【劳特里奇当代哲学导论系列】语言哲学

【劳特里奇当代哲学导论系列】语言哲学

\nPhilosophyofLanguagePhilosophyofLanguage:aContemporaryIntroductionintroducesthestudenttothemainissuesandtheoriesintwentiethandtwenty-first-centuryphi-losophyoflanguage,focusingspecificallyonlinguisticphenomena.Topicsarestructuredinfourpartsinthebook.PartI,ReferenceandReferring,includestopicssuchasRussell’sTheoryofDescriptions,Donnellan’sdistinction,problemsofanaphora,thedescriptiontheoryofpropernames,Searle’sclustertheory,andthecausal–historicaltheory.PartII,TheoriesofMeaning,surveysthecompetingtheoriesoflinguisticmean-ingandcomparestheirvariousadvantagesandliabilities.PartIII,PragmaticsandSpeechActs,introducesthebasicconceptsoflinguisticpragmatics,includesadetaileddiscussionoftheproblemofindirectforceandsurveysapproachestometaphor.PartIV,newtothisedition,examinesthefourtheoriesofmetaphor.FeaturesofPhilosophyofLanguageinclude:•newchaptersonFregeandpuzzles,inferentialism,illocutionarytheoriesofmeaning,andrelevancetheory•chapteroverviewsandsummaries•clearsupportiveexamples•studyquestions•annotatedfurtherreading•glossaryPraisefortheFirstEdition:“Thisexceptionaltextfulfilstwoessentialcriteriaofagoodintroduc-torytextbookinthephilosophyoflanguage:itcoversabroadrangeoftopicswell,allofwhicharethebasisofcurrentactiveresearch,anddoessoinanaccuratemanneraccessibletoundergraduatestudents.”MikeHarnish,UniversityofArizona“...anexcellenttextbookforteaching.Theexamplesthroughoutaredelightfulandstudentswilllovethem.”EdwinMares,VictoriaUniversityofWellingtonWilliamG.LycanisWilliamRandKenanProfessorofPhilosophyattheUniversityofNorthCarolina.Heistheauthorofover150articlesaswellassevenbooks.\nRoutledgeContemporaryIntroductionstoPhilosophySeriesEditor:PaulK.MoserLoyolaUniversityofChicagoThisinnovative,well-structuredseriesisforstudentswhohavealreadydoneanintroductorycourseinphilosophy.Eachbookintroducesacoregeneralsubjectincontemporaryphilosophyandoffersstudentsanaccessiblebutsubstantialtransitionfromintroductorytohigher-levelcollegeworkinthatsubject.Theseriesisaccessibletonon-specialistsandeachbookclearlymotivatesandexpoundstheproblemsandpositionsintroduced.Anorientatingchapterbrieflyintroducesitstopicandremindsreadersofanycrucialmaterialtheyneedtohaveretainedfromatypicalintroductorycourse.Considerableattentionisgiventoexplainingthecentralphilosophicalproblemsofasubjectandthemaincompetingsolutionsandargumentsforthosesolutions.Theprimaryaimistoeducatestudentsinthemainproblems,positionsandargumentsofcontemporaryphilosophyratherthantoconvincestudentsofasingleposition.ClassicalPhilosophyPhilosophyofMindChristopherShieldsSecondEditionJohnHeilEpistemologySecondEditionPhilosophyofReligionRobertAudiKeithE.YandellEthicsPhilosophyofScienceHarryGenslerAlexRosenbergMetaphysicsSocialandPoliticalSecondEditionPhilosophyMichaelJ.LouxJohnChristmanPhilosophyofArtPhilosophyofPsychologyNoëlCarrollJoséBermudezPhilosophyofLanguageClassicalModernSecondEditionPhilosophyWilliamG.LycanJeffreyTlumakPhilosophyofMathematicsPhilosophyofBiologySecondEditionAlexRosenbergandDanielJamesR.BrownW.McShea\nPhilosophyofLanguageaContemporaryIntroductionSecondEditionWilliamG.Lycan\nFirstpublished2000byRoutledgeThiseditionfirstpublished2008byRoutledge270MadisonAve,NewYork,NY10016SimultaneouslypublishedintheUKbyRoutledge2ParkSquare,MiltonPark,Abingdon,OxonOX144RNRoutledgeisanimprintoftheTaylor&FrancisGroup,aninformabusinessThiseditionpublishedintheTaylor&Francise-Library,2008.“TopurchaseyourowncopyofthisoranyofTaylor&FrancisorRoutledge’scollectionofthousandsofeBookspleasegotowww.eBookstore.tandf.co.uk.”Firstedition©2000Taylor&FrancisGroupSecondedition©2008Taylor&FrancisGroupAllrightsreserved.Nopartofthisbookmaybereprintedorrepro-ducedorutilisedinanyformorbyanyelectronic,mechanical,orothermeans,nowknownorhereafterinvented,includingphotocopyingandrecording,orinanyinformationstorageorretrievalsystem,withoutpermissioninwritingfromthepublishers.TrademarkNotice:Productorcorporatenamesmaybetrade-marksorregisteredtrademarks,andareusedonlyforidentificationandexplanationwithoutintenttoinfringe.LibraryofCongressCataloginginPublicationDataLycan,WilliamG.Philosophyoflanguage:acontemporaryintroduction/WilliamG.Lycan.–2nded.p.cm.–(Routledgecontemporaryintroductionstophilosophy)Includesbibliographicalreferencesandindex.1.Languageandlanguages–Philosophy.I.Title.P106.L8862008401–dc222007038223ISBN0-203-93000-2Mastere-bookISBNISBN10:0-415-95751-6(hbk)ISBN10:0-415-95752-4(pbk)ISBN10:0-203-93000-2(ebk)ISBN13:978-0-415-95751-9(hbk)ISBN13:978-0-415-95752-6(pbk)ISBN13:978-0-203-93000-7(ebk)\nContentsPrefaceixAcknowledgementsxiAcknowledgementsfortheSecondEditionxii1Introduction:meaningandreference1Overview1Meaningandunderstanding1TheReferentialTheory3Summary6Questions6Furtherreading6PartI:Referenceandreferring2Definitedescriptions9Overview9Singularterms10Russell’sTheoryofDescriptions12ObjectionstoRussell’stheory19Summary29Questions30Furtherreading303Propernames:theDescriptionTheory31Overview31Fregeandthepuzzles31Openingobjections36Searle’sClusterTheory38Kripke’scritique39Summary43Questions43Furtherreading44\nviContents4Propernames:DirectReferenceandtheCausal–HistoricalTheory45Overview45Possibleworlds46Rigidityandpropernames47DirectReference49TheCausal–HistoricalTheory53ProblemsfortheCausal–HistoricalTheory55Natural-kindtermsand“TwinEarth”58Summary60Questions61Furtherreading61PartII:Theoriesofmeaning635Traditionaltheoriesofmeaning65Overview65ThePropositionTheory68Summary74Questions74Furtherreading746“Use”theories76Overview76“Use”inaroughlyWittgensteiniansense77Objectionsandsomereplies79Inferentialism83Summary84Questions85Furtherreading857Psychologicaltheories:Grice’sprogram86Overview86Grice’sbasicidea86Speaker-meaning88Sentencemeaning92Summary96Questions96Furtherreading978Verificationism98Overview98Thetheoryanditsmotivation98Someobjections100\nContentsviiThebigone105Summary108Questions108Furtherreading1089Truth-ConditionTheories:Davidson’sprogram109Overview109Truthconditions109Truth-definingnaturallanguages114ObjectionstotheDavidsonianversion117Summary123Questions124Furtherreading12510Truth-ConditionTheories:possibleworldsandintensionalsemantics126Overview126Truthconditionsreconceived126AdvantagesoverDavidson’sview129Remainingobjections131Summary133Questions133Furtherreading133PartIII:Pragmaticsandspeechacts13511Semanticpragmatics137Overview137Semanticpragmaticsvs.pragmaticpragmatics138Theproblemofdeixis139Theworkofsemanticpragmatics141Summary143Questions143Furtherreading14312Speechactsandillocutionaryforce144Overview144Performatives144Infelicitiesandconstitutiverules149Cohen’sproblem151Illocutionarytheoriesofmeaning153Summary154Questions154Furtherreading155\nviiiContents13Implicativerelations156Overview156Conveyedmeaningsandinvitedinferences157Conversationalimplicature158Presuppositionandconventionalimplicature163RelevanceTheory166Indirectforce168Summary170Questions171Furtherreading171PartIV:Thedarkside17314Metaphor175Overview175Aphilosophicalbias175Theissues,andtwosimpletheories176Davidson’scausaltheory177TheNaiveSimileTheory179TheFigurativeSimileTheory180ThePragmaticTheory183Metaphorasanalogical187Summary189Questions190Furtherreading190Notes191Glossary201Bibliography205Index217\nPrefaceAsitstitleslylysuggests,thisbookisanintroductiontothemainissuesincontemporaryphilosophyoflanguage.Philosophyoflanguagehasbeenmuchinvoguesinceearlyinthetwentiethcentury,butonlysincethe1960shavetheissuesbeguntoappearinhighresolution.Onecrucialdevelopmentinthepastfortyyearsistheattentionofphiloso-phersoflanguagetoformalgrammarorsyntaxasarticulatedbytheoreticallinguists.Ipersonallybelievethatsuchattentionisvitaltosuccessinphi-losophizingaboutlanguage,andinmyownworkIpayasmuchofitasIamable.Withregret,however,Ihavenotmadethatathemeofthisbook.Underseverespacelimitations,Icouldnotexpendasmanypagesaswouldbeneededtoexplainthebasicsofformalsyntax,withouthavingtoomitpresentationofsomephilosophicalissuesIconsideressentialtocompetenceinthefield.Sincearound1980,somephilosophersoflanguagehavetakenaturntowardthephilosophyofmind,andsomehaveengagedinmetaphysicalexplorationoftherelationorlackthereofbetweenlanguageandreality.Theseadversionshavecapturedmanyphilosophers’interest,andsomefinetextbookshavefocusedononeorboth(forexample,Blackburn1984;DevittandSterelny1987).ButIhavechosenotherwise.Whateverthemeritsofthosesortsofwork,Ihavenotfoundthateitherhelpsussufficientlytounderstandspe-cificallylinguisticmechanismsorthecoreissuesofphilosophyoflanguageitself.Thisbookwillconcentrateonthosemechanismsandissues.(Readerswhowishtopressonintometaphysicsorphilosophyofmindshouldconsult,respectively,MichaelJ.Loux’sMetaphysicsandJohnHeil’sPhilosophyofMind,bothintheRoutledgeContemporaryIntroductionsseries.)Manyofmychaptersandsectionswilltaketheformofpresentingdatapertinenttoalinguisticphenomenon,expoundingsomeone’stheoryofthatphenomenon,andthenlistingandassessingobjectionstothattheory.Iemphasizehere,becauseIwillnotalwayshavethespacetodosointhetext,thatineachcasewhatIwillsummarizeforthereaderwillbeonlytheopeningmovesmadebythevarioustheoristsandtheiropponentsandobjectors.Inparticular,Idoubtthatanyoftheobjectionstoanyofthetheoriesisfatal;championsoftheoriesareremarkablygoodatavoidingorrefutingobjec-tions.Therealtheorizingbeginswherethisbookleavesoff.\nxPrefaceIhaveusedsomenotationofformallogic,specificallythepredicatecalcu-lus,forthosewhoarefamiliarwithitandwillfindpointsmadeclearerbyit.ButineachcaseIhavealsoexplainedthemeaninginEnglish.Manyofthewritingstobediscussedinthisbookcanbefoundinthefollowinganthologies:T.Olshewsky(ed.),ProblemsinthePhilosophyofLanguage(Austin,TX:Holt,RinehartandWinston,1969);J.F.RosenbergandC.Travis(eds.)ReadingsinthePhilosophyofLanguage(EnglewoodCliffs,NJ:Prentice-Hall,1971);D.DavidsonandG.Harman(eds.),TheLogicofGrammar(Encino,CA:Dickenson,1975);R.M.Harnish(ed.)BasicTopicsinthePhilosophyofLanguage(EnglewoodCliffs,NJ:Prentice-Hall,1994);A.Martinich(ed.)ThePhilosophyofLanguage,5thedn.(Oxford:OxfordUniversityPress,2006),andearliereditionsaswell;P.Ludlow(ed.)ReadingsinthePhilosophyofLanguage(Cambridge,MA:BradfordBooks/MITPress,1997);A.Nye(ed.),PhilosophyofLanguage:TheBigQuestions(Oxford:BasilBlackwell,1998);M.Baghramian(ed.),ModernPhilosophyofLanguage(NewYork:CounterpointPress,1999).\nAcknowledgementsIthankmyeditor,MoiraTaylor,forherbracingencouragementand(espe-cially)forherpatience.Thelatterwasseverelytried.MikeHarnish,GregMcCulloch,andEdMareseachverykindlyreadanearlydraftandsuppliedmewithmanythoughtfulcommentsandsuggestions.Ibelievethebookismuchimprovedasaresult,andIammostgrateful.PeterAlwardandLauraMorganproducedmuchoftheearlydraftbytranscribingmanyhoursoflecturesfromverybadaudiorecordings.IthankthemwarmlyandIhopethateachofthemwillsoonmakeafullrecovery.SeanMcKeever’smonthsofeditorialhelpandadvicehavebeeninvaluable.(Hesufferedthroughsometranscribingaswell.)ThanksespeciallytoSeanforsuggestingsomeneededcuts,andfororganizingthebibliography.ThelastfewchaptersofthisbookwerecompletedduringmytenureasaFellowoftheNationalHumanitiesCenter,in1998–99.IthanktheCenteranditswonderfulstafffortheirgeneroussupport.ForadditionalfundingIamindebtedtotheNationalEndowmentfortheHumanities(#RA–20169–95).\nAcknowledgementsfortheSecondEditionThankstoeditorKateAhlforherpatienthelp,andtoMegWallaceforagooddealofresearchaswellaseditingandindexing.Thanksalsotomanyreadersofthefirstedition,whohaveofferedcommentsandsuggestionsfromaroundtheworld.IamespeciallygratefultoMikeHarnish(again!),PatrickGreenough,andMarkPhelan,whogavemeverydetailedcommentsthathaveledtomanyimprovements,includingthecorrectionofsomeerrors.ManyofthecommentsIhavereceivedurgedmetoaddachapterorsectiononthisorthatadditionaltopic.Thoseweregoodideas,butspacelimitationsforbademyadoptingmorethanahandfulofthem;myapologies.\n1Introduction:meaningandreferenceOverviewThatcertainkindsofmarksandnoiseshavemeanings,andthatwehumanbeingsgraspthosemeaningswithouteventhinkingaboutit,areverystrik-ingfacts.Aphilosophicaltheoryofmeaningshouldexplainwhatitisforastringofmarksornoisestobemeaningfuland,moreparticularly,whatitisinvirtueofwhichthestringhasthedistinctivemeaningitdoes.Thetheoryshouldalsoexplainhowitispossibleforhumanbeingstoproduceandtounderstandmeaningfulutterancesandtodothatsoeffortlessly.Awidespreadideaaboutmeaningisthatwordsandmorecomplexlin-guisticexpressionshavetheirmeaningsbystandingforthingsintheworld.Thoughcommonsensicalandatfirstattractive,thisReferentialTheoryofmeaningisfairlyeasilyshowntobeinadequate.Foronething,comparativelyfewwordsdoactuallystandforthingsintheworld.Foranother,ifallwordswerelikepropernames,servingjusttopickoutindividualthings,wewouldnotbeabletoformgrammaticalsentencesinthefirstplace.MeaningandunderstandingNotmanypeopleknowthat,in1931,AdolfHitlermadeavisittotheUnitedStates,inthecourseofwhichhedidsomesightseeing,hadabriefaffairwithaladynamedMaxineinKeokuk,Iowa,triedpeyote(whichcausedhimtohal-lucinatehordesoffrogsandtoadswearinglittlebootsandsingingtheHorstWesselLied),infiltratedamunitionsplantnearDetroit,metsecretlywithVice-PresidentCurtisregardingsealskinfutures,andinventedtheelectriccanopener.Thereisagoodreasonwhynotmanypeopleknowallthat:noneofitistrue.Buttheremarkablethingisthatjustnow,asyoureadthroughmyopeningsentence—letuscallitsentence(1)—youunderstooditperfectly,whetherornotyouwerereadytoacceptit,andyoudidsowithouttheslight-estconsciouseffort.Remarkable,Isaid.Itprobablydoesnotstrikeyouasremarkableorsur-prising,evennowthatyouhavenoticedit.Youareentirelyusedtoreadingwordsandsentencesandunderstandingthematsight,andyoufinditnearly\n2PhilosophyofLanguageasnaturalasbreathingoreatingorwalking.Buthowdidyouunderstandsentence(1)?Notbyhavingseenitbefore;Iamcertainthatneverinthehistoryoftheuniversehasanyoneeverwrittenorutteredthatparticularsentence,untilIdid.Nordidyouunderstand(1)byhavingseenaverysimilarsentence,sinceIdoubtthatanyonehaseverproducedasentenceevenremotelysimilarto(1).Youmaysaythatyouunderstood(1)becauseyouspeakEnglishand(1)isanEnglishsentence.Thatistruesofarasitgoes,butitonlypushesthemysterytoarm’slength.Howisitthatyouareableto“speakEnglish,”giventhatspeakingEnglishinvolvesbeingabletoproduceandunderstand,notonlyelementaryexpressionslike“I’mthirsty,”“Shutup,”and“Moregravy,”butnovelsentencesascomplexas(1)?Thatabilityistrulyamazing,andmuchhardertoexplainthanhowyoubreatheorhowyoueatorhowyouwalk,eachofwhichabilitiesisalreadywellunderstoodbyphysiologists.Oneclueisfairlyobviousuponreflection:(1)isastringofwords,Englishwords,thatyouunderstandindividually.Soitseemsthatyouunderstand(1)becauseyouunderstandthewordsthatoccurin(1)andyouunderstandsomethingabouthowtheyarestrungtogether.Asweshallsee,thatisanimportantfact,butfornowitisonlysuggestive.Sofarwehavebeentalkingaboutahumanability,toproduceandunder-standspeech.Butconsiderlinguisticexpressionsthemselves,asobjectsofstudyintheirownright.(2)wgfjsdkhjjiobfglglfud(3)It’sdangeroustosplashgasolinearoundyourlivingroom.(4)Goodofoffprimlytheathethewhy.(1)–(4)areallstringsofmarks(orofnoises,ifutteredaloud).Buttheydifferdramaticallyfromeachother:(1)and(3)aremeaningfulsentences,while(2)and(4)aregibberish.(4)differsfrom(2)incontainingindividuallymeaningfulEnglishwords,butthewordsarenotlinkedtogetherinsuchawayastomakeasentence,andcollectivelytheydonotmeananythingatall.Certainsequencesofnoisesormarks,then,haveafeaturethatisbothscarceinnatureandurgentlyinneedofexplanation:thatofmeaningsome-thing.Andeachofthosestringshasthemorespecificpropertyofmeaningsomethinginparticular.Forexample,(3)meansthatitisdangeroustosplashgasolinearoundyourlivingroom.Soourphilosophicalstudyoflanguagebeginswiththefollowingdata.•Somestringsofmarksornoisesaremeaningfulsentences.•Eachmeaningfulsentencehaspartsthatarethemselvesmeaningful.•Eachmeaningfulsentencemeanssomethinginparticular.\nIntroduction:meaningandreference3•Competentspeakersofalanguageareabletounderstandmanyofthatlanguage’ssentences,withouteffortandalmostinstantaneously;theyalsoproducesentences,inthesameway.Andthesedataallneedexplaining.Invirtueofwhatisanysequenceofmarksornoisesmeaningful?Invirtueofwhatdoessuchastringmeanwhatitdis-tinctivelydoes?Andhow,again,arehumanbeingsabletounderstandandproduceappropriatemeaningfulspeech?TheReferentialTheoryThereisanattractiveandcommonsensicalexplanationofalltheforegoingfacts—soattractivethatmostofusthinkofitbythetimewearetenorelevenyearsold.Theideaisthatlinguisticexpressionshavethemeaningstheydobecausetheystandforthings;whattheymeaniswhattheystandfor.Onthisview,wordsarelikelabels;theyaresymbolsthatrepresent,designate,name,denoteorrefertoitemsintheworld:thename“AdolfHitler”denotes(theperson)Hitler;thenoun“dog”referstodogs,asdotheFrench“chien”andtheGerman“Hund.”Thesentence“Thecatsatonthemat”representssomecat’ssittingonsomemat,presumablyinvirtueof“Thecat”designatingthatcat,“themat”designatingthematinquestion,and“saton”denoting(ifyoulike)therelationofsittingon.Sentencesthusmirrorthestatesofaffairstheydescribe,andthatishowtheygettomeanthosethings.Forthemostpart,ofcourse,wordsarearbitrarilyassociatedwiththethingstheyreferto;some-onesimplydecidedthatHitlerwastobecalled“Adolf,”andtheinscriptionorsound“dog”couldhavebeenusedtomeananything.ThisReferentialTheoryofLinguisticMeaningwouldexplainthesig-nificanceofallexpressionsintermsoftheirhavingbeenconventionallyassociatedwiththingsorstatesofaffairsintheworld,anditwouldexplainahumanbeing’sunderstandingasentenceintermsofthatperson’sknowingwhatthesentence’scomponentwordsreferto.Itisanaturalandappealingview.Indeeditmayseemobviouslycorrect,atleastsofarasitgoes.Andonewouldhaveahardtimedenyingthatreferenceornamingisourcleanest-cutandmostfamiliarrelationbetweenawordandtheworld.Yet,whenexam-ined,theReferentialTheoryverysoonrunsintoseriousobjections.OBJECTION1Noteveryworddoesnameordenoteanyactualobject.First,therearethenamesofnonexistentitemslikePegasusortheEasterBunny.“Pegasus”doesnotdenoteanything,becausethereisinrealitynowingedhorseforittodenote.(Weshalldiscusssuchnamesatsomelengthinchapter3.)Orconsiderpronounsofquantification,asin:(5)Isawnobody.\n4PhilosophyofLanguageItwouldbeatiredjoketotake“nobody”asanameandrespond,“Youmusthaveverygoodeyesight,then.”(LewisCarroll:“Whodidyoupassontheroad?”...“Nobody”...“...Soofcoursenobodywalksslowerthanyou.”1Ande.e.cummings’poem“anyonelivedinaprettyhowtown”2makeslittlesensetothereaderuntils/hefiguresoutthatcummingsisperverselyusingexpressionslike“anyone”and“noone”asnamesofindividualpersons.)Second,considerasimplesubject–predicatesentence:(6)Ralphisfat.Though“Ralph”maynameaperson,whatdoes“fat”nameordenote?Notanindividual.CertainlyitdoesnotnameRalph,butdescribesorcharacter-izeshim(fairlyorno).Wemightsuggestthat“fat”denotessomethingabstract;forexample,itandotheradjectivesmightbesaidtorefertoqualities(or“properties,”“attri-butes,”“features,”“characteristics,”andthelike)ofthings.“Fat”mightbesaidtonamefatnessintheabstract,orasPlatowouldhavecalledit,TheFatItself.Perhapswhat(6)saysisthatRalphhasorexemplifiesorisaninstanceofthequalityfatness.Onthatinterpretation,“isfat”wouldmean“hasfat-ness.”Butthen,ifwetrytothinkofsubject–predicatemeaningasamatterofconcatenatingthenameofapropertywiththenameofanindividualusingthecopula“is,”wewouldneedasecondabstractentityforthe“is”tostandfor,saytherelationof“having,”asintheindividual’shavingtheproperty.Butthatwouldinturnmake(6)meansomethinglike,“Ralphbearsthehav-ingrelationtofatness,”andsowewouldneedathirdabstractentitytorelatethenew“bears”relationtotheoriginalindividual,relationandproperty,andsoon—andon,andon,foreverandever.(TheinfiniteregressherewaspointedoutbyBradley1930:17–18.)Third,therearewordsthatgrammaticallyarenounsbutdonot,intui-tively,nameeitherindividualthingsorkindsofthings—notevennonexistent“things”orabstractitemssuchasqualities.Quine(1960)givestheexamplesof“sake,”“behalf,”and“dint.”Onesometimesdoessomethingforsomeoneelse’ssakeoronthatperson’sbehalf,butnotasifasakeorabehalfwereakindofobjectthebeneficiaryledaroundonaleash.Oroneachievessome-thingbydintofhardwork;butadintisnotathingorkindofthing.(Ihaveneverbeensurewhata“whit”ora“cahoot”is.)Despitebeingnouns,wordslikethesesurelydonothavetheirmeaningsbyreferringtoparticularkindsofobjects.Theyseemtohavemeaningonlybydintofoccurringinlongerconstructions.Bythemselvestheybarelycanbesaidtomeananythingatall,thoughtheyarewords,andmeaningfulwordsatthat.Fourth,manypartsofspeechotherthannounsdonotevenseemtorefertothingsofanysortorinanywayatall:“very,”“of,”“and,”“the,”“a,”“yes,”and,forthatmatter,“hey”and“alas.”YetofcoursesuchwordsaremeaningfulandoccurinsentencesthatanycompetentspeakerofEnglishunderstands.\nIntroduction:meaningandreference5(NoteveryoneisconvincedthattheReferentialTheoryissodecisivelyrefuted,eveninregardtothatlastgroupofthemostclearlynonreferentialwordsthereare.Infact,RichardMontague(1960)setouttoconstructaverysophisticated,highlytechnicaltheoryinwhichevenwordslikethoseareassignedreferentsofahighlyabstractsort,anddohaveameaning,atleastinpart,byreferringtowhattheysupposedlyreferto.WeshallsaymoreofMontague’ssysteminchapter10.)OBJECTION2AccordingtotheReferentialTheory,asentenceisalistofnames.Butamerelistofnamesdoesnotsayanything.(7)FredMarthaIrvingPhylliscannotbeusedtoassertanything,evenifMarthaorIrvingisanabstractentityratherthanaphysicalobject.Onemightsupposethatifthenameofanindividualisconcatenatedwiththenameofaquality,asin(8)Ralphfatnesstheresultingstringwouldhavenormalsubject–predicatemeaning,saythatRalphisfat.(Earlyinhiscareer,BertrandRussellsuggestedthat,bywritingdownalistofnamesfortherightsortsofthingsintherightorder,onewouldformthecollectivenameofastateofaffairs.)Butinfact(8)isungrammatical.Forittotakeonnormalsubject–predicatemeaning,averbwouldhavetobeinserted:(9)Ralph{has/exemplifies}fatnesswhichwouldlaunchBradley’sregressagain.OBJECTION3Asweshallseeanddiscussinthenexttwochapters,therearespecificlin-guisticphenomenathatseemtoshowthatthereismoretomeaningthanreference.Inparticular,coreferringtermsareoftennotsynonymous;thatis,twotermscansharetheirreferentbutdifferinmeaning—“JosephRatzinger”and“thePope,”forexample.Itlooksasthoughweshouldconcludethattheremustbeatleastonewayofbeingameaningfulexpressionotherthanbynamingsomething,pos-siblyevenforsomeexpressionsthatdonamethings.ThereareanumberoftheoriesofmeaningthatsurpasstheReferentialTheory,eventhougheachtheoryfacesdifficultiesofitsown.WeshalllookatsomeofthetheoriesandtheirbesettingdifficultiesinPartII.Butfirst,inthenextthreechapters,weshalllookfurtherintothenatureofnaming,referring,andthelike,inpart\n6PhilosophyofLanguagebecause,despitethefailingsoftheReferentialTheoryofMeaning,refer-enceremainsimportantinitsownright,andinpartbecauseadiscussionofreferencewillhelpusintroducesomeconceptsthatwillbeneededintheassessmentoftheoriesofmeaning.Summary•Somestringsofmarksornoisesaremeaningfulsentences.•Itisanamazingfactthatanynormalpersoncaninstantlygraspthemeaningofevenaverylongandnovelsentence.•Eachmeaningfulsentencehaspartsthatarethemselvesmeaningful.•Thoughinitiallyattractive,theReferentialTheoryofMeaningfacesseveralcompellingobjections.Questions1CanyouthinkofanyfurtherobjectionstotheReferentialTheoryasstatedhere?2Areobjections1and2entirelyfair,orarethereplausiblerepliesthatthereferentialtheoristmightmake?Furtherreading•ProbablythemostpersistentcriticoftheReferentialTheoryisWittgenstein(1953:PartI).AmoresystematicWittgensteinianattackisfoundinWaismann(1965a:ch.8).•Argumentsofthesortlyingbehindobjection3arefoundinFrege(1892/1952a)and(1892/1952b).•Bradley’sregressisfurtherdiscussedbyWolterstorff(1970:ch.4)andbyLoux(1998:ch.1).\nPartIReferenceandreferring\n\n2DefinitedescriptionsOverviewEveniftheReferentialTheoryofMeaningdoesnotholdforallwords,onemightthinkitwouldapplyatleasttosingularterms(termsthatpurporttorefertosingleindividuals,suchaspropernames,pronouns,anddefinitedescriptions).ButGottlobFregeandBertrandRussellarguedpowerfullythatdefinitedescriptions,atleast,donotmeanwhattheymeaninvirtueofdenotingwhattheydenote.Rather,hecontended,asentencecontainingadefinitedescription,suchas“Thewomanwholivesthereisabiochemist,”hassubject–predicateformonlysuperficially,andisreally—logically—atrioofgeneralizations:itisequivalentto“Atleastonewomanlivesthere,andatmostonewomanlivesthere,andwhoeverlivesthereisabiochemist.”Russellarguesforthisanalysisbothdirectlyandbyshowingthatitaffordssolutionstoeachoffourvexinglogicalpuzzles:theProblemofApparentReferencetoNonexistents,theProblemofNegativeExistentials,Frege’sPuzzleaboutIdentity,andtheProblemofSubstitutivity.AvarietyofobjectionshavebeenraisedagainstRussell’sTheoryofDescriptions.P.F.Strawsonpointedoutthatitisatoddswithourusuallinguistichabits:thoughasentencehaving“thepresentKingofFrance”asitssubjectpresupposesthatthereisatleastoneKingofFrance,itisnotfalseforlackofaKing;rather,itcannotbeusedtomakeaproperstatementatall,andsoithasnotruth-value.AndRussell’stheoryignoresthefactthatmostdescriptionsarecontext-bound,anddenoteuniquelyonlywithinacircum-scribedlocalsetting(“Bringmethebookonthetable”).StrawsonarguesmoregenerallythatRusselltreatssentencesandtheirlogicalpropertiesintooabstractanddisembodiedafashion,forgettinghowtheyareactuallyusedbyflesh-and-bloodpeopleinconcreteconversationalpractice.KeithDonnellannotesthat,evenifRussellisrightaboutsomeusesofdescriptions,hehasignoredacommonsortofcaseinwhichadescriptionisused“referentially,”merelytoindicateaparticularpersonorthing,regard-lessofthatreferent’sattributes.Finally,therearefurtherusesofdescriptions,called“anaphoric”uses,whichmaydefyRusselliantreatment.\n10ReferenceandreferringSingulartermsInEnglishoranyothernaturallanguage,theparadigmaticreferringdevicesaresingularterms,expressionsthatpurporttodenoteordesignateparticularindividualpeople,places,orotherobjects(asopposedtogeneraltermssuchas“dog”or“brown”thatcanapplytomorethanonething).Singulartermsincludepropernames(“Jane,”“WinstonChurchill,”“Djakarta,”“7,”3:17p.m.”),definitedescriptions(“theQueenofEngland,”“thecatonthemat,”“thelastdepartmentmeetingbutone”),singularpersonalpronouns(“you,”“she”),demonstrativepronouns(“this,”“that”),andafewothers.EveniftheReferentialTheoryofMeaningisnottrueacrosstheboard,onemightreasonablyexpectittobetrueofsingularterms.ButGottlobFrege(1892/1952a,1892/1952b)and,followinghim,BertrandRussell(1905/1956,1918/1956,1919/1971)showeddefinitivelythatitisnottrueofdefinitedescriptions,andraisedseriousdoubtswhetheritistrueofotherordinarysingulartermseither.FregeandRussellsetforthfourpuzzlesaboutsingularterms,thefirstthreeofwhichgobacktoobjectionsraisedinchapter1againsttheReferentialTheoryofmeaning.THEPROBLEMOFAPPARENTREFERENCETONONExISTENTSConsider:(1)JamesMoriartyisbald.