Sie sind auf Seite 1von 27

9/20/2016 ConversationalImpliciture

ConversationalImpliciture
KENTBACH


Confusionintermsinspiresconfusioninconcepts.Whenarelevantdistinctionisnotclearlymarkedornotmarkedatall,itisapt
tobeblurredorevenmissedaltogetherinourthinking.Thisistrueinanyareaofinquiry,pragmaticsinparticular.Noone

disputesthattherearevariouswaysinwhichwhatiscommunicatedinanutterancecangobeyondsentencemeaning,[1]The
problemistocatalogtheways.Itisgenerallyrecognizedthatlinguisticmeaningunderdeterminesspeakermeaningbecauseof
theneedfordisambiguationandreferenceassignmentandbecausepeoplecanspeakfigurativelyorindirectly.Butphilosophers
andlinguistsarecomingtorecognizethatthesearenottheonlyways.ThesituationmaybedescribedinGriceanterms:the
distinctionbetweenwhatissaidandwhatisimplicatedisnotexhaustive.Chartingthemiddlegroundbetweenthetwowill
requireattendingtospecificexamples,notingtheirdistinctivefeatures,andarticulatingtherelevantconcepts.ThatiswhatIaim
todohere.Thebasicideawillbetodistinguishnotonlytheimpliedfromtheexplicitbuttheimplicitfromtheimplied.
.
1.Introduction:BeingInexplicit
Ifusinglanguageamountedtonothingmorethanputtingonesthoughtsintowords,thenunderstandinganutterancewouldbe
merelyamatterofdecodingthewordsuttered.Nooneseriouslybelievesthat,andyeteventodayphilosophersoftenpretendthat
everysentenceexpressesapropositionandthateveryutteranceistheassertionofthatproposition.Therearevariousphenomena
thatsuchasimplisticviewfailstotakeintoaccount.Fortunately,thankstotheworkandinfluenceofAustinandGrice,mostof
thesephenomenaarenowquitefamiliarandfairlywellunderstood.However,therearetwosuchphenomenawhich,though
pervasive,havereceivedlittlenotice.
Considerthefollowingexample,inwhichawomansays(1)toherhusband.
(1)Shehastakenenoughfromyou.
Themeaningsofthewordssheandyoudonotbythemselvesdeterminewhotheyarebeingusedtoreferto,sayacertainfemale
employeeandthespeakershusband,whoisalsotheemployeesboss.Moreover,becausethewordtakenisambiguous
(ambiguityissemanticoverdetermination),oneofthisword'slinguisticmeaningsmustbeselected.Dependingonwhichoneis

operative,takencouldbeusedmeanappropriated,received,tolerated,orsuffered.[2]Supposeitisbeingusedtomean
suffered.Then(1)wouldbeusedtosaythattheemployeeinquestionhassufferedenoughfromthespeaker'shusband.Perhaps
thewifeisnotonlyassertingthatpropositionbutalsourgingherhusbandtostopdoingwhathehasbeendoingtotheemployee
ortowarnhimaboutherpossiblyretaliating.Ormaybehiswifedoesnotreallymeanwhatsheissayingbutisbeingodiously
ironic,urgingherhusbandtomistreattheemployeestillmore.Butleaveasidepossibleindirectorfigurativeusesof(1).Just
noticethatnothinginthesentenceindicateswhatitisthattheemployeeisbeingsaidtohavesufferedenoughoforwithrespect
towhatitisenoughnordoesanythinginthesentenceindicatethattheutteranceisintendedtoapplyonlytothetimesincethe
husbanddevelopedapersonalinterestintheemployee.
Thisexampleillustratesthetwowaysinwhichaspeakercan,independentlyofusinganyambiguousorindexical
expressionsandwithoutspeakingfigurativelyorindirectly,meansomethingwithoutmakingitfullyexplicit.Thefirstway
ariseswheneveranutterance,evenafterdisambiguationandreferencefixing,doesnotbyvirtueoflinguisticmeaningexpressa
completeproposition.Whenasentenceisinthiswaysemanticallyunderdeterminate,understandinganutteranceofitrequiresa
processofcompletiontoproduceafullproposition.Thesecondwayoccurswhentheutterancedoesexpressacomplete

http://userwww.sfsu.edu/kbach/impliciture.htm 1/27
9/20/2016 ConversationalImpliciture

proposition(possiblyastheresultofcompletion)butsomeotherproposition,yieldedbywhatIcalltheprocessofexpansion,is
beingcommunicatedbythespeaker.Inbothcasesthespeakerisnotbeingfullyexplicit.Rather,heintendsthehearertoread
somethingintotheutterance,toregardasifitcontainedcertainconceptualmaterialthatisnotinfactthere.Theresultof
completionand/orexpansioniswhatIcallconversationalimpliciture.
ImplicitureistobedistinguishedfromGrices(1967a)conversationalimplicature.Inimplicatureonesaysand
communicatesonethingandtherebycommunicatessomethingelseinaddition.Impliciture,however,isamatterofsaying
somethingbutcommunicatingsomethingelseinstead,somethingcloselyrelatedtowhatissaid.Asourexampleswillmake
clear,unlikemetaphoricalandothersortsofnonliteralutteranceimplicitureisnotacaseofusingparticularwordsinsome
figurativeway.Rather,partofwhatiscommunicatedisonlyimplicitinwhatisexplicitlyexpressed,eitherbecausetheutterance
issemanticallyunderdeterminateandcompletionisrequiredorbecausewhatisbeingcommunicatedisanexpandedversionof
thepropositionexpressed.Examplesofthesetwowaysofbeingimplicitarepresented,respectively,insections2and3.These
examplesshowthatGrice'sdistinctionbetweenwhatissaidandwhatisimplicatedisnotexhaustive.Section4explainsthe
differencebetweenimplicitureandimplicature.Section5takesupthenotionofwhatissaidthatentersintoGricesaccountof
implicatureandintomyaccountofimpliciture.Hisnotionhasbeenregardedbysomeastoorestrictive,bothintuitivelyand
theoretically,butwithsomeneededmodificationsithas,Iwillsuggest,bothintuitiveappealandtheoreticalvalue.Grice's
contrastbetweenwhatissaidandwhatisconventionallyimplicatedistakenupinsection6,whereitisarguedthatthelatter
notioncanbedispensedwith.Section7takesupthecaseofcompletionandexpansionatthelexicalandphrasalasopposedto
thesententiallevel.Finally,insection8,generalissuesareaddressedconcerningtherelationbetweenouraccountof
conversationalimplicitureandthepsychologyofinferencesinvolvedinrecognizingthem.

2.UnderdeterminationandCompletion
Ingrammarschoolonewastaughtthatasentenceexpressesacompletethought.Thephenomenonofsemantic
underdeterminationshowsotherwise.Itgainedmodestrecognitioninthelateseventiesandeightiesundersuchlabelsas
semanticgenerality(Atlas,1977)andnonspecificity(Bach,1982)andismorewidelyknownthesedaysassemantic

underdetermination.[3])Itisakintotheoldernotionofsensegeneralityofwords(suchasdeep,take,andbefore),whichlexical
semanticistsdistinguishfromhomonymy,ambiguity,andvagueness(seeAtlas,1989,ch.2),butsemanticunderdeterminationis
afeatureofsentences.Forexample,sentences(1)and(2),
(1)Steelisntstrongenough.
(2)Williealmostrobbedabank.
thoughsyntacticallywellformed,aresemanticallyorconceptuallyincomplete,inthesensethatsomethingmustbeaddedforthe
sentencetoexpressacompleteanddeterminateproposition(somethingcapableofbeingtrueorfalse).With(1)weneedtoknow
strongenoughforwhat.Noticethat(1)doesnotexpresstheweakpropositionthatsteelisntstrongenoughforsomethingor
other.Theproblemwith(2),duetothewordalmost,isthis:whatiscommunicatedcouldbethatWillietriedandnearly
succeededatrobbingabank,thathebarelyrefrainedfromrobbingabank,orthat,intentonrobbingsomething,hereluctantly
decidedagainstrobbingabankandrobbedsomethingelseinstead(ordidsomethingelsetothebank).Inthesecasesthe
conventionalmeaningofthesentencedeterminesnotafullpropositionbutmerelyapropositionalradicalacomplete
propositionwouldbeexpressed,atruthconditiondetermined,onlyifthesentencewereelaboratedsomehow.
Becausetheutteranceofasemanticallyunderdeterminatesentencerequirescompletion,thespeakercannotmeanjustwhat
isdetermined(evenwithanyneededdisambiguationandreferenceassignment)bywhathiswordsmean.Evenso,anutteranceof

http://userwww.sfsu.edu/kbach/impliciture.htm 2/27
9/20/2016 ConversationalImpliciture

suchasentencecanstillbeliteralnoneoftheconstituentsof(1)or(2),forexample,isbeingusednonliterallyitsjustthat
thisisnotthewholeofwhatthespeakermeans.Whathemeansmustbeacompleteproposition.
(1)and(2)illustratetwodifferentsourcesofpropositionalincompleteness:constituentandstructuralunderdetermination.

In(1)anadditionalpropositionalconstituent[4]isneededtocompletetheproposition,whereasin(2)somethinglikescopemust
beassigned.Anutteranceof(1)mustbetakentomeanthatsteelisntstrongenoughinsomecontextuallyidentifiablerespect,
e.g.forbuildinga500storybuildingortoresistbendingbySuperman.Thespeakercouldhavemadetheadditionalconceptual
materialexplicitbyincludingthecorrespondinglexicalmaterialinhisutterance.With(2)somecontextuallyidentifiable
contrastisintendedbytheuseofalmost.Thepossibilitiescorrespondtosomethinglikescope(ofalmost),butIhesitatetosay
thatitisscopebecauseitdoesnotseemtobeamatterofstructuralambiguityatanysyntacticlevel.Ifitisnot,thentherelevant
differencebetween(1)and(2)isthatwhereasanunderstandingofanutteranceof(1)requirestheinsertionofadditional
conceptualmaterial,(2)requiresthearticulationofstructuralrelationsamongexistingmaterial.
Examplesofconstituentunderdeterminationcanbemultipliedindefinitely.Hereisasample,withpossiblecompletions
giveninbrackets:
(3)Thatlampischeap.[relativetootherlamps]
(4)Gentlemenpreferblondes.[tobrunettes]
(5)Mutualknowledgeisrelevant.[tocommunication]
(6)Stromistooold.[tobeagoodsenator]
(7)Evencowgirlssingtheblues.[inadditiontocowboys]
(8)Gregorwasmerelyabookkeeper.[asopposedtoanaccountant]

Ineachcaseaconstituentisneededtospecifywhatcompletestheproposition[5]thatisonlyincompletelyexpressed,some
relevantclass,respect,orcontrastwithrespecttowhichtheutteranceisintendedtobeunderstood.Othercasesinvolvean
implicitsituation,location,oraction,asin
(9)Theprincessislate.[fortheparty]
(10)Tipperisready.[todance]
(11)Thekinghasarrived.[atthepalace]
(12)Alhasfinished.[speaking]
Noticethatincontrastto(11)and(12),(13)and(14)areungrammatical.
(13)*Thekinghasreached.
(14)*Alhascompleted.
Thedifferencehereseemsentirelylexical.Thereisnosemanticorconceptualexplanationforwhy(11)and(12)areallrightand
(13)and(14)arenot.Andthereissurelynopragmaticexplanation.Forexample,tobepragmaticallyacceptableanutteranceof
(12)hastobemadeincircumstanceswhereitisinferablewhatthespeakermeansAlhasfinished,butthesearetheverysame
cricumstancesinwhichitwouldbeinferablewhatAlhascompleted.
Severalcasesofconstituentunderdeterminationareofspecialphilosophicalinterest.Forexample,counterfactual
conditionalsarenotcategoricallytrueorfalsebutonlyrelativetoasetofimplicitbackgroundassumptions.Utterancesofboth
ofthefollowingcouldbetrueifdifferentbackgroundassumptionswereheldfixed.
(15)IfLincolnhadntgonetothetheater,hewouldnthavebeenassassinated.
(16)IfLincolnhadntgonetothetheater,hewouldhavebeenassassinatedanyway.
Thissuggeststhattheseconditionalsdonotexpresscompletepropositionsastheystand.Similarly,contrastiveexplanations,
suchasthosegivenbythefollowingbecausesentences,differastotherelevantbutimplicitexplanatorycontrast.

http://userwww.sfsu.edu/kbach/impliciture.htm 3/27
9/20/2016 ConversationalImpliciture

