussedearlier.Recall

thatprimarymetaphorsarebasedontheexperientialcorrelationofsourceandtarget.

For instance, the classicMORE ISUP is grounded in our experience that an increase in

amount often causes an increase in hight (Lakoff and Johnson 2003: 16). Likewise,

painting and its descriptive/characterizing function are part of the same experiential

domain.Therefore, ifweunderstandmetonymyasamentalaccessoperationwithina

givenexperientialdomain(andnotmerelyatooloflinguisticreference)itcanindeedbe

claimedthatallmetaphorshaveametonymicbasis(Barcelona2000:51).Thereasoning

behind this is as follows: If complex metaphors (e.g. LOVE IS A JOURNEY) can be

decomposed into primary metaphors (e.g. STATES ARE LOCATIONS), and if primary

metaphorshaveametonymicbasis(e.g.thecorrelationbetweenbeingincertainstates

and being in certain locations), then all metaphors are ultimately motivated by

metonymy.15

7.PolysemyandLexicalNetworks

7.1.TheCaseforPolysemy

Whenalinguisticitemhasseveralinterrelatedmeanings,wespeakofpolysemy.Whenit

hasonlyasinglemeaning,wespeakofmonosemy.ThedominantviewwithinCognitive

Linguisticsisthatpolysemyisthenorm,nottheexception(e.g.Langacker1987,1991;

Lakoff1990b;Rice1992).Whenwelookatsomethinglikethedictionaryentryofpaint,

it seems innocuous enough to suggest that the word has more than one meaning.

However, instead of simply taking this view for granted, we should inquirewhy it is

preferableoverastrongmonosemyposition.Suchapositionwouldamounttotheclaim

thateven linguistic itemswithmanydifferentestablishedusageshaveonlyonehighly

schematic meaning that subsumes all the variants (see Rice 1992: 89). That is, the

variantsarenotfull-fledgedmeaningsintheirownright,butcreatedandunderstoodon

theflyinaccordancewithpragmaticprinciplesonthebasisofasinglesemanticvalue.

There are several problemswith this account. First, this viewof semantic structure

requires an all-encompassing meaning that is schematic enough for all category

members to be subsumed under. Inmany cases this seems extremely ambitious. For

15Notethatthisalsoappliestotheone-shotimagemetaphors(e.g.hourglass-waist,barcodehairstyle)mentionedearlier,sincethesearebasedonthecorrelationofobjectandshape.Missingfromthisdiscussionaresynaestheticmetaphors(e.g.loudcolor,sweetmusic).ButseeBarcelona(2000:35ff.)fora

metonymy-basedaccount.

29

example,itisunlikelythatthereisasinglemeaningofpaintthatisschematicforboth

HepaintedthehousegreenandHisstorypaintedherasthevillain.Yet,aswehaveseen

above,bothusagesarerelated. Inothercases, findingsuchasuper-schemaisoutright

impossible. Wittgenstein’s observation about the category game comes to mind as a

famouscounterexampleagainststaunchlymonosemicaccounts:

Considerforexampletheproceedingsthatwecall"games".Imeanboard-games,card-games,ball-games,

Olympicgames,andsoon.Whatiscommontothemall?–Don'tsay:"Theremustbesomethingcommon,ortheywouldnotbecalled'games'"–butlookandseewhetherthereisanythingcommontoall.–Forifyou lookat themyouwillnotseesomethingthat iscommontoall,butsimilarities, relationships,anda

wholeseriesofthematthat.[...](1953:section66)

Icanthinkofnobetterexpressiontocharacterizethesesimilaritiesthan"familyresemblances"; forthe

variousresemblancesbetweenmembersofafamily:build,features,colourofeyes,gait,temperament,etc.

etc.overlapandcriss-cross inthesameway.–AndIshallsay: 'games' forma family. [...](1953:section

67)

Secondly,inmanycaseswhereasuper-schemadoesexist,wecanreasonablydoubtits

semanticsignificanceonthegroundsofinsufficientcognitiveentrenchment.Itissimply

not plausible that a highly schematic conceptwhich abstracts away frommost of the

characteristicfeaturesofitsinstancesshouldbestoredinlong-termmemory,whilesaid

instances–whichwearegenerallymuchbetteracquaintedwith–arenot.AsLangacker

(1987:381)notes,“evenifanall-subsumingsuperschemacanplausiblybepositedfora

category, it may well be only minimally entrenched and have very little cognitive

salience.” Empirical support for this view comes from studies such as Rosch (1977,

1978),who suggests that categories are structured around prototype effects and that

theseeffectsaremostlikelytoemergeatintermediatespecificity(i.e.the“basiclevel”),

notatthelevelofhighestschematicity.16

Lastly, itshouldbenotedthatthestrongmonosemicviewhasbeensoattractivefor

generativeandformalsemanticistsnotleastbecauseofitsallegedeconomy:Whyposit

additionalsemanticentitieswhentheycanintheorybe“computed”onthefly,basedon

asinglevalue?Asarguedabove,thisconcernfliesinthefaceofcognitivereality.Foran

encyclopedicsemanticsthatembracesredundancyandoverlapinconceptualstructure

itisanon-issue(Langacker1987:275).

16Forexample,thefeaturesprototypicallyassociatedwiththecategorybird(abilitytofly,feathers,

averagesize,etc.)emergeatanintermediatelevelthatsubsumesonly“typical”memberssuchas

sparrows,doves,andsoforth.Asuper-schemathatsubsumesallmembers(penguins,ostriches,etc.)hastoabstractawayfromthesefeaturesandisthereforenotaswell-entrenchedinthecognitivesystem.

30

7.2.MakingSenseofSenses–SomeProposals

Given these considerations, and based on the general theoretical commitments of

Cognitive Linguistics outlined earlier, a lexical network model promises the most

accurate representation of conceptual structure (and therefore semantic structure).

According to this approach, linguistic categories are natural bundles of senses held

together by family resemblances (e.g. Brugman 1981, Lakoff 1990b). Of course, this

raisesthedelicatequestionofhowtoidentifyanddifferentiatethesensesofaword(or

anylinguisticconstruction,forthatmatter).Asmaybeexpected,thisisatopicofmuch

heateddebateamongcognitive linguists.Consider, forexample,someof thecriticisms

directed towards the seminalBrugman/Lakoff analysis ofover (Lakoff 1990b), one of

the earliest lexical network proposals. According to Lakoff (1990b: 420ff.), all of the

followingusagesofoverconstitutediscretesenses(i.e.meaningvariants):

(33) Thebirdflewovertheyard (extendedLM,nocontact)

(34) Theplaneflewoverthehill (vertical&extendedLM,nocontact)

(35) Thebirdflewoverthewall (verticalLM,nocontact)

(36) Samdroveoverthebridge (extendedLM,contact)

(37) Samwalkedoverthehill (vertical&extendedLM,contact)

(38) Samclimbedoverthewall (verticalLM,contact)

Thus, in the Brugman/Lakoff model each different combination of dimensional

parameters is granted the statusof ameaningvariant. Some linguists (e.g.Vandeloise

1990; Dewell 1994) have criticized this “full-specification approach” (Lakoff 1990b:

420)asarelapseintocompositionalfeatureanalysis.Forexample,insteadofrelyingon

LM specifications (i.e. the bracketedparameters above)Dewell suggests that the only

semanticextensionmechanismsrelevantforoverareimageschematransformationand

metaphor.He furtherargues that thecategorialprototype isnot theabove-and-across

schemapositedbyLakoff(1990b:419)butacurvedarcschema(Dewell1994:352ff.).

Others (e.g. Kreitzer 1997; Tyler and Evans 2001, 2003) have objected that the

Brugman/Lakoff analysis is methodologically unconstrained and vastly inflates the

number of senses by downplaying the role of context and on-line inference. Kreitzer

argues that many of the parameters used by Brugman/Lakoff (contact, extended,

vertical, etc.) actually belong to the component level of schematic structure, which is

below “the basic level of granularity at which individual prepositions are defined”

(1997:304).Atthisbasicrelationallevelherecognizesonlythreeschemasfromwhich

theentirerangeofusagetypesofovercanthenbederived(Kreitzer1997:308ff.):

31

over1 staticrelationwithoutocclusion (e.g.‘Thepictureisoverthefireplace.’)

over2 dynamicrelation (e.g.‘Thecatjumpedoverthepost.’)

over3 staticobjectwithocclusion (e.g.‘Thetableclothisoverthetable.’)

Inasimilarspirit(i.e.ofprovidingaconstrainingmethodology),TylerandEvans(2001:

105)proposetwocriteriafordifferentiatingthesensesofover:

[...]forasensetocountasdistinct,itmustinvolveameaningthatisnotpurelyspatialinnatureand/orin

which thespatial configurationbetween theTRandLM ischangedvis-a-vis theothersensesassociatedwithaparticularpreposition.Secondly,theremustbeinstancesofthesensethatarecontext-independent,

instancesinwhichthedistinctsensecouldnotbeinferredfromanothersenseandthecontextinwhichit

occurs.17

In summary, then, the discussion of over shows that there is considerable dissent

when it comes to representing polysemy in a lexical network. Depending on which

linguistyouask,thenumberofsensesproposedvariesbetweenthree(Kreitzer1997),

sixteen(TylerandEvans2001)andtwenty-four(Lakoff1990b).Sowhoisright?

7.3.PolysemyasaFuzzyNotion–TheLangacker/TuggyModel

Muchofthecontroversysurroundingpolysemyandlexicalnetworksingeneralhinges

onourunderstandingofwhatconstitutesameaningvariant.However,whenstrivingfor

cognitiverealismthereseemstobeonlyoneplausiblewayofcharacterizingthenotion

of sense, i.e. as a semantic structure that has achievedunitstatuswithin the cognitive

system.Langacker(1987:57)characterizesaunitasfollows:

Aunitisastructurethataspeakerhasmasteredquitethoroughly,totheextentthathecanemployitin

largelyautomaticfashion,withouthavingtofocushisattentionspecificallyonitsindividualpartsortheir

arrangement.Despiteitsinternalcomplexity,aunitconstitutesforaspeakera“pre-packaged”assembly;

becausehehasnoneedtoreflectonhowtoput it together,hecanmanipulate itwitheaseasaunitary

entity.Itiseffectivelysimple,sinceitdoesnotdemandtheconstructiveeffortrequiredforthecreation

ofnovelstructures.Psychologistswouldspeakofa“habit”,orsaythat“automization”hasoccured.

In otherwords, whether a structure has unit status or not depends on its degree of

entrenchment. As Tuggy puts it, entrenchment is best thought of as the “enduring

salience” of a structure as a result of that structure’s repeated usage (1993: 279).

Crucially, there can be no “nonarbitrary cutoff point” for unit status, since cognitive

entrenchmentisclearlyamatterofdegree(Langacker1987:59).Thus,ifweacceptthat

17Recallthattheissueofcontextualinferencehasalsobeenraisedbyproponentsofthemonosemicview.However,itisimportanttonotethatbothKreitzer(1997)andTylerandEvans(2001)rejecttheclaim

thatallusagesofoverareinferredon-linefromasinglesuper-schema.

32

senses are units,wemust conclude that there is no nonarbitrary line of demarcation

separatingsensesfromlesswell-entrenchedsemanticstructures.

BasedontheinsightsofLangacker(1987,1991),Tuggy(1993)proposesamodelthat

embracestheindeterminacyofpolysemybyplacingthenotiononacontinuumbetween

thepolesofambiguity(homonymy)andvagueness.Aprototypicalexampleofambiguity

wouldbethecaseofbank1(riverbank)andbank2(financialinstitution).Botharevery

wellentrenchedandthereisnosalientschemasubsumingthem.Thestructuresarenot

evenetymologicallyrelated,andifoneweretolookforacommonschema,itwouldhave

tobesomethinghighlyabstractlikething.Suchaschemawouldalsoshowahighdegree

ofelaborativedistancefrombothinstances,i.e.itwouldhavetoignorealmostalloftheir

characteristic specifications. Conversely, aunt1 (mother’s sister) and aunt2 (father’s

sister) is givenas anexampleofprototypical vagueness.Here, the subsuming schema

(parent’ssister)ismuchbetterentrenchedthanits instances.Additionally,elaborative

distance is minimal, i.e. schema and instances are identical, except that the schema

ignores theparent’s gender specifications.18In thismodel polysemy constitutes an in-

between case. Consider, for instance, paint1 (artistic painting) and paint2 (utilitarian

painting):Bothstructuresarewell-entrenched,butsoisthesubsumingschema(apply

paint to surface) which is located at intermediate elaborative distance (Tuggy 1993:

283).Thisisexactlythekindofcasewherewewouldspeakofpaint1andpaint2(aswell

asthesubsumingschema)asdifferentsensesofpaint.

To summarize,we speakof ambiguity if twoormore semantic structureshaveunit

statuswhilethesubsumingschemadoesnot;ofvaguenessiftheylackunitstatuswhile

thesubsumingschemahasunit status;andofpolysemy ifboth thestructuresand the

subsuming schema have unit status. Yet, as Tuggy (1993: 282) observes, due to the

dynamicandgradualnatureofsalienceitis“impossibletodrawabsoluteboundarylines

betweenthecategoriesofambiguity,polysemyandvagueness.”

SohowdothevariousanalysesofoverfareinlightoftheLangacker/Tuggymodeland

how canwe account for the differences in results? According to the Brugman/Lakoff

analysis, eachminimal distinction inTR/LM specifications qualifies as a sense. But as

Tyler andEvanspoint out (2001: 99), the linguistic expressions corresponding to the

18AsTuggy(1993:283)notes,thereisaninverse(albeitnotcompletelyparallel)correlationbetweenaschema’selaborativedistanceanditsentrenchment.Thatis,schemasarelesslikelytobecomeentrenched

vis-a-vistheirinstancesoncetheysurpassacertainthresholdofabstractness,i.e.whatRosch(e.g.1977)

callsthebasiclevel(seealso7.1.).Thisis,ofcourse,theresultofusage-basedreinforcement:Highlyabstractstructurestendtobelessfrequentlyused/activatedthantheirelaborations.

33

TR/LM often underspecify these minimal distinctions. Consider the following two

examplesfromLakoff(1990b:421):

(39) Thebirdflewovertheyard. (extended,non-verticalLM;nocontact)

(40) Theplaneflewoverthehill. (extended,verticalLM;nocontact)

IntheBrugman/Lakoffmodel(39)and(40)areminimalvariants.Everythingelsebeing

equal, the former features a horizontally extended LM, whereas the latter features a

horizontallyandverticallyextendedLM.Nowcomparethisto(41):

(41) Thebird/planeflewoverthearea. (extended,?verticalLM;nocontact)

Unlike yard and hill, the noun area underspecifies whether the LM is also vertically

extendedornot. Considering theminimaldegreeof elaborativedistance from (41) to

(39)and(40),andontheassumptionthattheschemain(41)ismorefirmlyentrenched,

the Langacker/Tuggymodelwould predict thatover in (39) and (40) is vague rather

than polysemous. Accordingly, one might argue that the Brugman/Lakoff account

ascribespolysemy toseveral cases thataremoreaptlycharacterizedasvague.On the

other hand, estimates of a structure’s degree of entrenchment basedon introspection

arebesttakenwithagrainofsaltandexperimentaldataindeedsuggeststhat“subjects

seemtomakedistinctionsofaratherfine-grainednature”(SandraandRice1995:122-

123) when confronted with semantic decision tasks. While a given structure is not

automatically guaranteed unit status in virtue of its TR/LM specifications, it is still

plausiblethatlanguageusersmakedistinctionsatthislowlevelofgranularity.

Assumingtheentrenchment-basedvagueness-polysemyclineoftheLangacker/Tuggy

model,Kreitzer(1997)aswellasTylerandEvans(2001)postulatecriteriaforpolysemy

thatultimatelyseembothrigidandarbitrary.RecallfromaboveKreitzer’s(1997:304)

claimthatBrugman/LakoffstyleTR/LMspecificationsarebelowthelevelofgranularity

atwhichprepositionsaredefined,andthusirrelevant.LikeTylerandEvans,heobserves

thatlinguisticexpressionssuchasThemanwentoverthefenceareoftenunderspecified:

“Here,itisunimportantwhetherthemanjumpedoverthefenceorclimbedthefence–

over remains grammatical either way as long as (1) there is motion and (2) the

trajectorytraversestheboudariesofthelandmark”(Kreitzer1997:304).Inotherwords,

Kreitzer suggests that a specification is only relevant if there are contexts where an

expression’s grammaticality hinges on its presence or absence. But there is no

psychologicalevidencetosuggestthatasemanticstructurecannotattainunitstatusin

34

thecognitivesystemofa languageusermerelybecause itsspecificationsdonotcause

ungrammaticality in certain linguistic contexts.19Likewise, Tyler and Evans’s (2001:

105)criteriumthatthere“mustbe[...]instancesinwhichthedistinctsensecouldnotbe

inferredfromanothersenseandthecontextinwhichitoccurs”begsthequestion.Even

inthemostextremescenariowhereausagetypecanalwaysbe inferredfromanother

usagetypeorfromcontextualfactors,wecannotconcludethatthestructureinquestion

isinsufficientlyentrenched.Ifastructureismentallyaccessedtimeandtimeagainitwill

attain unit status, i.e. the language user will produce it without “constructive effort”

(Langacker 1987: 57) – regardless of whether such effort could theoretically be

employedtoinferitornot.

7.4.ImplicationsforthePresentStudy

The above considerations have several implications for the following case studies.

While I agree that lexical networks are themost suitable tool available for accurately

representingsemanticstructure,itisalsoimportanttobeawareoftheirlimitations.As

the Langacker/Tuggy model reminds us, the representation of polysemy is an

approximationatbest.Itisnotpossibletodrawahardandfastboundarylinebetween

vagueness and polysemy. Although there are certain indicators for clear cases of

polysemy,noneofthemcanserveasnecessaryorsufficientconditions.Forexample,if

usage type B is a metaphorical extension of usage type A, chances are that A and B

representdistinctsenses.Yet,wecannotconcludethatmetaphoricalextensionalways

entails polysemy. As Tuggy (1993: 285) points out, during the process of meaning

extensionagivensemanticstructure“canstraddlethethefence[i.e.betweenthepoles

of vagueness and ambiguity] indefinitely, shifting its weight back and forth, before

graduallymovingmoretoonesidethantheother.”Thisappliestometaphor,metonymy,

andanyothermechanismofsemanticextensionalike.Ifwewantsomethingresembling

a“perfectinventory”ofsensesforsomeconstructionatsomespecificpointintime,our

bestbetwouldbetocollectlargeamountsofdataonlanguageuserintuitionsregarding

entrenchment–whatevertheexactnatureofsuchanexperimentmightbe.20Itshould

also be noted, that these experiments would somehow have to take into account

19Infact,thefine-graineddistinctionsattestedbySandraandRice(1995)provideevidencetothecontrary.20SemanticdecisiontaskssuchastheonespresentedinSandraandRice(1995)mightbeagoodstartingpoint.

35

individual as well as cultural variation among test subjects. For instance, Langacker

(1987:376)remarksthat“[e]lmsandmaplesmaynotsurviveasprototypicaltreesfora

speakerwhohaslivedforfortyyearsinthedesert.”

Meanwhile,theaimofmycasestudiesisnottopresentaperfectinventoryofsenses

for theverbsunderscrutiny. Instead,mygoal is togive thereaderan ideaofhowthe

different usage types are connected via family resemblances and tomake explicit the

mechanismsofmeaningextensionbywhichtheyarederivablefromoneanother.Inthe

end it should hopefully become clear that, in the case of these particular verbs, the

respective image schemasare thegluewhich “holds the family together”, so to speak.

Thereadershouldkeepinmindthatthesensespostulatedinthisthesisaretheresultof

introspection checked against the intuitions of native speakers. Consequently, the

proposedcategorystructuresarebestviewedasapproximationsandtobetakenwitha

grainofsalt.Whileforsimplicity’ssakethetermsenseisusedforeachproposedusage

type, the reader should be well aware that several cases are probably closer to

vaguenessthanpolysemy.

8.SpatialExpressionsinCognitiveLinguistics:AnOverviewofthe

Literature

Giventheimagisticstanceonlanguageoutlinedearlier,itcomesasnosurprisethatthe

analysis of spatial terms has been a staple of cognitive linguistic research since the

earliestdays.Thisisevidencedbyavarietyofstudiesfeaturingnetwork-typeanalyses

of prepositional polysemy and verb particle constructions. Of these I have already

mentioned the pioneering contributions of Brugman (1981) and Lindner (1981). In a

comprehensive case study Brugman has successfully shown that the various uses of

English over are not a random aggregate, but instead constitute a systematically

interrelatedcategoryofsenses.Thiswork,arevisedversionofwhichappearedinLakoff

(1990b), illustrates how semantic structure is dependent on andmotivated by image

schematic structure. In the same spirit, but from the perspective of Langackarian

Cognitive Grammar (then called “Space Grammar”), Lindner’s analysis of the English

verb particle constructions V-out and V-up makes a strong case for the substantial

semanticcontributionoftherespectiveparticles.Alternativeanalysesofoverwithinthe

cognitive framework include Dewell (1994), Kreitzer (1997) and Tyler and Evans

(2001). Morgan (1997) proposes a metaphor-based account of verb particle

36

constructions with out. Notable book-length studies of English prepositions and/or

particles include Hawkins (1984), Herskovits (1986), Lindstromberg (1998), Hampe

(2002), and Tyler and Evans (2003). Outside of English, spatial terms have been

exploredfromacognitiveperspectivebySmith(1987)(Germantwo-wayprepositions),

Cuyckens (1991) (Dutch prepositions), Vandeloise (1991) (French prepositions), and

Delbeque(1996)(Spanishporandpara),tonamebutafewexamples.

All of these studies agree in one central respect. Namely, that the image schematic

structure of the preposition/particle is inherently meaningful. Insofar I am greatly

indebted to these works for providing the major working hypothesis of the present

thesis. Although Japanese has no verb particle constructions, the grammatical V2s

considered in this thesis fulfill a similar function by contributing an abstract spatial

meaning–whichmaythenserveasbasisforvariousmechanismsofsemanticextension.

Sincethereiscurrentlynoconvenientwaytorefertothisspecificgroupofverbs(akin

tospatialpreposition),Ihavecoinedthetermimageschemaverbforpracticalpurposes.

Asbrieflymentionedearlier, theverb-verbcompound(fukugôdôshi) is likelyamong

the most widely studied phenomena in Japanese linguistics.21 Nontheless, we can

roughly divide the vast amount of research on the topic into theory neutral and

generativeapproaches.Theformercamp–eitherrootedintraditional“schoolgrammar”

(gakkôbunpô)orassumingnospecifictheoreticalframework–isgenerallyopentothe

possibilitythatgrammaticalV2ssuchas-kakaruor-derumaycarryconceptualcontent

related to their simplex counterparts. However, the details of this relation remain

implicitandnosophisticatedexplanationisusuallyoffered.Mostoftheearlierresearch

on V-V compounds falls into this category (e.g. Teramura 1969;Nagashima 1976), as

well as the insightful series of studies by Himeno (e.g. 1976, 1977, 1979, 1980).22

Yamamoto’s (1984) famous essay on “case government” (kaku shihai) marks a shift

towards syntactocentric accounts of V-V compounds, predominantly occupied with

questions of “argument structure”. Then, ever since Kageyama’s (1993) highly

influential introduction of the lexical vs. syntactic distinction – postulating two

fundamentally different kinds of V-V compounds assumed to emerge in separate

“components”of thegrammar– , researchonV-Vcompoundshasbeendominatedby

21AnextensivebibliographyisavailablefromthehompageoftheNationalInstituteforJapaneseLanguageandLinguistics(NINJAL):http://pj.ninjal.ac.jp/lexicon/files/bunken_v3.pdf;retrieved22Oct.

2016.22ButnotethatHimenoadoptsthegenerativedistinctionbetweenlexicalandsyntacticcompoundsinherlaterwork(e.g.1999).

37

generative approaches (e.g. Kageyama 1996, 2009; Yumoto 1996, 2008; Matsumoto

1998a;Fukushima2005).

Sowheredowecurrently stand?AlthoughCognitiveLinguisticshasbecomeawell-

established research paradigm among Japanese scholars (e.g. Yamanashi 2000, 2009;

Taniguchi 2003, 2005; Momiyama 2014), and despite their many insightful

contributions to the field, Japanese image schema verbs have rarely been a focus of

interest. 23 This is somewhat surprising, considering the amount of attention

prepositionsandverbparticleconstructionshavereceivedfromacognitiveperspective

since the early 1980s. Meanwhile, the dominant generative, and thus formalist,

approaches to Japanese V-V compounding make little to no attempt to draw a

connectionbetweenthemeaningofgrammaticalV2sandtheirsimplexcounterparts–

at leastnotinthesenseofwhatwehavecalledencyclopedicsemantics. Inthiscontext,

thepresent study is intendedas amodest first step towardsestablishing the studyof

JapaneseimageschemaverbsasaresearchtopicinCognitiveLinguistics.

23However,seeMatsuda’s(2001a,2001b)schema-basedstudiesonV-komu.

38

9.1.TheSensesofKAKARU

9.1.1.Sense(Ia):PhysicalSupport(fig.2)

(1) Kabe-ni e-ga kakat-teiru.

Wall-DAT picture-NOM KAKARU-RES ‘Apictureishangingonthewall.’

(2) Monchû-ni hyôsatsu-ga kakat-teiru.

Gatepost-DAT nameplate-NOM KAKARU-RES ‘Thereisanameplateon(fixedto)thegatepost.’

(3) Yôfuku-ga hanga-ni kakat-teiru.

Clothes-NOM hanger-DAT KAKARU-RES ‘Theclothesareonthehanger.’

(4) Kabe-ni hashigo-ga kakat-teiru.

Wall-DAT ladder-NOM KAKARU-RES ‘Aladderisleaningagainstthewall.’

Both TR and LM are concrete objects. The TR exerts force on the LM, which the LM

resists.ThisSUPPORTconfiguration is typicallyvertical, rarelyhorizontal as in (4). In

either case the force is gravitational. Since the TR is an inanimate entity incapable of

self-propelledmovement,construction(B)isusuallynotrealized.

FIGURE2

9.1.2.Sense(Ib):ImaginedSupport(fig.3)

(5) Tsuki-ga sora-ni kakat-teiru.

Moon-NOM sky-DAT KAKARU-RES ‘Themoonishanginginthesky.’

(6) Sora-ni kumo-ga kakat-teiru.

Sky-DAT clouds-NOM KAKARU-RES ‘Cloudsarehanginginthesky.’

(7) Yama-no chôjô-ni moya-ga kakat-teiru.

Mountain-LK summit-DAT mist-NOM KAKARU-RES ‘Mistishangingoverthemountaintop.’

Thissenseisavailablefrom(Ia)viaimagemetaphor(Lakoff2006:215ff.).Duetosimilar

TR-LM arrangements, the force dynamics of scenes like (1) and (2) are left intact

althoughnoactualSUPPORTis involved in(5)-(7).Asshownbelow,(1)and(5)share

thesamebasicspatialconfiguration.

40

FIGURE3

9.1.3.Sense(Ic):PhysicalForce(fig.4)

(8) Paipu-ni atsuryoku-ga kakat-teiru.

Pipe-DAT pressure-NOM KAKARU-RES ‘Thereispressureonthepipe.’

(9) Migiashi-ni taijû-wo kakeru

Rightfoot-DAT bodyweight-ACC KAKERU ‘Toshiftonesweightontotherightfoot’

Thissenseisavailableviametonymicshift from(Ia):Thefocusisnotonthesourceof

physical force,buton the force itself.This forceneednotbeverticallyorientated,but

canincludevariouskindsofinternalorexternalpressure,asexemplifiedby(8).

FIGURE4

9.1.4.Sense(Id):PsychologicalBurden

(10) Seijika-ni fuhai-no utagai-ga kakat-teiru.

Politician-DAT corruption-LK doubt-NOM KAKARU-RES

‘Thepoliticianissuspectedofcorruption.’

(11) Wakashachô-ni kitai-ga kakat-teiru.

YoungCEO-DAT expectations KAKARU-RES ‘ExpectationsrestontheyoungCEO.’

(12) Kimi-ni meiwaku-wo kake-te, môshiwake na-ku omot-teiru.

You-DAT trouble-ACC KAKERU-TE excuse exist.NEG-INF think-RES

‘Iamsorryfortroublingyou.’

(13) Jukensei-ni puresshâ-ga kakat-teiru.

Testcandidates-DAT pressure-NOM KAKARU-RES ‘Thereispressureonthetestcandidates.’

41

Viametaphor this sense can be obtained from (Ia) or (Ic), depending onwhether the

psychologicalburden isconstruedasanobject (10-12)ora force(13).Therespective

metaphorsarePSYCHOLOGICALBURDENSAREPHYSICALBURDENSandPSYCHOLOGICALFORCESARE

PHYSICALFORCES. In either case, the sensation ofweight ismapped onto the domain of

psychologicalstates.

9.1.5.Sense(Ie):PreconditionforSuccess(fig.5)

(14) Boku-no shôrai-ga kyô-no kaigi-ni kakat-teiru.

I.M-LK future-NOM today-LK meeting-DAT KAKARU-RES

‘Myfuturedependsontoday’smeeting.’

(15) Rôjin-ga musuko-ni kakat-teiru.

Oldman-NOM son-DAT KAKARU-RES ‘Theoldmandependsonhisson.’

(16) Kare-ga keiba-ni zenzaisan-wo kake-ta.

He-NOM horseracing-DAT wholefortune-ACC KAKERU-PAST

‘Hebethisallofhisfortuneonhorseraces.’

This sense involves the application of force dynamics to the domain of abstract

reasoning. (Ie) is available from (Ia) via twometaphors: PRECONDITIONSARESUPPORTING

OBJECTSandthemoregeneralSTATESOFAFFAIRSAREPHYSICALOBJECTS(ofwhichtheformer

is an instance). The underlying cognitive principle that allows us to conceive of non-

things as things has been variously discussed –most prominently under the label of

reification (e.g. Talmy 2003a: 43f.). Langacker (1991: 35) refers to reification in the

contextofnominalizationandrelativeclauses,noting that the latterallowus “to ‘step

back’fromthesituation[...]andconstrueitasanabstractobjectorpropositioncapable

ofbeingmanipulated,evaluated,andcommentedon.”Abstractobjects, in turn, canbe

construed metaphorically as concrete objects, making them compatible with force

dynamic notions of the physical domain. We therefore speak of claims supported by

evidenceorcertainassumptionsrestingonotherassumptionsetc.In(14)-(16)onestate

ofaffairsissupportedbyanother.In(14)theTRboku-noshôraimetonymicallystands

for a proposition like “Iwill not be fired (ormight even get promoted)”whilekyôno

kaigi metonymically stands for something along the lines of “the outcome of today’s

meetingwillbepositive”.3Letuscall the formerqand the latterp.The implication in

(14)-(16)is:Ifpturnsoutfalse,qwillbefalse.Ifpturnsouttrue,qwillbetrue.Inother

words,q(reifiedasaTHING)issupportedbyp(alsoreifiedasaTHING).Thus,(Ie)isan

3Putinanotherway,theovertnominalsin(14)-(16)havepropositionalactivezones(Langacker1991:456).

42

example of construing the logical principle of entailment in terms of the embodied

schemaSUPPORT.4Thediagrambelowillustratesthis.

FIGURE5

9.1.6.Sense(If):OnotologicalDependence

(17) Kuruma-ni hoken-ga kakat-teiru.

Car-DAT insurance-NOM KAKARU-RES ‘Thereisinsuranceonthecar.’

(18) Subete-no shina-ni yu’nyûzei-ga kakat-teiru.

All-LK goods-DAT importtax-NOM KAKARU-RES

‘Allgoodsaresubjecttoimporttax.’

(19) Dokusaisha-ga hangyakusha-no kubi-ni shôkin-wo kake-ta.

Dictator-NOM rebel-LK head-DAT bounty-ACC KAKERU-PAST

‘Thedictatorputabountyontherebels’heads.’

Thisisanothercaseofabstractsupportwhichismetaphoricallyavailablefrom(Ia).The

TR is supportedby the LM insofar as theTR’s existencepresupposes – and therefore

dependson–theLM’sexistence.TherelationbetweenTRandLMissomewhatakintoa

mereologicalone.Althoughonewouldbehesistanttodescribecarinsuranceaspartofa

car,itiscertainlyimpossibletograsptheconceptofcarinsurancewithouttheconcept

of a car. Neither can one conceive of import tax without the concept of goods. Prize

moneyisonlymeaningfulagainstthebackgroundofsomechallengeorcompetition,and

soforth.InLangackarianterms,TRandLMin(17)-(19)areconnectedbyaprofile-base

relationship,wherethebaseisacognitivedomain(ordomainmatrix)andtheprofilea

salient substructure within that domain (Langacker 2006: 34f). This explains why in

(17)-(19)theTRisontologicallyandconceptuallydependentontheLM.

9.1.7.Sense(II):ElicitedEffect(fig.6)

(20) Kare-ga pureiyâ-ni rekôdo-wo kake-ta.

He-NOM player-DAT record-ACC KAKERU-PAST ‘Heputarecordontheplayer.’

4ForfurtherdiscussionofhowlogicalnotionsareunderstoodintermsofimageschemasseeJohnson(1990:63f.)andSweetser(1991:58ff.)onmodality.

43

(21) Jazu-no kyoku-ga kakat-teiru.

Jazz-LK song-NOM KAKARU-RES ‘Ajazzsongisplaying.’

(22) Doa-ni kagi-ga kakat-teiru.

Door-DAT key-NOM KAKARU-RES ‘Thedoorislocked.’

(23) PDF-ni rokku-ga kakat-teiru.

PDF-DAT lock-NOM KAKARU-RES ‘ThePDFfileislocked.’

(24) Kono ken-ni mahô-ga kakat-teiru.

This sword-DAT magic-NOM KAKARU-RES ‘Aspellrestsonthissword.’

(25) Yatto kuruma-no enjin-ga kakat-ta.

Finally car-LK engine-NOM KAKARU-PAST ‘Thecar’senginefinallycaught(on).’

(26) Ocha-ni akami-ga kakat-teiru.

Tea-DAT redness-NOM KAKARU-RES ‘Theteahasareddishhue.’

This usage type exploits the experiential correlation betweenCONTACT and resulting

effect. It seems plausible to suppose that (II) gradually emancipated itself from

SUPPORTsensessuchas(Ia)and(Id).Notethat(20)canbereadasaninstanceof(Ia),

sincetherecordplayerphysicallysupportstherecord.However, ifwebackgroundthe

physicalsupportarrangementand focusourattentionon theeffectelicitedbyputting

the record in contactwith the player,we arrive at something like (21). There are no

moretracesoftheSUPPORTschemain(21),sincethesentenceisfelicitousevenifthe

music comes fromadevice such as anMP3player. Similar observations canbemade

about (22) and (23). Read as an instance of (Ia), the former describes the physical

SUPPORTarrangementbetweenkeyholeandkey.Again,focussingonthelockingeffect

instead,wearriveatsceneslike(23),devoidofanySUPPORTconfiguration.Considering

thesentencesunder(Id),wecanseehow(24)isrelatedtoabstractSUPPORT.InEnglish

aswell,aspellcanbeplacedonaswordandthenrestontheswordasaresult.Infact,

thelinebetweenabstractSUPPORTandelicitedeffectisnotclear-cutatall,sincecases

like(24)(andsomeinstancesof[Id])exhibitboth.Ofcourse,suchinbetween-casesare

to be expected in the gradual and dynamic process of meaning extension. Once the

elicited effect sense of KAKARU is established, it can be extended to non-SUPPORT

scenes such as (25) and (26) – since the the relevant experiential correlation obtains

between CONTACT and effect, rather than SUPPORT and effect. That is, CONTACT

betweenfunctionalpartselicitsaneffect,whetherSUPPORTisinvolvedornot

Beforemovingon,itisworthnotingthat(25)(enjingakakaru)isneverrealizedwith

ani-argument.SowhereistheLMin(25)?Andwhataboutthearrangementin(21)?

44

Langacker (1991:232)distinguishesbetween the “internal structureofapredicate”

and “its combinatorial properties”. The former pertains to the more general level of

conceptual organization, the latter to the linguistic level of clause structure. It is

thereforepossibleforapredicatetohaveaninternalTR/LM(atthelevelofconceptual

structure)thatisnotrealizedasanovertnominal.In(21)and(25)thematterisfurther

complicatedbymetonymyandidiomaticity.Asstatedabove,usagetype(II)focusseson

theeffectelicitedwhenonefunctionalpart(theTR)ofasystemcomes intoCONTACT

withanotherfunctionalpart(theLM).Thatis,theverbkakaruprofilestherelationship

betweenkeyandkeyholeorrecordandrecordplayeretc.Nowconsider(27)-(29):

(27) Rekôdo-ga pureiyâ-ni kakat-teiru.

Record-NOM player-DAT KAKARU-RES ‘Therecordis(playing?)ontherecordplayer.’

(28) Rekôdo-ga kakat-teiru.

Record-NOM KAKARU-RES ‘Therecordisplaying.’

(29) (Ribingu-ni) Shûberuto-ga kakat-teiru.

(Livingroom-DAT) Schubert-NOM KAKARU-RES ‘Schubertisplaying(inthelivingroom).’

(27)makesovert linguistic reference to the functionalparts recordandplayer,which

correspond to the schematic TR and LM of kakaru. This sentence is still ambiguous

between usage types (Ia) and (II), although it will most likely be interpreted

metonymically(CONTACT-->effectofCONTACT)infavorofthelatter.(28)invitesthis

metonymicalinterpretationevenstrongerbylinguisticallyomittingtherecordplayeras

afunctionalpart.Finally,asstatedabove,(29)isfelicitiousevenifthemusiccomesfrom

an MP3 player. On this interpretation, kakaru has already assumed the idiomatic

meaningbelow(seefig.6):

(30) X-ga kakat-teiru

[Musicalentity] isplaying

FIGURE6

45

In (27) we have two easily identifiable functional parts which, when brought in

CONTACT, cause music to play. An MP3 player with no physical medium however,

causesmusic to play in an entirely differentway – sowe have no TR and LM in the

CONTACT sense of (27) anymore. Instead, kakaru in (29) is interpreted idiomatically

according to (30) and the most salient participant (i.e. the music represented

metonymicallybyitscomposer) ispromotedtoclausalTR.WhileaclausalLMsuchas

ribingucanbeintruduced,suchaLMdoesobviouslynotcorrespondtotheinternalLM

ofkakaruasvia(27)(i.e.,thelivingroomisnotafunctionalpartofaCONTACT-systema

larecord–recordplayer,key–keyhole,etc.).

Now let us consider (25), which is quite similar. Our layman’s knowledge or “folk-

model” of howmachinery works involves the CONTACT of functional parts: Entity A

comes into CONTACT with entity B and something happens. This is no less true for

starting up a car.However, exactlywhich parts of amechanical systemneed to be in

CONTACTwithone another is usually expert knowledge andbeyond the graspof our

folk-model. And even if we can identify those parts, their salience is usually

overshadowedbyotherentities.Comparethefollowingsentences:

(31) Enjin-ga kakat-ta.

Engine-NOM KAKARU-PAST ‘Theenginecaught(on).’

(32) Witharoartheenginecaught(on).

Herebothkakaru andcatchonhaveanon-transparentargumentstructure.What is it

the engine caught(on)?What functional parts are involved in both cases?Andwould

theyberelevantfromalayman’sperspectivewhenstartingupacar?Theupshotisthat

kakaruin(31)hasaschematicTRandLMcorrespondingtofunctionalparts,whichwe

mayormaynotbeabletoidentify(dependingonourtechnicalexpertise).Butbecause

of their low salience as far as our everyday experiencewith cars is concerned, these

partsremainconceptuallyschematicandnebulous.Asaconsequence,kakarugainsan

idiomaticmeaninganalogousto(30)andthemostsaliententity(car,engine)takesthe

placeofclausalTR/subject.This,ofcourse, isametonymicprocess,as it constitutesa

domaininternalshift(i.e.WHOLEFORPART).

9.1.8.Sense(III):Covering(fig.7)

(33) Hanako-ga sarada-ni doresshingu-wo kake-ta.

Hanako-NOM salad-DAT dressing-ACC KAKERU-PAST ‘Hanakoputdressingonthesalad.’

46

(34) Beddo-ni beroa-no kabâ-ga kakat-teiru.

Bed-DAT velour-LK cover-NOM KAKARU-RES ‘Thereisavelourcoveroverthebed.’

(35) Kawazura-ni kiri-ga kakat-teiru.

Riversurface-DAT mist-NOM KAKARU-RES ‘Theriveriscoverdbymist.’

The COVERING sense is obtained by backgrounding the force dynamics of SUPPORT

fromcertainusagesof(Ia)or(Ib),suchas(1),(5),or(6).Therearetworequirements

for(III).ThefirstisconcernedwiththeshapeoftheTR,whichmustbeplanarandcover

atleastanextendedportionoftheLM.Secondly,thesalienceoftheCOVERINGschema

mustbehighervis-a-visthesalienceoftheSUPPORTschema.

Both applies to (33)-(35). The TR is a planar entity which makes a considerable

portion of the LM visually inaccessible. This object can be either discrete (individual

specks of dressing covering a salad) or continuous (a blanket covering a bed ormist

covering a river). In all of the above sentences the supporiting function of the LM is

stronglybackgrounded(33,34)ornon-existent(35).

FIGURE7

It shouldbenoted thatboth theTR’s shapeaswell as the salienceofCOVERINGvs

SUPPORTareamatterofdegree.Consequently,categorialfringecaseslike(36)aretobe

expected.

(36) Mado-ni kâten-ga kakat-teiru.

Window-DAT curtain-NOM KAKARU-RES ‘Thereisacurtainoverthewindow.’

Therearetwopossiblewaystoconstruethisscene.Wecaneitherfocusonthecurtain

covering thewindow,yieldingaCOVERINGreading.Or, alternatively,we focuson the

curtain track supporting the curtain, yielding a physical SUPPORT reading (see Ia). In

thelattercasemadowouldbemetonymicallyrepresentingthecurtaintrack.Depending

onthecontext,eitherschema’ssaliencecanbehightenedvstheother:

(37) Mado-ni kâten-ga kakat-tei-te, heya-ga kurai.

Window-DAT curtain-NOM KAKARU-RES-TE room-NOM dark

‘Thereisacurtainoverthewindow,sotheroomisdark.’

47

(38) Mado-ni kâten-wo kake-yôtoshi-ta ga, uma-ku kakara-nakat-ta.

Window-DAT curtain-ACC KAKERU-INT-PAST but good-INF KAKARU-NEG-PAST

‘Itriedtoputacurtainoverthewindowbutcouldn’tattachitproperly.’

FocussingonthebrightnessoftheroomwillraisethesalienceoftheCOVERINGschema.

Incontrast, ifwedirectourefforts towards fixing thecurtain to thecurtain track, the

SUPPORTschemawillgainasalienceboost.

9.1.9Sense(IVa):PhysicalRestraint(fig.8)

(39) Kitsune-ga wana-ni kakat-teiru.

Fox-NOM trap-DAT KAKARU-RES ‘Thefoxiscaughtinthetrap.’

(40) Sakana-ga hari-ni kakat-teiru.

Fish-NOM hook-DAT KAKARU-RES ‘Thefishisonthehook.’

(41) Sêtâ-ga kugi-ni kakat-ta.

Sweater-NOM nail-DAT KAKARU-PAST ‘Thesweatergotcaughtonanail.’

(42) Oki-ni fune-ga kakat-teiru.

Offshore-DAT ship-NOM KAKARU-RES ‘Ashipisanchoredofftheshore.’

Thisusagetypeislinkedto(Ia)viaaperspectivalshift,sincethenotionsSUPPORTand

RESTRAINT refer to different construals of the same force dynamic arrangement.

AdoptingTalmy’s(2003a:409ff.)frameworkandterminology,thisarrangementcanbe

characterized as follows: An agonist with a tendency towards action is blocked by a

strongerantagonist(seefig.8).However,SUPPORTandRESTRAINTdifferastowhether

the presence of the antagonist’s counterforce is deemed favorable from the agonist’s

perspective.TheEnglishverbkeepillustratesthisforthephysical–andbymetaphorical

extension – for several abstract domains. Byway of example, consider the sentences

below:

(43) Thefencekeepsthewolvesfromattackingthesheep. (RESTRAINT)

(44) Thetailkeepsthekangaroofromtopplingover. (SUPPORT)

(45) Coffeekeepsmefromfallingasleeponthejob. (SUPPORT)

(46) Toomuchcoffeekeptmefromgettingagoodnight’ssleep. (RESTRAINT)

Withthisinmind,considertheSUPPORTandRESTRAINTreadingsofKAKARU:

(47) E-ga kabe-ni kakat-teiru.

Picture-NOM wall-DAT KAKARU-RES ‘Thepictureishangingonthewall.’

(48) Kitsune-ga wana-ni kakat-teiru.

Fox-NOM trap-DAT KAKARU-RES ‘Thefoxiscaughtinthetrap.’

48

In(47)thepictureissubjecttotheforceofgravityandthereforehasatendencytofall

to the ground.However, some other object (a nail etc.) is blocking this tendency.We

consider this BLOCKAGE a case of SUPPORT, since from the (admittedly

anthropocentric) “perspective of the picture” falling to the ground and shattering to

piecesisanunfavorableoutcome.In(48)thefox’self-propelledmotioniscounteracted

onbythetrap.Sincethiscounterforceisobviouslyunfavorablefromtheperspectiveof

the fox, it is interpreted as an instance of RESTRAINT. Now recall (42), which is

particularlyinteresting,becauseitcanbeinterpretedeitherway.Ifweconceiveofships

primarilyasvehiclescontrolledandoperatedbypeople,(42)yieldsaSUPPORTreading.

But ifwechoosetoviewthemasentitieswhicharecapableofdriftingaway“ontheir

own” (backgrounding the current etc.) – i.e. construe the ship as capable of “self-

propelled”motion–wearriveataRESTRAINTreading.

FIGURE8

9.1.10.Sense(IVb):AbstractRestraint

(49) Kare-wa teki-no keiryaku-ni kakat-ta.

He-TOP enemy-LK scheme-DAT KAKARU-PAST

‘Hefellvictimtotheenemy’sscheme.’

(50) shiken-no koto-ga ki-ni kakat-te, nemur-e-nai.

Test-LK thing-NOM mind-DAT KAKARU-TE sleep-POT-NEG

‘(I’m)worriedaboutthetestandcan’tsleep.’

(51) O-me-ni kakaru no-wo tanoshimi-ni shi-tei-masu.

HON-eye-DAT KAKARU NMLZ-ACC pleasure-DAT do-PROG-POL

‘I’mlookingforwardtomeetingyou.’

As stated under (IVa), force dynamic notions such as RESTRAINT are often

metaphoricallyextendedtoanumberofabstractdomains.In(49),forinstance,physical

RESTRAINT ismappedonto thesocial,psychological,orother limitations, imposedon

the agonist. Notable in particular are the idiomatic uses that construe sense data or

49

propositional content as “free-floating” entities,which are “caught on” the respective

perceptiveormentalfacultiesusedforprocessingthem.

9.1.11.Sense(Va):ExternalControl(fig.9)

(52) Tarô-ga wazawai-ni kakat-ta.

Tarô-NOM misfortune-DAT KAKARU-PAST ‘Tarôsufferedamisfortune.’

(53) Hanako-ga byôki-ni kakat-teiru.

Hanako-NOM illness-DAT KAKARU-RES ‘Hanakoissick.’

(54) Kanja-ga isha-ni kakat-teiru.