(ProfessorMoriartyisSherlockHolmes’arch-enemy,describedmostfullyinConanDoyle’sstory“TheFinalProblem.”1)Thefollowingsetofstate-mentsisinconsistent(thatis,onpainoflogicalcontradiction,thestatementscannotallbetrue):J1(1)ismeaningful(significant,notmeaningless).J2(1)isasubject–predicatesentence.J3Ameaningfulsubject–predicatesentenceismeaningful(only)invirtueofitspickingoutsomeindividualthingandascribingsomepropertytothatthing.J4(1)’ssubjecttermfailstopickoutordenoteanythingthatexists.J5If(1)ismeaningfulonlyinvirtueofpickingoutathingandascribingapropertytothatthing(J1,J2,J3),andif(1)’ssubjecttermfailstopickoutanythingthatexists(J4),theneither(1)isnotmeaningfulafterall(contrarytoJ1)or(1)picksoutathingthatdoesnotexist.But:J6Thereisnosuchthingasa“nonexistentthing.”TherubisthateveryoneofJ1–J6seemstrue.\nDefinitedescriptions11THEPROBLEMOFNEGATIvEExISTENTIALSThisisaspecialcaseoftheforegoingpuzzlebut,asweshallsee,anaggra-vatedone.Consider:(2)Pegasusneverexisted.(2)seemstobetrueandseemstobeaboutBellerophon’ssteed,Pegasus.Butif(2)istrue,(2)cannotbeaboutPegasus,forthereisnosuchentityforittobeabout.Likewise,if(2)isaboutPegasus,then(2)isfalse,forPegasusmusttheninsomesenseexist.ItisworthnotingaprevioussolutiontotheProblemsofApparentReferencetoNonexistentsandNegativeExistentials,rejectedbyFregeandlaterevenmorevehementlysobyRussell.J1isuncontroversial;J2seemsobvious;J4isjustafact;andJ5istriviallytrue.AlexiusMeinong(1904/1960)hadboldlyleapttodenyJ6,insistingàlaSt.Anselmthatanypossibleobjectofthought—evenaself-contradictoryone—hasbeingofasorteventhoughonlyafewsuchthingsaresoluckyastoexistinrealityaswell.Moriartyhasbeingofthatsortandcanbereferredto,eventhough—fortunatelyforEnglandandtheworld—helacksthepropertyofexisting.2Withthatotherwiseunexplaineddistinctioninhand,Meinongcoulddealhandilywithnegativeexistentialsinparticular.Suchasentencesays,ofanentitythat(ofcourse)hasbeing,thatthatentitylacksexistence.Secretariat,SeabiscuitandSmartyJoneswerehorsesthatexistedbutlackedwings;Pegasushadwingsbutfailedtoexist.Ithappens.Lessimplausibly,FregehimselfdealtwithApparentReferencetoNonexistentsbyrejectingJ3:Hepositedabstractentitiesthathecalled“senses”andarguedthatasingulartermismeaningfulinvirtueofhavingoneofthoseoverandaboveitsreferent—orinthecaseofanonreferringsingularterm,insteadofareferent.Thatis,sincethesingulartermexpressesasense,itismeaningfulwhetherornotitactuallyrefers.Frege’ssolutionstoNegativeExistentialsandtheothertwoproblemswillbebrieflysurveyedinthenextchapter.FREGE’SPuzzLEABOuTIDENTITYAnidentitystatementsuchas(3)MarkTwainisSamuelLanghorneClemenscontainstwosingularterms,bothofwhich(ifthestatementistrue)pickoutordenotethesamepersonorthing.Itseems,then,thatwhatthestate-mentsaysissimplythatthatpersonisidenticalwiththatperson,thatthatpersonisidenticalwithherself.Ifso,thenthestatementistrivial;(3)saysnomorethan“MarkTwainisMarkTwain.”Yet(3)seemsnontrivial,ineach\n12Referenceandreferringoftwoways:first,(3)isinformative,inthatsomeonemightlearnsomethingnewuponreading(3)(eithersomethingaboutTwain’sactualidentityorthatClemenswasthefamousauthor);second,(3)iscontingent,asphilosopherssay—thefact(3)statesisonethatneednothaveheld;realitycouldhavebeenotherwise.Soitseemsthatatleastoneofthesingulartermsfiguringin(3)musthaveandcontributesomekindofmeaningoverandaboveitsreferent.THEPROBLEMOFSuBSTITuTIvITYThefunctionofasingulartermistopickoutanindividualthingandintro-ducethatthingintodiscourse.EvenifonestopsshortoftheentireReferentialTheoryofMeaning,onemightthinkitisinvirtueofthatdenotingrolethatsingulartermsaremeaningfulatall.Therefore,wewouldexpectthatanytwosingulartermsthatdenoteoneandthesamethingwouldbesemanticallyequivalent:wecouldtakeanysentencecontainingoneofthetermsandsub-stitutetheotherofthetwoforthefirstterm,withoutchangingthemeaningoratleastwithoutchangingthetruth-valueofthesentence.Butconsider:(4)AlbertbelievesthatSamuelLanghorneClemenswaslessthan5feettall.andsuppose(4)istrue.Now,AlbertisunawarethatClemenswrotenov-elsandstoriesunderthepenname“Twain.”Wecannotsubstitutetheterm“MarkTwain”for“SamuelLanghorneClemens”in(4)withoutchang-ing(4)’struth-value;theresultisafalsesentence,since(wemaysuppose)AlberthasseenaphotoofTwainandbelievesthathewasofnormalheight.InW.V.Quine’s(1960)terminology,thesententialpositionoccupiedbythenamein(4)isreferentiallyopaque—“opaque”forshort—asopposedtoreferentiallytransparent(“opaque”meansjustthatsubstitutingadifferentsingulartermintothatpositionmaychangethetruth-valueofthecontainingsentence).Whatcausestheopacityisthe“believesthat”construction,sincethesentence“SamuelLanghorneClemenswaslessthan5feettall,”standingalone,istransparent:IfClemenswaslessthan5feettall,thensoofcoursewasTwain,“they”beingtheverysameperson.Russell’sTheoryofDescriptionsRussellinitiallyposedthefourpuzzlesintermsofdefinitedescriptionsratherthanpropernames,becausehewasinterestedinthelogicoftheword“the.”(“Itmaybethoughtexcessivetodevotetwochapters[ofhisIntroductiontoMathematicalPhilosophy]tooneword,buttothephilosophicalmathemati-cianitisawordofverygreatimportance:likeBrowning’sgrammarianwiththeencliticδε,IwouldgivethedoctrineofthiswordifIwere‘deadfromthewaistdown’andnotmerelyinaprison”3(1919/1971:167).)\nDefinitedescriptions13Nottoosurprisingly,Russellarguedonthebasisofthepuzzlesthatdefinitedescriptionsdohaveandcontributemeaningsthatgobeyondtheirreferentsalone.HisTheoryofDescriptions,asithassincebeencalledandcapitalized,takestheformofacontextualdefinitionoftheword“the”asitoccursintypicaldefinitedescriptions.Thatis,ratherthandefiningthewordexplicitly(howwouldyougoaboutcompletingtheformula,“The=def...”?),Russelloffersarecipeforparaphrasingstandardtypesofwholesentencecontaining“the,”insuchawayastoexhibittheroleof“the”indirectly,andtorevealwhathecalledthesentences’“logicalforms.”(Hedoesnotheretreatpluralusesof“the,”orthegenericuseasin“Thewhaleisamammal.”Noticethatdefinitedescriptionscanbeformedwithoutuseof“the,”forexamplebywayofpossessives,asin“mybrother”or“Doris’eggsaladsandwich,”thoughperhapswemightparaphrasethosealongthelinesof“thebrotherofme.”)HereisRussell’scontextualdefinitionof“the.”Letustakeaparadigmaticsentence,oftheform“TheFisG.”(5)TheauthorofWaverleywasScotch.4(5)appearstobeasimplesubject–predicatesentence,referringtoanindividual(SirWalterScott)andpredicatingsomething(Scottishness)ofhim.Butappearancesaredeceiving,Russellsays.Noticethattheostensiblesingularterm,“TheauthorofWaverley,”consistsofourtroublesomeword“the”pastedontothefrontofapredicativeexpression,andnoticetoothatthemeaningofthatexpressionfigurescruciallyinourabilitytorecognizeorpickouttheexpression’sreferent;tofindthereferentwehavetolookforsomeonewhodidwriteWaverley.Russellsuggeststhat“the”abbreviatesamorecomplexconstructioninvolvingwhatlogiciansandlinguistscallquanti-fiers,wordsthatquantifygeneralterms(“allteenagers,”“somebananas,”“sixgeesea-laying,”“mostpoliceofficers,”“nolightbulbs,”andthelike).Indeed,hethinksthat(5)asawholeabbreviatesaconjunctionofthreequantifiedgeneralstatements,noneofwhichmakesreferencetoScottinparticular:(5a)AtleastonepersonauthoredWaverley.(5b)AtmostonepersonauthoredWaverley.(5c)WhoeverauthoredWaverleywasScotch.Eachof(5a)–(5c)isintuitivelynecessaryforthetruthof(5).IftheauthorofWaverleywasScotch,thentherewassuchanauthor;ifthereweremorethanoneauthor,“the”shouldnothavebeenused;andiftheauthorwasScotchitfollowstriviallythatwhoeverdidtheauthoringwas.And(5a)–(5c)takentogethercertainlyseemsufficientforthetruthof(5).Soweseemtohaveasetofindividuallynecessaryandjointlysufficientconditionsfor(5);thatinitselfisapowerfulargumentforRussell’sanalysis.\n14ReferenceandreferringInstandardlogicalnotation:LetWrepresentthepredicate“...authoredWaverley”andSrepresent“...wasScotch.”ThenRussell’sthreeconditionsare:(a)(∃x)Wx(b)(x)(Wx→(y)(Wy→y=x))(c)(x)(Wx→Sx)(a)–(c)areconjointlyequivalentto(d)(∃x)(Wx&((y)(Wy→y=x)&Sx))Russell’spositionisthat(d)correctlyexpressesthelogicalformof(5),asdistinctfrom(5)’ssuperficialgrammaticalform.Wehavealreadyencounteredanexampleofthisdistinction,inchapter1,illustratedbythesentence“Isawnobody.”Superficially,thatsentencehasthesameformas“IsawMartha”—Subject+TransitiveVerb+Object.Yetthetwodiffersharplyintheirlogicalproperties.“IsawMartha”entailsthatIsawsomeone,while“Isawnobody”entailspreciselytheopposite;itisequivalentto“It’snotthecasethatIsawanyone”andto“ThereisnoonethatIsaw.”ThoughsomeonejustbeginningtolearnEnglishmighttakeitasone,“nobody”isnotreallyasingularterm,butaquantifier.Inlogicalnotation,lettingArepresent“saw”andirepresent“I,”“Isawnobody”isexpressedas~(∃x)Aixor,equivalently,(x)~AixandtheexplicitinferencerulesgoverningthisformalnotationexplainthelogicalbehavioroftheEnglishsentencethustranslatedintoit.Sotoo,Russellmaintained,theapparentsingulartermin(5),“TheauthorofWaverley,”isnotreally(thatis,attheleveloflogicalform)asingulartermatall,butaconvenient(ifmisleading)abbreviationofthemorecomplicatedquantificationalstructuredisplayedin(a)–(c).Asheputsit,theapparentsingularterm“disappearsonanalysis.”Ourpuzzleshaveariseninfactfromapplyingprinciplesaboutsingularreferencetoexpressionsthatarenotreallysingulartermsatallbutonlymasqueradeassuch.LetusnowgothroughthefourpuzzlesandexhibitRussell’ssolutionsonebyone.APPARENTREFERENCETONONExISTENTSRussellputtheProblemofApparentReferencetoNonexistentsintermsof(6):(6)ThepresentKingofFranceisbald.\nDefinitedescriptions15Weturn,then,totheinconsistentsetofstatementscorrespondingtoJ1–J6above,replacing“(1)”with“(6)”andreletteringthestatementswith“K”insteadof“J.”(Thus,statementK1is“(6)ismeaningful(significant,notmeaningless),”K2is“(6)isasubject–predicatesentence,”andsoon.)Now,letusparaphrase(6)accordingtotheforegoingmethod:AtleastonepersonispresentlyKingofFrance[moreperspicuously,presentlykingsFrance],andatmostonepersonispresentlyKingofFrance,andwhoeverispresentlyKingofFranceisbald.Noproblem.Thefirstoftheforegoingthreeconjunctsissimplyfalse,sincenoonekingsFranceatpresent;so(6)itselfcomesoutfalseonRussell’sanalysis.Whenwefirststatedthepuzzle,itlookedasthoughonehadtorejecteitherJ3/K3or(outrageously)J6/K6,sinceJ2seemedasobviousastheotherundeniableJstatements.ButnowRussellingeniouslydeniesstatementK2,“(6)isasubject–predicatesentence,”sincehedeniesthat“ThepresentKingofFrance”is“really”asingularterm.Ofcourse(6)hassubject–predicateforminthesuperficialgrammaticalway.ButnoticeagainthatourthreeconjunctsareallgeneralstatementsandthatnonementionsanyspecificindividualcorrespondingtotheallegedKing;“theKing”figuresnowhereinlogicalformasasubject.(Alternativelyandlessdramatically,wecouldkeepK2,understandingitasalludingtosuperficialgrammaticalform,andrejectK3onthegroundsthatasuperficiallysubject–predicatesentencecanbemeaningfulwithoutpickingoutanyparticularindividualbecauseitabbreviatesatrioofpurelygeneralstatements.)NEGATIvEExISTENTIALSLetusapplyRussell’sanalysisto(7):(7)ThepresentKingofFrancedoesnotexist.Now,thereisaRussellianparaphraseof(7)thatleaves(7)justasanomalousasitseemstothenaivehearer.Thatistheparaphrasethattakes“exist”tobeanordinarypredicatelike“wasScotch”or“isbald,”andtakes“not”tomodifyorapplytothatpredicate:AtleastonepersonispresentlyKingofFrance,and\n16ReferenceandreferringatmostonepersonispresentlyKingofFrance,andwhoeverispresentlyKingofFrancedoesnotexist.TheanomalyisthatthefirstconjunctassertstheexistenceofapresentKing,whilethethirdconjunctdeniesit.Nowonder(7)soundspeculiartous.Tomakesenseof(7),wemustunderstand“not”,notasmodifyingtheverb“exist”,butasapplyingtotherestof(7)asawhole,thus:Not:(ThepresentKingofFranceexists).[Thatis,itisfalsethat:thepresentKingofFranceexists],whichisobviouslywhatwouldbemeantbysomeonewhouttered(7)seriously.ThenweapplyRussell’spatternofanalysisinsidethe“not,”asfollows.Not:(AtleastonepersonispresentlyKingofFrance,andatmostonepersonispresentlyKingofFrance,andwhoeverispresentlyKingofFranceexists).Insymbols:~(∃x)(Kx&((y)(Ky→y=x)&Ex))where“E”represents“exists.”(Actually,“exists”isitselftreatedasaquanti-fierinlogicaltheory,andsotheconjunctExoughtproperlytobereplacedby(∃z)(z=x),whichisredundant.)Theintuitivecontentof(7)isjust,“NooneisuniquelyKingofFrance,”or“NooneuniquelykingsFrance,”andRussell’sparaphrasehasthevirtueofbeingpreciselyequivalenttothat.NowhereinRussell’sanalysisdowepickoutanindividualandsayofthatindividualthathedoesnotexist,sotheProblemofNegativeExistentialsvanishes,atleastforthecaseofdefinitedescriptions.Inthispreferredunderstandingof(7),thedescriptionoccursinwhatRussellcalled“secondary”position;thatis,wehaveconstrueditsunderlyingquantifiers“atleast,”“atmost,”and“whoever”asfallinginsidethe“not.”Theprevious,dispreferredparaphrasegavethedescription“primary”posi-tion,placingitfirstinthelogicalorderwiththe“not”insideandgovernedbyit.Ameaningdistinctionofthiskindiscalledascopedistinction:inmorecontemporaryterminology,thesecondaryreadingisthatonwhichthequan-tifierstake“narrow”scope,fallinginsidethescopeof“not”;ontheprimaryreadingthequantifiersareoutsidethescopeof“not,”and“not”isintheirscope.\nDefinitedescriptions17FREGE’SPuzzLEAnexamplefeaturingadefinitedescription:(8)ThepresentQueenofEnglandis[oneandthesameindividualas]ElizabethWindsor.(8)’sleft-handtermisadefinitedescription,soletusparaphraseitawayinRussell’smanner:AtleastonepersonispresentlyQueenofEngland[presentlyqueensEngland],andatmostonepersonispresentlyQueenofEngland,andwhoeverispresentlyQueenofEnglandis[oneandthesameas]ElizabethWindsor.Insymbols:(∃x)(Qx&((y)(Qy→y=x)&x=e)).Nowweseeeasilywhyouroriginalidentitystatementisnontrivial.OfcoursewelearnsomethingwhenwehearRussell’sparaphrase,somethingsubstantiveaboutElizabethandthepresentQueenboth.Andofcoursetheidentitystatementiscontingent,sincesomeoneelsemighthavebeenQueen(theremightevenhavebeennoQueenatall),Elizabethmighthaverunawayfromhomeandformedarockbandratherthanbecrowned,orwhatever.TheTheoryofDescriptionsseemstogiveacorrectaccountoftheidentitystate-ment’sintuitivecontent.NotethatonRussell’sviewthestatementisonlysuperficiallyanidentitystatement;reallyitisapredicationandattributesacomplexrelationalpropertytoElizabeth.Thatleavesuswiththeproblemofhowarealidentitystatementcouldmanagetobebothtrueandinformative,moreofwhichinchapter3.SuBSTITuTIvITYLetusreturntoAlbert.Hehasbeenreadingphilosophy,and:(9)AlbertbelievesthattheauthorofNothingandBeingnessisaprofoundthinker.\n18ReferenceandreferringNow,AlbertisunawarethattheauthorofNothingandBeingnessmoonlightsbywritingcheap,disgustingpornography.Wecannotsubstitutetheterm“theauthorofSizzlingVeterinarians”for“theauthorofNothingandBeingness”in(9)withoutchanging(9)’struth-value;theresultisafalsesentence,sinceAlbertbelievesthattheauthorofSizzlingVeterinariansisadroolingmoron.(IamafraidthisrevealsthatAlberthasreadSizzlingVeterinarians.)Thepositionoccupiedbythedefinitedescriptionin(9)isopaque.In(9),thedefinitedescriptionoccursaspartofwhatAlbertbelieves,soweshallstartourRussellianparaphrasewith“Albertbelieves”andthenapplyRussell’spatternofanalysis,givingthedescriptionsecondaryoccurrenceornarrowscope:Albertbelievesthefollowing:(AtleastonepersonauthoredNothingandBeingness,andatmostonepersonauthoredNothingandBeingness,andwhoeverauthoredNothingandBeingnessisaprofoundthinker).ThisisaprettygoodaccountofwhatAlbertbelieves.5Andnowitisobvi-ouswhywemaynotsubstitute“theauthorofSizzlingVeterinarians”into(9),forthecorrespondinganalysisoftheresultingsentencewouldcomeout:Albertbelievesthefollowing:(AtleastonepersonauthoredSizzlingVeterinarians,andatmostonepersonauthoredSizzlingVeterinarians,andwhoeverauthoredSizzlingVeterinariansisaprofoundthinker).SincethisattributesanentirelydifferentbelieftoAlbert,itisnowonderthatitisfalseeventhough(9)istrue.(Ofcourse,attheleveloflogicalformwehavenotmadeasubstitutionatall,forthesingulartermshave“disap-pearedonanalysis”andarenolongertheretobesubstituted.)Thefourpuzzlesmadeitclearthatdefinitedescriptionsdonothookontotheworldbydirectlynamingandnothingelse.6Butweneededapositivetheoryofhowtheydohookontoit.Russellhasprovidedoneverywellmoti-vatedtheory.Noticethateventhoughdefinitedescriptionsarenotassignedreferentsinthewaythatnamesare,andeventhoughtheyarenot“really”\nDefinitedescriptions19singulartermsatall,theystillpurporttohavesingleindividualsthatanswertothem;whenadescriptiondoesinfacthavethecorrespondingindividualthatitpurportstohave—thatis,whentheredoesexistauniqueso-and-so—Ishallspeakofthedescription’ssemanticdenotatumorsemanticreferent.Butthe“hook”betweenadefinitedescriptionanditssemanticreferentis(onRussell’sview)farlessdirectthanisthehookbetweenasimplenameanditsbearer.ObjectionstoRussell’stheoryImpressiveasRussell’sachievementis,anumberofobjectionshavebeenbroughtagainsttheTheoryofDescriptions,chieflybyStrawson(1950).Beforewetakethoseup,Inoteanimportantcriticismthatmightbemadeatjustthispoint,thoughRussellquicklymovedtoforestallit.WhenIsetoutthefourpuzzleswithwhichwebegan,Icalledthempuz-zles“aboutsingularterms.”Ihavesinceexpoundedeachofthembyusingexamplesfeaturingdefinitedescriptions,andwieldedRussell’sTheoryofDescriptionsagainstthem.Buttheyareindeedpuzzlesaboutsingulartermsacrosstheboard,notjustdescriptions.Wehavealreadyusedpropernamestomakeapparentreferencetononexistents,andwecouldevenusepronouns(“you,”saidbyScroogetoMarley’sghost);Frege’sPuzzleandSubstitutivityofcourseariseforpropernames.TheseseemtobeexactlythesameproblemsasthosethatRussellhasstatedintermsofdescriptions.ItlooksasthoughRussellhassimplymissedtheboat,becausehehasgivenatheorythatbyitsnatureappliesonlytooneveryspecialsubclassofsingularterms,whileanyadequatesolutiontothepuzzlesoughttogeneralize.Russell’ssolutiontothisproblemwasifanythingevenmoreingeniousthantheTheoryofDescriptionsitself.Inbrief,itwastoinvokeanotherdistinctionbetweensurfaceappearanceandunderlyinglogicalreality,andclaimthatwhatweordinarilycallpropernamesarenotreallypropernamesatall,butrathertheyareabbreviationsfordefinitedescriptions.ButIshallpostponeexaminationofthatthesisuntilthenextchapter.Strawson’scritiquewasradicalandsearching.Indeed,RussellandStrawsonwererespectivelyfigureheadsfortwoverydifferentapproachestothestudyoflanguage(andtoalesserdegreefortwogreatrivalsystemsoftwentieth-centuryphilosophy),thoughweshallnotgointothatuntilchapter6.TosetthestageforStrawson’sobjections,IshallmerelynotethatwhereasRussellthoughtintermsofsentencestakenintheabstractasobjectsinthemselves,andtheirlogicalpropertiesinparticular,Strawsonemphasizedhowthesentencesareusedandreactedtobyhumanbeingsinconcretecon-versationalsituations.Russell’smostfamousarticle(1905/1956)wascalled“OnDenoting,”andinitdenotingwastakentobearelationbetweenanexpression,consideredinabstraction,andthethingthatistheexpression’sreferentordenotatum.Strawson’stitlewas“OnReferring,”whichhemeantironically,becausehethoughtofreferringnotasanabstractrelationbetween\n20Referenceandreferringanexpressionandathingbutasanactdonebyapersonatatimeonanoccasion.ThiswayoflookingatthingsgaveStrawsonquiteanewslantonthefourproblems.Strawsonholdsthatexpressionsdonotreferatall;peoplerefer,usingexpressionsforthatpurpose.Thisisreminiscentofthe(U.S.)NationalRifleAssociation’sslogan,“Gunsdon’tkillpeople,peoplekillpeople.”CertainlythereisanobvioussenseinwhichStrawsonisright.Touseanexampleofhis,ifIwritedown,“Thisisafineredone,”“This”doesnotrefertoanything—andnodeterminatestatementhasbeenmade—untilIdosomethingtomakeitrefer.AnexpressionwillcometoreferonlyifIuseitinasuitablywell-engineeredcontext,sothatitdoesrefertoaparticularthingorperson.Butthatisamatteroftheexpressionbeingused,andwhenIdouseit,itisIthatamdoingthework,nottheexpression.OBJECTION1AccordingtoRussell,sentence(6)(“ThepresentKingofFranceisbald”)isfalseowingtothelackofanysuchKing.Strawsonpointsoutthatthatver-dictisimplausible.Supposesomeonecomesoutandasserts(6).Wouldthatperson’shearersreactbysaying“That’sfalse”or“Idisagree”?Surelynot.Rather,Strawsonmaintains,thespeakerhasproducedanonlyostensiblyreferringexpressionthathasmisfired;thespeakerhassimplyfailedtorefertoanythingandsohasfailedtomakeacompletestatement.Thespeaker’sutteranceiscertainlydefective,butnotinthesamewaythat“ThepresentQueenofEnglandhasnochildren”isdefective.Itisnotincorrectbutabor-tive;itdoesnotevengetachancetobefalse.Sincenoproperstatementhasbeenmadeinthefirstplace,itfollowsthatnothingeithertrueorfalsehasbeensaid.Ahearerwouldeitherjustnotcomprehendorwouldsay“Backup”andquestiontheutterance’spresupposition(“I’mnotfollowingyou;Francedoesn’thaveaking”).7StrawsonthereforesolvestheProblemofApparentReferencetoNonexistentsbydenyingK3:(6)ismeaningful,inthatithasalegitimateuseinthelanguageandcouldbeusedtosaytrueorfalsethingsiftheworld(ortheFrench)weremorecooperative,butnotbecauseitsucceedsinpickingoutanyindividualthing.Russellthoughtofameaningfulsentenceasasentencethathasameaning,or,asheputit,asentencethatexpressesaproposition.Asentence’slogicalform,onhisview,isreallythatofthepropositionthesentenceexpresses.Butpropositionsbytheirnatureareeithertrueorfalse.Strawsoneschewstalkof“propositions,”anddeniesthatsentencesarethekindofthingsthatcanbetrueorfalseatall.Whatbearsthepropertiestruthandfalsityareratherthestatementsmadewhenspeakerssucceedinsayingsomething,andnoteveryactofutteringdoessucceedinthatway,fornoteverymeaningfulsentenceisalwaysusedtomakeastatement.\nDefinitedescriptions21Russellianshaveastandardreplytoobjection1,butitdependsonsomenotionsthatIshallnotdevelopuntilchapter13,soIshallpostponeituntilthen.OBJECTION2Strawsonfurthercriticizestheclaim,whichheattributestoRussell,that“partofwhat[aspeaker]wouldbeasserting[inuttering(6)]wouldbethatthereatpresentexistedoneandonlyonekingofFrance”(1950:330).Thatclaimtooisimplausible,foralthoughthespeakerpresupposesthatthereisoneandonlyoneking,thatiscertainlynotpartofwhatthespeakerasserts.Butthatisamisunderstanding:Russellhadmadenosuchclaim.Hesaidnothingatallaboutactsofasserting.PerhapsStrawsonisassumingonRussell’sbehalfthatwhateverislogicallyimpliedbyasentenceisnecessarilyassertedbyaspeakerwhouttersthatsentence.Butthelatterprincipleisfalse:ifIsay“FatTommycan’trunorclimbatree,”IdonotassertthatTommyisfat,eventhoughmysentencelogicallyimpliesthatheis;ifIsay“Tommyisfivefeetseveninchestall,”IdonotassertthatTommyiseithergreaterthantwomilesorlessthaneighteenmilestall.OBJECTION3Strawsonpointsoutthatmanydescriptionsarecontext-bound.Heofferstheexampleof:(10)Thetableiscoveredwithbooks.Presumablythesubjecttermisadefinitedescription,usedinastandardwayratherthaninanydifferentorunusualway.ButifweapplyRussell’sanalysisweget“Atleastonethingisatableandatmostonethingisatableandanythingthatisatableiscoveredwithbooks”—whichbywayofitssecondconjunctentailsthatthereisatmostonetable,intheentireuniverse.Thatcannotbeshruggedoff.Howeverunwillingly,Russellisgoingtohavetotakesomenoticeofthecontextofutterance.Hehasseveraloptions.Afterall,Strawsonhasnomonopolyonthefactthatwhensomeonesays“Thetable,”wehearersgenerallyknowwhichtableismeant,becausesomethinginthecontexthasmadeitsalient.Itmaybetheonlytableinsight,ortheonlyoneintheroom,ortheonewehavejustbeentalkingabout.Russellmaysaythatthereisellipsishere;thatinthecontext,“Thetable”isshortforamoreelaboratedescriptionthatisuniquelysatisfied.(Asweshallseeinthenextchapter,Russellwasnoenemyofellipsishypotheses.)Theellipsisviewhassomedisturbingimplications.Russellthinksoflogi-calformasobjectivelyreal—thatsentencesreallydohavethelogicalforms\n22Referenceandreferringheposits.Soif“Thetable”iselliptical,therehasgottobeadeterminateanswertothequestion,“Whatmaterialisellipsized?”Andtheanswerwillmatterbecause,dependinguponwhichcandidateyoupick,(10)willturnouttosaysomethingcompletelydifferent.Ifwesaythat“Thetable”meansthetableinthisroom,thenwehaveintroducedtheconcept“room,”andconstrued(10)asbeingliterallyaboutaroom,indeedashavingthepredicate“room”hiddeninitsunderlyinglogicalstructure.Perhapsabetterapproachwouldbeanappealtorestrictedquantification(asinLycan1984andNeale1990).Oftenwesaythingslike“Everyonelikesher,”meaning,noteverypersonintheuniverse,buteveryoneinacertaincontextuallyindicatedsocialcircle.Or“Nobodygoestothatrestaurantanymore,”whichisunlikelytomeanthatnohumanbeingatallgoesthere;itwouldmorecommonlymean,nooneofoursort(whateversortthatis).8Whatlogicianscallthedomainsoverwhichquantifiersrangeneednotbeuniversal,butareoftenparticularclassesroughlypresupposedinthecontext.Infact(youcancheckthisforyourself),practicallyallquantificationthatoccursinEnglishisrestrictedquantification:“I’lleatanythingonpizza,”“There’snobeer,”andeven“Iwouldn’ttradethiscarforanythingintheworld.”OfcoursetheusualRusselliananalysisstartswithaquantifier:“Atleastonethingisatable....”Letussimplyregardthatquantifierasrestrictedintheappropriateway.Thesamerestrictionwillapplytothe“atmostonething,”andsowelosetheunwantedimplicationthatthereisatmostonetableintheuniverse;(10)willnowimplyonlythatthereisatmostonetableofthecontextuallyindicatedsort,whichisfine.Theappealtorestrictedquantificationdiffersfromtheellipsishypothesis,inthatitdoesnotrequirethatexplicitconceptualmaterialbeclandestinelymentionedin(10).Thequantifierrestrictionismorelikeasilentdemonstra-tivepronoun:“Atmostonetableofthatsort,”wherethecontextfixesthereferenceof“that.”SoweseemtohavesolvedthetableproblemonRussell’sbehalf.Buttherearemoreaggravatedproblemcases.Consider(11):(11)Ifabishopmeetsanotherbishop,thebishopblessestheotherbishop.