(17)BilldemotedMickey[ratherthanRon]becausehemadetheblunder
(18)BilldemotedMickey[ratherthanfirehim]becausehewasstillneeded.
Finally,ofoccasionalphilosophicalinterestarethewordsalsoandeven.Insentenceslike(19)and(20),
(19)Psychologistscanalsoattendthemeeting.
(20)Iwillalsostudylinguistics.
thereisanimplicitallusiontosomereferenceclass(intheseexamples,toothergroupsorotheracademicsubjects).Witheven,
(21)Evenaccountantsmakemistakes.
(22)Illeventakewater.
themembersofthereferenceclassareorderedonacertainscale,ononeextremeofwhichlietheitemsmentioned.
Structuralunderdeterminationisofteninducedbyadverbs,likealmost,asin(2),andtoo.Myfavoriteexampleis(23),
(23)Iloveyoutoo.
whichcanbeusedinatleastfourdifferentways,therelevantcontrastdependingonwhethertooappliestoI,love,you,orIlove
you.Andtherearecertainconstructionsofphilosophicalandlinguisticinterestwhichseemtoinducestructural
underdetermination.Thesearesentencesinvolvingtheinteractionofsuchelementsasquantifiednounphrases,modaland
temporaloperators,andbeliefcontexts..Becauseofthecomplexissuestheyraise,suchasscopeandlogicalformandthe
distinctionbetweendedictoandderereadings,Iwillnotargueforsemanticunderdeterminationhere(Idefenditinthecaseof
beliefsentencesinBach,1987a,pp.210214)butwillmerelysuggestafewexamples.Inthecaseofsentenceslikethefollowing
(alternativewaysoftakingthemaregiveninthebrackets),
(24)Afewarsonistsdestroyedmanybuildings.[each/together]
(25)Thenumberofplanets[nine/whicheveritis]maybeeven.
(26)In1996the[now/then]presidentoftheU.S.willbeaRepublican.
(27)GyrobelievesthattheinventoroftheYoYo[whoeverthatmaybe/DonaldDuncan]
isrich.
theusualviewisthattheyexemplifyscopeambiguity.Yetunlessthereisshowntobeastructuralphenomenonatsomesyntactic
level,thisviewrestsonanidleappealtoananalogywithlogicalformulae.Ontheotherhandif,say,thelevelofLFinGB
theoryisgenuineandappliestoscopephenomenagenerally,thentheabovesentencesarenotsemanticallyunderdeterminate
afterallbutambiguous.Neale(1993)hasdefendedLFforthecaseofquantifiednounphrases,asin(24),butnotforthevarious
operatorsthatoccurin(25)(27).
Nowthereareseveralnaturalobjectionstothenotionofsemanticunderdetermination.OneobjectionisthatthesentencesI
regardassemanticallyunderdeterminatearereallysemanticallycomplete.Arelatedobjectionisthatutterancesofallegedly
semanticallyunderdeterminatesentencesareelliptical,inthatthesyntacticstructureofthesentencecontainsaslotforeach
elementneededtocompletetheproposition.Athirdobjectionisthatalthoughthesesentencesareindeedincomplete,theyareso
onlyinthewaythatsentencescontainingindexicalsareincomplete.
Theideabehindthefirstobjectionisthattheutteranceofasocalledsemanticallyunderdeterminatesentence,evenifthe
utteranceistakenstrictly,reallydoesexpressacompleteproposition,albeitaminimalpropositionmuchtooweak(ormuchtoo
strong)tobewhatthespeakermeans.Supposedly,theminimalpropositionsexpressedby(11)and(12),forexample,are(11MP)
and(12MP).
(11)Tipperisready.
(11MP)Tipperisreadyforsomethingorother.
(12)Alhasfinished.

http://userwww.sfsu.edu/kbach/impliciture.htm 4/27
9/20/2016 ConversationalImpliciture

(12MP)Alhasfinisheddoingsomethingorother.
Therearetwodifficultieswiththissuggestion.First,ontheassumptionthatanysentencecanbeusedliterally,itwouldfollow
fromthissuggestionthat(11)and(12)couldbeusedtomeanwhat(11MP)and(12MP)indicate.Butitseemsthattheycouldnot
besoused.Anexplicitcompletionintheformof(11MP)or(12MP)themselves,isnecessarytoconveysuchaproposition.
Similarly,itdoesnotseemthatthenegationof(11)orof(12)couldbeusedtoconvey,respectively,thatTipperisnotreadyfor
anythingorthatAlhasnotfinisheddoinganything.Secondly,thissuggestionjustassumeswithoutargumentthatbecausean
utteranceofasentencelike(11)or(12)mustexpressacompleteproposition,theremustbe,correspondingtowhatappearsonthe
surfacetobeamissingsemanticorconceptualgap,anunderlyingsyntacticslotinthestructureofthesentence.Asyntactic
argumentisneededtosustainthissuggestion,anargumentcapableofreckoningwiththedifferencenotedabovebetween,for
example,finishandcomplete(orbetweeneatanddevourorconfessandadmit).Itmustexplainwhy(12)isgrammaticalbut(13)
(13)Alhascompleted.

isnot,[6]eventhough,bothfinishingandcompletingrequiresomethingtofinishorcomplete,contrarytowhatissuggestedby
thelexicaldifferencebetweenfinishandcomplete.
Noteherethataspecificationofwhatissaidintheutteranceofasemanticallyunderdeterminatesentencecanpreservethe
underdetermination.Inthecaseof(11)and(12),forexample,thefollowingspeechreportsorindirectquotations
(11IQ)SsaysthatTipperisready.
(12IQ)SsaysthatAlhasfinished.
arestrictlyaccurate.Ifastandalonesentencecanexpressmerelyapropositionalradical,itcandolikewisewhenembeddedina
thatclauseinindirectquotation.Eitherway,thereseemstobenosyntacticreasonwhyeverythingneededtodeliveracomplete
propositionshouldcorrespondtosomethinginthesyntacticstructureofthesentence.
Forthisreasonitwouldbemisleadingtoassimilate,asthesecondobjectiondoes,usesofsemanticallyunderdeterminate
sentencestothecategoryofellipticalutterances.Utterancesareelliptical,strictlyspeaking,onlyifthesuppressedmaterialis
recoverable,atleastuptoambiguity,bygrammaticalmeansalone,asintagquestionsandinsuchreducedformsasconjunction
reduction,VPellipsis,andgapping:
(28)Billishappiestwhenworking.
(29)BilllikesworkingandsodoesAl.
(30)BillwantspiefordessertandAlpudding.
Noticethat(28)unequivocallyentailsthatAlishappiestwhenAlisworking,(29)thatAllikesAlworking,notBillworking,
and(30)thatAlwantspuddingfordessert.Nocontextuallysalientsubstitutesareallowed.Asentencelike(31)canbetakenin
morethanoneway,
(31)IknowarichermanthanRossPerot.(isorknows?)
butthatisbecause(31)issyntacticallyambiguousitisnotsemanticallyunderdeterminate.Inalltheseexamplesindirect
quotationcanlegitimatelyincludeaparaphrasethatspellsoutthesuppressedmaterial,
(28IQ)SsaysthatBillishappiestwhenBillisworking.
(29IQ)SsaysthatBilllikesworkingandAllikesworking.
(30IQ)SsaysthatBillwantspiefordessertandAlwantspuddingfordessert.
and,ifnecessary,disambiguate,inthecaseof(31).
(31IQ)SsaysthatheknowsarichermanthanRossPerotknows.
Sincetherecoveredmaterialcorrespondstosomethinginthesentence,thoughnotnecessarilytosomethingthatis

phonologicallyrealized,thereisnoreasontodenythattheparaphrasespecifieswhatissaid.[7]Thisisnotthecasewithreports

http://userwww.sfsu.edu/kbach/impliciture.htm 5/27
9/20/2016 ConversationalImpliciture

thatincludethecompletionofanutteranceofasemanticallyunderdeterminatesentence,forinthatcasetheinsertedmaterialis
notonlyunheard,itisnoteventheresyntactically.Linguisticallyspeaking,itisnottheretoberecovered.Forthisreason,there
isnolinguisticbasisforincludingsuchmaterialinwhatissaid.

Thethirdobjectionisthatsemanticunderdeterminationisjustakindofindexicality.[8]Presumably,thisisnotbasedon
thefatuousbutoccasionallyvoicedviewthatanysortofcontextsensitivitycountsasindexicalitybutonthespecificfactthat
indexicalsarereferentiallyunderdeterminate.Asaresult,acompletepropositionisnotsuppliedprovidedbythelinguistic
contentofthesentencecontaininganindexical.Indexicalreferencesmustbeassignedbeforethereisacompleteproposition,

and,likecompletion,thisisapragmatic,notasemanticmatter.[9]Therationaleforthisobjectionisthatindexicalsdonotin
themselvesspecifytheirreferentsbutaremerelyusedtoindicatethem.So,forexample,inutteringShereturnedlastweek,one
doesntreallysaywhoreturnedwhen.Thatistruebutirrelevant.Theobjectionassumesthatthereisnorelevantdifference
betweenindexicalreferenceandfillinginconceptualgaps.Butthereis:indexicalreferencefixestheinterpretationofanelement
thatoccursintheutterance,beitapronoun,ademonstrativephrase,atemporalorlocationaladverb.Anindexicalislikeafree
variableneedingtobeassignedavalue.Ontheotherhand,theconceptualgapsinutterancesofsemanticallyunderdeterminate
sentencesdonotcorrespondtoanythinginthesentencesthemselves,notevenemptysyntacticcategories.Notbeingsentence
constituents,theyenterinnotatthelinguisticlevelbutattheconceptuallevel.Anindexicalisthereinthesentence.

3.ExpansionandSentenceNonliterality
Becausetheutteranceofasemanticallyunderdeterminatesentenceleavesoutaconceptualelement(orarelationbetween
conceptualelements),theprocessofcompletionisrequiredbeforeapropositionisyielded.Theprocessofexpansionisnot
requiredinthissenseitismandatednotconceptuallybutmerelypragmatically.Forinthiscasethereisalreadyacomplete
proposition,somethingcapableofbeingtrueorfalse(assuminglinguisticallyunspecifiedreferenceshavebeenassignedandany
ambiguitieshavebeenresolved),albeitnottheonethatisbeingcommunicatedbythespeaker.Thepropositionbeing
communicatedisconceptuallyenrichedorelaboratedversionoftheoneexplicitlyexpressedbytheutteranceitself(Ileaveaside
thecasewherewhatisexpandedisthecompletionofasemanticallyunderdeterminateutterance).So,forexample,ifamother
utters(1)tohercryingsonupsetaboutacutfinger,
(1)Yourenotgoingtodie.
sheislikelytomeanthatheisnotgoingtodiefromthatcut,notthatheisimmortal.Exampleslike(1)and(2),wherethe
implicitconceptualmaterialtobeinsertedappearsincurlybrackets,
(2)Ihaveeatenbreakfast.{today}
suggestthatexpansionisamatteroflogicalstrengthening,butingeneralthisisnotso.Thepropositionbeingcommunicatedcan
belogicallyweakerthantheoneexpressed,
(3)Ihaventeatenbreakfast.{today}.
anapproximationofit,
(4)Franceishexagonal.{roughly}
aprecisificationofit,
(5)Andreweighed500pounds.{exactly}
orevenlogicallyequivalenttoit,asin(6).
(6)Ihaveeatencaviar.{before}
Soexpansioninvolvesnotlogicalstrengtheningbutwhatmightbecalledlexicalstrengthening,inthatwhatisbeing

communicatedcouldhavebeenmadefullyexplicitbytheinsertionofadditionallexicalmaterial.[10]However,Iprefertothink
http://userwww.sfsu.edu/kbach/impliciture.htm 6/27
9/20/2016 ConversationalImpliciture

thisprocessasconceptualstrengthening,[11]becauseitisnotnecessaryfortheheareridentifytheexactwordsthespeakerhasin
mindbutonlywhatthosewordswouldcontributeiftheywereused.SowhenRecanatidescribestheunstrengthenedproposition
astheminimalpropositionexpressibleby[an]utterance(1989/1991,p.102,304),weshouldunderstand,contrarytowhathe
suggests,thatitisnotnecessarilyminimalinalogicalorinformationalsense.Itisminimalonlyinthesensethatitsconstituents
allcorrespondtoconstituentsofthesentence.Whatisminimalhereisthenottheinformationcontentofthepropositionbutits
departurefromthemeaningofthesentence.Maybeitshouldbecalledaskeletalproposition.
Caseslike(6)and(7)
(7)Johnhasthreecars.{atleast}
arenoteworthy,becausetheyillustratethatexpansioncanbelogicallyredundant.Theinclusionofanadditionalelement(likeat
least)canprecludealogicalstrengthening(liketheforceofexactly).Alogicallyredundantexpansionnullifiesthenormal,but
notstrictlyliteral,forceoftheutterance.Althoughthesentencein(7)wouldordinarilybeusedtostate(exactly)howmanycars
Johnhas,itcouldalsobeused,withanimplicitatleast,toindicate,forexample,whyJohnisineligibletoenteracarlottery
opentopeoplewithfewerthanthreecars.
Howshouldwecharacterizetheexpansionistuseoftheabovesentences,insofarastheyareeachusedtocommunicatea
propositionthatisconceptuallymoreelaboratethantheonethatisstrictlyexpressed?Firstofall,itseemsthatthedifference
betweenthetwopropositionsisnotattributabletoanyparticularconstituentofthesentence.Inthecaseof(1),forexample,
(1)You'renotgoingtodie.{fromthiscut}
themotherisusingeachofherwordsliterallybutisomittinganadditionalphrasethatcouldhavemadewhatshemeantfully
explicit.Ifhersonhadreplied,YoumeanI'mgoingtoliveforever,Mom?,itwouldnotbebecauseshewasbeingobscurebut
becausehewasbeingobtusehewouldbetakingherutterancestrictlyandliterally,notasshemeantit.Butifshewereusing
eachofherwordsliterally,howcouldshenotbespeakingliterally?Thisexampleillustratesacommonbutnotwidely
recognizedkindofnonliterality,wherebyasentenceisusednonliterallywithoutanyofitsconstituentsbeingsoused.Ifthe
motherhadsaid,Thiscutwillnotmakeyoucroak,shewouldhavebeenusingthewordcroaknonliterally.Butherutteranceof
(1)isacaseofsentencenonliterality(Bach,1987a,pp.7172).Itisunlikemetaphor,metonymy,andothersortsofconstituent
nonliterality,foritdoesnotinvolvethefigurativeuseofanyparticularwordorphrase.Evenso,leavingwordsoutisakindof
nonliteralityinitsownright.Youcanuse(2),forexample,
(2)Ihaventeatenbreakfast.
tomeanyouhaventeatenbreakfasttodayoreven(8)
(8)Ihaven'teaten.
tomeanyouhaventeatenbreakfasttoday.Similarly,youmightuse(9)
(9)Ihavenothingtowear.
tomeanthatyouhavenothingappropriatetoweartoacertainwedding,withouthavingtoincludeextraphrasesinyour
utterance.Usingsentencesnonliterallyinthiswayissocommonthatwetendneithertobeawareofdoingitnortothinkofitas
notliteralwhenothersdoit.Butwedoitallthetime(asIdidjustthenIleftoutwhenwespeak).Ratherthaninsertextra
wordsintoourutterancesinordertomakefullyexplicitwhatwemean,weallowourlistenerstoreadthingsintowhatwesay.
Eventhough,wemaynotintuitivelythinkofthisphenomenonasnonliteralitybecausenospecificwordsarebeingused
figuratively,itisawayofnotbeingliteral,becausewhatthespeakersaysisonethingandtheexpandedversionofittobe
identifiedbythehearerisanother.Aswithanyothersortofnonliterality,forthehearertounderstandwhatthespeakeris
conveyingintheutterance,hemustrecognizethatthespeakercannotbeplausiblybetaken,andthereforedoesnotintendtobe
taken,tomeanwhatheissaying(BachandHarnish,1979,pp.6570).Whereexpansionisinvolved,whatismeantisclosely