Patient-NOM doctor-DAT KAKARU-RES ‘Thepatientisconsultingadoctor.’

This sense is linked to (IVb) via a subtle metonymic shift, since we experience

RESTRAINT usually in tandem with external control. The latter notion, however, is

broaderandnotlimitedtoRESTRAINT.I.e.,anillness,amisfortuneoraschemewillnot

justlimitourscopeofactionbutcanaffectandmanipulateourbehaviorinvariousways.

(Since the difference is gradual, [49]would be somewhere inbetween [IVb] and [Va],

depending on what exactly keiryaku denotes.) Note that (50) constitutes a curious

unprototypical case, because theTR intentionally surrenders itself to external control.

Thiswillbediscussedinthenextsection.

FIGURE9:AweakerTRinthesphereofinfluenceofastrongerLM

9.1.12.Sense(Vb):AgentiveControl(fig.10)

(55) Sate, shigoto-ni kakar-ô.

Now work-DAT KAKARU-VOL ‘Now,let’sgettowork.’

(56) Hanako-ga kodomo-no sewa-ni kakat-teiru.

Hanako-NOM children-LK care-DAT KAKARU-RES ‘Hanakoistakingcareofthechildren.’

(57) Shôsetsuka-ga shinsaku-ni kakat-teiru.

Novelist-NOM newwork-DAT KAKARU-RES ‘Thenovelistisworkingonanewbook.’

When comparing (53), (54), and (55), onewill noticewhatmight be called a clineof

agentivity.Allofthesescenesshareacommonimageschematicstructure:TheTRmoves

50

along a PATH, spatial or virtual, which terminates at the LM (recall the schematic

constructions [B] and [C] from before). They differ, however, in regards to the TR’s

intentionality and the perceived relative strength of TR and LM. Specifically, the TR’s

“degreeofagentivity“graduallyincreasesfrom(53)to(55):

TR’smovementtowardsLM relativestrength degreeofagentivity

(52-53) unintentional TR<LM low

(54) intentional TR<LM intermediate

(55-57) intentional TR>LM high

In other words, the cline of agentivity amounts to the following (reverse) hierarchy:

unintentional movement towards stronger LM --> intentional movement towards

stronger LM --> intentionalmovement towardsweaker LM. In each case, theweaker

participantisconstruedasbeinginthesphereofinfluenceofthestrongerparticipant.

FIGURE10:AweakerLMinthesphereofinfluenceofastrongerTR

9.1.13.Sense(VIa):PhysicalArrival(fig.11)

(58) Mori-wo deru to tôge-ni kakaru.

Forest-ACC getout COND mountainpass-DAT KAKARU

‘Onceoutoftheforest,we’llarriveatthemountainpass.’

The RESTRAINT and control senses (IV and V) have a strong tendency to feature an

animate (or quasi-animate) TR capable of self-propelled motion, thereby raising the

relativesalienceofPATHtraversal inherent inconstructions(A),(B)andpresupposed

byconstruction(C).5(VIa),aswell,foregroundsPATHtraversalbutabstractsawayfrom

any forcedynamicnotions. Inotherwords, theTR’sPATHterminatesas itcomes into

CONTACTwiththeLM,butneitherexertsanyforceontheother.

5I.e.(C)istheresultofPATHtraversal.

51

FIGURE11

9.1.14.Sense(VIb):Transmission

(59) Hanako-ga Tarô-ni koe-wo kake-ta.

Hanako-NOM Tarô-DAT voice-ACC KAKERU-PAST 'HanakosaidhellotoTarô.’

(60) (Tarô-kara) Hanako-no ie-ni denwa-ga kakat-ta.

(Tarô-ABL) Hanako-LK house-DAT phonecall-NOM KAKARU-PAST

‘Hanako’shousegotaphonecall(fromTarô).’

(61) (Taisa-kara) shôsa-ni meirei-ga kakat-ta.

(Captain-ABL) major-DAT command-NOM KAKARU-PAST

‘Themajorreceivedacommand(fromthecaptain).’

Thissenseisaslightvariationof(VIa).Here,theTRisaninformationalentitytravelling

alongaPATHfromsender(SOURCE)torecepient(GOAL).Astheaboveexamplesshow,

theSOURCEcanbelinguisticallyprofiledeitherassubjectofacausedmotionevent(59)

or,iftheTR’smotionisconstruedasself-propelled,asanoblique(60,61).Asin(VIa),

therelationshipbetweenTRandLMisforcedynamicallyneutral.

9.1.15.Sense(VIc):TemporalArrival

(62) Shingata terebi-no kaihatsu-ga oikomi-ni kakat-ta.

Newmodel TV-LK development-NOM finalstage-DAT KAKARU-PAST

‘DevelopmentofthenewTVmodelhasreachedthefinalstage.’

(63) Koko-wa môsugu uki-ni kakaru.

Here-TOP soon rainyseason-DAT KAKARU

‘We’reheadedfortherainyseason.’

Thisisastraightforwardmetaphoricalextensionof(VIa)intothetemporaldomain.The

TIME IS SPACE metaphor is extremely pervasive cross-linguistically (Radden 2006). For

example,inEnglishonecan“approach”adeadlineorbeworriedaboutan“approaching”

deadline.NotethatthissensemayfeatureanextendedTR(suchaskaihatsuin[62])

9.1.16.Sense(VII):ResourceRequirement(fig.12)

(64) Kono shôsetsu-wo kaku no-ni gonenkan kakat-ta.

This novel-ACC write NMLZ-DAT fiveyears KAKARU-PAST

52

‘Writingthisnoveltookfiveyears.’

(65) Ie-wo kau-Ø-ni-wa takusan okane-ga kakaru.

House-ACC buy-NMLZ-DAT-TOP alot money-NOM KAKARU

‘Tobuyahouseoneneedsalotofmoney.’

(66) Kanojo-wa tema-wo kake-te, sono e-wo kai-ta.

She-TOP effort-ACC KAKERU-TE that picture-ACC draw-PAST

‘Sheput(alotof)effortintothedrawing.’

Theresourcerequirementsenseisavariantofthearrivalsense,featuringanextended

TR which incrementally “grows” from a SOURCE (0%) along a PATH, until it makes

CONTACTwithitsGOAL/LM(100%).Thisimageschematicstructureappliestoallkinds

of resources, suchas time,money, effort,orability. Ineachcase, there is someGOAL

whichrepresentstheend-pointofaSCALE.Inordertoreach(i.e.makeCONTACTwith)

thisGOAL, theTRneeds tobeof sufficient length.Theconceptualizationofamount in

termsofphysicallengthisbasedontheexperientialcorrelationbetweenbothdomains

andconstitutesaprimarymetaphor.6Likewise,thereisaveryrealexperientialbasisfor

understandingtheachievementofagoalintermsestablishingphysicalcontact.

As is to be expected, then, expressions that construe resource requirement via the

SOURCE-PATH-GOALschemaarenot limitedto Japanese.Bywayofexample,consider

theGermanverb(aus)reichen(reach)anditsdialectalvariantlangen,derivedfromthe

adjectivelang(long).

(67) MeinGeldreichtnicht(aus),umeinBootzukaufen.

Idon’thaveenoughmoneytobuyaboat.

(68) DieVorrätelangennochbisMonatsende. Theprovisionswilllastuntiltheendofthemonth.

FIGURE12

9.1.17Sense(VIIIa):PhysicalLink(fig.13)

(69) Murabito-tachi-ga kawa-ni hashi-wo kake-ta.

Villager-PL-NOM river-DAT brigde-ACC KAKERU-PAST

‘Thevillagersbuiltabridgeacrosstheriver.’

6ConsiderthesimilarcaseofMOREISUP(LakoffandJohnson2003:16).

53

(70) Toguchi-ni kumo-no su-ga kakat-teiru.

Doorway-DAT spider-LK web-NOM KAKARU-RES

‘Thereisaspiderwebinthedoorway.’

(71) sora-ni niji-ga kakat-teiru.

Sky-DAT rainbow-NOM KAKARU-RES

‘Arainbowspansacrossthesky.’

TerminativePATHsandLINKsarealternativeconstrualsofoneanother.Ifaroadruns

betweenXandY, it connectsYwithX.WeconstrueLINKswhen followinganobject’s

trajectoryfromSOURCEtoend-point.Imaginesomeoneswimmingacrossariver.Then

mentally connect all the “dots“ on the swimmer’s PATH. This will yield a LINK from

shoretoshore.WecanalsogotheoppositerouteandconstrueLINKsasPATHs:

(72) ThatmountainrangegoesfromCanadatoMexico. (Talmy2003a:104)

Thisoperationinvolveswhathasvariouslybeencalledvirtualmotion(Talmy1983)or

abstract motion (Langacker 1987: 168ff.), i.e. mentally scanning an extended entity

along a trajectory in successive fashion.7In a way, then, (terminative) PATHs are

“dynamically“construedLINKswhereasLINKsare “statically“construed(terminative)

PATHs.

KeepingthePATH-LINKtransformationinmind,wecannowseehow(VIIIa), too, is

relatedtothearrivalsense(VI).In(69)-(71)theLINKistheresultoftheTR’sextension

fromSOURCEtoGOAL.I.e.,thebridge,thespiderweb,andtherainbowareconstruedas

graduallymovingfromonesidetotheother.However,liketheTRsin(VII)theyarenot

punctualbutextendedandthereforemoveby“growing”.AstheymakeCONTACTwith

theirGOAL,theseTRshaveevolvedintostaticstructuresextendingfromtheirpointof

departuretotheirpointoftermination.

FIGURE13

7InmorerecentworkTalmy(2003a:138f)usesthetermcoextensionpath.

54

(VIIIa) also features what we may call a salience based argument shift. (Recall our

discussionofenginegakakaruunder[II].)Thiskindofmetonymicshiftoccurswhenthe

entity most directly involved in the profiled relation is outranked in prominence by

some other participant of the scene and thus fails to appear as an overt nominal.8In

(69)-(71)thisentitycorrespondstothepointatwhichtheTR’sPATHterminates,i.e.the

“other side“ of the river (69), the “other side“ of the door frame (70), and the point

where the rainbow terminates (71). However, since in each case the TR ends up

occupying the whole PATH instead of just the GOAL and its vicinity, attention is

redistributedaccordinglyandthePATHreplacestheGOALasclausalLM/DAT-Obj.

9.1.18Sense(VIIIb):RelevanceLink

(73) Hanbaikeiyaku-ni kakaru shôhin

Salescontract-DAT KAKARU goods ‘Goodssubjecttothesalescontract’

(74) Shôgai-no aru kodomo-ni kakaru kyôiku sôdan

Disability-NOM exist children-DAT KAKARU education advice

‘Educationaladviceconcerningchildrenwithdisabilities’

The metaphorical construal of relevance in terms of the LINK schema is cross-

linguistically quite common. This is not suprising, since our most basic embodied

experienceofrelevance involvesphysicalLINKsbetweenobjects: “twopiecesofwood

arenailedtogether,thechildholdstheparent’shand,thesnapsonthechild’scoatare

connected,thelampispluggedintothewallsocket.Inthesesimplephysicalcasesthere

isaspatialcontiguityandclosenessofthelinkedobjects,andtheconnectedobjectsare

related via the link” (Johnson 1990: 118). Likewise, TR and LM in (73) and (74) are

metaphoricallyconstruedasphysicallyconnectedentitiesinvirtueofbeingrelevantto

oneanother.Asin(VIIIa),wemayviewtheTRasanextendedobjectmakingCONTACT

withtheLM.

9.1.19.Summary

The above analysis shows that the semantic structure of KAKARU consists of at least

eightclusters,allofwhicharelinkedtotheCONTACTschemabyvirtueofexperiential

correlation.TheseclustersareSUPPORT(I),effect(II),COVERING(III),RESTRAINT(IV),

CONTROL(V),arrival(VI),resourcerequirement(VII),andLINK(VIII).Inthefollowing

8Again,thesecasescanbeanalyzedasactivezonephenomena.Compare,forexample,Iblinkedvs*Myeyelidblinked(Langacker1987:272).

55

sectionsofthischapterIwillarguethatthesensesofV-kakaru/V-kakeru,too,arelinked

totheCONTACTschemaandconsequentlyshouldbeviewedaspartofthesamelexical

network.

9.2.TheSensesofV-KAKARU

Himeno (1979, 1999), who offers the most comprehensive treatment of V-KAKARU,

recognisestwogeneralmeanings:“dotowards”(shikô)and“start(andbeinterrupted)”

(shidô)(1979:61).Shefurthersubdividestheseintothefollowingsenses.

(i) DOTOWARDS(shikô)

(i1) contactbyfalling(rakkasesshoku)-kakaru Ha-ga atama-ni chiri-kakaru.

Leaf-NOM head-DAT fall-KAKARU ‘Theleavefallsontothehead.’

-kakeru ---

(i2) supportivecontact(ikyosesshoku)

-kakaru Hito-ga kabe-ni yori-kakaru.

Person-NOM wall-DAT movetowards-KAKARU ‘Someoneleansagainstthewall

-kakeru Kabe-ni ita-wo tate-kakeru

Wall-DAT board-ACC put-KAKERU ‘Toputaboardagainstthewall’

(i3) orientedcontact(shikôsesshoku)

-kakaru Inu-ga hito-ni osoi-kakaru.

Dog-NOM person-DAT attack-KAKARU ‘Thedogpouncesattheperson’

-kakeru Hito-ni tsuba-wo haki-kakeru

Person-DAT spit-ACC spit-KAKERU ‘Tospitatasomeone’

(i4) psychologicalorientation(shinritekishikô)

-kakaru ---

-kakeru Hito-ni warai-kakeru

Person-DAT smile-KAKERU ‘Tosmileatsomeone’

(i5) orientedmovement(shikôidô)

-kakaru ---

-kakeru Kaijô-ni tsume-kakeru

Assemblyhall-DAT cram-KAKERU ‘Tocrowd(into)anassemblyhall’

(i6) grasping(hasoku)

-kakaru ---

-kakeru Inu-wo oi-kakeru

Dog-ACC chase-KAKERU ‘Tochaseafteradog’

(i7) encounterinpassing(tsûkasôgû)

-kakaru Hito-ga mise-no mae-wo tôri-kakaru.

Person-NOM store-LK front-ACC pass-KAKARU

‘Apersonpassesbyinfrontofastore.’

-kakeru ---

(ii) START(shidô)

(ii1) beginning(shidôtai)

-kakaru ---

56

-kakeru Hon-wo yomi-kakeru

Book-ACC read-KAKERU ‘Tobeginreadingabook’

(ii2) emergence(shôgentai)

-kakaru Jiko-ni at-te, shini-kakaru

Accident-DAT meet-TE die-KAKARU

‘Tobeonthevergeofdyingafteranaccident.’

-kakeru Jiko-niatte,shini-kakeru

(sameasabove)

(basedonHimeno1979:59)

Is there reason to suggest conceptual links between the grammatical V2s -kakaru/-

kakeru and theirsimplexcounterparts? Ibelieve there is,and thatcrosslinguisticdata

canprovideevidenceforthis.Aswehaveseen, theCONTACTschemaiscentral tothe

semanticsofKAKARU,while-kakaru/-kakerucanbeparaphrasedas“doVtowards”and

“begin to V/be about to V”. Curiously, this resembles the relationship between the

Germanprepositionan–whichprototypicallyinvolvesCONTACTbetweenTRandLM–

and the derived verb particle construction an-V, which among its various meanings

includestwosensessimilarto(i)and(ii).Letussuppose,forthesakeofargument,the

relationship between KAKARU and V-KAKARU is conceptually arbitrary. Then why

wouldweencounterahighlysimilarrelationshipinadrasticallydifferent language?It

would be quite a coincidence, to say the least. Consequently, it seemsworthwhile to

explore the possibility of conceptual links between the CONTACT schema and the

notions of directedness and inchoative aspect. In the following, I will (1) discuss the

prepositionaninrespecttotheCONTACTschema,(2)compareV-KAKARUandan-Vin

respecttoCONTACTanddirectedness,and(3)compareV-KAKARUandan-Vinrespect

toCONTACTandinchoativeaspect.

9.2.1.TheGermanPrepositionanandtheCONTACTSchema

AsSmith(1987:94)notesinhisstudyofGerman2-wayprepositions(i.e.prepositions

whichoccurwithbothdativeandaccusativecase),an“prototypicallyprofilesarelation

betweenaTRandaverticalLMinwhichtheTRmakescontactwiththesurfaceofthe

LMitself.”Heillustratesthiswiththefollowingpairofsentences:

(75) DasBildhängtander(DAT)Wand.

Thepicturehangsonthewall.

(76) HanshängtdasBildandie(ACC)Wand.

Hanshangsthepictureon(to)thewall.

(Smith1987:95)

57

TherehasbeensomedebateintheliteratureregardingtheimportanceofCONTACTfor

themeaningofan.Whilesome(e.g.Brinkmann1962;Saile1984)arguethatanentails

someformofCONTACT,othersrejectthisviewmaintainingthatCONTACTisimpliedby

the verb rather than the prepositon itself (Li 1994: 76) or that “ ‘contact’ and ‘non-

contactreadings’[...]areduetosizerelationships,edgeproperties,andsimilarfeatures

ofbothrelatumandtheme”(Nüse1999:16).Firstofall,itistruethatneitherCONTACT

nor a vertical LM are necessary features of an. This is illustrated by the examples

below:

(77) AnnawartetanderHaltestelle.

Annawaitsatthebusstop.

(78) DasHausstehtam(andem)See.

Thehousestandsbythesea.

Mystanceontheissue,however,isnotthatannecessarilyentailsCONTACTbutrather–

following Smith (1987) – that an is prototypically associated with CONTACT. This

positioniscompatiblewiththepossibilitythattheissueofCONTACTvsnon-CONTACT

dependsonfactorssuchasthechoiceofverband/ortheconfigurationalpropertiesof

TR and LM. For an encyclopedic view of meaning this is of little relevance. If an

frequently profiles CONTACT relations, then CONTACT will become conventionally

associatedwithan.Furthermore,CONTACTispresentinthosecaseswhereanprofiles

thekindsofrelationsthataremostdeeplyentrenchedinoureverydayexperience.I.e.,

an-relationsthatinvolvephysicalCONTACTbetweenmundaneentitiesfromthedomain

ofdailylifearearguablythemostbasicintermsofcognitiveentrenchment:

(79) DerZettelam(andem)Kühlschrank

Thenoteonthefridge

(80) DerRingam(andem)Finger

Theringonthefinger

(81) DerKnopfam(andem)Mantel

Thebuttononthecoat

9.2.2.CONTACTandDirectedness:TheDOTOWARDSSenseofan-VandV-KAKARU

OneofthesensesoftheGermanparticleverbconstructionan-Vcanbecharacterizedas

“directedactivity”.AccordingtoFleischerandBarz(2012:402),“[a]n-indicatesthatthe

actiondenotedbythesimplexverbisdirectedtowardsapersonorthing.”They(2012:

402) further observe that these simplex verbs characteristically belong to semantic

58

fieldssuchasseeingandspeaking,motion,measuringandtargeting,aswellastouching,

attaching,andresistance.Someexamplesare:

(82) HanssiehtHelgaan.(ansehen)

HanslooksatHelga.

(83) HelgalächeltHansan.(anlächeln)

HelgasmilesatHans.

(84) DerHundspringtdenJungenan.(anspringen)

Thedogjumpsattheboy.

(85) DieYachtsegeltdenHafenan.(ansegeln)

Theyachtsailstowardstheharbor.

(86) DerJägervisiertdenHirschan.(anvisieren)

Thehuntertakesaimatthestag.

(87) KlarakämpftgegendieLangeweilean.(ankämpfen)

Klarafightsagainstboredom.

Soisthereaconceptuallinkbetween(79)-(81)and(82)-(87)–betweenCONTACTand

directedness?Forillustrativepurposes,Isuggestthatweconsiderthenotionofasearch

domain. The concept was originally introduced by Hawkins (1981) and subsequently

adopted into Langacker’s framework of Cognitive Grammar. In Langackarian terms,

“[t]he search domain (sd) of a locative predication (LOC) is defined as the region to

which it confines the trajector”(Langacker 1987: 286). Based on the notion of search

domain,Smith(1987:91)postulates the following“configurationally-baseddefinitions

forDATandACCinthe2-wayprepositionalrealm.”

DAT:theTRoftheprepositionisconfinedthroughouttheprocesstoasetofpointssatisfyingthelocative

specificationofthepreposition(i.e.theSDofthepreposition).Inthisrespect,thesituationisdescribedas

unchanging.

ACC:theTRoftheprepositonisNOTalwaysconfinedtotheSDofthepreposition,butenterstheSDat

somepointalongapath.Thesituationinvolveschangeofsometypewithrespecttothelocative

configuration.

(Smith1987:92)

Bywayofillustrationconsiderthesentenceswithaufandinbelow:(88a) DieKinderspringenaufdieParkbank.(ACC)

Thechildrenjumpon(to)thebench.

(88b) DieKindersindaufderParkbank.(DAT)

Thechildrenareonthebench.

(88c) DieKinderspringenaufderParkbank.(DAT)

Thechildrenarejumping(upanddown)onthebench.

(89a) DieKinderrennenindenLaden.(ACC)

59

Thechildrenrunintothestore.

(89b) DieKindersindimLaden.(DAT)

Thechildrenareinthestore.

(89c) DieKinderrennenim(indem)Laden.(DAT)

Thechildrenrun(around)insidethestore.

Thedativeversionsconfine theTR–whetherstationary(b)or inmotion(c)– to the

searchdomainoftheLM,whereastheaccusativeversionsinvolvemotionoftheTRinto

the LM’s search domain (Smith 1987: 93). So while the accusative versions “involve

goal-orientedmovement”,theDATversionsdonot(Smith1992:391).Ofcourse,thisis

not thewhole story as far as German 2-way prepositions are concerned. Smith notes

thattherearemanyinstanceswhereaccusativecasedoesnotentailspatialmotion,but

rathersomekindofabstractchangeonpartoftheTR:

(90) DieTablettelöstsichindasWasserauf.

Thetabletdissolvesintothewater.

(91) ErhateinZitatindenTexteingefügt.

Heputaquoteintothetext.

(fromSmith1995:312,314)

He(e.g.Smith1995:319f.)thereforearguesinfavorofamoreschematicchangevsno

change distinction which would subsume the motion vs location distinction, while

accounting for these more abstract cases as well.9For our purposes though, it is

important to emphasize, firstly, that the goal oriented spatial motion sense is the

categorialprototypeforaccusativecaseinGermanand,secondly,that instanceswhich

involve change still exploit themotion concept via the EVENTSTRUCTUREMETAPHOR, e.g.

STATES ARE LOCATIONS, CHANGES ARE MOVEMENTS, etc. (Lakoff 2006: 204). In fact, Smith

(1992: 387) holds that “[German] ACC is motivated whenever a grammatical

construction[...]evokesanaspectofthepath-goalschema.”

HavingdiscussedGerman2-wayprepositionsintermsofconfinementtovsmovement

intoaLM’ssearchdomain,wecannowreturntothespecificcaseofan.AsIhaveargued

above,thesearchdomainofanprototypicallyencompassestheexternalboundariesof

the prepositional object (see also Leys 1989: 101). Again, consider theDAT/ACCpair

frombefore:

(75) DasBildhängtander(DAT)Wand.

9Inasimilarvein,Leys(1989)arguesforanaccountintermsofanincipient(ACC)vsanexisting(DAT)relationship.

60

Thepicturehangsonthewall.

(76) HanshängtdasBildandie(ACC)Wand.

Hanshangsthepictureon(to)thewall.

In(75)theTR(dasBild)isatalltimesconfinedtothesearchdomainoftheLM.In(76)

theTRstartsoutoutsidetheLM’ssearchdomain,thenmovesalongaPATHtowardsit,

andfinallycomestorestwithinthesearchdomain.Where(76)construestheLMasa

GOAL, (75)construes it asaPLACE.Furthermore, since (75) iseasily interpretatedas

the result of (76), the two configurations are closely interconnected by experiential

correlation.Inthisway,asearchdomainanalysisofanwithdativevsanwithaccusative

revealsaconceptuallinkbetweenCONTACTanddirectedness.

Ifweextendourscopefromconcretespatial to fictivemotion (Talmy2003a:103ff.),

wecanaccountfordirectionalan-Vinanalogousfashion.Inotherwords,(82)-(87)are

allinstanceswhereanentitymovesintothesearchdomainspecifiedbythepreposition

an–eitherviaself-propelledorviacausedmotion.(82)and(83)canbeunderstoodas

involvingcausedmotionofreifiedsensedata(e.g.agazeorasmile)alongaPATHthat

terminatesuponCONTACTwiththeLM.10(84)and(85)straightforwardlyinvolveself-

propelledspatialmotion into theLM’ssearchdomain. (86) traces the “targetingpath”

(Talmy2003a:109f.)ofthebulletuntilCONTACTwiththestagisestablished.And(87)

involves the TR’s psychological motion towards an abstract antagonist. I.e.,

prepositional an with ACC and directional an-V share the same image schematic

topology.

Atthispointitshouldcomeasnosurprisethatthesameanalysisservestoaccountfor

therelationbetweenKAKARUanddirectional -kakaru/-kakeru.Consider the following

pairsofsentencesinlightoftheabovediscussion.

(92a) Kôto-ga kabe-ni kakat-teiru.

Coat-NOM wall-DAT KAKARU-RES ‘Thecoatishangingonthewall.’

(92b) Tarô-ga kabe-ni kôto-wo kake-ta.

Tarô-NOM wall-DAT coat-ACC KAKERU-PAST ‘Tarôhungthecoaton(to)thewall.‘

(93a) Sakana-ga hari-ni kakat-teiru.

Fish-NOM hook-DAT KAKARU-RES ‘Thefishiscaughtonthehook.’

(93b) Sakana-ga hari-ni kakat-ta.

Fish-NOM hook-DAT KAKARU-PAST ‘Thefishcaughtthehook.’

10SeealsoFelfe(2012:155),whopointstotheconceptualsimilarityofErlächeltsiean(Hesmilesather)andSeinLächelngelangtansie(Hissmilereachesher).

61

SinceKAKARUprototypically involvesCONTACTbetweenTR andLM,we can assume

that KAKARU and an specify roughly the same search domain, i.e. the external

boundariesoftheLM.Now,comparingthe(a)and(b)versions,itisplainlyevidentthat

in terms of TR-LM arrangement kakat-te iru (see schema C) corresponds to an with

dative, while kakeru and kakaru (see schemas A and B) correspond to an with

accusative:

KAKARU an imageschematictopology

kakat-teiru anw/DAT confinementofTRtoSDofLM

kakaru anw/ACC(intransitive) (self-propelled)motionofTRintoSDofLM

kakeru anw/ACC(transitive) (caused)motionofTRintoSDofLM

Consequentlytherelationbetweensimplexkakaru/kakeruanddirectionalV-kakaru/V-

kakerucanbeaccountedforinanalogytotherelationbetweenanwithaccusativeand

directionalan-V:

anw/ACC à directionalan-V}motionofTRintoSDofLMkakaru/kakeru à directionalV-kakaru/V-kakeru

Withthis,wecanstraightforwardlyaccountformostofthesenseslistedbyHimeno

(1979):

(i1) Ha-ga atama-ni chiri-kakat-ta.

Leaf-NOM head-DAT fall-KAKARU-PAST

‘Theleaflandedonthehead.’

(i2) Tarô-ga kabe-ni yori-kakat-ta.

Tarô-NOM wall-DAT movetowards-KAKARU-PAST ‘Tarôleanedaginstthewall.’

Hanako-ga kabe-ni ita-wo tate-kake-ta.

Hanako-NOM wall-DAT board-ACC put-KAKERU-PAST

‘Hanakoputtheboardagainstthewall.’

(i3) Inu-ga Tarô-ni osoi-kakat-ta.

Dog-NOM Tarô-DAT attack-KAKARU-PAST ‘ThedogpouncedatTarô.’

Tarô-ga Jirô-ni tsuba-wo haki-kake-ta.

Tarô-NOM Jirô-DAT spit-ACC spit-KAKERU-PAST ‘TarôspatatJirô.’

(i4) Hanako-ga Jirô-ni warai-kake-ta.

Hanako-NOM Jirô-DAT smile-KAKERU-PAST ‘HanakosmiledatJirô.’

(i6) Inu-ga neko-wo oi-kake-ta.

Dog-NOM cat-ACC chase-KAKERU-PAST ‘Thedogchasedafterthecat.’

62

ItshouldbeobvioushowalloftheaboveexamplesinvolvetheTR’smovementintothe

searchdomainspecifiedbyKAKARU–especially in lightofourpreviousdiscussionof

(82)-(87). Analogous to the case of directional an-V, this limits the choice of V1 to

semanticfieldswhichallowforspatialorfictivemotion.11

However,wemightaskwhysomeverbs thatnormally takeadirectobject (marked

withwo)mark the correspondingnominalwithni insteadwhenappearing in tandem

with-kakaru.AsHimenonotes,verbswithassault-likemeaningsareparticularlyprone

tothisshift:

(94a) Tarô-ga Jirô-wo osot-ta.

Tarô-NOM Jirô-ACC attack-PAST

‘TarôattackedJirô.’

(94b) Tarô-ga Jirô-ni osoi-kakat-ta.

Tarô-NOM Jirô-DAT attack-KAKARU-PAST

‘TarôdartedatJirô.’

(95a) Tarô-ga Jirô-wo nagut-ta.

Tarô-NOM Jirô-ACC hit-PAST

‘TarôhitJirô.’

(95b) Tarô-ga Jirô-ni naguri-kakat-ta.

Tarô-NOM Jirô-ACC hit-KAKARU-PAST

‘TarôtookaswingatJirô.’

She(1973:43)suggeststhatsuchV1sare“influenced”bythedirectionalmeaningof

-kakaru, so that the compound as a whole marks the argument in question with ni

insteadofwo.Buildingonthis,wecanfurtherspecifytheissueintermsofalternative

construal.Whilenimarks an argument as indirect object andGOAL12,womarks it as

directobject–agrammaticalroleprototypicallyassociatedwiththenotionenergysink

(Langacker1991:292).Theformerisacharacterizationintermsofwhatwemightcall

thematicrole (suchasrecepient,experiencer, etc.), the latteraprimarily forcedynamic

characterization in termsof the action chainmodel.13In otherwords, the choice ofni

overwobytheabovecompoundsraisesthesalienceofthePATH-GOALschemafavored

by -kakaru vis-a-vis the force dynamic construal favored by the V1s (osou, naguru).

11Anexampleforthelatterwouldbe(i4),whichcorrespondsto(83).12Atleastintheallativesenserelevanthere.ForanetworkanalysisofniseeKabataandRice(1997).13SeeLangacker’s(1991:304ff.)discussionofbasicgrammaticalrelations.Whilehesuggeststhatsubjectsareprototypicalaction-chainheadsandobjectsprototypicalaction-chaintails,Langackerseems

torejectaprimarilyforcedynamiccharacterizationforindirect/obliqueobjects.Instead,hearguesthat

indirectobjectsarebestunderstoodintermsoftheexperiencerrole.Relatednotionssuchasrecipientor

goalmaybeviewedassemanticextensionsofthisthematicrole.

63

There is a fairly simple reason for this. Many of the compounds with assault-like

meaningsdonotstrictlyentailthattheLMisactuallysubjectedtotheactiondenotedby

theV1.Considerthesesentencepairs:

(96a) Tarô-ga Jirô-ni naguri-kakat-ta ga, Jirô-ga kawashi-ta. Tarô-NOM Jirô-DAT hit-KAKARU-PAST CONJ Jirô-NOM dodge-PAST

‘TarôtookaswingatJirô,butJirôdodged(theblow).’

(96b) *Tarô-ga Jirô-wo nagut-ta ga, Jirô-ga kawashi-ta.

Tarô-NOM Jirô-ACC hit-PAST CONJ Jirô-NOM dodge-PAST

*TarôhitJirô,butJirôdodged(theblow).’

(97a) Samurai-ga teki-ni kiri-kakat-ta ga, teki-ga kawashi-ta.

Samurai-NOM enemy-DAT slash-KAKARU-PAST CONJ enemy-NOM dodge-PAST

‘Thesamurailashedhisswordattheenemy,buttheenemydodgedaway.’

(97b) *Samurai-ga teki-wo kit-ta ga, teki-ga kawashi-ta.

Samurai-NOM enemy-ACC slash-PAST CONJ enemy-NOM dodge-PAST

*‘Thesamuraicut(down)theenemywithhissword,buttheenemydodgedaway.’

In otherwords, in the (a) versions the effect of the simplex verb/V1 on the LM (Jirô,

Hanako)canbefelicitouslycancelled.Incomparisontothe(b)versionsthismakesthe

LMa relativelypoor energy sink, but abetter candidate for the role ofexperiencer or

goal.14ThiskindofalternativeconstrualisnotlimitedtoJapanese.Similarconstructions

canbeencounteredinEnglishandGerman:

(98a) Hestabbedatme,butIdodgedtheknife.

(98b) *Hestabbedme,butIdodgedtheknife.

(99a) Erhatnachmir(DAT)getreten,aberichbinausgewichen.

Hekickedatme,butIdodged.

(99b) *Erhatmich(ACC)getreten,aberichbinausgewichen.

Hekickedme,butIdodged.

Still,twosensesfromHimeno’slistmaystrikeusassomewhatproblematic:

(i5) Hito-ga kaijô-ni tsume-kake-ta.

People-NOM assemblyhall-DAT cram-KAKERU-PAST

‘Peoplecrowded(into)theassemblyhall.’

(i7) Hito-ga mise-no mae-wo tôri-kakat-ta.

Person-NOM store-LK front-ACC pass-KAKARU-PAST

‘Someonepassedbyinfrontofthestore.'

14ThefactthatnophysicalCONTACTisentaileddoesnotaffectoursearchdomaininterpretation.Rather,suchcasesarebestunderstoodasinvolvingatargetingpath(Talmy2003a:109f.)alongthelinesof(86).

64

(i5) seems to not quite fit our interpretation, since it emphasises the interior region

ratherthantheexternalboundariesoftheLM.However,theissueiseasilyresolvedonce

we realize that this particular example involves a metonymic shift triggered by the

gestaltpropertiesof theLMrather thanbytsume-kakeru itself.Consider the following

examples:

(100) Hôdôjin-ga joyû-ni tsume-kake-ta.

Press-NOM actress-DAT cram-KAKERU-PAST

‘Thepressbesiegedtheactress.’

(101) Hitogomi-ga ie-no iriguchi-ni tsume-kake-ta.

Crowd-NOM house-LK entrance-DAT cram-KAKERU-PAST

‘Thecrowdbesiegedtheentranceofthehouse.’

(102) Hitogomi-ga ie-ni tsume-kake-ta.

Crowd-NOM house-DAT cram-KAKERU-PAST.

‘Thecrowdbesiegedthehouse.’Or:

‘Thecrowdpouredintothehouse.’

(100)iscompletelyconsistentwithanexternalboundaryinterpretation.Thisistruefor

(101)aswell.Although,since iriguchi is theBOUNDARYofaCONTAINER, the interior

region’s salience is hightened. (102), finally, is vague between a BOUNDARY and an

INTERIORreading.Thatis,tsume-kakeruwillbydefaultspecifytheLM’sBOUNDARYas

itssearchdomain,butaCONTAINER-likeLM–whichconsistsofbothBOUNDARYand

INTERIOR – can metonymically override the BOUNDARY reading in favor of an

INTERIORreading.

(i7) issomewhat idiosyncraticbecause tôri-kakaru takesaPATHargument (marked

bywo)although-kakaru/-kakeru typically favorsaGOALargument. Itseemsplausible

tosuggestthatthespecificusagecontextoftôri-kakaruisresponsibleforthis.According

to Himeno (1979: 44) tôri-kakaru typically appears within the limited syntactic

environmentofcertaintemporalclausesexpressingcoincidence.Someexamplesare:

(103) Heya-no mae-wo tôri-kakat-tara, hito-ga de-te ki-ta.

Room-LK front-ACC pass-KAKARU-when person-NOM moveout-TE come-PAST

‘AsIpassedby(infrontof)theroom,someonecameout.’

(104) Kafeteria-wo tôri-kakaru to, shiriai-ga koe-wo kake-ta.

Cafeteria-ACC pass-KAKARU as acquaintance-NOM voice-ACC KAKERU-PAST

‘AsIpassedby(infrontof)thecafeteria,anacquaintancegreetedme.’

(105) Kôsaten-wo tôri-kakat-ta tokoro, kôtsujiko-ga oki-ta.

Crossing-ACC pass-KAKARU-PAST moment trafficaccident-NOM happen-PAST

‘AsIcrossedtheintersection,atrafficaccidentoccured.’

65

That is, we might characterize the majority of constructions containing tôri-kakaru

informallyas:

(106) AsXpassesbyY[LOC],XencountersZ[EVENT]

I therefore suggest that tôri-kakaru does in some abstract sense indeed have a GOAL

argument, namely the event Z.On this interpretation, onemight paraphrase (103) by

somethinglike:

(107) [Heya-nomae]LOC-

wo

tôt-tara, [hito-gade-tekuruno]EVENT-ni sashi-kakat-ta.

[Room-LKfront]-

ACC

pass-

when

[Person-NOMmoveout-TEcomeNMLZ]-

DAT

comeacross-

PAST

‘AsIpassedby(infrontof)theroomIencounteredtheeventofsomeonecomingout.’

Ofcourse,thisisjustaroughparaphraseandshouldinnowaybeunderstoodassome

sort of underlying structure. It merely serves to illustrate how tôri-kakaru might be

understoodasimplyingaGOALalthoughnoGOALargumentislinguisticallyrealized.15

Butwhataboutourinitialexample?

(i7) Hito-ga mise-no mae-wo tôri-kakat-ta.

Person-NOM store-LK front-ACC pass-kakaru-PAST

‘Someonepassedby(infrontof)thestore.’

Iwouldarguethatevenwithoutthekindofclauselinkagefoundin(103)-(105)sucha

sentencestilltendstoimplysomesortofschematiceventZ.Thisseemsplausible,since

expressions of coincidence from English and German invite the same interpretation.

Comparethefollowingsentencepairs:

(108a) Ipassedbyafurniturestore.

(108b) Ihappendtopassbyafurniturestore.

(109a) IchbinaneinemMöbelgeschäftvorbeigekommen.

(correspondsto110a)

(109b) IchbinzufälliganeinemMöbelgeschäftvorbeigekommen.

(correspondsto110b)

Here the (b) versions, more so than the (a) versions, seem to suggest some kind of

unforseen event or state of affairs. I.e., (108b) and (109b)may be followed upmore

15ThisisconsistentwithChen’s(2013)observationthatimportantframeelements–e.g.“corearguments“ofV1orV2–arenotalwaysovertlyrealizedasargumentsofthecompound.

66

naturallybysomethinglike:“Andguesswhat–theyhadthatlampyou’vebeenlooking

for!”16

9.2.3.CONTACTandInchoativity:TheSTARTSenseofan-VandV-KAKARU

In this section I will make the case for a conceptual relation between CONTACT and

inchoative aspect. While an account of directional V-KAKARU in terms of image

schematictopologyagreeswithourpretheoreticintuitions,thesamecanhardlybesaid

forinchoativeV-KAKARU.Inherstudyonthegrammaticalizationofinchoative

-kakaru/-kakeru,Kikuta(2008:118)remarksthatdirectedness–butnotinchoativity–

canbetracedbacktothesemanticsofthesimplexverb.Dismissingthepossibilityofany

conceptual links between inchoative -kakaru/-kakeru and its simplex counterpart

(2008:157,165),sheproposesastrongversionofaninvitedinferencingaccountbased

ontheusageofthecompoundkure-kakaru:

Accordingtothedataexamined,inthecaseofV-kakar,theinchoativeinterpretationwasmadeavailable

insuchcollocationsaskure-kakar,whichoriginallymeant‘thesunsetsandthedarknesshangslow.’Since

becomingdark takesawhileafter thesunset, thisphrase implies somekindof indeterminacyas to the

exact time denoted. In addition, the phrase apparently became idiomatic, referring to dusk in general,

whichisattheonsetofnight.Theinchoativesenseassociatedwithkure-kakarwasgraduallygeneralized

and schematized until around 12-13C, when V-kakar became a productive pattern of inchoatives

applicable tootherverbs thankure. Thegradual, context-basedprocessof the emergenceof inchoative

usagedirectlyfollowsthepredictionmadebytheusage-basedapproachtogrammaticalization(Traugott

&Dasher2002;Hopper&Traugott2003).[…][T]heinchoativeusageofV-kakestartedanalogicallyafter

its intransitivecounterpartV-kakarhadbeensufficientlygrammaticalizedandrecognizedas inchoativeconstruction.(Kikuta2008:164)

I agree that invited inferencing can play a substantial role in semantic change and

consequently take no issue with the idea that certain items may feature more

prominently thanothers in theearlystagesofgrammaticalization.However, thereare

twopointstoconsiderhere–onegeneralandonemorespecific.First,anaccountlike

the above,which relies onpragmatic reanalysis alone, runs the riskof trivializing the

role of image schematic topology and conceptual metaphor in meaning extension.17

Secondly, Kikuta’s emphasis on kure-kakaru in this specific case begs the question in

lightofthelinguisticevidencesheprovides.Tobackupherclaimthatfrequencyofuse

alongwithcontextualreinterpretationofkure-kakaruledtotheemergenceofinchoative

V–kakaru,shecitesonlyahandfulofoccurrencesfromclassicalliterature(2008:141ff.).

16However,itshouldbenotedthatthisunforseeneventreadingisacancellableconversationalimplicature(Grice1975)ratherthananentailment.17Fortheroleofmetaphoringrammaticalizationseee.g.Heineetal.(1991)andSweetser(1991).

67

Foragrass-rootslevelphenomenonlikepragmaticstrengtheningthisisarathershaky

empiricalbasis–leavingroomforthepossibilitythatotherfactorsmighthaveplayeda

roleafterall. In the following Iwillargue,not fromahistoricalbut fromasynchronic

andcross-linguisticviewpoint,thattheconceptuallinksdismissedbyKikutadoindeed

existandmaywellhaveprovidedthemacro-structureforthemeaningextensionfrom

simplexverbtoinchoativemarker(andinthecaseofGerman:fromspatialpreposition

toinchoativemarker).

IwillbeginbyexaminingtheinchoativesenseofGermanan-V.Considerthefollowing

examples:

(110) IchhabedasBuchangelesen.(anlesen)

Ihavestartedreadingthebook.

(111) HansbrätdasFleischan.(anbraten)

Hanscooksthemeatgently.(i.e.,untilitisrare/medium-rare)

(112) LisahatdasAutoangezahlt.(anzahlen)

Lisahasmadeadownpaymentonthecar.

(113) WenndudieBretteranbohrst,hastduesspäterleichter.(anbohren)

Ifyoupartlydrill/pre-drilltheboards,you’llhaveiteasierlater.

(114) DasBrotschimmeltan.(anschimmeln)

Thebreadstartstogetmoldy.

(115) DasAutofuhrhupendan.(anfahren)

Honking,thecarstartedmoving.

AccordingtoFelfe(2012:156ff.),an-Vintheaboveexamplesexpressestransitive(110-

113)or intransitive(114-115)partiality(Partialität), i.e. thebeginning,weakintensity

or weak after-effect of an event or action. In any case, the process denoted by the

simplexverbhasenteredbutnotsurpassedtheinitialstage,whichiswhyIwillreferto

allsuchinstancesas“inchoativean-V”.18

Felfe (2012: 164) argues that this sense is related to directional an-V in twoways.

Firstly, he notes that establishing physical contact with a concrete object often

18Heretheterm“initialstage”pertainstoanidealscriptofgoingthroughallpossiblestagesofaprocessfromstarttofinish,regardlessofwhetherthesestagesareactuallyrealizedornot.Inthissenseanbratenreferstotheinitialstageofcookingsomethinguntilitiswelldone(German:durchbraten,lit.“cook

through”)–nottheinitialstagesoftheactualprocess(whichmightendprematurely).

68

Itwouldbeinappropriatetospeakofa“surface”inone-dimensionalspace.Weshould

rathersaythat,bydefault,processeshaveuptotwoexternalboundaries(dependingon

telicity) – a point of entry and a point of exit. As the name suggests, we come into

CONTACTwiththepointofentry first,sincewehavenochoicebut toenteraprocess

from the “front”. Now recall that the external boundaries of an object constitute the

searchdomain ofan. Viewed in this light the usage ofan as inchoativemarker is not

surprisingat all:Theprocess (e.g.braten, cook) is reifiedas aone-dimensionalobject

located in front of the TR. Consequently, the starting point of the process – which

correspondstotheobject’sfrontalboundary–comprisesthesearchdomainofan.The

mappingsare:

(non-punctual)process IS one-dimensionalobject

startingpoint à frontalboundary

(duration à middleportion)

(endpoint à rearwardboundary)

etc.

Onthisview,inchoativean-Visametaphoricalextensionofdirectionalan-V.Bothshare

the same image schematic topology, i.e. movement of the TR into the search domain

specified by an. However, in case of inchoative an-V the LM of an is the process

expressedbythesimplexverb.Toillustratethis,consider(111)fromabove:

(111) HansbrätdasFleischan.(anbraten)

Hanscooksthemeatgently.(i.e.,untilitisrare/medium-rare)

TheschematicLMoftheprefixaniselaboratedbytheprocessbraten.Thisisnottobe

confused with the LM of the compound anbraten, which is elaborated by the object

nominaldasFleisch.TheTRofbothanandanbrateniselaboratedbyHans.Insummary,

aninanbratenexpressesthatHansestablishesCONTACTwiththe“frontalboundary”of

theprocessbraten.Keepinmindthatthisschematic“boundary”isnotahardandfast

lineofdemarcation.Assearchdomainofan itmereleyservesas therelevantpointof

referenceforentryintotheprocess,andthusallowsforsuperficialingression.

But is theaboveaccountpsychologicallyplausible? Iwouldarguethat it is,andthat

converging evidence comes in form of all kinds of charts that we naturally use to

visualizeplanningandprogress.Consider,forexample,figure15whichshowsapopular

typeofprojectscheduleknownas“GanttChart”.Theplannedactivitiesinthischartare

visualizedexactlyashypothesizedabove–asone-dimensionalextents.Accordingly,the

70

I will now demonstrate that everything which applies to the relation between

directional and inchoative an-V applies to the relation between directional and

inchoativeV-KAKARUaswell.Considerthecaseofyomi-kakeru:

(120) Tarô-ga hon-wo yomi-kake-ta tokoro-e denwa-ga nat-ta.

Tarô-NOM book-ACC read-KAKERU-PAST moment-ALL phonecall-NOM ring-PAST

‘AsTarôbegantoreadthebook,thephonerang.’

ForreasonsIwillelaborateonlater(see14.4.)theV2-kakeruininchoativeV-kakeruis

assumed to be intransitive. Recall from our previous discussion that KAKARU andan

specify roughly thesamesearchdomain, i.e. theexternalboundariesof theLM.Recall

further,thattheimageschematictopologyofkakaru(andanwithaccusative)hasbeen

characterizedasmovementoftheTRintotheSDoftheLM.IntheaboveexampletheLM

of kaketa is the process expressed by yom(u), its TR is Tarô (again, applying the

metaphoricalviewofdurativeprocessesasone-dimensionalextents).Consequently,the

interpretationofkakeru in (120) is analogous to thatofan inanbraten: both indicate

CONTACTwith the frontal boundary of (and subsequently superficial ingression into)

the process designated by the simplex verb. That is, in the above example kaketa

indicatesthattheTR(Tarô)hascomeintoCONTACTwith(andmovedslightlybeyond)

thefrontaltemporalboundaryofyom(u)whenthetelephonerang.