(Heim1990)Forfurtherexamples,seeReimer(1992),StanleyandSzabó(2000),LudlowandSegal(2004)andLepore(2004).Also,thereisstillageneralproblemofhowquantifiersgetrestrictedincontext,whatdeterminestheexactrestricteddomains(whicharealmostalwaysvaguetoboot),andhowonearthhearersidentifytherightdomainsasquicklyandaseffortlesslyastheydo.Butwehavethatgeneralproblemanyway;itposesnospecialobjectiontoRussell’sTheoryofDescriptions.IpausetoofferapartialrebuttalofStrawson’snotionthatpeopleratherthanexpressionsrefer.RecalltheNationalRifleAssociation’sslogan,“Gunsdon’tkillpeople,peoplekillpeople.”Anappropriateresponseis,“Yes,but\nDefinitedescriptions23theykillthemmuchmoreeasilyandefficientlyusingguns,”andthereisaperfectlygoodsenseinwhichthegundidkillthevictim.Sotoothereisatleastasecondarysenseinwhichexpressionsdorefer.Thereisnothingatallwrongwithsayingthat,inaparticularcontext,theexpression“Thetable”referstothesalientpieceoffurniture.Moreover,wehavealreadyintroducedthenotionofthe“semanticreferent”ofadescription:incontext,remember,adescription’ssemanticreferentiswhateverobject(ifany)infactuniquelysatisfiesthedescription.NoticethatRusselltoohasanobjectiontotalkingaboutthereferentofadescription.Hewantstoinsistthatdescriptionsarenotreallyreferringexpressionsatall;asentencecontainingoneabbreviatesamassofquantifi-cationalmaterialthatisentirelygeneralandnotaboutanyoneinparticular.ButmynotionofasemanticreferentappliesequallyagainstRussellonthispoint.Thereisatleastthatsecondarysenseinwhichadescriptioncanhaveareferent.AnditisperfectlyharmlessforaRusselliantograntthatdefi-nitedescriptionsdorefer,solongass/heremembersthattheydonotdoitdirectly,inthewaywemayhavethoughtpropernamesdo.IturntoanobjectionmadebyKeithDonnellan(1966).OBJECTION4Donnellannoticedcasesinwhichwedoseemtousedefinitedescriptionsasiftheyarejusttagsornames,solelytorefertoindividuals.AndinsuchcasestheRusselliananalysisdoesnotcapturewhatseemstobesaidwhentherelevantsentencesareuttered.ThoughDonnellanintendedhisarticlemodestlyasanadjudicationoftheRussell–Strawsondispute,hisinsighthaswiderapplication,andIshallexpounditinitsownterms.DONNELLAN’SDISTINCTIONDonnellancalledattentiontowhathecalledthereferentialuse,asopposedtotheattributiveuse,ofadefinitedescription.Themostobvioustypeofref-erentialuseiswhenadescriptionhasgrowncapitallettersandisreallyusedasatitle.Aclassicexampleis“TheHolyRomanEmpire,”whosereferent,asVoltaireobserved,wasneitherholy,norRoman,noranempire.Or“TheGratefulDead,”whichisthenameofarockband;sentencescontainingthattitledonotmeanthatatleastonethingisgratefulanddeadand...Russellmightfairlyretortthat,asthecapitallettersshow,thosetitlesarenotbeingusedasdescriptionsatall,but(ofcourse)asfusedtitles.“TheSwan”isthenameofapieceofinstrumentalmusicbySaint-Saëns,andsentencescontainingthattitleareaboutmusic,notaboutwaterfowl.ButDonnellanshowsthattherearelessformalcasesinwhichweusedescriptionssolelytofocusonaparticularindividualregardlessofthatpersonorthing’sattributes.\n24ReferenceandreferringForcontrast,hereisastandardRussellianexample.WecomeacrossthehideouslymurderedbodyofSmith,andIassert(12)Smith’smurdererisinsanemeaningthatwhoevercommittedthisterriblecrimeisinsane.DonnellanhasnoquarrelwithRussellhere;thisiswhathecallstheattributiveuseofthedescription.Butsupposeinsteadthatwehavenotseenthebodyandhavenootherdirectknowledgeofthematter;Joneshasbeenarrestedandchargedwiththecrimeandweareattendinghistrial.Theprosecution’scaseisexcellent,andweareprivatelypresumingthatJonesisguilty;also,heisrollinghiseyesanddroolinginahomicidalmanner.HeretooIsaytoyou(12),“Smith’smurdererisinsane.”InthecontextIamonlyusingthedescription“Smith’smurderer”torefertothepersonwearelookingat,thedefendant,regardlessofwhatattributeshehas.Moreover,whatIsaidistrueifandonlyifthedefendantisinsane,regardlessofhishavingcommittedthemurder.ThisiswhatDonnellancallsthereferentialuse.Donnellan’sobjectiontotheTheoryofDescriptionsisjustthatthetheoryoverlooksthereferentialuse;Russellwritesasifalldescriptionswereusedattributively.But,againstStrawson,Donnellancomplainsthatequallyhedidnotseetheattributiveuse,thatStrawsonwritesasifalldescriptionswereusedreferentially,inacontext,todrawsomebody’sattentiontoaparticularperson,placeorthing.ThusbothStrawsonandRussellweremistakeninthinkingthatdefinitedescriptionsalwaysworkinoneway,becausethereisanambiguityacknowledgedbyneither.Donnellandoesnottakeapositionastowhatkindofambiguityitis;inparticularhedoesnottrytodecidewhetherthesentence(12)itselfhastwodifferentmeaningsexplainingthedescription’sevidentlydistinct“uses.”Donnellangivesseveralinformalcharacterizationsofthenewreferentialuse:“Aspeakerwhousesadefinitedescriptionreferentiallyinanasser-tion...usesthedescriptiontoenablehisaudiencetopickoutwhomorwhatheistalkingabout”(p.285).Thedescriptiondoesnot“occuressentially,”butis“merelyonetoolfordoingacertainjob—callingattentiontoapersonorthing—andingeneralanyotherdevicefordoingthesamejob,anotherdescriptionoraname,woulddoaswell”(ibid.).“[W]eexpectandintendouraudiencetorealizewhomwehaveinmind...and,mostimportantly,toknowthatitisthispersonaboutwhomwearegoingtosaysomething”(pp.285–6).Thisallsoundsstraightforwardlycorrect,forthe“Smith’smurderer”case.9However,Donnellangoesontoaddafurthercharacterization:intheattributiveuseof“TheøisY,”“ifnothingistheøthennothinghasbeensaidtobeY,”whereasinthereferentialcase“thefactthatnothingistheødoesnothavethisconsequence”(p.287).HetakesthispointfromLinsky(1963),whoofferstheexampleofsomeone(perhapsataparty)whoobservesawomanandhermalecompanionandsays,“Herhusbandiskindtoher.”\nDefinitedescriptions25DonnellanandLinskyagreethat,evenifthewomanisinfactunmarried,itisthecompanionthatisreferredto,andthatwhatissaidisthatthatpersoniskindtoherregardlessofhisnotactuallybeingherhusband.Onthisview,therealreferentdiffersfromwhatIhavebeencallingthesemanticreferent,therebeingnosemanticreferentinLinsky’sexample.OrsupposethatintheSmithcase,againstalltheevidence,Jonesisinno-cent;Smithcommittedsuicideandthereisnomurderer.(OrperhapsSmithisnotevendead,buthasbeenlanguishinginastateofdeepsuspendedani-mation.)Intuitively,Donnellanmaintains,thatdoesnotchangewhatIsaid.AndwhatIsaidistrueifandonlyifJonesisinsane,regardlessoftherebeingnomurderer.Donnellangivesthefurtherexampleofapartyguestseeinganinteresting-lookingpersonsippingfromamartiniglass;theguestasks,“Whoisthemandrinkingamartini?”Infacttheglassholdsonlywater,but,Donnellanmaintains,theguest’squestionisabouttheinteresting-lookingman,andnotabout(say)Dino,offinthebilliardroom,whoisinfacttheoneandonlymanatthepartydrinkingamartini.Exampleslikethese,sometimescalled“near-miss”cases,aredisputed.FollowingGrice(1957)andfloutingStrawson,Kripke(1979a)distinguishesbetweenwhatalinguisticexpressionitselfmeansorreferstoandwhataspeakermeansorreferstoinusingtheexpression.Forexample,takenliter-ally,thesentence“Albert’sanelegantfellow”meansthatAlbertisanelegantfellow,butaspeakermightuseitsarcasticallytopointoutthatAlbertisarevoltingslob.(Weshallsaymuchmoreofdisparitiesbetweenspeaker-meaningandliteralexpressionmeaninginchapters7and13.)Sotoo,Imaysay“Smith’smurderer,”whichphrasetakenliterallymeanswhoevermur-deredSmith,and(myself)honestlymeanJonesonthestandandaccuratelybetakentomeanJones.Linsky’sspeakerhimselfmeansthelady’scompanion,buttheexpression“Herhusband”accordingtotherulesofEnglishmeanswhoever(ifanyone)ismarriedtoher;Donnellan’spartyguestobviouslymeanstheinteresting-lookingman,thoughthephrase“themandrinkingamartini”literallymeanswhatevermanisinfactdrinkingamartini.Speakersin“near-miss”casesdomeanwhatDonnellansaystheymean,andmeantruethings,but(aswith“Albert’sanelegantfellow”)theydothosethingsbyutteringsentencesthatareinfactfalse.Letusdefinespeaker-referencealittlemoreformally,tocontrastwithsemanticreference.Thespeaker-orutterer’sreferentofadescriptiononanoccasionofitsuseistheobject,ifany,towhichthespeakerwhousedthedescriptionintendedtocalltotheattentionofher/hisaudience.(Thespeaker-referentistheobjectthattheutterermeanstobetalkingabout.)Fortunately,communicationgoesbyspeaker-meaningandspeaker-refer-ence:ifI(speaker-)meanJoneswhenIsay“Smith’smurderer”andyoutakemetomeanJonesandunderstandmetohavesaidthatJonesisinsane,thenyouhaveunderstoodmecorrectlyandcommunicationhassucceeded;itdoesnotmatterthatthesentenceIutteredwasaccordingtoitsliteralmeaninguntrue,anymorethanitmattersthat“Albert’sanelegantfellow”isliterallyfalse.\n26ReferenceandreferringThus,accordingtoKripke,Donnellanhasfailedtoshowthatasentencecontainingadefinitedescriptioncanbetrueevenifnothing(orsomethingextraneous)isthedescription’ssemanticreferent.EvenifKripkeisrightaboutthenear-missexamples,itisimportanttoholdontosomeversionofDonnellan’sdistinction.Thedistinctionisamplyillustratedbytheoriginal“Smith’smurderer”exampleandothers,evenifDonnellaniswrongaboutthemeaningsandtruth-valuesofthenear-misssentences.Donnellan’spaperraisesthequestionofspecifyingthecircum-stancesunderwhichonesucceedsinreferring,byusingadescription,tothepersonorthingoneintendstoreferto,andhehasshownthatthisdoesnotalwaysgobysemanticreferent.Further,thedistinctionunmistakablymat-terstothetruth-valueofsentencesthatembeddescriptionswithinclausesofcertainkinds.SupposeIweretosay:(13)Iknowthat’srightbecauseIhearditfromthetowndoctor.Youmighthavetoaskme,“Youmeanbecauseshe’sadoctorandthisisamedicalmatter,ordoyoumeanbecauseyouhearditfromherandshe’salsoanauthorityontruecrime?”(13)’struth-valuemaydependonwhether“thetowndoctor”isusedattributivelyorreferentially.Orconsider:(14)Iwishthatherhusbandweren’therhusband.Themostnaturalreadingof(14)istotakethefirstoccurrenceofthedescriptionreferentiallybutthesecondattributively;whatthespeakerwishesisthatthemaninquestionwerenotmarriedtothewomaninquestion.But(14)hasseveralotherreadings,dependingonwhichwaythedescriptionsaretaken,eventhoughtheyarefairlysilly.InlightofKripke’sdistinctionbetweenspeaker-orutterer’sreferentandsemanticreferent,onemightbetemptedsimplytowriteoffDonnellan’sissueasverbal,andmaintainthattheTheoryofDescriptionsisstillcorrectasanaccountofthetruth-valuesofsentencestakenliterally,whileDonnellanisoftenrightaboutspeaker-referentandspeaker-meaning.Buttheambiguityofsentenceslike(13)and(14)stillseemstoeludeRussell’sanalysis.10Also,evenifoneispersuadedbyKripkeandhasdiscountedthenear-missexamples,itremainscontroversialwhether,forthereferentialcase,theactualreferentisalwaysthespeaker-referent.Noticethatthisquestionpresupposesathirdnotion,thatof“actual”referent,thatisconceptuallydistinctfromtheothertwo.Whatismeantseemstobethattheactualreferentistheobjectaboutwhichthespeakeractuallysucceededinmakingastatement(askingaquestion,issuingacommand,andsoon),itbeingleftopenwhetherthistrackstheliteralsemanticinterpretationofthesentenceuttered.(Ofcourse,iftheTheoryofDescriptionsiscorrect,eithertheactualreferentisalwaysthesemanticreferentor,sinceaccordingtoRusselldefinitedescriptionsdonotreallyreferatall,thereisnoactualreferent.)\nDefinitedescriptions27MacKay(1968)arguesthatinsomecases,evenifonemisspeaks,one’sactualreferentmaybethesemanticreferentratherthanthespeaker-referent.Supposetherearearockandabookonthetable,and,wantingyoutobringmethebook,Ifumblemouthandsay,“Bringmetherockonthetable,”using“therock”referentiallyandspeaker-referringtothebook,Ihavestillaskedyoutobringmetherock,andyouwouldnotbecomplyingifyoubroughtmethebookinstead.OrsupposeIsaytoyou,“Ibetyou$5thatthegloriouswinner[ofabigautorace]isoverfortyyearsofage.”Iamusing“thegloriouswinner”referentially,thinkingofDaleEarnhart,completelyconfidentthathehastheracewon,andIhavehimverymuchinmind,clearmentalimageandall.But,althoughhecrossedthefinishlinefirst,Earnhartdoesnotinfactwin;byalittle-knowntechnicalityhecomesinsecondtoFatFreddyPhreak,whohasgotlooseagainandenteredtheraceatthelastminute.FatFreddyisonlytwenty-two.Ioweyou$5.MacKaymakesthegeneralpointthataspeaker’sintentionsmaybearbi-trarilycrazy.SupposeIhaveformedtheinsanebeliefthatKeithDonnellanistheillegitimatesonofSantaClausandMargaretThatcher.Usingthedescrip-tionreferentially,Isay“MrsThatcher’sChristmasbastardwroteaclassicarticleondescriptions.”Ifyouknowenoughaboutmyweirdbeliefs,youwillpickouttherightindividualandunderstandwhatImeant,butnoonecouldcorrectlydescribemeashavingsaidthatKeithDonnellanwrotetheclassicpaper.Itshouldbequestionedwhetherthereisanyrightfullyseparatenotionofan“actualreferent.”Theconceptofasemanticreferentisclear,andthetheoryofcommunicationrequiresthatofaspeaker-referent,butperhapstheideaofan“actualreferent”isjustaconfusionofthetwobasedonourfailuretoseethedifferencebetweenliteralsentencesemanticsandthetheoryofcommunication.Thenwewouldhavetoexplainawaythefactofourhavingintuitionsabout“actualreferents”incaseslikesomeoftheforegoing.Kripketakesroughlythatline,makinguseofanideaofGrice’sthatweshalldiscussinchapter13.ANAPHORAOnefinalobjectiontotheTheoryofDescriptionsmustbementioned.Aswehavenoted,Russelldealsonlywithwhatheconsidersthecentraluseof“the,”andexemptsthetheoryfromhavingtoexplainpluralusesorthegenericuse.Onemaythinkthatatheoryofdefinitedescriptionsoughtnottobepam-peredthusforever.ButinanycaseRusselldoesnotmentionanaphoricuses,andithascertainlybeenwonderedwhythetheoryshouldnotberequiredtocoverthose,sinceunlikepluralandgenericuses,anaphoricdescriptionsareostensiblysingularreferringexpressions.Ingeneral,ananaphoricexpressioninheritsitsmeaningfromanotherexpression,itsantecedent,usuallythoughnotalwaysoccurringearlierinthesentenceorinaprevioussentence.Forexample,in\n28Referenceandreferring(15)Themanwholivedaroundthecornerwaseccentric.Heusedtosnackonturtleheads“he”refersbacktothemanwholivedaroundthecorner.Geach(1962)calledsuchaterma“pronounoflaziness”andsuggestedthatitmerelyabbreviatesaboilerplaterepetitionoftheantecedentphrase,sothat(15)’ssecondclauseispreciselyequivalentto“Themanwholivedaroundthecornerusedtosnackonturtleheads.”Geach’ssuggestionisonlyoneamongseveraltheoriesofanaphoricpronouns,butthegeneralideaisthatthepronounhasthereferentthatitdoesonlyinvirtueofitsrelationtotheantecedentphrase.IfGeachisright,then(15)posesnoproblemfortheTheoryofDescriptions;itssecondclausewouldbeanalyzedintheusualmannerandthatanalysisseemsatleastascorrectasothercentralRussellianparaphrases.But,asEvans(1977)pointsout,aparalleltreatmentfailswhentheantecedentisaquanti-fierphraseoranindefinitedescription:(16)Justoneturtlecamedownthestreet.Itwasrunningasifitwerebeingpursuedbyamaniac.(17)Arabbitappearedinouryardafterdinner.Itseemedunconcerned.(16)’ssecondclauseisnotequivalentto“Justoneturtlewasrunningasifitwerebeingpursued...,”becausethelattermightbefalseevenwhen(16)istrue(ourownpetturtle,inthediningroomwithus,mayhavebeenrunningaswell).(17)’ssecondclauseisnotequivalentto“Arabbitseemeduncon-cerned,”forthatparaphrasemissesthefactthattheoriginal“It”referredtotheparticularrabbitthatappearedintheyard.Russellmayfairlyrejointhatwhatheofferedwasatheoryofdefinitedescriptions,andneither(16)nor(17)containsadefinitedescription.Butifthe“It”sin(16)and(17)arenotpronounsoflaziness,whyshouldwethinkthat“He”in(15)isone?Also,definitedescriptionscanthemselvesbeanaphors:(18)Justoneturtlecamedownthestreet.Theturtlewasrunningasifitwerebeingpursuedbyamaniac.(19)Arabbitappearedinouryardafterdinner.Therabbitseemedunconcerned.Itisplausibleenoughtotake“Theturtle”in(18)asabbreviating“Theturtlethatcamedownthestreet,”inwhichcase(18)doesnotthreatenRussell’sanalysis.Butthesamewillnotdofor(19):ifwetrytosupposethat“Therabbit”abbreviates“Therabbitthatappearedinouryardafterdinner,”thenbytheusualuniquenessclause,(19)wouldentailthatatmostonerabbit\nDefinitedescriptions29appearedintheyard,and—notice—(19)itselfdoesnotentailthat,but(sinceitsopeningphraseisonly“Arabbit”)islogicallyconsistentwithmorethanonerabbit’shavingappearedintheyard.True,aspeakerwhoutters(19)doessomehowsuggestthattherewasjustone.Butnoticethatitwouldnotbecontradictorytoutter(19)andthenadd,“Infact,therewereseveralrabbits,andnoneofthemlookedveryworried.”Neale(1990)hastriedtoaccommodateanaphorawithinaconservativeRusselliantheory;Heim(1990),KampandReyle(1993)andothershavearguedthatabroadersemanticformatisrequired.ButIshallleavethetopicatthispoint.Afewfurtherissueshaveariseninrecentyears.Forexample,ithasbeenquestionedwhethertheuseofadefinitedescriptiondoesreallyentailevencontextualuniqueness(seeSzabó2000,2003;Abbott2003).Moregenerally,therelationsbetweendefiniteandindefinitedescriptionshavecomeunderscrutiny(Szabó2000again;LudlowandSegal2004.)PluraldescriptionshavebeeninvestigatedbySharvy(1980),Neale(1990)andBrogaard(2007)amongothers.Summary•Singulartermsrefertoindividualobjectsintheworld.Buttosupposethatthatisalltheydoleadstologicalpuzzles.•Russellarguedthatsentencescontainingdefinitedescriptionsshouldbeanalyzedastriplesofgeneralstatements.•RusselldefendedthisTheoryofDescriptionsbothdirectlyandbyappealtoitsaffordingsolutionstothefourlogicalpuzzles.•StrawsonarguesthatRussellviewssentencesandtheirlogicalproper-tiestooabstractlyandignorestheirstandardconversationalusesbyrealpeopleinreallife.•Inparticular,Russellmissesthefactthatsentencescontainingnon-denotingdescriptionsarenotregardedasfalse,butlacktruth-valuealtogetheronaccountofpresuppositionfailure.Also,Russellignorescontext-bounddescriptions.•Donnellancallsattentiontothereferentialuseofdescriptions,alsoignoredbyRussell,andtries,notaltogethersuccessfully,todistin-guishitfromtheattributiveuse.•ItisnotobviousthatRussell’stheorycanaccommodatealltheana-phoricusesofdescriptions.Questions1Given(forthesakeofargument)thattheTheoryofDescriptionsisotherwiseplausible,areyouconvincedbyRussell’ssolutionstothefourpuzzles?\n30Referenceandreferring2IsStrawson’scritiquemorepersuasivethanIhavegranted?DevelopitabitfurtheragainstRussell.3Towhatextentdoesthetheorycorrectlypredictandexplaintheentireuseof“the”inEnglish?4WhatdoyoumakeofDonnellan’sdistinction?Canitberenderedmoreprecise?TrytorefinetheintuitivecontrastwithwhichDonnellanbegins.5DisputeordefendanyofDonnellan’sinterestingintuitivejudgementsabout“actualreferents”inparticularhypotheticalspeechsituations.Thencommentonthesignificance,forDonnellan’sprogram,ofyourownpositiononsuchacase.6DonnellanregardshisarticleasacontributiontotheRussell–Strawsondispute.Buthedoesnotsaymuchinthearticleaboutthefourpuzzleswithwhichthatwholeissuebegan.Doeshistheory,asyouconstrueit,solveanyorallofthefourpuzzles?7CanyouhelpRussellextendhistheorytocoverourexamplesofana-phora?Arethereotheranaphoricexamplesthatcreatefurtherproblemsforhim?Furtherreading•Kaplan(1972)isanexcellentdetailedexpositionoftheTheoryofDescriptions.SeealsoCartwright(1987)andNeale(1990).Thejour-nalMindpublishedanentirespecialissue(Vol.144,October2005)inhonorof“OnDenoting”’sone-hundredthbirthday.•Russell(1957)repliedtoStrawson’scritique.•Linsky(1967)surveystheRussell–Strawsondisputewell.•DespiteRussell’scontemptforit,Meinong’sviewhasbeenstoutlydefendedbyRoutleyetal.(1980)andbyParsons(1980).•Donnellan(1968)repliedtoMacKay.Donnellan(1979)isamoreextendedtreatmentandalsotakesupquestionsofanaphora.•Taylor(1998:ch.2)offersafullerbutstillaccessiblesurveyofana-phoricphenomena.•Ostertag(1998)containsmanyimportantpapersondefinitedescrip-tions,asdoesBezuidenhoutandReimer(2004).SeealsoLudlow(2007),afinesurveyarticle.\n3Propernames:theDescriptionTheoryOverviewRussellseemstohaverefutedtheReferentialTheoryofMeaningfordefinitedescriptions,byshowingthatdescriptionsarenotgenuinelysingularterms.Perhapsthatisnotsosurprising,sincedescriptionsarecomplexexpressionsinthattheyhaveindependentlymeaningfulparts.Butonemightnaturallycontinuetothinkthatordinarypropernamesaregenuinelysingularterms.Yetthefourpuzzles—aboutnonexistents,negativeexistentials,andtherest—arisejustasinsistentlyforpropernamesastheydidfordescriptions.Fregeofferedsolutionstothepuzzlesbyproposingthatanamehasasenseinadditiontoitsreferent,thesensebeinga“wayofpresenting”theterm’sreferent.Buthesaidfartoolittleaboutwhat“senses”areandhowtheyactu-allywork.Russellsolvedthisproblembyarguing,fairlypersuasively,thatordinarypropernamesarereallydisguiseddefinitedescriptions.ThishypothesisallowedhimtosolvethefourpuzzlesforpropernamesbyextendinghisTheoryofDescriptionstothem.YetRussell’sclaimthatpropernamesaresemanticallyequivalenttodescriptionsfacesseriousobjections:forexample,thatitishardtofindaspecificdescriptiontowhichagivennameisequivalent,andthatpeopleforwhomthesamenameexpressesdifferentdescriptionswouldbetalkingpasteachotherwhentheytriedtodiscussthesamepersonorthing.JohnSearleproposedalooser,“cluster”descriptiontheoryofpropernamesthatavoidstheinitialobjectionstoRussell’sview.ButSaulKripkeandothershaveamassedfurtherobjectionsthatapplyasmuchtoSearle’sloosertheoryastoRussell’s.FregeandthepuzzlesWemayhaveagreedwithRussellthattheReferentialTheoryofMeaningisfalseofdescriptionsbecausedescriptionsarenotreally(logically)singularterms,butwemaycontinuetoholdtheReferentialTheoryforpropernamesthemselves.Surelynamesarejustnames;theyhavetheirmeaningssimplybydesignatingtheparticularthingstheydesignate,andintroducingthosedesignataintodiscourse.(LetuscallsuchanexpressionaMillianname,since\n32ReferenceandreferringJohnStuartMill(1843/1973)seemedtodefendtheviewthatpropernamesaremerelylabelsforindividualpersonsorobjectsandcontributenomorethanthoseindividualsthemselvestothemeaningsofsentencesinwhichtheyoccur.)ButrecallourinitialobjectiontoRussell’sTheoryofDescriptions:that,althoughitwasmotivatedentirelybythefourpuzzles,thepuzzlesarenotatallspecifictodefinitedescriptions,becausetheyarosejustasinsis-tently—nottomentionfirst—forpropernamesaswell.FregeprecededRussellinofferingsolutionstothepuzzles.WehavealreadyseenwhathesaidaboutApparentReferencetoNonexistents:(1)JamesMoriartyisbaldismeaningfulbecausethename“JamesMoriarty”hasa“sense”overandaboveitsputativereferent,evenifthereisinfactnoreferent.Infact,nothingisreferredtoordenotedbythename,butthesenseis“expressed”byit.ForFrege,the“sense”was,roughly,aparticular“wayofpresenting”theterm’sputativereferent.Thoughitselfanabstractentityratherthanamen-talorpsychologicalone,thesensereflectsaperson’sconceptionorwayofthinkingofthereferent.Fregesometimesexpressedsensesintheformofdefinitedescriptions;forexample,thesenseofthename“Aristotle”mightbe“Plato’sdiscipleandtheteacherofAlexandertheGreat,”or“theStagiriteteacherofAlexander”(Frege1892/1952b:58n).Asensedeterminesauniquereferent,butmultiplesensesmaydeterminethesamereferent.LetusnowseehowFregeattackedtheotherthreepuzzles.NEGATIvEExISTENTIALS(2)Pegasusneverexisted.Asbefore,(2)seemstobetrueandseemstobeaboutPegasus,butif(2)istrue,(2)cannotbeaboutPegasus....NoticethatthereisaworsecomplicationherethanisraisedbytheProblemofApparentReferencetoNonexistentsalone:whereas(1)ismeaningfuldespitethenonexistenceofJamesMoriarty,(2)isnotonlymeaningfuldespitePegasus’nonexistencebutactuallyandimportantlytrue.TheideaofsensesasparticularmodesofpresentationaffordsFregeatleastanimpressionisticsolutiontotheProblemofNegativeExistentials(thoughwhetherthiswasactuallyhisviewandhowitmightbemadepreciseareunclear):(2)canbetakentomeanroughlythatthesenseof“Pegasus,”theconceptionofawingedhorseriddenbyBellerophon,failstofindareferent—notevena“nonexistent”one.Nothinginrealityanswerstothatsense.1ThereasonthisideaisnotstraightforwardisthatforFregeanameonly“expresses”anddoesnotdenoteitsownsense.So(2)isnotliterallyaboutthesenseof“Pegasus,”anddoesnotout-and-outsayofthatsensethatitlacksa\nPropernames:theDescriptionTheory33referent,eventhoughthelatterisonethingwephilosophersknowwhenweknowthat(2)istrue.FREGE’SPuzzLE(3)MarkTwainisSamuelLanghorneClemenscontainstwopropernames,bothofwhichpickoutordenotethesameper-sonorthing,andso—ifthenamesareMillian—shouldbetriviallytrue.Yetasbefore,(3)seemsbothinformativeandcontingent.(Afictionalexampleis“SupermanisClarkKent”;accordingtoMr.JerrySiegel’scomic-booksaga,dilettantemillionairesspenttimeandmoneytryingtodiscoverSuperman’ssecretidentity.)OnFrege’sview,althoughthetwonamesin(3)pickoutacommonrefer-ent,they“present”thatindividualindifferentways;theyhaveimportantlydifferentsenses.Andwhathecalls“cognitivesignificance”goeswithsense,ratherthanwithreference.Thisiswhathewrites:Whenwefound‘a=a’and‘a=b’tohavedifferentcognitivevalues,theexplanationisthatforthepurposeofknowledge,thesenseofthesentence,viz.,thethoughtexpressedbyit,isnolessrelevantthanitsreference...Ifnowa=b,thenindeedthereferenceof‘b’isthesameasthatof‘a,’andhencethetruth-valueof‘a=b’isthesameasthatof‘a=a.’Inspiteofthis,thesenseof‘b’maydifferfromthatof‘a’,andtherebythethoughtexpressedin‘a=b’differsfromthatof‘a=a.’Inthatcasethetwosentencesdonothavethesamecognitivevalue.(1892/1952b:78).(Butwearenottoldhowitisthat“a=b”canbecontingent.)SuBSTITuTIvITY(4)AlbertbelievesthatSamuelLanghorneClemenswaslessthan5feettall.Butsubstituting“MarkTwain”for“SamuelLanghorneClemens”in(4)producesafalsehood;asinthepreviouschapter,thesingular-termpositiongovernedby“believesthat”isreferentiallyopaque.IfthenameswereMillian,andcontributednothingtomeaningbesidestheintroductionoftheirrefer-entsintodiscourse,thesubstitutionshouldmakenodifferenceatallandthepositionwouldbetransparent.HereFregemakesaningeniousmove.Theproblem,werecall,wasthattheopacitywasinducedbythe“believesthat”construction,sincewhatfol-lowsitisnotitselfopaque.Sincebeliefisacognitivematter,Fregesupposed\n34Referenceandreferringthatwhatdetermineabeliefsentence’struth-valuearethesensesratherthanmerelythereferentsoftheexpressionsthatfollowthebeliefoperator.Hethereforesuggeststhatwhattheoperatordoesistoshiftthereferenceofthenameinparticular:Inside“believesthat,”thenamerefersnot,asusually,toClemens/Twaintheperson,buttoitsownsense.Thatiswhytheresultofsubstituting“MarkTwain”into(4)hasadifferenttruth-value:Inthebeliefcontext,“MarkTwain”referstoitssense,adifferentonefromthatof“SamuelLanghorneClemens.”ThusFrege’sdistinctionbetweenreferenceand“sense”enableshimtoaddresseachofthepuzzles.Andhissolutionssoundright,sofarastheygo:Namescontributemeaningofsomesortoverandabovetheirreferents,andthatiswhatmakesthedifferenceswhereweseedifferences,ofcourse.Butthesolutionssoundright,Isuspect,becauseoftheirschematicnature.Fregecallstheaddedmeaning“sense,”butsayslittlemoreaboutit(likewiseabout“expressing”asopposedtodenoting,“cognitivesignificance,”andthelike).Inparticular,hedoesnotsaywhatsortofmeaningitisorwhatpositivecontributionitmakes.Thishasanairoflabelingratherthansolvingtheproblem.(ButweshallconsideramuchmoresubstantiveandtestableimplementationofFrege’sviewinchapter10.)PerhapswecouldpickuponFrege’sfurtherhintthatnamescanhavethesensesofdescriptions.ThatisjustwhatRusselldid,anditledhimtoaveryrichapproachtothepuzzles.2RuSSELL’SNAMECLAIMRussell’sresponseisbothbrilliantandstronglydefended.Heturnsaroundandoffersanewthesis,whichIwillcalltheNameClaim.Theclaimisthateverydaypropernamesarenotreallynames,atleastnotgenuineMilliannames.Theylooklikenamesandtheysoundlikenameswhenwesaythemoutloud,buttheyarenotnamesattheleveloflogicalform,whereexpres-sions’logicalpropertiesarelaidbare.Infact,Russellmaintains,theyareequivalenttodefinitedescriptions.Indeedhesaysthey“abbreviate”descrip-tions,andheseemstomeanthatfairlyliterally.ThusRussellintroducesasecondsemanticappearance–realitydistinction;justasdefinitedescriptionsaresingulartermsonlyinthesenseofsurfacegrammar,thesame—moresurprisingly—istrueofordinarypropernamesthemselves.Here,ofcourse,thedifferenceismoredramatic.Ifyoulookatadefinitedescriptionwithoutreferentialistbias,youcanseethatithasgotsomeconceptualstructuretoit,intheformofindependentlymeaningfulwordsoccurringinitthatseemtocontributetoitsownoverallmeaning.Soitisnottoobigasurprisetobetoldthatunderlyingthemisleadinglysimpleappearanceoftheword“the,”thereisquantificationalmaterial.Butnowwearetoldthesameaboutakindofexpressionthatlooksconceptuallysimple.IftheNameClaimistrue,thenRussell’ssolutiontothefourpuzzlesdoesgeneralizeafterall—becausewejustreplacethenamesbythedefinite\nPropernames:theDescriptionTheory35descriptionstheyexpressandthenproceedasinchapter2;theRusselliansolutionsapplyjustasbefore(whetherornotwethinktheyaregoodonesinthefirstplace).ThusnamesdohavewhatFregethoughtofas“senses,”thatcandifferdespitesamenessofreferent,butRussellgivesananalysisoftheseratherthantakingthemasprimitiveitemsofsomeabstractsort.ItisimportanttoseethattheNameClaimisentirelyindependentoftheTheoryofDescriptionsitself.