http://userwww.sfsu.edu/kbach/impliciture.htm 7/27
9/20/2016 ConversationalImpliciture

relatedtowhatissaid,sincetheformerpropositionisderivedfromthelatterbytheinsertionofconceptualmaterial,butisnot
identicaltoit.
Noticethatifwhatisconveyedimplicitlyhadbeenmadeexplicit,itdoesnotfollowthatthespeakersmeanngwouldhave
beenthesame.Thatisbecausetheveryactofmakingsomethingexplicitcanprecludesomethingthatwouldotherwisebeleft
open.Forexample,ifsomeoneutters(7)andintendsittobeexpandedasindicated,
(7)Johnhasthreecars.{atleast}
hewouldnotbeimplicatingorevensuggestingthatthiswasnottheexactnumber.Butifthepersonhadutteredthefull
expansion(7EX),
(7EX)Johnhasatleastthreecars.
therebymakingfullyexplicitwhatwaspartlyimplicitin(7),hecouldimplicatethatthreeisprobablynottheexactnumber.
Similarly,itisnoobjectiontomyaccount,onwhich(3)islogicallyequivalentto(3EX),
(3)Ihaventeatenbreakfast.
(3EX)Ihaventeatenbreakfastbefore.
thatsomeonewhoutters(3)canmeanthathehadnteatenbreakfastthatdaybutcouldnotmeanthatifhehaduttered(3EX)
instead.Forhowtheminimalpropositionisexpressedmakesadifference.JustasaGriceanimplicaturethatexploitsthemaxim
ofmannerisdetachable(Grice,1967/1989,p.39),soisanimpliciture.
NowRecanatihasarguedthatwhatIcalltheexpansionofwhatissaidiswhatissaid.Accordingtohisavailability
principle,incharacterizingwhatissaidweshouldalwaystrytopreserveourpretheoreticintuitions(1989,p.310),sothat
whatissaidinthecasesof(8)and(9)wouldbespecifiednotas
(8IQ)Ssaysthathehasnteaten.
(9IQ)Ssaysthathenothingtowear.
butas
(8EQ)Ssaysthathehasnteatendinnertoday.
(9EQ)Ssaysthathenothingtoweartothewedding.
Looselyspeakingthisiscorrect,andifonesintuitionsareinsensitivetothedistinctionbetweenwhatissaidlooselyspeaking
andwhatissaidstrictlyspeakingor,worse,iftheyareinsensitivetothedistinctionbetweenthe(locutionary)levelofwhatis
saidandtheillocutionarylevelofwhatisstated(seesection5),ofcourseonewillfind(8EQ)and(9EQ)acceptable.Nodoubtthe
intuitionsofthemanonthestreetarenotsensitivetosuchniceties.However,itdoesntfollowthathisinferencestospeakers
communicativeintentionsareasinsensitive.Forsuchinferencesmustbesensitivetothesemanticcontentofsentencesif
sentencesaretoprovidethelinguisticbasisforidentifyingspeakerscommunicativeintentions.Besides,theseuntutored
intuitionsareeducable.Manypeople,ifaskedtocompareanutteranceof(8EX)
(8EX)Ihaventeatendinnertoday.
withanutteranceof(8),willsaythattheydonotsaythesamething,contrarytowhatRecanatisintuitionssuggest.
Also,Recanatisexplicationofhisavailabilityprinciple,despiteitslipservicetointuition,ignorestheintuitionthatthe
constituentsofwhatissaidmustcorrespondtotheconstituentsoftheutterance.Ifsomethingdoesnot,itisnotpartofwhatis

SAID.[12]RecanatisrejectionofwhatCarstoncallsthelinguistic(orgrammatical)directionprinciple(1988,p.163)doesnot
takethisintuitionintoaccount.HismainreasonfornotacceptingGricesintuitiveunderstandingofthemeaningofsay,
accordingtowhichwhatissaidmustcorrespondtotheelementsof[thesentence],theirorder,andtheirsyntacticcharacter
(1969/1989,p.87),isthatthisleadstotheconsequencethatutteranceslike(1)(9)arenonliteral.Buthisonlyreasonforfinding
thatobjectionableisthatgenuinecasesofnonliteralitymustbelikemetaphor(1989,p.313),i.e.befigurative.However,as

http://userwww.sfsu.edu/kbach/impliciture.htm 8/27
9/20/2016 ConversationalImpliciture

explainedabove,sentencenonliteralityisnotlikefigurativeusesoflanguage,exceptinthegenericrespectthatwhatismeantis
distinctfromwhatissaid.
OnereasonRecanatisuggeststhatintuitionresistsreportingtheminimalpropositionaswhatissaidin(2)isthatitwould
bereportedas(2MQ)
(2MQ)Sissayingthathehaseatenbreakfastbefore.
HereRecanatiisrightaboutintuitionbutwrongtosupposethat(2MQ)givesaminimalistreportofwhatissaidin(2).Rather,it
givesaconceptualstrengthening,includingthewordbefore,ofwhatissaid,albeitonethatislogicallyequivalent.Noticethat
onRecanatisliberalpositiononwhatissaidsyntacticallyparallelsentencesdonotgetparallelsemantictreatment.Compare(2)
with(6)
(2)Ihaveeatenbreakfast.
(6)Ihaveeatencaviar.
or(9)with(10),
(9)Ihavenothingtowear.
(10)Ihavenothingtorepair.
forexample.(6)and(10)illustratethatwhatiscommunicatedcanbetheminimalproposition.Ifwetakesyntacticparallelism
seriously,weneednotbeconcernedthatontheviewthatwhatissaidincaseslike(2)and(9)isaminimalproposition,thisis
notwhatisnormallyortypicallycommunicated.OnRecanatisview,whatisimplicitintheutterancesof(2)or(9)ispartof
whatissaid,inwhichcasewhatissaidincludespragmaticallydeterminedelementsthatarenotassociatedwithconstituentsof
theutterance.Recanatiwouldnotobjecttothisresult,butitseemstome,forreasonsthatwillbecomeclearerinthenexttwo
sections,thatoncewerecognizeimplicitureasamiddlegroundbetweenexplicitcontentandimplicature,thereisnoreasonto
retainthisliberalposition.Bynotallowingintuitivedifferencesinmeaningtooverridesyntacticparallelisms,ifwedonot
recognizeaslotin(6)thatrepresentsatimeperiod,weshouldnotrecognizeonein(2).Similarly,despitethecommonsuggestion
thatquantifiednounphrasesrangeoverunspecifiedrestricteddomains,ifthereisnoreasontorecognizeadomainslotin(11),
(11)Johnisreadingabook.
thereisnoreasontorecognizeonein(12),
(12)Johnisreadingthebook.
eventhoughsomeonewhoutters(12)wouldundoubtedlyhaveinmindalimitedrangeofreadingmaterial.
Finally,intuitionsdonotseemtofavorRecanatisinflationaryconceptionofwhatissaidincaseswheretheliterally
expressedpropositionisnotallthatminimal(inthelogicalsense).Inthefollowingcases(adaptedfromHarnish,1976),
(13)Jacksonsquirtedthepaintonthecanvas.{intentionally}
(14)Georgesquirtedthegrapefruitjuiceonthetable.{unintentionally}
(15)JackandJillaremarried.{toeachother}
(16)Mr.JonesandMrs.Smitharemarried.{butnottoeachother}
(17)Mr.JonesandMrs.Smithareinlove.{witheachother}
eventhoughtheunuttered(bracketed)materialisunderstood,peoplearedisinclinedtoincludeitinthespecificationofwhatis
said.Theyappreciatethefactthatalthoughwhatisnotutteredisinferable,itisnotthere.Allinall,Idoubtthatthereisasmuch
intuitivesupportforRecanatisliberalconceptionofwhatissaidasheimagines.Asnotedabove,whatinclineshimtowardthis
conceptionishisassumptionthattruenonliteralitymustbelikemetaphor,i.e.befigurative,therebyrulingoutwhatIclassifyas
sentencenonliterality.Bethatasitmay,Iamnotdisputingtheideaunderlyinghiscontentionthattherearepragmaticaspectsto

http://userwww.sfsu.edu/kbach/impliciture.htm 9/27
9/20/2016 ConversationalImpliciture

whatissaidaswellastowhatisimplicated,butinmyviewtheseaspectsareproperlyregardedaspertainingtowhatisimplicit

inwhatissaid.[13]
Sofarwehavedistinguishedtwotypesofimpliciture,dependingonwhetherthehearermustdosomeconceptualfillingin
ofapropositionalradicalorfleshingoutofaminimalpropositioninordertoascertainwhatthespeakermeans.Fillinginis
neededifthesentenceissemanticallyunderdeterminate,andfleshingoutwillbeneededifthespeakercannotplausiblybe
supposedtomeanjustwhatthesentencemeans.Infact,bothprocessescanoccurwithinasingleutterance,aswith(18),
(18)Everybodyiscoming.
whichmightbecompletedtoyield
(18CM)Everybodyiscoming[tomyparty].
andthenexpandedtoyield
(18EX)Everybody{inmyclass}iscoming[tomyparty].

4.TheExplicit,theImplied,andtheImplicit
TheexamplesofthelasttwosectionsarequitedifferentfromGrices(1967a)wellknownexamplesofconversational
implicature.Implicaturesare,asGriceobserved,cancelableandcanbevagueorindeterminate,butthesameistrueof
implicitures.What,then,isthedifferencebetweenthetwo?Althoughbothimplicitureandimplicaturegobeyondwhatis
explicitintheutterance,theydosoindifferentways.Animplicatumiscompletelyseparatefromwhatissaidandisinferredfrom
it(moreprecisely,fromthesayingofit).Whatissaidisonepropositionandwhatiscommunicatedinadditiontothatisa
conceptuallyindependentproposition,apropositionwithperhapsnoconstituentsincommonwithwhatissaid.Forexample,
onemightuse(1)
(1)It'safter10.
notjusttogivethetimebuttoimplicatethatacertainrestaurantisclosed.Onemightuse(2)
(2)It'sraining.
notjusttodescribetheweatherbuttoimplicatethatonewillnotbemowingthelawnthatday.Incontrast,implicituresarebuilt
upfromtheexplicitcontentoftheutterancebyconceptualstrengtheningorwhatSperberandWilson(1986)callenrichment,
whichyieldswhatwouldhavebeenmadefullyexplicitiftheappropriatelexicalmaterialhadbeenincludedintheutterance.
Implicituresare,asthenamesuggests,implicitinwhatissaid,whereasimplicaturesareimpliedby(thesayingof)whatissaid.
Thefollowingexample(duetomystudentCindyHall)illustrateshowbothanimplicitureandanimplicaturecanbeproduced
bytheutteranceofthesamesentence.
(3)Maryhasaboyfriend.
AlikelyimplicitureisthatMaryhasexactlyoneboyfriend,andpossibleimplicatures,dependingonthecircumstances,arethat
thehearershouldntaskMaryout,thatMaryisnotalesbian,thatMaryisgettingadivorce,orthatMarywillgetadivorce.
IshouldpointoutthatmyuseofthetermexplicitismorerestrictivethanSperberandWilsons.Theycountasexplicit
anythingcommunicatedthatisadevelopmentofthelogicalformencodedby[theutteredsentence](1986,p.182).Thusthey
regardwhatIcallexpansionsandcompletionsasexplicatures,asexplicitcontentsofutterances.Ifindthisuseoftheterm
misleading,inasmuchastheconceptualstrengtheninginvolvedinexpansionorcompletionisnotexplicitatall.Includingthe
requisitelexicalmaterialwould,ofcourse,explicatewhatthespeakeriscommunicating,butonlythenwouldwhatisbeing
communicatedbemadefullyexplicit.
Implicituresgobeyondwhatissaid,butunlikeimplicatures,whichareadditionalpropositionsexternaltowhatissaid,
implicituresarebuiltoutofwhatissaid.Evenifnowordsorphrasesarebeingusedfigurativelyandevenafteranyambiguities

http://userwww.sfsu.edu/kbach/impliciture.htm 10/27
9/20/2016 ConversationalImpliciture

orindexicalreferencesareresolved,inimpliciturewhatthesentencemeansdoesnotfullydeterminewhatthespeakermeans
(whetherbecauseofsentencenonliteralityorsemanticunderdetermination).SofarasIcantell,theonlyexplanationforthefact
thatGricescriticscountimplicituresasexplicitcontentsofutterances,oridentifythemwithwhatissaid,isthatthey
uncriticallyassume,alongwithGrice,thatthereisnomiddlegroundbetweenwhatissaidandwhatisimplicated.Itiscuriousto
notethatGricehimselfoccasionallyalludedtowhatIamcallingimpliciture,aswhenheremarkedthatitisoftenunnecessaryto
putinqualificatorywords(1967b/1989,p.44).Althoughhediddescribesuchcasesasimplicatures,heappearedtohave
somethingdistinctiveinmind:strengtheningonesmeaningbyachievingasuperimposedimplicature(1967b/1989,p.48my
italics).Bystrengtheningheappearstohavemeantincreasingtheinformationcontentofwhatissaid,notaddingawhole
separatepropositiontoit.Nevertheless,Gricedidgivetheimpressionthatheintendedthedistinctionbetweenwhatissaidand
whatisimplicatedtobeexhaustive.Accordingly,sinceexpansionsandcompletionsarenotrelatedcloselyenoughto
conventionalmeaningtofallunderwhatissaid(inGricesfavoredsense),itdoesseemthatforhimtheywouldhavetocountas
implicatures.SperberandWilson,Carston,andRecanatiallfindthisresultunintuitive.Iagreewiththem,butratherthansuppose
thatwhatissaid(theexplicit)andwhatisimplicated(theimplicit)exhaustthe(propositional)significanceoftheutterance
(Carston,1988,p.155myitalics),Isuggestthatwesimplyrecognizeadistinctcategory,theimplicit,betweentheexplicitand
theimplicated.