Butwhatofpunctualprocessesexpressedbyverbslikeshinu?

(121) Tarô-ga jiko-ni at-te, shini-kake-tei-ta.

Tarô-NOM accident-DAT meet-TE die-KAKERU-RES-PAST

‘Tarôgotintoanaccidentandwasonthevergeofdying.’

Here,too,asearchdomainanalysisalongwithametaphoricalinterpretationyieldsthe

correct results. If durative processes are one-dimensional extents then punctual

processes are zero-dimensional objects. I.e., if yomu is conceived of as a line, shinu is

conceived of as a point. Since there can be no superficial ingression into (or partial

overlap with) zero-dimensional objects, the “start V-ing” interpretation is rendered

impossiblebyimageschematictopology.Instead–inkeepingwiththeone-dimensional

tozero-dimensionalLMtransformation–CONTACTwiththefrontalboundaryofshinu

can only correspond to immediate temporal adjecency of TR and process. Therefore,

kake-tei-tain(121)indicatesthatTarôwasconfinedtoastateimmediatelyadjecentto

thatofdeath.

72

Inthis wayametaphorical interpretationcaneasilyaccommodateboththeshidôtai

and the shôgentai readings via image schema transformation. Note, however, that

exampleslikethefollowingchallengetheprototypicalshidôtai=durativeandshôgentai

=punctualpairings(Himeno1979:52f.).

(122) Kare-wa nanika-wo ii-kake-ta ga, futakoto-mikoto-de yame-ta.

He-TOP something-ACC say-KAKERU-PAST CONJ, fewwords-INS quit-PAST

Hewasabouttosaysomething,butstoppedafterafewwords.

(123) Kare-

wa

nanika-wo ii-kake-ta ga, yahari kotoba-wo nonde-shimat-ta.

He-TOP something-

ACC

say-KAKERU-

PAST

CONJ after

all

words-

ACC

swallow-IRR-

PAST

‘Hewasabouttosaysomething,but(instead)keptsilentafterall.’

(124) Hana-ga yuru-ku hiraki-kake-teiru.

Flower-NOM slow-INF open-KAKERU-RES

‘Theflowerhasopenedupslightly.’

Examples (122) and (123) show that durative processes are compatible with both

shidôtaiandshôgentaireadings.AccordingtoHimeno(1979:53)shôgentaifocusseson

thebeginningofaprocesswhileshidôtaifocussesonthebeginningofthecontinuation

ofaninterruptedprocess.Theabovesearchdomainanalysissuggestsadifferentpicture.

Inchaotive-kakaru/-kakeruspecifiesthevicinityoftheV1’s frontalboundaryas itsSD

andisthereforeslightlyvaguewithrespecttotheexactlocationoftheTR:IftheTRis

immediatelyinfrontoftheboundarywegettheshôgentaireading.IftheTRisslightly

past the boundary (superficial ingression) we end up with the shidôtai reading. This

vaguenessisusuallyresolvedbythecontext,asin(122)and(123).Punctualprocesses,

ontheotherhand,limitV-kakerutotheshôgentaireading,sincetheyareincompatible

withsuperficialingression.However,asHimeno(1979:53)notes,processeswhichare

typically punctual can be construed as durative under certain circumstances. For

example, theverbhiraku (open) isusually construedas apunctual event, butmaybe

interpreteddurativelywhenreferringtoslow,gradualprocessesasisthecasein(124).

Inconclusion,then,durativeprocessesarecompatiblewithboththeshidôtaiandthe

shôgentaireading(althoughthelatterpairingislesscommon),whilepunctualprocesses

force a shôgentai interpretation.23The possible pairings are illustrated in figure 16

below:

23Asforthequestionwhyinchoativean-Vhasnointerpretationcorrespondingtotheshôgentaireading:Sinceimageschematictopologydoesnotprecludesuchareading,itsabsenceisbestregardedasa

contingentphenomenon.I.e.,noteverythingthatistheoreticallypossiblebecomesalinguisticreality.

73

FIGURE16:ThehatchedareaindicatesthesearchdomainoftheLM.

74

10.DERUandtheEXITSchema

The intransitive/transitive verb pair deru/dasu – at its most schematic level – is

characterizedbythefollowingconstructions.

(A) X-ga Y-wo Z-kara dasu

XCAUSE YMOVEOUTOF CONTAINER

(B) Y-ga Z-kara/Z-wo deru

YMOVEOUTOF CONTAINER

Thatis,deruprofilesarelationbetweensomething(theTR)andaCONTAINER(theLM),

wherebytheTRmovesfromtheINTERIORtotheEXTERIORoftheLM.Ishallreferto

this image schematic configuration as EXIT. The transitive variant dasu introduces a

thirdparticipantwhichcausestheTR’smovement.

IwilluseDERU(uppercase)asanumbrellatermforboth(A)and(B).

10.1.TheSensesofDERU

10.1.1.Sense(I):SpatialEXIT(fig.1)

Asthefollowingexamplesshow,thesemanticsofDERUcoverawidevarietyofTR-LM

configurationsinthespatialdomain.Toillustratethis,Ihavechoseneightparametersin

respect towhich configurationsmight differ. These parameters are based on Talmy’s

work on configurational structure (e.g. Talmy 2006) and notmeant to be exhaustive.

Rather, the following is intended to give the reader an idea of the vast number of

configurational possibilities, without spelling each one out individually. In fact, If we

were tocounteachpossiblecombinationofparametersettingsasacaseof full-blown

polysemy,wewouldendupwithan intuitivelyquestionableamountofsenses.This is

nottosay,however,thatspatialDERUexhibitsnopolysemyatalleither.Sincemymain

goal is to show how image schematic structure allows for meaning extension into

abstractdomains(andnottogiveanexhaustiveinventoryofsensesforagivenitem),I

willsimplygroupallspatialconfigurationsofDERUunderthe label“spatialEXIT”and

remainagnosticinregardstotheissueofpolysemyvsvagueness(see7.4.).

75

FIGURE1:Theschematicprototypeforspatialexit

--Parameter1:DimensionalityoftheTR(anddegreeofenclosure)--

0DmovingTR:(1) Nezumi-ga ana-kara de-ta.

Mouse-NOM hole-ABL DERU-PAST ‘Amousecameoutofthehole.’

1DstaticTR:(2) Hana-kara ke-ga de-teiru.

Nose-ABL hair-NOM DERU-RES ‘Anosehairisstickingout.’

1DmovingTR:(3) Hari-ga ude-kara de-ta.

Needle-NOM arm-ABL DERU-PAST ‘Theneedlecameout(wasremoved)fromthearm.’

Note that (2) is available from (1) via the image schema transformation 0DMTR <->

1DTR described by Lakoff (1990: 442), i.e. the construal of a one-dimensional TR by

mentallyconnectingthepositionsoccupiedovertimebyazero-dimensionalmovingTR.

(4) Shewenttothetopofthemountain.(0DMTR)

(5) Theroadwenttothetopofthemountain.(1DTR)

(Lakoff1990:442)

A third configuration involves a one-dimensional moving TR. Consider the following

sentences,whichareallsupposedtofeatureapartiallyenclosed1DTR:

(6a) Hana-kara ke-ga de-teiru.

Nose-ABL hair-NOM DERU-RES ‘Anosehairisstickingout.’

(6b) (??)Ude-kara hari-ga de-teiru.

Arm-ABL needle-NOM DERU-RES (??)‘Aneedleisstickingoutofthearm.’

(6c) Kabe-kara kugi-ga de-teiru.

Wall-ABL nail-NOM DERU-RES ‘Anailisstickingoutofthewall.’

Here the questionable status of (6b) can only be understood in respect to the ideal

default locationsof the respectiveTRs.Nosehairsaswell asnails are supposed tobe

76

fully enclosed by their respective LMs (or almost fully in case of the nail). Therefore,

partial enclosurewill be construed as an instance of EXIT. A foreign object such as a

needle, however, is not supposed to even partially enter the body. Therefore, in (6b)

partial enclosure is construedasENTRY,notasEXIT. Inotherwords,whetherpartial

enclosureofa1DTRwillbeconstruedasEXITorENTRYseemstodependontheTR’s

ideal default location. As (6c) shows, depending on its function (and possibly other

factors),aTR’sidealdefaultlocationcanbeinsidetheLM,eventhoughitsinitiallocation

iscompletelyoutsidetheLM.

Insummary:Whenpartialenclosure isconstruedasEXIT,nomovement isrequired

and DERU takes a 1D static TR. By constrast, when partial enclosure is construed as

ENTRY,thenEXITrequiresmovementofthecompleteTRtotheexterioroftheLM–in

whichcaseDERUtakesa1DmovingTR.

--Parameter2:DimensionalityoftheLM--

3DLM(volume):(7) Tarô-ga heya-wo de-ta.

Tarô-NOM room-ACC DERU-PAST ‘Tarôlefttheroom.’

2DLM(plane):(8) Tarô-ga machi-wo de-ta.

Tarô-NOM town-ACC DERU-PAST ‘Tarôleftthetown.’

--Parameter3:BoundednessoftheTR--

BoundedTR:(09) Entotsu-kara kômori-ga de-te ki-ta.

chimney-ABL bat-NOM DERU-TE come-PAST ‘Abatcameoutofthechimney.’

UnboundedTR:(10) Entotsu-kara kemuri-ga de-ta

chimney-ABL smoke-NOM DERU-PAST ‘Smokecameoutofthechimney.’

--Parameter4:NatureoftheLM’sboundaries--

Environmentalboundaries:(11) Shukudai-wo wasure-ta mono-wa kyôshitsu-wo de-nasai.

homework-ACC forget-PAST persons-TOP classroom-ACC DERU-IMP

‘Thosewhoforgottheirhomework,leavetheclassroom!’

Mentallyimposedboundaries:(12) Shukudai-wo wasure-ta mono-wa mae-ni de-nasai.

homework-ACC forget-PAST persons-TOP front-DAT DERU-IMP

‘Thosewhoforgottheirhomework,cometothefront!’

77

--Parameter5:LM’sphaseofmatter--

Empty:(13) Tarô-ga kissaten-wo de-ta.

Tarô-NOM cafe-ACC DERU-PAST ‘Tarôleftthecafe.’

Liquid:(14) Mizu-no naka-kara awa-ga de-ta.

Water-LK interior-ABL bubbles-NOM DERU-PAST ‘Bubblessurfacedfromthewater.’

Solid:(15) Yubi-kara chi-ga de-ta.

Finger-ABL blood-NOM DERU-PAST ‘Bloodoozedfromthefinger.’

--Parameter6:TR’sphaseofmatter--

Empty:(16) Taiya-kara kûki-ga de-ta.

Tire-ABL air-NOM DERU-PAST ‘Airleftthetire.’

Liquid:(17) Jaguchi-kara mizu-ga de-ta.

faucet-ABL water-NOM DERU-PAST ‘Watercameoutofthefaucet.’

Solid:(18) Hako-wo furu to hyakuendama-ga de-te ki-ta.

Box-ACC shake when hundredyencoin-NOM DERU-TE come-PAST

‘WhenIshookthebox,ahundredyencoincameout.’

--Parameter7:PlexityoftheLM--

UniplexLM:(19) Tarô-ga ie-kara de-ta.

Tarô-NOM house-ABL DERU-PAST ‘Tarôleftthehouse.’

MultiplexLM:(20) Tarô-ga hitogomi-kara de-ta.

Tarô-NOM crowd-ABL DERU-PAST Tarôemergedfromthecrowd.’

--Parameter8:Part-wholerelation--

TRisnotapartoftheLM:(21) Ushi-ga koya-kara de-ta.

Cow-NOM barn-ABL DERU-PAST ‘Thecowcameoutofthebarn.’

TRisasub-partoftheLM:(22) Ushi-ga mure-kara de-ta.

Cow-NOM herd-ABL DERU-PAST ‘Thecowemergedfromtheherd.’

78

10.1.2.Sense(II):Activity

(23) Tarô-ga niwa-ni de-ta.

Tarô-NOM garden-DAT DERU-PAST ‘Tarôwenttothegarden.’

(24) Hanako-ga mise-ni de-ta.

Hanako-NOM store-DAT DERU-PAST ‘Hanakowenttothestore.’

(25) Tarô-ga kaisha-ni de-ta.

Tarô-NOM office-DAT DERU-PAST ‘Tarôwenttotheoffice.’

(26) Hanako-ga gakkô-ni de-ta.

Hanako-NOM school-DAT DERU-PAST ‘Hanakowenttoschool.’

(27) Tarô-ga shigoto-ni de-ta.

Tarô-NOM work-DAT DERU-PAST ‘Tarôwenttowork.’

The activity sense is available via metonymic shift from spatial usages that profile a

GOALoutsidetheLM.Accordingtoourencyclopedicknowledgeoftheworld,placesare

usuallyassociatedwithcertainactivities:Schoolswithlearningandteaching,companies

withworking,andsoon.Further,whenweleaveourhomesinordertocarryoutsome

activitysomeplaceelse,weEXITourprivatedomain.Activitiesinthepublicdomainare

not just spatially outside our homes, but also likely to be of a more public nature.

Compare, for example, knitting at home for leisure with knitting at a company for

commercialpurposes.

Astheabovesentencesshow,spatialEXITandactivityarebestthoughtofaspoleson

a continuumwith no clear cut line of demarcation. Take (23), for instance. Here the

metonymic implicatureplace-->activityatplace isratherweakandeasilycancellable:

I.e.,(23)isconsistentwithaninterpretationonwhichTarôtakesaleisurelystrollinhis

own backyard. In such a case, the change of location implies little activity, let alone

publicactivity.Inotherwords,(23)isarathermediocreexamplefortheactivitysense.

(24) is a better example, since going to the store is conventionally understood as

involvingeitherwork(Hanakoasemployeescenario)orshopping(Hanakoasshopper

scenario),bothofwhicharegoodexamplesforactivities.

Generally, then, places that are notmerely physical locations, but also instances of

institutions,tendtoinvitetheplace-->activityreadingmoststrongly.Whilecompanies

andschoolsarerepresentedbybuildingsthatoccupyphysicallocations,theyareatthe

same time abstract purpose-bound entities in the social realm – and as such

conventionally associated with prototypical activities such as learning, teaching or

79

working.1Thisiswhyin(25)and(26)theactivityinterpretationishardertocancelthan

in (23). Finally, in (27), the activity lexically replaces the locus of action as GOAL

argument.Althoughchangeof locationisstill implied, it isclearlybackgroundedvis-a-

vistheactivityitself.

10.1.3.Sense(III):Incubation(fig.2)

(28) Kono daigaku-kara yumei-na gakusha-ga takusan de-ta.

This university-ABL famous-COP.ATT scholar-NOM many DERU-PAST

‘Thisuniversitybroughtforthmanyfamousscholars.’

(29) Kono go-wa ratengo-kara de-teiru.

This word-TOP Latin-ABL DERU-RES ‘ThiswordisderivedfromLatin.’

(30) Sono shûkan-wa ikyôto-no matsuri-kara de-te ki-ta.

This custom-TOP heathen-LK festival-ABL DERU-TE come-PAST

‘Thiscustomdevelopedoutofaheathenfestival.’

AnimportantaspectofourexperiencewithphysicalCONTAINERSistheknowledgethat

the INTERIOR isoftenqualitativelydifferent from theEXTERIOR.Take the insideof a

house,which iswarmandsheltered fromtheelements.But thenwestepoutside into

thecoldrain.Cavesaredark,bodiesofwaterarewet,theinteriorofrefrigeratorscold,

andsoforth.WhenmoreabstractentitiesaremetaphoricallyconstruedasCONTAINERS,

thisdifferenceinqualityisfrequentlymappedontothenewdomain.Onesuchexample

isthecaseofsocialrestrictionandobligation(Lindner1981:104f.;Johnson1990:35).

Entities likeagreementsandcontractsareconstruedasboundedentities.Withintheir

boundaries we are subject to social, legal, or moral forces. But once we get out of a

contractoranagreement,wearefreefromtheseforces.

This notion of qualitative difference gives rise to the concept of incubation. If, over

someperiodof time,something issubjectedtoand influencedbycertain forceswhich

obtainwithinaCONTAINER,thenitwillmostlikelyemergefromtheCONTAINERwith

characteristic features. Consider the case of baking: You put dough into an oven. The

doughissubjectedtotheheatinsidetheoven.Afteracertainamountoftimethedough

emerges as bread. This concept of incubation is frequently applied to abstract

CONTAINERSaswell.Supposesomeonegrowsupinacertainculture.Thatpersonwill

likely be shaped by that culture, even if they end up living in another cultureoutside

1SeealsoRadden(2000:101):“Theassociationbetweensuchman-designedspacesandtheactivitiestypicallyperformedthereissotightthatthementionoftheplacesufficestoevoketheimplicatureofa

specialactivity.”

80

theirown.Nowconsider (28).The implicationhere is that theacademic forceswithin

theinstitutionshapestudentsinsuchawaythattheyemergeaseminentscholars.The

same reasoning applies to the remaining examples.A loanword is locatedoutside the

language itoriginated in,but stillbears characteristicsof that language (phonological,

etc.),andsoforth.

FIGURE2

10.1.4.Sense(IV):Transfer(fig.3)

(31) Kono ryô-wa asagohan to bangohan-ga deru.

This dormitory-TOP breakfast and dinner-NOM DERU

‘Thisdormitoryoffersbreakfastanddinner.’

(32) Kono sagyô-wa kyûryô-ga deru.

This work-TOP payment-NOM DERU ‘Thisispayedwork.’

(33) Shachô-kara kyoka-ga de-ta.

CEO-ABL approval-NOM DERU-PAST ‘TheCEOgavehisapproval.’

(34) Akutagawashô sakka-no hon-ga yo-ku deru.

AkutagawaPrize authors-LK books-NOM good-INF DERU

‘BooksbyAkutagawaPrizewinnerssellwell.’

(35) Gakuhi-wa kikin-ga dasu.

Tuition-TOP foundation-NOM DASU ‘Tuitionispayedforbythefoundation.’

In the above examples the TR is a concrete or abstract object undergoing change of

ownershiporcontrol.AsLindner(1981:105)notes,theLMinsuchcases“isconstrued

asanabstractneighborhoodaroundaperson,asortof sphereof influence, such that

items owned are IN it and items transferred to someone else are OUT.” This

metaphorical construal of transfer in terms of the EXIT schema is hardly surprising,

seeing how the concepts of ownership and control are experientially grounded in the

sensationof spatialproximity.Prototypically, ifAowns/controlsB, thenB is spatially

81

proximal to A. If B leaves the proximity of A and becomes proximal to some other

individualC,thenchangeofownership/controlofBfromAtoCisimplied.2

Insummary,whatthescenesdescribed in(31)-(35)have incommon, is that theTR

leaves the sphereof influence3of someentityandcoincidentially enters the sphereof

influenceofanotherentity.Note thatneither theoldnor thenewowner/controllerof

theTRneedstobementionedexplicitly.Often,asin(31),(32)and(34),bothareonly

impliedandremainschematic. In(31), for instance, thedormitoryasan institutionor

thekitchenpersonnel concedeownershipof/controloverbreakfast anddinner to the

inhabitants,althoughnoneoftheseentitiesarelinguisticallyrealized.

FIGURE3

10.1.5.Sense(V):Access(fig.4)

(35) Negi-no kaori-ga de-te ki-tara, shio-wo furi-ireru.

Scallion-LK aroma-NOM DERU-TE come-when salt-ACC sprinkle

‘Oncethescallionsbecomefragrant,addsalt.’

(36) Supiikâ-kara oto-ga de-nai.

Speaker-ABL sound-NOM DERU-NEG ‘Nosoundcomesfromthespeaker.’

(37) Kinô shiriai-ga terebi-ni de-ta.

Yesterday acquaintance-NOM TV-DAT DERU-PAST

‘YesterdayanacquaintanceappearedonTV.’

(38) Kono yashiki-ni yûrei-ga deru-rashii.

This mansion-DAT ghosts-NOM DERU-EVI ‘Thismansionissaidtobehauntedbyghosts.’

(39) Nakushi-ta saifu-ga de-te ki-ta.

Lose-PAST wallet-NOM DERU-TE come-PAST ‘(My)lostwalletturnedup.’

2SeealsoTaylor(1996:340),whosuggeststhatproximityisanaspectofthe“possessiongestalt”:“Inorderthatthepossessorcanhaveeasyaccesstothepossessed,thepossessedistypicallylocatedinthe

proximityofthepossessor.Insomecasesthepossessedmaybeapermanent,oratleastregular

accompanimentofthepossessor.”Notethatownershipandcontrolarespecificinstancesofpossession.

Foranoverviewofpossessionrelationsseee.g.Langacker(1991:169).3IhaveoptedforLindnder’s(1981:105)term“sphereofinfluence”overLangacker’s(1991:170)term“dominion”inthiscasetoexcludepossessionrelationswherethepossessordoesneitherownnorcontrol

thepossessed(e.g.Thedoghasfleas).ThetransfersenseofDERUalwaysentailsownershiporcontrol.

82

(40) Yo-ku kangae-tara kitto kotae-ga deru.

Good-INF think-COND surely answer-NOM DERU

‘Ifyouthinkhard,theanswerwillsurelycometoyou.’

(41) Rainen shingata pasokon-ga deru-rashii.

Nextyear newmodel PC-NOM DERU-EVI

‘Reportedly,anewPCmodelwillreleasenextyear.’

(42) Mada honki-wo dashi-tei-nai dake da.

Yet seriousness-ACC DASU-RES-NEG just COP ‘I’vejustnotgottenseriousyet.’

Among ourmost fundamental experienceswith CONTAINERS is the inaccessibility of

theirINTERIOR.Oneofthemostprototypicalconceptsofinaccessibility,OCCLUSION,is

acquired during early infancy: „Presumably infants are attracted to containment and

occlusioneventsbecausetheobjectstheyarewatchingdisappearfromsight;peoplego

outoftheroom,objectsgointopansandcupboards.Itmaybetheseactsofdisappearing

thatmakecontainersthefirstobjectswearesurethatinfantsconceptualize(otherthan

peopleandtheireyes)“(MandlerandPagánCánovas2014:6).BesidesOCCLUSION,we

experience acoustic and olfactoric inaccessibility, e.g.when a roomholds in noises or

smells. Inotherwords, sensory inaccessibility is anessential aspectofourknowledge

concerning CONTAINERS. Further, since knowledge is ultimately grounded in

perception, sensory inaccessibility entails epistemic inaccessibility. (We understand

metaphors such as KNOWING IS SEEING [I see your point, He showed me the truth, etc.]

preciselybecauseofthiscorrelationbetweenperceptionandknowledge.4)

Based on our experience with physical CONTAINERS, it makes sense, then, to

metaphorically construe inaccessibility in terms of CONTAINMENT and – by image

schematicentailment–thechangetoaccessibilityintermsofEXIT.Theimageschematic

topology of the metaphor equates inaccessibility with being located inside and

accessibilitywithbeing locatedoutsideofaCONTAINER.Thisexplainscases like(39):

Here,DERUindicatesthatthewallet,whileremaininglost,wasperceptuallyinaccessible

toitsowner–astatethatchangedbyitsbeingfound.Likewise,theTRsin(40)and(41)

become epistemically and economically accessible, respectively. Also note that

sometimes various kinds of accessibility are conflated. Someone who appears on

televisioncanbeseenandheard.Anewcomputerbecomesavailable forsaleanduse,

etc. Finally, as (42) perhaps showsbest, TRs exist as unrealized “potentials” (Lindner

1981:109)foraslongastheyarelocatedinsidetheCONTAINER.

4SeealsoSweetser’s(1991:32ff.)studyofEnglishandIndo-Europeansense-perceptionverbs.

83

FIGURE4

10.1.6.Sense(VI):Excess(fig.5)

(43) Hiyô-wa sanmanen-wo de-nai.

Expenses-TOP 30.000yen deru-NEG ‘Theexpensesdon’texceed30.000yen.’

(44) Tarô-no ryôri-wa shirôto-no iki-wo de-teiru.

Tarô-LK cooking-TOP amateur-LK level-ACC deru-RES

‘Tarô’scookingispasttheamateurlevel.’

(45) Hanako-wa rokujussai-wo de-teiru.

Hanako-TOP sixtyyears-ACC deru-RES ‘Hanakoisoversixty.’

If STATESARELOCATIONS5and CONTAINERS (by their nature) are bounded regions, then

normativitycanbeconstruedasaCONTAINER(oraSCALE,i.e.a1Dboundedregion).In

the construal of non-normativity there are twomajor image schematic variants: One

with a uniplex TR and one with a mass TR. Consider the following examples from

English:

(46) Hisbehaviorliesoutsidethenorm.

(47) Herskillexceedsthenorm.

In (46) the TR his behavior is conceptualized as a zero-dimensional object (a point)

whichislocatedoutsidetheboundariesoftheLMthenorm.Thus,thereisnooverlapat

allbetweenTRandLM.Incontrast,themassTRin(47)“spillsacross”theboundariesof

ascalarLM.That is, theTRbothoccupiesandexceedsthe interiorof theLM.It is this

schemaofEXCESSwhich isatwork in(43)-(45).Forexample, in(45)Hanako’sage is

understood to occupy all locations on the age scale up to sixty, plus some amount

exceeding the scale. Note that the EXCESS schema (an instance of EXIT) is firmly

5Recallourdiscussionoftheeventstructuremetaphorunder4.1.

84

grounded in our experience with the physical world, e.g. overflowing or bursting

CONTAINERS,linearobjectsexceedingotherobjectsinheightorlength,etc.6

FIGURE5

10.1.7RelationsBetweenSensesandCategorialFringeCases

From the above analysis it is plain that several examples fall somewhere “inbetween

senses”. Since the senses themselvesare related systematically, this is tobeexpected.

Considerthefollowingsentence:

(48) Senshu-ga shiai-ni de-ta.

Athlete-NOM game-DAT DERU-PAST ‘Theathleteappearedin(tookpartin)thegame.’

Thiscanbecategorizedaseitheractivityoraccessorboth.Thereasonisthatshiai isa

cluster model in Lakoff’s sense (1990: 74ff.), which means that the word can be

understood in reference to more than one background frame or idealized cognitive

model(ICM).Inparticular,whatconcernsushereistheconflationofthefollowingtwo

ICMs:

Thework/professionICM:Partakinginsportingcontestsistheathlete’sjob.Itisan

activitywhichearnshimmoneyandasocialobligation.

6AtthispointitisinteresingtonoteinpassingthatthesemanticsofDERUasoutlinedaboveshowconsiderableoverlapwiththesemanticsof“out”inEnglishverbparticleconstructions.Infact,manyof

thesenseswehaveencounteredabovearefamiliarfromLindner’s(1981)seminalanalysisof“out”.These

include:LMasabstractneighborhoodofpossession,LMasprivacy,andchangefromhiddennessto

accessibility.DERUand“out”arefarfromsynonymous,however,anddifferinimportantrespects.For

example,whiletheparticle“out”canindicatechangefrominaccessibilitytoaccessibility(findoutthe

truth)aswellaschangefromaccessibilitytoinaccessibility(ruleoutapossibility),DERUisunidirectional.

WhileDERUcanindicatescalarexcess(i.e.exceedacertainamountofquantity),Englishusesoverrather

thanoutinsuchcases,etc.

Ofcourse,overlapistobeexpectedconsideringtheuniversalnatureoftheembodiedschemaEXIT.But

atthesametimeconceptualizationisflexibleandweare,inthevastmajorityofcases,notforcedto

construeacertainconceptintermsofoneandonlyonecorrespondingschema.

85

ThespectatorsportsICM:InJapaneseculturesportingeventsareapopularformof

publicentertainment.Thatis,baseballandsoccerare“spectatorsports”.

What tomakeof this?Note that theabove ICMsare (a) intrinsically relatedwhile (b)

presupposingdifferent viewing arrangements. In thediscussionof theactivity sense I

have already mentioned that activity implies some degree of publicness. Now,

publicness in turn implies accessibility. To engage in a sporting contest, an athlete

movesoutoftheprivatedomainandtherebybecomespubliclyaccessibletoaspectator

crowd.Butnotethat“movingoutoftheprivatedomain”istiedtotheathletesinternal

perspective,whereas“becomingpubliclyaccessible” is tiedtotheexternalperspective

of the spectator. I.e., there is a cline from activity to publicness to accessibility,

accompaniedbyashiftfrominternaltoexternalperspective.Consequently,whetherwe

categorizeaparticularuseofDERU(suchas48)asactivityoraccesswilloftendepend

on whether we adopt an internal or an external viewing arrangement where human

activityisconcerned.7Infact,ifoneweretoemploya“god’seyeview”theactivitysense

couldbesubsumedundertheaccesssense,sincehumanactivityusuallyimplies(some

degreeof)publicaccess,whereasnotallkindsofaccessimplyhumanactivity.However,

since the access sense presupposes an external viewing arrangement, instances of

activitywhichemployaninternalviewingarrangementshouldnotbesubsumedunder

access.

Muchof thesamecouldbereiterated for the transfer sense. Ifwedefine transferas

changeofownership/controlandacceptTaylor’sviewthatpossessiontypicallyimplies

“easy access” (1996: 340), then it follows that transfer implies access. Recall the

followingexamplesfromabove:

(31) Kono ryô-wa asagohan to bangohan-ga deru.

This dormitory breakfast and dinner-NOM DERU

‘Thisdormitoryoffersbreakfastanddinner.’

(34) Akutagawashô sakka-no hon-ga yo-ku deru.

AkutagawaPrize authors-LK book-NOM good-INF DERU

‘BooksbyAkutagawaPrizewinnerssellwell.’

As these sentences show, change of ownership/control coincides with change from

inaccessibility to accessibility from the perspective of the new owner/controller (and

7Notethatthegrammatical“firstperson“isnotnecessarilyanindicatorofaninternalviewingarrangement.AsDewell(1997:24)observes,theuttererofthesentenceWelooklikeabunchofidiots

sittinghereneverthelessadoptsanexternalviewingarrangement.

86

(49) Tarô-ga Shinjuku-kara Shinagawa-made arui-ta.

Tarô-NOM Shinjuku-ABL Shinagawa-ALL walk-PAST

‘TarôwalkedfromShinjukutoShinagawa.’

(50a) Tarô-ga Shinjuku-wo arui-ta.

Tarô-NOM Shinjuku-ACC walk-PAST ‘Tarôwalkedthrough(wanderedabout)Shinjuku.’

(50b) Tarô-ga chûôdôri-wo arui-ta.

Tarô-NOM centralstreet-ACC walk-PAST ‘Tarôwalkeddownthecentralstreet.’

While (49)construes theLMasazero-dimensionalpointofdeparture, (50)construes

theLMasaplanar(50a)orlinear(50b)traversableextent.

In this section Iwill attempt to give an explanation of the principles governing the

felicitoususeofkaraandwointandemwithderu.Letusstartwiththeobservationthat

theparticlescannotalwaysbeusedinterchangeably:

(51a) Kemuri-ga entotsu-kara de-ta.

Smoke-NOM chimney-ABL DERU-PAST ‘Smokecameoutofthechimney.’

‘Smokecameoutofthechimney.’

(51b) (??)Kemuri-ga entotsu-wo de-ta.

Smoke-NOM chimney-ACC DERU-PAST

(52a) Nezumi-ga ana-kara de-ta.

Mouse-NOM hole-ABL DERU-PAST ‘Themousecameoutofthehole.’

(52b) (??)Nezumi-ga ana-wo de-ta.

Mouse-NOM hole-ACC DERU-PAST.

(53a) Tarô-ga ie-kara de-ta.

Tarô-NOM house-ABL DERU-PAST ‘Tarôleftthehouse.’

(53b) Tarô-ga ie-wo de-ta.

Tarô-NOM house-ACC DERU-PAST ‘Tarôleftthehouse.’

Himeno (1977:76)has suggested thatwo requires theTRofderu tobe animate.The

questionablefelicityofsentencessuchas(52b)andthesuccessfulsubstitutionin(53b)

seems to indicate that the TR need not only be animate but also human (or at least

anthropomorphic).IwillarguethatindeedhumanTRshavethebestcompatibilitywith

wo.However, thisdoesnotentail thatanimacy is thecentral issuehere.Moreover,we

should be able to give an account of the difference between (53a) and (53b). In

summary,wemustasktwoquestions:

• WhyarehumansthebestcandidatesforTRofderuintandemwithwo?

• Incaseswhere[N-karaderu]and[N-woderu]canbeusedinterchangeably,what

isthesemanticdifference?

88

Again, consider the difference between external and internal viewing arrangement

discussedintheprevioussection.Ifweassume–basedonourexperienceswithhouses,

caves,boxes,etc.–thatnon-transparencyisaprototypicalfeatureofCONTAINERS,then

thechangefromexternaltointernalviewpoint(andviceversa)isquitedrastic.Imagine

ahousefromanexternalviewpoint.AsDewell(1997:24)putsit,thehousewillbe“like

a black box“ hiding everything inside it. By contrast, imagine an interior path going

throughit:

ALMsuchasahousecontainsaheterogenouscollectionofdistinctobjectsorparts,includingstructurally

inherentroomsandhallwaysanddoors,avarietyofpiecesoffurniture, inhabitants,etc.Apaththrough

such as space [...] is naturallydefinedas a continuousprogression fromone suchvisible component to

another,forexamplefromadoorthroughahallwaypastanotherdooranddisappearingatanotherdoor.

(Dewell1997:26)

Applythisto(53b).Thissentenceismostfelicitouslyutteredbysomeonelocatedinside

thehouseatthetimeofTarô’sleaving.Asacommentfromsomeoneontheoutsidewho

sawTarôemergingfrom,say,thefrontdoorintothestreets(53b)wouldsoundrather

odd.Inthecaseof(53a),theexactoppositeapplies.

Inotherwords,wointandemwithderuconstruestheLMasaninternalPATH.Inan

wo-construal we trace the TR’s PATH through the interior of the CONTAINER. The

tracingstopsastheTRexitstheCONTAINERandtheworldontheoutsideisbeyondthe

scopeoftheconceptualizer’sview.Akara-construal,ontheotherhand,impliesthatthe

conceptualizer is located outside the CONTAINER. This is the “black box” perspective

withnoknowledgeof theLM’s internal structure. Instead, all attention is focussedon

thepointofemergence,fromwhichtheTRwillmakeitsappearance.Thedifferencein

construalisillustratedinfigure7:

FIGURE7:kara-constural(left)vswo-construal(right)

89

The view that [N-wo deru] presupposes an internal viewing arrangement is also

supportedbytheincompatibilityofwowiththesomewhatlexicalizedvariantde-tekuru

(‘comeout’):

(54) Tarô-ga ie-kara de-te ki-ta.

Tarô-NOM house-ABL DERU-TE come-PAST ‘Tarôemergedfromthehouse.’

(55) (??)Tarô-ga ie-wo de-te ki-ta.

Tarô-NOM house-ACC DERU-TE come-PAST

Theconstruction[V-tekuru]indicatesadeicticpathtowardstheconceptualizer.I.e.,de-

te kuru requires the TR to exit a CONTAINER and subsequently move towards the

conceptualizer – which is impossible if the conceptualizer is located inside that

CONTAINER.8

Notethataperspective-basedexplanationcanaccountfortheselectionofkaravswo

intheabstractdomainaswell.Considerthefollowingsentencepairs:

(56a) Kono daigaku-kara yumei-na gakusha-ga takusan de-ta.

This university-ABL famous-COP.ATT scholars-NOM many DERU-PAST

‘Thisuniversitybroughtforthmanyfamousscholars.’

(56b) (??)Kono daigaku-wo yumei-na gakusha-ga takusan de-ta.

This university-ACC famous-COP.ATT scholars-NOM many DERU-PAST

(57a) Tarô-ga kyonen daigaku-wo de-ta.

Tarô-NOM lastyear university-ACC DERU-PAST

‘Tarôgraduatedfromuniversitylastyear.’

(57b) (??)Tarô-ga kyonen daigaku-kara de-ta.

Tarô-NOM lastyear university-ABL DERU-PAST

(56a)emphasizesthequality(famous)oftheTRandimpliesalinkbetweenthisquality

andsomequalityoftheLM.ItissuggestedthattheTRhasqualityFbecauseitemerged

fromtheLM,whichhassomequalityG.Itisthiscausallink–theLMastheSOURCEof

andtheTRasthebeareroffamousness–whichisemphasizedhere,nottheprocessby

which the TR acquired said quality. In other words, (56a) construes the LM as the

SOURCEofsomeoutput(famousscholars),whilebackgroundingtheinteriorPATHtaken

bytheTR(i.e.howthescholarsmaketheirwaythroughtheinstitution).Sincenosuch

“causal frame“ is evoked by (57a), the sentence is interpreted against a default

8Granted,thisisaslightoversimplificationforillustrativepurposes.Tobesure,onecanassumeaninternalperspectivewithoutactuallybeinglocatedinsideaCONTAINER.Thepointisthatonecannotat

thesametimeassumeboththeexternalperspectivepresupposedby[V-tekuru]andtheinternal

perspectivepresupposedby[N-woderu].Hencetheinfelicityof(55).

90

“graduationframe”whichforegroundstheinternalstructureoftheLM.Forexample,we

knowthatgraduationisthefinalcomponentofanacademicprogramwhichconsistsof

ca. four years,with each year consisting of several semesters. The semesters, in turn,

consistofcourses.Coursesrequirecourseworkandexamsforwhichcreditpointsare

awarded, etc. The default “graduation frame”, then, causes us to trace the students

interiorpaththroughanacademicprogram.

Sofarwehaveaccountedforthedifferencebetweenkaraandwointermsofexternal

vsinternalviewingarrangement.Butwhydoestheinternalvariant[N-woderu]seemto

favor a human TR? The answer is quite straightforward: An internal perspective

typically requires a CONTAINER-internal conceptualizer – and conceptualizers are

human. For a human conceptualizer and a non-human TR to be both located

CONTAINER-internally is simply a very rare state of affairs.9The following examples

illustratethis:

(58) (??)Kanjûsu-ga jidôhanbaiki-wo de-ta.

Canofjuice-NOM vendingmachine-ACC DERU-PAST

Intendedmeaning:‘Acanofjuicecameoutofthevendingmachine.’

(59) (??)Kemuri-ga entotsu-wo de-ta.

Smoke-NOM chimney-ACC DERU-PAST Intendedmeaning:‘Smokerosefromthechimney.’

(60) (??)Nezumi-ga kabe-no ana-wo de-ta.

Mouse-NOM wall-LK hole-ACC DERU-PAST

Intendedmeaning:‘Amousecameoutofthehole.’

(61) (??)Kuma-ga dôkutsu-wo de-ta.

Bear-NOM cave-ACC DERU-PAST Intendedmeaning:‘Abearcameoutofthecave.’

(62) (??)Namida-ga me-wo de-ta.

Tears-NOM eyes-ACC DERU-PAST Intendedmeaning:‘Tearscameoutof(his/her)eyes.’

Theproblemwithallofthesesentencesisthatitwouldbeextremelyoddforahuman

conceptualizertohaveaCONTAINER-internalperspective.Humansareusuallylocated

outsideofvendingmachines,chimneys,animalhabitats,andsoon.Ofcourse,specialists

9Notehowever,thatthisconstellationisnotimpossible.Forexample,(61)mightbefelicitouslyutteredbyaspelunker.Furthermore,aquicksearchontheinternetbroughtupthefollowingexample:

Kyôgaku-nokyû-byô!Heya-wodetainu-gaikinarikaidan-wo...

Ninesecondsofastonishment!Dogleavesroomand(dashesdown)thestairs...

(http://petfilm.biz/?p=5760,retrieved22Apr.2015)

Sincethekeepingofpetsconstitutesoneoftherarecasesinwhichhumansandanimalsshareahabitat,

usageof[N-woderu]isunproblematichere.

91

mayhavean“internalperspective”asfarassomeoftheseCONTAINERSareconcerned.

But such highly atypical construal has little impact on the entrenchment of linguistic

constructionsingeneralusage.

10.2.TheSensesofV-DERU

ThesemanticsofV-DERUfollowstraightforwardlyfromthesemanticsofthesimplex.I

willbrieflytouchuponthecorrespondingsensesof–DERUbeforediscussinginchoative

-dasu in some more detail. The question why the transitive variant -dasu is often

attachedtointransitiveverbswillbeaddressedinalaterchapter(see14.4.).

10.2.1.SpatialV–DERU

(63) Chi-ga shatsu-kara nijimi-de-ta.

Blood-NOM shirt-ABL ooze-DERU-PAST ‘Bloodoozedfromtheshirt.’

(64) Me-kara Namida-ga kobore-de-ta.

Eye-ABL Tears-NOM drop-DERU-PAST ‘Tearsdroppedfrom(his/her)eyes.’

(65) Tamago-ga su-kara korogari-de-ta.

Egg-NOM nest-ABL roll-DERU-PAST ‘Theeggrolledoutofthenest.’

(66) Tarô-ga heya-wo tobi-de-ta.

Tarô-NOM room-ACC dash-DERU-PAST ‘Tarôdashedoutoftheroom.’

AsHimeno(1977:75)notes,-DERUcanbeattachedtobothpathverbs(63and64)and

motionverbs(65and66).WhiletheformerincludetheEXITschemaaspartoftheir

meaning,thelatterexpressapath-neutralmannerofmotion.Therefore-deruin(63)

and(64)isbestunderstoodasadditionalemphasisoftheEXITschema,whereasin(65)

and(66)-deruservesasakindofframingsatellite.10

10.2.2.Activity

(67) Hanako-ga Tarô-ni shien-wo môshi-de-ta.

Hanako-NOM Tarô-DAT support-ACC speak.HUM-DERU-PAST

‘HanakoaskedTarôforsupport.’

(68) Jirô-ga kaisha-ni jishoku-wo negai-de-ta.

Jirô-NOM company-DAT resignation-ACC wish-DERU-PAST.

‘Jirôsubmittedhisresignationtothecompany.’

10AccordingtoTalmy(2003b:221f.),languagestendtoeitherexpressmotionpathsviathemainverborviaasatellite(e.g.aparticlesuchasoutinmoveout),i.e.languagesareeitherverb-framedorsatellite

framed.WhilehecountsJapaneseamongtheverb-framedkind,examplesliketheaboveserveasa

reminderthatthedistinctionisnotabsolute.

92

In sentences like the above the TR leaves the private domain for communicative

purposes. Since the TR thereby makes itself accessible, it is a matter of perspective

whethertocategorizethesecompoundsasinstancesofactivityoraccess(seediscussion

above). As a rule of thumb, if the compound is primarily associatedwith an internal

viewingarrangement,itshouldbecategorizedunderactivity.

10.2.3.Incubation

(69) Tarô-ga ii hôhô-wo kangae-dashi-ta.

Tarô-NOM good method-ACC think-DASU-PAST ‘Tarôcameupwithagoodmethod.’

(70) Kaihatsubu-ga shinseihin-wo tsukuri-dashi-ta.

Developmentdepartment-NOM newproduct-ACC create-DASU-PAST

‘Thedevelopmentdepartmentcreatedanewproduct.’

Note that the direct objects in these sentences are understood as having existed as

unrealized potentials within the mental domain of the TR from which they are then

“releaseduntotheworld”.Theseentitiesexist“insidetheminds”oftheircreatorsforan

incubationperiodbeforetheyemergewithcharacteristicfeatures(e.g.,Tarô’sideawill

beshapedbyTarô’swayofthinking).Again,thisusagetypeisconflatedwiththeaccess

sense.

10.2.4.Transfer

(71) Tarô-ga Hanako-ate-ni tegami-wo okuri-dashi-ta.

Tarô-NOM Hanako-aim-DAT letter-ACC send-DASU-PAST

‘TarôsentaletteraddressedtoHanako.’

Here-dasucanbeunderstoodasemphasizingthechangeofcontrol/ownershipencoded

bytheV1.Therearehardlyanyinstancesoftransferapartfromokuri-dasu.

10.2.5.Access

(72) Hanako-ga Tarô-wo kissaten-ni yobi-dashi-ta.

Hanako-NOM Tarô-ACC cafe-DAT call-DASU-PAST ‘HanakocalledTarôtothecafe.’

(73) Keiji-ga hannin-no dôki-wo saguri-dashita.

Detective-NOM criminal-LK motive-ACC search-DASU-PAST

‘Thedetectivesoughtoutthecriminal’smotive.’

(74) Tarô-ga tsuri-ni ik-ô to ii-dashi-ta.

Tarô-NOM fishing-DAT go-VOL QT say-DASU-PAST ‘Tarôsuggestedtogofishing.’

(75) Akari-ga otoko-no kao-wo terashi-dashi-ta.

93

Lamplight-NOM man-LK face-ACC illuminate-DASU-PAST

‘Thelamplightilluminatedtheman’sface.’

(76) Hanako-ga kami-ni namae-wo kaki-dashi-ta.

Hanako-NOM paper-DAT name-ACC write-DASU-PAST

‘Hanakowrotehernameonthepaper.’

AmongtheabstractsensesofV-DERUaccessisbyfarthemostproductiveone.Interms

ofimageschematicstructure,everythingthatappliestotheaccesssenseofthesimplex

applies here as well. The acces sense of -dasu is represented by the following

constructionalschema:

X-ga Y-wo/to V-dasu

XCAUSE YBECOMEACCESSIBLE BY(X’s)DOINGV

Asmentionedabove, it isnotuncommonfora lexical itemtobelongtomorethanone

category.Forexample,(69)canbethoughtofasinstantiatingbothincubationandaccess.

10.2.6InchoativeV-dasu(fig.8)

(77) Hanako-ga Tarô-no fukusô-wo mi-te, warai-dashi-ta.

Hanako-NOM Tarô-LK clothes-ACC see-TE laugh-DASU-PAST

‘HanakosawTarô’soutfitandburstintolaughter.’

(78) Tarô-ga shiken-no kekka-wo shit-ta totan, watto naki-dashi-ta.

Tarô-NOM test-LK results-ACC learn-PAST moment suddenly cry-DASU-PAST

‘ThemomentTarôlearnedofhistestresultsheburstintotears.’

(79) Enjin-no oto-ga nari, kuruma-ga ugoki-dashi-ta.

Motor-LK sound-NOM makesound.CONJ car-NOM move-DASU-PAST

‘Theengineroaredandthecarstartedtomove.’

(80) Kyû-ni ame-ga furi-dashi-ta.

Suddenly rain-NOM fall-DASU-PAST ‘Suddenly,itbegantorain.’

In this section Iwill argue that inchaotive V-dasu is an extension of theaccess sense.

Therearetwomainargumentstoconsiderinsupportinthisview.

(i)Theargumentfromperceptibility

Note that there is a good amount of experiential correlation between access and

inchoativity. The beginning of a process often coincides with a perceptible change of

state.Obviously,thisisthecasewhensomeonebeginstolaugh,cry,move,etc.Inother

words, the process described by the V1s in (77)-(80) becomes externally accessible.

However,manyprocesseslackthiskindofperceptiblechange,e.g.mentalorattentional

94

activities such as thinking or listening. In light of this, the collocational behavior of

inchoative -dasu isquite telling.Consider the findingsof tworecentcorpusstudiesby

Yamaguchi (2009) and Ishikawa (2010). The most frequent collocational V1s of

inchaotive-dasufrombothstudiesarelistedbelow:

collocationalV1sofinchaotive-dasu(totalfrequency,descendingorder)

Yamaguchi(2009) Ishikawa(2010)

Corpusused:

43novelswrittenbyJapanesenative

speakersandreleasedbetween1898and

2009

Corpusused:

BalancedCorpusofContemporaryWritten

Japanese(2009)11

ugoku(move) iu(say)

aruku(walk) aruku(walk)

naku(cry) warau(laugh)

suru(do) hashiru(run)

okoru(becomeangry) naku(cry)

hanasu(speak) kakeru(run)

warau(laugh) ugoku(move)

utau(sing) furu(rain)

naru(sound) hanasu(speak)

odoru(dance) naru(sound)

Considering that different corpora have been used, both studies yield fairly similar

results. As Ishikawa (2010: 26) observes, inchoative -dasu has a strong tendency to

occurwithV1sthatprofilea“clear-cutphysicalaction”(hakkirishitabutsuritekidôsa).