(Peopleoftenusethephrase“Russell’stheoryofdescriptions”aslumpingtogetheranumberofdifferentthingsRussellbelieved,includingtheNameClaim.)Butonemightaccepteitherdoctrinewhilerejectingtheother:sometheoristsholdtheTheoryofDescriptionsasatheoryofdefinitedescriptionsthemselves,whilerejectingtheNameClaimentirely;lesscommonly,onecouldembracetheNameClaimbutholdatheoryofdescriptionsdifferentfromRussell’s.InsupportoftheTheoryofDescriptions,Russellgaveadirectargument;thenhetoutedthetheory’spowerinsolvingpuzzles.HemakesasimilarexplanatorycasefortheNameClaim,inthattheclaimlendshistheoryofpropernamesthesamepowertosolvepuzzles—puzzlesthatlookedconsider-ablynastierfornamesthantheydidfordescriptions.Buthealsogivesatleastonedirectargument,andasecondiseasilyextractedfromhiswritings.First,recallRussell’sdirectdefenseofhistheoryofdefinitedescriptions:Hemaintainsthatasentencecontainingadefinitedescriptiondoesintui-tivelyentaileachofthethreeclausesthatmakeuphisanalysisofit,andthethreeclausesjointlyentailthesentence.Henowarguesthatthesameistrueofsentencescontainingpropernames.Takeoneofthetoughestcasesofall,anegativeexistential.(2)(“Pegasusneverexisted”)isactuallytrue.What,then,coulditmean?Itdoesnotpickoutanexistingthingandassertfalselythatthethingisnonexistent;nordoesitpickoutaMeinongianentityanddenyexistenceofit.Itmerelyassuresusthatinfacttherewasnosuchwingedhorse.Similarly,“SherlockHolmesneverexisted”meansthatthereneveractuallywasalegendaryEnglishdetec-tivewholivedat221BBakerStreetandsoon.Thisisveryplausible.Theseconddirectargument(nevergivenexplicitlysofarasIknow)callsattentiontoakindofclarificatoryquestion.Supposeyouhearsomeoneusinganame,say“LiliBoulanger,”andyoudonotknowwhothespeakeristalkingabout.Youaskwhothatis.Thespeakerreplies,“Oh,thefirstwomanevertohavewonthePrixdeRome,in1913,withhercantataFaustetHélène”;andthatisaproperanswer.Youaskedbecause,sotospeak,youdidnotunder-standthenameyouheard.Inordertocometounderstandit,youhadtoaska“who”question,andtheanswerhadtobeadescription.(MerelygivingasecondpropernameofBoulangerwouldnothavedonethetrick,unlessyouhadpreviouslyassociatedthatnamewithadescription.)Orwecoulduse“who”questionsasakindoftesting,whichmightbecalledthe“spot-checktest.”Supposeyouusedthename“WilfridSellars,”andIwhiparoundandsay“Who’sthat?”Allyoucanreply,allthatcomesout,is“Um,thefamousphilosopheratPittsburghwhowrotethosereally\n36Referenceandreferringdensebooks”orthelike.Ingeneral,whenasked“Who[orwhat]doyoumean?”afteronehasjustusedaname,oneimmediatelyandinstinctivelycomesupwithadescription,asanexplanationofwhatonemeant.JohnSearle(1958)madeasimilarappealtolearningandteaching:howdoyouteachanewpropernametoachild,andhowdoyoulearnthereferentofaparticularnamefromsomeoneelse?Inthefirstcase,youproduceoneormoredescriptions;inthelatter,youelicitthem.Theseareveryrobustphenomena;sotheNameClaimisnotjustades-peratelungemadeinordertosolvetheproper-nameversionsofthefourpuzzles.Russellspeaksaggressivelyofnames’“abbreviating”descriptions,asiftheyweremerelyshortforthedescriptionsas“theU.S.A.”isshortfor“theUnitedStatesofAmerica.”Thatistoostrong.AllRussellactuallyneedsforhisanalyticalpurposesistheweakercontentionthatnamesaresomehowequivalentinmeaningtodescriptions(letuscallthatweakerthesistheDescriptionTheoryofpropernames).YeteventhelessambitiousDescriptionTheoryhassincecomeinforseverecriticism.OpeningobjectionsOBJECTION1Searle(1958)complainedthat,ifpropernamesareequivalenttodescrip-tions,thenforeachnametheremustbesomeparticulardescriptionthatitisequivalentto.Forexample,ifIunreflectivelymuse,(5)WilfridSellarswasanhonestman,whatamIsaying,giventhatIknowafairnumberofindividuatingfactsaboutSellars?Searletriesoutacoupleofcandidatedescriptiontypes,andfindsthemwanting.Wemightsupposethat“WilfridSellars”isformeequivalentto“TheoneandonlythingxsuchthatxisFandxisGand...,”whereF,G,andtherestareallthepredicatesthatIwouldapply(orbelievetrulyappli-cable)tothemaninquestion.Butthiswouldhavethenastyconsequencethat(5)asIuseitentails(6)ThereisatleastonephilosopherwithwhomIhadafairlyviolentargumentinGeorgePappas’livingroomin1979—and(5)surelydoesnotentail(6),formeorforanyoneelse.Now,thespot-checktestoughttosupplyamorelocalanswerforeachuseofaname,andaswehaveseen,itisplausibletothinkthataspeakercannormallycoughupafairlyspecificdescriptionwhenprodded.Butitisunclearthatthisisalwaysbecausethedescriptionwasonethespeaker\nPropernames:theDescriptionTheory37alreadyhaddeterminatelyinmind.Ifyouaskme,“WhoisSellars?,”Imightmakeanyofanumberofanswersthatcometomind,dependingonwhatsortofinformationIthinkyoumaywantabouthim.IthardlyfollowsthattheanswerIdoproduceistheprecisedescriptionthatmyuseof“Sellars”antecedentlyexpressed.Notice:Thecomplaintisnotmerelythatitwouldbehardtofindoutwhichdescriptionaspeaker“hadinmind”inutteringsomename.Thestrongerthesisisthatatleastinmanycasesthereisnosingledeterminatedescrip-tionthatthespeaker“hasinmind,”eitherconsciouslyorsubconsciously.Iseelittlereason(independentofthesemanticalpuzzles)forthinkingthatthereisafactofthematterastowhether“WilfridSellars”isusedasequivalentto“Theauthorof‘PhilosophyandtheScientificImageofMan’,”or“Pittsburgh’smostfamousphilosopher,”or“Theinventorofthe‘Theory’theoryofmentalterms,”or“ThemanonwhosepaperIhadtocommentattheTenthChapelHillColloquiumin1976,”notforgetting“ThevisitingphilosopherwithwhomIhadafairlyviolentargumentinGeorgePappas’livingroomin1979.”Ineedhavehadnoneoftheseinparticular(eventacitly)inmindwhenIunreflectivelyuttered(5).OBJECTION2Undeniably,differentpeopleknowdifferentthingsaboutotherpeople.InsomecasesX’sknowledgeaboutZandY’sknowledgeaboutZmaynotevenoverlap.Assumingthatthedescriptionswithwhichnamesaresupposedtobesynonymousareinspeakers’mindsasrevealedbythespot-checktest,itfollowsfromtheNameClaimthatthesamenamewillhave(many)differentsensesfordifferentpeople;everynameismultiplyandunfathomablyambigu-ous.For,ifnamesareequivalenttodefinitedescriptions,theyareequivalenttodifferentdefinitedescriptionsindifferentpeople’smouths,andforthatmattertodifferentdescriptionsinthesameperson’smouthatdifferenttimes,bothbecauseone’sknowledgekeepsfluctuatingandbecausewhatispsycho-logicallyprominentaboutonepersonforanotherkeepsfluctuatingtoo.Andthingsgetworse.SupposethatIamthinkingofWilfridSellarsas“theauthorof‘PhilosophyandtheScientificImageofMan’,”andsupposeyouarethinkingofSellarsas“Pittsburgh’smostfamousphilosopher.”ThenwewouldbecuriouslyunabletodisagreeaboutSellars.IfIweretosay,“Sellarsusedtotiehisshoeswithonehand,”andyousaid“That’sridiculous,Sellarsdidnosuchthing,”wewould(onRussell’sview)notbecontradictingoneanother.ForthesentenceIhadutteredwouldbeageneralization:(7)Oneandonlyonepersonwrote“PhilosophyandtheScientificImageofMan”,andwhoeverwrote“PhilosophyandtheScientificImageofMan”usedtotiehisshoeswithonehandwhileyourswouldbejustadifferentgeneralization:\n38Referenceandreferring(8)OneandonlyonepersonwasaphilosophermorefamousthananyotherinPittsburgh,andwhoeverwasaphilosophermorefamousthananyotherinPittsburghdidnosuchthingastiehisshoeswithonehand.Andthetwostatementswouldbeentirelycompatiblefromalogicalpointofview.Whatlookedlikeaspiriteddispute,vergingonfistfight,isnorealdisputeatall;wearemerelytalkingpasteachanother.Butthatseemsquitewrong.3Searle’sClusterTheoryInlightofthesetwoobjections(andseveralothers)toRussell’sversionoftheDescriptionTheory,JohnSearleofferedalooserandmoresophisticatedvariant.Hesuggestedthatanameisassociated,notwithanyparticulardescription,butwithavagueclusterofdescriptions.Asheputsit,theforceof“ThisisN,”whereNisreplacedbyapropername,istoassertthatasufficientbutsofarunspecifiednumberof“standardidentifyingstatements”associ-atedwiththenamearetrueoftheobjectdemonstratedby“this”;thatis,thenamereferstowhateverobjectsatisfiesasufficientbutvagueandunspecifiednumber(SBVAUN)ofthedescriptionsgenerallyassociatedwithit.(SearleaddsthemetaphysicalclaimthattobethepersonNistohaveaSBVAUNoftherelevantproperties.)Thevaguenessisimportant;Searlesaysitispreciselywhatdistinguishesnamesfromdescriptions,andinfactiswhywehaveandusenamesasopposedtodescriptions.Noticethat,iftheNameClaimwerecorrect,thenpropernames’onlyfunctionwouldbetosavebreathorink;theywouldbejustshorthand.Searleinsiststhat,ratherthanbeingequivalenttoasingledescription,anamefunctionsasa“peg...onwhichtohangdescriptions”(1958:172),andthatiswhatenablesustogetalinguistichandleontheworldinthefirstplace.Wewouldneedtomakesomerefinements.Forexample,ifoneisaSearleanitseemsnaturaltorequirethata“sufficientnumber”beatleastoverhalf—otherwisetwoobviouslydistinctindividualscouldbothbethename’sreferent.Also,wewouldsurelywanttosaythatsomeofaperson’sidentifyingpropertiesaremoreimportantthanothersindetermininghisorheridentity;somewayofweightingtheidentifyingdescriptionsisinvolved.ThisClusterTheoryallowsSearletoavoidthetwoobjectionswehaveraisedforRussell’sview.Objection1ismootedbecauseSearlehasaban-donedthecommitmentthatforeachnametheremustbesomeoneparticulardescriptionthatitexpresses.Thenameistiedsemanticallyjusttoalooseclusterofdescriptions.Objection2isblunted(Searlebelieves)bythefactthatdifferentpeoplecanhavedifferentsubclustersofdescriptivematerialinmind,yeteachhaveaSBVAUNofidentifyingdescriptionsandtherebysuc-ceedinreferringtothesameindividual.4\nPropernames:theDescriptionTheory39ThusSearletriedtomitigatetheopeningobjectionstoRussell’stheorybyofferinghislooserclusterversionofthedescriptionapproach.ThisversionseemstoqualifyasasensiblemiddlewaybetweenRussell’sviewandtheMillianconceptionofnamesapparentlydiscreditedbythefourpuzzles.But,buildingonsomeimportantideasofRuthBarcanMarcus(1960,1961),SaulKripke(1972/1980)wentontosubjectRussell’sNameClaimandSearle’sClusterTheorytogethertoamoresustainedcritique.HearguedthatSearlehadnotbackedfarenoughawayfromRussell,forSearle’sviewinheritsproblemsofmuchthesamekinds;rather,thewholeDescriptivistpictureofpropernamesismisguided.Thetheoryofreferencehasneverbeenthesame.Kripke’scritiqueOBJECTION3Supposethat“RichardNixon”isequivalentto“thewinnerofthe1968U.S.Presidentialelection.”Andnowconsideraquestionaboutpossibil-ity.(Questionsaboutpossibilityandnecessityarecalledmodalquestions;moreabouttheseinthenextchapter.)CouldRichardNixonhavelostthe1968election?Theanswerseemsunequivocallytobe“Yes,”assumingthat“could”hereexpressesmerelytheoretical,logical,ormetaphysicalpossibilityratherthansomethingaboutthestateofourknowledge.ButaccordingtotheDescriptionTheory,ourquestionmeansthesameas(9)Isitpossiblethat:oneandonlyonepersonwonthe1968electionandwhoeverwonthe1968electionlostthe1968election?theanswertowhichisclearly“No.”Searle’sClusterTheorymayseemtoofferanimprovement,becauseitispossiblethatapersonwhosatisfiesaSBVAUNofthedescriptionclusterassociatedwith“RichardNixon”nonethelessdoesnotsatisfytheparticulardescription“winnerofthe1968election.”But,Kripkepointsout,humanpossibilityextendsfurtherthanthat:Nixontheindividualpersonmightnothavedoneanyofthethingsgenerallyassociatedwithhim.Hemighthaveapprenticedhimselfatagetwelvetoasandalmakerandgoneontomakesandalsallhislife,nevergoinganywherenearpoliticsorpubliclifeatallandneveroncegettinghisnameinanynewspaper.Yet,obviously,itisnotpossiblethatapersonwhosatisfiesaSBVAUNofthedescriptionclusterassociatedwith“RichardNixon”nonethelessdoesnotsatisfyanyatallofthedescriptionsinthatcluster.OnSearle’sview,thecharacterwhowentintosandalmakingwouldnothavebeenthereferentof“RichardNixon”andforthatmatterwouldnothavebeenRichardNixon.Andthatseemswrong.MichaelDummett(1973)hasprotestedthatobjection3issimplyinvalidasitstands;atleast,itrestsonahiddenfalseassumption.Wemayinferthatourmodalquestionissynonymouswith(9)onlybyassumingthat,if“Richard\n40ReferenceandreferringNixon”isequivalenttoadescriptionatall,itisequivalenttoonethathasnar-rowscope;intheterminologyofchapter2,thatisa“secondary”occurrencewithrespectto“Itispossiblethat.”Whatiftherelevantdescriptionhaswidescope?Thenouroriginalquestionissynonymous,notwith(9),butwith(10)Oneandonlyonepersonwonthe1968election,and,concerningwhoeverwonthe1968election,isitpossiblethatthatpersonlost?(10)isclumsy;also,thereareother,irrelevantdisambiguationsofourquestionduetothefactthattheinterrogativeoperatoritselfhasscope,soletmemakethepointmoresimplyusingjusttheindicativeversionsofthetworeadings.Thesentence(11)ItispossibleforRichardNixontohavelostthe1968election,presumingthat“RichardNixon”isequivalentto“thewinnerofthe1968election,”isambiguousasbetweenthenarrow-scopereadingPossible:(∃x)(Wx&(y)(Wy→y=x)&(z)(Wz→~Wz))whichcorrespondsto(11)andisfalse(Ihaverepresented“lost”as“didnotwin”),andthewide-scopereading(∃x)(Wx&(y)(Wy→y=x)&(z)(Wz→Possible:~Wz))whichpresumablyistrue.Colloquially,(11)meansthatoneandonlyoneper-sonwontheelectionandwhoeverwonitissuchthats/hecouldhavelost.5Inasimilarbutmoresophisticatedmove,somephilosophershavefinessedobjection3by“rigidifying”thedescriptionsintermsofwhichtheyexplicatenames:Understand“RichardNixon,”notas“thewinnerofthe1968elec-tion,”butas“theactualwinnerofthe1968election.”Seethenextchapter.OBJECTION4Kripke(1972/1980:83–7)offersan(utterlyfictional!)exampleregard-ingGödel’sIncompletenessTheorem,afamousmetamathematicalresult.InKripke’sfiction,thetheoremwasprovedinthe1920sbyamannamedSchmidt,whodiedmysteriouslywithoutpublishingit.KurtGödelcamealong,appropriatedthemanuscript,andscurrilouslypublisheditunderhisownname.6Now,mostpeopleknowGödel,ifatall,asthemanwhoprovedtheIncompletenessTheorem.Yetitseemsclearthat,wheneventhosewhoknownothingelseaboutGödelutterthename“Gödel,”theydorefertoGödelratherthantotheentirelyunknownSchmidt.Forexample,whentheysay“GödelprovedtheIncompletenessTheorem,”theyarespeakingfalsely,howeverwelljustifiedtheymaybeintheirbelief.\nPropernames:theDescriptionTheory41ThisobjectiontoogoesagainstSearle’sClusterTheoryaswellasagainsttheclassicalRussellianview.SupposenooneinfactprovedtheIncompletenessTheorem;Schmidt’sallegedproofwasirreparablyflawed,orperhapstherewasnotevenanySchmidt,but“theproofsimplymaterializedbyarandomscatteringofatomsonapieceofpaper”(p.86).Hereitisevenmoreobvi-ouslytruethatmostpeople’susesof“Gödel”refertoGödelratherthantoanyoneelseatall;yetthoseusesarenotevenbackedbyanySearleancluster.OBJECTION5Considerthesentence(12)SomepeopleareunawarethatCiceroisTully.(12)isostensiblytruebut,iftheNameClaimiscorrect,(12)ishardtointer-pret,for“thereisnosinglepropositiondenotedbythe‘that’clause,thatthecommunityofnormalEnglishspeakersexpressesby‘CiceroisTully’”(Kripke1979b:245).Since“Cicero”and“Tully”areequivalenttodifferentdescriptionsfordifferentpeople,thereisnosinglefactofwhich(12)sayssomepeopleareunaware.Now,ifIassert(12),presumablyitscomplementclauseexpresseswhat“CiceroisTully”meansinmyspeech.ButsinceIknowthatCiceroisTully,Iassociatethesamesetofdescriptions(whatevertheymightbe)withbothnames.Supposethat,likemostphilosophers,Iassociateboth“Cicero”and“Tully”with“thefamousRomanoratorwhodenouncedCatilineandwhofiguresinsomefamousexamplesofQuine’s.”Then(12)isequivalentto:(13)SomepeopleareunawarethatoneandonlyonepersonwasafamousRoman...[etc.]andoneandonlyonepersonwasafamousRoman...[etc.]andwhoeverwasafamousRoman...[etc.]wasafamousRoman...[etc.].Thatmassivelyredundantsentenceisequivalentto:(14)SomepeopleareunawarethatoneandonlyonepersonwasafamousRomanoratorwhodenouncedCatilineandwhofiguresinsomefamousexamplesofQuine’s.Nodoubt(14)istrue,butsurelyitdoesnotexpresswhat(12)means,evenwhen(12)isutteredbyme.ItisfarfromobvioushowSearlemighthandleobjection5,either.\n42ReferenceandreferringOBJECTION6IftheNameClaimistrue,theneverynameis“backed”byadescriptionthatappliesuniquelytothename’sreferent.Butmostpeopleassociate“Cicero”onlywith“afamousRomanorator”orsomeotherindefinitedescription,and,say,“RichardFeynman”onlywith“aleading[then]contemporarytheo-reticalphysicist”;yetthesepeoplesucceednotonlyinusingthosenamescorrectlybutalsoinreferringtoCiceroandtoFeynmanrespectivelywhentheydoso.Moreover,twonamesofthesameperson,suchas“Cicero”and“Tully,”maywellhavethesameindefinitedescriptionasbackingand,whentheydo,noRusselliantheorycanexplaintheircontinuingfailuretosubsti-tuteinbeliefcontexts(Kripke1972/1980:80ff.,1979b:246–7).Moregenerally,itdoesnottakemuchtosucceedinreferringtoaperson.KeithDonnellan(1970)offersanexampleinwhichachildwhohasgonetobedandtosleepisawakenedbrieflybyhisparents.TheyhavewiththemTom,anoldfriendofthefamilywhoisvisitingandwantedjusttoseethechild.Theparentssay,“ThisisourfriendTom.”Tomsays,“Hello,youngster,”andtheepisodeisover;thechildhasonlybarelywoken.Inthemorning,thechildwakeswithavaguememorythatTomisaniceman.Butthechildhasnodescriptivematerialatallassociatedwiththename“Tom”;hemaynotevenrememberthatTomwasthepersonthathewassemi-awaketomeetduringthenight.Yet,Donnellanargues,thatdoesnotpreventhimfromsucceedinginreferringtoTom;thereisapersonwhoisbeingsaidtobeaniceman,anditisTom.OBJECTION7Russellemphaticallywantedhistheorytoapplytofictionalnamessuchas“Hamlet”and“SherlockHolmes”and“thefreelunch.”IftheNameClaimiscorrect,then,anysentencecontainingafictionalnameina“primary”orwide-scopepositionwillcomeoutfalse.Forexample,(15)SherlockHolmeslivedat221BBakerStreetwillcomeoutfalsebecauseitissupposedtobeequivalentto(16)Oneandonlyonepersonwas[thatis,thereexistsexactlyonepersonwhowas]afamousdetectivewho...[etc.]andwhoeverwasafamousdetectivewho...[etc.]livedat221BBakerStreet,and(16)isfalse(therehavingexisted,infact,nosuchperson).Butsomefictionalsentences,suchas(15)itselfand“HamletwasaDane,”aretruesentences,oratanyratenotfalseones.Russellwouldnothavebeenmuchswayedbythisargument,sincehehadnoinclinationtocallittrue,asopposedtomerely“make-believe-true”or\nPropernames:theDescriptionTheory43“true-in-fiction,”thatHolmeslivesatBakerStreetorwhatever.(NB:ifitweretruethatHolmeslivedinBakerStreet,thenitwouldbetrueofBakerStreet,arealplacetothisday,thatithadhadHolmeslivinginit.Also,ifsuchsentencesweretruejustinvirtueofsomeone’shavingwrittentheminpopularbooksorstories,thenitwouldbeequallytruethatHolmesexisted,Hamletexisted,andsoon,sincepeoplesaythosethingsinbooksandstoriestoo;thispointisstrangelyoverlooked.)Yetsomepeoplewanttoinsistthatfictionalsentencesareliterallytruth-valuelessratherthanfalse;ifyouaresympathetictothis,youwillwanttoholdaKripkeantheoryoffictionalnamesratherthanRussell’s(Kripke1972/1980:156–8).Donnellan(1974)defendssuchatheoryinmoredetail.KripkehasafurtherandinawaymorefundamentalobjectiontotheDescriptionTheory,butitrequiresabitoftechnicalapparatus.Thatappa-ratusisonewewillbeneedingagainanyway.Ishalldevelopitinthenextchapter.Summary•Thefourlogicalpuzzlesaboutreferencearisejustasinsistentlyforordinarypropernamesastheydidfordefinitedescriptions.•Fregeofferedsolutionsintermsofwhathecalled“senses,”butthesolutionsdonotreallyexplain.•Inresponse,RussellextendedhisTheoryofDescriptionsbydefend-ingtheNameClaim.•ButtheNameClaimfacesatleasttwopowerfulobjections.•Searleoffersalooser,“cluster”versionoftheDescriptionTheoryofnames,whichavoidstheinitialobjections.•ButKripkemarshalsahostoffurtherobjectionsthatapplytoSearle’sviewastrenchantlyastheydotoRussell’sstrictertheory.Questions1AreFrege’ssolutionstothepuzzlesreallysolutions,afterall?Whatdotheyexplain,absenttheassumptionthat“senses”taketheformofdescriptions?2SupposeyourejectRussell’sNameClaim.Howmightyouthensolvethefourpuzzles,inregardtonames?3RespondonRussell’sbehalftooneormoreofthetwoopeningobjec-tions;orcomeupwithafurtherobjection.4DoesSearle’sClusterTheoryreallyavoidobjections1and2,inwaysthatRussell’sstricterversionofDescriptivismdidnot?5CanyouthinkofanobjectiontoSearle’stheorythatdoesnotapplytoRussell’soriginaltheory?6CanRussellrebutanyofKripke’sobjections3–7?EvenifRussellcan-not,canSearle?\n44ReferenceandreferringFurtherreading•Russell’sNameClaimisdefendedmostaccessiblyin“ThePhilosophyofLogicalAtomism”(1918/1956).•ForsomecriticismsoftheNameClaimsimilartoKripke’s,seeDonnellan(1970).•Searleaddressesthematteroffictionalnamesinchapter3ofSearle(1979a).HerepliestosomeofKripke’sobjectionsinchapter9ofSearle(1983).Moregenerally,thereisahugeliteratureonfictionalnames;see,forexample,EverettandHofweber(2000),Braun(2005),andthereferencestherein.•Burge(1973),Loar(1976),Bach(1987)andothershavedefendedmorespecificsortsofDescriptiontheoryagainstKripke,versionsthatavoidsomeoftheobjections.\n4Propernames:DirectReferenceandtheCausal–HistoricalTheoryOverviewInafurtherargumentagainstdescriptiontheoriesofpropernames,Kripkeappealedtothenotionofa“possibleworld”oruniversealternativetoourown.AdefinitedescriptionofRussell’ssortchangesitsreferentfromworldtoworld;although“theworld’sfastestwomanin1998”actuallyreferstoMarionJones,itdesignatesdifferentindividualsinotherworlds,sinceJonesmighthavebeenslower(ornotevenhaveexisted)andotherwomenmighthavebeenbetterrunners.Buttypically,apropernamesuchas“MarionJones”referstotheverysameindividualineveryworldinwhichthatindividualexists.Sometheoristsclaimthatnamesaredirectlyreferential,inthatanamecontributesnothingbutitsbearerorreferenttothemeaningofasentenceinwhichitoccurs.InlightofKripke’sargumentsagainstdescriptiontheories,thisviewishighlyplausible.Butthefourpuzzlesreturntohauntit.Soweareleftwithsomethingofaparadox.Aseparatequestionis,invirtueofwhatdoesapropernamedesignateitsbearer?Kripkeofferedacausal–historicalpictureofreferring,accordingtowhichagivenuseof“MarionJones”referstoMarionJonesinvirtueofacausalchainthatgroundsthatutteranceeventintheceremonyinwhichJoneswasfirstgiventhename.But,inlightofsomeexamplesthatclearlydonotfitthatmodel,considerablerefinementisneededtoworkupthatpictureintoanadequatetheoryofreferring.Kripke,andHilaryPutnam,extendedthecausal–historicalviewtocovernatural-kindterms,like“water,”“gold,”and“tiger,”aswellaspropernames.Ifweassumethebasiccorrectnessofthatmove,ithasanunexpectedcon-sequence:Putnam’sfamous“TwinEarth”examplesseemtoshowthatthemeaningofsuchatermisnotdeterminedsolelybywhatisintheheadsofspeakersandhearers;thestateoftheexternalworldmakesacontributionaswell.Thus,twospeakerscouldbemolecule-for-moleculeduplicatesandyetmeandifferentthingsbytheirwords.\n46ReferenceandreferringPossibleworldsIshallnowsetuptheapparatusneededtostateKripke’sfundamentalcriti-cismofdescriptiontheoriesofpropernames.Ibeginwiththenotionofa“possibleworld.”(ItgoesbackatleasttoLeibniz,thoughitwasincorpo-ratedintophilosophicallogiconlyinthetwentiethcentury.)Considertheworldwelivein—notjusttheplanetEarth,butthewholeuniverse.Ourtalkaboutthingsinouruniverseistalkaboutwhatactuallyexists,whatthingstherereallyare:GordonBrowntheBritishPrimeMinister,myleftelbow,Bolivia,thesandwichonyourplate,theAndromedaGalaxy,andsoon,butnotHamlet,theEasterBunny,theBigRockCandyMountain,orthefreelunch.Andwhatistrueinthisuniverseisofcourseactuallytrue.Buttherearethingsthatareinfactfalse,yetmighthavebeentrue.Thingsmighthavegoneotherwise;theworldcouldhavebeendifferentfromthewayitis.SomeoneotherthanBrownmighthavesucceededTonyBlairasPrimeMinister,Imighthavemarriedadifferentperson(whichwouldhavebeenamistake),andIknowIcouldhavefinishedwritingthisbooksoonerifIhadhadaprivatesecretaryandaretinueofservantsincludingapersonalchefandacoupleofhiredkillers.Thusthereareanumberofwaystheworldmighthavebeen.Toputitslightlymorefancifully,therearealternativeworlds.Differentworlds,worldsthatcouldhavebeenours,butthatareonlypossibleandnotactual.Thinkofanarrayofpossibleuniverses,correspondingtotheinfinitelymanywaysinwhichthings,verybroadlyspeaking,mighthavegone.Allthesepossibleworldsrepresentnonactualglobalpossibilities.Now(obviously)asentence’struth—evenwhenweholdthesentence’smeaningfixed—dependsonwhichworldweareconsidering.“BrownisPrimeMinister”istrueintheactualworldbut,sinceBrownneednothavebeenPrimeMinister,therearecountlessworldsinwhich“BrownisPrimeMinister”isfalse:inthoseworlds,BrowndidnotsucceedTonyBlair,orneverwentintopolitics,orneverevenexisted.Andinsomeotherworlds,someoneelseisPrimeMinister—DavidCameron,P.F.Strawson,me,Madonna,orDaffyDuck.Instillothers,thereisnosuchofficeasPrimeMinister,ornotevenaBritain;andsoonandsoforth.Soagivensentenceorpropositionvariesitstruth-valuefromworldtoworld.(Fornow,letustakeallthistalkof“alternateworlds”intuitively,asamet-aphororpicture,aheuristicforseeingwhatKripkeisgettingat.Consideredasseriousmetaphysics,itraisesmanycontroversialissues,1butwemayhopethatthoseissueswillnotmuchaffectKripke’suseofthepossible-worldspictureforhispurposesinthephilosophyoflanguage.)Justassentenceschangetheirtruth-valuesfromworldtoworld,agivensingulartermmayvaryitsreferentfromworldtoworld:inouractualworldin(late)2007,“ThepresentBritishPrimeMinister”designatesGordonBrown.But,asbefore,Brownmightnothavesucceeded,orevengoneintopoliticsinthefirstplace,orevenexistedatall.Soinsomeotherworldsthesame\nPropernames:DirectReferenceandtheCausal–HistoricalTheory47description,meaningwhatitdoeshereinourworld,designatessomeoneelse(Cameron,Strawson,...),ornooneatall—sinceinsomeotherpossibleworldstheConservativeswontheelection,andinsometherewasnoelec-tion,andsoforth.Thisiswhythedescription’sreferentchangesfromworldtoworld.Letuscallsuchasingularterm,onethatdesignatesdifferentthingsindif-ferentworlds,aflacciddesignator.ItcontrastsspecificallywithwhatKripkecallsarigiddesignator:atermthatisnotflaccid,thatdoesnotchangeitsreferentfromworldtoworld,butdenotestheverysameitemineveryworld(atleastineveryworldinwhichthatitemexists.2)RigidityandpropernamesNowweareabletostateKripke’sfurtherobjectiontodescriptiontheoriesofpropernames(1972/1980:74ff.):adefinitedescriptionofthesortRussellhadinmindisflaccid,ashasjustbeenillustrated.Yetpropernames,Kripkesays,donot(usually)varytheirreferenceacrossworldsorhypotheticalsituationsinthatway.IfweimagineaworldinwhichAristotledoessuch-and-such,itisoneinwhichAristotledoesthatthingandhassomedifferentpropertiesfromthosehehashereintherealworld.Ourname“Aristotle”denoteshimthere,notsomeoneelse.Namesare(normally)inthatsenserigiddesignators,keep-ingthesamereferentfromworldtoworld,whereasRusselliandescriptionsareflaccid.Thus,namesarenotequivalenttoRusselliandescriptions.