5.GriceOnWhatIsSaid
ThenotionofwhatissaidplaysapivotalroleinGricesaccountofconversationalimplicatureand,byextension,inmyaccount
ofimpliciture.Heacknowledgesthathisfavoredsenseofsayisstipulativeandadmittedlyartificial(1968/1989,p.118).Still,
howoneusestheterm,eveninthecontextofatheory,shouldnotbeentirelyarbitrary,andGriceintendshisusetocomportwith
theintuitiveunderstandingofthemeaningofsay(1967a/1989,pp.2425),onwhichwhatissaidmustcorrespondtothe
elementsof[thesentence],theirorder,andtheirsyntacticcharacter(1969/1989,p.87).Herehementionsthathowsomethingis
putmayenterintowhatissaid.Hisexampleconcernsreference.Heallowsthatsomeonewhoutters(1)andsomeonewho(in
1967)utters(2)mightnotsaythesamething,
(1)HaroldWilsonisagreatman.
(2)TheBritishPrimeMinisterisagreatman.
buthedoesnotcommithimselfonthispoint.However,ifwordorderandsyntaxreallydoaffectwhatissaid,thenutterancesof
(3),(4),and(5)donotsaythesamething,
(3)JohnlovesMartha.
(4)MarthaislovedbyJohn.
(5)ItisJohnwholovesMartha.
eventhoughtheyaretruthconditionallyequivalent.Hereisanotherwayinwhichlinguisticcontentconstrainswhatissaid.
SupposeTomutters(6),
(6)Iregretgoinghome.
therebysayingthatheregretsgoinghome.Thisentailsthathebelievesthathewenthome(toregrethavingdonesomethingisto
wishyouhadn'tdoneit,andthatrequiresbelievingthatyoudidit),butwearedisinclinedtoregardthataspartofwhathesaid.
Itisnotingeneraltruethatanythingentailedbywhatissaid,thoughinasensepartofwhatissaid,issaidinitsownright.
Itseems,then,thatonGrice'snotionthecontentsofwhatissaidarenottobeindividuatedmerelybytruthconditions.
Theyneedtobeindividuatedmorefinelythanintermsofpropositions.Letmesuggest,eventhoughIhavebeenandwill
remainneutralabouttheontologicalcharacterofpropositions,thatforGricethecontentsofwhatissaidarestructured

http://userwww.sfsu.edu/kbach/impliciture.htm 11/27
9/20/2016 ConversationalImpliciture

propositions,propositionsassociatedwithsyntacticforms.IwillnottrytoexplicatethisnotionbutjustfollowwhatItaketobe
Grice'sintuitiveunderstandingofthemeaningofsayandsimplyassumethatacriterionofclosesyntacticcorrelation
distinguisheswhatissaidineachof(3),(4),and(5),despitetheirtruthcondtionalequivalence.Bythiscriterion,itisclearthat
thereisalwaysadifferencebetweenwhatissaidinanutteranceandwhatwouldbesaidinanexpandedorcompletedversionof
thatutterance.
SoitwouldseemthatforGriceanythingcommunicatedinanutterancethatcloselycorrespondstoitsformcountsaswhat
issaidandthatanythingelsecountsasbeingimplicated.However,heisledtocomplicatehisdistinctionsomewhat,onaccount
oftwoconsiderations(notexpansionandcompletion,ofcourse).First,hedistinguishedsayingfrommerelymakingasiftosay
(1967a/1989,p.30),asinironyandmetaphor(p.34),andallowedthatmakingasiftosay,likesaying,cangenerateimplicatures.
Second,heinsistedthatpartofwhataspeakermeanscanbecloselyrelatedtoconventionalmeaningandyetnotbepartofwhat
issaiditisimplicatednotconversationallybutconventionally.Wewilltakeupthisnotioninthenextsection.
ForGricesayingsomethingentailsmeaningit.Thisiswhyheusesthelocutionmakingasiftosaytodescribeirony,
metaphor,etc.,sinceinthesecasesonedoesnotmeanwhatoneappearstobesaying.Mostofuswoulddescribethesemore
straightforwardlyascasesofsayingonethingandmeaningsomethingelseinstead.Thatswhatitistospeaknonliterallyat
leastifonedoessointentionally.Onecanalsounintentionallysaysomethingotherwhatonemeans,owingtoaslipofthe
tongue,amisuseofaword,orotherwisemisspeaking(Harnish,1976/1991,p.328).Finally,onecansaysomethingwithout
meaninganythingatall,asincasesoftranslating,reciting,orrehearsinginwhichoneuttersasentencewithfullunderstanding
(oneisntjustpracticingonespronunciation)andyetisnotusingittocommunicate.Toreckonwiththesevariouswaysof
sayingsomethingwithoutmeaningit,GricecouldhaveinvokedAustinsdistinctionbetweenlocutionaryandillocutionaryacts,
buthedidnot.Austin,itmayberecalled,definedthelocutionaryact(specificallytherheticact)asusingcertainvocableswith
acertainmoreorlessdefinitesenseandreference(1962,p.95).ThatsoundsalotlikeGricesnotionofsaying,exceptthatfor
Gricesayingsomethingentailsmeaningit:theverbsay,asGriceusesit,doesnotmarkaleveldistinctfromthatmarkedbysuch
illocutionaryverbsasstate,tell,ask,etc.,butratherfunctionsasagenericillocutionaryverb.Itdescribesanyillocutionaryact
whosecontentismadeexplicit.SincevirtuallyallofGricesexamplesinvolveindicativeutterances,inpracticeheusessayto
mean'explicitlystate'.Indeed,theoriginalversionofhisdistinction(inGrice,1961)wasbetweenstatingandimplying.Clearly

Griceoptedforthewordsayinordertobroadenthescopeofhisdistinctionbeyondstatements.[14]
Consideringthathedescribesnonliteralutteranceslikeironyandmetaphornotassayingbutasmakingasiftosay,itis
puzzlingthatGriceshouldhaveassimilatedthesetoimplicature.Intuitively,onethinksofimplicatingasstatingormeaningone
thing(i.e.sayingsomethinginGricesfavoredsense)andmeaningsomethingelseinaddition,notasmeaningsomethingelse
instead.Havingdeniedthatironyandmetaphorarecasesofsayinginhissense,hedidnotneedtodescribetheirnonliteral
contentsasimplicatures.Sinceimplicatureisakindofindirectspeechactwhereasironyandmetaphorarespeciesofnonliteral
butdirectspeechact(BachandHarnish,1979,ch.4),thelattershouldnotbeclassifiedasimplicature.Unfortunately,thisis
donebybothGriceandmanyofhiscritics.
Iamsuggesting,then,twowaystoimproveonGricestaxonomicschemewhileretainingthecriterionofclosesyntactic
correlation.WereplaceGricesdistinctionbetweensaying(inhisfavoredsense)andmerelymakingasiftosaywiththe
distinction(inindicativecases)betweenexplicitlystatingandsayinginAustinslocutionarysense,andwedistinguish
nonliterality(includingsentencenonliterality)fromimplicature.Inthiswaywehaveanotionofwhatissaidthatapplies
uniformlytothreesituations:(1)wherethespeakermeanswhathesaysandsomethingelseaswell(implicatureandindirect
speechactsgenerally),(2)wherethespeaker(intentionally)saysonethingandmeanssomethingelseinstead(nonliteral
utterances),and(3)wherethespeakersayssomethinganddoesntmeananything.Whatissaid,beingcloselytiedtothe(ora)

http://userwww.sfsu.edu/kbach/impliciture.htm 12/27
9/20/2016 ConversationalImpliciture

meaningoftheutteredsentenceprovides(allowingforindexicalityandambiguity)thehearerwiththelinguisticbasisfor
inferringwhat,ifanything,thespeakermeansinadditionorinstead.Also,itallowsforthecaseinwhichthespeakerdoesnot
saywhatheintendstosay,asinthemisuseofawordoraslipofthetongue.Forthisreason,thespeakerisnotthefinalauthority
onwhathesaid.
Giventhecriterionofclosesyntacticcorrelation,onwhichwhatissaidneednotbeacompleteproposition,impliciturecan
beamatterofeitherfillinginorfleshingoutwhatissaid.Completionisthefillinginofapropositionalradical,andexpansionis
thefleshingoutoftheminimalpropositionexpressiblebyanutterance.IagreewithGricescriticsthatneitherisacaseof
implicature,althoughbothinvolvebasicallythesamesortofpragmaticprocessasinimplicatureproper,butIseenoreason,as
theydo,toextendthenotionofexplicitcontent,ofwhatissaid.Formethereisinexplicitmeaningbutnoinexplicitsaying.

6.ConventionalImplicature?
Gricealsodisallowedinexplicitsaying,buthedidrecognizeacategoryofexplicitnonsaying.Hethoughttherecanbeelements
ofwhatiscommunicatedcorrespondingtothemeaningofelementsinthesentencethatdonotenterintowhatissaid.Becauseof
thiscorrespondencetheyleadtoconventionalratherthanconversationalimplicatures,propositionswhicharemerelyindicated.
Gricesexamplesofproblematicelementsareconnectives,suchastherefore(1967a/1989,p.251968/1989,p.120)andbut
(1969/1989,p.88).Suchaconnectivemakesacertaincontribution,givenbyitsconventionalmeaning,towhatisbeing

communicated.[15]Itindicatesacertainrelationbetweenthetwoitemsitconnects,e.g.thatoneisaconsequenceoftheotheror
thatthereisacontrastbetweenthetwo.Gricedeniesthatthislinguisticallyspecifiedrelationentersintowhatissaid.
Gricesbriefdiscussionsofconventionalimplicatureareintendedtonarrowdownthesenseofsaythathefavorsbecauseof
itstheoreticalutility(1968/1989,p.121),inthiscasetoprovideforanelementofliteralcontentthatisnottruthconditional.
So,forexample,hedeniesthatanutteranceofthesentence
(1)HeisanEnglishmanheis,therefore,brave.
wouldbe,strictlyspeaking,falseshouldtheconsequenceinquestionfailtohold(1967a/1989,pp.2526).Butthisis
implausiblethespeakerdoesseemtobesayingthatthesecondclaimisaconsequenceofthefirst.Justbecauseconnectives
likethereforeandbutare,unlikeand,nottruthfunctionaldoesnotmeanthattheydonotenterintotruthconditions.Sowhat
mighthaveledGricetosupposethattheirimportismerelyimplicated?WhenanutteranceoftheformpCONJ qimpliespandq
butconveysmorethanmereconjunction,thereisnowaytoexplicateitsimportoverandabovepandqwithoutusingathird
clause,anadditionalconjunct,e.g.totheeffectthatthereisarelationofconsequenceorcontrastbetweenpandq.A
specificationofwhatissaidintheaboveexamplewould,accordingtoGrice,taketheform
(1GIQ)SsaidthatacertainmaleisanEnglishman,thatheisbrave,
andthatbeingbraveisaconsequenceofbeingEnglish.
Nowifthethirdconjunctispartofwhatissaid,whatissaidwouldcontainonemoreclausethaniscontainedinthesentence
usedtosayit.Thisfurtherconjunctwouldnotcorrespondtoaclauseinthatsentenceandcouldnotcountaspartofwhatissaid.
Fortheelementsofwhatissaid(inGricesfavoredsense)mustcorrespondtoelementsinthesentence.Thefurtherconjunct,not
beingsuchanelement,cancountonlyasa(conventional)implicature.
ThetroublewithallthisisthatwhatissaidinutterancesoftheformpCONJ q,eventhoughitimpliespandq,doesnot
havetobespecifiedinthreeclausesbysomethingoftheform,pandqand.Intheabovecase,forexample,thereisnoreason
whythewordthereforecannotgodirectlyintoatwoclausespecificationofwhatissaid:
(1IQ)SsaidthatacertainmaleisanEnglishman[and]thereforeheisbrave.

http://userwww.sfsu.edu/kbach/impliciture.htm 13/27
9/20/2016 ConversationalImpliciture