Henotesthatinchoative-dasuprefersthoseV1sthatimplyastarkcontrastbetweenthe

poles of inactivity and activity. For example, as far as verbs of verbal expression are

concerned, -dasushowsstrongcollocationaltieswithhanasu(speak)andkataru(talk,

narrate), but never attaches to sasayaku (whisper) or tsubuyaku (murmur) (Ishikawa

2010:27).Further,Himeno (1977:90)pointsout that inchoative -dasu frequently co-

occurs with expressions indicating sudden and abrupt change, such as fui ni

(unexpectedly)andkyûni(suddenly).Infact,shesuggeststhatinchoative-dasuisbest

understoodasaspontaneousoutburstofinternalenergyandthereforeasanextension

ofthecentralsenseofoutwardsmovement(Himeno1977:89).

AlthoughIamingeneralagreementwithHimeno’sassessment,myviewdiffersinone

respect: Inchaotive -dasu is not a direct extension of the spatial sense but rather

11http://pj.ninjal.ac.jp/corpus_center/bccwj/,linkretrieved24Apr.2015

95

mitigated via the access sense. What the most frequent V1s from the above corpus

studieshaveincommonistheirexternalaccessibility:Whensomeonebeginstospeak,

dance,laugh,cry,etc.,thereisanimmediateanexternallyperceptiblechangeofstate,i.e.

othershavesensoryaccess to thestateof speaking,dancing, laughing, andsoon.And

themore sudden and abrupt the change of state is, the higher the chance of external

sensory access becomes. Contrast this with the above verbs of “low key” verbal

expression (whisper,murmur)orverbsdenoting internalprocesses. It canbehard to

tellwhensomeonestartswhispering,becausethechangefromsilencetonon-silenceis

gradual and subtle.Whisperingby its verynatureposes a challenge toperception.Or

consider an internal process such as listening: How can one be sure when exactly

someonestartslisteningtoaradioprogram(tokenreading)?Thecuestopickuponare

limitedandsubtle.Consideranexample:

(81) Rajiokôza-wo kiki-dashi-ta.

Radiolectures-ACC listen-DASU-PAST

AsSuk(2004:159)notes,thedefaultinterpretationof(81)isahabitualreadingalong

the lines of (One day) she started listening to radio lectures or (One day) she started

listeningto(someparticularseriesof)radiolecture.Ontheviewthatinchoative-dasu is

conceptually linkedtotheaccesssensewecaneasilyexplainwhyweendupwiththis

interpretation: The beginning of a particular listening event is difficult to discern,

whereasachange inhabit isoftenplainlynoticeable toone’s surroundings (e.g.when

someonewhonevershowedaninterestinradiolecturesbecomesanardentfollower).

(ii)Theargumentfromnon-intentionality

It has often been pointed out (e.g. Himeno 1977; Morita 1991) that inchoative -dasu

seemstobeincompatiblewiththeexpressionofintentionality:

(82) (??)Ronbun-wo kaki-dashi-tai.

Thesis-ACC write-DASU-DES Intendedmeaning:‘Iwanttostartwritingthethesis.’

(83) (??)Ronbun-wo kaki-das-ô.

Thesis-ACC write-DASU-VOL Intendedmeaning:‘Let’sstartwritingthethesis.’

(84) (??)Ronbun-wo kaki-das-e.

Thesis-ACC write-DASU-IMP Intendedmeaning:‘Startwritingthethesis!’

Such expressions require the TR of the V1 to be in a mental state that is about the

processprofiledby theV1, prior to that process’ realization. For example,wanting to

96

writesomethingentails thinkingaboutwritingbeforeactuallydoing it.Nowrecall the

conceptual metaphor(s) underlying the access sense of DERU: OUT IS ACCESSIBLE/IN IS

INACCESSIBLE. According to the “logic“ of this metaphor, access is inevitably tied to an

externalviewpoint.Appliedtoinchaotive-dasu thismeans:Realizedprocesses, insofar

as they are accessible by our sensory faculties, are OUT. Unrealized processes (i.e.

processesthatexistonlyasintentionalobjects“inthemind”,aspossibilitiesorinsome

othernon-actualstate)areinaccessiblebyoursensoryfacultiesandthereforeIN.Ifthis

isso,wecaneasilyaccountfortheinfelicityof(82)-(84):Intentionalitypresupposesan

internal viewpoint and entails access to unrealized processes (i.e. in the form of

intentionalobjects),butaccordingtotheabovemappingsunrealizedprocessesareIN–

andthereforeinaccessible.Inotherwords,themetaphoricalstructurewhichunderlies

themeaningofinchaotive-dasuisatoddswiththeconceptofintentionality.

In summary, then, both the fact that inchoative -dasu entails perceptible change as

well as its incompatibility with expressions of intentionality are straightforwardly

accountedforifwetreatthissenseasanextensionoftheaccesssense.

FIGURE8.1.:furi-dasu(begintorain)

FIGURE8.2.:hanashi-dasu(begintospeak)

97

11.KiruandtheSPLITSchema

For the majority of its senses the verb kiru can be schematically characterized asfollows:

X-ga Y-wo kiru

XCAUSE YSPLIT

11.1.Thesensesofkiru

11.1.1.Sense(Ia):PhysicalDiscontinuity-LMisaSolidExtentofMatter(fig.1)

(1) Tarô-ga niwa-no ki-wo kit-ta.

Tarô-NOM garden-LK tree-ACC KIRU-PAST ‘Tarôcutthetreeinthegarden.’

(2) Hanako-ga tsume-wo kit-ta.

Hanako-NOM fingernails-ACC KIRU-PAST ‘Hanakocutherfingernails.’

(3) Kêki-wa yottsu-ni kit-te kudasai.

Cake-TOP fourpieces-DAT KIRU-IMP please ‘Pleasecutthecakeintofourpieces.’

ThissensefeaturesaspatiallyextendedsolidLMwhichissegmentedintonumerically

distinctpartsasaresultoftheTR’saction.

FIGURE1(POSstandsforpointofsegmentation)

11.1.2.Sense(Ib):UnintentionalSelf-injury-NoSegmentation(fig.2)

(4) Hanako-ga naifu-de yubi-wo kit-teshimat-ta.

Hanako-NOM knife-INS finger-ACC kiru-IRR-PAST ‘Hanakocutherfingerwithaknife.’

(5) Tarô-ga kamisori-de kao-wo kit-teshimat-ta.

Tarô-NOM razor-INS face-ACC kiru-IRR-PAST ‘Tarôcuthisfacewitharazor.’

(Ib) is related to (Ia) in that, here too, the structure of the LM is altered by a cutting

instrument. Apart from this, (Ib) is much more narrow in its application due to the

followingconstraints:

98

• TheLMmustbeabodypart.

• Thecuttingofthebodypartmustbeunintentional.

• Thecuttingdoesnotresultinsegmentation.

Although this sense is related to (Ia)via family resemblance, itdoesnotqualify as an

instanceoftheSPLITschemaduetothethirdconstraint.Notethatthenosegmentation

readingisnotavailablefornon-animateLMsinthephysicaldomain:

(6) (??)Hanako-ga kêki-wo kit-teshima-ta.

Hanako-NOM cake-ACC KIRU-IRR-PAST

FIGURE2

11.1.3.Sense(Ic):Opening-LMisaCONTAINER(fig.3)

(7) Ryôrinin-ga futa-wo kit-ta.

Cook-NOM lid-ACC KIRU-PAST ‘Thecookliftedthelid.’

(8) Shachô-ga kaigi-de kuchi-wo kit-ta.

CEO-NOM meeting-LOC mouth-ACC KIRU-PAST ‘TheCEObrokethesilenceatthemeeting.’

SplittingaCONTAINERentailsopeningit.Theexamplesaboveinvolvemetonymicshifts.

(7)isacaseofpart-wholemetonymy,wherebypartoftheLMstandsfortheLM.Thelid

stands for the whole of the CONTAINER (the pot-lid structure), because it is the

substructureof theCONTAINERtheTRdirectly interactswith incausingtheSPLIT. In

Langackarianterms,thelidistheactivezoneoftheLM(Langacker1987:271).Notethat

(7) is somewhatuncommon in that theactive zone takes linguisticprecedence (i.e., is

overtlyrealizedasanargument)overthesuper-structure.Inthemajorityofactivezone

phenomenathesuper-structureisdeemedmoresalientthanthesubstructure/facet:

(9) Weallheardthetrumpet.(instrumentforsound)

(10) Ifinallyblinked.(personforeyelid)

(fromLangacker1987:271)

99

The idiomatic expression in (8) – kuchi-wo kiru – exploits the metonymy ONE ACTION

STANDSFORANOTHERACTION.That is, theactofopeningone’smouthstandsfortheactof

speaking.Thispresupposesmetaphorical construalof themouthas aCONTAINER for

words.SplittingthelipsapartcausestheCONTAINERtoopenandallowsthewordsto

flowout.

FIGURE3:schematicdepictionoffuta-wokiru

11.1.4.Sense(Id):TraversalofNon-solid,UnboundedLM(fig.4)

(11) Jûdan-ga kaze-wo kit-ta.

Bullet-NOM wind-ACC KIRU-PAST ‘Thebulletcutthroughtheair.’

(12) Fune-ga kanâru-no mizu-wo kit-te susun-da.

Ship-NOM canal-LK water-ACC KIRU-TE proceed-PAST

‘Cuttingthroughthewatertheshipadvancedthoughthecanal.’

This sense is available from (Ia) but differs in respect to phase of matter and

boundedness.Here theLM is anunboundedextentof emptyor liquidmatter through

whichtheTRmovesalongalinearpath,resultingintheLM’sspatialsegmentation.This

construalofsegmentationexploits the imageschematransformationmoving0DTR-->

static1DTR,asdescribedbyLakoff(1990:442).I.e.,thepathtakenbytheTRconstitutes

adividinglinethroughtheLM(althoughnopermanentsegmentationisachieved).

FIGURE4

100

11.1.5.Sense(Ie):Disconnection-LMisanAssemblyofFunctionalParts(fig.5)

(13) Tarô-ga enjin-wo kit-ta.

Tarô-NOM engine-ACC KIRU-PAST ‘Tarôturnedofftheengine.’

(14) Hanako-ga terebi-wo kit-ta.

Hanako-NOM TV-ACC KIRU-PAST ‘HanakoturnedofftheTV.’

(15) Tarô-ga dengen-wo kit-ta.

Tarô-NOM powersupply-ACC KIRU-PAST ‘Tarôcutthepowersupply.’

This sense is closely associated with the “turning off” of electrical appliances. Such

appliances rely on theCONTACTof certain components and anuninterrupted flowof

electricitytofunctionproperly.Therefore,(Ie)canbeunderstoodasthecounterpartof

theelicitedeffectsenseofKAKARUdiscussedearlier(see9.1.7.).Considerthefollowing

pair:

(16a) Enjin-ga kakaru

Engine-NOM KAKARU ‘TheEnginecatches(on).’

(16b) Enjin-wo kiru

Engine-ACC KIRU ‘Toturnofftheengine.’

WhereKAKARUcodestheconnectionoffunctionalpartsviatheCONTACTschema,kiru

codestheirdisconnectionviatheSPLITschema.LiketheelicitedeffectsenseofKAKARU,

(Ie) is an extension via the metonymic shift ACTION FOR EFFECT OF ACTION (i.e., the

disconnectionoffunctionalpartsstandsfordisablingtheappliance).

FIGURE5

11.1.6.Sense(If):Disconnection-LMisanAbstractRelation(fig.6)

(17) Hanako-ga Tarô-to-no kankei-wo kit-ta.

Hanako-NOM Tarô-COM-LK ties-ACC KIRU-PAST ‘HanakoseveredtieswithTarô.’

(18) Tarô-ga mae-no jinsei-to en-wo kit-ta.

Tarô-NOM former-LK life-COM relationship-ACC KIRU-PAST

‘Tarôputhisformerlifebehind.’

101

Similar to (Ie), this sense presupposes the LINK schema and then applies the SPLIT

schemato it.However,(If) isarrivedatviaametaphoricalextension,whichconstrues

the target domain of abstract relations in terms of physical connection. As Johnson

(1990: 117) observes, “physical linking is never the full story of humanity, which

requiresacertainnonphysicallinkingtoourparents,oursiblings,andoursocietyasa

whole.”ThisisevidencedbymanymetaphoricalexpressionsexploitingtheLINK/SPLIT

schemas,suchasIseveredtieswithhim,Sheburnedallbridgestoherpast,Icuthimloose,

etc.

FIGURE6

11.1.7.Sense(Ig):TemporalDiscontinuity-LMisanActivity(fig.7)

(19) ‘Sore-wa hitei deki-nai ga...’ to Hanako-wa ii-kake-te, kotoba-wo kit-ta. That-

TOP

denial can-

NEG

CONJ QT Hanako-

TOP

say-KAKERU-

TE

words-

ACC

KIRU-

PAST

‘’Well,thatcan’tbedenied...’saidHanakoandpaused.’

(20) Toriaezu konohen-de shigoto-wo kit-te, ashita tsuzuki-mash-ô.

First aroundhere-LOC work-ACC KIRU-TE tomorrow continue-POL-VOL

‘Let’scallitadayandcontinuetomorrow.’

Due to the homologous structure of the categories space and time (seeTalmy2003a:

47ff.), temporal entities are frequently construed in terms of spatial entities. For

instance, both action and matter are quantifiable amounts. (Ig), in particular, is an

extension via the high-level ontologicalmetaphor ACTIVITY IS AONE-DIMENSIONAL SPATIAL

EXTENT. In other words, the discontinuation of an activity is analogous to the

segmentationofaphysicalobject(see[Ia]).Itisworthyofmentionthat(Ig)mayinclude

expected but unrealized activity. In (19), for example, Hanakomight not continue to

speakatall.Inthiscase,theapplicationoftheSPLITschemastillmakessense,becauseit

isourgeneralunderstandingthatHanakowasexpectedtouttermorethansheactually

did. (For the samereason it is felicitous to sayYouinterruptedmeinthemiddleofthe

sentence,evenifthesentencehasnotbeencompletelyuttered.)

102

FIGURE7

11.1.8.Sense(Ih):Reduction-LMisanAbstractScalarExtent(fig.8)

(21) Tarô-ga kyôsô-de jûbyô-wo kit-ta.

Tarô-NOM race-LOC tenseconds-ACC KIRU-PAST

‘Attherace,Tarôshavedtensecondsofftherecord.’

(22) Sekai-no kiga jinkô-ga hachiokunin-wo kit-ta.

World-LK hunger population-NOM 800millionpeople-ACC KIRU-PAST

‘Thehungeringworldpopulationdroppedbelow800million.’

(23) Kono shôhin-no nedan-ga ichimanen-wo kit-ta.

This product-LK price-NOM 10.000yen-ACC KIRU-PAST

‘Thepriceofthisproductdroppedbelow10.000yen.’

In (Ie), (If), and (Ig) we have seen semantic extensions that emphasize the aspect of

discontinuityordisconnectivity.Thisisespeciallyplainin(If)and(Ig),wheretheSPLIT

schemaisimposedontheLINKschema.

(Ih), like (If) and (Ig), is the result of metaphorical extension. But here, SPLIT is

imposed on the SCALE schema rather than the LINK schema.Whatmakes the SCALE

schemacompatiblewiththeSPLITschemais itsone-dimensionality.Thisfollowsfrom

what Johnson (1990: 122) calls the “fixed directionality” of SCALEs. I.e., the amount,

number, degree, etc.measured by a SCALE is always organized along a single salient

dimension. Examples of this are the vertical UP-DOWN axis underlying theMORE ISUP

metaphor (Johnson 1990: 121f.) and the horizontal LEFT-RIGHT axis used to express

politicalalignment.

Given the above, it follows that the SPLIT schema can be imposed on SCALEs to

metaphorically express the reduction of an abstract extent (such as an amount or

degree).SinceSCALESareconstruedasone-dimensional–andsinceaone-dimensional

objectsplitinmultiplesyieldsseveralone-dimensionalobjectsoflesserlengththanthe

original – a split SCALE will yield at least two parts of lesser length. By way of

illustration,consider (21).TheLMhere is jûbyô (10seconds),ascalarextentwith the

103

value0atonepoleandthevalue10attheother.Anewrecord,letussayat9.8seconds,

willSPLITthisSCALEintotwoparts.Theportionfrom0to9.8replacestheoldSCALEas

the new standard for record attempts, while the portion from 9.8 to 10 becomes

obsoleteforthatpurpose.

Insummary, then, theSPLIT in (21)–and(Ih) ingeneral–divides theLM into two

qualitativelydistinctparts,onerelevantandoneobsolete.By implication,reductionof

theLMisachievedbygettingridoftheobsoletepart.

FIGURE8

11.1.9.Sense(II):FocusonObsoletePortionofLM(fig.9)

(24) Hanako-ga mikan-no kusat-ta bubun-wo kit-ta.

Hanako-NOM mandarineorange-LK rot-PAST part-ACC KIRU-TE

‘Hanakocuttherottenpartfromthemandarineorange.’

(25) Seijika-ga tônai-no hantaiha-wo kir-ôtoshi-ta.

Politician-NOM partyinternal-LK opposition-ACC KIRU-INT-PAST

‘Thepoliticiantriedtogetridoftheparty-internalopposition.’

Itispartofourencyclopedicknowledgethatobjectsareoftendividedforthepurposeof

getting ridof (orotherwise singlingout) some specificportion.As shownabove, (1h)

heavily relies on this implication. In English and German, this cutting off sense is

expressedbyverbparticleconstructionssuchasV-offandab-V:

(26) Thewoundfestered,sohisfoothadtobecutoff.

(27) SiehatsichvondemBrotlaibeinStückabgeschnitten.

Shecutoffsomebreadfromtheloaf(forherself).

Similarly,inJapanese,focusshiftfromtheobjectwhichissplittotheobjectwhichissplit

offisusuallyachievedbyattachingoneofseveralV2s:

(28) Zentai-kara 10cm-wo kiri-hanasu

Whole-ABL 10cm-ACC KIRU-setapart ‘Tocut10cmofffromthewhole’

(29) Sentan-wo kiri-toru

Tip-ACC KIRU-take ‘Tocutthetipoff’

104

11.1.11.Sense(IV):FocusonManner

(32) Tarô-ga handoru-wo migi-ni kit-ta.

Tarô-NOM steeringwheel right-DAT KIRU-PAST

‘Tarôturnedthesteeringwheeltotheright.’

(33) Hanako-ga bôru-wo kit-te, tsuyoi kaiten-wo kake-ta.

Hanako-NOM ball-ACC KIRU-TE strong spin-ACC KAKERU-PAST

‘Hanakohittheballinaslice,puttingastrongspinonit.’

(34) Shinpu-ga jûjika-wo kit-ta.

Priest-NOM cross-ACC KIRU-PAST ‘Thepriestmadethesignofthecross.’

Like (II) and (III), this sense is an extension viametonymic shift. None of the above

objectsareactuallysplitintomultiples.(IV)isnotevenaninstanceoftheSPLITschema.

Itisconnectedtotheothersensesmerelybyexperientialcorrelation,i.e.theknowledge

thatthesplittingofobjectsusuallyinvolvesacertaintypeof“cutting”motion.Here,this

latteraspectcompletelyreplacestheformeraspectwithintheframe,givingrisetothe

newschema[XaffectsYviaacuttingmotion].

Insummary, then,ouranalysisofkiruyields foursenseclustersassociatedwith the

SPLITschema,basedontheirrespectivefocusproperties:(I)focusontheLMasawhole,

(II) focuson theobsoletepartof theLM, (III) focuson thePOS,and (IV) focuson the

mannerinwhichtheTRaffectstheLM.Withthisinmindletusnowturntothesensesof

V-kiru.

11.2.TheSensesofV-kiru

11.2.1.PreviousSuggestions

Over the past decades linguists havemade several suggestions onhow to classify the

senses of V-kiru. The following is a short breakdown of the categorization attempts

basedonSugimura(2008:64ff.).1

Morita[1977](1989)

• Thephysicalcuttingsense:TheV1expressesamannerofcuttingaphysical

objectinmultiples.

• Theperfectivesense(kanryô):-kiruexpressesthattheactionprofiledbytheV1iscarriedoutcompletelyandproperly.

• Theaugmentativesense:-kiruexpressesthattheactionprofiledbytheV1iscarriedoutwithsufficientconfidenceandintensity,precludinganyneed

1Forexamplesentencesseethediscussioninthenextsection.

106

forfurtheraction.

• Thelimitsense:-kiruexpressesthemaxingoutofascalarprocess.

Himeno(1980)

• Thephysicalcuttingsense(setsudan):SeeMorita(1989)above.• Theconclusionsense(shûketsu):-kiruexpressesdeterminationandgoal-

orientednessontheagent’spart;correspondstoMorita’s(1989)

augmentativesense.

• Theaccomplishmentsense(kansui):-kiruexpressesnotonlythecompletionofanaction,butitscompletiontothesatisfactionofitsagent.

• Thelimitsense(kyokudo):RoughlycorrespondstoMorita(1989)above.

Lee(1997)

• Thephysicalcuttingsense(mononosetsudan):SeeMorita(1989)above.• Theaccomplishmentsense(kansui):SeeHimeno(1980)above.• Thelimitsense(kyokugen):SeeHimeno(1980)above.• The“fullofconfidence”sense(jishinmanman):Roughlycorrespondsto

Morita’s(1989)augmentativesense.

• Lexicalizedsenses(goika):InstancesofV-kiruthatarenoteasilyanalyzable

Sugimura(2008)

Sensesthatpreservetheseverance(setsudan)meaningofthesimplexverb:

• Thephysicalcuttingsense(setsudan):SeeMorita(1989)above.• Theconclusionsense(shûketsu):Astateofaffairs(jitai)isdiscontinuedby

theactioncorrespondingtotheV1.

Sensesthatdonotseemtopreservetheseverancemeaningofthesimplexverb;-kirufunctionsasagrammaticalsuffix:

• Theaccomplishmentsense(kôinokansui):-kiruexpressesthattheactionprofiledbytheV1hasbeencarriedoutcompletely.

• Thecompletionofchangesense(henkanotassei):-kiruexpressesthattheprocessofchangeprofiledbytheV1hasreacheditsfinalstate(i.e.non-Fto

max-F).

• Thelimitsense(kyokugenjôtai):-kiruexpressesthemaxingoutofanalreadyongoingscalarprocess(i.e.Ftomax-F).

11.2.2.Discussion

Although theaboveclassificationsarenotentirely inagreementwithoneanother,we

canneverthelessdistill roughly three sensesofV-kiru from them–plus several “odd”

107

items that seem somewhatdifficult to categorize. (The following are onlypreliminary

sketches,amalgamsbasedontheaboveclassifications–notmyfinalsuggestions.)

Thephysicalcuttingsense(mononosetsudan)(35) Nezumi-ga dengenkôdo-wo kami-kit-ta.

Mouse-NOM powercord-ACC bite-KIRU-PAST ‘Themousechewedthroughthepowercord.’

(36) Tarô-ga rôpu-wo tachi-kit-ta.

Tarô-NOM rope-ACC cut-KIRU-PAST ‘Tarôcuttherope.’

(37) Hanako-ga niku-wo tataki-kit-ta.

Hanako-NOM meat-ACC hit-KIRU-PAST ‘Hanakochoppedthemeat.’

TheV1profilesawayofsegmentingaphysicalentity.

Thelimitsense(genkai,kyokudo,kyokugenjôtai):

(38) Kuchi-ga kawaki-kit-teiru.

Mouth-NOM dry-KIRU-RES ‘(My)mouthisalldriedup.’

(39) Sora-ga sumi-kit-teiru.

Sky-NOM becomeclear-KIRU-RES ‘Theskyiscloudless.’

(40) Tarô-ga tsukare-kit-teiru.

Tarô-NOM tire-KIRU-RES ‘Tarôiscompletelyexhausted.’

The V1 is typically intransitive and telic (i.e. goal-oriented). The subject is typically

eithernon-humanorhumanbutnon-intentional.

As Himeno (1980: 29) notes, the V1 often belongs to the domains of natural

phenomena, physiological, emotional, or psychological change. In case of a human

subject, there is a rather strong tendency for the process to be non-intentional and

beyondthesubject’scontrol.

Sugimura (2008:74-76) furtherdistinguishesbetweenprocessesof “limit reaching”

(kyokugen jôtai) and those that indicate “completion of change” (henkano tassei). An

exampleoftheformerwouldbehieru(becomecold):Somethingcanbedescribedusing

theresultativehie-teiru(cool/cold)evenwhentheinherenttelicityscaleoftheprocess

is notmaxedout (i.e.,when it could still get colder). In contrast, one cannot describe

somethingasnaot-teiruunlessthescaleofnaoru(heal)ismaxedout.Thatis,awound

cangraduallyheal,but itcanonlybedescribedasnaot-teiruwhenthereisnofurther

roomforbetterment.Althoughthis isan interestingobservation,oneshouldnotethat

bothvariantshavetheirinherentlytelicnatureincommon.

108

Theaccomplishmentsense(kansui):(41) Tarô-ga gyûdon-no tokumori-wo hitori-de tabe-kit-ta.

Tarô-NOM gyûdon-LK extralargeserving-ACC alone eat-KIRU-PAST

‘Tarôate(up)anextralargeservingofgyûdonallbyhimself.’

(42) Hanako-ga senpêji-no chôhenshôsetsu-wo yomi-kit-ta.

Hanako-NOM thousandpages-LK full-lengthnovel-ACC read-KIRU-PAST

‘Hanakoreadafull-lengthnovelofathousandpages(totheend).’

(43) Tarô-ga sanjûkiro-no kyori-wo hashiri-kit-ta.

Tarô-NOM 30km-LK distance-ACC run-KIRU-PAST ‘Tarôranthe(whole)30kmdistance.’

TheV1istypicallytransitiveandatelic,thesubjectanintentionalhumanagent.Thereis

mostlikelyascalarmodifier–eitherovertorimplicit–thatimposestelicityontheatelic

V1.Forexample,senpêji-nochôhen-shôsetsuservesasatelicmodifierforyomu,whichby

itselfisanon-goal-oriented,potentiallyopen-endedprocess.

As Himeno (1980: 27) notes, the agent usually feels some kind of satisfaction for

havingsuccessfullyhandledaquantifiable“workload”(sagyôryô).

11.2.2.1.SomeRemarksonLimitvsAccomplishment

Whileitmightbepracticaltodistinguishbetweena“limit“and“accomplishment“sense,

the two are best thought of as poles on a continuum rather than as having clear-cut

boundaries.Forexample,Omata(2007:213)pointsout that tsukai-kiru ishardlyever

accompaniedbyafeelingofachievementonpartoftheagent.Consider(44):

(44) Tarô-ga okane-wo tsukai-kit-ta.

Tarô-NOM money-ACC use-KIRU-PAST ‘Tarôusedupallofhismoney.’

Inasimilarvain,shinji-kiru requiresan intentionalhumanagentbutcanoftencarrya

negativeconnotation,asin:

(45) Hanako-ga uso-wo shinji-kit-teiru.

Hanako-NOM lie-ACC believe-KIRU-RES ‘Hanakoisutterlyconvincedofalie.’

Whiletheagentinsuchscenescouldbedescribedasintentional,onemightarguethat

thedegreeofcontrolexercisedinactsofresourceusageorbeliefistypicallylowerthan

inothertypesofactivities,suchasrunning,reading,etc.

Furthermore, Himeno (1980) places moe-kiru in the accomplishment (kansui)

category,statingthattheaccomplishmentsenseexpresses“notonlytheendofanaction,

109

but that the action has been carried out completely (in both quantity and quality) in

accordancewiththeagent’sexpectations”(1980:27).Thisseemsabitoddconsidering

thefollowingsentences:

(46) Tarô-ga yonjûnikiro-no furumarason-wo hashiri-kit-ta.

Tarô-NOM 42km-LK fullmarathon-ACC run-KIRU-PAST

‘Tarôranafullmarathonof42km.’

(47) Hanako-ga saigo-made tatakai-kit-ta.

Hanako-NOM end-ALL fight-KIRU-PAST ‘Hanakofoughttotheend.’

(48) Rôsoku-ga moe-kit-ta.

Candle-NOM burn-KIRU-PAST ‘Thecandleburnedout.’

AccordingtoHimeno,allthreeoftheaboveareaccomplishmentverbs.However,inthe

case ofmoe-kiru, agency, intentionality, and sense of achievement seem to be much

lowerthanintheothertwoexamples.Infact,moeru–denotinganaturalphenomenon–

seemsmoreakin to limit verbs suchashieru(becomecold)orkawaku(becomedry).

Andwhile it canbeargued thatmoeru, unlike theseverbs, isnot inherently telic, it is

typically understood as being temporally bounded. In other words, something might

burn indeterminately long(say,aceremonial fire thatmustnotgoout)–but typically

thereisalimitedburningsubstanceinvolvedthatwillburndownafteracertainamount

oftimehaspassed.

The upshot is that cases like shinji-kiru, tsukai-kiru and moe-kiru are peripheral

exampleslocatedsomewherebetweenthelimitandtheaccomplishmentsense.

Althoughthelimitvsaccomplishmentdistinctionhasitsmerits,Ibelievethat–forthe

purpose of this study – one can make a more useful distinction based on the V1’s

ontological specifications. In his study on the historical development of V-kiru, Aoki

(2004)makesan interestingobservation in this respect.Hepointsout that inherently

telicverbs(genkaidôshi)donotappearasV1untilthelatemiddleages(chûseikôki)and

non-telic verbs (higenkai dôshi) not until earlymodern times (kinsei ikô). In order of

historicalemergenceAoki(2004:39)liststhefollowingsenses:

Aphysicalsegmentation(mono-nosetsudan)Someactionverbs(ichibu-nodôsadôshi)(i-kiru,tachi-kiru,kaki-kiru,...)

A’spatialpartitioning(kûkan-noshadan)Someactionverbs

110

(shi-kiru,tate-kiru,seki-kiru,...)

Bclosure–emphasis(shûketsu–kyôchô)Verbsofthoughtandspeech(hatsuwa-,shikôdôshi)

(ii-kiru,omoi-kiru,furi-kiru,...)

Cextremestate/limit(kyokudo-nojôtai)Verbsofchange,telicverbs(henkadôshi,genkaidôshi)(sumi-kiru,shizumari-kiru,kawaki-kiru,...)

Daccomplishment(dôsa-nokansui)Actionverbs,atelicverbs(dôsadôshi,higenkaidôshi)

As we can see, this classification is not much of a divergence from the rest of the

literature.A is thephysical cutting sense.A’ is an extensionofA fromsolidmatter to

emptymatter.CandDcorrespondtothelimitandaccomplishmentsenses,respectively

(indeed,theyarelabeledassuch).However,Aokimakestheimportantobservationthat

the limit sense involves a telicV1whereas the accomplishment sense involves anon-

telicV1.Hisdistinctionisthereforenotonlyhistoricallysoundbutalsogroundedinthe

V1s configurational structure. And since configurational (i.e. schematic) structure is

what mainly concerns us, we can from here on dismiss the limit vs accomplishment

distinctioninlieuofthetelicvsatelicdistinction.

11.2.2.2.‘Odd’Cases

What to make of Aoki’s category B though? This is the class of “odd cases“ I have

mentioned earlier. Aoki calls this sense closure (shûketsu) and emphasis (kyôchô).

Himeno,too,usesthetermshûketsu todescribe ii-kiru(assert).Morita(1989)andLee

(1997),however,placeii-kiruinaclassof“confidence”verbs(jishin,jishinmanman).As

for furi-kiru (shakeoff, decline) andomoi-kiru(giveup), Lee (1997) categorizes these

intotheirownclassof“lexicalized”verbs(goika).

In conclusion, there seems to be a group of verbs – variously termed “closure”,

“emphasis”, “confidence” or “lexicalized” – which fit neither the cutting, limit or

accomplishment categories. Historically, these verbs follow the cutting sense and

precedethelimitandachievementsenses.

Tosummarize,wehaveidentifiedthreemajorsensesofV-kiru:

• Sense1:TheV1profilesawayofphysicalsegmentation.

• Sense2:TheV1profilesaninherentlygoal-orientedprocess.

111

• Sense3:TheV1profilesanon-goal-orientedprocess.TheLMoftheV1functions

asatelicmodifier.

Additionally,wewillneedtoaddressseveral“odd”caseswhichseemtofitneitherofthe

aboveverywell.Theseinclude:omoi-kiru,ii-kiru,furi-kiru,etc.

Iwillnowproceedtoanalyzethetopologicalstructureofthesesenses.

11.2.3.RevisitingV-kiru:ACategorizationBasedonSchematicTopology

11.2.3.1.Sense(I):TheV1ProfilesaWayofPhysicalSegmentation

(seeexamples35-37)

This is thephysical cutting sensewhichhas alreadybeenadequately characterized in

theliterature.TheLMisanextendedquantityofmatter.Itissplitintomultiplesbythe

TRinthemannerprofiledbytheV1.

11.2.3.2.Sense(IIa):TheV1ProfilesanInherentlyGoal-orientedProcess(“Limit”Sense)

(fig.11)

(seeexamples38-40)

Recall from our discussion of the simplex verb the various implications of the SPLIT

schema, i.e. consider our encyclopedic knowledge about splitting things. Specifically,

recall the caseswhere the location of the POS is non-arbitrary, because itmarks the

divisionbetweenqualitativelydistinctpartsoftheLM.Forexample,wemightcutstalks

ofasparagusalongatender/non-tenderdivide,thenusethetenderpartforcookingand

throwawaythenon-tenderpart.Similarly,whensettingadeadline(seeexample[30])

wedividethetimelineintotwoqualitativelydistinctparts–withthePOSmarkingthe

divisionbetweentimelyandtoolate.

NowconsiderV-kiruinasentencelike(38):

(38) Kuchi-ga kawaki-kit-teiru.

Mouth-NOM dry-KIRU-RES ‘(My)mouthisalldriedup.’

ThequalitativeSPLITisquiteobvious.TheV1kawakuprofilesagoal-orientedprocess.

Letussay0isnotdryatall,5somewhatdry,and10isdevoidofanyliquid.Itisworth

emphasizingthattheV1sparticipatingin(IIa)donotrepresentopenscalesbutclosed

ones. That is, if something is completely devoid of liquid, it is impossible for it to get

drier.Thesameistruefortsukareru(becomeexhausted),sumu(becomeclear),andso

112

forth. It is easy to seehow theendpointof a sucha scale functionsasanaturalPOS,

since itmarks the temporaldivisionbetweenagoal-orientedprocessand the stateof

havingreachedthatgoal.2Thisleavesuswiththefollowingspecifications:

LMspecificationsfor(IIa):TheLMisatimeline(anunboundedextentoftime).TheV1

hasthestructureofaclosedtemporalSCALE(aboundedaxialextentoftime),whichis

imposedon a portionof this timeline. The end-point of the SCALE (i.e. theGOALof a

SOURCE-PATH-GOAL schema) functions as the POS. The LM is thus segmented into a

pre-GOALandapost-GOALportion.Thepre-GOALportion isdynamic: it isabounded

extent of time during which change occurs. The post-GOAL portion is static: it is an

unboundedextentoftimeduringwhichnochangeoccurs.

In other words, the end-point of the scalar process profiled by the V1 divides the

timelineintoadynamicsegmentandasteady-statesegment.

FIGURE11

11.2.3.3.Sense(IIb):TheV1ProfilesaNon-goal-orientedProcess;TheLMoftheV1

FunctionsasaTelicModifier(“Accomplishment”Sense)(fig.12)

Bynowitshouldbeobviousthat–intermsofimageschematictopology–thissenseis

merelyavariantof(IIa).WhiletheV1sparticipatingin(IIa)areinherentlygoal-oriented,

theV1sparticipatingin(IIb)arepotentiallyopen-ended.However,bytakingaLMwith

closedscalestructure,open-endedprocessesbecomegoal-oriented.

LMspecificationsfor(IIb):TheLMisatimeline(anunboundedextentoftime).TheV1

profilesanactivity(anunboundedtemporalquantity).Atelicmodifier(eitherexplicitor

implicit) imposes a SOURCE-PATH-GOAL structure on this activity, yielding a closed

temporalSCALE(aboundedaxialextentof time).This temporalscale is imposedona

2ThisviewisconsistentwithNakashima’s(2006:43ff)suggestionthatthegrammaticalsensesofV-kiruaretheresultofasemanticextensionviatheconceptofdiscontinuity(hirenzokusei).

113

portionof thetimeline.Theend-pointof theSCALE(i.e. theGOALofaSOURCE-PATH-

GOALschema) functionsas thePOS.TheLMis thussegmented intoapre-GOALanda

post-GOAL portion. The pre-GOAL portion is dynamic: it is a bounded extent of time

duringwhichchangeoccurs.Thepost-GOALportionisstatic:itisanunboundedextent

of time during which no change occurs. Again, the end-point of the scalar process

profiled by the V1 divides the timeline into a dynamic segment and a steady-state

segment.

Example:Consider(41)fromabove.

(41) Tarô-ga gyûdon-no tokumori-wo hitori-de tabe-kit-ta.

Tarô-NOM gyûdon-LK extralargeserving-ACC alone eat-KIRU-PAST

‘Tarôate(up)anextralargeservingofgyûdonallbyhimself.’

In this sentencewehaveanactivity tabe(ru) anda telicmodifiergyûdonnotokumori.

Thelatterimposesthescalarstructureof0%depletion–100%depletionontheprocess

ofeating.I.e.,attheSOURCEpointoftheprocessthereis0%depletionofthequantity

profiled by gyûdon no tokumori, while at the GOAL point there is 100% depletion.

Consequently,theGOALfunctionsasPOS,devidingthetimelineintoadynamicsegment

during which depletion occurs and a steady-state segment during which no more

depletionoccurs.

FIGURE12

11.2.3.4.OtherSenses

Whatabouttheaforementioned“odd”cases?Itturnsoutthatmanyofthesearesimply

lexicalized metaphorical extensions of the physical cutting sense. Take the following

usesoffuri-kiru,forexample:

(42) Hanako-ga Tarô-no te-wo furi-kit-te nige-ta.

Hanako-NOM Tarô-LK hand-ACC shake-KIRU-TE run-PAST

‘HanakoshookoffTarô’sgraspandranaway.’

(43) Hanako-ga Tarô-no tanomi-wo furi-kit-ta.

Hanako-NOM Tarô-LK request-ACC shake-KIRU-PAST

‘HanakodeclinedTarô’srequest.’

114

Earlier in this chapter, in our discussion of (If), we have already come across the

conceptualmetaphorABSTRACTRELATIONSAREPHYSICALCONNECTIONS. In a similarmanner,

themetaphoricaluseoffuri-kiruconstruessocialforces(43)intermsofphysicalforces

(42).Bydenyingarequestetc.we“freeourselvesfromthegrasp“ofsocialforces.Itcan

reasonablybesuggestedthatsuchforcedynamicmetaphorsunderlyothercasesaswell:

hari-kiru(stretchout–KIRU) --> workhard,bevigorous

oshi-kiru(push–KIRU) --> forceone’swaythroughsth.

fumi-kiru(treadon–KIRU) --> makeupone’smind,takemeasurestowardssth.

The compounds ii-kiru and omoi-kiru are highly lexicalized and therefore not

straightforwardly analyzable. Nonetheless we can try to give a tentative account in

termsoftheLMspecificationsof-kiru:

(44) Tarô-ga ‘machigai ari-masen!’ to ii-kit-ta.

Tarô-NOM mistake-TOP exist-POL.NEG QT say-KIRU-PAST

‘‘Iamabsolutelypositive!’Tarôsaidwithconfidence.’

LMspecificationsforii-kiru:TheLMisastretchofdiscourse(coveringanextentoftime).

Anutterance(apunctualevent) functionsasPOS.Thesegmentofdiscoursepreceding

the utterance is characterized by uncertainty or dispute regarding the utterance’s

content or the speaker’s belief. The segment of discourse succeeding the utterance is

supposedtobefreefromanysuchuncertainty.

(45) Hanako-ga shôsetsuka-no yume-wo omoi-kit-ta.

Hanako-NOM novelist-LK dream-ACC think-KIRU-PAST

‘Hanakogaveupherdreamofbecominganovelist.’

LMspecificationsforomoi-kiru:TheLMisthecognitivesubject’smentalstate(covering

anextentoftime).ApunctualmentaleventfunctionsasPOS.Thementalstatepreceding

theevent ischaracterizedbybeingaboutacertainentity.Thementalstatesucceeding

theeventischaracterizedbynotbeingaboutthisentity.

Thus,ii-kiruandomoi-kiruareinstancesoftheSPLITschema,inwhichthePOSdivides

theLMintoqualitativelydistinctparts.

115

11.3.RelatedConstructions

Before concluding this chapter, let usbriefly consider some related constructions and

seehowtheyfitinwiththeaboveanalyses.

11.3.1.V-kiri/V-takiri(da)(fig.13)(46) Tarô-ga byôki-de ne-ta kiri da.

Tarô-NOM illness-INS sleep-PAST KIRI COP ‘Tarôisbedriddenwithanillness.’

(47) Hanako-ga ‘shira-nai’ to it-te, damari-kiri dat-ta.

Hanako-NOM know-NEG QT say-TE besilent-KIRI COP-PAST

‘Hanakosaid‘Idon’tknow’andthenremainedsilent.’

(48) Tarô-ga tabi-ni de-ta kiri kaera-nai.

Tarô-NOM journey-DAT leave-PAST KIRI return-NEG

‘Tarôwentonajourney,nevertoreturn.’

(49) ‘Mata kake-naosu’ to iw-are-ta kiri renraku-ga ko-nai. Again call-repeat QT say-PASS-PAST KIRI contact-NOM come-NEG

‘I’llcallagain’Iwastoldbutneverheardback(fromhim/her).’

What is importanttonotehere, is thatourknowledgeabouttheworldequipsuswith

certainexpectationsor mental “scripts”ofhowthingswillnormallyplayout (seee.g.

Schank andAbelson 1977). For example, if you lie down (because you are exhausted

etc.),youwilleventuallygetupagain.Aconversationischaracterizedbyasteadyflowof

utterances. Someone who goes on a journey eventually returns. Someone makes a

promiseandthenkeepsit,andsoon.

InthecaseofV-takiri,however,thesescriptsaredisrupted.ThisiswhyV-takirioften

appears as part of the larger construction V-ta kiri … V-nai, where the second verb

profilesthedefaultcontinuationofagivenmentalscript(48,49).

LMspecificationsforV-takiri(da):TheLMisatimeline(anunboundedextentoftime).

TheV1eitherprofilesanact(apunctualevent)oranactivity(anextendedevent).The

POSiseitherthepuncutaleventorthestartingpointoftheactivity.Thesegmentoftime

preceding the POS is characterized by conforming to the expectations of the

conceptualizer.ThesegmentoftimesucceedingthePOSischaracterizedbydefyingthe

expectationsoftheconceptualizer.

116

FIGURE13

11.3.2.NumeralClassifier+kiri(50) Shinrai dekiru no-wa nijûroku-nin-no uchi tatta shichi-nin kiri da.

Trust can NMLZ-TOP 26-persons-LK among just seven-persons KIRI COP

‘Outofthe26personsameresevencanbetrusted.’

(51) DVDmedia-wa ik-kai kiri-no kaki-komi da.

DVDmedia-TOP one-time KIRI-LK datawriting COP

‘YoucanonlyburndatatoaDVDonce.’

(52) Machi-ni dekake-ta no-wa ni-do kiri da.

Town goout-PAST NMLZ-TOP two-times KIRI COP

‘Ionlywenttothetowntwice(andneveragainsince).’

LM specifications for numerical classifier + kiri: The LM is a numerical ray (an open

SCALE).ThePOSisapointonthisray.OnthesegmentuptoandincludingthePOSall

numerical instancesarerealized.Onthesegmentsucceeding thePOSno instancesare

realized.

11.3.3.kiri-ganai(fig.14)(53) Sonna koto-wo kinishi-tei-tara, kiri-ga nai.

Suchasthat things-ACC worry-PROG-COND KIRI-NOM exist.NEG

‘Ifyouworryaboutsuchthings,therewillbenoendtoit.’

(54) Itsumade mat-temo kiri-ga nai.

Forever wait-evenif KIRI-NOM exist.NEG ‘It’snousewaiting(here)forever.’

(55) jirei-wo kazoe-ageru to kiri-ga nai.

examples-ACC count-raise(enumerate) COND KIRI-NOM exist.NEG

‘(I)couldgoonlistingexamplesforever.’

Thisconstructionconstitutesanother“scriptviolation”.Kiri-ganaiusuallyoccurswithin

thescopeofaconditionalorcounterfactual(e.g.V-tara,V-eba,Vto,V-temo)andmarks

theutteranceitappearsinasawarning:OncecarriedouttheactionprofiledbyVwill–

tothedismayoftheconceptualizer–continueorre-occuradnauseam.

117

Ina“script-conforming”timelineV-ingwouldbefollowedbynon-V-ing.Forexample,

someonewouldcount,finishcounting,andthenbedonecounting.Thatis,theendofV-

ingwoulddividethetimelineintoqualitativelydistinctparts.However,inthescenario

describedbykiri-ganai,V is conceptualized as infinite andno such segmentation can

occur.

FIGURE14

118

12.AGARUandtheUPSchema

The intransitive/transitive pair agaru/ageru can be schematically characterized as

follows:

(A) X-ga Y-wo ageru

XCAUSE YMOVEUP

(B) Y-ga agaru

YMOVEUP

AGARUprototypicallycodestheTR’smovementalongtheverticalaxis.However,aswe

willsee,somesenseshaveabandonedtheverticalityaspectinfavorofdirection-neutral

GOAL-orientedmovement.

12.1.TheSensesofAGARU

12.1.1.Sense(Ia):SpatialAscensionofaZero-dimensionalTR(fig.1)

(1) Taiyô-ga higashi-kara agaru.

Sun-NOM east-ABL AGARU ‘Thesunrisesfromtheeast.’

(2) Hikôki-ga sorataka-ku agat-ta.

Plane-NOM sky-high-INF AGARU-PAST ‘Theplanetooktothesky.’

(3) Tarô-ga yane-no ue-ni agat-ta.

Tarô-NOM roof-LK top-DAT AGARU-PAST ‘Tarôclimbedontotheroof.’

(4) Manshon-no erebêtâ-ga yonkai-made agaru.

Mansion-LK elevator-NOM fourthfloor-ALL AGARU

‘Themansion’selevatorgoesuptothefourthfloor.’

(5) Seijika-ga endan-ni agat-ta.

Politician-NOM podium-DAT AGARU-PAST ‘Thepoliticianassumedthepodium.’

Thissensecorrespondstoapunctual(zero-deminsional)TR’sverticalmotioninspace.

TheTR’smotioniseitherGOAL-directedornot.Inconfigurationslike(1)and(2),there

isnoentityinprofilewhichcouldserveasGOALoftheTR’smotion.Rather,theLMhere

isimplicitandbestthoughtofas“theprojectionontheverticalaxisoftheorderedseries

of spatialpoints that the trajectoroccupies,oftensuccessively through time” (Lindner

1981:148).Sentences(3)-(5),ontheotherhand,profileaLM–markedbyniormade–

towards which the TR’s path of motion is directed, i.e. a GOAL. Note that AGARU is

neutral in respect to the TR’smanner (float, jump, etc.) and path ofmotion (zig-zag,

swerve,etc.)–althoughastraightverticallineisarguablythemostprototypicalPATH.