(Ofcourse,ifadescriptionisusedreferentiallyinDonnellan’ssense,itmaygorigid.)Theforegoingparenthesizedqualifications(“usually,”“normally”)areimportant.Kripkedoesnotholdanystrictuniversalthesisaboutpropernames.Heisgeneralizingaboutnormalusesofordinarypropernamesandsayingonlythat,forthemostpart,suchnamesareusedrigidly.Soheisnottoberefutedbycomingupwithunusualflaccidnames,whichcertainlyexist:occasionally,adescriptionisofferedasconventionallyfixingthemean-ingandnotjustidentifyingthereferentofanapparentpropername.“JacktheRipper”isanexample.AndinpopularwritingsaboutScotlandYardorBritishdetectivecultureofthe1950s,forexample,thename“Chummy”wasusedasameresynonymfor“theculprit”;itmeant,attributivelyorflaccidly,just“whoevercommittedthecrime.”Forthatmatter,probablyanypropernamehasoccasionalflacciduses.Frege(1892/1952a)offersafamousexample:“TriesteisnoVienna,”where“Vienna”functionsnotasthenameofacity,butasabbreviatingalooseclusterofexcitingculturalpropertiesthatViennahas.Inthesametone,onanoccasionwellrememberedbyAmericanvoters,1988Vice-PresidentialcandidateLloydBentsentoldhisrivalDanQuayle,“Senator,you’renoJackKennedy.”Butthosearehardlystandardusesofthenames“Vienna”and“JackKennedy.”3Kripkeoffersafurtherlittleintuitivetestfortellingwhetheratermisrigid:trytheterminthesentenceframe,“NmightnothavebeenN.”Ifwe\n48Referenceandreferringplugin,forN,adescriptionlike“thePresidentoftheUnitedStatesin1970,”weobtain“ThePresidentoftheUnitedStatesin1970mightnothavebeenthePresidentoftheUnitedStatesin1970”;andthelattersentenceisclearlytrue,atleastonitsmostnaturalreading:thepersonwhowasPresidentin1970mightnothavebeenPresidentthen(oratanyothertime).Thetruthofthatsentenceshowsthedescriptiontorefertodifferentpeopleindifferentworlds,hencetobeflaccid.Butifweputinthepropername“Nixon,”weget“NixonmightnothavebeenNixon,”atbestaverystrangesentence.ItmightmeanthatNixonmightnothaveexistedatall,whichisperhapsthemostobviouswayinwhichNixoncouldhavefailedtobeNixon.ButgiventhatNixonexisted,howcouldhehavefailedtobeNixon?Hecouldhavefailedtobenamed“Nixon,”butthatisnottohavefailedtobeNixonhimself(because,ofcourse,Nixonneednothavebeennamed“Nixon”).Hecouldhavefailedtohavethepropertiesste-reotypicallyassociatedwithNixon,hencefailedto“beNixon”inthesensethatTriestefailsto“beVienna,”butaswesawinthepreviouschaptersuchflaccidusesofnamesareunusual.Kripkearguesthatwhenoneusesthename“Nixon”torefertoapersoninthisworldandthenstartsdescribinghypotheticalscenariosoralternativepossibleworlds,continuingtousethename,oneistalkingaboutthesameperson.Soifyouask,“MightNixonhavejoinedtheBlackPanthersratherthanbecomingPresident?,”theanswermaybeyesormaybeno,butthesce-narioyouareconsideringisoneinwhichNixon,thatveryperson,isaBlackPanther—notoneinwhichwhoeverorwhateverwastheU.S.PresidentwasaPanther.YouarenotimaginingaworldinwhichaBlackPantherisPresidentoftheUS.ButwhatofRussell’sspot-checkargument?Inresponseto“Whomdoyoumeanby‘LiliBoulanger’/‘WilfridSellars’?”youpromptlycoughupadescriptionorclusterofdescriptions.LikewiseSearle’sappealtoteachingandlearning:Theyalsoproceedbyequatingthenameinquestionwithadescriptionorcluster.Thesefactsseemundeniableandinsuperable.Inresponse,Kripkeintroducedanimportantdistinction.RussellandSearlehadbothassumedthat,ifanamehasadescriptionorclusterassociatedwithitinthewaystheyhaverespectivelypointedout,thenthenamemustsharethemeaningofthatdescriptivematerial(fromnowonIshallsayjust“description”forshort).Butthisassumptionisunwarranted,becausethereisaweakerrelationthatthedescriptionmightbeartothenameandstillexplainthespotcheckandpedagogicaldata:eventhoughthedescriptiondoesnotgivethelinguisticmeaningofthename,itiswhatisusedtodeterminethename’sreferenceonanoccasion.Althoughthename“LiliBoulanger”isnotsynonymouswith“thefirstwomanevertowinthePrixdeRome,”thelatterdescriptioncanbeusedtoindicatethepersononeisreferringtowhenoneuses“LiliBoulanger.”Anditcanbeusedaspartofanexplanationtoapupil,toidentifytheindividualtowhichthenameisattached.\nPropernames:DirectReferenceandtheCausal–HistoricalTheory49Thus,evenifanameinsomeone’smouthatatimehasafirmpsychologicalassociationwithaparticulardescriptioninthatperson’smind,itdoesnotfollowthatthenameisequivalenttothedescriptioninmeaning.Forallthathasbeenshown,whenthepersonobliginglycoughsupthedescriptioninresponsetoaspotcheck,thepersonismerelyidentifyingthename’srefer-ent.Similarly,ifItellasmallchildwho“GordonBrown”is,identifyingthatname’sreferentbysaying“GordonBrownistheBritishPrimeMinister,”itdoesnotfollowthatthename“GordonBrown”simplymeans“theBritishPrimeMinister.”(Ofcourse,thisisnotanargumentagainsttheNameClaimitself;itonlyunderminesRussell’suseofthespot-checktestasanargumentfortheNameClaim.)DirectReferenceRussellusedthefourpuzzlesand(implicitly)hisspot-checkargumenttoattacktheviewthatordinarypropernamesareMilliannames,infavoroftheDescriptionTheory.Inturn,KripkeattackedtheDescriptionTheoryinfavoroftheclaimthatordinarypropernamesarerigiddesignators.ButthelatterclaimdoesnotquiteamounttoMillianism,fornotallrigiddesignatorsareMilliannames.AMillianname,remember,isonethatmakesnopropositionalcontribu-tionbutitsbearerorreferent.Itssolefunctionistointroducethatindividualintodiscourse;itlendsnothingelsetothemeaningofasentenceinwhichitoccurs.Ifwesay“Jasonisfat,”and“Jason”isastandardpropername,thenthemeaningofthatsentenceconsistssimplyofthepersonJasonhimselfconcatenatedwiththepropertyofbeingfat.BeingMilliancertainlyimpliesbeingrigid.Butthereversedoesnothold.AlthoughKripkecitesMillandarguesthatnamesarerigid,rigiditydoesnotimplybeingMillian.Fordefinitedescriptionscanberigid.Supposewefallinwiththeprevalentviewthatarithmeticaltruthsareallnecessarytruths.Thentherearearithmeticaldescriptions,suchas“thepositivesquarerootofnine,”thatarerigid,becausetheydesignatethesamenumberineverypossibleworld,butarecertainlynotMillianbecauseinordertosecuretheirreferencetheyexploittheirconceptualcontent.Indeed,theyseemtoRussellize:“Thepositivesquarerootofnine”seemstomeanwhateverpositivenumberyieldsninewhenmultipliedbyitself.SothatdescriptionisnotMillianeventhoughitisrigid,becauseitdoesnotsimplyintroduceitsbearer(thenumberthree)intothediscourse;italsocharacterizesthreeasbeingsomethingthatwhenmultipliedbyitselfyieldsnine.Thus,indefendingtherigidityofnames,Kripkedidnottherebyestablishthestrongerclaim.(Nordidheintendto;hedoesnotbelievethatnamesareMillian.)4However,otherphilosophershavechampionedtheMillianconception,whichhascometobecalledtheDirectReferencetheoryofnames.ThefirstoftheseinourcenturywasRuthMarcus(1960,1961),citedbyKripkeashav-ingdirectlyinspiredhiswork.SubsequentDirectReference(DR)theoriesof\n50ReferenceandreferringnameshavebeenbuiltonMarcus’andKripke’swork(forexample,Kaplan1975;Salmon1986).ThelattertheoristshaveextendedDRtocoversomeothersingularterms,notablypersonalanddemonstrativepronounssuchas“I,”“you,”“she,”“this,”and“that,”aswellasnames.(AnobviousproblemaboutextendingDRtopronounsisthatanynormalspeakerofEnglishknowswhattheymean,whetherornotthespeakerknowswhomtheyarebeingusedtodesignateonagivenoccasion;ifyoufind“Iamillandwillnotholdclasstoday”writ-tenontheblackboardinanemptyclassroom,youunderstandthesentenceevenifyoudonotknowwhowroteitoronwhatday.Thisproblemwillbeaddressedinchapter11.)Ofcourse,DRmustconfrontthefourpuzzles.And,obviously,theDRtheoristcannotsubscribetoRussell’ssolutionoranythingverylikeit,for,accordingtoDR,namesdonothingsemanticallybutstandfortheirbearers.LetusreconsidertheSubstitutivitypuzzlefirst.Recalloursentence:(1)AlbertbelievesthatSamuelLanghorneClemenswaslessthan5feettall.(1)goesfalsewhen“MarkTwain”issubstitutedfor“SamuelLanghorneClemens.”HowcanDRexplainoreventoleratethatfact?DRtheoristsemployatwo-prongedstrategy.Thereisapositivethesisandthereisanegativethesis(thoughthesearenotoftenexplicitlydistinguished).ThepositiveDRthesisisthatthenamesinquestionreallydosubstitutewithoutalteringthecontainingsentence’struth-value.Onthisview,(2)AlbertbelievesthatMarkTwainwaslessthan5feettall.istrue,notfalse.Attheveryleast,beliefsentenceshavetransparentreadingsorunderstandings,onwhichreadingsthenamesthatfallwithinthescopeof“believes”reallydojustrefertowhattheyreferto.Wenaturallythinkotherwise;(2)doesnotseemtruetous.Thatisbecausewhenweseeabeliefsentence,weusuallytakeitscomplementclausetoreproducethewaysinwhichitssubjectwouldspeakorthink.IfIassert(2),ItherebysomehowimplythatAlbertwouldacceptthesentence“MarkTwainwaslessthan5feettall”orsomethingfairlyclosetoit.IfIsay,“Albertdoesn’tbelievethatMarkTwainwaslessthan5feettall,”Iamsuggestingthatiffacedwiththesentence“MarkTwainwaslessthan5feettall,”Albertwouldsayeither“No”or“Iwouldn’tknow.”ButtheDRtheoristspointoutthatsuchsuggestionsarenotalwaystrue,perhapsnotevertrue.Consider:(3)ColumbusbelievedthatCastro’sislandwasChina.(AttributedtothelateRoderickChisholm.)Weallknowwhatonewouldmeaninasserting(3);thespeakerwouldmeanthatwhenColumbussighted\nPropernames:DirectReferenceandtheCausal–HistoricalTheory51CubahethoughtthathewasnearingtheEastIndiesandwasdirectlyapproachingChina.Ofcourse,being450yearsearly,ColumbusdidnotknowanythingaboutFidelCastro;yetwecanassert(3)withnopresump-tionthatitscomplementclauserepresentsthingsinthewaythatColumbushimselfrepresentedthem.ThespeakermakesthisreferencetoCubawithoutatallassumingthatColumbuswouldhavereferredtoCubainthatwayorinanyparalleloranalogousway.OrsupposeyouandIareamongthefewpeoplewhoknowthatouracquaintanceJacquesisinfactthenotoriousjewelthiefwhohasbeenter-rorizingParis’wealthyset,called“LeChat”inthepopularpressandbythegendarmes.Wereadinthenewspaperafteraparticularlydaringbutflawedrobberythatpolicebelieve“LeChatdroppedthefistfulofanchoviesasheorsheran.”Wesaytoeachother,“ThepolicethinkJacquesdroppedtheanchoviesasheran.”Soitseemsundeniablethattherearetransparentpositionsinsidebeliefsentences,inwhichthereferringexpressiondoesjustrefertoitsbearer,withoutanyfurthersuggestionaboutthewayinwhichthesubjectofthebeliefsentencewouldhaverepresentedthebearer.Singulartermscanbeandareoftenunderstoodtransparently.Wemightevensay:(4)SomepeopledoubtthatTullyisTully.meaningthatsomepeoplehavedoubtedofthemanCicerothathewasalsoTully.Thatwouldperhapsbeaminorityinterpretationof(4),butwecanatleasthear(4)asassertingthatthepeopledoubtofCicerothathewasTully.5VirtuallyalltheDRliteraturehasbeendevotedtoestablishingtheposi-tivethesis,thatnamesdohaveMillianreadingseveninbeliefcontexts.ButthepositivethesisisfarfromallthattheDRtheoristneeds.For,althoughwemaybepersuadedthateverybeliefsentencedoeshaveatransparentreading,mostofusalsoremainconvincedthateverybeliefsentencealsohasanopaquereading,thatonwhichsomesubstitutionsturntruthsintofalsehoods:inonesenseColumbusbelievedthatCastro’sislandwasChina,butinanother,hebelievednosuchthing,fortheobviousreasonthathehadneverheard(andwouldneverhear)ofCastro.Similarly,inonesensethepolicebelievethatJacquesdroppedtheanchovies,butinanothertheydonot,andlikewiseforpeopledoubting“thatTullyisTully.”YetitseemsDRcannotallowsomuchasasenseinwhichbeliefcontextsareopaque.ThatisDR’snegativethesis:thatnamesdonothavenon-Millianreadings,eveninbeliefcontexts.Theproblemgetsworse:itishardtodenythattheopaquereadingsaremorereadilyheardthanthetransparentreadings.Indeed,thatisimplicitlyconcededbytheDRtheorists,inthattheyknowtheyhavehadtoworktomakeushearthetransparentreadings.TheDRtheoristsmusttrytoexplainthefactawayasaparticularlydramaticillusion.Thatis,theymustholdthat,infact,sentenceslike(1)–(4)cannotliterallymeanwhatwecanandusuallywouldtakethemtomean;thereissomeextraneousreasonwhyweareseducedintohearingsuchsentencesopaquely.Afewsuchputativeexplanationshave\n52Referenceandreferringbeensketched,usingmaterialsweshallencounterinchapter13(Salmon1986;Soames1987,2002;Wettstein1991;andseeMarcus1981).Buthere,inmyopinion,theDRtheoristshavecomeupshort;atleast,noneofthesketchesproducedtodatehasstruckmeasveryplausible,thoughperhapsSoames(2002)isthemostpromising.Asisimpliedbyexample(4),Frege’sPuzzleisevenworsefortheMillian.AccordingtoDR,asentencelike“SamuelLanghorneClemensisMarkTwain”canmeanonlythatthecommonreferent,howeverdesignated,ishimself.Yetsuchasentenceisvirtuallyneverunderstoodasmeaningthat.AndanyonemightdoubtthatClemensisTwain,seeminglywithoutdoubtinganyone’sself-identity.Hereagain,theDRtheoristbearsamassiveburden,ofexplainingawayourintuitivejudgementsasillusory.TheProblemsofApparentReferencetoNonexistentsandNegativeExistentialsareifanythingworseyet.Ifaname’smeaningissimplytorefertoitsbearer,thenwhataboutallthoseperfectlymeaningfulnamesthathavenobearers?(ButfordoughtyrecentDRattemptstomeetthosetwoprob-lems,seeSalmon1998;Soames2002;Braun2005.)Wehavecometoadeepdilemma,nearlyaparadox.Ontheonehand,inchapter3wesawcompellingKripkeanreasonswhynamescannotbethoughttoabbreviateflacciddescriptions,orotherwisetohavesubstantivesensesorconnotations.Intuitively,namesareMillian.Yet,becausetheoriginalpuzzlesarestillbristlingasinsistentlyasever,italsoseemsthatDRisprettywellrefuted.Thisisadilemma,orrathertrilemma,becauseitfurtherappearsthatwearestuckwithoneofthosethreepossibilities:eitherthenamesareMillian,ortheyabbreviatedescriptionsoutright,orinsomelooserwaysuchasSearle’stheyhavesomesubstantive“sense”orcontent.Butnoneoftheseviewsisacceptable.Afewtheoristshaveclaimedtofindwaysbetweenthethreehorns.Asnotedinchapter3,Plantinga(1978)andAckerman(1979)haveappealedtorigidifieddescriptions,suchas“theactualwinnerofthe1968election”asopposedtojust“thewinnerofthe1968election”;theformerdescriptionisrigid,because“theactualwinner”meansthewinnerinthis(our)world,andreferstothatsamepersonatanyotherworldregardlessofwhetherhewonthere.Onthisview,namesarerigidthoughtheyarenotMillian.(SeealsoJackson1998.)MichaelDevitt(1989,1996)hasofferedaradicalrevisionofFrege’snotionofsense.Imyself(Lycan1994)haveofferedamuchsubtler,morebeautifulandmoreeffectiveweakenedversionofDR,butitwouldbeimmodestofmetotoutithere.6Wemustnowmakeacrucialdistinction.Sofarinthischapterwehavebeentalkingaboutthesemanticsofpropernames,thatis,abouttheoriesofwhatnamescontributetothemeaningsofsentencesinwhichtheyoccur.DRinparticulartakesforgrantedtheideaofaname’sreferentorbearer.Butthenaseparatequestionis.Invirtueofwhatisathingthereferentorbearerofaparticularname?Semanticsleavesthatquestiontophilosophicalanalysis.Aphilosophicaltheoryofreferringisahypothesisastowhatrelationitisexactly\nPropernames:DirectReferenceandtheCausal–HistoricalTheory53thattiesanametoitsreferent—morespecifically,ananswertothequestionofwhatittakesfortheretobeareferentiallinkbetweenone’sutteranceofanameandtheindividualthatgetsreferredtobythatutterance.Semanticaltheoriesofnamesandphilosophicalaccountsofreferringvaryindependentlyofeachother.ThedifferencewasblurredbyRussellandbySearle,7becauseeachgaveaverysimilaranswertobothquestions.Russellsaidthatanamegetsitsmeaning,andcontributestooverallsentencemean-ing,byabbreviatingadescription;also,whatmakesathingthebearerofthenameisthatthethinguniquelysatisfiesthedescription.LikewiseforSearleandhisclusters.Butnownoticethat,ifoneisaDRtheorist,thatalonetellsusnothingaboutwhatattachesanametoitsreferent.ThesamegoesforKripke’sweakerrigiditythesis;uptillnow,hehasbeentalkingsemanticsonly,andwehaveseennothingofhistheoryofreferring.Tothatwenowturn.TheCausal–HistoricalTheoryAsyoucanverifyforyourself,mostofKripke’sobjectionstotheNameClaimandtodescriptionsemanticsgenerallywillalsotranslateintoobjectionstotheDescriptionTheoryofreferring;theDescriptionTheorywillpredictthewrongreferent(thinkoftheGödel/Schmidtexampleinobjection5,chapter3)ornoreferentatall(aswhenthereisnoparticulardescriptionthespeakerhasinmind(objection1)orinindefinitecases,asinobjection6).Kripkesketchesabetteridea.Hebeginsmemorably(1972/1980:91):“Someone,let’ssay,ababy,isborn....”(IthinkweshouldgrantKripke’sassumptionthattheneonateisababy.Thereissuchathingasbeingtoopicky.)Hecontinues:[Thebaby’s]parentscallhimbyacertainname.Theytalkabouthimtotheirfriends.Otherpeoplemeethim.Throughvarioussortsoftalkthenameisspreadfromlinktolinkasifbyachain.Aspeakerwhoisonthefarendofthischain,whohasheardabout,sayRichardFeynman,inthemarketplaceorelsewhere,maybereferringtoRichardFeynmaneventhoughhecan’trememberfromwhomhefirstheardofFeynmanorfromwhomheeverheardofFeynman.HeknowsthatFeynmanwasafamousphysicist.Acertainpassageofcommunicationreachingultimatelytothemanhimselfdoesreachthespeaker.HethenisreferringtoFeynmaneventhoughhecan’tidentifyhimuniquely.Theidea,then,isthatmyutteranceof“Feynman”isthemostrecentlinkinacausal–historicalchainofreferenceborrowings,whosefirstlinkistheeventoftheinfantFeynman’sbeinggiventhatname.Igotthenamefromsomebodywhogotitfromsomebodyelsewhogotitfromsomebodyelsewhogotitfromsomebodyelse...,allthewaybacktothenamingceremony.IdonothavetobeinanyparticularcognitivestateofRussell’sorSearle’s\n54Referenceandreferringsort.NorneedIhaveanyinterestingtruebeliefaboutFeynman,orastohowIacquiredthename.Allthatisrequiredisthatachainofcommunicationinfacthasbeenestablishedbyvirtueofmymembershipinaspeechcommunitythathaspassedthenameonfrompersontoperson,whichchaingoesbacktoFeynmanhimself.Ofcourse,whenanewuserfirstlearnsanamefromapredecessorinthehistoricalchain,itcanonlybebythenewbie’sandthepredecessor’ssharingapsychologicallysalientbackingofidentifyingdescriptions.But,asbefore,thereisnoreasontoassumethatthatparticularbackingofdescriptionsfixesthename’ssense.Itisneededonlytofixreference.Solongasthenewbiehasacorrectidentificatoryfixonthepredecessor’sreferent,thenewbiecanthenfreelyusethenametorefertothatperson.Takenatfacevalue,thiscausal–historicalviewmakestherightpredictionsaboutexamplessuchasDonnellan’sTom.Ineachexample,referringsucceedsbecausethespeakeriscausallyconnectedtothereferentinanappropriatehistoricalway.Kripke(1972/1980:66–7)offersthefurthercaseofthebiblicalcharacterJonah.Itissimilartothe“Nixon”example(objection3inchapter3).Kripkepointsoutthatweshoulddistinguishbetweenstoriesthatarecompleteleg-endsandstoriesthatare,rather,substantiallyfalseaccountsofrealpeople.Supposehistoricalscholarsdiscoverthatinfactnoprophetwaseverswal-lowedbyabigfish,ordidanythingelseattributedbytheBibletoJonah.ThequestionremainsofwhethertheJonahcharacterwassimplymadeupinthefirstplace,orwhetherthestoryisgroundedultimatelyinarealperson.Actuallytherearesubcases:someonecouldhavemadeupandspreadahostoffalsestoriesaboutJonahimmediatelyafterhisdeath;orbecauseJonahwasanexcitingindividual,allsortsofrumorsandstoriesbegantocirculateabouthim,andtherumorsgotoutofhand;ortheremighthavebeenaverygraduallossofcorrectinformationandaccretionoffalseattributionsoverthecenturies.But,inanyofthesecases,itseemsthattodaytheBibleissayingfalsethingsabouttherealperson,Jonah.8Itmightbethoughtthatambiguousnames—namesbornebymorethanoneperson—poseaproblemforthecausal–historicalview.(“JohnBrown”isambiguousasbetweentheformerScotsghilliewhobefriendedQueenVictoriaafterAlbert’sdeath,themonomaniacalfailedfarmerwhoinvadedHarper’sFerryin1859,anddoubtlessthousandsofothermalesoftheEnglish-speakingworld.Until1994,eventhehighlydistinctivename“WilliamLycan”appliedtomorethanoneperson.Isupposethevastmajorityofnamesareambiguous;anameisunambiguousonlybyhistoricalaccident.)Thisisnoproblematallfordescriptiontheoriesbecause,accordingtodescriptiontheories,ambigu-ousnamessimplyabbreviatedifferentdescriptions.(Ifanything,descriptiontheoriesmakepropernamestoorichlyambiguous.)ButwhatifyouendorseDR,andyoudenythatnameshavesensesordescriptiveconnotationsinanysenseatall?\nPropernames:DirectReferenceandtheCausal–HistoricalTheory55Iaskedthatlastquestiononlytoseeifyouhadbeenpayingattentionearlier.Itflagrantlyignorestheimportantdistinctionbetweenthesemanticsofnamesandthetheoryofreferring.TheCausal–HistoricalTheoryofrefer-ringhasastraightforwardanswertothequestionofambiguousnames:ifanameisambiguous,thatisbecausemorethanonepersonhasbeengivenit.Whatdisambiguatesaparticularuseofsuchanameonagivenoccasionis—whatelse?—thatuse’scausal–historicalgrounding,specificallytheparticularbearerwhosenamingceremonyinitiateditsetiology.Kripkeemphasizesthathehasonlysketchedapicture;hedoesnothaveaworked-outtheory.Thetrickwillbetoseehowonecantakethatpictureandmakeitintoarealtheorythatresistsseriousobjections.Theonlywaytomakeapictureintoatheoryistotakeitoverliterally,totreatitasifitwereatheoryandseehowitneedstoberefined.Kripkedoesjustthat,thoughheleavestherefinementtoothers.ProblemsfortheCausal–HistoricalTheoryThecausal–historicalview’skeynotionisthatofthepassingonofreferencefromonepersontoanother.Butnotjustanysuchtransferwilldo.First,wemustruleoutthe“namingafter”phenomenon.MyboyhoodfriendJohnLewisacquiredasheepdog,andnamedit“Napoleon”aftertheemperor;hehadthehistoricalNapoleonexplicitlyinmindandwantedtonamehisdogafterthatfamousperson.“Namingafter”isalinkinacausal–historicalchain:itisonlybecausetheemperorwasnamed“Napoleon”thatJohnLewisnamedhisdogthat.Butitisthewrongkindoflink.Toruleitout,Kripkerequiresthat“[w]henthenameis‘passedfromlinktolink’,thereceiverofthenamemust...intendwhenhelearnsittouseitwiththesamereferenceasthemanfromwhomheheardit”(1972/1980:96).ThisrequirementwasclearlynotmetbyJohnLewis,whowasdeliberatelychangingthereferentfromtheemperortothedogandmeanthisfriendstobewellawareofthat.Second,Kripkeadducestheexampleof“SantaClaus.”Theremaybeacausalchaintracingouruseofthatnamebacktoacertainhistoricalsaint,probablyarealpersonwholivedineasternEuropecenturiesago,butnoonewouldsaythatwhenchildrenuseittheyunwittinglyrefertothatsaint;clearlytheyrefertothefictionalChristmascharacter.Butthen,howdoes“SantaClaus”differfrom“Jonah”?WhyshouldwenotsaythattherewasarealSantaClaus,butthatallthemythologyabouthimisgarishlyfalse?Instead,ofcourse,wesaythatthereisnoSantaClaus(apologiestoany-onewhodidnotknowthat).Weusethename“SantaClaus”asthoughitabbreviatesadescription.Asimilarexamplewouldbethatof“Dracula.”ItiswellknownthatthecontemporaryuseofthatnamegoesbacktoarealTransylvaniannoblemancalled“Vlad”(commonly,“VladtheImpaler,”invirtueofhiscustomarytreatmentofpeoplewhohadannoyedhim).Butofcoursewhenwenowsay“Dracula”wemeanthefictionalvampirecreatedbyBramStokerandportrayedbyBelaLugosiinthefamousmovie.\n56ReferenceandreferringHavingmerelyraisedtheproblem,Kripkedoesnottrytopatchhisaccountinresponse,butmoveson.Probablythemostobviousfeaturetonoteisthat“SantaClaus”and“Dracula”asweusethosenamesareassociatedwithverypowerfulstereotypes,indeedculturaliconsintheUnitedStates.Theirsocialrolesaresoprominentthattheyreallyhaveossifiedintofictionaldescriptions,inawaythat“Jonah”hasnotevenamongreligiouspeople.Inaway,Jonah’siconicpropertiesaresidebysidewithhishistoricalproper-tiesintheOldTestament,butaswemightsay,“SantaClaus”and“Dracula”arepureicons.AndfortheaverageAmerican,themythutterlydwarfsthehistoricalsource.AsKripkesays,muchworkisneeded.Devitt(1981a)offersafairlywelldevelopedviewthatdoesqualifyasatheoryratherthanonlyapicture.However,hereareafewobjectionsthatwouldapplytoanyversionoftheCausal–HistoricalTheoryasdescribedabove.OBJECTION1Wehavebeenofferedthenotionofacausal–historicalchainleadingbackintimefromourpresentusesofthenametoaceremonyinwhichanactualindividualisnamed.Buthow,then,cantheCausal–HistoricalTheoristaccommodateemptynames,namesthathavenoactualbearers?Perhapsthebestbethereistoexploitthefactthatevenemptynamesareintroducedtothelinguisticcommunityatparticularpointsintime,eitherthroughdeliberatefictionorthrougherrorofonekindoranother.Fromsuchanintroduction,asDevitt(1981a)andDonnellan(1974)pointout,causal–historicalchainsbeginspreadingintothefuturejustasifthenamehadbeenbestowedonanactualindividual.Soreferenceor“reference”tononexistentsisbycausal–historicalchain,butthechain’sfirstlinkisthenamingeventitselfratherthananyputativedoingsofthenonexistentbearer.9OBJECTION2Evans(1973)pointsoutthatnamescanchangetheirreferenceunbeknownst,throughmishaporerror,buttheCausal–HistoricalTheoryaspresentedsofarcannotallowforthat.AccordingtoEvans,10thename“Madagascar”originallynamed,notthegreatAfricanisland,butaportionofthemainland;thechangewasultimatelyduetoamisunderstandingofMarcoPolo’s.Or:Twobabiesareborn,andtheirmothersbestownamesuponthem.Anurseinadvertentlyswitchesthemandtheerrorisneverdiscovered.Itwillhenceforthundeniablybethecasethatthemanuniversallyknownas“Jack”issocalledbecauseawomandubbedsomeotherbabywiththename.(Evans1973:196)\nPropernames:DirectReferenceandtheCausal–HistoricalTheory57Wedonotwanttobeforcedtosaythatouruseof“Madagascar”stilldes-ignatespartofthemainland,orthat“Jack”continuestorefertotheotherformerbabyratherthantothemaneveryonecalls“Jack.”Inreply,Devitt(1981a:150)suggestsamovetomultiplegrounding.Anamingceremony,hesays,isonlyonekindofoccasionthatcangroundanappropriatehistoricalchain;otherperceptualencounterscanservealso.Insteadoftherebeingjustthesinglelinearcausalchainthatgoesbackfromone’sutterancetotheoriginalnamingceremony,thestructureismangrove-like:theutteranceproceedsalsooutoffurtherhistoricalchainsthataregroundedinlaterstagesofthebeareritself.Onceouruseof“Madagascar”hasalargepreponderanceofitsgroundingsintheislandratherthanthemainlandregion,ittherebycomestodesignatetheisland;onceouruseof“Jack”isheavilygroundedinmanypeople’sperceptualencounterswiththemancalledthat,thosegroundingswillovermasterthechainthatbeganwiththenamingceremony.Thisisvague,ofcourse,perhapsobjectionablyso.OBJECTION3Wecanmisidentifytheobjectofanamingceremony.SupposeIamseekinganewpetfromtheAnimalShelter.IhavevisitedtheShelterseveraltimesandnoticedanappealinggreytabby;Idecidetoadopther.OnmynextvisitIpreparetonameher.TheattendantbringsoutatabbyofsimilarappearanceandIbelievehertobethesameoneIintendtoadopt.Isay,“Hereweareagain,then,puddy-tat;yournameisnow‘Liz’,afterthecomposerElizabethPoston,andI’llseeyouagainafteryou’vehadallyourshots”(tactfullyIdonotmentionthemandatoryneutering).Theattendanttakesthecatawayagain.Butunbeknownsttomeitwasthewrongcat,notmyintendedpet.Theattendantnoticesthemistake,withouttellingme,recoverstherightcat,andgivesherhershots(andtherest).Ipickherupandtakeherhome,naturallycallingher“Liz”everthereafter.Theproblemisofcoursethatmycatwasnotgiventhatnameinanyceremony.Theimposterwasgivenit,evenifIhadnorighttonameher.Yetsurelymyowncatisthebearerof“Liz,”notjustaftersubsequentmultiplegroundingshavebeenestablished,butevenjustafterthenamingceremonyIdidperform.(ItwouldbedifferentifIhadtakentheimposterhomeandcontinuedtocallher“Liz.”)Themultiple-groundingstrategydoesnotseemtohelphere.Rather,whatmattersiswhichcatIhadinmindandbelievedIwasnamingintheceremony.(Devitt(1981a:section5.1)speaksof“abilitiestodesignate,”construingtheseasmentalstatesofacertainsophisticatedtype.)Ifso,thenrepairoftheCausal–HistoricalTheoryonthispointwillrequireasignificantforayintothephilosophyofmind.