WhatissaidistruejustincasetherelevantmaleisanEnglishmanandisbrave,andbeingbraveisaconsequenceofbeing
English,butofcoursewhatissaidisnotidentical,thoughitisequivalent,totheexplicationofitstruthcondition.Toappreciate
this,consideranentirelydifferentsortofcase,example(6)oftheprevioussection,whereTomutters'Iregretgoinghome',
therebysayingthatheregretsgoinghome.Thisentailsthathebelievesthathewenthome,butthatisnotpartofwhathesaid.It
isnotingeneraltruethatanythingentailedbywhatissaidisitselfsaid.Butthatdoesn'tmeanitismerelyimplicated
(conventionally).Sincewearetakingstructuredpropositionstobewhatissaid,acomplex,multiclauseanalysisofwhatissaid
isnotidentical,thoughconceptuallyequivalent,towhatissaid.Gricesownexamplessuggestthatwhathecallsconventional
implicatureisreallyentailment.Entailmentsareimpliedbywhatissaid,buttheyarenotimplicatedbythesayingofwhatis
said.
Ibelieve,then,thatinGricesallegedexamplesofconventionalimplicature,whichallinvolvenontruthfunctional
connectives,thespecificationofwhatissaid,asillustratedby(1IQ),canandshouldincludetherelevantconnectivebutnotin
aseparateclause.Griceisledtoconventionalimplicatureineachcaseonlybecausehearbitrarilyinsistsonforcingthese
specificationsintosetofindependentconjuncts,whereuponthespecificationmusteitherincludeoneclausetoomanyoromit
theconventionalforceoftheconnective.Withthisisinmind,wecandealwithvarioussortsofexpressionsnottakenupby
Gricewhoseuse,asLarryHornhasremindedme,hasbeenthoughttoyieldconventionalimplicatures.Theseincludeparticles
likeevenandtoo,implicativeverbslikemanageandfail,factiveverbslikeforgetandrealize,andcleft(Itwaswho)and
pseudocleft(WhatXdidwas)constructions.Ithinkthatevenandtooenterintotruthconditionsinwaysanalogoustobut
somesortofcontrastispartofthetruthconditionalcontent,althoughtheprecisecontrastisunspecified,asinacaselike(2),
(2)EvenBilllikesMary.
KarttunenandPetershavearguedthattheembeddingof(2)in(3),
(3)HejustnoticedthatevenBilllikesMary.
doesnotmeanthathehasjustnoticedthatotherpeoplelikeMaryorjustnoticedthatBillistheleastlikelypersontodoso
(1979,p.13).Butallthisshowsisthattherelationofnoticingisnotdistributive,asillustratedby(4),
(4)InoticedthatBillhasthreecars.
whichcouldbetrueevenifIalreadyknewthathehastwo.Hornpointsoutthatanimplicativeverblikemanage(to)seemsto
addsomesortofadverbialcontentwhilefunctioningsyntacticallylikeamainverb,butitisnotcleartomewhythisshould
suggestthatconventionalimplicatureisinvolved.Theadverbialcontentofsuchverbscanjustenterstraightforwardlyintothe
wholetruthconditionalcontentofthesentencesinwhichtheyoccur.Forexample,in(5),
(5)Billmanagedtofinishhishomework.
thetruthconditionalcontentincludesboththefinishingandtheentaileddifficulty.Withfactiveverbs,asin(6),
(6)Billforgotthathehadanappointment
thathehadanappointmentispartofthetruthconditionalcontentofwhatissaiditisnotsomethingsaidinitsownright(and
certainlynotthecontentofanillocutionaryactofassertion).Ananalogouspointappliestocleftandpseudocleftconstructions,
althoughtheirform,likecontrastivestress,marksaspecial,illocutionaltopiccommentrelation(Atlas,1989,pp.8191).
Gricesfavoredexamples,theconjunctionsthereforeandbut,constraststrikinglywithhisgeneraldiagnosisofwhatgives
risetoconventionalimplicatures:Theelementsintheconventionalmeaningofanutterancewhicharenotpartofwhathasbeen
saidarelinkedwithcertain[noncentral]speechacts(1968/1989,p.122).Herehegivestheexampleofmoreover,whichis
linkedtothespeechactofadding,anactthatrequirestheperformanceofacentralspeechact,likereportingorpredicting.Grice
doesnotindicatewhichnoncentralspeechactsthewordsbutandthereforearelinkedtopresumablytheseareactsofmakinga
contrastanddrawingaconclusion(orgivinganexplanation).However,moreoverisrelevantlydifferentfromthetwoother

http://userwww.sfsu.edu/kbach/impliciture.htm 14/27
9/20/2016 ConversationalImpliciture

connectives:unlikemoreover,butandthereforesignifyrelationshipsbetweenthe(putative)factsexpressedbytheclausesthey
connect.ThesameistrueofvariousotherconjunctionsandadverbialsnotmentionedbyGrice,suchasafterall,although,
anyway,because,despite,evenso,for,however,nevertheless,since,so,still,thus,andyet.Takethecaseofbecause,whichis
linkedtotherathercentralspeechactofexplaining.Inanutteranceof(7),
(7)Becausetheverdictwasunjust,ariotbrokeout.
whatissaidisspecifiedby(7IQ).
(7IQ)Hesaidthatbecausetheverdictwasunjust,ariotbrokeout.
Becauseisusedtoexpressanexplanatoryrelationofsomesortbetweenthefactsdescribedbythetwoclauses.Butthereis
anotheruseofbecausethatdoesfitGricesparadigm.Compare(7)with
(8)Becauseyoullfindoutanyway,yourwifeishavinganaffair.
In(8)becauseisnotbeingusedtoexpressanexplanatoryrelationbetweenthefactsdescribedbythetwoclauses.Rather,the
speakerisusingthebecauseclausetoexplainhisspeechactofinformingthehearerofthefactexpressedbythesecondclause.
Accordingly,thehearercannotplausiblyspecifywhatisbeingsaidasaconjunction,asin
(3IQ)SissayingthatbecauseIllfindoutanyway,mywifeishavinganaffair.
Gricesdiagnosisiscorrect:specificationsofnoncentralspeechactsdonotfitcomfortablyintospecificationsofwhatissaid.
Thisholdsforthefollowingassortmentoflocutions,atleastwhenused(astheygenerallyare)toperformnoncentralspeechacts:

accordingly,afterall,allinall,allthingsconsidered,although,anyway,asitwere,atanyrate,besides,bethatasit
may,bytheway,consideringthat,disregarding,evenso,finally,firstofall,frankly,ifImaysayso,ifyouwantmy
opinion,incontrast,inconclusion,inshort,inviewofthefactthat,leavingaside,looselyspeaking,nevermind
that,nevertheless,nottointerruptbut,nottomentionthat,nowthatyoumentionit,ontheotherhand,sotospeak,
speakingformyself,strictlyspeaking,takingintoaccountthat,tobebluntaboutit,tobeginwith,tochangethe
subject,todigress,togetbacktothesubject,togettothepoint,tooversimplify,toputitmildly

Thereisastraightforwardexplanationwhythesediscourseconnectivesdonotfitcomfortablyintospecificationsofwhatissaid:
theyareinconstructionsyntacticallybutnotsemanticallywiththeclausestheyintroduce.Syntacticallytheyaresentence
adverbialsbuttheyfunctionasillocutionaryadverbials(BachandHarnish,1979,pp.219228),modifyingnotthemainclause
butitsutterance.Theresultisasitwereasplitlevelutterance.Thatiswhywecannotreportutteranceslike(9)and(10)inthe
formof(9IQ)and(10IQ).
(9)Frankly,Jerryismakingabigmistake.
(9IQ)#HesaidthatfranklyJerryismakingabigmistake.
(10)Incontrast,Danwouldneverdoathinglikethat.
(10IQ)#HesaidthatincontrastDanwouldneverdoathinglikethat.
Allinall,Idonotbelievethatweneedtoresorttothenotionofconventionalimplicaturetodescribetheconventional
importoftheabovelocutions,theseillocutionaryadverbials.Rather,asGricehimselfputsit,theyareusedtoperformnoncentral
speechacts,suchassimplifying,qualifying,andconcluding.Oneisnotconventionallyimplyinganythinginusingsucha
locutionrather,oneisprovidingsomesortofglossorrunningcommentaryononesutterance,e.g.,concerningits
conversationalrole.Soitseemsthatwecandowithoutthenotionofconventionalimplicaturealtogether:inGricesexamplesof
connectiveswithtruthconditionalimport,theconventionalmeaningoftheproblematicelementdoesenterintowhatissaid

http://userwww.sfsu.edu/kbach/impliciture.htm 15/27
9/20/2016 ConversationalImpliciture

andinthewideassortmentoflocutionsusedtoperformnoncentralspeechacts,theproblematicelementdoesnotenterintowhat

issaid,butitdoesnotgenerateaconventionalimplicatureeither.[16]

7.LexicalCompletionandExpansion
Theillustrationsofimplicitureinsections2and3requiredcompletionofutterancesofsemanticallyunderdeterminatesentences
orexpansionofsentencenonliteralutterances.Itappearsthatsimilarphenomenacanoccuratthelexicalandatthephrasallevel.
[17]Wewillfocusonthelexicalcase,whichhasbeeninvestigatedindepthbyRuhl(1989).

Ruhlcontendsthataconsiderablepartofallegedlexicalmeaningisactuallysuppliedbyothermeans(1989,p.86)and
thatdictionariesareinahabitofoverspecifying,ofattributingtowordsmeaningthatinpartissuppliedbythecontext(p.1).
TocounteractthistendencyRuhladoptsamonosemicbias:startwiththehypothesisthatawordhasasinglemeaning,then

supposethatitsmeaningsarerelatedbygeneralrules,andonlyasalastresortacceptambiguityorhomonymy(p.4).[18]He
suggeststhatwithmanycommonlyused,multipurposewords,liketheverbsget,hit,put,andtakeandtheprepositionsat,in,
on,to,andwith,linguisticmeaningishightlyabstract[i.e.,schematic]andremotefrompracticalusefulness(p.7),sothatwhen
wehearanyoftheabovewordsinaparticularlinguisticenvironmentandconversationalcontext,weimportextralinguistic
informationintoourunderstandingoftheutterance.Comparetheoccurrencesofputandonin(1)and(2),forexample.
(1)Alputthebeeronthetable
(2)Alputtheburdenonthelawyer.
Weinvokeextralinguisticknowledge,aboutbeerandtablesandaboutburdensandlawyers,tointerpret(1)and(2)intheway
wedo.Itisnotasemanticfactthatoneisnotlikelytomeanwith(2),forexample,thatAlphysicallyplacedsomethingon
someone.
Ruhlrejectstheviewthatconcretesensesaremorebasicthanabstract[i.e.notspatiotermporal]sensesorthat
concrete/abstractisthecontrastofliteralversusmetaphoric(p.168).Rather,hemaintains,wordsarehighlyabstract[i.e.
schematic]ininherentmeaning,oftentoomuchsoforconsciousunderstanding(p.86).Ourintuitionsaboutwordmeaningsare
unreliablebecauseselective:initstendencyforpolarextremes,consciousnesscannotholdallthepossibilitiesatonce(p.
125),henceourtendencyforoversemanticizing(p.96).Tosupportshiscontentions,Ruhlpresentshundredsofusesofeachof
thewordsheinvestigates.Cruse(1992)hasarguedthatRuhloverstateshiscaseincertainways,butCrusedoesacknowledgethe
methodologicalbenefitsofthemonosemicbias.Forusitwillsufficetoassumethatthereareatleastsomewordswhose
meaningsareschematicorabstractinthewayRuhldescribes.
Howmeaningcanbeschematicisnicelyillustratedbyafamiliarexampleofsemanticunderdeterminationatthephrasal

level,namelythegenitiveconstruction.[19]Usingtheexample,Petersbat(nevermindtheambiguityofbat),Sperberand
Wilsonlistvariouspossiblerelationsthatthisphrasemightbeusedtopickout(beingownedby,beingchosenby,beingkilled
by,beingmentionedby,etc.),andremark,Itishardtobelievethatthegenitiveisambiguous,withasmanysensesasthereare
typesofrelationshipsitmaybeusedtodenote,orthatalltheserelationshipsfallunderasingledefinitionwhichistheonly
meaningexpressedbyuseofthegenitiveonanygivenoccasion(1986,p.188).Ontheotherhand,if,asRecanatisuggestsin
regardtotheexampleJohnsbook,theonlyconstraintlinguisticallyimposedontherelationbetweenJohnandthebookisthat
itbearelationbetweenJohnandthebook(1989,p.298),thentherewouldbenoexplanationfortheapparentfactthatone
cannotuseJohnsbooktomeanthebookthatJohndroppedorusePetersbattomeanthebatthatPetergothitby.Ifthereisa
constraintontherelation(perhapstheconstraintvarieswiththetypesofrelata)butif,asSperberandWilsonsuggest,itcannot

http://userwww.sfsu.edu/kbach/impliciture.htm 16/27
9/20/2016 ConversationalImpliciture

begivenbyadefinition,thenRuhlisrightabouthowabstractlinguisticmeaningcanbe.Atanyrate,onethingisclear:in
uttering(3)
(3)Petersbatisgrey.
onecannotbesayingthatthebattowhichPeterbearssomerelationorotherisgrey(orworse,beexpressingadisjunctionof
propositions,eachinvolvingadifferentrelation).Rather,(3)issemanticallyunderdeterminate.UnlessthephrasePetersbatis
enriched,anutteranceof(3)is,asSperberandWilsonobserve,lessthanfullypropositional(1986,188).
ThesamesortofthingwouldseemtoholdtrueforlexicalitemsfittingRuhlsaccount.If,forexample,theseemingly
distinctsensesofgetorinarereallypragmaticspecializationsofasinglesense,thenunderstandingutterancesofsentenceslike
(1)and(2)or(4)and(5)
(4)Youllgetasandwichinabag.
(5)Youllgetagoodideainanhour.
requiresnotdisambiguationofthosewordsbutidentifyingtheirintendedpragmaticspecializations.
Ruhlapplieshisaccounttoamatterofspecialphilosophicalinterest,thetypetokendistinction.Hearguesthatthis
distinctiondoesnotgiverisetoanambiguityinwordslikecar,dog,andbookrather,theirspecificandgenericsensesare
modulationswordsareunspecifiedforthisdistinction(1989,p.106).In(6),forexample,
(6)EnzodrivesthesamecarIusedtodrive.
theuseofcarislikelytobespecific,whereasin(7)
(7)EnzodrivesthecarIusedtodrive.
itislikelytobegeneric,but,claimsRuhl,thisdifferenceisnottraceabletoanyambiguityintheword.Itstwouses(actually,
Ruhloftenmisleadinglyusessenseratherthanuse)arepragmaticspecializationsofitssinglemeaning.Onemajorconsideration
insupportofRuhlspositionisthatahugenumberofwords,namelyallwordsforkindsofthings,exhibittheallegedambiguity.
Toregardthedoubleuseasalexicalfactabouteachwordwouldbetomissthatgeneralization.Thesamepointwouldseemto
applytoanotheritemofphilosophicalinterest,thesocalledactobjectorprocessproductambiguity,apparentlyexemplifiedby
wordslikeattachment,building,creation,description,statement,belief,andmeaning.Tocallthisanambiguityandtotreatitas
lexicalwouldbetomissthegeneralityofthephenomenon.
Therearemanyothercandidateexamplesofsemanticallyunderdeterminatewordsandphrases.Particularcasesmaybe
disputed,andRuhlsclaimsofmonosemymaybeexaggerated,butinsofarastherearegenuinecasesofit,theirpresenceinan
utterancemeansthatpragmaticspecializationisneededforadeterminatepropositiontobeexpressed.Whatisrequiredheremay
becalledlocalcompletion.Also,evenifwordslikegetandwitharenot,asRuhlsuggests,reallymonosemous,theycouldstill
besemanticallyunderdeterminatewithrespecttoeachoftheirsenses.Inthatcaseonceasenseisselected,amorespecific
construalofthewordisneededbeforeadeterminatepropositionisreached.Inanyevent,itseemsthatweshouldincludelocal
completion,alongwithdisambiguationandreferenceassignment,amongthepragmaticprocessesthatenterintothe
determinationoftheexplicitcontentofanutterance.
Theothermainway,accordingtoRuhl,thatlinguisticmeaningunderdeterminesuseispragmaticgeneralization,as
exemplifiedbymetonymy.Heoffersexampleslikethefollowing(1989,pp.9899),wheretheitalicizednoun(togetherwithits
determiner)isnotmeantliterallybutasthecompletiongivenincurlybrackets:
(8)Itookherpoemoutofherpocket.{apaperonwhichherpoemwaswritten}
(9)Thiscaniscontaminated.{thecontentsofthiscan}
(10)Turnupthehifi.{thevolumeofthehifi}
(11)Turnoffthecereal.{theheatunderthecereal}

http://userwww.sfsu.edu/kbach/impliciture.htm 17/27
9/20/2016 ConversationalImpliciture