119

FIGURE1

12.1.2.Sense(Ib):AbstractAscensionofZero-dimensionalTR(SocialAscension)

(6) Hanako-ga kotoshi-no shiken-de jûban agat-ta.

Hanako-NOM thisyear-LK test-INS tenplaces AGARU-PAST

‘Hanakowentuptenplacesinthisyear’sexam.’

(7) Kono jôhô-ga shachô-ni-made agat-ta.

This information-NOM CEO-DAT-ALL AGARU-PAST

‘ThisinformationmadeitswayuptotheCEO.’

(8) Tarô-ga shusse shi-te, takai chii-ni agat-ta.

Tarô-NOM success do-TE high position-DAT AGARU-PAST

‘Tarôhasreachedahighpositioninhiscareer.’

(9) Hanako-ga kodomo-wo ii gakkô-ni ageru to kesshin shi-ta.

Hanako-NOM children-ACC good school-DAT AGERU QT determination do-PAST

‘Hanakoisdeterminedtosendherchildrentoagoodschool.’

Thisisanextensionof(Ia)whichmapsriseinthesocialdomainontoverticalmotionin

space,whilepreservingthezero-dimensionalnatureoftheTR.I.e.,Hanako,jôhô,etc.are

allconceivedofaspunctualentities.Again,thereisaGOAL-directed(7,8,9)andanon-

GOAL-directedvariant(6).

12.1.3.Sense(IIa):SpatialExtensionalongtheVerticalAxis(fig.2)

(10) Entotsu-kara kemuri-ga agat-ta.

Chimney-ABL smoke-NOM AGARU-PAST ‘Smokerosefromthechimney.’

(11) Shio-ga hiza-made agat-te ki-ta.

Tide-NOM knee-ALL AGARU-TE come-PAST ‘Thetidehasrisenknee-high.’

Incontrastto(Ia)and(Ib),thissensedoesnotinvolveachangeoflocationoftheTRin

itsentirety,butrathertheTR’sextensionalongtheverticalaxis.Inotherwords,theTR

successively comes to occupy additional points along the vertical axis without

abandoning theones it isalreadyoccupying.Althoughverticality is certainly themost

120

salientdimensionhereitwouldseemoddtospeakofa“one-dimensional”TR(e.g.,the

TRin[11]isclearlyhorizontallyextendedaswell).

The connectionbetween this sense and (Ia) becomesobviousoncewe consider the

conceptofactivezones:

“Entitiesareoftenmultifaceted,onlycertainfacetsbeingabletointeractwithaparticulardomainorplay

adirectrole ina particularrelationship.Those facetsofanentitycapableof interactingdirectlywitha

givendomainorrelationarereferred toas theactive zoneof theentitywithrespect to thedomainor

relationinquestion.”(Langacker1987:272-73)

I.e.,in(10)and(11)itisonlyafacet(asubstructure)oftheTRthatactuallyundergoesa

changeoflocation.Consequently,ifwefocusourattentiononlyonthetopportionofthe

respectiveTRs(e.g.justthesurfaceinsteadofthewholebodyofwaterin[11]),weend

upwithsense(Ia).However,thewaterandthesmokeasawholearemoresalientthan

their respective active zones – presumably because they are more coherent gestalts.

Thus,thelargerstructuresareelevatedtothelevelofTR,resultinginanimageschema

transformationofthetypepunctualTR-->extendedTR.

Liketheprevioussenses,(IIa)varieswithrespecttoGOAL-orientation.

FIGURE2

12.1.4.Sense(IIb):AbstractExtensionalongtheVerticalAxis

(12) Kion-ga 37-do-made agat-ta.

Temperature-NOM 37degrees-ALL AGARU-PAST. ‘Thetemperaturehasrisento37degrees.’

(13) Kono terebidorama-no ninki-ga agat-te, fan-mo fue-ta.

This TVseries-LK popularity-NOM AGARU-TE fans-also increase-PAST

‘ThisTVserieshasriseninpopularityandgarneredmorefans.’

(14) Tarô-ga kodomo-no seiseki-wo ageru tame-ni kateikyôshi-wo yatot-ta.

Tarô-NOM children-LK grades-ACC AGERU sake-DAT privateteacher hire-PAST

‘Tarôhiredaprivateteacherinordertoimprovehischildren’sgrades.’

(15) Irairasuru to, ketsuatsu-ga agat-teshimau.

Getirritated COND bloodpressure AGARU-IRR

‘Ifyougetirritated,yourbloodpressurewillgoup.’

121

Thissenseisanextensionfrom(IIa)viatheconceptualmetaphorMOREISUP.AsLakoff

and Turner (1989: 83) have pointed out, verticality and quantitative increase are

experientially correlated, since “we constantly encounter cases where an increase in

substance(e.g.,pouringmorewaterinaglass)increasestheheightofthesubstance(e.g.,

thelevelofthewaterintheglass).”Forimageschematictopology,thismeansthatthe

abstractTR(temperature,bloodpressure,etc.)isconceivedofasanobjectsuccessively

extendingalongtheverticaldimension.Quantitativeincreaseisthereforeincompatible

withazero-dimensionalTR(obviouslya0DTRlacksheight).

12.1.5.Sense(III):SubtractiveCompletion(fig.3)

(16) Ame-ga agat-tara, soto-ni de-yô.

Rain-NOM AGARU-when outside-DAT goout-VOL ‘Let’sgooutsideoncetherainhasceased.’

(17) Kuruma-no batterii-ga agat-teshimat-ta.

Car-LK battery-NOM AGARU-IRR-PAST ‘Thecarbatteryisdrained.’

(18) Kono shigoto-ga kotoshi inai-ni agaru to ii nâ.

This work-NOM thisyear within-DAT AGARU COND good EMPH

‘Isurehopewecanfinishthisworkwithintheyear.’

In the above sentencesagaru expresses completion.We find a similar extension from

verticalmotiontocompletioninEnglishandGermanverbparticleconstructions(Vup

andauf-V,respectively).AsLindner(1981)haspointedoutinherdiscussionoftheVup

construction,verticalmotionisoftenGOAL-directed:“[…]mostextensionsfromvertical

UPincorporatewithincreasedsalienceafinal,boundingloacationorstate[…]Themore

salientlythefinallocatonorstatefiguresinthemeaningofUP,themorethislocationor

statemay be thought of as directing or defining the path” (Lindner 1981: 180). This

applies to AGARU as well. As we have seen, senses (I) and (II) may involve GOALs

(markedbyniormade)whichindicatethefinallocationoftheTRortheendpointofthe

TR’s extension. She further notes that “[a]s UP extends into abstract domains, literal

verticalitybecomeslesssalient”(Lindner1981:180).Here,too,weseeaparallel.While

the vertical dimension of AGARU is preserved throughout the mappings in the non-

spatialsenses(Ib)and(IIb),itssalienceiscertainlylowered.Forexample,althoughwe

tendtothinkofa“rise”intemperatureintermsofextensionalongtheverticalaxis,itis

not impossible to conceptualize such quantitative increases in terms of horizontal

extension as well (e.g. form left to right). In contrast, the take-off of a plane etc.

necessarily involves theverticaldimension.Wecan thereforeconclude thatverticality

122

gradually“bleachesout”asAGARUextends intoabstractdomains. (III)seemstomark

theendpointofthisdevelopment:Theverticaldimensionsimplyplaysnorolein(16)-

(18).

In summary, then,wecanattribute theextensionofAGARU fromverticalmotion to

completiontotwomajorsalienceshifts:

• TheheightenedsalienceofGOAL-directedness

• Theloweredsalienceoftheverticaldimension

LetusnowexaminewhatkindofcompletiveprocessAGARUexpresses.Lindner(1981:

194)recognizes“atleasttwowaysaprocesscanreachagoal–byactingontheentire

substanceof itsobjectandbyeffectingasufficientlysalientstatechangein itsobject.”

Toillustrate,considerthefollowingsentences(basedonLindner1981):

(19) Johnateupthesandwich. (congruencebetweenprocessandprocessedregion)

(20) Sallytightenedupthescrew. (achievementofgoalstate)

Themajordifferencebetween(19)and(20) lies inhowtheprocessexpressedbythe

simplex verb acts on its object. As Lindner (1981: 204) points out, processes such as

eatingarespecialinthatthey“measurecompletenessagainsttheamountoftheobject

affected; UP codes the gradual spreading out of the abstract processed region as it

reaches itsgoal,which is the capacityorboundaryor limitsof theobjectaffected.” In

otherwords,onecanmeasuretheprogressofeatingbytrackingthestateofwhateveris

beingeaten.ThesameobservationunderliesDowty’snotionofincrementaltheme:

[…]ifItellmysontomowthelawn(rightnow)andthenlookatthelawnanhourlater,Iwillbeableto

concludesomethingaboutthe“aspect”oftheeventofhismowingthelawnfromtthestateofthelawn,viz.,

thattheeventiseithernotyetbegun,partlydonebutnotfinished,orcompleted,accordingtowhetherthe

grassonthe lawn isall tall,partlyshortorall short.OntheotherhandIwillnotnecessarilybeable to

inspectthestateofmysonandconcludeanythingatallabouthiscompletionofhismowingthelawn.In

thisevent,mysonistheAgentandthelawnistheTheme,infacttheIncrementalTheme.(Dowty1991:

567)

Incontrast,thescrewin(20)isnotanincrementalthemeintheabovesense,sincethere

is no portion of the screw which is gradually consumed or encroached upon by the

processoftightening.

Returning toagaru (note,by theway, that there isno transitivevariantof [III]),we

cannowseethattherelevantentitiesin(16)-(18)–rain,battery,andwork–aresimilar

in their behavior to incremental themes. Take (18) for example: Here, we have a

123

quantifiable amount of a certain (abstract) substance, i.e. work. This substance is

implicitly assumed to be bounded, its boundaries constituting the LM of agaru.

Consequently,agaruprofileswhatLindner(1981:194)callsa“subtractiveprocess”,i.e.

aprocesswhichgraduallyencroachesuponasubstancetowardsitsintrinsicboundaries

– theTRofagaru being theprocessed region. Inotherwords, asweareworking, the

“substanceofwork”graduallydepletesuntil thedepletedregion finallycoincideswith

theinitialworkload.Thesametopologyiseasilyapplicabletothedepletionofabattery,

so(17)isstraightforwardlyaccountedfor.Butwhatabout(16)?Tobesure,rainseems

tobe lacking in incrementality,sincethecorrespondingprocess isnot inherentlytelic,

i.e.goal-orientedlikethedrainingofabattery.Byinspectinghowmuchrainhasfallen

wecannot(atleastnotreliably)tellhowmuchrain“isleft”.However,telicity,wherenot

inherent, is usually supplied by our knowledge that potentiallyopen-endedprocesses

(like working or raining) are not actually endless. Therefore, considering our

encyclopedicknowledge,theschematictopologyofagaruisperfectlycoherentincases

like(16):Sinceraindoesnotcontinueendlessly,itmakessensetoimaginethatthereis

some limited amount of rain (althoughwe do not know howmuch),which gradually

depletes until the depleted amount coincideswith the initial amount (howevermuch

thatmaybe).Viewedinthislight,thequestionoftelicitybecomesanepistemicone.

FIGURE3:ThehatchedareamarkstheunprocessedregionoftheLM.

12.1.6.Sense(IV):Access(fig.4)

(21) Konkai-no torihiki-de hyakumanen-no rieki-ga agat-ta.

Last-LK deal-INS 1.000.000yen-LK profit-NOM AGARU-PAST

‘Thelastdealyieldedaprofitof1.000.000yen.’

(22) Gutaiteki-na shôko-ga agaru-made matsu shikanai.

Concrete-COP.ATT evidence-NOM AGARU-ALL wait nochoicebut

‘Wehavenochoicebuttowaituntilsomeconcreteevidenceemerges.’

(23) Kôhosha toshite kare-no namae-ga agat-teiru.

Candidate as he-LK name-NOM AGARU-RES

‘Hisnamepoppedupforthecandidacy.’

124

(24) Betsu-no rei-wo age-te kudasai.

Different-LK example-ACC AGERU-IMP please ‘Pleasegiveadifferentexample.’

Wehaveaccesstotheworldthroughourperceptivefaculties–oursenseofvisionbeing

of paramount importance. However, the structure of our bodies (assuming a

prototypical upright pose) limits the scope of our visual field.What is on the ground,

belowthewaist,etc.isoutsidethefieldofvision(again,assumingaprototypicalpose)

andcannotbeaccessedunlessweeitherlookdownortheentityinquestioniselevated

to the lineofvisualaccess.1Therefore, ifwewantsomeonetoconsidersomething,we

needto“bringitup“.Thisconstrualofaccessintermsofverticalelevationsubtlydiffers

fromthealternateconstrual in termsofCONTAINMENTwhichIhavediscussed inthe

studyofDERU:InthecaseofAGARUthereisnoparticularobstacle(e.g.aCONTAINER)

which is blocking theTR frombeing accessed. In this respect, the differencebetween

DERU and AGARU corresponds to the difference between out and up in English verb

particleconstructions.AsLindnerputsit:

OUTprofilesasitsLMitstrajector’soriginalprivateorconcealedstate,whereasUPcodesitstrajector’s

comingintoviewwithnosalientprevioushistoryorsource.Compare100peopleturnedoutfortheparty

to100peopleturnedupfortheparty.The formerhasa feelingthat thepeoplecamefromtheprivacyof

their homes into the domain of possible interaction with others. The latter suggests that they simply

appeared,withthefeelingthattheymayhavecomespontaneouslyoffthestreet.(Lindner1982:319)

Considerthefollowingpairs:

(25a) Tarô-ga shôko-wo dashi-ta.

Tarô-NOM evidence-ACC DASU-PAST

(25b) Tarô-ga shôko-wo age-ta.

Tarô-NOM evidence-ACC AGERU-PAST

‘Tarôgaveevidence.’

(26a) Tarô-ga jirei-wo dashi-ta.

Tarô-NOM example-ACC DASU-PAST

(26b) Tarô-ga jirei-wo age-ta.

Tarô-NOM example-ACC AGERU-PAST

‘Tarôgaveanexample.’

1SeeLindner(1981:163):“Fromourexperience,weknowthatwhenanobjectislow,itoftenisamongotherobjectswhichobscureitfromview;whenitisprone,wearelikelytooverlookit.Anobjectthatis

highappearslarger,isunobstructedandeasiertosee.”

125

Both(a)and(b)haveincommonthatTarômakestheTRaccessible,butdifferinhow

thisisachieved.Inthe(a)versionstheevidenceandtheexampleareconstruedasbeing

broughtoutofan initiallyconcealedstate. Incontrast, the(b)versionsconstrue these

entitiesas“public”.Inthesecasestheyaremadeaccessiblemerelybypointingthemout,

i.e.bydirectingsomeoneelse’sattentiontowardswhatisalreadyinthepublicdomain.

FIGURE4

12.2.TheSensesofV-AGARU

12.2.1.SpatialAscensionofaZero-dimensionalTR

(27) Sore-wo kiku to Tarô-ga tobi-agat-te, yorokon-da.

That-ACC hear when Tarô-NOM jump-AGARU-TE rejoice-PAST

‘WhenTarôheardthat,hejumpedforjoy.’

(28) Hanako-ga jûwaki-wo tori-age-ta.

Hanako-NOM receiver-ACC take-AGERU-PAST ‘Hanakotookupthereceiver.’

(29) Tarô-ga bôru-wo taka-ku keri-age-ta.

Tarô-NOM ball-ACC high-INF kick-AGERU-PAST ‘Tarôkickedtheballhigh.’

Thisusagecorrespondstosense(Ia)ofthesimplexverb.TheV1expressesthemanner

or cause of the TR’s ascension. Analogous senses of the V-up and German auf-V

constructionsaregivenbelow:

(30) Johnpickedupthereceiver.

(31) Therocketshotupintotheair.

(32) Maryclimbedupthewall.

(33) HanshobdenBallvomBodenauf. (aufheben)

Hanspickeduptheballfromthefloor.

(34) DieRaketeschossindenHimmelauf. (aufschießen)

Therocketshotupintothesky.

(35) AnnascheuchtedieTaubenauf. (aufscheuchen)

Annascaredupthepigeons

126

Note that themetaphor ELEVATIONTOHANDLEVEL IS AVAILABILITY FORUSE/ACTION (Lindner

1981:161)appliescross-linguistically:

(36) Iinkai-ga kaiin-no teian-wo tori-age-ta.

Committee-NOM member-LK proposal-ACC take-AGERU-PAST

‘Thecommitteetookupthemember’sproposal.’

(37) Johntookuptennisasahobby.

(38) DerPhysikerhatseinealteTheoriewiederaufgegriffen.(aufgreifen)

‘Thephysicisttookuphisformertheoryagain.’

AsHimeno(1976:101)pointsout,someusesofV-agerudonotinvolvetheascensionof

aconcreteobjectbutratherindicatetheverticaldirectionoftheprocessitself:

(39) Tarô-ga sora-wo mi-age-ta.

Tarô-NOM sky-ACC look-AGERU-PAST ‘Tarôgazedatthesky.’

TheTRin(39)isTarô’sgazeitself,reifiedasthevectorofhislineofvision.Sincealineis

obviouslyone-dimensional,itseemscounterintuitivetospeakofazero-dimensionalTR

in this case.However, the relevantmotion isnot extensionalong thevertical axisbut

ascensionof the individualpoints(zero-dimensionalentities)whichtogethermakeup

hislineofvision.Intermsofschematictopology,thisissimilartothecaseofflipupthe

lever,wheretheendoftheleverservesasactivezoneforup(Lindner1981:153).The

difference between a lever and a line of sight is, of course, that the line of sight is

unbounded (i.e., there is no single point corresponding to the end of the lever).

Consequently,theroleofactivezoneisassumedbytheentiretyofpointsonthelineof

sight, rather than one salient point in particular (although one could single out a

particularpointatrandom).

Asfarasargumentstructureisconcerned,notethat(39)canroughlybeparaphrased

as:

(40) Tarô-ga shisen-wo age-te, sora-wo mita.

Tarô-NOM gaze-ACC ageru-TE sky-ACC look ‘Tarôdirectedhisgazeupwardstothesky.’

Incontrastto(27)-(29),theentitywhichundergoesascensionin(39)isquiteabstract

and thereforenotsalientenough toberealizedasanovertnominal.Thisexplains the

infelicityof*Tarô-gasora-woage-ta.

127

12.2.2.AbstractAscensionofaZero-dimensionalTR(SocialAscension)

(41) Fuka-ku o-rei-wo môshi-age-masu.

Deep-INF HON-thank-ACC speak.HUM-AGERU-POL ‘Ithankyoudeeply.’

(42) Seifu-ga nômin-kara kome-wo kai-age-ta.

Government-NOM farmers-ABL rice-ACC buy-AGERU-PAST

‘Thegovernmentboughtricefromthefarmers.’

(43) Shachô-ga hikui mibun-kara nari-agat-ta.

CEO-NOM low status-ABL become-AGARU-PAST

‘TheCEOmadehiswayupfromamodestbackground.’

Thissenseisanalogoustosense(Ib)ofthesimplexverbinthatitmapsverticalmotion

inspaceonto thesocialdomain.Tobesure,entitiessuchaskome (rice)arenotzero-

dimensional per se (i.e., rice is an unbounded mass). However, kau (buy) evokes a

commercial frame in which bounded quantities are exchanged. Secondly, on a more

generalnote,AGARUintheabovesentencesentailschangeofplaceinthesocialdomain

–whichinturnrequiresthattheTRbeconstruedasaboundedmovingentitywithno

salient dimensions. Therefore “ascension of 0DTR” is to be understood as follows:

AlthoughtheTRmayhavemorethanzerodimensions,noneofthemaresalientinterms

of image schematic topology. Therein lies the contrast to spatial extension, which

requirespreciselyonesalientdimension.

Since structuring the social domain along the vertical axis is cross-culturally

widespread(althoughIwouldnotventuretosay“universal”),itishardlysurprisingto

findparallelconstructionsinEnglishandGerman:

(44) Johnclimbeduptheladdertovicepresident.

(45) HansistbiszumVizepräsidentenaufgestiegen.(aufsteigen)

Hansmadeittovicepresident.

12.2.3.GOAL-orientedSpatialMovement(BleachedVerticality)(fig.5)

(46) Nibanme-no uma-ga kyûsoku-ni oi-age-te ki-ta.

Second-LK horse-NOM rapidly chase-AGERU-TE come-past

‘Thehorseinsecondplacecaughtuprapidly.’

Recall from (III) above that heightened salience of GOAL-directedness can eventually

“bleachout”theverticalityaspectofAGARU.Thisappearstobethecasein(46),where

oi-ageru profiles GOAL-directed motion along the horizontal axis. Note that the

128

phenomenon of bleached verticality applies to English and German particle verbs as

well:

(47) Amanwalkeduptomeandaskedforacigarette.

(48) Johnmetupwithhissisterinacoffeeshop.

(49) DerLäuferhatwiederzurGruppeaufgeschlossen.(aufschließen)

Therunnercaughtupwiththefieldagain.

(50) DerLastwagenfahreristzudichtaufgefahren.(auffahren)

Thetruckdriverdrovetoclosetothecarinfront.

Althoughatfirstglancetheverticaldimensionseemstoplaynoroleinanyoftheabove

scenes, Lindner (1981: 181) suggests that GOAL-directedness may be rooted in

verticalityafterall,pointingoutthatentitiessubjectivelyincreasealongtheverticalaxis

asdistancefromtheconceptualizerdecreases.Considerthefollowingsentencepair:

(51) Marywalkeduptome.

(52) MarywalkeduptoJohn.

In (51) the TR of walked up (Mary) moves closer towards the observer, thereby

occupyingmorespace in thevisual field–with themostsalient increase insizebeing

the one along the vertical axis. As GOAL of Mary’s PATH, the observer perceives an

increasedheightintheTR.In(52),ontheotherhand,nosuchincreaseisperceivedby

the observer directly. Here the experience of increased height can only be obtained

“virtually”,i.e.bytheobservermentallyputtinghimselfinJohn’splace.Naturally,then,

this experience of vertical increase becomes less and less salient where GOAL and

conceptualizerdonotcoincide.

FIGURE5

However, returning to (46), it is worth noting that oi-ageru seems to be the only

instance of V-AGARU indicating non-vertical GOAL-orientedmotion in physical space.

Furthermore,oi-ageruhasasecondusedifferentfromtheonein(46):

129

(53) Inu-ga hitsuji-wo oka-no ue-ni oi-age-ta.

Dog-NOM sheep-ACC hill-LK top-DAT chase-AGERU-PAST

‘Thedogchasedthesheepupthehill.’

Intheabovesentenceoi-ageruindicatescausedmotionintoanupwardsdirection.The

verticalityaspectisfullyintacthere.Consideringthis,itispossiblethattheusagein(46)

has developed by invited inferencing from the one in (53), i.e. from chaseXY intoan

upwardsdirectiontosimplycatchuptoXY.Inotherwords,themeaningextensioncould

be more contextual than conceptual in nature. In fact, this seems to be the most

plausible account, since it would also explain why the transitive form -ageru is used

insteadof-agaru.

Neverthelessitwasworthpointingouttheconnectionbetweenincreaseinhightand

GOAL-orientedmotion–firstly,forthesakeofcross-linguisticcomparisonandsecondly,

becauseitmaypartlyexplainthefollowingsenseofV-AGARU.

12.2.4.GOAL-orientedNon-spatialMovement(BleachedVerticality)(fig.7)

(54) Tarô-ga yotei-wo isshûkan kuri-age-ta.

Tarô-NOM plans-ACC oneweek reelin-AGERU-PAST ‘Tarômoveduphisplansbyaweek.’

Theverbkuruinitsoriginalspatialsenseprofilescausedmotionofathin,linearobject

towardstheconceptualizer:

(55) Ito-wo kuru

Thread-ACC coilup ‘Tocoilupathread’

Whilesuchmotionprototypically involveselevationtohandlevel, itwillalsoaffectthe

shapeofanobject, i.e. a threadofwoolwill take theshapeofball, a linewill take the

shape of a coil, and so forth. In these cases GOAL-directed motion (towards the

conceptualizer) coincides with an increase along the salient vertical axis (see fig. 6).

Bothaspectsarepreservedthroughoutthemetaphoricalmapping:

• Theentityis“elevatedtohandlevel“andthereforemovedintothedomainof

actionandactuality(cf.TheymovedupthegrandopeningtoJanuary).

• Theentity“increasesinsize“,andthereforeimportance,asitbecomesless

distantfromtheconceptualizer.

130

FIGURE6:schematicdepictionofito-wokuru

FIGURE7:schematicdepictionofyotei-wokuri-ageru

12.2.5.SpatialExtensionalongtheVerticalAxis

(56) Tarô-ga hako-wo tsumi-age-ta.

Tarô-NOM boxes-ACC pile-AGERU-PAST ‘Tarôpileduptheboxes.’

(57) Hanako-ga nobi-agat-te, hon-wo tor-ôtoshi-ta.

Hanako-NOM stretch-AGARU-TE book-ACC take-INT-PAST

‘Hanakostretchedandtriedtotakethebook.’

(58) Tarô-ga isu-kara tachi-agat-ta.

Tarô-NOM chair-ABL stand-AGARU-PAST ‘Tarôstoodupfromthechair.’

Thissensecorrespondsto(IIa)ofthesimplexverb.Here,-AGARUencodesvariousways

inwhich an entitymay extend along the vertical axiswithout undergoing a complete

change of location. In (56), for instance, several uniplex entities (discrete boxes) are

piledontooneanother,resultingintheverticalextensionofamultiplexentity(thepile).

(57) and (58), on the other hand, encode vertical extensionof theTR into anupright

position.

Note that canonical standing posture is commonly associated with readiness for

activity,givingrisetothemetaphorREADYISUP:

(59) Tarô-ga chôsen-ni tachi-agat-ta.

Tarô-NOM challenge-DAT stand-AGARU-PAST ‘Tarôstooduptothechallenge.’

131

(60) Pasokon-ga tachi-agat-ta.

PC-NOM stand-AGARU-PAST ‘ThePCbootedup.’

ItisthusunsurprisingthatweshouldfindsimilarexpressionsinEnglishandGerman:

(61) Isanyoneupforsomeicecream?

(62) SeineFreundehabenihnwiederaufgerichtet.(aufrichten)

Hisfriendsliftedhisspiritsagain.

12.2.6.AbstractExtensionalongtheVerticalAxis

(63) Korekutâ-tachi-ga ôkushon-de kaiga-no nedan-wo seri-age-ta.

Collector-PL-NOM auction-INS painting-LK price-ACC makebid-AGERU-PAST

‘Thecollectorsbidupthepainting’spriceattheauction.’

Thiscorrespondstosense(IIb)ofthesimplexverb.I.e.,theabstractTRisconceivedof

asanobjectsuccessivelyextendingalongtheverticaldimension.

12.2.7.MultidimensionalSpatialExtension(fig.8)

(64) Ashi-ga hari-agat-teiru.

Foot-NOM swell-AGARU-RES ‘Thefootisswollenup.’

(65) Zenshin-no kekkan-ga fukure-agari, hageshi-ku myaku-wo ut-ta.

Wholebody bloodvessel swell-AGARU.CONJ intense-INF pulse-ACC beat-PAST

‘Thebloodvesselsinhis/herbodyswelledupandtheheartbeatintensely.’

AspointedoutbyLindner(1981:152),verticalextensionissometimesaccompaniedby

a simultaneous extension into other dimensions. Put in anotherway, when an object

grows in size, its increase along the vertical dimension is typically the most salient

change.Consequently,verticalextensioncanstandmetonymicallyforoverallgrowth:

(66) Myfootisallswollenup.

(67) DieKrötehatsichaufgebläht.(aufblähen)

Thetoadblewitselfupintoabloatedstate.

FIGURE8

132

12.2.8.VerticalEncroachment(fig.9)

(68) Hanako-ga Tarô-no kami-wo kari-age-ta.

Hanako-NOM Tarô-LK hair-ACC trim-AGERU-PAST

‘HanakocroppedthehairatthebackofTarô’shead.’

(69) Atama-ga hage-agat-teiru node, fuke-te mieru.

Head-NOM bald-AGARU-RES because growold-TE look

‘Becausehishairhasreceeded,helooksold.’

Wehavealreadydiscussedsomesimilarprocessesunder(III).Theaboveexamples(68)

and(69)featurearegionmadeupofsomesubstance(inthiscasehair),whichisthen

subsequentlyencroacheduponalongtheverticaldimensionbytheprocessencodedby

the V1. As such, this sense bears a strong family resemblance to what I have called

substractivecompletion.However,itdiffersfromthelatterinthatitactuallyinvolvesthe

verticaldimensioninphysicalspace.

FIGURE9:ThehatchedareamarkstheunprocessedregionoftheLM.

12.2.9.Completion1:SubtractiveCompletion

(70) Arubamu-wa hyakumanmai-wo uri-age-ta.

Album-TOP 1.000.000units-ACC sell-AGERU-PAST ‘Thealbumsold1.000.000units.’

(71) Tarô-ga isshûkan kake-te, hon-wo yomi-age-ta.

Tarô-NOM oneweek spend-TE book-ACC read-AGERU-PAST

‘Tarôfinishedthebookwithinaweek.’

(72) Hanako-ga Tarô-no ketten-wo kazoe-age-ta.

Hanako-NOM Tarô-LK shortcomings-ACC count-AGERU-PAST

‘HanakoenumeratedTarô’sshortcomings.’

This sense is analogous to sense (III) of the simplex verb. Here, too, we have some

abstract region which is gradually being processed until the processed region

completelycoincideswiththeinitialunprocessedregion.Recallfromourdiscussionof

(III) thattheexactcapacityofthe initialregionisnotalwaysknownfromtheonsetof

the process. For example, (70) does not necessarily entail that the album sold out

completely.Whatitmeansisthatonemillionunitsmakeupthefinalsalesfigure,afixed

133

amount which is then (i.e. after the fact) construed as the capacity of the initial

unprocessedregion.Inothercases,suchas(71),itisnaturaltoassumethattheregion’s

capacityisknownbeforeitsbeingprocessed(i.e.,itiseasyforTarôtoknowhowmany

pages his book consists of and to track his reading progress vis-a-vis the book’s

thickness).

As for the topic of incrementality:The entities in (70)-(72)behave like incremental

themes insofar as progress is measured against howmuch of the entity’s region has

been processed. On the other hand,we are not necessarily able to tell the processed

amountbyinspectingtheentity.I.e.,ifthereisnobookmarkinTarô’sbookwewillnot

be able to deduce his reading progress, since reading does not cause any perceptible

changeofstateintheaffectedobject(unlikemowingoreating).

BelowaresomecorrespondingexamplesfromEnglishandGerman:

(73a) Heateupthewholepizza.

(73b) ErhatdieganzePizzaaufgegessen.(aufessen)

(sameas73a)

(74a) Sheusedupallhermoney.

(74b) SiehatihrganzesGeldaufgebraucht.(aufbrauchen)

(sameas73a)

(75) HanshatdasGedichtaufgesagt.(aufsagen)

Hansrecitedthepoem(completely).

(76) HelgahatallePräsidentenaufgezählt.(aufzählen)

Helgarecited(thenamesof)allthepresidents.

It is interesting to note that the German verbs in (75) and (76) not only imply

completive aspect but at the same time public accessibility. I.e., there seems to be a

conflationofthecompletiveandaccesssenses,bothrootedinthespatialmeaningofthe

prepositionauf.

On a related sidenote, all German examples exhibit some leniency in respect to

completiveaspect:

(77) HanshatdiePizzanurzurHälfte/halbaufgegessen.

Hansateonlyhalfthepizza.

(78) HelgahatdasGedichtnurzueinemDrittelaufgesagt.

Helgaonlyrecitedathirdofthepoem.

Incontrast,thefollowingstrikeusasveryodd,ifnotoutrightinfelicitous:

134

(79) (??) Johnateuponlyhalfofthepizza.

(80) *Tarô-ga hon-wo tochû-made yomi-age-ta.

Tarô-NOM book-ACC halfway read-AGERU-PAST

Itseemsthatsentence(77)construesthecompleteconsumptionofthepizzaaspartof

thenon-actualdomainand itspartial consumptionaspartof therealworld.Sentence

(79),ontheotherhand,construesbothcompleteconsumptionandpartialconsumption

asactual,resultinginaparadoxicalstateofaffairs.Thesameappliesmutatismutandisto

(78)and(80), respectively.Theexactreason for this isunclear.Wemightsuspect the

semanticsofauf-orzurHälfte/halborsomecombinationthereoftoberesponsiblefor

thisphenomenon.

12.2.10.Completion2:AchievementofSufficientState(fig.10)

(81) Men-ga yude-agat-tara, utsuwa-ni moru.

Noodles-NOM cook-AGARU-when bowl-DAT pile.

‘Oncethenoodlesarecooked,putthemintoabowl.’

(82) Tarô-ga karada-wo kitae-age-ta.

Tarô-NOM body-ACC train-AGERU-PAST ‘Tarôbuiltuphisbody.’

(83) Isha-ga kanja-no jôtai-wo tetteiteki-ni shirabe-age-ta.

Doctor-NOM patient-LK condition-ACC thoroughly examine-AGERU-PAST

‘Thedoctorexaminedthepatient’sconditionthoroughly.’

This sense is related to the preceding one in that V-AGARU codes a GOAL-directed

process. Here, however, progress is not achieved in a manner of encroaching

consumption.Instead,theGOALisdefinedassomesalientstateonavaluescale.In(81)

this is the state of being sufficiently cooked2, in (82) presumably the state of being

sufficientlyfit,andin(83)thestateofbeingsufficientlycertain.AsimpliedbyLindner,it

isplausibletothinkofthesufficientstatesenseasanextensionviathemetaphorMOREIS

UP: „Scalar organization immediately calls tomind an extensionof verticalUP-1 – the

brightersomethingis,thehigheritsstateisonthescaleofbrightness.Byvirtueofthis

extension of verticality, UP will code any increase in degree, that is, any positive

increment along a given scale“ (Lindner 1981: 204). Note that the GOAL state is not

necessarilyabsolute–whatisdeemed“sufficient“mayvaryaccordingtopersonaltaste,

needs, or from situation to situation (e.g., some like their noodlesmorealdente than

others; see [81]). In contrast, the substractive completion sense will not admit to the

2AsthedatainHimeno(1976)shows,verbsoffoodpreparationareprototypicalforthissense.

135

setting of ad hoc GOALS (e.g., it is non-debatable at which point a book is read

completely).

FIGURE10

We find an interesting subtype of the sufficient state sense in German, where auf-

codestherestorationofanentitytoitsformerdesirablestate(seeGüler1986:96):

(84) ErhatdieSuppeaufgewärmt.(aufwärmen)

Hereheatedthesoup.

(85) SiehatdasSofaaufgepolstert.(aufpolstern)

Shereupholsteredthesofa.

(86) IchmussdieBatterieaufladen.

Ihavetorechargethebattery.

Nowrecall(17)fromabove:

(87) Kuruma-no batterii-ga agat-teshimat-ta.

Car-LK battery-NOM AGARU-IRR-PAST ‘Thecar’sbatteryisdrained.’

Since the two completion senses may code GOAL-directed processes into opposite

directions,sentenceslikethefollowingareeasilyaccountedfor:

(88) DieBatterieistaufgebrauchtundmusswiederaufgeladenwerden.

Thebatteryisusedupandneedstobechargedupagain.

Here aufbrauchen codes the battery’s depletion process in terms of substractive

completion,whileaufladencodesit’srestorationtofullcapacityintermsofthesufficient

statesense.

12.2.11.SomeNotesonReflexiveTRs

As Lakoff (1990b: 430) has pointed out, Lindner’s 1981 study is renowned for the

discoveryofreflexivetrajectors.Sincetheaboveanalysishasshownthatthesensesof

Japanese V-AGARU and English V up (as well as German auf-V) are often similarly

136

motivated in termsof image schematic topology,wemight expect to find instancesof

reflexive-AGARU.Arethereany?

Considerthefollowingexamplesofreflexiveupandauf-:

(89) Maryrolledupthecarpet.

(90) MariehatdenTeppichaufgerollt.(aufrollen)

(sameas89)

ReflexiveTRs are special in that “the trajector andLMare identifiedwith eachother,

thatis,thetrajectoryoftheobjectisdefinedrelativetotheobjectitself”(Lindner1981:

186).I.e.,intheaboveexamples“eachsubpartservesasbothtrajectorandLMtoother

subparts, which amounts to a change in the object’s shape.When each of an object’s

subparts has other subparts as its goal of appraoch, the object becomes compact, its

subpartsmoretightlyintegrated”(Lindner1981:186).

Interestingly,Himeno(see1976:103)listssomeinstancesofV-AGARUthatappearto

codeasimilarchangeofshape,namelya“reductioninform”(katachinoshukushô):

(91) Tarô-ga rôrukâten-wo maki-age-ta.

Tarô-NOM rollershade-ACC roll-AGERU-PAST ‘Tarôrolleduptherollershade.’

(92) Hanko-ga sode-wo makuri-age-ta.

Hanako-NOM sleeve-ACC roll-AGERU-PAST ‘Hanakorolleduphersleeve.’

(93) Tarô-ga kami-wo ori-age-ta.

Tarô-NOM paper-ACC fold-AGERU-PAST ‘Tarôfoldedupthepaper.’

Althoughtheentities inallof thesescenesdobecomemorecompactasaresultof the

processcodedby-AGARU,Iwouldhesitatetocallthem“true”reflexiveTRsinLindner’s

sense.That is, in(91)-(93)theLMseemstobethegeneralverticalPATHtakenbythe

entity’sactivezonerather thantheentity itself. In (91), forexample, there isasalient

subpart–thelowerendofthecurtain–whichmovesintoanupwarddirection.Incases

like (92), where the active zone is salient enough to assume the role of grammatical

object,thismetonymicshiftisevenmoreobvious:

(92a) Hanko-ga sode-wo makuri-age-ta.

Hanako-NOM sleeve-ACC roll-AGERU-PAST ‘Hanakorolleduphersleeve.’

(92b) Hanko-ga sode-no suso-wo makuri-age-ta.

Hanako-NOM sleeve-LK cuff-ACC roll-AGERU-PAST

‘Hanakorolledupthecuffofhersleeve.’

137

Ithereforesuggesttocategorize(91)-(93)ascasesofspatialascension,exhibitingapart

-->wholemetonymicshift.

138

13.TÔRUandthePATHTRAVERSALSchema

Theintransitive/transitivepairtôru/tôsucanbeschematicallycharacterized

asfollows:

(A) X-ga Y-wo Z-ni tôsu XCAUSE YTRAVERSE Z(PATH) (B) Y-ga Z-wo tôru YTRAVERSE Z(PATH)

TRAVERSAL is the relation between a moving entity and its PATH, leading from a

SOURCE to a GOAL. In other words, the conceptual content of TÔRU is roughly

equivalentwiththenotionofmovementalongaterminalPATH.NotethatTRAVERSAL

islessspecificthanthePENETRATIONimageschema,whichcharacterizessemantically

related prepositions such as English through and German durch. For instance, the

TRAVERSAL of a flat, 2-dimensional “floor” surface – compatible with TÔRU, but not

withthroughordurch– isnotan instanceofPENETRATION.Thiswillbediscussed in

moredetailbelow.

13.1.TheSensesofTÔRU13.1.1.Sense(Ia):LMisaVolumeinPhysicalSpace(fig.1)(1) Tarô-ga rôka-wo tôt-ta. Tarô-NOM corridor-ACC TÔRU-PAST ‘Tarôwentthroughthecorridor.’

(2) Hanako-ga shatsu-no sode-ni te-wo tôshi-ta. Hanako-NOM shirt-LK sleeve-DAT hand-ACC TÔSU-PAST ‘Hanakoputherhandthroughtheshirtsleeve.’

(3) Tankensha-ga fukai mori-ya numa-wo tôt-ta. Explorer-NOM deep forests-and swamps-ACC TÔRU-PAST ‘Theexplorermadehiswaythroughdeepforestsandswamps.’

(4) Kaibô-de dangan-ga kanzô-wo tôt-ta koto-ga wakat-ta. Autopsy-INS bullet-NOM liver-ACC TÔRU-PAST NMLZ-NOM understand-PAST ‘Theautopsyrevealedthatthebulletpiercedtheliver.’

(5) Tarô-ga nohara-wo tôt-te, machi-e mukat-ta. Tarô-NOM field-ACC TÔRU-TE town-ALL headfor-PAST ‘Tarôcrossedthefieldandheadedforthetown.’

This sense profiles the PATH of the TR through a three-dimensional volume in the

spatial domain. Note the gradual differences along the parameters of enclosure and

139

phaseofmatter.Thetube-likeLMsin(1)and(2)areverticallyboundedandconstitute

prototypicalcasesofthree-dimensionalPATHs.TheLMin(3) isunboundedatthetop

andthereforelacksfullenclosure.However,theactivezoneactuallybeingtraversedstill

encompassestheTRalongallthreedimensions,sincethetreesexceedtheTRinheight.

Thisisnotthecasein(5):WhiletheLMisnotexactlyaflattwo-dimensionalsurface–

wecanimaginethegroundcoveredbyshrubsandgrassesofvaryingsizes–ahumanTR

willtypicallyexceeditinheight.Itisworthpointingoutthatthisratioofheightbetween

LMandTR constitutes a cut-off point for the categoriesdescribedbyEnglish through

andGermandurch,whichpartiallyoverlapwithTÔRU.

(6) Annaranthroughtheforest.(7) Annaranthrough/??overthecornfield.(8) Annaranover/*throughthelawn.

(9) AnnaliefdurchdenWald.(10) Annaliefdurch/??überdasKornfeld.(11) Annaliefüber/*durchdenRasen. (basedonKaufmann1993:227)

As these examples show, through/durch in the domain of three-dimensional space

requirestheLMtobeofsufficientheightvis-a-vistheTR.AsKaufmann(1993:227)puts

it, “a two-dimensional object such as a lawn cannot include a three-dimensional one.”

What consitutes “sufficient height” is, of course, amatter of degree. In regards to the

similar caseof in(to) vson(to)Hawkins (1988:254)observes fuzzy cases suchas the

following:

(12) SheplacedMaryJane’sdrinksecurelyinitscoaster

“When confronted with this particular example, one native speaker of English (not

myself!)notedthattheprepositioninisappropriateonlyifthecoasterhasaperceivable

‘lip’.[…]Withoutsuchalip,theappropriateprepositionwouldbeon”(Hawkins1988:

254).Wewillseeinthenextsection,however,thattheheightcriterionisnotcrucialin

thecaseofTÔRU,i.e.thatTÔRUcross-cutsthecategoriesthrough/durchandover/über.

Another salientparameter isphaseofmatter,which refers to the consistencyof the

LM, ranging fromcompletelyempty to completely solid.For instance,while theLM in

(1) is empty, the forest and swamp in (3) are interspersedwith solid objects or of a

higher overall density, thereby posing impediments to PATH traversal. Thus,phaseof

matteroften featuresprominently in thetargetdomainofmetaphoricalmappings,e.g.

140

when someone has to hack their way through a “thicket of regulations” to obtain a

businesslicense.

FIGURE1

13.1.2.Sense(Ib):LMisa“Floor”SurfaceinPhysicalSpace(fig.3)(13) Tarô-ga hashi-wo tôt-ta. Tarô-NOM bridge-ACC TÔRU-PAST ‘Tarôcrossedthebridge.’

(14) Kuruma-ga kôdô-wo tôt-ta. Car-NOM highway-ACC TÔRU-PAST ‘Thecardrovedownthehighway.’

(15) Kyûkyûsha-ga basuyûsenrên-wo tôt-ta. Ambulance-NOM busprioritylane-ACC TÔRU-PAST ‘Theambulancedroveonthebusprioritylane.’

Asmentionedabove,theLMsin(13)-(15)areincompatiblewiththroughordurch.One

cannotgothroughabridgeorlaneinEnglish,sincetheseobjectslackverticalextension.

Itmightbearguedthatthisisnottrueif,forexample,thebridgehashandrailsoneach

side.However,suchhandrailswouldnotbepartofthebridge’sactivezone,i.e.itisthe

flat surfaceof thebridge that isbeing traversed,not therails.That is, ifwe imaginea

prototypical 3D-volume as a cube, then a 2D-surface corresponds to the cube’s

undersideor“floor”(seefig.2).InthiswaythevolumeandsurfacesensesofTÔRUare

connectedviaa3D-cube-->2D-squareimageschematransformation.

FIGURE2:volumeto“floor”transformation

141

Still, it isworthnoting that the scenes in (13)-(15) include thepossibilityof theTR

facingsomesortofresistanceorchallengeonitsPATH.Forexample,itisdangerousto

cross a shaky bridge spanning a river, evenwith no physical obstacle in theway. As

mentioned in the preceding section, this concept of resistance or challenge is often

metaphoricallyexpressedintermsofdensephaseofmatterinthephysicalrealm.This

isnot tosaythatanyof thesentences in(13)-(15)areactuallymetaphorical. Imerely

suggestthatthereareconceptualparallelsbetweenthetraversalofadensevolumeand

facing resistance/challenge –which by associationmay allow for the construal of the

aboveLMsassomewhat“abstractly3-dimensional“.

FIGURE3

13.1.3.Sense(Ic):LMisa“Wall”SurfaceinPhysicalSpace(fig.6)

(16) Hanako-ga tera-no mon-wo tôt-ta. Hanako-NOM temple-LK gate-ACC TÔRU-PAST ‘Hanakopassedthroughthetemplegate.’

(17) Tarô-ga hari-ni ito-wo tôshi-ta. Tarô-NOM needle-DAT thread-ACC TÔSU-PAST ‘Tarôthreadedtheneedle.’

(18) Kono kappa-ga ame-wo tôsa-nai. This raincoat-NOM rain-ACC TÔSU-NEG. ‘Thisraincoatiswaterproof.’

Again,ifwevisualizeaprototypicalvolumeasacube,thecube’sboundaryprecedingthe

GOALcorrespondstoa2D“wall”(seefig.4).Thisisanothervariantofthe3D-cube-->

2D-square image transformation postulated in the section above. The term “wall” is

somewhat inadequate (and therefore put in quotationmarks) because the 2D surface

LMcanbesolid(18)aswellasempty(17).1Conversely,ifonezoomsinona2D-“wall”

oneendsupwitha3D-volume.Thus, the image-schematransformationworks inboth

ways.Ifgranularityissufficientlyincreased(viaamagnifyingglass,amicroscope,etc.),

evena“2D”LMofonemillimetercanbeconstruedasa3D-volume(seefig.5).1WhethertheLMin(16)isconstruedassolidoremptydependsontheexistenceorabsenceofacloseddoor.

142

13.1.4.Sense(Id)LMisaMassofUnspecifiedDimensionalityinPhysicalSpace(19) Kinzoku-ga denki-wo tôsu. Metal-NOM electricity-ACC TÔSU ‘Metalconductselectricity.’

Asexamplesliketheseshow,theLMdoesnotnecessarilyneedtobespecifiedalongthe

dimensionality parameter. Though, as our encyclopedic knowledge tells us, electrical

conductivity involvesaPATHand thereforeanat leastone-dimensionalLM.Fromthe

fact that TÔRU is inherently dependent on a PATH, we can conclude the verb’s

incompatibilitywithzero-dimensionalpoint-likeLMs.