\n58ReferenceandreferringOBJECTION4Peoplecanbecategoricallymistakenintheirbeliefsaboutreferents.EvanscitesE.K.Chambers’ArthurofBritain11asassertingthatKingArthurhadasonAnir“whomlegendhasperhapsconfusedwithhisburialplace.”Aspeakerinthegripofthelatterconfusionmightsay“Anirmustbeagreenandlovelyspot”;theCausal–HistoricalTheorywouldinterpretthatsentenceassayingthatahumanbeing(Arthur’sson)wasagreenandlovelyspot.Lessdramatically,onemightmistakeapersonforaninstitutionorviceversa.(AformercolleagueofmineusedtousethenameofEmersonHall—thebuild-ingthathousestheHarvardphilosophydepartment—asawayofreferringtothedepartment,asin“EmersonHallisn’tgoingtolikethis.”Acasualhearermighteasilyhavegottentheideathat“EmersonHall”namesaperson.)Oronemightmistakeashadowforalivehumanbeingandgiveitaname.Innoneofthesecasesisitplausibletosaythatsubsequentusesofthenameinquestionreallyrefertothecategoriallyerroneousitem.DevittandSterelny(1987)callthisthe“qua-problem.”Theyconcedethatthecelebrantatanamingceremony,orotherpersonresponsibleforanyofaname’sgroundings,mustnotbecategoricallymistakenandmustindeedintendtorefertosomethingoftheappropriatecategory.Thisisamildcon-cessiontoDescriptivism.Therearemoreobjections(someofthemfurtheronesofEvans’).ThemajoritypositionseemstobethatKripkeinitiallyoverreactedtotheDescriptivistpicture.Hewasrighttoinsistthatcausal–historicalchainsofsomekindarerequiredforreferringandthatdescriptionsdonotdonearlyasmuchworkasRussellorevenSearlethoughttheydid;but(ascritics,includ-ingKripkehimself,maintain)therestillaresomedescriptiveconditionsaswell.ThetrickistomovebackinthedirectionofDescriptivismwithoutgoingsofarasevenSearle’sweakDescriptivistdoctrine.Butthatdoesnotleavemuchroominwhichtomaneuver.Onepromisinglinehascometobecalled“causaldescriptivism”(Kroon1987):Theideaistoacceptthecausal–historicaltheorist’scausal–historicalproposalasgettingthecasesright,butmakeitintoadescriptivecondition.Kroondefendsthisprimarilybyextend-ingthe“namingafter”objection.Natural-kindtermsand“TwinEarth”Kripke(1972/1980)andHilaryPutnam(1975a)wentontoextendboththesemantictheoryofrigiddesignationandtheCausal–HistoricalTheoryofreferringfromsingulartermstosomepredicatesorgeneralterms,chieflynaturalkindterms,commonnounsofthesortthatrefertonaturalsubstancesororganisms,like“gold,”“water,”“molybdenum,”“tiger,”and“aardvark.”Suchexpressionsarenotsingularterms,sincetheydonotpurporttoapplytojustonething.ButKripkeandPutnamarguedthattheyaremorelikenamesthantheyarelikeadjectives.Semanticallytheyarerigid;eachrefers\nPropernames:DirectReferenceandtheCausal–HistoricalTheory59tothesamenaturalkindineveryworldinwhichthatkindhasmembership.AndsomeversionoftheCausal–HistoricalTheorycharacterizestheirrefer-ringuse.Thisviewsharplyopposedalong-heldDescriptivisttheoryofnaturalkindterms,whichassociatedeachsuchtermwithadescriptivestereotype.Forexample,“water”wouldhavebeenanalyzedasmeaningsomethinglike“aclear,odorless,tastelesspotableliquidthatfallsfromtheskyasrainandfillslakesandstreams,”and“tiger”assomethinglike“aferocious,carnivorousjunglefeline,tawnywithdistinctiveblackstripes.”KripkeandPutnamurgedmodalargumentsagainstsuchanalyses,similartoobjection3fromthepre-viouschapterandtotherigidityargumentthatbeganthisone.Forexample,therecouldhavebeenwatereveniftherehadneverbeenrain,lakesorstreams,andunderdifferentcircumstanceswatermighthavehadanodororataste.Tigersmighthavebeenborntame,andwemightevenfindoutthatnotigerhaseverinfacthadstripes(aworldwideWonderland-styleconspiracymighthavehadallthestripespaintedon).Whatdoesmakesomethingatiger,then,orasampleofwater,ifnotthecommonsensestereotype?KripkeandPutnamadvertedtothescientificnaturesofnaturalkinds.Whatmakeswaterwaterisitschemicalcomposi-tion,thatitisH2O;whatmakestigerstigersistheirdistinctivegeneticcode.Ineverypossibleworld,waterisH2O,butinsomeworldsH2Ohasanodororataste.Itmaybeobjectedthatthechemicalcompositionofwaterandthegeneticcharacteristicsoftigerswerehighlysubstantiveempiricaldiscoveries;sosurelyitwaspossiblethatwaternotbeH2O,andsothereareworldsinwhichwaterisnotH2O.ButKripkeandPutnamrejoinedthatthealleged“possibil-ity”hereisonlyamatterofscientificignorance,notagenuinemetaphysicalpossibility;onceonedoesdiscoverthescientificessenceofanaturalkind,onehasdiscoveredthetruemetaphysicalnatureofthatkind,andthekindhasthatnatureineverypossibleworldinwhichitismanifested.Whatchangefromworldtoworldaretheelementsofthecommonsensestereotype.Ifthisviewiscorrect,12ithasasomewhatstartlingimplicationabouttherelationbetweenlinguisticmeaningandthemind:asPutnamputsit,thatmeaning“ain’tinthehead.”Putnamimaginesthatsomewhereinanothergalaxythereisaplanet,called“TwinEarth,”whichisanearlyexactduplicateofourEarth,runningalongexactlyinparallelwithourownhistory.Itcon-tainsaTwinPutnam,aTwinBrooklynBridge,aTwinLycanandaTwinyou,allmolecularduplicatesoftheircounterpartshere.Ifonewereabletowatchbothplanetssimultaneously,itwouldbelikewatchingthesametelevisionprogramontwodifferentscreens.(ButitisimportanttonotethatTwinEarthisnotadifferentpossibleworld;itisonlyanotherplanet,withinthesameworldasEarth.Thoughexactlylikeyoutolookatandembeddedinanalmostexactlysimilarplanetarycontext,ofcourseTwinyouisnotyou,butanumericallydifferentperson.)\n60ReferenceandreferringIsaidthatTwinEarthisanearlyexactduplicateofEarth.Thereisonedif-ference:whatlooksandbehaveslikewateronTwinEarthisnotwater—thatis,H2O—butadifferentsubstancethatPutnamcallsXYZ.XYZisodorlessandtastelessandhastheothersuperficialpropertiesthatwaterdoes,butitisonly“fool’swater”(asin“fool’sgold”).Ofcourse,theTwin-English-speakingTwinEarthlingscalltheXYZ“water,”sincetheyareotherwisejustlikeus,13butthatisanequivocation;“water”inTwinEnglishmeansXYZ,notwater,justas(Iamtold)thekindterm“chicory”inBritishEnglishmeansadifferentplantfromtheonemeantbythesamewordinAmericanEnglish.Now,considerapairoftransworldtwins,sayGordonBrownandTwinGordon.Afteranaturaldisaster,Brownemphasizestheurgencyofgettingfoodandwatertothevictims.Naturally,atthesametime,TwinGordonemphasizestheurgencyofgettingfoodand“water”tothevictims.Buttheword-for-word-identicalsentencestheyuttermeandifferentthings.Brown’ssentencemeansthatthevictimsmustbeprovidedwithfoodandH2O,whileTwinGordon’smeansthatthevictimsmustbeprovidedwithfoodandXYZ.YetBrownandTwinGordonarephysicalduplicates.GivenPutnam’sbackgroundassumptions,thisshowsthatthemeaningsofBrown’sandTwinGordon’sutterancesarenotdeterminedbythetotalstatesoftheirbrains,orevenbythetotalstatesoftheirbodies.Fortheirbrainstatesandbodilystatesareidentical,yettheirutterancemeaningsdiffer.Thatisperhapsnobigsurprise.Afterall,languageispublicproperty;anygivenlanguageisusedbyacommunity,forcommunicationbetweendifferentpeople,notoftenforthemerearticulationofsomeone’sprivatethoughts.Butinfact(againgiventhebackgroundassumptions),Putnam’sexampleshowsmore:thatthelinguisticmeaningsofsentencesarenotdeterminedevenbythetotalityofspeakers’brainstatesandbodilystates,indeedevenbytheentirecommunity’spatternofusage.ForEnglishspeakersandTwinEnglishspeakersareallexactlyalikeintheirphysicalcompositionandinthepublicdeploymentofwordsthatsoundjustthesame;yetthesentencesoftheirotherwiseidenticallanguagesmeandifferentthings.14Weshallreturntothispointinchapter6.Nowitistimetobranchoutandtakeupthewholematterofmeaningandtheoriesofmeaning.Summary•Kripkearguedthatpropernamesnormallyfunctionasrigiddesigna-tors,thatanamedenotesthesameindividualineverypossibleworldinwhichthatindividualexists.•Takingamoreambitiousline,theDRtheoristsdefendtheMillianviewthataname’ssolecontributiontothemeaningofasentenceinwhichitoccursistointroduceitsbearerintothediscourse.\nPropernames:DirectReferenceandtheCausal–HistoricalTheory61•Butourfourlogicalpuzzlesaboutreferencestillarisejustasinsis-tentlyasbefore,andseemtomakeDRuntenable.Weareleftwithsomethingofaparadox.•Turningtothetheoryofreferring,Kripkeofferedhiscausal–histori-calpictureasareplacementforDescriptionTheories.MichaelDevittandothershaverefinedandramifiedthecausal–historicalviewinresponsetoinitialobjections.•KripkeandPutnamextendedtheCausal–HistoricalTheorytocovernatural-kindterms.•IftheCausal–HistoricalTheoryiscorrect,thenPutnam’s“TwinEarth”examplesseemtoshowthatthemeaningsofaspeechcommu-nity’swordsarenotentirelydeterminedbythecontentsofspeakers’andhearers’heads;theexternalworldmakesacontributionalso.Questions1SomephilosophersareuneasyaboutKripke’snotionofa“rigiddesig-nator”andhisancillarydistinctionbetween“fixingsense.”Ifyoutooareuneasyabout“rigidity,”canyouarticulatetheproblem?2DofictionalnamesposeaspecialproblemforKripke’srigiditythesis?Howmighthetreatfictionalnames?3NowthatKripkehasrejectedtheNameClaim,howmightheaddressoneormoreofthefourpuzzles?4CanyouhelpDRaddressoneormoreofthefourpuzzles(ahardertask)?5CanyourespondmorefullyonbehalfoftheCausal–HistoricalTheorytoobjections1–4?6Offeryourowncriticismsofthecausal–historicalpicture.7AssesstheKripke–Putnamviewthatnatural-kindtermsrigidlydesig-natescientificallycharacterizedkinds.8AreyoupersuadedbyPutnam’s“TwinEarth”examplesthatmeanings“ain’tinthehead”?Furtherreading•FurtherrepresentativepapersonDirectReferencemaybefoundinAlmog,PerryandWettstein(1989);asurveyandcritiqueisofferedinDevitt(1989).SeealsoRecanati(1993).•Kvart(1993)alsoelaboratesaversionoftheCausal–HistoricalTheoryofreferring.•Evans(1973)offersfurtherobjectionstoKripke’spicture,andaninterestingrevisionofit.Evans(1982)makesconcessionstoKripkebutinsiststhattheideaofa“name-using(social)practice”mustbeintroducedasafurtherelement.McKinsey(1976,1978)hasmoved\n62Referenceandreferringfurtherbacktowardtheancienrégime.FurtherobjectionsaremadebyErwin,KleimanandZemach(1976)andLinsky(1977).•Salmon(1981)surveyssemanticviewsofkindterms.Schwartz(1977)containsrelevantpapers.CriticismsoftheKripke–PutnamlineareofferedbyFine(1975),Dupré(1981),Unger(1983),andothers.Boër(1985)respondstosomeofthosecriticisms.•Theimpactof“TwinEarth”examplesonthetheoryofmeaninggen-erallyisexploredinHarman(1982)andLycan(1984:ch.10).\nPartIITheoriesofmeaning\n\n5TraditionaltheoriesofmeaningOverviewIftheReferentialTheoryofMeaningisfalse,whattheoryistrue?Anytheoryofmeaningmustaccountfortherelevantfacts,whichwemaycall“themeaningfacts”:thatsomephysicalobjectsaremeaningful(atall);thatdistinctexpressionscanhavethesamemeaning;thatasingleexpressioncanhavemorethanonemeaning;thatthemeaningofoneexpressioncanbecontainedinthatofanother;andmore.Wetendtotalkof“meanings”asindividualthings.Meaningshavebeenthoughttobeparticularideasinpeople’sminds.Butseveralobjectionsshowthatthiscannotmeanactualthoughtsinthemindsofparticularpeopleatparticulartimes.Atbest,meaningswouldhavetobemoreabstract:typesofideathatmight(ormightnot)occurinthemindofsomebeingsomewhere.Accordingly,meaningshavealsobeentakentobeabstractthingsinthem-selves,alternatelycalled“propositions.”Thesentence“Snowiswhite”meansthatsnowiswhite;equally,wemaysayit“expressesthepropositionthat”snowiswhite.Othersentences,eveninotherlanguagessuchas“Laneigeestblanche”and“DerSchneeistweiss”expressthatsameproposition,andarethereforesynonymous.ThisPropositionTheoryfitsthevarious“meaningfacts”well,since“proposition”isessentiallyanotherwordfor“meaning.”Butcriticshavequestionedwhetheritexplainsthemeaningfactssatisfacto-rily,orindeedatall.Whenthisbookbegan,thetopicsofreferenceandmeaningwerenotsepa-rate,becausethemostcommonnaiveideapeoplehaveaboutmeaningisthatmeaningisreference.Inchapter1wedisparagedthecommonsensicalbutuntenableReferentialTheoryofMeaning.Sowemustnowconfrontmean-ingdirectly,andlookatsomemoresophisticatedtheoriesofmeaning.Likeanytheory,atheoryofmeaninghastohaveaproprietarysetofdata.Whataretheprimarydataforatheoryofmeaning?Iwillrefertothemcor-poratelyas“themeaningfacts.”First,asweemphasizedinchapter1,thereismeaningfulnessitself.Somestringsofmarksornoisesintheairarejuststringsofmarksornoisesintheair,whereasothers—particularlywholesentences—aremeaningful.Whatisthedifference?Perhapsthatisthebasicquestionforthetheoryofmeaning.\n66TheoriesofmeaningSecond,wesometimessaythattwodistinctexpressionsaresynonymous.Third,wesometimessayofasingleexpressionthatitisambiguous,thatis,thatithasmorethanonemeaning.(Soexpressionsandmeaningsarenotcorrelatedonetoone.)Fourth,wesometimessaythatoneexpression’smeaningiscontainedinthatofanother,asfemaleanddeerarecontainedinthemeaningof“doe.”Animportantspecialcasehereisthatofonesentence’sentailinganother:“HaroldisfatandBenisstupid”entails“Benisstupid.”(Thereisjointentailmenttoo:“Grannieiseitherintheholdingcellorincourtalready”and“Grannieisnotintheholdingcell”togetherentail“Grannieisincourtalready,”eventhoughneithersentencealoneentailsthat.)Therearemoreexoticmeaningfactsaswell.Forexample,somedisputesorallegeddisputesaremerelyverbalor“onlysemantic,”unlikesubstantivedisagreementsoverfact.XandYdonotdisagreeaboutwhatactuallyhap-pened;theydisputeonlyoverwhetherwhathappenedcountsasa“so-and-so.”Onlookerssay,“Oh,they’rejusttalkingpasteachother.”(Thathappensalotinphilosophy.)Instatingtheforegoingmeaningfacts,Ihaveatleasthalf-heartedlytriedtoavoid“reification”ofthingscalledmeanings;thatis,talkingabout“mean-ings”asiftheywereindividualthingslikeshoesorsocks.Ihavetalkedofsentenceshavingsuchfeaturesasbeingmeaningful,beingsynonymous,beingambiguous,thoughIdideventuallyslipintoalludingto“meanings.”Icouldhavereifiedthroughout,andsaid“hasameaning”insteadof“ismeaningful,”“havethesamemeaning”insteadof“aresynonymous,”andsoon,orperhapsevenusedexplicitquantifierexpressions,asin“Thereisameaningthatthesentencehas”and“Thereexistsameaningthatiscommontoeachofthesesentences.”Philosophershavemadeanissueofthis.Letususetheterm“entitytheory”tomeanatheorythatofficiallytakesmeaningstobeindividualthings.Andthereissomeconsiderablesupportforentitytheoriesinthewayweordinarilytalk.Wenotonlyseemtorefertothingscalledmeaningsusingthewordasacommonnoun,butweseemtousequantifierexpressionsinthatconnection.Wesometimesevenseemtocountthem:“Thiswordhasfourdifferentmeanings.”Soitisnotunnaturaltoturnfirsttoentitytheories.Thereareatleasttwodifferentkindsofentitiesthatmeaningsmightbetakentobe.First,onecouldtaketheentitiestobementalitems.Theoriesofthatkindaresometimescalledideationaltheories.IDEATIONALTHEORIESThewhippingboyhereisusuallyJohnLocke(1690/1955),sinceLockeseemstohaveheldthatthemeaningsoflinguisticexpressionsareideasinthemind.Onthissortofview,whatitisforastringofmarksornoisestobemean-ingfulisforthestringtoexpress,orsomehowsignificantlycorrespondto,acontent-bearingmentalstatethatthespeakerisin,anidea,animage,orperhapsathoughtorabelief.Whatischaracteristicofideationaltheoriesas\nTraditionaltheoriesofmeaning67Iamusingthetermisthatthementalstatesinquestionareactualstatesofparticularpeopleatparticulartimes.Ifastringismeaningfulinthatitexpressesanidea,onemaythensaythatfortwoexpressionstobesynonymousisforthemtoexpressthesameidea.Foranexpressiontobeambiguousisfortheretobemorethanoneideathatitcouldexpress,andsoon.Andregardingthephenomenonofmerelyverbaldisagreement,theideationaltheoristmaysay:Itisnotthatonepartyhasonethoughtandtheotherhasadifferent,conflictingthought;theybothhavethesamethought,butareconfusinglyputtingitindifferentwordsthatsoundincompatible.Soanideationaltheoryseemstogiveusanintuitivewayofexpressingourmeaningfactsmoreprecisely.Nonethelessideationaltheorieshavenotbeenpopularinthisorthepastcentury(thoughweshallseeinchapter7thatPaulGricedefendsadescendantofone).Hereareseveralofthereasonsfortheirdisrepute.OBJECTION1Ifanideationaltheoryistobepreciseenoughtotest,itmust(eventually)specifywhatsortofmentalentityan“idea”is.Andthenitwillrunintotrouble.Mentalimageswillnotdoatall,asamatteroffact,forimagesaremoredetailedthanmeanings.(Animageofadogisnotjust,generically,ofadog,butofadogofsomeparticularshapeandsize,possiblyofaparticularbreed;animageofatriangleisofsomeparticulartypeoftriangle,equilateralorrightorwhatever.)Abettercandidatewouldbeamoreabstractmental“concept,”butthatsuggestionwouldbecircularuntilsomeonemanagedtotelluswhata“concept”is,independentlyofthenotionofmeaning.Also,aconceptsuchasthatofdogortriangleisnottrueorfalseonitsown,andsocannotserveasthemeaningofacompletesentence.Awholethoughtmightdo,asthemeaningofacompletesentence.Butnoteverysentenceexpressesanyone’sactualthought.Andif“thought”ismeantinamoreabstractsense,asitwasbyFrege,thenwearetalkingaboutaverydifferentsortoftheory(seebelow).OBJECTION2AswiththeReferentialTheory,therearejusttoomanywordsthathavenoparticularmentalimagesorcontentsassociatedwiththem:“is,”“and,”“of.”Indeed,ifimagesarewhatareonoffer,therearecertainlywordsthatpsycho-logicallycouldnothaveimagesassociatedwiththem,forexample“chiliagon”or“nonentity,”andevenwhenaworddoeshaveanassociatedimage,as“red”does,wedonotalwayscallthatimagetomindintheeverydaycourseofunderstandingthewordasitgoesby;indeed,wemayvirtuallynevercallittomind.\n68TheoriesofmeaningOBJECTION3Meaningisapublic,intersubjective,socialphenomenon.AnEnglishwordhasthemeaningitdoesfortheentirecommunityofEnglishspeakers,evenifsomemembersofthatcommunityhappennottounderstandthatword.Butideas,images,andfeelingsinthemindarenotintersubjectiveinthatway;theyaresubjective,heldonlyinthemindsofindividualpersons,andtheydifferfrompersontopersondependingonone’stotalmentalstateandback-ground.Therefore,meaningsarenotideasinthemind.(OnemightreplybyappealingtowhatiscommontoallEnglishspeakers’ideasof“dog,”say,butwhatiscommontoall“dog”ideasisnotitselfanideabutatypeofidea,auniversaloranabstract“quality”inthesenseofchapter1.)OBJECTION4Therearemeaningfulsentencesthatdonotexpressanyactualideaorthoughtormentalstate.For,aswesawinchapter1,therearequitelongandcomplicatedsentencesofEnglishthathaveneverbeenuttered,andsomeofthosewillneverbeuttered.(Ofcourse,assoonasIgaveyouanexampleofone,itwasnolongeranexampleofone,becauseassoonasIhadwrittenitdown,itbecameasentencethathadbeenuttered.Butwecanextrapolate;therearemorewheremyfancifulHitlersentencecamefrom.)Sotherearesentencesthatareorwouldbeperfectlymeaningfulbutwhosecontentshaveneverbeenthoughtbyanyoneorevenoccurredtoanyone.Thus,therearemeaningfulsentencesthatdonotcorrespondtoanyactualmentalentities.Muchmorecommonlyinthiscentury,theentitiesinvokedbyentitytheoriesofmeaninghavebeenabstractratherthanmental.Themeaningsofsentencesinparticularhavebeencalled“propositions”(asbyRussell,wesawinchapter2).ThePropositionTheoryLikeideas,theseabstractitemsare“language-independent”inthattheyarenottiedtoanyparticularnaturallanguage.Butunlikeideas,theyarealsopeople-independent.Mentalentitiesdependuponthemindsinwhichtheyinhere;amentalstatehastobesomebody’smentalstate,astateofsomeparticularperson’smindataparticulartime.Propositionsareentirelygen-eraland,ifyoulike,eternal.(Russellhimselfhadlittlefurthertosayabouttheirnature;hiscolleagueG.E.Moorewasclearerandmoreforthcom-ing,oratleastmoreforthright.1FregehadpreviouslyconstructedquiteanelegantPropositionTheory,butseemstohaveheldthatthereisnothingtounderstandingwhatapropositionisbutunderstandingtheroleplayedby“propositions”inthetheory.)Considerapossiblereplytoobjection4above:someonemighttrytosavetheideationaltheorybysuggestingthatweneednotrestrictourselvesto\nTraditionaltheoriesofmeaning69actualideas;wecanappealtomerelypossibleideas—ideasthatsomeonemighthaveormighthavehad.Butthatwouldbetopositabstractcontentsthatarepossiblecontentsofthoughtbutarenotrelatedtoanyone’sactualthoughts.Enterthepropositiontheorist:“Right,let’scallsuchthinkables‘propositions.’”Andso(iftheideationaltheoristdoesmakethemoveunderdiscussion),theideationalviewsimplycollapsesintothePropositionalTheory.ThePropositionalTheoryoffersagraphicpicture.Supposewehaveonestringofwords,S,thatismeaningful,alongsideanotherstring,g,thatisonlygibberish.Whatisthedifference?AccordingtoRussellandMoore,itisthatthereisanabstractcontentorproposition,callitP,towhichSstandsinacertainspecialrelation.Sisasentenceofaparticularlanguage.Poorgdoesnotbearthatrelationtoanysuchitem.Therelationisoftencalledexpression;philosopherscommonlytalkofsentencesexpressingpropositions.(Thoughherethetermismorebloodlessthaninideationaltheories.Ideationaltheoriststhinkofsentencesalmostasbeingpushedoutfrominsideusbythepressureofourthoughts,butpropositionsareabstract,changelessandpowerlessanddonotpushorpull.)SoSismeaningfulinvirtueofexpressingtheparticularpropositionP;g’sfailingisthatitexpressesnopropositionatall.Theothermeaningfactsareneatlydepictedfromthepresentpointofview.ForsentencesS1andS2tobesynonymousisjustforS1andS2toexpressthesameproposition.Theyaredistinctlinguisticexpressions—theycouldbedifferentexpressionsinoneandthesamenaturallanguageortheycouldbecorrespondingexpressionsfromdifferentlanguages.Whattheyhaveincom-monissolelythattheybeartheexpressingrelationtothesameproposition.Sotooforambiguity.AsentenceSisambiguousifandonlyifthereareatleasttwodistinctpropositionsP1andP2,andthesingleexpressionSbearstheexpressingrelationtoeachofP1andP2.Inthecaseofmerelyverbaldisputes,wecansaythatthepartiesdonotdisagreeoveranyproposition;theyaremerelyusingdifferentformsofwordstoexpressthesameproposi-tion,andtheparticularformsofwordslookasthoughtheyareinconflicteventhoughtheyarenot.Weknowsomepositivethingsaboutwhatpropositionsaresupposedtobe,besidestheirbeingexpressedbysentences.Theyareidentifiableintermsof“that”clauses:wespeakofthepropositionthatsnowiswhite,anddedicateourselvestothepropositionthatallmen[sic]arecreatedequal.“Snowiswhite,”“Laneigeestblanche”and“DerSchneeistweiss”aresynonymousbecauseeachofthemexpressesthepropositionthatsnowiswhite.Althoughwhatfollowsthe“that”clauseisjustanothersentenceofoneparticularnaturallanguage,theonewehappentobespeaking,thefunctionofthe“that”—creatingindirectdiscourse—istofreethereferencetotheproposi-tioninquestionfromitsparticularexpression.Propositionsarealsoobjectsofmentalstates.PeopleallovertheworldmaybelievethatAsianmarketsarecollapsing,doubtthatAsianmarketsarecollapsing,hopeorfearthatAsianmarketsarecollapsing.Heretoo,the\n70Theoriesofmeaning“that”servestoremovetheimplicationthattheyallthoughtthatthoughtinEnglish.Theycouldhavethoughtitinanylanguage;itwouldstillbetruethattheybelieved,doubtedorwhateverthatAsianmarketsarecollapsing.Further,propositionsarethefundamentalbearersoftruthandfalsity.Whenasentenceistrue/false,itissoonlybecausethepropositionitexpressesistrue/false.Oneargumentforthisclaimisthatsentenceschangetheirtruth-valuesfromtimetotimeandfromcontexttocontext.(1)ThepresentQueenofEnglandisbald.Webelieve(1)tobefalse,assumingthatElizabethWindsorisnotfollowingRussell’sadviceandwearingawig.Butwhatabouttheotherqueens,pastorfuture,whomayhavebeenormaybebald?If(1)hadbeenutteredduringthereignofapreviousqueenwhowasbald,itwouldhavebeentrue,andifitshouldbeuttereddecadesfromnowduringthereignofasubsequentqueen,itmightbetrueorfalse.Sowhether(1)isfalseortruedependsonwhenitisuttered.Whatmakesaparticularutteranceofasentencetrueorfalseisthepropositionitexpressesonthatoccasion.Thereason(1)changesitstruth-valueisthatitexpressesdifferentpropositionsondifferentutteranceoccasions.Sentencesderivetheirtruth-valuesfrompropositions;proposi-tions’truth-valuesarepermanent.MostPropositionTheoristsholdthatpropositionshaveinternalstruc-ture;theyarecomposedofabstractconceptualparts.Theword“snow”isameaningfulexpression,butitisnotmeaningfulinvirtueofexpressingaproposition;justbyitselfitdoesnotexpressafullproposition.Onlyasen-tenceexpressesapropositionor,astheyusedtosayinmygrammarschooldays,acompletethought.“Snow”doesnotexpressacompletethought,butitexpressessomethingthatispartofmanythoughts—aconcept,oratype,oran“idea”intheabstractratherthanthementalsense.“Concept”istheusualtermusedtomeananequallyabstractconstituentofalargerabstractproposition.2Thereare“meaningfacts”aboutthepartsorconstituentsofsentencesaswell,andtheycanbetreatedanalogously.Wordsthataresynonymouswith“snow”canbesaidtoexpressthesameconcept;if“snow”isambiguous,asitis,itisambiguousinvirtueofexpressingdifferentconcepts:Sometimesitmeansthechillywhitestuffthatfallsfromtheskyandatothertimesitmeansacertaincontrolledsubstance.ThePropositionTheoryavoidsallfourofourobjectionstoideationaltheories,thoughonemorenarrowlythantheothers.Wehavealreadyseenthatiteludesobjection4.Itavoids1becausepropositionsandconceptsarenotmentalentities,anditavoids3because,unlikementalentities,propo-sitionsandconceptsareintersubjective,independentofparticularpeople,languages,andevenwholecultures.Itonlybarelyevadesobjection2.ThePropositiontheoristcaninsistthatwordslike“is,”“and,”“of,”“chiliagon,”and“nonentity”expressconcepts\nTraditionaltheoriesofmeaning71(“chiliagon”especially,whichisawell-definedgeometricterm).ButasIsaidinresponsetoobjection1,ifthisisnottosoundemptyandperhapsevencircular,thepropositiontheoristwillhavetogiveussomefurthercharacter-izationoftherelevantconcepts,onethatdoesnotquietlypresupposesomenotionoflinguisticmeaning.(Weshallseeinchapter10thatasophisticatedversionofthePropositionTheorycandothis.)ThePropositionviewistheleadingentitytheoryofmeaning.Likeanytheoryofmeaningitaimsatexplainingthemeaningfacts.Itattemptstodothatbypositingacertainrangeofentities;thatishowweoftenexplainthings,especiallyinscience.Wepositsubatomicparticles,unobservableentitiesofacertainrangeandkind,toexplainthebehaviorofobservablechemicalsubstancesandtheratiosinwhichtheycombine.Afirstproblemforthetheoryasstatedsofariscreatedbyasortofmean-ingfactthatIhavenotmentioneduptillnow.Somephilosophersconsiderthissortevenmoreimportantthanalltheoneslistedabove:WeunderstandasentenceS,inanimmediateway,whereaswedonotunderstandagibberishstringofwords.Somestringsofwordsareintelligibleandotherstringsarenot.Thisbringsanothertermintotherelation.Tillnow,thePropositionTheoryhasfocusedjustonlinguisticexpressionsandonpropositions,withtheexpressingrelationdefinedonthem.Nowitmustletinhumanbeings.WhatisitforapersontounderstandasentenceS?TheclassicMooreanansweris:forthatpersontobearacertainrelationtoapropositionandtoknowthatSexpressesthatproposition.ThisrelationMoorecalled“grasping”(orsometimes“apprehending”).TounderstandSistograspsomeproposi-tionPandtoknowthatSexpressesP.ThePropositionTheorytooiscongenialtocommonsense.Itiseasytoagreethatcertainsentencesofvariousdifferentlanguagesallhavesomething(theirmeanings)incommon,alanguage-independentcontent,anditiseasyandnaturaltocallthatcontent“thepropositionexpressedby”thevariousdifferentsentences.Moreover,thePropositionTheoryisahandytoolfordescribinganddiscussingtheothersortsof“meaningphenomena”wehavementioned,tosaynothingofentailment,meaninginclusion,antonymy,redundancy,andmore.Finally,asweshallseeinchapters10and11,thePropositionTheorylendsitselftoelegantmathematicalelaboration,inthehandsof“possibleworlds”semanticistsandintensionallogicians.But,asalways,thereareproblems.OBJECTION1Wehavesaidthat“propositions”areabstractentities,eventhoughsentencesarenowbeingsaidto“express”themratherthantonamethemasintheReferentialTheory.Consideredasentities,theseabstractitemsaresomewhatweird.Theyarenotlocatedanywhereinspace,and,sincetheycouldnotbecreatedordestroyed,theyarealsotemporallyeternaloratleasteverlasting.Theyexistedlongbeforeanylivingbeingdid,eventhoughtheircontents\n72Theoriesofmeaninghavetodowithhighlyspecificstatesofhumanaffairs,suchasFred’shavingdownedfourquickMalagaCoolersatHe’sNotHereduringtheearlyeveningofTuesday19September1995.Thepropositionswillexistlongafterthelastsentientcreatureisdead.And(necessarily,sincetheyarenotinspacetime)theyhavenocausalproperties;theydonotmakeanythinghappen.AREPLYItisrightandpropertobewaryofpositingweirdentities.Butperhapsthisdirectappealto“Occam’sRazor”ispremature.ThemedievalphilosopherWilliamofOccamtoldusnottomultiplypositedentitiesbeyondexplana-torynecessity.Butwecouldknowthatpropositionswereunnecessaryforexplanationonlyifwehadanalternativetheoryofmeaningthatexplainedthemeaningphenomenajustaswellbutwithoutdragginginpropositions.And(sofar)wehavenosuchcompetitor.OBJECTION2“Propositions”areinasenseunfamiliarandalientoourexperience.IhearorseewordsandIunderstandthem,butthisishardly,orseemshardly,acaseofmydoingsomethingcalled“grasping”thatputsmeintouchwithasupra-empiricalnonspatial,indestructible,eternalobject.