WhatRuhlcallspragmaticgeneralizationinthesecasesofmetonymy[20]mayberegardedasaprocessoflocalexpansion.Itis
local,unliketheexamplesofsection3,becausethenonliteralityisattributabletoaspecificexpression.
Ruhldoesnotmakeclearthatmetonymyisaratherspecialcase,forothercasesoflocalexpansiondonotinvolve
pragmaticgeneralization.Contrast(12)with(13)or(14)with(15),forexample.
(12)Thethieftookawatch.
(13)Thethieftookhisownwatch.
(14)Thethieffelltothefloorwhenorderedto.
(15)Thethieffelltothefloorwhenshot.
Oneislikelytousetooktomeanstealin(12)butnotin(13),orfelltomeanfellintentionallyin(14)butnotin(15).
However,asRuhlwouldsay,thesespecializedusesarenotdistinctsenses.Onlythemoregeneralsenseissemantic.The
specializedusewouldseemtobeacaseofpragmaticspecialization,butunlikeexamples(1)(7),sentences(12)and(14)arenot
semanticallyunderdeterminate.Rather,theyareusedtocommunicatepropositionsmorespecificthantheonesstrictlyexpressed.
Sothesearecasesoflocalexpansionratherthancompletion.However,incontrasttometonymytheyarenotcasesofpragmatic
generalization.Itappears,then,thatthedistinctionbetweenlocalcompletionandexpansioncutsacrossRuhlsdistinction
betweenpragmaticspecializationandgeneralization.
Letusconcludethissectionwithtwospecial,becausemetalinguistic,casesoflocalexpansion.Firstthereisthesocalled

usementiondistinction.[21]Utterancesof(16)and(17),forexample,
(16)Californiaisalongstate.
(17)Californiaisalongname.
arelikelytobeabouttheplaceandtheword,respectively(quotationmarkswereomittedin(17)becausetheydonotoccurin
speech).However,theusementiondistinctioninvolvesneitherambiguitynorsemanticunderdetermination.Forthedifference
betweenutterancesof(16)and(17)doesnotcorrespondtoanythingspecifictothenameCaliforniaonecanuseany
expressiontorefertothatveryexpression.Isuggestthatweviewmentioningasinvolvingaspecialsortoflocalexpansion:an
expressionEisusedasshortfortheexpressionE.Thehearercanrecognizesuchausewhen,forexample,Eoccursin
subjectpositionandthepredicateisnotplausiblyapplicabletoit(thiswillbecompletelyobviousifEisnotevenanoun
phrase,asinApproximatelyisalongword).
Finally,thereisthecaseofmetalinguisticnegation,investigatedatlengthbyHorn(1989,pp.362444).Utterancesof
sentenceslike(18)and(19)
(18)Itdoesnt{only}lookexpensiveitisexpensive.
(19)Thechefdoesntcreate{mere}mealsbutworksofart.
areamenable,asindicatedinthebrackets,toastraightforwardexpansionisttreatment.Butinsertionsofonlyormeredonotwork
forutterancesofsentenceslikethefollowing:
(20)IdidnttraptwomongeeseItrappedtwomongooses.
(21)Hesnotananimaldoctorhesaveterinarian.
(22)IwasreferringnottoDonnELLanbuttoDONNellan.
(23)Imnothisbrotherhesmybrother.
(22),forexample,wouldobviouslynotbeusedtomeanIwasreferringnotonlytoDonnELLanbuttoDONNellan.Whatis
goingonintheseutterancesisthatthespeakerisobjectingtoonewayofputtingsomethingandputsitanotherway.Because
theforceofthismetalinguisticnegationisquitedifferentfromthatofordinarynegation,Hornmaintainsthatnegationis
effectivelyambiguous,albeitacaseofpragmaticambiguity,abuiltindualityofuse,becausemetalinguisticnegationisnot

http://userwww.sfsu.edu/kbach/impliciture.htm 18/27
9/20/2016 ConversationalImpliciture

reducibletoasuitablyplacedItisnotthecasethat(1989,p.370).However,hedoesnotconsiderthepossibilitythat

metalinguisticnegationisacaseofnonliteralityorthatunderstandingitinvolveslocalexpansion.[22]Isuggestthatineachcase
expansionyieldsthedesiredresultifaphraselikewhatIwoulddescribe/pronounceas"isinserted.Suchaphraseseemsto
specifywhataspeakermeansintheseexamples.Noticethatitisonlyuponhearingthecorrectedformulation(butto
DONNellan)ofthenegatedproposition(IwasreferringnottoDonnELLan),thatthehearercanfigureoutthatthespeakerisnot
assertingthenegativeproposition.

8.InformationandProcessinginImpliciture
Completionandexpansionarebothprocesseswherebythehearersuppliesmissingportionsofwhatisotherwisebeingexpressed
explicitly.Withcompletionapropositionalradicalisfilledin,andwithexpansionacompletebutskeletalpropositionisfleshed
out.Thecharacteroftheinferenceinthesecasesisdistinctfromthatoftheinferencetothecontentofanindirectspeechact(such
asanimplicature)orthefigurativecontentofanonliteralutterance.Inthesecases,insteadofbuildingonwhatthespeakerhas
madeexplicit,thehearerinfersadistinctproposition.NowGriceisoftenchargedwithfocussingonimplicatures(andfigurative
utterances)totheexclusionoflessobliqueutterances,butclearlyhetooktheapplicationofhisapparatustothelattercasestobe
straightforwardthesejustdontinvolveostensiblefloutingofthemaxims.Heintendedthisapparatus,supportedbyhis
analysisofcommunicationorspeakermeaningingeneral,toapplyacrosstheboard.BachandHarnish(1979)adaptedGrices
apparatus,albeitwithvariousmodificationsandembellishments,tospeechactsgenerally,butweretainedthebasicideathat
communicationessentiallyinvolvesanaudiencedirectedreflexiveintention,sothat,whetherornottheutterancecanplausibly
betakenatfacevalue,thehearermustrelyonthesuppositionthatthespeakerintendswhatheiscommunicatingtobe

recognized(partlyonthebasisofthatintention).[23]Communicationisakindofcoordinationgame(Schelling,1960,pp.54ff.
and89ff.,andBach,1993).
Nowtodescribethegeneralcharacterofcommunicationisnottoexplainhowitsucceeds.SperberandWilson(1986,pp.
20and6970)rightlypointoutthatwepragmatistshavenotsuppliedmuchinthewayofpsychologicaldetailabouthowthe
processofunderstandingutterancesworks.Theycouldhavemadethesamepointabouttheprocessofproducingutterances.
Providingsuchdetailwouldrequireageneraltheoryofrealworldreasoningandatheoryofsalienceinparticular.Researchin
thepsychologyofreasoninghasidentifiedmanysortsoflimitationsinandconstraintsonhumanreasoningandAImodelsof
welldemarcatedtaskshavebeendeveloped,butageneralpredictiveandexplanatorytheoryisnotevenonthehorizon.And,
accordingtogametheoristsIhaveconsulted,althoughthenotionofsaliencehascontinued,eversinceitsintroductionby
Schelling,toberelieduponintheorizing,thereisstillnotheoryofsalience,nogeneralaccountofwhatitisinvirtueofwhich
certainitemsintheperceptual,cognitive,orconversationalenvironmentaresalient,muchlessmutuallysalient.Andyetour
abilitytocommunicate,toexpresspropositionalattitudes,exploitssuchinformation.
SperberandWilsonofferrelevancetheoryasaviablealternative.Theyeschewsuchallegedlyproblematicnotionsas

reflexiveintention,mutualbelief,[24]andmaximsofconversation.[25]Theysuggestthattheprincipleofrelevanceandthe
presumptionofoptimalrelevancecanpickuptheslack,whererelevance,intheirtechnicalsense,isamatterofmaximizing

contextualeffectsandminimizingprocessingeffort.[26]Interestingly,however,whentheytakeupspecificexamplesindetail,
theyrelyonthenotionofwhatthespeakermightreasonablybeexpectedtointend,theverynotionthatentersintoGricean
accountsofthehearersinference.ThatthisnotionisunavoidableisclearfromGricesinsightthatcommunicativeintentionsare
reflexiveinacertainspecialway,suchthatunderstandingutterancesreliesonthesuppositionthattheintentionsbehindthem
areintendedtoberecognized.Ontheotherhand,SperberandWilsonarerighttocomplainthatreconstructionsofhearers

http://userwww.sfsu.edu/kbach/impliciture.htm 19/27
9/20/2016 ConversationalImpliciture

inferences,howevermuchtheyringtrue,willinevitablyappearadhocintheabsenceofanexplanationofhowitisthatcertain
informationemergesasmutuallysalient(or,inSchellingsphrase,obviouslyobvious)sothatitmightbeexploitedbythe
hearer.Forthatveryreason,tosuggestthatprocessingtakesplaceonlyifitisworththeeffortandisamatterofsettlingonthe
firsthypothesisthatsatisfiestheprincipleofrelevance(SperberandWilson,1986,p.201)doesnotsaymuchabout
thishypothesisisarrivedat.Equally,tosaythatinferenceistoanunopposedplausibleexplanationofthespeakers
communicativeintention(BachandHarnish,1979,p.92)isnottosayhowthatisarrivedat.Theyspeakofoptimizingandwe
speakofdefaultreasoning,buttospeakofeitherisnottosaywithanydeterminatenesshowtheseprocesseswork.Norisitto
explainhoworwhycertainthoughts,suchashypothesesaboutspeakersintentions,cometomindwhentheydo,.Nooneis
preparedtodothat.
Allinall,IbelievethatthedisputebetweenGriceansandrelevancetheoristsisorthogonaltothequestionofhowto
demarcatetheregionbetweenstrictlyexplicitcontentandimplicature.ThemerefactthatGricehimselfdidnotrecognizethis
intermediateregionobviouslydoesnotshowthatonlyrelevancetheoryiscapableofdemarcatingit.Ontheotherhand,toshow
thatitfitsintoaGriceanaccountisnottoexplainhowparticularimplicituresareachievedorwhysomeareachievedandnot
others.
Letusconcludebytakingupsomeissuesrelatedspecificallytotheprocessesofexpansionandcompletion.Recanatihas
criticizedmyaccountofexpansion,atleastitwasfirstsketchedinBach,1987a(pp.72ff.),byappealingtotheideaoflocal
pragmaticprocesses.Herecognizesthatthenotionoflogicalstrengtheninghaslimitedexplanatoryscopeandacknowledges
thatthenotionofexpansionappearstohaveamuchwiderrangeofapplication(1993,pp.2612).Nevertheless,hehashis
doubtsaboutit,forthreereasons.Oneobjectionisthatsincewhatisenrichedisnotanaturallanguagesentencebutasemantic
representation[and]expansionisasyntacticoperation,theexpansiontheoristmusttreatsemanticrepresentationsassyntactic
entitiesasmentalsentences(1993,p.267).Recanatisuggeststhatweneedtoknowmuchmoreaboutthelanguageof
thoughtbeforewecanmakethisproposalprecise.InreplyIwouldsayfirstthatIdonotviewexpansionasasyntacticprocess.
ThatiswhyIdescribeitasamatternotoflexicalstrengtheningbutofconceptualstrengthening.Ofcoursethespeakercan
lexicallystrengthenhisutteranceandtherebymakefullyexplicitwhatheiscommunicating.Butthisisnotwhatthehearerdoes
whenheconceptuallystrengthensthespeakersutterance.Moreover,itisfarfromcertainthatthesemanticrepresentationsof
sentencesbelongtothelanguageofthought(assumingthereisone).Theycouldbepsychologicallyrealbut,assemantically
interpretedoutputsofthegrammar,functiononlyasinputstotheconceptualsystem.Indeed,thisiswhatonewouldexpect,
consideringthatsentencesareoftensemanticallyunderdeterminate.Sincethecontentsofthoughtscannotbeconceptually
incomplete,theconceptualrepresentationsthatcomprisethemcannotbesemanticallyunderdeterminateinthewaythat