13.1.5.Sense(II):LMisaTemporalExpanse(fig.7)(20) Hitotsu-no kimono-de natsu fuyu tôsu hito-ga iru. One-LK garment-INS summer winter TÔSU people-NOM exist ‘Therearepeoplewhowearthesamegarmentthroughsummerandwinter.’

(21) Tarô-ga sanjikan tôshi-te hon-wo yon-da. Tarô-NOM threehours TÔSU-TE books-ACC read-PAST ‘Tarôreadbooksforthreehoursstraight.’

(22) Hanako-ga hiru-mo yoru-mo tôshi-te hatarai-ta. Hanako-NOM day-also night-also TÔSU-TE work-PAST ‘Hanakoworkeddayandnight.’

In thismetaphorically derived usage type the TR follows a linear PATH through time

ratherthanspace.AlthoughintheabovesentencesthetemporalLMcanbeconstruedas

a one-dimensional SOURCE-PATH-GOAL structure (i.e. a horizontal line), it should be

mentioned that the parameter phase ofmatter – which normally only applies to 3D-

volumes–canbeofsomerelevanceinthetemporaldomain.Ahardwinterandaneasy

youthareobviouslyverydifferentinrespecttohowmuchresistanceaTRencounterson

itswaythroughthem.However,this isanobservationofamoregeneralnatureandis

notnecessarilyrelevanttothespecificexamplesabove.

FIGURE7

144

Note further that this sense is incompatible with tôru and instead requires a

somewhatgrammaticalizedformoftôsu.Asimilargrammaticalizedvariant,thecomplex

postpositionwotôshite,willbediscussedfurtherbelow.

13.1.6.Sense(III)LMisaNon-spatial,Non-temporalExpanse

(23) Hôan-ga teikô-ni at-ta ga, kekkyoku gikai-wo tôshi-ta. Bill-NOM resistance-DAT meet-PAST CONJ eventually congress-ACC TÔSU-PAST ‘Thebillmetwithresistancebuteventuallypassedcongress.’

(24) Hanako-ga ganko-ni jibun-no iken-wo tôshi-ta. Hanako-NOM stubbornly self-LK opinion-ACC TÔSU-PAST ‘Hanakosubbornlypushedthroughheropinion.’

(25) Kono sakka-ga ‘kitsuneudon’ to iu pennêmu-de tôt-teiru. This author-NOM ‘KitsuneUdon’ QT call penname-INS TÔRU-PROG ‘Thisauthorisknownbythepenname‘KitsuneUdon’.’

(26) Tarô-ga shinbunkiji-ni me-wo tôshi-ta. Tarô-NOM newspaperarticle-DAT eyes-ACC TÔSU-PAST ‘Tarôskimmedthroughthenewspaperarticle.’

InthissensetheSOURCE-PATH-GOALstructureofTÔRUismappedontootherabstract

domains.Asshownby(23), thephaseofmatterparametermayplayasignificantrole.

Here gikai (parliament) plainly denotes a political institution, not simply a locus in

physical space.Nevertheless, theTR encounters resistance on its PATH.Resistance in

physicalspacetypicallyentailsaLMofsomedensityoranemptyLMinterspersedwith

solidobstacles.In(23)thisnotionismappedontothetargetdomain,givingrisetothe

conceptofresistanceinthesocialrealm.Suchagonist–antagonistconstellationscanbe

observedwithrespecttothrough/durchinEnglishandGermanaswell:

(27) Thepresidentpushedthebillthroughparliament.(28) DieRegierungpeitschtedasneueGesetzdurchdenBundestag.

Inthesecasesthemetaphoricalmappingsareroughlyasfollows:

SOURCE:physicalspace TARGET:policymaking TR: -physicalobject(e.g.traveller) -abstractobject(e.g.bill)SOURCE: -pointofdeparture -draftingofthebillPATH(LM): -physicalPATHthroughspace -abstractPATHthroughinstitutionGOAL: -pointofarrival -passingofthebillImpediments:(LM’sphaseofmatter)

-physicalobstacles(e.g.mud,stones,trees,rivers)

-institutionalobstacles(e.g.politicalopposition,laws,regulations,etc.)

145

Therefore,whenconsideringtheuseofTÔRUintheabstractdomain,wecanmakethe

followingobservation regarding theparameterofdimensionality:Where thenotionof

resistanceissalient,suchas(23)and(24),phaseofmatterwillnecessarilyplayarole.In

thesecasestheabstractLMisconstruedasavolumecapableof includingobstaclesor

impedingtraversalthroughahigherdensity. Inothercases,suchas(25)and(26),the

dimensionalityparameterdoesnotfeatureprominentlyandtheLM’sdimensionalityis

thereforelessclear-cut.

Wecan furthernote that severalLMscannotbeovertly realizedby lexicalmaterial.

Examplesofsuchsub-lexicalLMsaregivenin(24)and(25):

(24’) Hanako-ga ganko-ni jibun-no iken-wo (??)-ni tôshi-ta. Hanako-NOM stubbornly self-LK opinion-ACC (??)-DAT TÔSU-PAST

(25’) Kono sakka-ga ‘kitsuneudon’ to iu pennêmu-de (??)-wo tôt-teiru. This author-NOM ‘KitsuneUdon’ QT call penname-INS (??)-ACC TÔRU-PROG

In both cases the LM corresponds to some vague communicative PATH. Here, the

impossibilityoffillingthelexicalgapreflectsthehighlyabstractnatureoftheLM.Such

constructions,beingnotreadilyanalyzable,usuallybecomeidiomatic.

13.1.7.Sense(IV):LMisanInstrument(wotôshite)(29) Oto-ga kabe-wotôshite mimi-ni todoi-ta. Sound-NOM wall-WOTÔSHITE ears-DAT reach-PAST ‘Thesoundwasaudiblethroughthewall.’

(30) Gakusha-ga bôenkyô-wotôshite hoshi-wo kansoku shi-ta. Scholar-NOM telescope-WOTÔSHITE stars-ACC observation do-PAST ‘Thescholarobservedthestarsthroughatelescope.’

(31) Ryôgawa-wa bengoshi-wotôshite kôshô shi-teiru. Bothsides-TOP lawyers-WOTÔSHITE negotiation do-PROG ‘Bothsidesarenegotiatingthroughtheirrespectivelawyers.’

(32) shujinkô-wa samazama-na keiken-wotôshite seichô suru. protagonist-TOP various-COP.ATT experiences-WOTÔSHITE growth do ‘Theprotagonistmaturesbygoingthroughvariousexperiences.’

As these examples show, the complex postposition wo tôshite marks its LM as an

instrument.SincewotôshiteisagrammaticalizedvariantwithintheTÔRUnetwork,itis

worth askingwhatPATHs and instrumentsmayhave in common. In order to answer

this question, consider the notion of an action chain. According to Langacker (1991:

292),anactionchainfollowstheflowofenergyfromanenergysource(orhead)toan

146

energysink (or tail).Forexample, in thesentenceAwaitercrackedtheicewitharock

(1991:292)thewaiteristheenergysourceandtheiceistheenergysink.However,the

energy is not transmitted directly from waiter to ice. There is an intermediary, an

instrument, through which the energy travels from waiter to ice, i.e. the rock. It is

therefore natural to construe instruments as PATHs, because the image schematic

structureofaprototypicalactionchainisSOURCE-PATH-GOAL.

Unsurprisingly,thesamephenomenoncanbeobservedcross-linguistically.Inregards

to the English preposition through, Radden (1989: 571) notes: “The spatial idea of

passing through a tunnel or a channel gives rise to the figurative meaning of a

determinate, ‘channelled’ means […]” In a similar vein Smith (1987) points out the

parallelsbetweenspatialandinstrumentalusesofGermandurch:

(33) ErwarfdenBalldurchdasFenster.[Hethrewtheballthroughthewindow.](34) DasHauswurdedurchFeuerzerstört.[Thehousewasdestroyedbyfire.] (fromSmith1987:445f.)

FollowingLangacker’s action chainmodel, henotes that “theLMof durch serves as a

conduitofsorts inbothof theseexamples:[…]asaconcreteobject throughwhichthe

ball moves, and […] as an instrument through which force or energy moves” (Smith

1987: 446). Therefore, returning to the case ofwotôshite, it seemsonly natural for a

verbwhichisstronglyassociatedwiththePATHschematograduallytakeontheroleof

aninstrumentalmarker.

Althoughwotôshite isaptlydescribedasaninstrumentalmarker, itshouldbenoted

thatinstrumentalityisnotaclear-cutconceptbutascalarone.Basedontheabovedata,

a scale of instumentalitywould look something like this (taking into account that the

followingispossiblyanoversimplification):

(35) SpatialPATH-->physicalinstrument-->abstractinstrument/manner-->cause

Agiven instancemaybe locatedonanypointon this continuum. (29), forexample, is

locatedtowardstheleftendofthisscale,sincetheLMisactuallyaspatialentitythrough

whichtheTRtravels.Ontheotherhand,asentence like(32)wouldgravitatetowards

therightendofthescale,sinceheretheabstractLMofwotôshite(experience)initiates

147

a maturing process. This conflation of instrument and cause is also reflected by the

competitionbetweenwotôshiteandtheSOURCEmarkerniyotte3:

(36a) Tarô-ga kono keiken-niyotte seichô shi-ta. Tarô-NOM this experience-NIYOTTE growth do-PAST

(36b) Tarô-ga kono keiken-wotôshite seichô shi-ta. Tarô-NOM this experience-WOTÔSHITE grow do-PAST

‘Tarômaturedthroughthisexperience.’

CausalusesofPATHprepositionsareknownfromEnglishandGermanaswell:

(37) Animmensenumberofpeoplearekilledthroughtrafficaccidentseveryyear. (38) JedesJahrverliertderStaatGelddurchSteuerhinterziehung. Everyyearthestatelosesmoneythroughtaxevasion. ([37]fromRadden1989:571)

Considering the image schematic structure of the action chain, the cause-instrument-

link ishardlysurprising. Inaprototypicalactionchain theSOURCE(theagent)affects

thePATH(theinstrument),whichinturnaffectstheGOAL(thetheme).I.e.,wehavea

causal chain of the type agent --> instrument --> theme. However, as Langacker (e.g.

1991: 295ff.) and Talmy (e.g. 2003a: 357ff.) have pointed out, construal changes

dependingonourdistributionof attention.Again, considerLangacker’s example from

above:

(39) Awaitercrackedtheicewitharock.

Thecausalchain inthisscene iswaiter-->rock-->ice.Accordingtothisconstrual, the

waiter, being an animate intentional entity, causes the ice to crack. Although he only

affects the ice indirectly through the rock, he is the original SOURCEof energy in the

causalchain.Butwhatifweeclipsethewaiterfromourconstrualofthescene?

(40) Therockcrackedtheice.

This sentence could be used in a context where the waiter’s contribution to the ice-

crackingeventisdeemedlessrelevantthantheinstrument’s.E.g.:Hetriedanicepick,a

paperweightandarock.Butonlytherockcrackedtheice. In thiscase, the instrument’s

3ForamoredetaileddiscussionofniyotteseeMatsumoto(1998b).Ashenotes,thecausaluseistheoldestamongtheabstractusesofniyotte,followedbyitsuseasamarkerofmeans(1998b:40).So,heretoo,wehaveevidenceforthecloseconnectionbetweencausalityandinstrumentality.

148

levelofagencyisheightened,whiletheoriginalenergysourceisbackgrounded.Weare

familiar with similar cases of “transferred agency“ from the so called middle

construction:

(41a) Theemployeesellsthecar.(41b) Thecarsellswell.(41c) Thecar(virtually)sellsitself.

Iwillnotdiscussthemiddleconstruction itselfhere,sincesuchadiscussionwouldbe

farbeyondthescopeofthischapter.Forthepresentpurpose,Iammerelyinterestedin

theshiftofagencyfrom(a)to(c).In(41a)theemployee’sactionsarethemaincauseof

the car’s selling. In (41b) the quality of the car seems to be more salient than the

employee’ssalespitch.Andfinally,in(41c)thecar’squalityistheonlysalientfactor–so

muchso,thatitisconstruedas“sellingitself”withoutanyhumanassistance.Theupshot

isthis: If theenergysourceisremovedfromthecausalchain, it isonlynatural forthe

nextelementdownstreamtotakeitsplace.

Thus, in terms of the action chainmodel, the conflation of instrument and cause is

straightforwardlyaccountedfor:Instruments,asintermediaryelements,areadjecentto

the energy source (the cause). Consequently, if the original energy source is

backgrounded/gapped,theinstrumentbecomesthe“next-bestenergysource”andmay

thereforebeconstruedascause.

13.2.TheSensesofV-TÔRU

13.2.1.LMisanX-dimensionalExpanseinPhysicalSpace(42) Ame-ga fuku-no ura-made shimi-tôt-ta. Rain-NOM clothes-LK backside-ALL soak-TÔRU-PAST ‘Therainsoakedthroughtotheliningoftheclothes.’

(43) Kawa-no mizu-ga suki-tôt-teiru. River-LK water-NOM becometransparent-TÔRU-RES ‘Thewateroftheriverisclear.’

(44) Tarô-ga ita-ni kugi-wo tsuki-tôshi-ta. Tarô-NOM board-DAT nail-ACC thrust-TÔSU-PAST ‘Tarôdroveanailthroughtheboard.’

(45) Kaze-ga ie-no naka-wo fuki-tôshi-ta. Wind-NOM house-LK inside-ACC blow-TÔSU-PAST ‘Windblewthroughthehouse.’

ThissenseofV-TÔRUisanalogoustoitssimplexcounterpartinthespatialdomain.That

is, the TR traverses the LM in themanner expressed by the V1. Aswith the simplex,

149

there is variance along the parameters of dimensionality and phase of matter. For

instance, the LM in (44) is two-dimensional and solid,while the LM in (45) is three-

dimensionalandempty.SinceIhavealreadycommentedontheseparametersabove,I

haveoptedforalessfine-graineddistinctioninthecaseofV-TÔRU.4

13.2.2.LMisaTemporalExpanse(ExtendedProcess)(46) Hanako-ga sannenkan onaji kutsu-wo haki-tôshi-ta. Hanako-NOM threeyears same shoes-ACC wear-TÔSU-PAST ‘Hanakoworethesameshoesforthreestraightyears.’

(47) Tarô-ga hyakkiro-no michi-wo aruki-tôshi-ta. Tarô-NOM hundredkilometers-LK road-ACC walk-TÔSU-PAST ‘Tarôwalkedaroadofhundredkilometersallthewaydown.’

(48) Kono hon-wo yomi-tôsu no-ni isshûkan kakat-ta. This book-ACC read-TÔSU NMLZ-DAT oneweek take-PAST ‘Ittookaweektoreadthroughthisbook.’

(49) Hanako-ga uso-wo tsuki-tôshi-te, iki-te ki-ta. Hanako-NOM lies-ACC tell-TÔSU-TE live-TE come-PAST ‘Hanakolivedalie(allherlife).’

This sense is analogous to its simplex counterpart in the temporal domain. However,

heretheLMof-tôsuistheprocessprofiledbytheV1.TheV1,inturn,takesasitsLMthe

direct object of the sentence (markedbywo). In (46), for example, theV1haku takes

kutsuasitsLM,yieldingthestructure(sannenkanonaji)kutsu-wohaku.Thisstructureis

atemporalexpanseandaninstanceoftheSOURCE-PATH-GOALimageschema:

SOURCE: Beginningofprocess,i.e.startwearingshoesPATH: Durationofprocess,i.e.temporalportionduringwhichshoesarewornGOAL: Endofprocess,i.e.stopwearingshoes

TheV2-tôsutakesthistemporalexpanseasitsLM.Thatis,intermsofimageschematic

structuretheTR(Hanako)isconstruedasmovingthroughthistemporalexpanseinthe

samewayitwouldmovethroughaspatialexpanse–fromSOURCEtoGOAL.

Further,-tôsu–likeitssimplexcounterpart–doesnotfocusoneithertheSOURCEor

theGOALbutonthePATHportion.Toillustratethispoint,considerthatbothV-kiruand

V-tôsupresupposethewholeSOURCE-PATH-GOALstructureasbackground(orbase,in

4Ofcourse,insuchmattersofgranularity–andthereforeschematicity–judgementvariesfromauthortoauthor.Forexample,Sugimura(2012:54f.)goessofarastopostulateseperatesensesforindividuallexicalitemssuchasfuki-tôsu(blowthrough)andmi-tôsu(seethrough).

150

Langackarianterms),butdifferastowhatportionofthestructureis“inspotlight”(see

alsofig.8):

(47a) Tarô-ga hyakkiro-no michi-wo aruki-kit-ta. Tarô-NOM hundredkilometers-LK road-ACC walk-KIRU-PAST

(47b) Tarô-ga hyakkiro-no michi-wo aruki-tôshi-ta. Tarô-NOM hundredkilometers-LK road-ACC walk-TÔSU-PAST

The V2 -kiru contributes the conceptual content of the SPLIT schema and therefore

profilestheportionwherediscontinuityemerges(theGOALaspointofsegmentation).

In contrast, -tôsu contributes the conceptual content of the PATH schema and thus

profiles the portion between SOURCE and GOAL. These different focus properties can

explainwhy(50a)isacceptable,while(50b)isnot:

(50a) Hanako-ga isshûkan hon-wo yomi-tôshi-ta. Hanako-NOM oneweek books-ACC read-TÔSU-PAST

(50b) *Hanako-ga isshûkan hon-wo yomi-kit-ta. Hanako-NOM oneweek books-ACC read-KIRU-PAST

According to image schematic structure, there is a fundamental difference in

dimensionalitybetweenPATHsontheonehandandSOURCEsandGOALsontheother:

A PATH has at least one dimension (its most schematic depiction is a line) while

SOURCEsandGOALsarezero-dimensional(theirmostschematicdepiction isapoint).

In other words, isshûkan requires an extended process, whereasV-kiru is temporally

punctual.Ofcourse,onecanobservetheoppositeofthisaswell:

(51a) Gojinijuppun-ni yatto kono hon-wo yomi-kit-ta. 17:20h-DAT finally this book-ACC read-KIRU-PAST

(51b) *Gojinijuppun-ni yatto kono hon-wo yomi-tôshi-ta. 17:20h-DAT finally this book-ACC read-TÔSU-PAST

FIGURE8:V-tôsuvsV-kiru

151

NoteinpassingthesimilaroppositionbetweencertainusesoftheGermanparticleverbs

durch-V(PATHfocus)andaus-V(GOALfocus):

(52a) IchhabedieganzeNachtdurchgeschlafen.(52b) *IchhabedieganzeNachtausgeschlafen.

13.3.RelatedConstructions:N-notôri(N-dôri)/V-tôri(53) Isha-no iu tôri-ni shi-nasai. Doctor-LK say TÔRI-DAT do-IMP ‘Doasthedoctorsays.’

(54) Yosô dôri-no tenkai dat-ta. Expectation TÔRI-LK development COP-PAST ‘Thingsplayedoutasexpected.’

(55) Setsumeisho-ni kai-tearu tôri-ni yat-ta ga uma-ku ika-nakat-ta. Instructionsmanual-DAT bewritten TÔRI-DAT do-PAST CONJ good-INF go-NEG-PAST ‘Ididitaccordingtotheinstructionsmanual,butitdidn’twork.’

Finally,letusbrieflyconsidertheN-notôri/V-tôriconstruction,whichmaybeviewedas

aninstanceofalargerfamilyofmetaphorical“way-constructions”foundacrossvarious

languages.Compare(53)-(55)withthefollowingexamplesfromEnglishandGerman:

(56) Severebudgetcutsaretheonlywaytosavethiscompanyfrombankrupcy.

(57) Ifollowedinthefootstepsofmyfatherandbecameanactor.

(58) IchkannDirnichtfolgen,bitteerkläreesnocheinmal.

Ican’tfollowyou,pleaseexplainitagain.

(59) DerWegzumErfolgistmitvielenSteinengepflastert.Theroadtosuccessisarockyone.

Again, such expressions show the pervasiveness of space-time-homology in language

and thought. Aswe have established, a terminal continuous process is understood in

termsoftheSOURCE-PATH-GOALschema.However,therearemultiplepossiblilitiesas

towhat happens between the onset and the end of a process, just as there are often

multiple paths leading to the same physical destination. Thus, doing something in a

certainmanner(orsomethinghappeninginacertainmanner)isanalogoustofollowing

a particular path to a physical destination. In other words, it follows from the EVENT

STRUCTUREMETAPHOR (Lakoff 2006: 204ff.) thatmanner should bemappedonto spatial

navigation.

152

CaseStudies:Summary

Ashypothesized,theprecedingstudieshaveshownthateachverbunderconsideration

isbaseduponaparticularimageschema.Butwhatexactlydoesitmeantosaythatthe

semanticstructureofKAKARUiscenteredaroundtheimageschemaCONTACT;orthat

thesemanticstructureofDERUiscenteredaroundtheEXITschema?

AsLakoff(1990b:438)cautions,itisimportanttodistinguishbetweenpredictionand

motivation.Thesuper-schemaspostulatedattheoutsetofeachcasestudydonotallow

us to predict the senses of the respective verbs. For example, the most schematic

meaningofDERU– movementoutofacontainer–underspecifieseventhecategorial

prototype (spatial exit), since it says nothing about the dimensionality of the TR, the

boundedness of the LM, and so forth. Likewise, the schema is too impoverished to

predictsensesthatinvolvemetonymyand/ormetaphor.Forinstance,wehaveanalyzed

theaccesssenseofDERUasanextensionoftheprototypeviatheprimarymetaphor(s)

INISINACCESSIBLE/OUTISACCESSIBLE(see10.1.5.).Butthisinformationisnotincludedinthe

super-schema;nordowehaveanyaprioriguaranteethatalanguagewillmakeuseofa

givenmetonymicalormetaphoricalextensionjustbecauseitistheoreticallyavailable.In

othercases,themostschematicmeaning,althoughpervasive,doesnotevencoverallthe

senses.Aswehaveseen,somesensesofAGARUsuchassubtractivecompletionhardly

involvetheverticaldimensionatall(see12.1.5).

What the case studies have given us instead is an account of the individual senses’

motivatedness. Image schematic structure in tandem with mechanisms of meaning

extensionputsus inaposition toexplain,aposteriori,whyagivensenseexists.From

thisperspective,thehigh-levelschemasoftheverbsconsideredhereareanindicatorof

theirvastsemanticpotential.Torecapitulate, letusconsiderthemajormechanismsof

meaningextensionandtheirrelationtoimageschematicity.

SchemassuchasEXIT,SPLIT,orPATHare idealcandidates formetaphorical source

domains in virtue of being among ourmost basic experiential gestalts. In some cases

theywillserveassourcedomainsformultipleprimarymetaphorsatonce.Forinstance,

theconceptsoftransfer,accessandexcessareallrelatedtoourexperienceswiththings

leaving CONTAINERS – and hence the EXIT schema (see chapter 10). Furthermore,

SPACE, as the primary domain of exprience, imposes its structure upon a variety of

other domains such as TIME or SOCIAL RELATIONS. Subdomains like TEMPORAL

DISCONTINUITYorSOCIALDISCONNECTIONaresubsequentlyunderstoodintermsof

153

the SPLIT schema (chapter 11), while “movement” through ACTIVITIES and

INSTITUTIONSisunderstoodintermsofPATHTRAVERSAL(chapter13),andsoforth.

Thepotentialformetonymyisnolessimpressive.Itisunsurprisingthatabasicspatial

schema such as CONTACT serves as a point of access for a multitude of associated

concepts, including SUPPORT, FORCE, ELICITED EFFECT, RESTRAINT, and CONTROL

(seechapter9).Although,consideringthecommonbasisofexperientialcorrelation,the

difference between metonymical mapping and primary metaphor is not always

apparent andmost likely amatter of degree. As a subtype ofmetonymy, active-zone

phenomena deserve particular attention. Exemplary in this regard are the profiling

propertiesofthevarioussensesofKIRU,wherefocalprominencechangesbetweenthe

LMasawhole,theobsoleteportionoftheLMandthepointofsegmentation(see11.1.).

Finally, we have seen how image schema transformations function as a source of

polysemy.This is illustratedbythesensesofKAKARU,whichvariouslyfeatureazero-

dimensional TRmoving towards a goal, a one-dimensional TR extending (“growing”)

towardsagoal,andaone-dimensional link-typeTR(see9.1.).Otherexamples include

thevolume, floorandwallLMsofspatialTÔRU(see13.1.)aswellastheascensionand

extensionsensesofAGARU(e.g.12.1.1.;12.1.3.).1

Perhaps most crucially, we have succeeded in showing that grammatical V2s are

inherently meaningful. Their meaning reflects the image schematic structure of their

simplex counterparts and is derived via the samemechanisms of semantic extension

responsible for the complex category strucuture of the simplex. It was argued, for

example,thattheconceptualstructureofinchaotiveV-dasu–ametaphoricalextension

of theEXITschema– iscloselyrelatedtotheaccesssenseof thesimplex(see12.2.6.).

Analogouscaseshavebeenmadefortheremainingimageschemaverbs.Inconclusion,

then, the senses of the simplex and the V2 are best understood as constituting an

internallycoherentsemanticcategory.

1Asinthelattercase,imageschematransformationsandactivezonephenomenaoftengohandinhand.E.g.,aone-dimensionalextendingTR,whenreducedtoitsactivezone,willyieldazero-dimensionalmovingTR.

154

PARTIII:BEYONDTHENETWORK

14.CompositionalDisparity

14.1.ButWhataboutSyntax?

The case studies in this thesis have mainly been concerned with image schematic

structure. Specifically, they showhow image schemas, in tandemwithother cognitive

phenomena such as conceptual metaphor and metonymy, play a vital role in

polysemisation processes. A particularly interesting aspect of this is the semantic

connectionbetweenthespatialsensesofsimplexverbsandthemoreabstractsensesof

grammaticalV2s.Forexample,wehaveseenhowspatialdasuandinchoativeV-dasuare

relatedtooneanotherviathesharedEXITschema.

“Butwhataboutsyntax?”,onemightask.Indeed,itisnoexaggerationtosaythatthe

majorityof the literatureon JapaneseV-Vcompounds isabout “classical”questionsof

argumentstructure–oftenexplicitlyorimplicitlypresupposingagenerativeframework

that treats syntax, semantics, and the lexicon as distinct linguistic components. How,

then,aresuchquestionsaddressedwithinameaning-basedapproachto languagethat

viewslexiconandgrammarasacontinuum?AlthoughIcannotpossiblyanswer(oreven

consider)allopenquestions,Ican–basedontheprecedingcasestudies–atleastshow

howanalternativewayofframingsomeofthemainissuesmightleadtonewinsights.I

will start by examining some earlier theory-neutral suggestions on how to classify

JapaneseV-Vcompoundsaccordingtotheircompositionalproperties,beforeeventually

discussingmorerecentwork, includingKageyama’s(e.g.1993,1996,2009) influential

distinctionofsyntacticvslexicalcompounds.

14.2.V-VCompoundsinTeramura(1969),Nagashima(1976),andYamamoto

(1984)

One of the earliest classifications of Japanese V-V compounds from a compositional

perspective is Teramura (1969), who recognizes two kinds of components: A given

V1/V2 is “independent“ (jiritsu)only if itpreserves itsoriginalmeaningaspartof the

compound, otherwise it counts as an “attached“ (fuzoku) element. The independent vs

attacheddichotomyyieldsthefollowingfourpermutations.

155

TypeI: independentV1 independentV2E.g.:hashiri-saru(run-leave)-->runaway;mochi-ageru(hold-lift)-->liftupTypeII: independentV1 attachedV2E.g.:hashiri-komu(run-inwardsmovement)-->runinto;mi-ageru(look-raise)-->lookup(at)TypeIII: attachedV1 independentV2E.g.:tori-osaeru(take-catch)-->catch(acriminal);uchi-nagameru(hit-gaze)-->lookatsth.whileabsorbedinthoughtTypeIV: attachedV1 attachedV2E.g.:tori-nasu(take-do)-->mediate(betweenparties);nori-dasu(ride-putout)-->start(todo),embarkon

In a similar vein, Nagashima (1976) suggests that a V-V compound consists of a

modifyingelement(shûshokuyôso)andamodifiedelement(hishûshokuyôso).Basedon

thisdistinction,hepostulatestwotypesofcompounds:

TypeI: v1(modifying,lowercasev) V2(modified,uppercaseV)Requirement:Both[N-gaN-wo/niv1]and[N-gaN-wo/niV2]areacceptable.E.g.: Tarô-ga ki-wo kiri-taosu. Tarô-NOM tree-ACC cut-knockdown ‘Tarôcutsdownthetree.’ --> Tarô-ga ki-wo kiru. --> Tarô-ga ki-wo taosu.

TypeII V1(modified) v2(modifying)Requirement:[N-gaN-wo/niV1]isacceptable,but[N-gaN-wo/niv2]isnot.E.g.: Inu-ga kodomo-ni kami-tsuku. Dog-NOM child-DAT bite-stickto ‘Thedogbitesthechild.’ --> Inu-ga kodomo-wo kamu. --> *Inu-ga kodomo-ni tsuku.

It isworthyofmentionthatNagashima’suseofconstructionaltemplatesmarksashift

towardssyntactictestsasamethodofcategorizingV-Vcompounds.WhereTeramura’s

dependent vs independent distinction is based on the somewhat vague notion of

“preserving theoriginalmeaning”of agivencomponent,Nagashima’s templatesmake

for a more technical approach – although he ultimately fails to specify what kind of

acceptabilityhehasinmind.1

1Presumably,Nagashimahasnativespeakerjudgementsinmind.However,anativespeakermayacceptordismissasentenceforvariousreasons.Heappleeatisstructurallyflawed(ungrammatical),whileShedranksolidstoneconflictswithwhatweknowaboutdrinking(Chomsky’sfamousexampleColorlessgreenideassleepfuriouslyexemplifiestheissue).

156

Critisizing Teramura’s meaning preservation criterion as too subjective and

Nagashima’s categorization attempt as not comprehensive enough (note that there

seems to be no place in his model for Teramura’s Type IV compounds), Yamamoto

(1984)suggestsanaccountofV-Vcompoundsbasedonthenotionof“casegovernment”

(kakushihai).AccordingtoYamamoto,everyverbhasafixednumberofargumentslots

for “case components” (kaku seibun), which define its valence (ketsugôka). The verb

hashiru(run),forexample,hasasubjectargumentslotforthecasecomponent[N-ga],as

in [Tarô-ga]hashiru(Tarô runs), and thereforeavalenceof1. Sincemiru (watch)has

two argument slots, it has a valence of 2: [Kodomo-ga] [terebi-wo]miru (The child

watchestelevision).Andoshieru(teach)withitsthreeargumentslotshasavalenceof3:

[Sensei-ga][seito-ni][rekishi-wo]oshieru (Theteacherteachesthestudenthistory).That

is, hashiru governs nominative case (N-ga),miru governs nominative and accusative

case (N-wo)andoshieru governsnominative,accusative,anddativecase (N-ni).Based

onhowthevalencepropertiesofthecompoundinteractwiththevalancepropertiesof

eachindividualverb,Yamamoto(1984)arguesforthefollowingcategorization:

--TypeI:BothV1andV2exhibitcasegovernment--Kodomo-ga naki-sakebu. Child-NOM cry-scream ‘Thechildcriesandscreams(criesintensely).’ -->Kodomo-ganaku. -->Kodomo-gasakebu.

Haha-ga kodomo-wo daki-kakaeru. Mother-NOM child-ACC embrace-hold ‘Themothercradlesthechild.’ -->Haha-gakodomo-wodaku. -->Haha-gakodomo-wokakaeru.

--TypeII:OnlyV1exhibitscasegovernment--Oyu-ga waki-tatsu. Water-NOM boil-stand ‘Thewaterseethes.’ -->Oyu-gawaku. -->*Oyu-gatatsu.

Hanako-ga okashi-wo tabe-sugiru. Hanako-NOM sweets-ACC eat-exceed ‘Hanakoeatstoomanysweets.’ -->Hanako-gaokashi-wotaberu. -->*Hanako-gaokashi-wosugiru.

Notethatinu-gakodomo-nitsuku,bydefault,isneither“ungrammatical“nornonsensical(ItcanbereadasThedogaccompaniesthechild).Itbecomesinfelicitousonlywhenreadasanalternativetoinu-gakodomo-nikami-tsuku–whichbringsusbacktoTeramura’snotionofmeaningpreservation.

157

--TypeIII:OnlyV2exhibitscasegovernment--Fuun-ga uchi-kasanaru. Misfortune-NOM hit-pileup ‘(I)haveastreakofhardluck.’ -->*Fuun-gautsu. -->Fuun-gakasanaru.

Fuchûi-ga jiko-wo hiki-okosu. Negligence-NOM accidents-ACC pull-cause. ‘Negligenceleadstoaccidents.’ -->*Fuchûi-gajiko-wohiku. -->Fuchûi-gajiko-wookosu.

--TypeIV:NeitherV1norV2exhibitcasegovernment--Tarô-ga (nyûsu-wo kii-te) tori-midasu. Tarô-NOM (news-ACC listen-TE) take-disturb ‘Tarôgetsupset(listeningtothenews).’ -->*Tarô-ga(nyûsu-wokii-te)toru. -->*Tarô-ga(nyûsu-wokii-te)midasu.

Keisatsu-ga inshuunten-wo tori-shimaru. Police-NOM drunkdriving-ACC take-tighten ‘ThePolicecrackdownondrunkdriving.’ -->*Keisatsu-gainshuunten-wotoru. -->*Keisatsu-gainshuunten-woshimaru.

Withinthismodela failedsubstitutiontestshowsthatagivenV1/V2doesnotexhibit

casegovernmentoverthecompound’sargument(s).Forexample,Yamamoto(1984:38)

claims that the result of substituting the simplex tatsu forwaki-tatsu inOyu-gawaki-

tatsu, i.e. *Oyu-ga tatsu, is “ungrammatical“ (hibun). However, this use of the term

“ungrammatical”begsthequestion,sincethesentenceisnotstructurallyflawedinthe

Chomskyansense.Inotherwords,inagenerativeframeworkthesentenceOyu-gatatsu

would not violate any “syntactic rules”. Sowhat else could Yamamoto havemeant by

“ungrammatical”? Ifwe lookat the sentencesmarkedbyanasterisk, it becomes clear

thattheV1/V2inquestioneitherrequiresadifferentinterpretationvis-a-visitsusein

thecompoundorbecomesdifficulttointerpretatall,thuschangingthemeaningofthe

sentence or rendering it outright incomprehensible. Viewed in this light, Yamamoto’s

grammatical/ungrammatical distinction is not so different from Teramura’s meaning

preservation criterion. In fact, if we take Teramura’s examples from above and run

Yamamoto’ssubstitutiontestswiththem,itturnsoutthatallofTeramura’sindependent

verbsexhibitcasegovernmentinYamamoto’ssense,whiletheattachedverbsdonot.

14.3.CompositionalDisparityasanUmbrellaTerm

The dichotomies independent vs attached and case government vs no case government

agree with our intuition that within some V-V compounds V1 and V2 behave in a

158

somewhat asymmetrical manner. Moreover, this asymmetry concerns the level of

discriptionwhichistraditionallyknownasargumentstructure.Considerthefollowing

examples:

(1) Kuchi-ga kawaki-kit-teiru. Mouth-NOM becomedry-KIRU-RES ‘(My)mouthisalldriedup.’ -->Kuchi-gakawai-teiru. -->*Kuchi-gakit-teiru.

(2) Hanako-ga uta-wo utai-dashi-ta. Hanako-NOM song-ACC sing-DASU-PAST ‘Hanakostartedsingingasong.’ -->Hanako-gauta-woutat-ta. -->*Hanako-gauta-wodashi-ta.

(3) Sagishi-ga okane-wo damashi-tot-ta. Scammer-NOM money-ACC deceive-take-PAST ‘Thescammertookthemoneybydeception.’ -->*Sagishi-gaokane-wodamashi-ta. -->Sagishi-gaokane-wotot-ta.

(4) Tarô-ga ji-wo kaki-nagut-ta. Tarô-NOM characters-ACC write-beat-PAST ‘Tarôwrotethecharactersinadisorderlymanner.’ -->Tarô-gaji-wokai-ta. -->*Tarô-gaji-wonagut-ta.

In eachof theseexampleseitherV1orV2 is somehow incompatiblewithat leastone

argument of the compound. I am deliberately using the vague gloss somehow

incompatible to indicate that we have little beyond the intuition that “something is

off“untilweareabletospecifywhatexactlythatis.Asaconvenientumbrellaterm,Iwill

refertotheasymmetrydisplayedbytheaboveexamplesascompositionaldisparity.

14.4.SalienceandAbstractEntities:SomeCompoundswithGrammaticalV2s

Letusbeginourinquiryintocompositionaldisparitybyexaminingthekindofverbsthat

havebeenthefocusofthisstudy:imageschemaverbsfunctioningasgrammaticalV2s.

RecallthattheseV2smayhavehighlyabstractTRsorLMs.

(5) Ame-ga furi-dasu. Rain-NOM fall-DASU

(6) Kuchi-ga kawaki-kiru. Mouth-NOM becomedry-KIRU

159

(7) Hon-wo yomi-kakeru. Book-ACC read-KAKERU

(8) Karada-wo kitae-ageru. Body-ACC train-AGERU

(9) Uso-wo tsuki-tôsu. Lie-ACC tell-TÔSU

Ihaveargued (see10.2.6.) that -dasu in (5) isanextension from theACCESSsenseof

DERU.Whatbecomesaccessiblein(5)isnotmerelyaTHING(rain)butratherastateof

affairs.Therefore,theTRoftheV2-dasucorrespondstotheabstractpropositionthatit

rains. The LM of -dasu is no less abstract. By application of the metaphor IN IS

INACCESSIBLE/OUT IS ACCESSIBLE, the LM of -dasu corresponds to the realm of sensory

inaccessibility. In other words, that it rains (i.e. the TR) emerges from the realm of

sensoryinaccessibility(i.e.theLM).ItiscrucialtonotethattheTRof-dasuandtheTRof

furi-dasuarenotthesameentity.Instead,theTRof furi-dasucorrespondstotheTRof

theV1furu,i.e.rain(ame).Thus,wemightsaythatincompetingfortheplaceofclausal

TRandgrammatical subject, theTRof furu prevailsover theTRof -dasu.Or, tousea

geneticsmetaphor,furi-dasu“inherits”itsTRfromfuru.

How,then,isthiscompetitionforsubjecthooddecided?Thestraightforwardansweris

that the element with the highest cognitive salience is expected to prevail. But what

exactly does “cognitive salience” refer to? According to Langacker (1991: 308) “a

prototypicalsubjectrankshighlywithrespecttoallfourtopicalityfactors:itisagentive,

human,definite,andthefigurewithintheprofiledrelationship.”Inthecaseof(5),both

candidatesaretiedintermsofthefirst,third,andfourthfactor:Itseemsdubioustosay

thatoneismoreagentiveordefinitethantheother,andbothareTRoftheirrespective

verbs furu anddasu.This leavesuswith the thirdcriteriumas themost relevantone:

their places on the empathy hierarchy. Langacker (1991: 307) – based on previous

research by Silverstein (1976), Deane (1987), and Kuno and Kaburaki (1977) –

postulatesthefollowingorderofentities,ranked“accordingtotheirpotentialtoattract

ourempathy”:

speaker>hearer>human>animal>physicalobject>abstractentity

According to this hierarchy, abstract entities offer the lowest potential for human

identification and therefore, all other things being equal, have the lowest cognitive

160

salience.Returning to (5), thismeans thatameteorologicalphenomenonsuchasrain,

whilenotanthropomorphicorevenveryconcrete, is stillmuch lessabstractand thus

moreempathy-enducing thana stateof affairs, i.e. (the fact, circumstance, etc.) thatit

rains.Noteinpassing,thatconceptualmetaphortheorylendsadditionalsupporttosuch

an assumption. A great many (though not all) metaphorical mappings illustrate the

abstractintermsofthemoreconcrete,makenon-sense-perceptibleentitiestangiblein

termsofsense-perceptibleones.Takethefollowingexamples:

(10) Ah,Iseeyourpoint. (UNDERSTANDINGISSEEING)(11) Ouruncleleftthisworldtooearly. (DEATHISDEPARTURE)(12) Heusedadirtytrick. (AMORALISDIRTY)(13) She’satoweringfigureinherfield. (SIGNIFICANTISBIG)

Themostobviousexamplesarecasesofpersonification:(14) TheGrimReapercameforhim. (DEATHISAREAPER)(15) Timeisnotonourside. (TIMEISANADVERSARY)

Fauconnier and Turner (2003: 322), in their book on conceptual blending, are quite

unambiguousonthematterofanthropocentrismandcognitivesalience:“Humanbeings

are evolved and culturally supported to deal with reality at human scale – that is,

throughdirectactionandperceptioninsidefamiliarframes[…].”

Furthermore,notethattheconceptualcontentoftheV1furuisalreadypartoftheTR

oftheV2dasu.Thus,promotionoftheTRofdasutosubjecthoodwouldresultinlexical

redundancy2:

(16) *[Ame-ga furu no]-ga furi-dasu. [Rain-NOM fall NMLZ]-NOM fall-DASU

Soinconclusion,whatcanbesaidaboutthecompositionaldisparityoffuri-dasu?Whyis

(17b)afelicitoussentencewhile(17c)isnot?

(17a) Ame-gafuri-dasu.

(17b) Ame-gafuru.

(17c) *Ame-gadasu/deru.

Asstatedabove,theTRoftheV2dasuisastateofaffairs,i.e.anabstractentityranked

extremely low on the empathy hierarchy. As such, it remains schematic and

unelaborated.Duetoitslowcognitivesalienceitisabadcandidateforsubjecthoodvis-

2See,however,theactivezoneaccountofinchoative-dasubelow(14.5.3.),whichoffersanelegantsolutiontotheredundancyissue.

161

a-vis the less abstract TR of the V1, i.e. ame. The infelicity of (17c) is now easily

explained:ThereisamismatchofTRs.Thecompoundfuri-dasurequirestheTRofdasu

to be a state of affairs. The reason for this is that under the metaphor IN IS

INACCESSIBLE/OUTISACCESSIBLE states of affairs can be conceptualized as physical things

movingoutofacontainer–therebyyieldinganinchoativereading.While(17c)couldbe

accommodatedsomehow(e.g.raincomingoutoftheclouds,etc.),wecannotexpectitto

have an inchoative reading without a state of affairs-type TR. Conversely, if we

substituteamewithsuchaTR,wedogetthedesiredreading:

(16) *[Ame-ga furu no]-ga deru. [Rain-NOM fall NMLZ]-NOM DERU

While (17d) is not generally considered felicitous either, the intended inchoative

meaningiseasilyrecognized.

Letusnowbrieflyconsidertheremainingsentences(6)through(9).AsIhaveargued

earlier(see11.2.3.2.), theLMoftheV2kiru inkawaki-kiru isthetimelineitself.Again,

thisisahighlyabstractentitywhichremainsunelaborated(i.e.linguisticallyunrealized)

andisnotinheritedaseithersubjectorobjectbythecompound.Thesubstitutiontestin

(18c)failsduetotheabsenceofatimeline-typeLM.

(18a) kuchi-gakawaki-kiru.

(18b) kuchi-gakawaku.

(18c) *kuchi-gakiru.

In(18a)theV1kawakuprofilesascalarprocesswhoseGOALportionservesaspointof

segmentation (POS), i.e. the end-point of the process profiled by kawaku marks the

transtionfromdynamictostatic.This iswhatmakesthetimelineitselfasalientframe

elementinthefirstplace.However,withnosuchV1thereisnothingin(18c)tosuggest

theexistenceofatimeline-typeLM.

Moving on to (7), the LM of the V2 kakeru is the process profiled by the V1.

Metaphorically,kakeruindicatesthattheTRmakesCONTACTwiththefrontalboundary

ofthatprocess(see9.2.3.).Sinceabookandtheprocessofreadingabookaretwovery

differententities,thesubstitutiontestcannotyieldthedesiredinchoativereading:

(19a) Hon-woyomi-kakeru

(19b) *Hon-wokakeru/*Hon-nikakaru

162

AswiththeV2-dasu,wegetclosertotheintendedreadingbyreplacingtheLMofthe

compoundwiththeappropriateabstractentity.(Though,again,theresultisnotentirely

felicitous.)

(19c) *[Hon-wo yomu no]-ni kakaru [Book-ACC read NMLZ]-DAT KAKARU

Sentence(8) isan instanceofwhat Ihavecalled–buildingonLindner(1981)– the

achievement of sufficient state sense of AGARU. Here the V1 kitaeru profiles a scalar

process. The TR corresponds to the degree of progress on that scale towards some

sufficientstate(i.e.theLM).Inthecaseofkitae-ageru,kitaeruevokesafitnessscale.The

TR thus corresponds to the degree of fitness whichmoves towards – and eventually

coincides with – the state of sufficient fitness. Note that both TR (degree) and LM

(sufficient state) are quite abstract and remain sublexical. Crucially, the TR is too

abstracttocompetewithkaradaforovertrealization.Analogoustotheaboveexamples,

thesubstitutiontestfailsduetoamismatchofentitytypes:

(20a) Karada-wokitae-ageru.

(20b) *Karada-woageru./*Karada-gaagaru.

Paraphraseclosesttotheintendedreading:

(20c) Karada-no kitaeguai-ga/wo (jûbun-na tokoro-made) agaru/ageru. Body-LK degreeoffitness-NOM/ACC (sufficient-COP.ATT point-ALL) AGARU/AGERU

Finally, -tôsu in (9) requires its LM tobe a temporalPATH (see13.2.2.). ThisPATH

correspondstotheprocessprofiledbytheV1(uso-wo)tsuku.I.e.,in(9)theTRtraverses

theprocessof telling lies. Inotherwords, theTRmetaphorically traverses thepathof

telling lies as if it were a spatial expanse. The closest paraphrase for the intended

readingwouldlooksomethinglikethis:

(21a) Uso-wotsuki-tôsu.

(21b) *[Uso-wo tsuku no]-wo (saisho-kara saigo-made) tôru. [Lies-ACC tell NMLZ]-ACC (beginning-ABL end-ALL) TÔRU

Again, themore abstract TR of -tôsu cannot competewith the TR of the V1 for overt

realization,andthesubstitutiontestfailstoevokethedesiredinterpretation,sinceuso

doesnotprofileaprocess/temporalexpanse:

163

(21c) *Uso-wotôsu/*Uso-wotôru

Tosummarize:TheasymmetrybetweenlexicalV1andgrammaticalV2in(5)-(9),or

more specifically, the “incompatibility“ of the V2 with at least one argument of the

compound,istheresultofamismatchbetweenentitytypes.Ineachcase,thecompound

initsentiretyevokesatleastonehighlyabstractframeelement(e.g.astateofaffairs,a

timeline, a degree of progress, etc.) not evoked by either the V1 or V2 alone. The V2

takesatleastonesuchentityasasublexicalTR/LM.However,becauseoftheirabstract

nature,theseentitiesarenotsalientenoughtocompetewiththeV1’sTR/LMforovert

realizationassubjectorobject.Inotherwords,ifaV2in(5)-(9)isnotcompatiblewith

someargumentof thecompound(as indicatedbya failedsubstitutiontest), there isa

simplereason:ThatargumentdoesnotmatchtheV2’sTR/LM(orevenitsentitytype).

The corresponding TR/LM of the V2 remains unelaborated, i.e. is “missing” from the

leveloflexicalrealization.