(Bringupspookymoodmusic.)MOORE’SREPLYItisquiteplain,Ithink,thatwhenweunderstandthemeaningofasen-tence,somethingelsedoeshappeninourmindsbesidesthemerehearingofthewordsofwhichthesentenceiscomposed.Youcaneasilysatisfyyourselvesofthisbycontrastingwhathappenswhenyouhearasentence,whichyoudounderstand,fromwhathappenswhenyouhearasentencewhichyoudonotunderstand....Certainlyinthefirstcase,thereoccurs,besidethemerehearingofthewords,anotheractofconsciousness—anapprehensionoftheirmeaning,whichisabsentinthesecondcase.Anditisnolessplainthattheapprehensionofthemeaningofonesentencewithonemeaning,differsinsomerespectfromtheapprehensionofanothersentencewithadifferentmeaning....Therecertainlyaresuchthingsasthetwodifferentmeaningsapprehended.AndeachofthesetwomeaningsiswhatIcallaproposition.(1953/1962:73–4)And,hemightadd,ifyousayyoudonotknowwhatheistalkingabout,youarealiar.Graspingissomethingyouhavedirectlyexperienced.\nTraditionaltheoriesofmeaning73ADIFFERENTREPLYGrantingthepremiseinsteadofchallengingit,onemightpointoutthatitiscommonnotonlyinphilosophybutinsciencetoexplainveryfamiliarphenomenaintermsofveryunfamiliar,perhapsquitearcanephenomena.Thatisnothingneworunusual.OBJECTION3FromGilbertHarman(1967–8).ThePropositionTheorydoesnotinfactexplainanything;itmerelyrepeatsthedatainafancierjargon.(“Whydo‘Snowiswhite’and‘Laneigeestblanche’havethesamemeaning?”—“Becausetheyexpressthesameproposition.”—“Oh,Isee.”)Itsoundsasthoughthephrase“expressesaproposition”isjustafancierwayofsaying“ismeaning-ful.”Atleastuntilweareshownsomeindependentwayofunderstandingpropositiontalk,thesuspicionwillremainthatitisonlyapretentiouswayofre-expressingthemeaningfacts.CompareMolière’sphysicianonopiumand“dormitivevirtue.”3REPLYWeneednotbetoodauntedbythisobjectioneither.ForwhenaPropositionTheoryiselaboratedandrefined,completewithanotionofaperson’s“grasp-ing”apropositionaswellasthatofasentence’sexpressingone,theapparatushasatleastabitofpredictivepowerandso(tothatextent)atleastabitofexplanatorypower.Whethertheresultingstoryisplausibleisadifferentquestion.ButperhapsHarmanwasreallygettingatthenextobjection.OBJECTION4Whatevermeaningis,itplaysadynamicroleinhumansociety.Someofyourbehavioriscausallytheresultofmysayingcertainwordsthatmeanwhattheydo,andsomeofmybehaviorresultsfromyoursayingmeaningfulwordslikewise.Legaldecisionsincapitalcasessometimesturnonthemeaningsofwords,andsoon.Thusmeaning,whateveritis,musthavesomecausalpower(somepushandpull,somepunch,somebiff).Butpropositions,asentirelyabstractentities,preciselydonothavecausalpowers.Theysitquiescentlyanduselesslyoutsidespacetime,anddonothing.Soitishardtoseehowproposi-tionscouldfigureintheexplanationofhumanlinguisticbehaviororcouldinanyotherwayhelptoaccountforthedynamicsocialroleofmeaning.Andthereforetheyseemtobeunnecessarypositsafterall.REPLYEvenifpropositionsdonothelpintheexplanationofhumanbehavior,humanbehaviorisnottheonlythingthatneedsexplaining.The“meaning\n74Theoriesofmeaningfacts”themselvesareourprimarydata,andpaceHarman,propositionsstillhelptoexplainthose.The“ordinarylanguage”philosophersofthe1950sdrewamoralfromearlyversionsofobjections1and4:thatwhatweneedisatheorythatexplainsmeaningphenomenaintermsthatdoconnectuptohumanbehavior.(Rememberthathumanbehaviorinvolvesactualphysicalmotion;meaningmustsomehowcontributetoliteralpushandpull.)Morespecifically,weneedtounderstandmeaningintermsoflanguageuse.Eversince,philosophershavespokenof“use”theoriesofmeaning.Butwearelittlethewiser,fortherearemanydifferentkindsofmodesof“use,”someofwhichareobviouslyirrelevanttomeaninginthecharacteristicallylinguisticsense.Differentspecificallylinguisticconceptionsof“use”leadtodifferentandcompetingtheoriesofmeaning.Summary•Atheoryofmeaningmustexplainthe“meaningfacts.”•“Meanings”haveoftenbeentakentobeentitiesorindividualthings.•Ideationaltheoristscontendthatmeaningsareparticularideasinpeople’sminds.•Butseveralobjectionsshowthat,atbest,meaningswouldhavetobemoreabstract:typesofidea,notactualthoughtsinthemindsofparticularpeople.•Propositiontheoriststakemeaningstobeabstractthingsinthemselves.•ButcriticshavequestionedwhetherthePropositionTheoryexplainsthemeaningfactssatisfactorily(oratall).Questions1IstheremoretobesaidinfavoroftheIdeationalTheory?And/orcanyoudefenditagainstoneormoreofourobjections?2DoesthePropositionTheoryreallyexplainthemeaningfacts?Whyorwhynot?3DefendthePropositionTheorymorethoroughlyagainstourobjec-tions.Orraiseanewobjectionofyourown.Furtherreading•Locke’sIdeationalTheoryisdiscussedinBennett(1971).•Frege(1918/1956)criticizedideationaltheoriesinfavorofthePropositionTheory.Wittgenstein(1953)criticizedthemfromaverydifferentperspective(seechapter6),asdidWaismann(1965a).•AclassicPropositionTheorywasofferedbyRussell(1919/1956).\nTraditionaltheoriesofmeaning75•Forsomediscussionofpropositionsandtheirrelationstosentencesandtoutterances,seeCartwright(1962)andLemmon(1966).•QuineancriticismofthePropositionTheoryisbestsummedupbyGilbertHarman(1967–8),particularlypp.124–7(pp.141–7arealsorelevant).Lycan(1974)isarejoinderonthetheory’sbehalf.SeealsoLoux(1998:ch.4).\n6“Use”theoriesOverviewThePropositionTheorytreatssentencesandotherlinguisticitemsasinertabstractentitieswhosestructurecanbestudiedasifunderamicroscope.ButLudwigWittgensteinarguedthatwordsandsentencesaremorelikegamepiecesortokens,usedtomakemovesinrule-governedconventionalsocialpractices.A“meaning”isnotanabstractobject;meaningisamatteroftheroleanexpressionplaysinhumansocialbehavior.Toknowtheexpression’smeaningisjusttoknowhowtodeploytheexpressionappropriatelyincon-versationalsettings.WilfridSellars’versionofthisideamakestheactofinferringcentral;itisthecomplexityofpatternsofinferencethatallowsthe“use”theoristtoaccommodatelong,novelsentences.Onthisview,onesentenceentailsanother,notbecausethetwo“express”“propositions”oneofwhichissome-how“containedin”theother,butbecauseitissociallyexpectedthatone’sneighborwouldperformtheactofinferringthesecondsentencefromthefirst.“Use”theoriesofthiskindfacetwomainobstacles:explaininghowlan-guageusediffersfromordinaryconventionalrule-governedactivities,suchaschessgames,thatgeneratenomeaning;andexplaininghow,inparticular,asentencecanmeanthatso-and-so(astheFrench“Laneigeestblanche”meansthatsnowiswhite).RobertBrandomhasrecentlyoffereda“UseTheory”thatclaimstoperformthesefeats.Aswesawinchapter2,Russell’shabitwastowriteasentenceontheblackboardandexamine(ashecontended)thepropositionexpressedbythesentence,treatingitasanobjectofinterestinitselfandtryingtodiscernitsstructure.LudwigWittgensteinandJ.L.Austinarguedthatthispictureofhowlanguageworksandhowitshouldbestudiediscompletelywrong.Languagesandlinguisticentitiesarenotbloodlessabstractobjectsthatcanbestudiedlikespecimensunderamicroscope.Rather,languagetakestheformofbehavior,activity—specificallysocialpractice.Sentencesdonothavelivesoftheirown.Thethingswewriteonblackboards,andthealleged“propositions”theyexpress,arefairlyviolentabstractionsfromtheutteringsperformedbyhumanbeingsinreal-worldcontextsonparticularoccasions.1\n“Use”theories77Andforapersontouttersomethingisfirstandforemostforthatpersontodosomething.Itisabitofbehaviorthatbyconventionhasgottensweptupintoarule-governedsocialpractice.Wehavealreadyencounteredaversionofthisideainchapter2,foritisfromthesameperspectivethatStrawsonwieldedhisseveralobjectionsagainstRussell’sinitiallyattractiveTheoryofDescriptions.Andwhetherornotweareultimatelyconvincedbytheobjections,theywerefreshandstrikingand,tomanypeople,stillintuitivelycompelling.Thatisagoodletterofrecommendationfortheperspectiveitself.“Use”inaroughlyWittgensteiniansenseWittgenstein(1953)andAustin(1961,1962)developedthissocial-behavioralideaindifferentways.HereIshallconcentrateonaWittgensteinianview,deferringAustin’suntilchapter12.Isayonly“aWittgensteinianview”because,forreasonsthatcannotdetainushere,Wittgensteinhimselfopposedsystematictheorizinginphilosophy,andhisfollowersobjectedtoanyphrasealongthelinesof“Wittgenstein’stheoryof...”or“Wittgenstein’sdoctrineregarding...”2IshallmerelytrytosketchanaccountbasedonWittgenstein’scontributions,withoutattributingthatoranyothertheorytoWittgensteinhimself.Ifmeaningitselfismysterious,onewaytoreducethemysteryistoenteritsdomainthroughsomethingwithwhichwearemoredirectlyfamiliar.Inordertogetahandleonmeaning,letusthinkofitfromthereceivingend,thegraspofmeaningorunderstandingoflinguisticexpressions.Andinordertounderstandunderstanding,letusthinkofitastheproductofourhavingbeentaughtourlanguage,andaswhatonelearnswhenonelearnsalanguage.Butassoonaswetrylookingatitthatway,somethingbecomesimmedi-atelyobvious:thatwhatislearnedandtaughtisacomplicatedformofsocialbehavior.Whatyoulearnwhenyoulearnalanguageistomakemoves,toengageincertainkindsofpractice,conversationalbehaviorinparticular.Andprimarily,whatistaughtistherightwaytobehavewhenotherpeoplemakecertainkindsofnoises,andwhatkindsofnoisestomakewhencir-cumstancesareappropriatefordoingso.Linguisticpracticeisgovernedbyhighlycomplexsetsofrules,eventhoughtherulesarerarelyarticulated;smallchildrenjustpickthemupatacolossalrate,learningtoobeythemwithoutrealizingthatthatiswhattheyaredoing.Thesehometruthsareobscuredbyentitytheories,whichtreatmeaningsasstatic,inertthings.BothWittgensteinandAustininveighedatlengthagainstentitytheories,thoughhereweshallbeconcernedwithapositiveaccountof“use.”Wittgensteinalsoscornedtheviewthatmeaningessentiallyinvolvesreferentialrelationsbetweenlinguisticexpressionsandthingsintheworld(thoughofcoursehedidnotdenythattherearesomesuchrelations).Wittgensteinofferedthekeyanalogyoflinguisticactivitytotheplayingofgames.(AccordingtothephysicistFreemanDyson,thenaCambridgeundergraduate,onedayWittgensteinwaswalkingpastafieldwhereafootball\n78Theoriesofmeaningmatchwasinprogress,and“thethoughtfirststruckhimthatinlanguageweplaygameswithwords.”)3Languageisnotamatterofmarksontheblackboardbearingthe“expressing”relationtoabstractentitiescalled“propositions”;languageissomethingthatpeopledo,anddoinahighlyrule-governedandconventionalway.Linguisticactivityisgovernedbyrulesinmuchthewaythattheplayingofagameisgovernedbyrules.Moreover,linguisticexpressionsthemselvesarelikegamepieces.Considerchessmen.A“pawn”ora“rook”isdefinedbythechessrulesthatgovernitsinitialpositionandsubsequentlegalmoves;whatmakesaknightaknightisthewayinwhichitcharacteristicallymovesaccordingtothegame’sconventionallyinstitutedrules.Sotoo,alinguisticexpression’smeaningisconstitutedbythetacitrulesgoverningitscorrectconversationaluse.Startwithexpressionslike“Hello,”“Damn”(or“Goodgracious”),“Oh,dear,”“Excuseme,”“Amen,”“Thanks,”“Stopit!,”“You’reon”(whenabethasbeenoffered),and“Blessyou.”Thesedonotseemtomeanwhattheydoinvirtueofstandingforanythingorinvirtueofexpressingpropositions.Theyarejustconventionaldevices,respectively,ofgreeting,evincingconsterna-tion,deploring,apologizing,endorsing,thanking,protesting,committingoneselftoabet,andblessing.Theyarenoiseswemakethathavesociallydefinedfunctionalroles;thereareappropriateoccasionsforusingthem,inap-propriateoccasionsforusingthem,andappropriateresponses.Whenwetalkoftheirmeanings,wemeanthefunctionstheycharacteristicallyperforminthecontextofourcurrentsocialpractices.OntheWittgensteinianview,thisisthelocusandnaturalhomeofallmeaning,thoughmostexpressionshavevastlymorecomplicatedsocialroles.Toemphasizeallthis,Wittgensteincoinedtheterm“language-game,”asinthemeetingandgreetinglanguage-game,theweddinglanguage-game,thearithmeticlanguage-game,andsoon.Wittgensteinoffersafurtheranalogy(1953:2):Abuilderandhisassistanthavejustfourkindsofbuildingstonesthattheyuse.Theyspeakalittleprimitivelanguagethathasjustfourcorrespondingwordsinit:“block,”“pil-lar,”“slab,”and“beam.”Theybuildthings,engagingintheirnonlinguisticactivitiesaidedbyacertainprimitivesortoflinguisticactivity:thebuildersays“slab,”andtheassistantbringsastoneoftheappropriateshape.Now,someonemightsay,“Ofcourse,thatword‘slab’bearsthereferringrelationtoablockofthisshape,anditsmeaningisthepropositionthattheassistantistobringsuchablocktothebuilder.”ButaccordingtoWittgensteinthiswouldbemissingthepoint.Inthislittleprimitivelanguage-game,theword“slab”doeshaveafunctionthatisobviouslyconnectedwithblocksofthatshape,butthepointisthefunctionandnotthereferringrelation.Thepointofthebuilder’smakingthenoise“slab”isjusttogettheassistanttodosomething,totriggerconventionally(theassistanthavinglearnedhistrade)apatternofusefulactivity.Theactivitydoesinvolvethingsofthisshape,buttheprimarypointistoinitiateaction,nottorefer,orto“express”aneternalproposition.\n“Use”theories79Ofcourse,itishardtoextrapolatethissimplepictureofmeaningasbrutelyconventionalsocialfunctiontolongandcomplexsentenceslike“ThepresentQueenofEnglandisbald”or“In1931,AdolfHitlermadeavisittotheUnitedStates,inthecourseofwhichhe...,”neitherofwhichhasanyeasilyidentifi-ableconventionalsocialrole(otherthan,unhelpfully,thoseofassertingthatthepresentQueenofEnglandisbaldandthatin1931...).Someadditionalmechanismmustbeintroducedtoaccomplishthatextrapolation.Thelogicalpositivistsappealedtothenotionofverification,butIshallsavediscussionofthatuntilchapter8.WilfridSellars(1963,1974)invokedtheideaofinferringasasocialact.Hespoketooof“language-entryrules,”and“language-exitrules,”thesebeingrespectivelyrulesgoverningwhatoneissupposedtosayinresponsetocertainsortsofnonlinguisticevents(suchasobservations)andwhatoneissupposedtodoinresponsetocertainlinguisticutterances,butmostimportantlyof“language–languagerules,”whichgovernwhatoneissupposedtosayastheproductofinferencefromsomethingelsethathaspreviouslybeensaid.LetuscallthistheInferentialTheoryofMeaning.Itishardtoseehowatheorythattook“Hello”or“Slab”asitspara-digmscouldsucceedinexplainingthemorerefinedofthemeaningfacts.Meaningfulness,synonymy,andambiguityarenotaproblem;butwhatofentailmentbetweencomplexsentences?TheInferentialTheory’sappealtoinferringhelps,forwhatmightseemtobethestaticabstractrelationof“entailment”betweentwosentencescanbereconstruedasarule-governedpracticeofinferringtheonefromtheother.“HaroldisfatandBenisstupid”entails“Benisstupid”because,ifsomeoneassertstheformerbutdeniesthelatter,weapplyseveresocialsanctions;indeed,weatleastraiseeyebrowsifsomeoneassertstheformerandthendoesnotgoontobehaveasifthelatteristrue.AccordingtoUsetheories,itisthispracticeitselfthatmakestheinferencevalid,not(aslogicbookswouldhaveit)anyindependentguaranteethattheinferencepreservestruth.ObjectionsandsomerepliesThebeautyoftheInferentialTheoryisitseffortlessavoidanceofeverysingleobjectionwehavemadetoeachofthethreetraditionaltheories(Referential,Ideational,andProposition).Inaddition,itisnaturalistic,inthatitfocusesattentionontheactualfeaturesoflanguageasusedintherealworld.Still,therearesomeformidableproblems.OBJECTION1Alllanguage-gamesareexactlythesameasbetweenEarthandTwinEarth,sincethoseplanetsarerunningexactlyinparallel;butwordsonTwinEarthandtherestdivergeinmeaningfromtheircounterpartsonEarth.OfanEarthutteranceanditsTwin,onemaybetrueandtheotherfalse;whatmore\n80Theoriesofmeaningcouldberequiredfordifferenceofmeaning?Thus,anexpression’smeaningisnotexhaustedbytheexpression’sroleinalanguage-game.REPLYOnecanclassify“language-games”morefinely,anddenythatweandourTwin-Earthdoppelgangersareplaying“thesame”game,eventhoughwhatwearedoingwouldlookexactlythesameifseenontelevision.Forexample,werespondtoandactuponwater(H2O),whileourTwinsdonot,butdealwithXYZ;differentrulesaltogether,yousee.(ThisactuallywasSellars’originalintention,thoughhehadnotyetheardofPutnam’sTwinEarth.)OBJECTION2Propernamesposeaproblemforthe“use”theorist.Trystatingaruleofuseforthename“WilliamG.Lycan,”orforthenameofyourbestfriend.Remember,ithastobearulethateverycompetentspeakerofyourlocaldialectactuallyobeyswithoutexception.Theonlycandidaterulesthatoccurtomepushthe“use”theoristintoadescriptiontheoryofmeaningfornames.Wittgensteinhimselffounddescriptivismcongenial,buthehadnotreadKripke.OBJECTION3TheWittgensteiniantheoryseemshelplessinthefaceofouroriginaldatum:speakers’amazingabilitytounderstandlong,utterlynovelsentencesatfirsthearingwithoutamoment’sthought.Chesspiecesandthelikearefamiliar,recurringtypesofobject,andtherulesfortheirusearedirectedone-to-oneuponthem.Andsimilarlyfor“Slab,”“Hello,”“Ouch,”“Ido,”andotherWittgensteinianexamplesofexpressionswhoseusesaredefinedbylocalritualsorcustoms.Butourabilitytounderstandandactonlongnovelsentencescannotbetheproductofourknowingconventionsdirecteduponthoseutterances,fornoconventionshaveeverbeendirecteduponthoseutterances.TheWittgensteinianmustgrantthatweunderstandnovelsentencescom-positionally,invirtueofunderstandingtheindividualwordsthatoccurinthemandworkingoutthesentences’overallmeaningsfromthewayinwhichtheindividualwordsarestrungtogether.(Weshallhaveagreatdealmoretosayaboutthisinchapter9.)Itfollowsthatwhatisunderstood,thatis,asentence’smeaning,isnotsimplyamatteroftherebeingconventionalnormsdirecteduponthatsentence’sdeployment,forthesentence’smeaningisinlargepartafunctionofitsinternalstructureaswell.\n“Use”theories81OBJECTION4CouldInotknowtheuseofanexpression,andfallinwithit,mechanically,withoutunderstandingit?Ihaveknownundergraduateswhoaregeniusesatpickingupacademicjargonofonesortoranotherandslingingitaroundwithgreatfacility,butwithoutunderstanding.Iknewonewhotookaphe-nomenologycoursetaughtbyavisitingParisian,understoodnoneofit,butlearnedtheknackofstringingthejargonexpressionstogethersowellthathistermpaperearned(or“earned”)anA.Useperfect(oratleastgradedA);meaningnil.OBJECTION5Manyrule-governedsocialactivities—sportsandgamesthemselvesinpartic-ular—donotcentrallyinvolvethekindofmeaningthatlinguisticexpressionshave.Certainlychessmovesandtennisshotsdonothavemeaningofthatsort.(Contrastthecasewherespiesareusingchessmovesasanactualsecretcode;forexample,N–Q3mayhaveconventionallybeenstipulatedtomean“TakethezircontoFoppaandtellhimwemovetonight.”)What,thenissupposedtodistinguishlanguage-gamesfromordinarygames?Supposesomecommunityagreestousecertainwords—oratanyratesoundsandmarks—inapeculiarway;saytheydecidetoputonly“words”withthesamenumberofsyllablesnexttoeachotherinthrees,ortheyutter“sentences”onlyinrhymingpairs,whereeachstringbeginswithaone-letterwordandaddsonelettersuccessivelytoeachensuingitem.(Thismightbeasortofcommunity-wideparlorgame.)Ifanewcomerhappeneduponthiswhimsicalsocietyandknewnothingofthearrangement,s/hewouldnotunderstandwhatwasgoingon.Thenewcomermight,intime,workoutalltherulesaccordingtowhichthevarioustokenswerebeingused,andyethavenonotionwhat,ifanything,wasbeingsaid.Andinthissimplecase,atleast,nothingisbeingsaid.Someonemightsuggestthatsuchagame,likethebuilders’language,isjusttoosimpleand/orprimitivetoqualify.Butitishardtoseehowthemereadditionoffurthercomplexitywouldhelp.REPLYSomeonemightarguethat,ifitsrulesarerichenoughandadvertoftenenoughtoambientconditions,referenceandpredicationwillberecoverablefromthegamedescription.Supposethereisarulethat,wheneverthewaitercomesin,everythirdplayershouts“Here,waiter,”andisgivenamartini;wheneveranyplayersays“Mixplease,”s/heispassedthebowlofsnacksbywhoeverisnearestit;andthelike.Onewouldthenbetemptedtoconcludethat“waiter”referstothewaiterand“mix”meanssnackfood.Sothegamemoveswouldhavemeaningafterall.\n82TheoriesofmeaningREjOINDERPerhaps,inthatcase,theutterancesspecifiedbythegameruleswouldhavemeanings—butonlybecausetheydostandfororrefertothingsandnotjustbecauseoftheirconventionaldeploymentbehavior.Letusthereforestipulatethat,nomatterhowcomplexthegamebecomes,theplayers’utterancesdonotrefertothingsexternaltothegame;theyareonlymovesinthegame.Butthenitseemsevenmoreobviousthatthegameisnoteventhebeginningofanactuallanguage,andthatthemovesdonothavemeaninginthesamewaythatutterancesofEnglishsentencesdo.Sothe“use”theorist’sexplicitconditionsarenotsufficientforsomething’sbeingalanguage.SECONDREPLYWaismann(1965a:158)anticipatesanobjectionofthiskind.Hehintsatacompetingreply:thatgenuinelanguage-gamesare“integrat[ed]...intolife.”Bycontrast,theparlor-gamewords,likechessmovesandtennisshots,“bearafarlesscloserelationtolifethanwordsusedinearnest.”Alanguage-gamecannotbeencapsulated,somethingthatwekeepatarm’slengthandplayjustwhenwefeellikeit.REjOINDERButsomelanguage-games,suchasthetellingofshaggy-dogjokes,areencapsulatedandplayedonlyoccasionallyandatwill.Also,evenifweagreethatmoreserious,multipurposelanguage-gamesarethoroughlyintegratedintolife,weusuallythinkofthatclose,integrativerelationasthatofrefer-ring,asourwordsbeingaboutthethingsintheworldthatconcernus.TheWittgensteiniandoesnotagreethatmeaningessentiallyinvolvesreferring,andsoWaismannneedstosaywhatthe“integration”isinstead.Theideaseemstobethatlanguage-gamesareintegratedwithothersocialpractices.ButitishardtoseehowtheWittgensteiniancanspellthatout(a)insuchawayastoexplainhowthelinguisticmovestakeonpropositionalcontent,but(b)withoutsecretlyintroducingreferring.Myusejustnowofthephrase“propositionalcontent”maysuggestanunsuccessfullytacitallegiancetothePropositionTheory.ButIamusingit,andwillcontinuetouseitthroughoutthisbook,inaweakersense,aswhateverpropertyofasentenceorotheritemissomehowexpressedbya“that”clause,asin“meansthatbroccoliwillkillyou.”Weneednottakethatpropertytobeamatterofbearingthe“expression”relationtoanabstractentitycalled“aproposition.”\n“Use”theories83OBJECTION6Oneclearsenseinwhichasocialpracticequalifiesasanactuallanguageisthat,accordingtoit,onecanmakenoisesorinscribemarksandtherebysaythatPforsomesuitablesentencereplacingP.Andoneofthethingsthatissurelyessentialtolanguageisthatwecansaythingsinit.Butnosuchindirectdiscourseislicensedjustinvirtueofsomepeople’splayingchessortheparlorgame;noneoftheplayershassaidoraskedorrequestedorsug-gested...thatanythingatall.Thereissomethingmissing.Weareplayingagame,andusingtokensaccordingtoasetofconventionalrules,andengaginginasocialpracticethatmaynotonlybefunbuthavesomelargerpoint;itmightevenbeinsomewayvitaltoourwayoflife.Thethingstheplayersinthesevariousgameshavedonemayhavesignificanceinsomesense,butnobodyhasmadeanyassertionsoraskedanythingoradvisedanyonetodoanything.InferentialismAtthispointitistemptingtomakesomeseriousconcessiontotheReferentialTheory.ButthatwouldbetooverlookthemostrecentincarnationofSellars’InferentialTheory:Brandom(1994),a700-pagemonsterpiece,whichatleasthasthepotentialforevadingsomeoftheforegoingobjections.Brandomdevelopsaparticularconceptionof“use,”anormativeconceptionaccord-ingtowhichasentence’suseisthesetofcommitmentsandentitlementsassociatedwithpublicutteranceofthatsentence.Hisparadigmisthatofasserting,consideredasanactualsocialact:whenoneuttersasentenceandtherebymakesanassertion,oneiscommittingoneselftodefendthatasser-tionagainstwhateverobjectionorchallengemightbemadebyahearer.Thedefensewouldtaketheformofgivingreasonsinsupportoftheassertion,typicallybyinferringitfromsomeothersentencewhoseutteringisnotsoreadilyopentochallenge.Andinmakingtheassertiononealsoconfersononeselftheentitlementtomakefurtherinferencesfromit.Thesocialgameofgivingandaskingforreasonsisgovernedbyrules,ofcourse,andscoreiskept.(NotionsofscorekeepingplayalargeroleinBrandom’ssystem.)ItisthereasonsthatwouldrightlybeofferedinsupportofasentenceS,andthenormsaccordingtowhichScouldrightlybegivenindefenseoffurthersentences,thatconstituteS’smeaning.AswouldpleaseWittgenstein,referenceplaysnoimportantroleinthistheory.ForBrandom,referenceisjustaconstructoutofinferentialpracticesdefinedonwholesentences,andnotafitsubjectforindependenttheorizing;theCausal–HistoricalTheoryisentirelybesidethepoint.(However,thisexacerbatesobjection2.)Brandom’ssystemisverycomplex,andwecannotexamineithere.ButInotethatitovercomessomeoftheobjectionsraisedsofaragainstthe\n84TheoriesofmeaningWittgensteinianview.Againstobjection5,itdoesdistinguishlinguisticutterancesfrom“Slab,”chessmoves,andsoon,sincethosearenotthesortsofthingsinsupportofwhichonegivesreasons,rebutschallenges,andsoon.(Onecanofcourseofferpracticalreasonsforhavingmadeaparticularchessmoveortennisshot,butBrandommeansevidentialreasons,utterancesthatgiveusreasontobelievesomestatementoffact.Again,hisparadigmisthatofaninferentialreason,andchessmovesandthelikearecertainlynotinferences.)Norisobjection6aproblem,forSellarshimselfgaveanelegantinferentialistaccountofthatclauses.ThoughBrandomholdsthatsubsen-tentialexpressions“havemeanings”onlyderivativelyfromwholesentencemeanings,healsorecognizesaweakkindofcompositionality,andsohemayevadeobjection3.And,admirably,headdressessomefairlydetailedseman-ticphenomena:propernames,descriptions,indexicals,quantification,andanaphora,intermsoftheircharacteristiccontributionstothecommitment/entitlementpotentialsofsentencesinwhichtheyoccur.4Inanycase,thecentralityofepistemologicalnotionstoinferentialisttheories—defense,support,justification,acceptance—suggeststhatthetheoriesarecloserinspirittoverificationaccountsthantoWittgenstein’soriginalidea.Seechapter8.Asomewhatdifferentsortof“use”theory(Alston1963,2000;Barker2004)isbasedonJ.L.Austin’snotionof“illocutionaryforce.”Butthelatterconceptwillnotbeintroduceduntilchapter12.Nowletusmoveonandlookataconsiderablydifferenttheoryofmean-ing.PaulGrice’stheorybeginswiththeoutrageousnotionthatlanguageisameansofcommunication.Summary•“Use”theorieshaveitthat“meanings”arenotabstractobjectslikepropositions;alinguisticexpression’smeaningisdeterminedbytheexpression’scharacteristicfunctioninhumansocialbehavior.