sentencescanbe.[27]
Recanatissecondandmoredevelopedobjectionisthatmyviewimpliesthatexpansionisinvariablyapsychologically
realprocessofgoingfrompropositiontopropositionratherthanalocalprocessoperatingatthephrasallevel.Hesuggeststhat
atleastinsomecases,theminimalpropositionexpressiblebyanutteranceisatheoreticalartefact,inthesensethatitneed
notbecomputedandhasnopsychologicalreality(1993,263).Thisseemstobesointhecaseof(1),
(1)Thehamsandwichisgettingrestless.
wherethehamsandwichisbeingusedbyawaitertorefertoapatronwhoorderedahamsandwich.Recanatisupposesthatthe
processofmetonymicaltransfertakesplacewithouttheintrusionofathoughtoftheabsurdpropositionassociatedwiththe
literalmeaningof(1).Thatis,thehearerdoesnothavetocomputethatthespeakerdoesnotmeanthatacertainculinaryitemis
gettingrestlessinordertodeterminewhatthespeakerdoesmean.But,Iask,howcanthehearergofromtheconceptofham

http://userwww.sfsu.edu/kbach/impliciture.htm 20/27
9/20/2016 ConversationalImpliciture

sandwichtothatofhamsandwichordererwithoutfirstentertainingtheabsurdminimalproposition?Whattriggersthelocal
processand,forthatmatter,keepsitfrombeingtriggeredinacaselike(2),utteredinsimilarcircumstances?
(2)Thehamsandwichisgettingeaten.
Recanatisaccountpredictsthatthehearerwouldentertainthepropositionthatthehamsandwichordererisgettingeaten,since
thelocalprocessitpositswouldgettriggeredbeforeafullpropositionwerereached.Andyet(2)couldbeunderstoodperfectly
well.
Iamnotdenyinganditisconsistentwiththeexpansionistaccounttoconcedethatlocalprocessingcanoccurinsome
cases.Forexample,ifanutteranceof(1)wereprecededbyanutteranceoftheburritowalkedoutwithoutpaying,thennodoubt
thelocalprocessofmetonymictransferwouldtakeplaceautomatically.Evenso,if(2)ratherthan(1)wereutteredinthiscontext,
theutterancecouldstillbeunderstood,albeitwithsomebacktracking.Thepointhereisthattheexpansionistaccountdoesnot
poseasatheoryaboutthetemporalorderoftheprocessofunderstanding.Rather,itisatheoryaboutthecharacterofthe
informationavailabletothehearerintheprocess,whateverthepsychologicaldetails,ofidentifyingwhatthespeakeris
communicating,thatis,whatisimplicitinanutterancethatexplicitlyexpressesonlyaminimalproposition.Thelevelofthe
minimalproposition,ofwhatisstrictlyandliterallysaid,isneededtoaccountforthehearersaccesstothelinguisticcontentof
anutterance.
Thethirdobjection,invitingasimilarreply,isthatminimalpropositionsarenotpropositionsthatpeoplespontaneously
acknowledgeastheexplicitcontentsofutteranceswhenwhatisbeingcommunicatedisanenrichedproposition.Recanati
claimsthatonmyviewtheminimalproposition(whatissaid)neednotbeconsciouslyaccessible(1993,p.245).Butmy
viewsuggestsnothingsostrongasthat.Whatitdoessuggestisthatwhatisstrictlyandliterallysaidneednotbeconsciously
accessed.Thisleavesopenthepossibilitythatitisaccessedunconsciouslyoratleastthatitbeavailabletothehearer,evenifnot
actuallyaccessed.Theissueofwhatthehearerconsciouslyaccessesor,forthatmatter,ofwhatthespeakerconsciouslyintends,
concernsthepsychologyofprocessing,nottheinformationavailabletothatprocess.Theclaimthattheenrichedpropositionis
anexpansionofthepropositionstrictlyandliterallyexpresseddoesnotentailthatbeforethehearerarrivesattheenriched

propositionhefirstentertainstheminimalpropositioninitsfullglory.[28]
Moreover,metonymy,asinthehamsandwichexample,isaspecialcase.Aswesawintheprevioussection,metonymy
involvesthenonliteraluseofaparticularwordorphraseandthereforerequireslocalexpansion.Ingeneral,however,thereisno
particularphrasewhosenonliteralusetriggersexpansion,asillustratedbyexamplesfromsection3like(3)and(4).
(3)Yourenotgoingtodie.{fromthatcut}.
(4)Ihavenothingtowear.{tothewedding}.
MetonymyandotherspecialcasesRecanaticonsidersdonotsupporthisblanketcontentionthattheminimalpropositionisnot
actuallycomputedandplaysnoroleintheinterpretationprocessasitactuallyoccurs(1993,p.318).Evenifitisnotcomputed,
itstillcanplayarole.Notonlyisitincludedintheinformationavailabletothehearerinunderstandinganutterance,butit
providesthedefaultvalueforthehearersinference.Evenifthehearerdoesnotexplicitlyruleitoutbut,inseekinganother
interpretation,merelymakestheimplicitassumptionthatitisincorrect,itstillplaysarole.Implicitassumptionsareanessential
ingredientindefaultreasoningingeneral(Bach,1984)andintheprocessofunderstandingutterancesinparticular.Andsuch
assumptionsmustbeaccessible,sinceimplicituresbasedonexpnsionsareobviouslycancelable.
Recanatialsohashisdoubtsaboutthesemanticminimalismbuiltintomyaccountofcompletion.HereheechoesCarstons
contentionthattheformalphilosophicalrequirementofminimaltruthevaluabilityisanarbitrarypriniciplewithnoforceina

psychologicallyadequateaccountofcommunication[29](1987,p.714).LikeCarston,Recanatirejectsthelinguisticdirection
principle,whichrequiresthattheelementsofwhatissaidbeinclosesyntacticcorrelationwiththeconstituentsoftheutterance.

http://userwww.sfsu.edu/kbach/impliciture.htm 21/27
9/20/2016 ConversationalImpliciture

Heclaimsthataconsiderationofwhatisconsciouslyintuitedinanutteranceoverridesanytheorybasedappealtominimalist
semantics.Heappreciatesbutultimatelyrejectsthemixedminimalistprinciple,accordingtowhichApragmatically
determinedaspectofmeaningispartofwhatissaidiff(i)itscontextualdeterminationistriggeredbythegrammar,thatis,the
sentenceitselfsetsupaslottobecontextuallyfilled,and(ii)theslotinquestionneedstobefilledfortheutterancetobetruth
evaluableandexpressacompleteproposition(1993,p.241).NowIhavetriednottogetboggeddowninapointless
terminologicaldisputeabouttheproperuseofthephrasewhatissaid.Whennecessarytoavoidconfusion,Ihaveusedwhatis
explicitlyexpressedorwhatisliterallyandstrictlysaid,andhavemeantthemtoapplytowhatdoessatisfythisprinciple.But
giventhatwhatissaidisusedinthisrestrictiveway,theninanutteranceofasemanticallyunderdeterminatesentencewhatis
said(explicitlyexpressed,strictlyandliterallysaid)isnotacompleteproposition.Thisisbecausenoslotforcompletingthe
propositionissetupbythegrammar.Evenso,thehearermustprocessthesentencebeforehecangoontoidentifythecomplete
propositionbeingcommunicated.Soweneedtorecognizethisstageoftheprocess.
Recanatigoesontoconsidertheminimaltruthevaluabilityprinciple,whichrelaxesthesyntacticrequirementandsays
simplythatApragmaticallydeterminedaspectofmeaningispartofwhatissaidiffitscontextualdeterminationisnecessaryfor
theutterancetobetruthevaluableandexpressacompleteproposition.Itprohibitsexpansionsfromcountingaspartofwhatis
said,therebydisallowingwhatRecanaticountsaswhatissaidandwhatCarston,followingSperberandWilson,countsasthe
explicitcontentoftheutterance.Presumablythedisputehereisnotoverterminologybutoverpsychologicallyrelevant
distinctionsneededforanaccountofunderstandingutterances.ThesecriticsofGricerightlyinsistthathiswayofdrawingthe
distinctionbetweenwhatissaid(orwhatisexplicit)andwhatisimplicatedisnotexhaustive.Butthesolutionisnottodrawthe
distinctiondifferentlywideningthescopeofwhatissaidandcallingthatexplicitcontentbuttokeepitexhaustive.Instead,
thewaytogoistodistinguishnotonlytheimpliedfromtheexplicitbuttheimplicitfromtheimplied.

DepartmentofPhilosophy
SanFranciscoStateUniversity
SanFrancisco
California94132

Notes

[1]Peopleuselargerchunksoflanguagethansentences,ofcourse,butfromthepointofviewofgrammar,thesentenceisthe

basicunit.Ourdiscussionwillnottakeuppragmaticphenomenathatinvolvemultisentenceutterances.
[2]ForthesakeofdiscussionIamassumingthesetobedistinctsensesoftheword.Butseesection7,whichtakesupRuhls

(1989)contentionthatinlexicographyandinlexicalsemantics,sensesofwordsaremultipliedfarbeyondnecessity.
[3]ItookitupagaininBach,1987a(pp.7477),whereIcalleditsemanticindeterminacy,andSperberandWilsonusetheterm

underdeterminetodescribetherelationbetweenthelogicalformofanutteranceandthepropositionexpressed(1986,p.
180)thatgenerallyobtainswhetherornotthelogicalformalreadydeterminesacompleteproposition.Whenitdoesnot,they
saythatthesentenceissemanticallyincomplete.
[4]ThisiswhatPerry(1986)callsan'unarticulatedconstituent'.Aswewillsee,unarticulatedconstituentsdonotcorrespondto

syntacticconstituentsliketheemptycategoriesofGBtheory.
[5]Noticethatthesemanticallyunderdeterminate(7)and(8)wouldexpresscompletepropositionswithoutevenormerely.

http://userwww.sfsu.edu/kbach/impliciture.htm 22/27
9/20/2016 ConversationalImpliciture

[6]Itmightbesuggestedthatfinishandcompleteeitherlose(atSstructure)andthenregain(atLF)theircomplementsduringthe

courseofsyntacticderivationorthattheircomplementsexistallalongbutonlyasemptycategories.Unfortunately,thefirst
optionviolatestheProjectionPrinciple,andthesecondignorestherequirementthatemptycategoriesbesyntacticallylicensed
(Chomsky,1986,pp.93101).
[7]So,forexample,adirectanswertoaWHquestionisaphrasethatwouldfitsyntacticallyintotheoriginalsiteoftheWH

wordinthequestion.Thepracticeofusingjustaphraseissodeeplyentrenchedthattheonlygoodreasonforusingawhole
sentenceistomakesurethatonehasunderstoodthequestioncorrectly.
[8]Itmightevenbeobjectedthatsemanticunderdeterminationisjustakindofambiguity,butasaclaimaboutsentencegrammar

thisisabsurd.Thereisnopositivereasontosupposethatandaverystrongreasonnotto,namelythatitwouldfollowthatany
sentencewithaconceptualgapisambiguousinasmanywaysasthereareconstituentsthatcouldfillit.Thesituationisnotas
badwithscopeunderdetermination(seeBach,1982)and,indeed,itisanopentheoreticalquestion,asnotedabovein
connectionwithsentenceslike(24)(27),whichsortsofscoperelationaremarkedatsomesyntacticlevel.Theonesthatareso
markedinduceambiguities.Settlingsuchaquestionmayrequirenotonlysyntacticargumentationbutalsopsycholinguistic
researchthatmightuncoversubtledifferencesintheprocessingofsentencesofthetwosorts.
[9]Thatis,itisnotamatterofsentencegrammar.Ishouldnoteanunfortunatedoubleuseoftheterm'semantic'.Itiscommonly

usednotonlyformeaningrelatedfeaturesofsentencegrammarbutalsoforfeaturespertainingtotruthandreference,even
whenthesegobeyondsentencegrammar.Thelattergobeyondgrammarbecausethemeaningofanindexical(withthe
exceptionofIandafewtemporaladverbslikeyesterday,today,andtomorrow)doesnotdetermineitsreference,indeednot
evenasafunctionofcontextcontextisnotawelldefinedsetofparameters,sothatindexical(ordemonstrative)referenceis
generallyamatterofthespeakersintention(seeBach,1987a,pp.176186,andBach,1992).Itiscontextrelativebutnot,as
sometimesthought,'fixedbythecontext'.However,thesemanticsofindexicalsmakeessentialreferencetotheirutterance
(henceReichenbachs(1947,p.284)descriptionofthemas'tokenreflexives').Thinkofanindexicalasintroducingavariable
intothesemanticrepresentationofthesentence.Eachonehasitsownsemanticallyspecifiedreferentialconstraintonhow,ina
givencontextofutterance,itcanbeusedtorefer,i.e.,onhowthevariableitintroducesacquiresavalueinagivencontextof
utterance(Bach,1987a,pp.186192).So,forexample,Iisusedtoreferthespeaker,yesterdaytothedaybeforetheutterance,
shetoacontextuallyidentifiablefemale,andtheretoacontextuallyidentifiableplace(otherthanthatoftheutterance)..
[10]ThatiswhyHarnishandIusedthephraselexicalomission(BachandHarnish,1979,p.231)andwhy,indiscussing

utterancesneedingexpansion(Bach,1987a,pp.7785),Ispokeofsuppressedmaterial.Buttherewasnosuggestionthatsuch
elementsarereallythere,inthesenseofbeingphonologicallyunrealizedconstituents..Rather,theycouldhavebeenusedto
makefullyexplicitwhatthespeakermeant.
[11]SperberandWilsoncallitenrichment(1986,p.181)andviewtheprocessasgoingfromthelogicalformofanutteranceto

thepropositionalformexpressed.However,theydonotspelloutwhatsortsofproceduresenrichmentcaninvolve.
[12]Idonotmeantoexcludethecaseofphonologicallynullconstituents,suchastheemptycategoriesofGBTheory.Theycan

contributetowhatissaideventhoughtheyarenotheard(orseen),theyarethere.
[13]Itisworthnotingthatpragmaticprocessesareinvolvedevenincaseswhereonemeansexactlywhatonesays.Fromthe

hearerspointofview,nottoreadanythingintoanutteranceandtotakethe(literal)meaningasdeterminingallthespeaker
meansisasmuchamatterofcontextualintepretationasexpandingtheutterance.

http://userwww.sfsu.edu/kbach/impliciture.htm 23/27
9/20/2016 ConversationalImpliciture