14.5.GrammaticalV2sand“FakeTransitivity”

UptothispointIhaveclaimedthatseveralgrammaticalV2s,suchas-dasuand-kakeru–

despite being morphologically marked as transitive – should really be considered

intransitive (see 9.2.3.; 10.2.). The reason for this has been touched upon in the

respective case studies and pertains to the image schematic topology and TR-LM

arrangementfortheseV2s:

(22) inchoative-dasu schema:EXIT(stateofaffairs[TR],CONTAINER[LM]) metaphoratwork:INISINACCESSIBLE,OUTISACCESSIBLE (23) inchoative-kakeru schema:CONTACT(thing/person[TR],process[LM]) metaphoratwork:STATESARELOCATIONS

ObservethatintheaboveschemastheLMisalocationratherthanaparticipant,which

is characteristic for an intransitive relation, whereas a transitive relation obtains

betweentwoparticipants(Langacker1991:343ff.).Infact,itisnotatalluncommonfor

verbstohavetransitiveformwithoutbeingtruetransitives.Oneexampleofthisisthe

setting-subjectconstructioninEnglish:

(24a) Thursdaysawyetanotherstartlingdevelopment.(24b) *YetanotherstartlingdevelopmentwasseenbyThursday. (fromLangacker1991:346)

164

Since the TR, elaborated byThursday, profiles a setting rather than a participant, the

clause fails topassivize.This isnot thecase ifbothTRandLMareparticipants (Mary

sawJohn-->JohnwasseenbyMary).

Inconclusion,then,thefactthatseveralgrammaticalV2sprofileabstractintransitive

relations despite being morphologically transitive is unproblematic in light of the

participantvssetting/locationdistinction.

14.6.OtherSourcesofCompositionalDisparity

AtthispointwehaveidentifiedtheabstractnatureoftheTR/LMofseveralgrammatical

V2sasonesourceofcompositionaldisparity.Ofcourse,therearemanyV-Vcompounds

that do not feature grammatical component verbs, but still display compositional

disparity.Howcanwe,forinstance,accountforthe“missingarguments”(Wittfeld2013)

of lexical–lexical typecompounds? In the following Iwillpostulateseveraladditional

sourcesofcompositionaldisparity:

14.6.1.“Subordination”,ConceptualAutonomy,andDiscourseContext

Considerthefollowingsentences:(25) Tarô-ga (Jirô-kara) okane-wo damashi-tot-ta. Tarô-NOM (Jirô-ABL) money-ACC deceive-take-PAST ‘Tarôtookmoney(fromJirô)bydeception.’ -->Tarô-gaokane-wotot-ta. -->Tarô-gaJirô-wodamashita. -->*Tarô-gaokane-wodamashi-ta.

(26) Tarô-ga (Jirô-kara) pasokon-wo yuzuri-uke-ta. Tarô-NOM (Jirô-ABL) computer-ACC yield-receive-PAST ‘Tarôreceivedacomputerby(Jirô’s)yielding.’ -->Tarô-gapasokon-wouke-ta. -->Jirô-gapasokon-woyuzut-ta -->*Tarô-gapasokon-woyuzut-ta.

(27) Sûtsu-ga ki-kuzure-ta. Suit-NOM wear-crumble-PAST ‘Thesuithaslostitsshapedueto(someone)wearingit.’ -->(??)Sûtsu-gakuzure-ta. -->Dareka[someone]-gasûtsu-woki-ta. -->*Sûtsu-gaki-ta.

In(25)theLMoftheV1isnotovertlyrealizedasdirectobject.In(26)and(27)theTR

of the V1 is not overtly realized as subject (although in the first two sentences the

165

“missing” participant can optionally appear as an oblique). In order to explain the

participant profiling properties of these compounds, it is important to note that they

instantiateacertaintypeofconstructioninwhichtheV1designatesamanner,means,

or cause that modifies the event profiled by the V2. While the classification of

compounds into various constructional schemas is hardly a new idea, their so-called

“argument selection” properties are usually accounted for within the list-and-rules

paradigmofGenerativeGrammar(e.g.Kageyama1993;Yumoto1996;Matsumoto1998;

Fukushima2005).InthissectionIwillsketchoutasalience-basedaccountthatdoesnot

dependonthepostulationofsemanticallyvacuousstructures.

LetusbeginwiththeobservationthattheV1sintheabovesentencesareequivalentto

Englishadverbialclausesinseveralkeyrespects.AccordingtoLangacker(1991,2008),

the following are some characteristics of subordinate clauses (of which adverbial

clausesareasubtype):

(i)Subordinateclausesserve tomodifyanotherstructure.Accordingly, the functionof

anadverbialclauseisto“qualifythemain-clauseprocesswithrespecttofactorssuchas

time,means,cause,andpurpose[...].”(Langacker2008:419)

(ii) One or more participants of the event designated by the subordinate clause will

oftenbeabsentfromtheleveloflinguisticrealization,i.e.theywillnotappearasovert

nominals.(Langacker1991:420)

(iii)Asubordinateclause isoftenmorphologicallymarkedby the lackofa finiteverb.

Suchaclausehasnotemporalprofileandisthusungrounded,i.e.itmakesnoreference

tothetimeofthespeechevent.(Langacker1991:421)

Thesepropertiesareillustratedby(28a)and(28b):

(28a) Johnmakesaliving[bystealingpuppies].(“missing”TRinsubordinateclause)(28b) Johnmakesaliving[bystealing].(“missing”TRandLMinsubordinateclause)

AsLangacker(2008:413)notes,“[a]temporalizationoftheclausalprocess–viewingit

holisticallyratherthansequentially–isonestepinthedirectionofitsnominalization.”

The main difference pertains to the level of autonomy: An atemporal relation still

contains a schematic TR/LM as part of its profile and is therefore conceptually

166

(30a) Climbingisagreatwaytostayinshape.(30b) [ClimbingMt.Rushmore]isagreatwaytostayinshape. (31a) Istayinshapebyregularclimbing.(31b) Istayinshape[byclimbingregularly].

In(30a)climbingisclearlyanominal(profilingathing).Heretheloadofsalienceison

the activity itself. Schematic participants hardly enter the picture and are therefore

relegated to the base. The sentence is about the beneficial effect of the activity –

regardlessofwhoclimbswhat.In(30b),ontheotherhand,itistheactivityofclimbinga

certain object that is claimed to be beneficial, not the act of climbingper se. In other

words, the schematic LM (elaborated byMt. Rushmore) is now salient enough to be

profiled and climbing is consequently understood as profiling a relation. Hence,

compared to (30a),climbingbecomessignificantlymore “clause-like”,while stillbeing

“thing-like” enough to function as subject of thematrix sentence.4Moving on to (31a)

and(31b),thedifferenceisquitesubtle:(31a)construesclimbingasathing5,(31b)asa

relation. Arguably, the salience of schematic participants is slightly higher in (31b),

albeitonlyminimally.In(31b),too,thefocusisclearlyontheactivityitself,leavingTR

and LM unelaborated. In this sense, the subordinate clause displays a characteristic

tendencytowardsconceptualautonomy.Thefactthatclimbingcouldbeinterpretedas

eithernominalor clausal ifweremoved themodifierregular(ly) furtherattests to the

closesemanticrelationshipbetweentherespectivevariants.6

Returning to the Japanese compounds, we encounter striking similarities. First, as

noted above, the V1 functions much like an adverbial clause in that it specifies the

manner,means,orreasonpertaining to theeventdesignatedby theV2.Secondly, like

English–ing (or the infinitival complement toV) the grammatical formof the V1, the

ren’yôkei(or“continuativeform”),hasatendencytowardsconceptualautonomy.Infact,

likeEnglish–ing,itis(I)alwaysungroundedand(II)hasasemanticvariantthatprofiles

athinginsteadofarelation.Thisisillustratedbelow:

4NotethattheTRstillremainsunelaborated.IsaschematicTRpartoftheprofilein(30b),butnotin(30a)?AtentativeansweristhatthehighsalienceoftheLMin(30b)willautomaticallyraisetheschematicTRtosomelevelofprominence.ThustheTRwouldpassthesaliencethresholdforprofiling,whilestillnotsalientenoughforelaboration.5Thenominalcharacterisindicatedbytheadjectiveregular.Asearchontheinternetrevealssimilarusages,suchastheheadline“Study:RegularRunningCanIncreaseYourLife”(http://running.competitor.com/2012/05/news/study-regular-running-can-increase-your-life_51953,retrieved16Oct.2016)6WecansaythatinsentenceslikeIstayfitbyrunning/climbing/lifting/etc.theconstructionV-ing“straddlesthefence“betweenclausalandnominal(seeTuggy’s[1993:285]remarksonpolysemy).

168

(32a) Tarô-ga Jirô-wo damashi, okane-wo nusun-da. Tarô-NOM Jirô-ACC deceive money-ACC steal-PAST ‘TarôdeceivedJirôandstole(his)money.’

(32b) Tarô-ga Jirô-wo damashi, okane-wo nusumu deshô ka.

Tarô-NOM Jirô-ACC deceive money-ACC steal COP.POL.CON Q ‘WillTarôdeceiveJirôandsteal(his)money?’

(33) Tarô-wa sagi-to damashi-de kut-teiru. Tarô-TOP fraud-and deception-INS eat-PROG ‘Tarômakesalivingbyfraudanddeception.’

Asshownin(32a)and(32b),theren’yôkeicausesatemporalization.Theclauseheaded

bydamashicontainsnoreferencetothetimeofthespeecheventandthusdependson

the second clause’s finite verb for temporal grounding. In sentence (33)damashi is a

nominal and profiles a thing, as indicated by the instrumental marker de and the

conjunction7with sagi (fraud). This semantic variant of the ren’yôkei is similar to the

nominalizer–ing(e.g.30a)inthatitreifiesaprocessandrelegatesitsparticipantstothe

base. Like a deverbal nominal in English, this variant can enter into N-N compounds

such askodomo-damashi (child’s play) ordamashi-e (trompe l’loeil) (cf. rockclimbing,

eatingcontest,etc.).

Given that atemporalization is the first step towards nominalization, and therefore

conceptual autonomy, thepolysemyof the ren’yôkei comes asno suprise. Considering

that the nominalized ren’yôkei directs the spotlight of prominence away from the

participantsandattheactivityitself,itstandstoreasonthatitsconjunctivevariant,too,

retainsthistendencytosomedegree.Inalloftheserespectstheargumentrunsparallel

totheoneregarding–ing.

On the basis of these considerations we are now in a position to account for any

“missing”participantsintermsofsalience.Insomecases,thefactthattheLMofdamashi

indamashi-toruisnotovertlyrealizedasdirectobjectcanbeinterpretedasareflection

ofitslowprominence.InthisrespecttheV1isnotunlikeitsnominalcounterpartin(33).

Observethat(33)isprimarilyaboutTarô’smethods,notabouthisvictims.

Ontheotherhand,onecanhardlyoverstatetheimportanceofdiscoursecontextwhen

talkingaboutsalience.Onmanyoccasions,thevictim’sidentitywillbeinferablewithout

explicit mention. If this is the norm rather then the exception, we can alternatively

interpret its inability to manifest as direct object as the syntactic consequence of a

7UnlikeEnglishand,theparticletocanonlylinknominals.

169

usage-basedeffect:InaccordancewiththeGriceanmaximsofquantity,informationthat

constitutes “common ground” need not be explicitly mentioned.8The same applies,

mutatis mutandis, to the TR of yuzuri in (26). Consider the following stretches of

discourse:

(34a) Tarô:Sonopasokon,saikinkattano? Tarô:Thatcomputer,didyoubuyitrecently? Hanako:Un,konoaidayasukuyuzuri-uketa. Hanako:Yeah,boughtit(offsomeone)forcheapacoupleofdaysago.

(34b) Tarô:Sonopasokon,Jirô-noyatsujanai? Tarô:Isn’tthatJirô’scomputer? Hanako:Un,konoaidayasukuyuzuri-uketa.Atarashiinokaukara,môiranaitte. Hanako:Yeah,boughtit(offhim)forcheapacoupleofdaysago.Hesaidhe’llbuyanewone,so

hedoesn’tneeditanymore.

In (34a) the V1 puts the load of salience on the manner of acquisition. While the

existenceofaschematicseller/yielderisvaguelypresupposed,itsidentityisconstrued

asirrelevant.I.e.,withregardstoitsTR,theV1’sbehaviorresemblesthatofadeverbal

nominal.9In (34b), on the other hand, the seller’s/yielder’s identity is easily inferred

fromtheprecedingutteranceanddoesnotneedtoberestated.Wecouldcomeupwith

analogous examples for ki-kuzureru, but I believe the gist of the argument should be

clear by now. In summary, when a participant remains unelaborated, it is either not

salientenough to be stated explicitly or it is already part of the discoursive “common

ground”(andthereforetoosalienttobestatedexplicitly).Asshownin(34a)and(34b),

bothcasescomplementeachother,andthuseveryutteranceneedstobeanalyzedonits

ownterms.

14.6.2.ActiveZones(35a) Hannin-ga higaisha-wo shime-koroshi-ta. Criminal-NOM victim-ACC choke-kill-PAST ‘Thecriminalchokethevictimtodeath.’

(35b) *Hannin-ga higaisha-wo shime-ta. Criminal-NOM victim-ACC choke

(35c) BUT: Hannin-ga higaisha-no kubi-wo shimeta. Criminal-NOM victim-LK throat-ACC choke ‘Thecriminalchokedthevictim’sthroat.’

8Thesecondmaximofquantitystates:“Donotmakeyourcontributionmoreinformativethanisrequired”(Grice1975:45).9Asmentionedabove,asubtledifferencebetweenclausalvsnominal-ing/conjunctivevsnominalren’yôkeiisthattheformervariantsprofileschematicparticipants,butoftenleavethemunelaborated,whereasthelattervariantsrelegateparticipantstothebase.Thus,thereisaprominence-clineforparticipantsalongthelinesof:partofthebase-->schematicallyprofiled,butunelaborated-->elaborated.

170

(36a) Sairen-ga nari-wataru. Siren-NOM sound-cross ‘Thesirenresounded(throughoutthearea).’

(36b) *Sairen-ga wataru. Siren-NOM cross

Wehavealreadyencounteredactivezonephenomenathroughoutthecasestudies(see

e.g. 12.1.3.). Recall that they serve “to accommodate the greater cognitive salience of

concreteobjectsoverabstractentities,wholesoverparts,andsoon”(Langacker1987:

272). In (35a) shime-korosu evokes the throat as the active zone of the victim (part -

wholerelation).Thevictimasawholeisdeemedmoresalientthanthebodypartandis

therefore licensedasdirectobject.(35b),however,showsthatshimeruwithoutkorosu

failstotriggerthepart-->wholemetonymicshiftandthereforedoesnotpermithigaisha

asobject.Likewise,nari-wataru in (36a)evokes the soundwavesasactivezoneof the

siren.Since thesiren is the lessabstractentity, it is– invirtueof itsgreatercognitive

salience–licensedassubject.Butwithoutnaruthesimplexwatarudoesnottriggerthe

metonymicshiftsound-->sourceand(36b)endsupinfelicitous.

14.6.3.TowardsanActiveZoneAnalysisofSomeGrammaticalV2sWhileonthetopicofactivezonephenomena,itisworthpointingoutthattheprofiling

propertiesofsomegrammaticalV2sbearanuncannyresemblancetothoseexhibitedby

so-called “raising”-constructions (see e.g. Langacker 1991, 1995). In this section

inchoativeV-dasuwillserveasanexampletoexaminetheparallels.

To briefly recapitulate, I have treated verb-verb compounds as an amalgamation of

twoprocesses,eachwiththeirownrespectiveTR/LMslots.Forexample,inthecaseof

furi-dasuIhaveclaimed(onthebasisoftheprecedingcasestudies)thatinchoative

-dasuhasasub-lexicalstateofaffairstypeTRalongthelinesofthatitrains,whiletheTR

of the V1 furu is elaborated by a thing, i.e. rain. Since the specifications for both TRs

clash and the compound as a whole can only have a single TR, the clausal subject is

chosenaccordingtoanempathyhierarchy,whichputsthingsabovestatesofaffairs.

Themostcontroversialpartof thisanalyisis, thepostulationofasub-lexicalstateof

affairs-typeTR,hingeson theargumentation inchapter10.2.6.:Here itwassuggested

thatV-dasuisbestunderstoodasavariantofthesimplexverb’saccesssense(recallthe

arguments from peceptibility and non-intentionality). This access sense of DERU is

arrivedatviatheconceptualmetaphorBECOMINGACCESSIBLEISEMERGINGFROMACONTAINER

171

with its entailments INACCESSIBLE IS IN and ACCESSIBLE IS OUT. The EXIT schema as

instantiatedbyDERUhasaschematicTRandaschematicLM:Anentity(theTR)moves

outofaCONTAINER(theLM).Under themetaphorical inchoativereading thisTR isa

stateofaffairs thatmovesoutof therealmofsensory inaccessiblity.At thevery least,

thisisthecorrectTR/LMconfigurationonthelevelofimageschematicstructure,i.e.as

pertainingtotheimageschemaEXITunderthescopeoftheaforementionedmetaphor.

Letusnowconsiderthepossibiltyofanactivezoneanalyisis.Langacker(1991:453ff.,

1995:21ff.)hasconvincinglyarguedforsuchananalysistoaccountforthephenomenon

knownas“raising”.Considerthefollowingsentences(fromLangacker1995:24):

(37) [ThatDonwillleave]SBJislikely.(38) [Don]SBJislikelytoleave.

In(37) likely takesapropositionalsubject(i.e.thatDonwillleave),whileintheraised

sentence (38) it takes a subject nominal designating a thing (i.e. Don). According to

Langacker (1995: 32), this shift is easily accounted for once we realize that the

propositional subject in (37) corresponds to the active zone of the raised sentence’s

subject in (38). As shown in figure 2 below, both sentences feature exactly the same

conceptualcontentbutdifferinregardstotheirprofilingproperties(asindicatedbythe

bold lines). In both cases, likely situates a process with respect to a region on a

probability scale. However, while (37) confers primary focal prominence (trajector

status)ontheprocessasawhole,(38)restrictsthisprominencetoDon,i.e.theprocess’

mostsalientparticipant(Langacker1995:24-25).Langackerpointsoutthatthissortof

metonymic shift is not limited to “raising” phenomena. In fact, the involvement of

processualactivezonesisnothingoutoftheordinaryatall:

Forexample,whenIsayThisbarberisfast,Idonotimplythatthebarberhimself–quapersonorphysicalobject–fallswithinacertainregiononascaleofrapidity.Itisratheracharacteristicactivityinwhichthebarber engages, such as shaving, cutting hair, or even running (the default for people in general), thatdirectlyinteractswiththescaleandisthusthesubject’sactivezonewithrespecttofast.(Langacker1991:456)

He further notes that a shift of prominence from process to participant is not

unexpected, considering that the latter –perdefault –makes for amore prototypical

subject(recalltheempathyhierarchyfromabove):

Becauseaprocessualparticipant isconceptuallyautonomousandusually lessabstract thantheprocessitself, the discrepancy between active zone andprofiled participant once again enables the spotlight ofmain-clausefocalprominencetofallonanentityofgreatercognitivesalience.(Langacker1991:456)

172

competing for subjecthood and the one with higher cognitive salience prevails in

accordancewiththeemapthyhierarchy.Themaindifference is that Ihavetreatedthe

processitselfasasub-lexicalTR(duetoitsprominencewithinthescopeofmetaphor),

whileIbelievethatLangackerwouldavoidsuchaninterpretation.Ontheotherhand,I

know of no comprehensive treatment of Japanese grammatical V2s within the

framework of Cognitive Grammar. I will therefore merely point out that this is a

potentiallypromisingareaforfutureresearch.Afterall,itstandstoreasonthatanactive

zone account couldwork for a variety of grammatical V2s, includingV-kakeru andV-

tôsu: Here, nominals likehon anduso would be considered to have focal prominence

instead of their processual active zones hon-wo yomu no and uso-wo tsuku no (see

examples[7]and[9]).

14.6.4.PartialMetaphoricalMappings(40a) Tarô-ga ji-wo kaki-nagut-ta. Tarô-NOM characters-ACC write-beat-PAST ‘Tarôwrotethecharactersinadisorderlymanner.’

(40b) *Tarô-ga ji-wo nagut-ta. Tarô-NOM characters-ACC beat-PAST

Consider the conceptual metaphor THEORIES ARE BUILDINGS, exemplified by expressions

suchasYourtheoryhasashakyfoundaton,Theargumentcollapsed,andsoon.AsLakoff

andJohnsonnote,onlycertainpartsof thesourcedomainaremappedontothetarget

domain:

ThepartsoftheconceptBUILDINGthatareusedtostructuretheconceptTHEORYarethefoundationandtheoutershell.Theroof,internalrooms,staircases,andhallwaysarepartsofabuildingnotusedaspartoftheconceptTHEORY.ThusthemetaphorTHEORIESAREBUILDINGShasa“used”part(foundationandoutershell)andan“unused”part(rooms,staircases,etc.).(LakoffandJohnson2003:52)

As (40) shows, this partial nature of metaphorical mappings can have bearing on

argumentrealization.Theaspectofnaguru(beat,hit)whichisinfocusthroughoutthe

mapping is the chaotic/disorderly/violent manner of the activity. The force dynamic

aspects ofnaguru on the other hand (i.e. energy transfer andpatient) arenon-salient

elementsof the sourcedomain that remain “unused”.The closestparaphraseof (40a)

usingnaguruwouldbesomethinglike:

174

(41) Tarô-ga ji-wo naguru yô-ni kai-ta. Tarô-NOM characters-ACC beat likemanner-DAT write-PAST ‘Tarôwrotethecharactersinabeating-likemanner.’

Thepoint is that there canbenoparaphraseof (40a) featuring theLMofnaguru (i.e.

somethingthatgetshit),becausethecorrespondingentityisexcludedfromthescopeof

themetaphoronwhichthecompoundisbased.

14.7.LexiconvsSyntax?TowardsaUnified,Schema-basedAccount

Upuntil nowwehave looked at various cases of compositional disparity and thereby

sketchedoutthefoundationsofanaccountofargumentstructurethatdoesnotviolate

thecontentrequirement(Langacker1987:53f.).Theargumentstructurephenomenawe

have encountered above are not explained in terms of procedural grammatical rules

(which are not themselves symbolic units), but rather in terms of cognitive salience.

Furthermore, we have assumed the lexicon-grammar continuum, which makes no

principled distinction between lexical and grammatical elements (see 1.2.; Langacker

1990:29).Onthisview,thedifferencebetweena“lexical”V2suchas-asaruinkai-asaru

(buy-scavange-->goaboutshoppingforsth.)anda“grammatical”V2suchas-dasu in

warai-dasu (laugh-DASU -->start laughing/burstout in laughter)pertains to thatunit’s

degreeofschematicity.Whiletheformercodesratherspecific,conceptuallyrichcontent,

thelattercodesconceptuallyleancontentofamorestructuraltype.

However,thisperspectiveisbynomeansprevalentinthecontemporarydiscussionof

Japanese V-V compounds. Instead, the vast majority of the literature assumes a

compartmentalized view of grammar with a sharp distinction between lexicon and

syntax(seee.g.Fukushima2005,Yumoto2008,Kageyama2009).Onthisview,thereare

two fundamentally different kinds of Japanese V-V compounds: “Lexical” compounds

(goitekifukugôdôshi)assembledinthelexicalcomponent(or“module”)ofthegrammar

and “syntactic” compounds (tôgôteki fukugô dôshi) assembled in the syntactic

component.Proponentsof thisdichotomyclaimthatevidencecomesfromavarietyof

linguistictests.Specifically,“lexical”and“syntactic”compoundsexhibitdistinctbehavior

inthefollowingcases(basedonYumoto2008:2):

175

--Test1:sôsuru--(42a) *Tarô-ga ason-de bakari iru no-wo mi-te, Jirô-

mo

sô shi-kurashi-

ta.

(lexical)

Tarô-NOM

play-PROG

only PROG NMLZ-ACC

see-TE

Jirô-too

so do-live-PAST

Intendedmeaning:‘SeeingthatallTarôdidwasplay,Jirô,too,idledhistimeaway.’

(42b) Tarô-ga mada hashit-te

iru

no-wo mi-te, Jirô-

mo

sô shi-tsuzuke-ta. (syntactic)

Tarô-NOM

still run-PROG NMLZ-ACC

see-TE

Jirô-too

so do-continue-PAST

‘AsJirôsawthatTarôwasstillrunning,hecontinueddoingsoaswell.’

--Test2:o-Vninaru--(43a) Shachô-wa o-asobi-kurashi ni-nat-ta. (lexical) CEO-TOP HON-play-live HON-PAST ‘TheCEOidledhistimeaway.’

(43b) *Shachô-wa o-asobi ninari-kurashi-ta. (lexical) CEO-TOP HON-play HON-live-PAST Intendedmeanding:sameas(36a)

(44a) Seinsei-wa tegami-wo o-kaki ninari-hajime-ta. (syntactic) Teacher-TOP letter-ACC HON-write HON-begin-PAST ‘Theteacherbeganwritingaletter.’

(44b) *Sensei-wa tegami-wo o-kaki-hajime ni-nat-ta. (syntactic) Teacher-TOP letter-ACC HON-write-begin HON-PAST Intendedmeaning:sameas(37a)

-->lexical:[o-V1-V2ninaru]-->syntactic:[o-V1ninari-V2]--Test3:Passivization--

(45b) *Kizôhin-ga mot-are-yot-ta. (lexical) Donation-NOM hold-PASS-drawnear-PAST Intendedmeaning:sameas(46a)

(46a) Tegami-ga kak-are-tsuzuke-ta. (syntactic) Letter-NOM write-PASS-continue-PAST ‘Theletterwascontinued.’

(46b) *Tegami-ga kaki-tsuzuker-are-ta. (syntactic) Letter-NOM write-continue-PASS-PAST Intendedmeaning:sameas(47a)

-->lexical:[V1-V2-PASS]-->syntactic:[V1-PASS-V2]--Test4:Lightverbconstruction(Nsuru)--(47a) *Kinyû shi-komu ; *Jisan shi-yoru (lexical) Entry do-[inwardsmovement] Bringing do-drawnear Intendedmeaning:‘Fillin(aformetc.)’Intendedmeaning:‘bringalong’

(45a) Kizôhin-ga mochi-yor-are-ta. (lexical) Donation-NOM hold-drawnear-PASS-PAST ‘Adonationwasbrought(along).’

176

(47b) Ensetsu shi-owaru ; Tôkan shi-wasureru (syntactic) Speech do-finish Mailing do-forget ‘Endaspeech’‘forgettodispatchsth.’

Yumoto (2008: 2) concludes: “The fact that Japanese V-V compounds can be

distinguished by clear-cut formal criteria provides strong evidence in support of the

modularityofmorphology.“

Thisinterpretationis,ofcourse,inconsistentwithamajorguidingassumptionofthis

thesis, namely the view that grammar is exhaustively characterized as a structured

inventoryof symbolicunits. It is therefore important to reconcile theabovedatawith

thebasicprinciplesofCognitiveLinguistics. Ibelieve that this is indeedpossible ifwe

considerwhatabottom-up,usage-basedapproachtogrammarentails.Letusbeginby

looking at frequency effects andentrenchment.10According toKageyama (2009:522),

there are approximately thirty syntactic V-V compounds in Japanese. He lists the

followingexamples:

V-hajimeru(begintoV),V-oeru(finishV-ing),V-tsuzukeru(continueV-ing),V-kakeru(beabouttoV),V-sokoneru(failtoV),V-kaneru(cannotaffordtoV),V-wasureru(forgettoV),V-naosu(Vagain),V-okureru(belateinV-ing),V-sugiru(Vexcessively)

As Kageyama notes, these compounds are all highly productive, as opposed to lexical

compounds,which“lackfullproductivity”(2009:522).Ifwelookatthesemanticsofthe

aboveV2s,thereisnothingmysteriousatallabouttheirhighlevelofproductivity.Take

V-hajimeruforinstance.TheV2hajimeruplacesvirtuallynosemanticrestrictionsonits

V1exceptthatitbeatemporallyextendedevent.Or,simplyput,-hajimeruiscompatible

with just about anything thathas abeginning.And this, in turn, results innear100%

productivity. However, one should be mindful of the fact that this is a bottom up

process: In real world communication -hajimeru is suffixed to various individual V1s

overthecourseofmyriadsofdifferentusageevents,resultingintheentrenchementof

compounds such as tabe-hajimeru(begin to eat),aruki-hajimeru(begin towalk),kiki-

hajimeru (begin to listen), odori-hajimeru (begin to dance), and so on. Through

continuous usage each compound becomes entrenched in the cognitive system, i.e. it

achieves unit status (see Langacker 1987: 57ff.). Once a large number of these

compoundsachieveunitstatus,thepartiallyfilledschema[V-hajimeru]will–asaresult

of inductive reasoning – achive unit status aswell. Keep inmind that unit status is a

10ThefollowingargumentsassumethatLangacker’sprinciplesofconventionandusage(1987:65f.)aswellasfullandpartialsanction(1987:66ff.)apply.

177

matter of degree. Since -hajimeru appears as V2 in hundreds of compounds, the

construction [V-hajimeru] is firmly entrenched in the cognitive system.Now, compare

this to -naguru.While thecompoundkaki-naguru (write-beat -->write inadisorderly

manner) has clearly achieved unit status through repeated usage (i.e. in virtue of its

relatively high token frequency), the partially filled schema [V-naguru] can hardly be

saidtobecognitivelyentrenched.Thereasonforthisisthat-nagurudoesnotappearin

theV2 slotof anyother compounds.Afterall, the semanticsof -naguru aremuch less

schematicthanthoseof-hajimeru,andthustheformerisalotmorespecificinitschoice

of V1. In conclusion, then, the cognitive entrenchment/unit status of lexically filled

constructionscanbemeasuredintermsoftokenfrequency,whiletheentrenchmentof

schematicconstructionsisbestmeasuredintermsofhownumeroustheirinstantiations

are,i.e.bytypefrequency.

Once a schema is well-entrenched, it will, in turn, serve as a template to sanction

specificinstancesintop-downfashion.I.e.,thecompoundyomi-hajimeru(begintoread)

is considered “well-formed”, because the schema [V-hajimeru] is a firmly entrenched

unit. In contrast, *nuri-naguru would be judged as “ill-formed”, since the schema [V-

naguru]isnotawell-entrenchedunitatall:

(48) *Tarô-ga penki-wo nuri-nagut-ta. Tarô-NOM paint-ACC apply-beat-PAST Intendedmeaning:‘Tarôappliedthepaintinadisorderlymanner.’

To be sure, it is not inconceivable for nuri-naguru to attain unit status. After all, the

construction is based on an already well-entrenched compound, and a speaker

community(orsomesub-community)might findtheexpressionamusingoruseful for

some reason or another. Through repeated usage the novel expression would then

graduallybecomeentrenchedaswell.ButasTuggy(2005:254)notes,extensionsofthis

type are “norm-bending and quite creative”. So as a rule of thumb, instances are

sanctionedbywell-entrenchedschemas.Conversely,ifaschemaisnotwell-entrenched

itislesslikelytosanctionaparticularinstance.Ontheotherhand,ifaunitisextremely

wellentrencheditisnotonlyfittosanctioninstancesinvirtueoffullschematicity(e.g.

[edible thing – soup] --> tomato soup), but also more likely to give rise to novel

expressionsinvirtueofpartialschematicity([ediblething–soup]-->primordialsoup).

Iwillnowarguethatthisinterplaybetweenentrenchmentandsanctionisresponsible

fortheresultsin(42)-(47).Onceagain,considertheexampleslistedabovebyKageyama.

178

Observethatallofthese“syntactic”compoundschemasareextremelywellentrenched.

Schemaslike[V-hajimeru],[V-tsuzukeru]or[V-wasureru]areinstantiatedbymyriadsof

different lexical items.For instance, [V-tsuzukeru] effortlessly servesasa template for

[hashiri-tsuzukeru] (continue to run), since [V-tsuzukeru] is deeply entrenched in the

cognitivesystemandstandsinarelationoffullschematicityto[hashiri-tsuzukeru].But

whatabout therelationbetween[V-tsuzukeru]and,say, the lightverbconstruction[N

suru]? This is the aforementioned case of partial schematicity. On the one hand, [V-

tsuzukeru] requires an antecedent of the type [V], which clearly clashes with the

specifications of [N suru]: [V] is a simplex verb whereas [N suru] is a composite

constructionconsistingofanounandthelightverbsuru.Ontheotherhand,[V]and[N

suru]arequitesimilar insofaras theyareboth instancesof the [PROCESS]schema. In

otherwords,itonlytakesasmallgeneralizationtolicense[N-suru]inthe[V]-slotof[V-

tsuzukeru],therebygivingrisetotheschema[Nshi-tsuzukeru].Inanalogousfashionthe

sameholdstruefor[sôsuru], [o-Vninaru],andthepassiveconstruction.Byextending

the category [V] to all processes and not just simplex verbs we arrive at the meta-

schema[PROCESS-tsuzukeru].

Now compare this to what adherents of the dichotomy call “lexical compounds“.

Schemasforthesecompoundscomeindifferentdegreesofentrenchment.Onthehigh

endofthespectrumwehavewell-entrenchedschemaslike[V-komu](asinhairi-komu

[enter])witha considerablenumberof instances.On the lowendof the spectrumwe

have schemas like [V-asaru] or [V-naguru]with only one or two instances (kai-asaru

[buy-scavange --> go around shopping for sth.]; yomi-asaru [read-scavange --> read

whatonecangetone’shandson];kaki-naguru[write-beat-->writedisorderly]).These

latter schemas are notwell-entrenched at all and lack unit status. They are therefore

unfittosanctionfurtherinstancesdespitearelationoffullschematicity.Naturallythen,

they are even less fit to sanction extensions of the schema via partial schematicity.

Simplyput,[V-asaru]cannotbeextendedto[PROCESS-asaru],since[V-asaru]doesnot

evenhaveunitstatus.Thisis,ofcourse,anextremeexample.[V-komu]isamuchbetter

candidate for unit status, but still not nearly aswell-entrenched as [V-tsuzukeru], [V-

hajimeru],etc.–andthusstillunfittogiveriseto[PROCESS-komu].

Fromausage-basedperspective,“lexical”compoundsaremoredeeplyentrenchedon

thenon-schematiclevelthanontheschematiclevel(e.g.kai-asaruisbetterentrenched

than[V-asaru]),whereas“syntactic”compoundstendtobemoredeeplyentrenchedon

theschematiclevel(e.g.[V-hajimeru]isbetterentrenchedthanitsinstances).Itiseasy

179

toseehowthisaffectscomposition.Toillustrate,considerkaki-hajimeru(begintowrite)

and the honorific construction [o-Vninaru]. Betweenkaki-hajimeru and [V-hajimeru],

the latter is much better entrenched. Consequently, it makes more sense for [o-V ni

naru] to elaborate the V-slot of [V-hajimeru], then for the less well-entrenched kaki-

hajimeru to elaborate the V-slot of [o-V ni naru]. That is, the composition works as

follows:TheverbkakuelaboratestheVslotof[o-Vninaru],yieldingo-kaki-ninaru.The

result,o-kakininaru,thenelaboratestheVslotof[V-hajimeru](forthis[V-hajimeru]is

extended to [PROCESS-hajimeru]). Note, that on this account the fully elaborated

compoundkaki-hajimerudoesnotevenpartake inthecomposition.Now,contrast this

withthecaseofkai-asaruand[o-Vninaru].Betweenkai-asaruand[V-asaru]theformer

ismuchbetter entrenched than the latter.Thus, itmakesmore sense forkai-asaru to

elaboratetheVslotof[o-Vninaru]thanfor[o-Vninaru]toelaboratetheV-slotofthe

non-unit[V-asaru].

Letussummarize:Themorphological“inseparability”ofso-calledlexicalcompounds

isstraightforwardlyaccountedforinlightoftheirlowschematicity.Acompoundsuchas

kai-asuru is filledwith lexicalmaterial inboth theV1- andV2-slot (bykauandasaru,

respectively).Incontrast,thepartiallyfilledconstruction[V-asaru]cannotlicenseother

constructionsdueto its insufficientdegreeofentrenchment.E.g., since[V-asaru] lacks

unitstatus,itsV1slotcannotbeelaboratedby,say,thelightverbconstruction[Nsuru]–

hencetheinfelicityof[Nshi-asaru]. Inthecaseof“syntactic”compounds,ontheother

hand, the partially filled construction is better entrenched than the fully elaborated

construction.E.g.,[V-hajimeru]isextremelywell-entrenched;arguablymoresothanthe

fullyelaborated[tabe-hajimeru].Consequently,“syntactic”compoundscanlicenseother

PROCESS-type constructions (such as the light verb-, the honrific-, or the passive-

construction)intheirV1-slotbywayoffullorpartialsanction–hencethefelicityof,say,

[N shi-hajimeru]. Of course, all of this is still rather programmatic. But the above

considerations show that the linguisticdatapresentedbyproponentsof the lexical vs

syntacticdichotomydoesnotcompelustoamodularview.

180

15.EmbodimentandtheScopeofMetaphorinGermanandJapanese

AtseveralpointsduringthecourseofthisthesisIhaveconsideredcross-linguisticdata.

Recall, for instance, the discussion of -kakaru and German an. A cross-cultural

perspective on embodied experience certainly seems interesting enough to warrant

closerinspection.Atthesametime,acomprehensivetreatmentofthetopicwouldbefar

beyondthescopeofthisthesis.Iwillinsteadrestrictmyselftoasmall-scalecomparative

studyofGermanandJapanese.Specifically,Iwillconsiderthreesourcedomainsdirectly

based in embodied experience: weight, edge properties, and surface properties. The

question, then, iswhether thesewillhavesimilarordifferentscope inboth languages.

Willwe end upwith the samemetaphors in German and Japanese? Since our source

domainsaredirectlyembodied,andgiventheuniversalnatureofhumanphysiology,we

shouldexpectonlysubtlevariance.

Ihavechosenmetaphorasatoolofanalysisforpracticalreasons.Eachsourcedomain

willlikelycorrespondtoanoverseeablenumberoftargetdomains.Thus,theamountof

data remains manageable, whereas “raw” image schemas such as CONTAINER,

CONTACT,etc.correspondtocountlesslinguisticexpressions.

Beforeproceeding,Ishouldmentionthevastbodyofexistingresearchonthetopicof

metaphorandculture.1Ofthesestudies,asignificantnumberfocusonspecificemotion

conceptssuchasanger(e.g.LakoffandKövecses1987;Munro1991;Matsuki1995;Yu

1995) or love (e.g. Kövesces 1988, Yang 2002). These works share a common

methodology in that they start with the target domain and work their way towards

varioussourcedomains.I.e.,thetherangeofmetaphorisaprimaryconcern.Incontrast,

Ihaveoptedfortheoppositedirectionfromdirectlyembodiedsourcetotarget,inorder

toexaminecross-linguisticvariance(orlackthereof)inmetaphoricalscope.

15.1.(I)TheWeightScale:HeavyandLight

15.1.1.(Ia)EFFORTFULACTIVITYISHANDLINGHEAVYOBJECTS

ConsiderthefollowingGermansentences:

(1) DasisteinschweresStückArbeit.

Thisisahard[heavy]pieceofwork.

(2) Wennwirunsanstrengen,könnenwirdieAufgabestemmen.(colloq.)

1SeeKövecses(2005)forarepresentativebook-lengthstudy.

181

Ifweworkhardwecanfinish[liftup]thetask.

(3) ErhörtschwerundkanndichnurmitMüheverstehen.

Hishearingisbad[‘hehearsheavily’],sohehastroubleunderstandingyou.

(4) DasRätselwarschwerzulösen.

Thepuzzlewashard[heavy]tosolve.

(5) IchhabeRückenschmerzenundkannnurleichteArbeitverrichten.

Ihaveabackacheandcanonlyperformlightwork.

(6) Wenndulautersprichst,kannichdichleichterverstehen.

IfyouspeakupIcanunderstandyoubetter[lighter].

(7) DieseMathematikaufgabeistleicht.

Thismathproblemiseasy[light].

Thismetaphorisexperientiallygroundedinametonymicalrelationshipbetweensource

and target domain.Dealingwithheavyobjects is a prototypical instanceof exercising

effort,givingrisetoastrongexperientialcorrelationbetweenthetwo.Consequently,the

source concept of weight is extended to other forms of effortful activity, e.g. sense-

perceptual (3,6) ormental (4,7). In (8)we encounter a seemingly similar expression

fromJapanese:

(8) Omoi shigoto-wo makas-are-te, sutoresu-ga tamaru.

heavy work-ACC entrust-PASS-TE stress-NOM accumulate

‘I’mentrustedwithhighresponsibilityjobsandstressisaddingup.’

However,notethattheweightcodedbyomoiintheabovesentenceisnotcoupledwith

the target domain of EFFORT but rather with the target domain of PSYCHOLOGICAL

BURDEN(seealso[16]below).

Duetothelackofexpressionscorrespondingto(1)-(9)weconcludethatthemetaphor

EFFORTFULACTIVITYISHANDLINGHEAVYOBJECTSdoesnotseemtoexistinJapanese.

15.1.2.(Ib)ABSTRACTBURDENSAREPHYSICALWEIGHTS

(09) GroßeVerantwortunglastetaufihrenSchultern.

Greatresponsibilityrestsonhershoulders.

(10) ErneigtzuSchwermut.

Heispronetodepression.

(11) IchmöchteniemandemzurLastfallen.

Idon’twanttobeaburdentoanyone.

(12) MitihrerTathatsieSchuldaufsichgeladen.

Sheincuredguilt[‘piledguiltontoherself’]bywhatshedid.

182

(13) Ein(e)leichte(s)Verletzung/Strafe/Erkältung/Vergehen

Aminor[light]injury/punishment/cold/offense

(14) WasfüreineErleichterung!MirfällteinSteinvomHerzen!

Whatarelief!That’saloadoffmychest!

(15) Omoi sekinin-wo seou

Heavy responibility-ACC shoulder

‘Tobearaheavyresponsibility’

(16) (Sekinin-no) omoi shigoto-wo makas-are-te, ki-ga omoi

(Responsibility-NOM) heavy work-ACC entrust-PASS-TE mind-NOM heavy

‘I’mbeingleftwithhigh-responsibilityjobsandIamfeelingdepressed.’

(17) Tanin-no onimotsu-ni nari-taku-nai.

Others-LK baggage-DAT become-DES-NEG

‘Idon’twanttobecomeaburdentoothers.’

(18) Hannin-ga jûhan-wo okashi-te, jûbatsu-wo uke-ta.

Criminal-NOM seriouscrime-ACC commit-TE severepunishment-ACC receive-PAST

‘Theperpetratorcommitedaserious[heavy]crimeandreceivedsevere[heavy]punishment.’

(19) Karui byôki/hanzai/sekinin

Light illness/crime/responsibility

(20) Toriaezu karui kimochi-de yat-temi-te.

Fornow light feeling-INS do-try-IMP

‘Fornow,trydoingitwithouttakingittoseriously.’

Asevidencedby theabovesentences, thescopeof thismetaphor isnearly identical in

German and Japanese. Both languages express psychological, emotional, and somatic2

burdens such as sadness and responsibility in terms of physical weight. Note at this

pointthatomoishigotoin(16)emphasizestheweightofsocialobligationandburdenof

expectation,whereaseinschweresStückArbeit in (1)merely emphasizes the required

degree of effort. Furthermore, German and Japanese share a common folk theory of

justiceasbalance(seeJohnson1990:90)inthedomainsoflawandmorality:Theweight

of the punishment/atonement must match the weight of the crime/guilt in order to

restoreequilibrium.

As for thegroundingof themetaphor,we canobserveparallelsbetween the effects

physicalweightandabstractburdenshaveontheir“bearer“.Notethatcarryingaheavy

objecthastwoimportantimplications:

• Incapacitation:Theweightlimitsthebearer’sabilitiesandpotentialforaction.

2Thecharacterizationofsomaticburdenssuchasinjuryorsickness(e.g.[15],[21])as“abstract“mayseemdubious.However,thepointisthatinjuryandsicknessdonotnecessarilyinvolvethesensationof

physicalweightonthebody.

183

• Exhaustion:Theweightwillincrementallyaffectthephysicalconditionofthe

bearer.

Nowcomparethistothe“symptoms”ofdepressionandresponsibility.

Depression

• Incapacitation:Adepressedpersonisunabletousehis/herabilitiesand

potentialtothefullest.

• Exhaustion:Adepressedpersonwilloftenfeelfatiguedandunmotivated.

Responsibility

• Incapacitation:Beingresponsibleentailsself-restraint.(Inabilityto“doasyou

please”)

• Exhaustion:Notlivinguptoone’sresponsibilitiesentailsnegativeconsequences.

Awarenessoftheseconsequencescanbepsychologicallyexhausting.

Thus,depression,responsibility,guilt,etc.arelikephysicalweightinsofarastheycause

incapacitation and exhaustion within their respective domains. In virtue of these

correspondencesthelattercanserveassourcedomainfortheformer.

15.1.3.(Ic)INTENSITYISWEIGHT

IfABSTRACTBURDENSAREPHYSICALWEIGHTS,thentheintensityoftheburdenisproportional

to theamountofweight.Wehaveseenthisentailmentatworkabove inbothGerman

andJapaneseinexpressionssuchasschwere/leichteKrankheitandomoi/karuibyôki.In

thecaseofGerman,itisinterestingtoobserve,however,thattheintensityentailmentof

weight has been extended from the domain of abstract burdens to themore general

domainofnegativelyevaluatedphenomena:

(21) Erhatsichschwerbetrunken.

Hegottotally[heavily]drunk.

(22) Allewarenschwerenttäuschtvonihm.

Everyonewasexremely[heavily]disappointedofhim.

AlthoughIsupposeitwouldstillbepossibletoview(21)and(22)asinstantiationsof

the abstract burden sense, the usage of schwer seems to have a more general

augmentativefunctioninthesecases.Thisbecomesevenmoreevidentinexampleslike

thefollowing:

(23) SieistschwerinOrdnung.

She’sarealtrouper[‘Sheisheavilydecent’].

184

(24) AllewarenschwerbeeindrucktvondieserLeistung.

Everyonewasextremely[heavily]impressedbythisaccomplishment.

(25) Dasmöchteichdochschwerhoffen!

Isure[heavily]hopeso!(utteredasawarning)

In these rather colloquial expressions schwer already functions as an augmentative

adverbalongthelinesofvery.Only(25)showstracesoftheabstractburdensenseand

itsnegativeconnotations,asthesentenceisusuallyutteredasathreatorwarning.Note

inpassingthatitisnotuncommonfornegativelyevaluatedcontentwordstotakeonan

augmentativefunction,especiallyincolloquialspeechstyles:

(26) Sickcar,dude!

(27) Kono konbini-no poteeto, yabai umai!

This conveniencestore-LK fries dangerously tasty

‘Thefriesatthisconveniencestorearethebomb!’

Insummary,thegrammaticalizationpathforschwerinGermancanberoughlysketched

outas follows,wherebytheoriginalmeaningofphysicalweightgradually fadesout in

favorof the intensityaspect: (intensityof)abstractburdens --> intensityofnegatively

evaluatedphenomena-->usageasaugmentativemarker.

15.1.4.(Id)IMPORTANCEISWEIGHT

(28) SeineStimmehatinderParteigroßesGewicht.

Hisvoicecarriesgreatweightwithintheparty.

(39) WirmüssenRisikenundVorteilegegeneinanderabwägen.

Wemustconsider[weigh]boththerisksandbenefits.

(30) DieseFaktorenfallennichtweiterinsGewicht.

Thesefactorsareirrelevant[donotcarryanyweight].

(31) Kono gakkô-wa bunkei-yori rikei-ni omoki-wo oku.

This school-TOP humanities-ABL sciences-DAT weight-ACC put

‘Thisschoollaysemphasisonthesciencesratherthanthehumanities.’

(32) Keizai-wo jûshi[‘heavy’-‘view’] shi-ta atarashii seisaku

Economics-ACC importance do-PAST.ATT new policy

Anewpolicyfocusingoneconomics

Earlier,wehave seen that there is a strongexperiential correlationbetweenhandling

heavy objects and exercising effort. In a similar vein, we can hypothesize that

importance andweight are tied inhumanexperience aswell.Heavyobjects generally

185

requiremoreresources,suchasforce,attention,andtime,tobedealtwith.Forexample,

movingaheavyrockoutofthewaywillrequireacertainamountofstrength,possibly

theapplicationoftools,andsoforth.Inshort,thetaskwillmakeitselfmoresalientvis-a-

visthehandlingofalight-weightobject.(Inthissense,theheavyobject“cannotbetaken

lightly.”) The metaphor is prominent in both German and Japanese and often

instantiatedbyparallelexpressionssuchasSchwer=punkt/jû=ten(“heavy=point”

-->emphasis).

15.1.5.(Ie)RESPECT/DIGNITYISWEIGHT

The fact that importance is conceptualized as weight has implications for the social

domainaswell.Someonewhosefunctionorstatusinsocietyisdeemedimportantisalso

more likely to be viewed as respectable and dignified.Metonymically, this evaluation

carriesover to thatperson’sactions, thoughts,mannerofspeech,andsoon.Similarly,

certain abstract entities such as ideals ormoral values are notmerely important, but

commandrespectfromasocialperspective.

(33) Yamada-sensei-wa omomi-no aru kata desu.

Yamada-teacher-TOP weight-NOM exist.ATT person COP.POL

‘Mr.Yamadacarriesanairofdignityabouthim.’

(34) Shachô-ga omoomoshii[omoi=heavy] kuchô-de ensetsu-wo hajime-ta.

CEO-NOM solemn tone-INS speech-ACC begin-PAST

‘Inasolemntone,theCEObeganhisspeech.’

(35) Kojin-no kenri-wo sonchô[‘respect-heavy’] suru.

Individual-LK rights-ACC value do

‘Tovaluetherightsoftheindividual’

Ontheotherendofthescalelackofrespectisconceptualizedaslackofweight:

(36) Hanako-ga Tarô-wo keibetsu[‘light-disregard’]-no me-de mi-teiru.

Hanako-NOM Tarô-ACC contempt-LK eye-INS view-PROG

‘HanakoviewsTarôwithcontempt.’

It is interesting to observe that thismetaphor seems to bemuchmore prominent in

JapanesethaninGerman,wherelinguisticinstantiationssuchasWürdenträger(“bearer

ofdignity”-->dignitary)arequiterare.

186

15.2.(II)EdgeProperties:SharpandDull

15.2.1.(IIa)SynaestheticMappings

(37) einscharfesChili

aspicy[sharp]chili

(taste)

(38) einscharfer/stechenderGeruch

apungent[sharp/piercing]smell

(scent)

(49) einscharfesZischen/eindumpfesGeräusch

asharphissing/adullnoise

(sound)

(40) einscharfes/stumpfesLicht

astrong/dulllight

(vision)

(41) shita-wo sasu aji

tongue-ACC pierce.ATT taste

‘aspicytaste’

(42) hana-wo sasu nioi

nose-ACC pierce.ATT smell

‘apungentsmell’

(43) surudoi/nibui oto

‘asharp/dull sound’

(44) surudoi/nibui hikari

‘aglaring[sharp]/dull light’

Synaestheticmetaphorischaracterizedbymappingsfromonesenseperceptualdomain

to another sense perceptual domain. Some metaphors are possibly grounded in

functional similaritiesbetween the respective senseperceptual faculties. For example,

eatingahotchilipeppercanresult inapain-likesensationnotdissimilartothetissue

damagecausedbyasharpobject.Thatis,someoneeatingspicyfoodmightexperiencea

sensationasifpiercedorcutbyasharpobject.Suchphenomenaarecloselyrelatedto

Sadamitsu’s (2001) Co-occurrence Condition of Sensations which, based on Lehrer

(1978),holdsthat“[t]hemappingbetweenthesensorymodalitieswhichcanco-occuris

preferabletothatbetweenthosewhichcannot”(Sadamitsu2001:126).Inotherwords,

mappingsbetweendomainssuchastouch-->tasteandtaste-->scent3aresocommon,

because the respective sense perceptual faculties often function in unison. However,

sinceeatingspicyfooddoesnotactuallyinvolvetissuedamagecausedbyasharpobject,

onemighthesitatetospeakofastrictexperientialcorrelationinGrady’s(1997a,1997b)

sense(see4.1.).

3ForadiscussionofdirectionalityseealsoYamanashi(1988)andShen(1997).

187

Note that the extensions to thedomainsof soundandvision likely involve ahigher

degreeofsemanticbleachingthantheaforementionedextensionstotasteandscent.For

example, a sharp object and a glaring light, despite causing different sensations and

being perceived in differentways, are still broadly similar in that both involve sense

dataofunpleasantintensity.

15.2.2.(IIb)UNPLEASANTINTENSITYISSHARPNESS

(45) Erversuchte,derDiskussiondieSchärfezunehmen.

Hetriedtoreducetheseverity[sharpness]ofthediscussion.

(46) DeinespitzenBemerkungensindunnötig.

Yourpointedremarksareunnecessary.

(47) ScharfeMunition/EinscharferHund

armed[sharp]munition/avicious[sharp]dog

(48) Surudoi goki-de aite-wo ii-makasu

Sharp tone-INS opponent speak-defeat

‘Toarguedownone’sopponentinasharptone’

(49) Toge-no aru iikata-wo suru

Thorn-NOM exist.ATT mannerofspeaking-ACC do

‘Touseharshlanguage.’

(50) Surudoi metsuki-no hito

Sharp gaze-LK person

‘Apersonwithasharpgaze’

Intheprevioussectionwehaveseenthattheconceptofsharpnesshasbeenextendedto

include general unpleasant intensity in the sense perceptual domain. As the above

examplesshow,thissensecanbefurtherextendedtoapplytomoreabstractdomainsas

well.InbothGermanandJapanesethemetaphorseemstoshowanaffinitytowardsthe

domain of verbal expression, often referring to a potentionally hurtful manner of

communication (and thereby piggybacking on the metaphor EMOTIONAL DISTRESS IS

PHYSICAL INJURY). Some applications, however, go beyond this central aspect. The

intensity in (47), for instance, refers to the state of being potentially dangerous. (50)

implies that the conceptualizer is unpleasantly affected on an emotional level, but

withoutreferencetotheverbaldomain.

188

15.2.3.(IIc)PRECISIONISSHARPNESS

(51) DieKonturensindgestochenscharf.

Theconturesarerazor-sharp.

(52) MitihrenscharfenAugenkannsiealleserkennen.

Withhersharpeyesshecanseeeverything.

(53) WirmüssendieUnterlageneinerscharfenPrüfungunterziehen.

Wemustsubjectthedocumentstoasevere[sharp]audit.

(54) Kare-no shinkei-ga hari-no yô-ni togat-teiru.

3S.M-LK nerves-NOM needle-LK likemanner-DAT becomesharp-RES

‘Heisextremelyperceptive.’

(55) Surudoi kansatsuryoku

Sharp observationskills

(56) Kankaku-no surudoi/nibui hito

Senses-LK sharp/dull person

‘Apersonwithsharp/dullsenses’

The experiential correlation here is that sharp objects lend themselves to precise

operations.Forexample,theuseofscissors,knifes,etc.astoolsusuallyresultsinclearly

demarcated boundaries. This, in turn, entails ease of distinction. And the better our

ability to distinguish becomes, the more likely we are to make correct judgements.

Consider(51):Themostsalientaspectofsharpcontours isaclear-cutdemarcationof

boundaries.Thismeansthatweareinanidealpositiontodifferentiatebetweenfigure

andgroundinavisualscene.Asharpphotowillleavenodoubtastowhereoneobject

ends and another one begins. The very same effect (i.e. the ability to make pricise

distinctions)isachievedbyhighvisualacuity(see52)and–viaextensiontotheother

faculties–bysenseperceptualacuity ingeneral (e.g.54,56).Similarly, a severeaudit

(see 53) entails making precise distinctions between relevant and irrelevant

information.

15.2.4.(IId)INTELLIGENCEISSHARPNESS

(57) IhrVerstandistscharfwieeinSkalpell

Hermindisrazor-sharp[sharplikealancet]

(58) EinemesserscharfeAnalyse(bymetonymy:RESULTforINSTRUMENT)

Arazor-sharpanalysis[Ananalysisassharpasaknife]

(59) Tarô-wa zunô-ga surudoi.

Tarô-TOP brain-NOM sharp

‘Tarôhasasharpmind.’

189

(60) Hanako-wa nakanaka-no kiremono[kiru=tocut] desu.

Hanako-TOP quite-LK brilliantperson COP.POL

‘Hanakoisquitebrilliant.’

BothGermanand Japaneseseem to sharea folk theoryaccording towhichANALYTICAL

THINKING IS DISASSEMBLING COMPLEX OBJECTS. Consider, for instance, the following

expressionsfromGerman:

(61) EinProbleminseineBestandteilezerlegen

Tobreakdownaproblemintosmallercomponents

(62) DiewesentlichenBausteinederTheorie4

Themajorcomponents[building-blocks]ofthetheory

In the case of Japanese, the most striking example is probably the verb wakaru

(understand, comprehend) which is etymologically related towakareru (divide, split

into).

Thus, if analytical thinking is conceptualized as decomposition, it follows by

metaphorical entailment that an able mind is a sharp instrument (since sharp

instumentsareidealfordisassemblingobjects).

15.2.5.Excursion:OverlapofMetaphorsinaSingleExpression

It is interesting to observe that some expressions seem to instantiate multiple

metaphorsatonce.Recall(53)fromabove:

(53) WirmüssendieUnterlageneinerscharfenPrüfungunterziehen.

Wemustsubjectthedocumentstoasevere[sharp]audit.

I have categorized this under theprecision reading, sincemost native speakers agree

that accuracy and precisionare themain aspects here. However,we can hardly deny

thattheothertargetdomainsdiscussedaboveplayaroleaswell.Theexpressionscharfe

Prüfung, best translated as severe audit, at least implies some amount of unpleasant

intensity. And since such an activity requires analytical rigor, the target domain of

intelligenceisactivatedaswell.Asimilarconflationcanbenotedregardingexpressions

suchas(57)and(59):

(57) IhrVerstandistscharfwieeinSkalpell

Hermindisrazor-sharp

4SeealsoGrady(1997a,1997b)ontheconceptualmetaphorTHEORIESAREBUILDINGS.

190

(59) Tarô-wa zunô-ga surudoi.

Tarô-TOP brain-NOM sharp

‘Tarôhasasharpmind.’

As Ihavearguedabove, the INTELLIGENCEISSHARPNESSmetaphor isanentailmentof the

metaphor ANALYTICAL THINKING IS DISASSEMBLING COMPLEX OBJECTS. But at the same time,

there are metaphorical links between sense perception and mental activity in both

GermanandJapanese:

(62) IchsehemomentankeineLösung.

Idon’tseeasolutionatthemoment.

(63) DasklingtnacheinemgutenPlan.

Soundslikeagoodplan.

(64) Anata-no keikaku-wa subete omitôshi da.

2S-LK plan-TOP all see-through COP

‘I’vecompletelyseenthroughyourplan.’

(65) Keiji-ga hannin-wo kagi-dashi-ta.

Detective-NOM perpetrator-ACC sniff-DASU-PAST

‘Thedetectivefoundoutwhotheperpetratorwas.’

These are all instances of a higher-levelmetaphorical system called theMind-as-Body

Metaphor(Sweetser1991:28ff.).Thepointisthatexpressionslike(57)and(59)canbe

seenasinstancesofboththeprecisionandtheintelligencereading.I.e.,theMIND-AS-BODY

metaphor and the ANALYTICAL THINKING IS DISASSEMBLING COMPLEX OBJECTS metaphor

simultaneously construe the mind as a sharp object. In fact, it might be possible to

subsume the intelligence under theprecision reading.However, I have chosen tokeep

themseperateheretodrawattentiontoarelativelyprominentfolkmodelofanalytical

thinking.Eitherway,theexactdistinctionisoflittleconcernforthepresentpurpose.

15.3.(III)SurfaceProperties:SmoothandRough

15.3.1.(IIIa)ABSTRACTREFINEMENTISSMOOTH,LACKOFABSTRACTREFINEMENT

ISROUGH

ThesemappingsareentailmentsofthehighlyschematicontologicalmetaphorABSTRACT

SUBSTANCE ISRAWMATTER – a prominent consequence ofwhich is the view of ABSTRACT

DEVELOPMENT AS AN ARTISANAL PROCESS. For instance, humans and their skills are

understood as being shaped by external and internal forces in a teleological manner

towardssomedesiredendstate.ThisisevidencedbyGermanexpressionssuchassich

191

bilden (lit. to form,shape,buildoneself),whichrefers to theprocessofself-education,

including personal maturation aspects. Similarly, in Japanese we have phrases like

seishin-wokitaeru(lit.forgeone’smind).

Given the above, it is not hard to seewhy abstract refinement is conceptualized as

smooth,andlackthereofasrough:Therawuntreatedmaterialwhichmarksthestarting

pointofthedevelopmentalpathiscoarseandrough,andgraduallybecomessmoothand

refined,asitissubjectedtotheartisanaltreatment.Toillustrate,considerthefollowing

examples.

Mannersandpersonality:

(66) ErbenahmsichwieeinungehobelterKlotz.

Heconductedhimselfuncouthly[likearough-hewnbrick].

(67) SiebeindrucktedurchgeschliffeneManieren.

Herpolishedmannersleftanimpression.

(68) arappoi kotobazukai

rough language

(69) Seikaku-no kado-ga tore-te, maru-ku ochitsuku.

Personality-LK edges-NOM comeoff-TE round-INF calmdown

‘(His/her)personalitybecamemoremellow.’

Skill:

(70) EristeinRohdiamant.

Thoughunrefined,hehasgreatpotential[Heisaroughdiamond].

(71) IchmussmeinEnglischaufpolieren.

Ihavetobrushupon[polish]myEnglish.

(72) Ude-wo migaku

arm-ACC polish

‘toimproveone’sskill’

(73) Arakezuri-no senshu da ga, mikomi-ga aru.

Roughhewn-LK athlete COP CONJ expectation-NOM exist

‘Theathelteisstillrougharoundtheedgesbutshowspromise.’

15.3.2.(IIIb)GOODDEVELOPMENTISSMOOTH,BADDEVELOPMENTISROUGH

These are entailments of the event structure metaphor (Lakoff 2006: 213). More

specifically,therelevantmappingsare:

192

• PURPOSEFULACTIVITIESAREJOURNEYS

• MEANSAREPATHS

• DIFFICULTIESAREIMPEDIMENTSTOMOTION

Inotherwords,ifmeansarepaths,thensmoothsurfacestructureispreferable,sinceit

posesnoimpedimentstomotion:

(74) SiehatzukünftigenGenerationendenWeggeebnet.

Shehaspavedthewayforfuturegenerations.

(75) VorunsliegteinsteinigerWeg.

Arockyroadliesaheadofus.[I.e.,difficultiesaretobeexpected]

(76) Kôshô-ga nameraka-ni shinkô shi-ta.

Negotiation-NOM smoothly progress do

‘Thenegotiationprogressedsmoothly.’

(77) Gengogakushû-wa dekobokomichi.

Languagelearning-TOP bumpyroad

‘Languagelearningisabumpyroad.’

Asindicatedbytheseexamples,aparticularlysalientaspectofthesmooth-roughscaleis

theforcedynamicnotionof friction.AsevidencedbytheCONFLICTISFRICTIONmetaphor,

frictionisgenerallyconsideredassomethingnegative:

(78) EskamzuReibungenzwischenRusslandundderTürkei.

FrictionensuedbetweenRussiaandTurkey

(79) Iken-ga masatsu shi-teiru.

Opionions-NOM friction do-PROG

‘Thereisaconflictofopinion.’

Note,however,thatsomeamountoffrictionisneededsinceMAKINGAMISTAKEISSLIPPING:

(80) EinbedauerlicherAusrutscher

Anunfortunateslip-up

(81) BeiihrerErklärunggerietsieinsschlingern.

Shehaddifficultygivinganexplanation[startedtoswerve].

(82) Shiken-ni suberu

Test-DAT slip

‘Tofailatest’

In summary, then, bothGerman and Japanesemetaphorically express the difficulty of

developmentalpathsintermsofsmooth,rough,orslipperysurfacestructure.

193

15.4.ObservationsandExplanations

In this chapter we have taken a brief look at embodiment from a cross-cultural

perspectivebyexaminingthreesourcedomainsandtheirmetaphoricalscopeinGerman

andJapanese.Theresults,summarizedintable1,canbebrokendownintothefollowing

observations:

• Allmetaphorsexcept(Ia)occurinbothlanguages.

• Cross-linguisticvariancestilloccurs,butismorelikelytobeencounteredatthe

specificleveloflinguisticrealizationratherthenatthegenerallevelofembodied

experience.

• As(Ia)shows,notallprimarymetaphorsareinstantiatedcross-linguistically.

How canwe account for these results? First, there is the almost identical scope of all

threesourcedomainsinGermanandJapanese.Whyisitthattwogeneticallyunrelated

languages have so many metaphors in common? As stated above, this result was

anticipated and is straightforwardly answeredby the choice of sourcedomains. Since

our focus inthischapter isontherelationbetweenembodimentandculture,all three

sourcedomainswerechosenfromasetofpropertiesthatdirectlypertaintothelevelof

embodied experience. Based on contemporary research (e.g. Grady 1997a, 1997b;

LakoffandJohnson1999;Kövecses2005),Yupointspointsoutthat“primarymetaphors

derivedirectlyfromourexperienceandveryoftenfromourcommonbodilyexperience

and therefore are more likely to be universal, whereas complex metaphors are

combinationsofprimarymetaphorsandculturalbeliefsandassumptionsand, forthat

reason, tend to be culture-specific” (Yu 2008: 248). In other words, metaphors are

located on a spectrum somewhere between being directly based and being only very

indirectlybasedonembodiedexperience.And since the sourcedomains considered in

thischapteraredirectilyembodied,andgiventheuniversalnatureofhumanphysiology,

it is hardly surprising that we ended up almost exclusively with cross-linguistically

viablemetaphors,manyofwhichareprimary.

Still, this is not to say that cross-linguistic variance is a non-issue. To illustrate this

point, consider (Ic). Themetaphor INSTENSITY ISWEIGHT is present in bothGerman and

Japanese. Yet, in Japanese it is only applicable to the domain ofABSTRACTBURDENS

while in German it is applicable to a much wider range of phenomena. This is also

reflectedonthelinguisticlevelbytheuseofschwer(heavy)asanaugmentativemarker.

Anotherobviousexampleofvarianceismetaphor(Ie).Again,RESPECT/DIGNITYISWEIGHT

isfoundinbothlanguages.Butassoonasweconsidertheleveloflinguisticexpression,

195

ConcludingRemarksandProspects

The main purpose of this thesis was to show that the V2s under consideration are

inherentlymeaningful.Itwasarguedthattheircontributiontothecompoundcanonly

be fully appreciated when considered as part of a complex lexical network that

subsumesbothgrammaticalV2sand their simplexcounterparts.The five case studies

elucidatedthestructureofthesenetworks.Allfiveverbsunderanalysiswereshownto

haveabasicimageschematicmeaningattherootoftheirhighlypolysemousstructure.

Mechanisms of semantic extension such as metaphor, metonymy, and image schema

transformation “latch onto” these basic spatial schemas to yield bundles of naturally

interrelated meaning variants. Thus, we were able to present evidence for the

motivatednessofindividualsensesinaccordancewithgeneralprinciplesofcognition.In

particular, thesensesoftheV2turnedouttobemetaphoricalextensions intoabstract

domainsbasedonthesameimageschematicstructuresasthesensesofthesimplex.

Giventheresultsofthecasestudies,Ihavearguedthatthemetaphoricalmotivationof

the respectiveV2s has important “syntactic” implications (chapter 14). Their peculiar

“argumentstructure”propertieswerereframedasprofilingphenomenaandexplained

intermsofsalience.Theproposedaccountholdsthatcertainparticipantsoftherelation

profiled by the V2 are too abstract, and therefore not prominent enough, to overtly

appearassubjectorobjectnominals.BasedonLangacker’s(e.g.1991,1995)treatment

of “raising” constructions, it was speculated that the absence of these abstract

participants might best be analyzed as an active zone phenomenon (14.6.3.).

Furthermore,theso-calledlexicalvs.syntacticdistinction–adominantparadigminthe

study of Japanese V-V compounds – was fundamentally called into question and the

contours of an alternative usage-based account were sketched out in terms of

schematicityandfrequencyeffects(14.7.).

Finally, we have observed some striking cross-linguistic similarities throughout the

thesis.TheparallelsbetweenDERUandout,AGARUandup,orKAKARUandGermanan

cometomind.Itseemsthatcertainconceptualmappings,especiallythosepertainingto

primarymetaphors,tendtobelesslanguage-specificthanothers.Inchapter15thiswas

investigated by comparing themetaphorical scope of three source domains –weight,

edgeproperties, and surfaceproperties – in Japanese andGerman. The results suggest

that directly embodied source domains such as these are likely to bemapped onto a

197

similar (though not necessarily identical) set of target domains across different

languagesduetotheuniversalnatureofbasicbodilyexperience.

In conclusion, then, we have presented a psychologically realistic account of the

semantic structure of image schema verbs, shown the inextricable relation between

simplexandV2,andarguedinfavorofameaning-basedapproachtosomelong-standing

syntacticquestions.Yet,astudysuchasthishasitslimitations,andinmanyrespectswe

havemerelyscrapedthetipoftheiceberg.Twoissuesinparticulardeservetobesingled

out. (I) The complete story of image schema verbs should undoubtedly include a

historical account of their polysemy, based on diachronic corpora. The present thesis

hastakenapredominantlytheoreticalperspectiveonpolysemy,andwhileIbelievethat

theargumentspresentedaresoundandcoherent,theyshouldbyallmeansbechecked

against a broad empirical basis – ideallyby tracing theprocessof semantic extension

andgrammaticalizationfromtheearliestwrittensourcestothepresentday.(II)Oneof

this thesis’ most serendipitous discoveries is probably the salience-based nature of

“argumentselection”.Twomajoravenuesforfurtherresearchpresentthemselves:First,

thepursuitofanactivezoneaccountregardingtheprofilingpropertiesofgrammatical

V2s.Secondly,theroleofconceptualautonomy,discoursecontext,andfrequencyeffects

in the case of non-grammatical compounds. And last but not least, the schema-based

alternative to the traditional lexicalvs.syntactic dichotomywill requirebacking in the

form of copious amounts of quantitative data. I hope to have layed out the basic

theoreticalgroundworkhere,sothatfuturestudiesmaytackletheseissues.

198

References

Abbreviations:BLS=ProceedingsoftheAnnualMeetingoftheBerkelyLinguisticsSocietyCLS=PapersfromtheRegionalMeetingoftheChicagoLinguisticSocietyAoki,Hirofumi(2004).Fukugôdôshi‘-kiru’notenkai.Kokugokokubun73:35-49.Bahnsen,Paul(1928).EineUntersuchungüberSymmetrieundAsymmetriebeivisuellenWahrnehmungen.ZeitschriftfurPsychologie108:129-154.Barcelona,Antonio(2000).OnthePlausibilityofClaimingaMetonymicMotivationforConceptualMetaphor.In:AntonioBarcelona(ed.),MetaphorandMetonymyattheCrossroads:ACognitivePerspective,TopicsinEnglishLinguistics30,31-58.Berlin:MoutondeGruyter.Benczes,Réka(2006).CreativeCompoundinginEnglish:TheSemanticsofMetaphoricalandMetonymicalNoun-NounCombination.Amsterdam:JohnBenjamins.Brinkmann,Hennig(1962).DiedeutscheSprache.GestaltundLeistung.Düsseldorf:Schwann.Brugman,Claudia(1981).TheStoryof‘over’:Polysemy,SemanticsandtheStructureoftheLexicon.MAthesis,UniversityofCalifornia,Berkeley.Chen,Yi-Ting(2013).Imifureemunimotozukunihongonogoitekifukugôdôshinokôkeisei.Handoutfromatalkgivenatthe38thannualmeetingoftheKansaiLinguisticsSociety(KLS):1-44.Chomsky,Noam(1957).SyntacticStructures.JanuaLinguarum4.TheHague:Mouton.-------(1965).AspectsoftheTheoryofSyntax.Cambridge,MA:MITPress.Clausner,TimothyandWilliamCroft(1999).DomainsandImageSchemas.CognitiveLinguistics10:1–31.Croft,William(1993).TheRoleofDomainsintheInterpretationofMetaphorsandMetonymies.CognitiveLinguistics4:335-370.Cuyckens,Hubert(1991).TheSemanticsofSpatialPrepositionsinDutch:ACognitive-linguisticExercise.PhDthesis,UniversityofAntwerp.Davidson,Donald(1967).TruthandMeaning.Synthese17:304-323.Deane,Paul(1987).EnglishPossessives,Topicality,andtheSilversteinHierarchy.BLS13:65-76.Delbeque,Nicole(1996).TowardsaCognitiveAccountoftheUseofthePrepositionsporandparainSpanish.In:EugeneCasad(ed.),CognitiveLinguisticsintheRedwoods:The

199

ExpansionofaNewParadigminLinguistics,CognitiveLinguisticsResearch6,249-318.Berlin:MoutondeGruyter.Dewell,Robert(1994).OverAgain:Image-schemaTransformationsinSemanticAnalysis.CognitiveLinguistics5:351–380.-------(1997).ConstrualTransformations:InternalandExternalViewpointsinInterpretingContainment.In:MarjolijnVerspoor,KeeDongLee,andEveSweetser(eds.),LexicalandSyntacticalConstructionsandtheConstructionofMeaning,17-32.Amsterdam/Philadelphia:JohnBenjamins.Dowty,David(1991).ThematicProto-rolesandArgumentSelection.Language67:547-619.Fauconnier,GillesandMarkTurner(2003).TheWayWeThink:ConceptualBlendingandtheMind’sHiddenComplexities.NewYork:BasicBooks.Felfe,Marc(2012).DasSystemderPartikelverbenmit“an”.EinekonstruktionsgrammatischeUntersuchung.Berlin/Boston:DeGruyter.Fleischer,WolfgangandIrmhildBarz(2012):WortbildungderdeutschenGegenwartssprache.Berlin/Boston:DeGruyter.(Originalworkpublished1969)Frege,Gottlob(1892).ÜberSinnundBedeutung.ZeitschriftfürPhilosophieundphilosophischeKritik,NeueFolge100:25–50.Fukushima,Kazuhiko(2005).LexicalV-VCompoundsinJapanese:Lexiconvs.Syntax.Language81:568-612.Gibbs,Raymond(2005).ThePsychologicalStatusofImageSchemas.In:BeateHampe(ed.),FromPerceptiontoMeaning:ImageSchemasinCognitiveLinguistics,CognitiveLinguisticsResearch29,113-135.Berlin:MoutondeGruyter.Goossens,Louis(1990).Metaphtonymy:TheInteractionofMetaphorandMetonymyinExpressionsforLinguisticAction.CognitiveLinguistics1:323–40.Grady,Joseph(1997a).FoundationsofMeaning:PrimaryMetaphorsandPrimaryScenes.PhDthesis,UniversityofCalifornia,Berkeley.-------(1997b).THEORIESAREBUILDINGSRevisited.CognitiveLinguistics8:267–90.-------(1999).ATypologyofMotivationforConceptualMetaphor:Correlationvs.Resemblance.In:RaymondGibbsandGerardSteen(eds.),MetaphorinCognitiveLinguistics,79-100.Amsterdam:JohnBenjamins.-------(2005).ImageSchemasandPerception:RefiningaDefinition.In:BeateHampe(ed.),FromPerceptiontoMeaning:ImageSchemasinCognitiveLinguistics,CognitiveLinguisticsResearch29,35-56.Berlin:MoutondeGruyter.

200

Grice,Paul(1975).LogicandConversation.In:PeterColeandJerryMorgan(eds.),SyntaxandSemantics,Vol.3:SpeechActs,41-58.NewYork:AcademicPress.Güler,Mungan(1986).DiesemantischeInteraktionzwischendempräfigierendenVerbzusatzunddemSimplexbeideutschenPartikel-undPräfixverben.Frankfurt/Main:PeterLang.Hampe,Beate(2002).SuperlativeVerbs:ACorpus-basedStudyofSemanticRedundancyinEnglishVerb-particleConstructions.Tübingen:Narr.Hawkins,Bruce(1981).VariableTemporalIntegrationbetweenMotionVerbsandLocalPrepositions.LinguisticNotesfromLaJolla10:98-127.-------(1984).TheSemanticsofEnglishSpatialPrepositions.PhDthesis,UniversityofCalifornia,SanDiego.-------(1988).TheNaturalCategoryMEDIUM:AnAlternativetoSelectionRestrictionsandSimilarConstructs.In:BrygidaRudzka-Ostyn(ed.),TopicsinCognitiveLinguistics,231-270.Amsterdam:JohnBenjamins.Heine,Bernd;UlrikeClaudiandFriederikeHünnemeyer(1991).Grammaticalization:AConceptualFramework.Chicago:UniversityofChicagoPress.Herskovits,Annette(1986).LanguageandSpatialCognition.Cambridge:CambridgeUniversityPress.Himeno,Masako(1976).Fukugôdôshi-agaru,-ageruoyobikakôwoarawasufukugôdôshirui.Nihongogakkôronshû3:91-122.-------(1977).Fukugôdôshi–deruto–dasu.Nihongogakkôronshû4:71-95.-------(1979).Fukugôdôshi–kakaruto–kakeru.Nihongogakkôronshû6:37-61.-------(1980).Fukugôdôshi-kiruto-nuku,-tôsu.Nihongogakkôronshû7:23-46.-------(1999).Fukugôdôshinokôzôtoimiyôhô.Tôkyô:Hitsujishobô.Ishikawa,Shin’ichirô(2010).Gendainihongokakikotobakinkôkôpasu(BCCWJ)niokerufukugôdôshi–dasunoryôtekibunseki.Tôkeisûrikenkyûjokenkyûrepôto238:15-34.Johnson,Mark(1990).TheBodyintheMind:TheBodilyBasisofMeaning,Imagination,andReason.Chicago:UniversityofChicagoPress.(Originalworkpublished1987)-------(2005).ThePhilosophicalSignificanceofImageSchemas.In:BeateHampe(ed.),FromPerceptiontoMeaning:ImageSchemasinCognitiveLinguistics,CognitiveLinguisticsResearch29,15-33.Berlin:MoutondeGruyter.Kabata,KaoriandSallyRice(1997).Japaneseni:TheParticularsofaSomewhatContradictoryParticle.In:MarjolijnVerspoor,KeeDongLee,andEveSweetser(eds.),

201

LexicalandSyntacticalConstructionsandtheConstructionofMeaning,107-127.Amsterdam/Philadelphia:JohnBenjamins.Kageyama,Tarô(1993).Bunpôtogokeisei.Tôkyô:Hitsujishobô.-------(1996).Dôshiimiron–gengotoninchinosetten.Tôkyô:Kuroshio.-------(2009).Isolate:Japanese.In:RochelleLieberandPavelStekauer(eds),TheOxfordHandbookofCompounding,513-526.Oxford:OxfordUniversityPress.Katz,Jerrold(1972).SemanticTheory.NewYork:Harper&Row.Kaufmann,Ingrid(1993).SemanticandConceptualAspectsofthePrepositiondurch.In:CorneliaZelinsky-Wibbelt(ed.),TheSemanticsofPrepositions:FromMentalProcessingtoNaturalLanguageProcessing,221-247.Berlin:MoutondeGruyter.Kikuta,Chiharu(2008).FukugôdôshiV-kakaruV-kakerunobunpôka.Kôbunnoseiritsutokakuchô.Dôshishadaigakueigoeibungakukenkyû81-82:115-162.Koffka,Kurt(1935).PrinciplesofGestaltPsychology.NewYork:HarcourtBrace.Kövesces,Zoltan(1988).TheLanguageofLove:TheSemanticsofPassioninConversationalEnglish.Lewisburg,PA:BucknellUniversityPress.-------(2005).MetaphorinCulture:UniversalityandVariation.Cambridge:CambridgeUniversityPress.Kövecses,ZoltanandGünterRadden(1998).Metonymy:DevelopingaCognitiveLin-guisticView.CognitiveLinguistics9:37–77.Kreitzer,Anatol(1997).MultipleLevelsofSchematization:AStudyintheConceptualizationofSpace.CognitiveLinguistics8:291-325.Kuno,SusumuandEtsukoKaburaki(1977).EmpathyandSyntax.LinguisticInquiry8:627-672.Lakoff,George(1990a).TheInvarianceHypothesis:IsAbstractReasonBasedonImage-Schemas?CognitiveLinguistics1:39–74.-------(1990b).Women,Fire,andDangerousThings:WhatCategoriesRevealabouttheMind.Chicago:UniversityofChicagoPress.(Originalworkpublished1987)-------(2006).TheContemporaryTheoryofMetaphor.In:DirkGeeraerts(ed.),CognitiveLinguistics:BasicReadings.CognitiveLinguisticsResearch34,185-238.Berlin:MoutondeGruyter.(Reprintedfrom:AndrewOrtony(ed.)(1993),MetaphorandThought,202-251.Cambridge:CambridgeUniversityPress.)Lakoff,GeorgeandMarkJohnson(1999).PhilosophyintheFlesh:TheEmbodiedMindandItsChallengetoWesternThought.NewYork:BasicBooks.

202

-------(2003).MetaphorsWeLiveBy.Chicago:UniversityofChicagoPress.(Originalworkpublished1980)Lakoff,GeorgeandZoltanKövecses(1987).TheCognitiveModelofAngerInherentinAmericanEnglish.In:DorothyHollandandNaomiQuinn(eds.),CulturalModelsinLanguageandThought,195-221.Cambridge:CambridgeUniversityPress.Lakoff,GeorgeandRafaelNúñez(2000).WhereMathematicsComesFrom:HowtheEmbodiedMindBringsMathematicsintoBeing.NewYork:BasicBooks.Lakoff,GeorgeandMarkTurner(1989).MorethanCoolReason:AFieldGuidetoPoeticMetaphor.Chicago:UniversityofChicagoPress.Langacker,Ronald(1987).FoundationsofCognitiveGrammar,vol.1:TheoreticalPrerequisites.Stanford:StanfordUniversityPress.-------(1990).Concept,Image,andSymbol:TheCognitiveBasisofGrammar.CognitiveLinguisticsResearch1.Berlin:MoutondeGruyter.-------(1991).FoundationsofCognitiveGrammar,vol.2:DiscriptiveApplication.Stanford:StanfordUniversityPress.-------(1993).Reference-pointConstructions.CognitiveLinguistics4:1–38.-------(1995).RaisingandTransparency.Language71:1-62.-------(2008).CognitiveGrammar:ABasicIntroduction.Oxford/NewYork:OxfordUniversityPress.Lee,KyungSoo(1997).Chûkantekifukugôdôshi‘kiru’noimiyôhônokijutsu:hondôshi‘kiru’tozenkôdôshi‘kiru’tokanrenzukete.Sekainonihongokyôiku7:219-232.Lehrer,Adrienne(1978).StructureoftheLexiconandTransferofMeaning.Lingua45:95-123.Leys,Odo(1989).AspektundRektionräumlicherPräpositionen.DeutscheSprache17:97-113.Li,Jie(1994).RäumlicheRelationenundObjektwissenamBeispielanundbei.Tübingen:Narr.Lindner,Susan(1981).ALexico-semanticAnalysisofEnglishVerbParticleConstructionswith‘out’and‘up’.PhDthesis,UniversityofCalifornia,SanDiego.-------(1982).WhatGoesupDoesn'tNecessarilyComedown:TheinsandoutsofOpposites.CLS18:305-323.Lindstromberg,Seth(1998).EnglishPrepositionsExplained.Amsterdam:JohnBenjamins.

203

Mandler,JeanandCristobalPagánCánovas(2014).OnDefiningImageSchemas.LanguageandCognition0:1–23.Matsuda,Fumiko(2001a).Fukugôdôshikôkô-komunoimi.Ningenbunkaronsô4:223-235.-------(2001b).Koazushikiwomochiitafukugôdôshikôkô-komunoninchiimirontekisetsumei.Nihongokyôiku111:16-25.Matsuki,Keiko(1995).MetaphorsofAngerinJapanese.In:JohnTaylorandRobertMacLaury(eds.),LanguageandtheCognitiveConstrualoftheWorld,137-151.Berlin:MoutondeGruyter.Matsumoto,Yô(1998a).Nihongonogoitekifukugôdôshiniokerudôshinokumiawase.Gengokenkyû114:37-83.-------(1998b).SemanticChangeintheGrammaticalizationofVerbsintoPostpositionsinJapanese.In:ToshioÔhori(ed.),StudiesinJapaneseGrammaticalization:CognitiveandDiscoursePerspectives,25-60.Tôkyô:Kurosio.Momiyama,Yôsuke(2014).Nihongokenkyûnotamenoninchigengogaku.Tôkyô:Kenkyûsha.Montague,Richard(1973).TheProperTreatmentofQuantificationinOrdinaryEnglish.In:JaakkoHintikka,JuliusMoravcsikandPatrickSuppes(eds.),ApproachestoNaturalLanguage,221-242.Dordrecht:Reidel.Morgan,Pamela(1997).FiguringOutfigureout.MetaphorandtheSemanticsoftheEnglishVerb-particleConstruction.CognitiveLinguistics8:327-357.Morita,Yoshiyuki(1989).Kisonihongojiten.Tôkyô:Kadokawa.(Originalworkpublished1977)Munro,Pamela(1991).ANGERISHEAT:SomeDataforaCross-linguisticSurvey.Manuscript,DepartmentofLinguistics,UCLA.Nagashima,Yoshio(1976).Fukugôdôshinokôzô.Tôkyô:Taishûkanshoten.Nakashima,Noriko(2006).Fukugôdôshinikansuruichikôsatsu:‘-kiru’‘-tôsu’‘-nuku’nohikakukara.Kokubungakutôsa18:262-271.Nüse,Ralf(1999).GeneralMeaningsforGermanan,auf,inandunter:Towardsa(Neo)classicalSemanticsofTopologicalPropositions.PhDthesis,Humboldt-UniversitätzuBerlin.Omata,Yoshihiro(2007).Fukugôdôshi‘-kiru’‘-nuku’‘-tôsu’nitsuite.Gaikokugogakkaishi37:211-222.Radden,Günter(1989).FigurativeUseofPrepositions.In:RenéDirven(ed.),AUser’sGrammarofEnglish,551-576.FrankfurtamMain:Lang.

204

-------(2000).HowMetonymicareMetaphors?.In:AntonioBarcelona(ed.),MetaphorandMetonymyattheCrossroads:ACognitivePerspective,TopicsinEnglishLinguistics30,93-108.Berlin:MoutondeGruyter.-------(2006).TheMetaphorTIMEASSPACEAcrossLanguages.In:ElzbietaGórskaandGünterRadden(eds.),Metonymy-MetaphorCollage,99-120.Warsaw:WarsawUniversityPress.Rice,Sally(1992).PolysemyandLexicalRepresentation:TheCaseofThreeEnglishPrepositions.In:ProceedingsoftheFourthAnnualConferenceoftheCognitiveScienceSociety,89-94.Hillsdale,NJ:LawrenceEarlbaum.Rohrer,Tim(2005).ImageSchemataintheBrain.BeateHampe(ed.),FromPerceptiontoMeaning:ImageSchemasinCognitiveLinguistics,CognitiveLinguisticsResearch29,165-196.Berlin:MoutondeGruyter.Rosch,Eleanor(1977).HumanCategorization.In:NeilWarren(ed.),StudiesinCross-lin-guisticPsychology,1-49.London:AcademicPress.-------(1978).PrinciplesofCategorization.In:EleanorRoschandBarbaraLloyd(eds.),CognitionandCategorization,27-48.Hillsdale,NJ:LawrenceErlbaum.Rubin,Edgar(1958).FigureandGround.In:DavidBeardsleeandMichaelWertheimer(eds.),ReadingsinPerception,194-203.Princeton,NJ:VanNostrand.(Originalworkpublished1915)Sadamitsu,Miyagi(2001).ACognitiveAcccountofSynaestheticMetaphor.OsakaPapersinEnglishLinguistics6:115-130.Saile,Günter(1984).SpracheundHandlung.EinesprachwissenschaftlicheUntersuchungvonHandhabe-Verben,Orts-undRichtungsadverbialenamBeispielvon

Gebrauchsanweisungen.Braunschweig:Vieweg.Sandra,DominiekandSallyRice(1995).NetworkAnalysesofPrepositionalMeaning:MirroringWhoseMind–TheLinguist’sortheLanguageUser’s?CognitiveLinguistics6:89–130.Schank,RogerandRobertAbelson(1977).Scripts,Plans,GoalsandUnderstanding:anInquiryintoHumanKnowledgeStructures.Hillsdale,NJ:LawrenceErlbaumAssociates.Shen,Yeshayahu(1997).CognitiveConstraintsonPoeticFigures.CognitiveLinguistics8:33-71.Silverstein,Michael(1976).HierarchyofFeaturesandErgativity.In:RobertDixon(ed.),GrammaticalCategoriesinAustralianLanguages,112-171.Canberra:AustralianInstituteofAboriginalStudies.Smith,Michael(1987).TheSemanticsofDativeandAccusativeinGerman:AnInvestigationinCognitiveGrammar.PhDthesis,UniversityofCalifornia,SanDiego.

205

-------(1992).TheRoleofImageSchemasinGermanGrammar.LeuvenseBijdragen82:385-410.Storch,Günther(1978).SemantischeUntersuchungenzudeninchoativenVerbenimDeutschen.Braunschweig:Vieweg.Sugimura,Yasushi(2008).Fukugôdôshi‘-kiru’noiminitsuite.Gengobunkakenkyûsôsho7:63-79.-------(2012).KôpasworiyôshitafukugôdôshiV1-tôsunoimibunseki.Gengobunkaronshû34:47-59.Suk,EunKyung(2004).Gendainihongoniokerufukugôdôshinikansurukôsatsu:Jidaitekisôwoarawasukôkôdôshinoimibunrui.Rikkyôdaigakunihonbungaku93:155-168.Sweetser,Eve(1991).FromEtymologytoPragmatics:MetaphoricalandCulturalAspectsofSemanticStructure.CambridgeStudiesinLinguistics54.Cambridge:CambridgeUniversityPress.(Originalworkpublished1990)Talmy,Leonard(1975).FigureandGroundinComplexSentences.BLS1:419-430.-------(1983).HowLanguageStructuresSpace.In:HerbertPickandLindaAcredolo(eds.),SpatialOrientation:Theory,ResearchandApplication,225-282.NewYork:Plenum.-------(1991).PathtoRealization:ATypologyofEventConflation.BLS17:480-519.-------(2003a).TowardaCognitiveSemantics,vol.1:ConceptStructuringSystems.Cambridge:MITPress.(Originalworkpublished2000)-------(2003b):TowardsaCognitiveSemantics,vol.2:TypologyandProcessinConceptStructuring.Cambridge:MITPress.(Originalworkpublished2000)-------(2006).TheFundamentalSystemofSpatialSchemasinLanguage.In:BeateHampe(ed),FromPerceptiontoMeaning:ImageSchemasinCognitiveLinguistics,CognitiveLinguisticsResearch29,199-234.Berlin:MoutondeGruyter.Taniguchi,Kazumi(2003).Ninchiimironnoshintenkai:Metafâtometonimii.Tôkyô:Kenkyûsha.-------(2005)Jitaigainennokigôkanikansuruninchigengogakutekikenkyû.Tôkyô:Hitsujishobô.Tarski,Alfred(2004).TheSemanticConceptionofTruthandtheFoundationsofSemantics.In:FrederickSchmitt(ed.),TheoriesofTruth,115-151.Oxford:Blackwell.(Originalworkpublished1944)Taylor,John(1996).PossessivesinEnglish:AnExplorationinCognitiveGrammar.Oxford:ClarendonPress.

206

Teramura,Hideo(1969).Katsuyôgobi,jodôshi,hojodôshitoasupekuto:Sonoichi.Nihongonihonbunka1:32-48.Tuggy,David(1993).Ambiguity,Polysemy,andVagueness.CognitiveLinguistics4:273–90.-------(2005).CognitiveApproachtoWord-formation.In:PavelStekauerandRochelleLieber(eds.),HandbookofWord-Formation,233-265.Dordrecht:Springer.Turner,Mark(1991).ReadingMinds:TheStudyofEnglishintheAgeofCognitiveScience.Princeton,NJ:PrincetonUniversityPress.Tyler,AndreaandVyvyanEvans(2001).ReconsideringPrepositionalPolysemyNet-works:TheCaseof‘over’.Language77:724–65.-------(2003).TheSemanticsofEnglishPrepositions:SpatialScenes,EmbodiedMeaningandCognition.Cambridge:CambridgeUniversityPress.Vandeloise,Claude(1990).Representation,PrototypesandCentrality.In:SavasTsohatzidis(ed.),MeaningsandPrototypes,403–437.London:Routledge.-------(1991).SpatialPrepositions:ACaseStudyfromFrench.Chicago:UniversityofChicagoPress.Wittfeld,Aron(2013).“Missing”Arguments:TowardsaCognitiveAccountofArgumentStructurePhenomenainJapaneseV-VCompound(VVC)Formation.ConferenceHandbookofthe14thannualmeetingoftheJapaneseCognitiveLinguisticsAssociation

(JCLA):193-196.Wittgenstein,Ludwig(1922).TractatusLogico-Philosophicus[C.K.Ogden,trans.].London:RoutledgeandKeganPaul.-------(1953).PhilosophicalInvestigations[G.E.Anscombe,trans.].NewYork:Macmillan.Yamaguchi,Maki(2009).Kaishino–dasu–hajimerunitsuite.Shôsetsusakuhinniokeruyôreibunsekiwotôshitezensetsusurudôshinokeikôwosaguru.Mitakokubun49:1-16.Yamamoto,Kiyotaka(1984).Fukugôdôshinokakushihai.Todaironkyû21:32-49.Yamanashi,Masaaki(2000).Ninchigengogakugenri.Tôkyô:Kuroshio.-------(2009).Ninchikôbunron:Bunpônogeshutaruto-sei.Tôkyô:Taishûkan.Yang,Gloria(2002).LoveandItsConceptualMetaphorsinMandarin:AspectualClassification.Manuscript,DepartmentofLinguistics,UniversityofCaliforniaatBerkeley.Yu,Ning(1995).MetaphoricalExpressionsofAngerandHappinessinEnglishandChinese.MetaphorandSymbolicActivity10:59-92.

207

-------(2008).MetaphorfromBodyandCulture.In:RaymondGibbs(ed.),TheCambridgeHandbookofMetaphorandThought,247-261.Cambridge:CambridgeUniversityPress.Yumoto,Yôko(1996).Gokeiseitogoigainenkôzô.In:Gengotobunkanoshosô:OkudaHiroyukikyôjutaikankinenronbunshû,105-118.Tôkyô:Eihôsha.-------(2008).Fukugôdôshiniokerukônogugen:Tôgotekifukugôdôshitogoitekifukugôdôshinosai.LexiconForum4:1-30.

208