[14]BachandHarnish,1979,pp.165172,arguethatGricesdistinction,whengeneralized,istantamounttothedistinction

betweendirectandindirectillocutionaryacts.Thepresentdiscussionofimpliciture,likeGricesofimplicature,willbelimited
toindicativecases.Takingupnonindicativecaseswouldintroducesomeminorcomplicationsconcerninghowtospecify
whatissaidandwhatismeant..
[15]Theuseofandtoconveyandthenorandasaresultisadifferentmatterbecause,asCarston(1988)argues,theseforcesdo

notcorrespondtospecialmeaningsofthewordand.ShepointsoutthattheywouldbeclassifiedbyGriceasgeneralized
conversationalimplicaturesbutarguesagainstthat,classifyingthemasenteringinto'explicitcontent'inthebroadsenseof
relevancetheorynotedearlier.Iwouldclassifythemasimplicitures,asenteringintoexpansions.
[16]Foraexaminationofthenontruthconditionalroleofdiscourseconnectivesfromtheperspectiveofrelevancetheorysee

Blakemore,1987.
[17]BydistinguishingtheselevelsIamnotimplyingthattheoperationsofcompletionandexpansionareonsentences.Theyare

operationsonwhatutterancesofsentencesexpress.
[18]Thedifferencebetweenambiguityandhomonymyonewordwithtwo(ormore)meaningsasopposedtotwo(ormore)

distinctwordsisnotasstraightforwardasitsounds,fortherearenoclearcriteriaforsamenessordistinctnessoflike
soundingwords.Forexample,cantheverbandcanthenounseemtobedistinctwords,butwhataboutcooktheverbandcook
thenoun?Istheadjectivelight(nevermindtheverborthenoun),asinthephraselightsuit,oneambiguouswordortwo
distinctwords?
[19]Whatfollowsappliesequallytocertainothersortsofphrases,suchastheseadjectivenounpairs:healthychildvs.healthy

food,fastcarvs.fasttrack,generousdonorvs.generousgift,andconsciousbeingvs.consciousstate.Inthefirstofeachof
pair,theadjectivedescribeswhatthenoundenotes,butinthesecondcaseitdoesnotatleastnotasthesephrasesarelikely
tobeusedortaken.Andhowtheyaretakendependsonone'sextralinguisticknowledge.Thereisnothinginthesemanticsof
thesephrasestopreventthemfrombeingtaken(absurdly),withtheadjectiveusedtodescribewhatthenoundenotes.Noun
nounpairscanalsobeunderstoodasexpressingvaioussortsofrelation:comparechickenpluckerwithchickenliverorsteak
knifewitharmyknife,forexample.
[20]Ruhlcallsthispragmaticmetonymyanddistinguishesitfromsemanticmetonymy,aswithwordslikeorangeand

tongue,whichhedeemspolysemicbecausetheirtwouses,thoughrelated,areentrenchedandindependent(1989,p.97).
[21]Thelabelforthisdistinctionisabitmisleading,inthattomentionanexpressionis,afterall,touseit,namely,torefertothat

veryexpression.Thisisnotitsnormaluse,ofcourse,andisnotpredictablefromitsmeaning.Generallythemeaningofthe
mentionedexpressionisirrelevant,butnotalways,aswhentheexpressionissaidtobeonomatopoeic.
[22]ItwouldtakeconsiderablespacetodojusticetoHornsposition.Forexample,hearguesthattherearethreesyntactic

diagnosticsformetalinguisticnegation(1989,pp.392413).Iwouldarguethatineachcasetheoccurrenceofanitemthatin
hisviewissyntacticallyincompatiblewithmetalinguisticnegationinfactpragmaticallyblockstheexpansionrequiredforthe
metalinguisticunderstanding.
[23]SperberandWilsonobjecttothecriterionofunderstandingintermsofidentifyingwhatisintendedbythespeaker.They

complainthatthiscannotbewhatthehearerusesbecauseifhealreadyknewthespeakersintention,hewouldhavenotaskof
identificationleft(p.183).Butthiscriterionisnotvacuous.Griceanaccountsdonotsaythatthehearerisamindreader.
Rather,thehearerconsiderswhat,giventheutteranceandthecircumstancesinwhichitwasmade,thespeakercould
reasonablyhaveintendedtobecommunicating.Thehearerhasaspecificbasistheutteranceandmutuallyavailable

http://userwww.sfsu.edu/kbach/impliciture.htm 24/27
9/20/2016 ConversationalImpliciture

informationtoinferthespeakersintention,takingintoaccountthatheisintendedsotoinferit.Thisdoesnotmeanthathe
hastoknowwhatthatintentionisbeforehecanidentifyit.
[24]SperberandWilsonobjecttothenotionsofreflexiveintentionandmutualbeliefonthegroundsthatthesenotionseach

involveaviciousinfiniteregressthatrendersthemifnotincoherentatleastpsychologicallyunrealistic.However,asIhave
arguedpreviously(seeBach,1987b),reflexiveintentionsarenotinfinitelyiterative.Asimilarpointappliestomutualbelief.
Thesetofmutualbeliefsthatarerelevanttoaspeaker'scommunicativeintentionandthehearer'srecognitionofitconsistsof
everythingthateachbelievestheotherbelieves.Membershipinthissetisdefinedrecursivelyamongthatwhicheach
believestheotherbelievesisthatwhichtheotherbelieveshebelievesbutthereisnoinfiniteregress.
[25]SeeHarnish,1976/1991,pp.330340,fordiscussionofGricesmaxims,theirweaknesses,theirconflicts,andgeneralgapsin

Gricestheoryofconversation.HarnishproposesaMaximofQuantityQuality:Makethestrongestrelevantclaimjustifiable
byyourevidence(p.340seealsonote46,pp.360361).
[26]Thisisafamiliarideaincognitivepsychology.ForexampleRosch,observingthatwordslikemonkeyorcararelearned

earlierthanwordslikeanimalandchimporvehicleandChevy,arguedthatthereisonebasiclevelofabstractionatwhichthe
organismcanobtainthemostinformationwiththeleastcognitiveeffort(1977,p.213).
[27]Iputthepointthiswaytoallowforindexicalityasopposedtounderdetermination.Forinmyviewsingularorsocalledde

rethoughts(thoughtsaboutparticularobjects)containelementsthatfunctionasmentalindexicals,sothattheirtruth
conditionsarecontextrelative(Bach,1987a,pp.1126).
[28]Comparethesituationherewithlexicaldisambiguation.Itiswellknown,asshownbysocalledcrossmodalpriming

experiments(seeForster,1990andGarrett,1990fordiscussionoftheseandotherexperiments),thatirrelevantsensesof
ambiguoustermsarebrieflyaccessedunconsciously.Uponhearinganambiguouswordinasentencewhereonesenseis
clearlyrelevantandtheotherisnot,subjectsaremoreapttorespondvisuallytoitemstowhichtheirrelevantsenseapplies
thantoitemsunrelatedtoeithersense.Iftheyhearasentencelike(i),
(i)Thesurgeonrealizedthattheorganneededrepair.
theywill,immediatelyafterhearingthewordorgan,brieflyaccesswithoutconsciouslyregisteringthemusicalinstrument
reading,whereaswith(ii),
(ii)Thesoloistrealizedthattheorganneededrepair.
theywillbrieflyaccessbutnotconsciouslyregisterthebiologicalpartreading.If,however,(i)iscontinuedwiththewords
becauseitsoundedawful,themusicalinstrumentreadingwillcomeintoplay.Sothealternativereadingoftheambiguous
wordisaccessedimmediatelybutnotconsciously,andsubsequentlyonlyifnecessary.Yetthealternativereadingoftheentire
sentence(i)isavailableallalong.
[29]Leavingasidetheirdoubtsabouttheroleofminimalpropositions'inthementallifeofthehearer'(Carston,1988,p.165),

whataboutthementallifeofthespeaker?Thespeaker'ssideoftheprocessofcommunicationmustbeconsideredaswell,lest
welapseintothinkingoflanguageasmerelyaninputmechanism,avehicleforinterpretingutterancesaspropositional
representations(Kempson,1988,16myitalics).Weshouldnotneglectitsroleinoutput,intheexpressionofpropositional
attitudes.Speakersdonotformattitudestobeexpressedbytheutterancestheyaredisposedtoproducerather,theyare
disposedtoproduceutterancesthatexpresstheattitudestheywishtoexpress.Nowtherearemanywaystoexpressagiven
propositionalattitude,butinordinaryspeechpeopledonotconsciouslyconsiderthealternatives.However,iftheirchoiceof
utterancewereinsensitivetothelinguisticcontentoftheutterance,then,onemightwonder,howcoulditmakeanydifference

http://userwww.sfsu.edu/kbach/impliciture.htm 25/27
9/20/2016 ConversationalImpliciture

whetheroneuttersasentencestrictlyandliterallyintepretableasexpressingaminimalpropositionandoruttersonethat
expressestheintendedexpansionofthefirst?






References

Austin,J.L.1962:HowToDoThingsWithWords.Oxford:OxfordUniversityPress.
Atlas,J.D.1977:Negation,Ambiguity,andPresupposition.LinguisticsandPhilosophy,1,321336.
Atlas,J.D.1989:PhilosophyWithoutAmbiguity.Oxford:OxfordUniversityPress.
Bach,K.1982:SemanticNonspecificityandMixedQuantifiers.LinguisticsandPhilosophy,4,593605.
Bach,K.1984:DefaultReasoning:JumpingtoConclusionsandKnowingWhenToThinkTwice.PacificPhilosophical
Quarterly,65,3758.
Bach,K.1987a:ThoughtandReference.Oxford:OxfordUniversityPress.
Bach,K.1987b:OnCommunicativeIntentions:AReplytoRecanati.Mind&Language,2,141154.
Bach,K.1992:IntentionsandDemonstrations.Analysis,52,140146.
Bach,K.1993:Meaning,SpeechActs,andCommunication.InR.M.Harnish(ed.),BasicTopicsinthePhilosophyofLanguage.
Herts,Eng.:HarvesterWheatsheaf.
Bach,K.andR.M.Harnish1979:LinguisticCommunicationandSpeechActs.Cambridge,Mass.:MITPress.
Carston,R.1987:BeingExplicit.BehavioralandBrainSciences,10,713714.
Carston,Robyn1988:Implicature,Explicature,andTruththeoreticSemantics.InR.M.Kempson(ed.),MentalRepresentations:
TheinterfaceBetweenLanguageandReality.Cambridge,Eng.:CambridgeUniversityPress.
Chomsky,N.1986:KnowledgeofLanguage.NewYork:Praeger.
Cruse,D.A.Monosemyvs.Polysemy(reviewofRuhl,1989).Linguistics,30,577599.
Forster,K.I.1990:LexicalProcessing.InD.OshersonandH.Lasnik(eds.),Language.Cambridge,Mass.:MITPress.
Garrett,M.F.1990:SentenceProcessing.InD.OshersonandH.Lasnik(eds.),Language.Cambridge,Mass.:MITPress.
Grice,H.P.1961:TheCausalTheoryofPerception.ProceedingsoftheAristotelianSociety,Supp.Vol.35,121152(abridgedas
chapter15ofGrice,1989).
Grice,H.P.1967a:LogicandConversation.WilliamJamesLectures.Photocopy(chapter2ofGrice,1989).
Grice,H.P.1967b:FurtherNotesonLogicandConversation.WilliamJamesLectures.Photocopy(chapter3ofGrice,1989).
Grice,H.P.1968:UtterersMeaning,Sentencemeaning,andWordmeaning.FoundationsofLanguage,4,225242(chapter6of
Grice,1989).
Grice,H.P.1969:UtterersMeaningandIntentions.PhilosophicalReview78,147177(chapter5ofGrice,1989).
Grice,H.P.1989:StudiesintheWaysofWords.Cambridge,Mass.:HarvardUniversityPress.
Harnish,R.M.1976:LogicalFormandImplicature.InT.Beveretal(eds.),AnIntegratedTheoryofLinguisticAbility.New
York:Crowell(reprintedinS.Davis(ed.),Pragmatics:AReader.NewYorkandOxford:OxfordUniversityPress,1991).
Horn,L.R.1989:ANaturalHistoryofNegation.ChicagoandLondon:UniversityofChicagoPress.

http://userwww.sfsu.edu/kbach/impliciture.htm 26/27
9/20/2016 ConversationalImpliciture

Karttunen,L.andS.Peters1979:ConventionalImplicature.InCKOhandD.A.Dinneen(eds.),SyntaxandSemanticsVolume
11,Presupposition.NewYork:AcademicPress.
Kempson,R.1988.Introduction:TheRelationbetweenLanguage,Mind,andReality.InR.M.Kempson(ed.),Mental
Representations:TheinterfaceBetweenLanguageandReality.Cambridge,Eng.:CambridgeUniversityPress.
Neale,S.1993:WhatisLogicalForm?Synthese00:000000.
Perry,J.1986:ThoughtWithoutRepresentation.ProceedingsoftheAristotelianSociety,Supp.Vol.60,137151.
Recanati,F.1989:ThePragmaticsofWhatisSaid.MindandLanguage,4,294328.
Recanati,F.1993:DirectReference,Meaning,andThought.Oxford:Blackwell.
Reichenbach,H.1947:ElementsofSymbolicLogic.London:Macmillan.
Rosch,E.1977:ClassificationofRealwordObjectsOriginsandRepresentationsinCognition.InP.N.JohnsonLairdandP.C.
Wason(eds.),Thinking:ReadingsinCognitiveScience.Cambridge,Eng.:CambridgeUniversityPress.
Ruhl,C.1989:OnMonosemy:AStudyinLinguisticSemantics.Albany:StateUniversityofNewYorkPress.
Schelling,T.1960:TheStrategyofConflict.Cambridge,Mass.:HarvardUniversityPress.
Sperber,DanandDeirdreWilson1986:Relevance.Cambridge,Mass.:HarvardUniversityPress.

http://userwww.sfsu.edu/kbach/impliciture.htm 27/27

Das könnte Ihnen auch gefallen