•AccordingtoWittgenstein,linguisticexpressionsarelikegametokens,usedtomakemovesinrule-governedconventionalsocialpractices.•Sellars’versionofthisideamakestheactofinferringcentral,anditisthecomplexityofpatternsofinferencethatallowsthe“use”theoristtoaccommodatelong,novelsentences.•“Use”theoriesfacetwomainobstacles:explaininghowlanguageusediffersfromordinaryconventionalrule-governedactivitiesthatgeneratenomeaning;andexplaininghowasentencecanmeanthatso-and-so.•Brandom’s“use”theoryovercomessomeoftheseobstacles.\n“Use”theories85Questions1CantheWittgensteinian“UseTheory”aswehavesketcheditbedefendedagainstoneormoreofobjections1–4?2Adjudicateobjection5.CanyoumakeabetterreplythanWaismann’s?3ComeupwithaWittgensteinianreplytoobjection6.4Canared/greencolor-blindpersonunderstandtheword“red”?Thinkaboutthisinregardto“use”theories.5IfyouhavereadBrandom,discusshisviews.Furtherreading•TheliteratureonWittgensteinissovastthatIhesitatetomentiononeortwoorthreeexegeticalworkstotheexclusionofothers.But:Rhees(1959–60);Pitcher(1964:ch.11);Hallett(1967);Kenny(1973:chs7–9).•ThelocusclassicusofSellars’FunctionalTheoryisSellars(1963);seealsoSellars(1974).AnexcellentexpositionanddefenseofthecentralthemesisgiveninRosenberg(1974).•PhilosophyandPhenomenologicalResearch57(1997)containsasymposiumonBrandom(1994),withaprécis,leadpapersbyJohnMcDowell,GideonRosen,RichardRorty,andJ.F.Rosenberg,andaresponsebyBrandom.Brandom(2000)isamoreaccessibleretroduc-tiontoBrandom1994).\n7Psychologicaltheories:Grice’sprogramOverviewH.P.Gricemaintainedthatalinguisticexpressionhasmeaningonlybecauseitisanexpression—notbecauseit“expresses”aproposition,butbecauseitmoregenuinelyandliterallyexpressessomeconcreteideaorintentionofthepersonwhousesit.Griceintroducedtheideaof“speaker-meaning”:roughlywhatthespeakerinutteringagivensentenceonaparticularoccasionintendstoconveytoahearer.Sincespeakersdonotalwaysmeanwhattheirsentencesstandardlymeaninthelanguage,Gricedistinguishedthisspeaker-meaningfromthesentence’sownstandardmeaning.Heofferedanelaborateanalysisofspeaker-meaningintermsofspeakers’intentions,beliefs,andotherpsy-chologicalstates,andrefinedthatanalysisinthelightofmanyobjections.Itisgenerallyagreedthatsomeversionoftheanalysismustberight.Moreimportantlyforourpurposes,Gricealsoofferedananalysisofasentence’s(standard)meaningintermsofspeaker-meaning.Herehefacesseveredifficulties,sincethereareseveralwaysinwhichsentencemeaningobstinatelyrefusestocooperatewithspeaker-meaning.Gricehasawayofovercomingsuchobstacles,butitseemslikelythatthatwayconcedestoomuchtocompetingtheoriesofsentencemeaning.Grice’sbasicideaWeareconcernedtoarriveatanaccountofmeaning,meaningconsideredasaremarkablefeatureoflinguisticexpressions,ofsentencesinparticular.Butsupposeweaskourselves,whataresentencesreally?Theyaretypesofmarksandnoises,individualtokensofwhichareproducedbypeopleonparticularoccasionsforapurpose.Whenyousaysomething,itisusuallyforthepurposeofcommunicating.Youdeliveryourselfofanopinion,orexpressadesireoranintention.Andyoumeantoproduceaneffect,tomakesomethingcomeofit.Soonemightbeginastheideationaltheoristsdidandinferthattherealnaturalgroundofmeaningfulutteranceisinwhatmentalstateisexpressedbytheutterance.Ofcoursewehavealreadyintroducedtheword“express”asdesignatingarelationbetweensentencesandpropositions,butheretheterm\nPsychologicaltheories:Grice’sprogram87hasamoreconcreteandliteraluse:sentencetokensareseenasexpressivelyproducedbyspeakers’beliefs,desires,andotherpropositionalattitudes.Grice(1957,1969)tookthesefactsasthebasisofhistheoryofmeaning.Hebelievedthatsentencemeaningisgroundedinthemental,andproposedtoexplicateitultimatelyintermsofthepsychologicalstatesofindividualhumanbeings.Wecanthinkofthisasnolessthanthereductionoflinguisticmeaningtopsychology.ThelinchpinofGrice’sprojectwasaslightlydifferentnotionofmeaning,whichdoesnotcoincidewiththatofsentencemeaning.(Thisishiscrucialdeparturefromtheclassicideationaltheories.)Herearethreeexamplestoillustratethedifference.First,recallStrawson’ssentencefromchapter2,“Thisisafineredone.”Aswesaw,themeaningofthatsentenceitselfisnotfullydeterminate;tounderstandit,weneedtoknowwhatthespeakerispointingto.Onespeakerinonecontextmaymeanthatthepearinher/hishandisafineredpear,whileadifferentspeakeronadifferentoccasionmaymeanthatthethirdfire-engineontheleftisafinefigureofaredfire-engine.Second,supposethatlikesomeunfortunatesIincorrectlybelievethattheword“jejune”meanssomethinglikecalloworpuerile,1andIsay“Mozart’s‘Piccolomini’Massisjejune,notgoodMozartatall,”meaningthatthe‘Piccolomini’Massiscallowandpuerile.But“jejune”actuallymeansmeagerandunsatisfying(itisfromtheLatinwordforfasting);thesentenceIutteredmeansthattheMassismeagerandunsatisfying,whichIwouldjudgetobefalseeventhoughIdofindtheMasscallowandpuerile.Third,considersarcasm,aswhenonesays“Thatwasabrilliantidea”,meaningthatsomeone’sideawasverystupid.Heretoo,wegetadivergencebetweenthemeaningofthesentenceutteredandwhatthespeakermeantinutteringit(sincethespeakermeanspreciselytheopposite).Themoralisthatwhataspeakermeansinutteringagivensentenceisaslightlydifferentkindofmeaningfromthesentence’sownmeaning.Gricecalledit“utterer’smeaning”;itisalsowidelycalledjust“speaker-meaning.”2Now,letusturntoGrice’sreductiveproject,theexplicationofsentencemeaninginpsychologicalterms.Itproceedsintwoimportantlydifferentstages.Inthefirststage,3Griceattemptstoreducesentencemeaningtospeaker-meaning.Inthesecond,hetriestoreducespeaker-meaningtoacomplexofpsychologicalstatescenteringonatypeofintention.Onthefaceofit,thefirststageisaplausibleidea.AsWittgensteinempha-sized,itisverystrangetothinkofsentencesashavingmeaningsontheirownandintheabstract,asopposedtothinkingofsentencesashavingmean-ingbecauseofwhatspeakersusethemtodo.Itdoesseemthatlinguisticexpressionshavetheconventionalmeaningstheydoonlyinvirtueofhumancommunicativepractices,andthatcommunicative“practices”boildowntosetsofindividualspeakers’communicativeacts.Griceamendsthatlastphrase,focusingonwhatspeakersusesentencestomean,inthesenseofwhatthespeakersmeaninutteringthesentenceswhentheydoutterthem.ForGrice,asentence’smeaningisafunctionofindividualspeaker-meanings.\n88TheoriesofmeaningButGriceconcentratedhisenergiesonthesecondstageofthereduc-tion.Thatspeaker-meaningshouldbeexplicatedintermsofmentalstatesisevenmoreplausiblethanthefirststage.If,whenIsay“Thatwasabril-liantidea,”whatImeanisthatSmedley’sideawasverystupid,surelythatspeaker-meaningissomethingpsychological,somethingaboutmymentalstate.Presumablyitisamatterofmycommunicativeintention,ofwhatIamintendingtoconveytoyou.Itdoesseemthat,ingeneral,individualcommu-nicativeactsareamatterofspeakers’havingcomplexintentionstoproducevariouscognitiveandotherstatesintheirhearers.Speaker-meaningLetusstartwithaplausibleandperhapsneedlesslyspecificversionofGrice’ssecond-stageanalysis,whichskipsoversomeoftheearlyfootworkcontainedinoroccasionedbyhisoriginal(1957)article.(Iofferaparaphraseratherthanadirectquotation,toavoidsomeofGrice’sownslightlytechnicaljar-gonandsomecomplications.)4Wewanttoexplicatestatementsoftheform,“Byutteringx,SmeantthatP,”asin“Byuttering‘The“Piccolomini”Massisjejune’,Lycanmeantthatthe‘Piccolomini’Massiscallowandpuerile.”Theanalysisrunsasfollows:(G1)SutteredxintendingthatAformthebeliefthatP[whereAisS’sheareroraudience]and(G2)SfurtherintendedthatArecognizeS’soriginalintention[asdescribedinG1]and(G3)SstillfurtherintendedthatAformthebeliefthatPatleastpartlyonthebasisofrecognizingthatoriginalintention.Thus,inourMozartexample,byuttering“The‘Piccolomini’Massisjejune,”ImeantthattheMassiscallowandpuerile,becauseIuttereditintendingthatyouformthebeliefthattheMassiscallowandpuerileatleastpartlyonthebasisofyourrecognizingthatIhadthatveryintention.Asadvertised,thecoreofspeaker-meaningisanintention,butothermen-talstatesfigureintheanalysisaswell,namelytheintendedfuturebeliefofyoursandtheintendedstateofrecognition.Itmaybethoughtimplausiblethatanordinaryspeakercouldhavesuchcomplexintentionsatall,muchlesshavethemeverysingletimes/hemakesanassertion.ButGriceisnotsupposingthatthesecommunicativeintentionsareconscious,orbeforethemind.Indeed,indailylifemostofourintentionsareonlytacit;weareonlyoccasionallyawareofthem.Sotoo,youusuallysaythingswithoutexplicitlythinkingaboutit,andoftenyouspeaker-meanthingsthatyouareunawareof.\nPsychologicaltheories:Grice’sprogram89Theforegoingsecond-stagetheoryhasbeenundernearlyconstantrevisionsince1969,inresponsetocounterexamplesofseveralkinds.Ishallreviewafewoftheobjectionsandrevisions,justenoughtogiveyoutheflavorofthissubproject.OBJECTION1Speaker-meaningdoesnotinfactrequireanactualaudience.SupposeIamgiventosoliloquizing.WhenIhaveaproblem,practicalortheoreticalorpersonal,Iworkthroughittalkingaloudtomyselfintheprivacyofmybase-mentbatcave.NotonlydoIintendnoeffectonanyaudience,IwouldbemortifiedifIweretofindoutthatsomeonehadbeenlistening.OrconsiderPaulZiff’s(1967:3–4)protagonistGeorgeandthesentence,“Claudiusmur-deredmyfather”:inasingleday,Georgemightutterthatsentencefirst“inthecourseofamorningsoliloquy,”again“intheafternooninthecourseofaconversationwithJosef,”andthenagain“intheeveningwhiledeliri-ouswithfever”andunawareofhisaudienceeventhoughtherewasone.YetGeorgemeantthesamethingby“Claudiusmurderedmyfather”eachtime.ButGrice’sanalysisrequiresnotonlyanaudiencebutthatthespeakerhaveveryspecificintentionswithrespecttothataudience,andthisisimplausibleatleastforthesoliloquyanddeliriumcases.Grice(1969:sectionV)addressestheaudiencelesscases.Heurgesasolu-tionintermsofhypotheticalorcounterfactualaudiences:ineffect,aspeakershouldintendthat,wereanyonepresentandenjoyingnormalperceptualandotherpsychologicalconditions,thatpersonwouldformthebeliefthatP.NeedI,asaspeaker,intendthis?Perhapsso,sincewhenIspeakeventomyselfImustassumethatwhatIsaywouldmakesensetosomeone.Ontheotherhand,furtherpotentialcounterexamplescometomind.SupposeIgrewuponadesertisland,andsomehowputtogetheralanguageallbymyself;yetIneverformedtheconceptof“anotherspeaker”orofan“audience.”ThenIcouldnotintendanythingaboutanaudienceevencounterfactually.Butthisisahighlycontroversialcase,sincemanyphilosophershavedeniedthatitwouldbeevenfaintlypossibleformetomakeupmyownlanguagewithouthavingformedtheconceptofspeakersandaudiences.OBJECTION2Evenwhenthereisanactualaudience,thespeakermaymeansomething,yetnotintendtoproducebeliefbymeansofintentionrecognition;requirements(G3)andeven(G2)maybetoostrong.Orthespeakermaynotevenintendtoproducethebeliefatall,sinceher/hisaudiencealreadyhasthebeliefinquestionandisknownbythespeakertohaveit.Hereisanexampleoftheformertypeofcase.Conclusionofargument:oneoffersanargument,perhapsproducesaproofofageometricaltheorem.Onecertainlyspeaker-meanstheargument’sconclusion,butdoesnotintendone’saudiencetoreachthatconclusioneveninpartonthebasisofrecognizing\n90Theoriesofmeaningone’soriginalintention.Onemayfirmlyintendthemnottodoso,butrathertoformthebeliefonthebasisoftheargument’smeritalone.Schiffer(1972:79–80)approachesthe(allegedly)audiencelesscases,andConclusionofargumentaswell,bystipulatingthatthespeakerisher/hisownaudience.(Ipersonallycannotdismissthisasfanciful,sinceithasbeensaidofmethatIoftenproduceutterancesforthesimplegratificationofhear-ingmyselftalk.)5Thismovemightdo,butforcasesofthesecondtype.AnexampleofthisisExaminee:Apupilwhocorrectlyanswersanexamquestionmeans,forexample,thattheBattleofWaterloowasfoughtin1815,butdoesnotintendtoinducethatbeliefintheexaminer(s).Gricemakesessentiallytworevisionsinresponsetotheseandaswarmofothercounterexamples.First,hesuggestsinvokingtheconceptof“activated”belief:thoughsomeoftheaudiencealreadybelievewhatthespeakerhasinmind,theirbeliefsmaynotbefullyconsciousandpsychologicallyactive,orevenconsciousatall.Ifwebeefup(G1),therequirementthattheaudiencebeintendedtobelievethatP,todemandthatSintendstoproduceactivatedbeliefinA,thatmayaccount(thoughnotverynaturally)forExaminee;itdoesbetteragainstsomeoftheothercases.Grice’ssecondrevisionisalsotoamend(G1),thistimereplacingitbytheweakerprovisionthattheaudiencebeintendedtobelieveonlythatthespeakerbelievesthatP.(Weakening(G1)inthiswayiscompatiblewithhav-ingstrengthened(G1)torequireactivatedbelief.)Thissecondrevisionseemsreasonable.AsGricesays,itdealsbrisklywithExaminee.Anditisnotimplausible.Tosaysomethingandmeanit,wemighthold,ismerelytoexpressabelief,usuallybutnotalwayshopingorintendingorexpectingthatone’saudiencewillcometosharethebelief.(Whenweinformpeopleofthingsbytellingthemthosethings,wenormallyexpectthatinformingtoworkbywhatinformallogicianscall“authority”:ourhear-erstakeourwordforwhatwearesayingandbelieveitbecausewedo.)YetasGricegrantsandSchifferemphasizes(p.43),Conclusionofargumentisnotalleviatedbyeitherthefirstorthesecondrevision.Moregenerally,notallcasesofcommunicationsucceedbecausetheaudiencetakesthespeaker’sword.Recallthegeometricalproof.Foranexampleclosertohome,Gricehimselfhascommunicatedhistheoryofmeaningtous,butnotbyvirtueofhavingintendedustoacceptitonthestrengthofhissay-so.ItistruethatwehavecometobelievethatGricebelieveshistheoryofmeaning,sothenewlyweakenedversionof(G1)issatisfied;butthatdoesnothelphere.(WecannotreallyevenassumethatGricedoesbelievethetheory;Iamafraidphilosophersarealwayswritingarticlesdefendingviewsthattheydonotthemselvesreallybelieve.)WhataboutSchiffer’sresponsetoConclusionofargument,thatthespeakerisher/hisownaudience?Ithinktherearestillcounterexamplesofthesametype.SupposeIproduceasecondproofofmytheoremwhilethefirstisstillsittingontheblackboard.Idonotinduceabeliefinmyself,orevenactivateonethatIalreadyheldquiescently.Another:Supposetwophilosophersare\nPsychologicaltheories:Grice’sprogram91havingalove-feastovertheDirectReferenceviewofpropernames.Theyaredancingaroundinacircleandshoutingjoyfullyateachother,“Namesonlyname!”overandover.Eachisinastateoffullyactivatedbeliefinthetruthofthisdubiousassertion,andeachknowsthattheotheris;soneithercanbesaidtohavetheintentionofeitherproducingoractivatingbeliefintheother.Yetsurelytheymeanbytheirutterancethatpropernamesonlyname;itisnotanonsensechant.Furthermovesareavailablehere,6butIshalldropobjection2atthispoint.ThefirsttwoobjectionswereintendedtoshowthatGrice’sanalysisistoodemanding.Thenexttwoaretotheeffectthatinotherrespectstheanalysisisnotdemandingenough.OBJECTION3Onbeinginductedintothearmy,Georgeiscompelledtotakeatestdesignedtoestablishsanity.Georgeisknowntobeanirritableacademic.Thetestheisbeinggivenwouldbeappropriateformorons.Oneofthequestionsaskedis:“Whatwouldyousayifyouwereaskedtoidentifyyourself?”Georgerepliestotheofficeraskingthequestionbyuttering“Ughblughblughughblugh”.(Ziff1967:2)Georgemeanttoshowhiscontempt,andmeanttheofficertorecognizehiscontemptonthebasisofrecognizinghisintentiontoshowit.But,althoughGrice’sconditionsaremet,Georgedidnotmeananythinginanylinguisticsense(thoughonemightcorrectlypointoutthatthereisawidersenseof“communication”thatGrice’sanalysisstillseemstocapture).7OBJECTION4DuringWorldWarIIanAmericansoldieriscapturedbyItaliantroops.HewantstogettheItalianstoreleasehim,byconvincingthemthatheisaGermanofficer.ButhedoesnotknoweitherGermanorItalian.HopingthathiscaptorsdonotknowGermaneither,he“asitwere,attemptstoputonashowoftellingthemthatheisaGermanofficer,”byofficiouslybarkingouttheonlyGermansentenceheknows,alineofpoetryhehadlearnedinschool:“KennstdudasLandwodieZitronenblühen?”(“Doyouknowthelandwherethelemontreesbloom?”)8(Searle1965:229–30).HerethesoldierutteredhissentenceintendingtogettheItalianstobelievethatheisaGermanofficer;hefurtherintendedthemtorecognizethatorigi-nalintention;andhestillfurtherintendedthemtoformthefalsebeliefinpartonthebasisofrecognizingtheintention.Butitdoesnotseemthatinsaying“KennstdudasLand...,”hemeansthatheisaGermanofficer.\n92TheoriesofmeaningGricerespondsbyrequiringthattheaudiencebeintendedtobelievetheretobea“modeofcorrelation”betweenfeaturesoftheutteranceandtheintendedbelieftype.Schiffer(1972)makesadifferentmove,intermsofhistechnicalnotion“mutualknowledge*.”Itseemsbetternottoforgeonintothesearcanafornow.DeterminedGriceanssuchasSchiffer(1972)andAvramides(1989)haveshownextraordinaryfortitudeandskillinmodifyingGrice’soriginalaccountinsuchawayastoaccommodatealltheforegoingproblemcasesandmore,withtheresultthat,despitetheprofusionofobjections,acom-plicated(!)versionofthetheoryremainstenable.Anditisgenerallyagreedthatspeaker-meaningmustbeinsomewayamatterofspeakers’intentionsandothermentalstates.ButnowwemustturnbacktothefirststageoftheGriceanprogram,thereductionofsentencemeaningtospeaker-meaning.SentencemeaningAsmaysurpriseyouafteryouhavereadtheprevioussection,Grice’s(1968)constructionofsentencemeaningoutofspeaker-meaningiselaborateandfulloftrickydetails.Ratherthanplungeintothem,Ishallrevealsomeobstaclesinadvance.ThenIshallonlyoutlinetheway(s)inwhichGricetriestosurmountthem.ItwouldbenaturaltostartbysupposingthatagivenEnglishsentencemeansthatPonlyinthesensethatwhenspeakersofEnglishutterthatsen-tence,theyalwaysoratleastnormally(speaker-)meanthatP.Butherecometheproblems.OBSTACLE1Ziff(1967)offeredthefollowingtwoexamples:Georgehashadhisheadtamperedwith:electrodeshavebeeninserted,platesmounted,andsoforth.Theeffectwascurious:whenaskedhowhefelt,Georgerepliedbyuttering...“Glytingellybeleg”.Whathemeantby[that],helaterinformedus,wasthathefeltfine.Hesaidthat,atthetime,hehadsomehowbelievedthat[“Glytingellybeleg”]wassynonymouswith“Ifeelfine”andthateveryoneknewthis.(pp.4–5;bynowyouwillhavegraspedthatGeorgeleadsalifemoreinterestingthanyoursormine)Amansuddenlycriedout“Gleeggleeggleeg!”,intendingtherebytoproduceacertaineffectinanaudiencebymeansoftherecognitionofhisintention.HewishedtomakehisaudiencebelievethatitwassnowinginTibet.Ofcoursehedidnotproducetheeffecthewasaftersincenoonerecognizedwhathisintentionwas.Nonethelessthathehadsuchanintentionbecameclear.Beingdeemedmad,hewasturnedoverto\nPsychologicaltheories:Grice’sprogram93apsychiatrist.Hecomplainedtothepsychiatristthatwhenhecried“Gleeggleeggleeg!”hehadsuchanintentionbutnoonerecognizedhisintentionandweretheynotmadnottodoso.(p.5)InthetextitisnotclearwhetherZifftookthesecasestobecounter-examplestoGrice’sanalysisofspeaker-meaning.ButIdonotunderstandhiminthatwayandIdonottakethemassuch.Itseemstomethat,inhisalteredstate,Georgedidmeanthathefeltfine;andthemadmanderangedlymeantthatitwassnowinginTibet.Rather,Itakethepointtobethat,ifGrice’stheoryofspeaker-meaningiscorrect,thenspeaker-meaningcomesverycheap:givenasuitablydisorderedmentalstate,anyspeakermightmeananythingatallbyanystringofnoisess/hehappenstoutter.IfGrice’sanaly-sisofspeaker-meaningiscorrect,thenalltheworseforthefirststageofhisproject,fortherewillthenbenoformalconstraintonwhatspeakersmightmeanbyanysentencetheyutter,butonlystatisticsabouthowoftenspeakersdomeanthisorthat.Inreallife,ofcourse,speaker-meaningisnotsoeasilyhad,fortworeasons.(a)MostpeoplearenotderangedinthemannerofZiff’spatients.Farmoreimportantly,(b)Englishsentenceshavethemeaningstheydohave,andonecannotjustmeananythingbythemonelikes.UnlessIamoddlymistakenaboutthemeaningoftheworditself,orsomemoreelaboratestage-settingisinplace,Icannotsay“It’scoldhere”andmeanbyit“It’swarmhere.”(TheexampleisWittgenstein’s.)Icouldbebeingsarcastic,ofcourse.ButIcouldnotverywellmean“IhavejustrentedthevideoofAgnesofGod,”or“Pigshavewings.”Theantecedentmeaningofasentencepartlycontrolswhataspeakercanmeanbyitinagivencontext.Reason(b)furtherembarrassesGrice’sfirststagesince,ifsentencemean-ingistobeanalyzedentirelyawayintospeaker-meaning,weshouldnothavetolooktosentencemeaningasconstrainingpossiblespeaker-meanings.(Perhaps“shouldnot”istoostrong.Thereisnoflat-outcircularityhere;anditiscertainlypossiblethatonespecialconstructoutofspeaker-meaningmightconstrainspeaker-meaningingeneral.ButtheGriceanwillstillhavetoexplainwhythishappenssorobustly.)OBSTACLE2Mostmeaningfulsentencesofalanguageareneverutteredatall.Thereforenoonehasevermeantanythingbythem.Thereforetheirmeaningscanhardlybedeterminedbywhatspeakers(normally,typically,andsoon)meanbythem(Platts1979:89).Itisnotmuchuse,thoughtempting,toappealtowhatspeakerswouldhavemeantbytheunutteredsentenceshadtheyutteredthem.Foronething,thevastmajorityofthosesentencesareonesthatthespeakerswouldneverhaveuttered.Evenforasentencethatthespeakersmighthaveutteredeven\n94Theoriesofmeaningthoughtheydidnot,theonlyhandlewehaveonwhatthespeakerswouldhavemeantinutteringitiswhatwealreadyknowthatsentencetomean.OBSTACLE3Novelsentencesagain.Evenwhenasentenceisactuallyuttered,itmaybewildlynovel,yetinstantlyunderstoodbyitsaudience.Butifitisnovel,thenthereis(asbefore,independentlyofwhatweknowthesentenceitselftomean)nopre-establishedfactofwhatspeakersnormallymeanorwouldnormallymeanbyit.Andnoticethatthefirst,novelusemaybe(a)alsothelastand(b)itselfnonliteral.(Iamprettysurethatthefollowingsentencehasneverbeenutteredbefore,thoughitmaybeutteredagain:“ThePresidentoftheUnitedStatesPhilosophyCorporation,whohasfinallybeenreleasedfromprisonandishurryingheretotheaviaryonwingedfeet,willsharetherichesofherspiritwithusat3:00p.m.tomorrow.”Insuchacase,eventhoughthesentencehadbeenuttered,noonewouldeveractuallyhavemeantbyitwhatitliterallymeans.)Blackburn(1984:ch.4)pointsoutthat,intherightcircumstances,agivensentencemaybeutteredwithpracticallyanyintentionandcertainlywithouttheintentionofdisplayingone’sactualbelief.(BlackburnbroachesthealternativeideathatasentenceSmeansPwhenitiseitheraconven-tionalregularityortheconsequenceofaconventionalregularitythatonewhouttersSwithassertiveforce“mayberegardedashavingdisplayed”thatP,thisregard-licensebeingasocialfactthatobtainsindependentlyofanyparticularutterer’sintentions.Thisisaninterestingidea,andcallsformuchunpackingof“may,”“beregarded,”and“display,”butitisnotaGriceanidea,foritself-consciouslyseverssentence-meaningfromspeakers’communica-tiveintentions.)OBSTACLE4Sentencesareoften,andnotjustabnormally,usedwithotherthantheirownliteralmeanings.Evenneglectingsarcasmandotherformsofindirectspeechacts(weshalltalkmoreaboutsuchthingsinchapter13),figurativeusageisveryprevalent(weshalltalkmoreaboutthatinchapter14).IfGriceshouldwanttosaythatasentence’sownmeaningiswhatspeakers“normally”meaninutteringthesentence,hewouldhavetosaywhat“normally”meansindependentlyofthesentence’sstandardmeaning,aswellasmotivatingtheclaim.Andthingsgetevenworse.Thereareprivatecodesinwhichagivensen-tenceisneverusedwithitsliteralmeaning.TheJapanesesignalforthe1941airattackonPearlHarborwas(theJapaneseexpressiontranslatedas)“Eastwind,rain,”whichsofarasIknowhasneverbeenusedtomeananythingbut“It’stimetogobombPearlHarbor.”Andevenapartfromprivatecodes,ineverydaylifetherearemanysentencesthatnormallyareutteredwithother\nPsychologicaltheories:Grice’sprogram95thantheirliteralmeanings,andperhapsareneverutteredwiththoseliteralmeanings.(“Allright,buddy,where’sthefire?”“Canyoutellmethetime?”“GeorgeandMarthaburiedthehatchet.”“Businessisbusiness.”)Andthereisthewholequestionofmetaphor,though,asweshallseeinchapter13,Gricehimselfthinksofmetaphorasaspeciesofwhathecalled“conversationalimplicature.”NowforasketchofGrice’sreductionofsentencemeaningtospeaker-meaning,andindicationsofhowhewouldhaveapproachedobjections1–4hadhebeenfullyawareofthem.9Hefirstconcentratesonthenarrownotionofsentencemeaningforaparticularindividual,thatisthemeaningthatthesentencehasinthatindividual’spersonal,distinctivespeechoridiolect.(NotwoEnglishspeak-ers’idiolectsareexactlyalike.)Andherestrictshisinitialtargetfurther,distinguishingstructuredutterancesfromunstructuredones.Astructuredutterancehasmeaningfulparts,suchasindividualwords,whichcontributetotheutterance’soverallmeaning;anydeclarativeEnglishsentenceisanexampleofthis,sinceitcontainswordsthatareindividuallymeaningfulanditmeanswhatitdoesinvirtueofthosewordsmeaningwhattheydo.Anunstructuredutteranceisasingleexpressionornonverbalgesture,suchas“Ouch”orabeckoningmotionthatmeans“Thisway,”whosemeaningisnotcompositionalinthatsense.(NotethatGriceusestheterm“utterance”verybroadly,asincludingnonverbalcommunicativeacts.)Aftersomebackingandfilling,Gricehypothesizesthatx[anunstructuredexpression]meansthatPinS’sidiolect,ifandonlyif(roughly)Shasinher/hisrepertoirethefollowingprocedure:toutterxif,forsomeaudienceA,SintendsAtobelievethatSbelievesthatP.(Thatlastclauseisasimplifiedversionof“Sspeaker-meansthatP”;Gricearguesthatthesimplificationisharmlesshere.)Nowheexpandsthisanalysistocoverutterancemeaningforagroupofspeakers:x[unstructured]meansthatPforgroupGifandonlyif(a)manymembersofGhaveintheirrepertoirestheprocedureofutteringxif,forsomeA,theywantAtobelievethattheybelievethatP;and(b)theretentionofthatprocedureisforthemconditionalontheassumptionthatatleastsomeothermembersofGhavethatsameprocedureintheirrepertoires.Ithinkwhatissupposedtoovercomeobstacle1isthecombinationof(a)and(b);thattherelevantprocedureiswidespreadinthecommunityandthatindividualmembersofthecommunityrelyontheothermemberstomaintainthatprocedureaswell.Thisseemsexactlyright.Butnowthetrickwillbetogofromtheanalysisofunstructured-utterancemeaningtoordinarysentencemeaning,sinceordinaryEnglishsentencesareallstructured.Gricebringsinthenotionofa“resultant”procedure.AtthispointGrice’sarticlebecomesdenseandobscure,butIthinktheideaisthis:JustasEnglishsentencesaremadeupofsmallermeaningfulparts—wordsandphrases—invirtueofwhichthewholesentencesmeanwhattheymean,anindividualspeakerwillhaveinher/hisrepertoireacomplex,abstract\n96Theoriesofmeaning“resultantprocedure”madeupoftheconcreteproceduresattachingtoitsrespectivecompositeparts.Thus,asentence’smeaningwillnotbedirectlyafunctionofspeaker-meaning,butratherafunctionoftheindividualutter-ancemeaningsofitsultimateparts.OnlythenwillthecoreGriceanidea,and(crucially)hisanalysisofutterancemeaningforagroup,beinvokedasexplicatingtheutterancemeaningsoftheparts.Iemphasize“abstractresultantprocedure,”becauseveryfewofthose“abstract”procedureswilleveractuallyoccur.AnditisthatfeaturethatwillhelpGricewithobstacles2–3.Forthethemeofthoseobstaclesisthatunut-teredandnovelsentencesdonotcorrespondtoanyactualspeaker-meanings.Butatleastarguably,theydocorrespondtothehypotheticalspeaker-mean-ingsthatwouldbegeneratedbyGrice’sabstractresultantprocedures.Theappealtoabstractproceduresmayalsohelptoovercomeobstacle4:Eventhoughacertainsentence’sliteralmeaningisnevermatchedbyanyactualspeaker-meaning,itmaystillcorrespondtoahypotheticalresultantspeaker-meaning.YetIbelievethatthisabsolutelynecessaryappealbetraysthespiritoftheGriceanprogram.Ineffect,itgivesthegameawaytoacompetingtheoryofmeaning;Ishallarguethatinchapter9.Summary•AccordingtoGrice,linguisticexpressionshavemeaningonlybecausetheyexpressideasorintentionsofthespeakerswhousethem.•“Speaker-meaning”is,roughly,whatthespeakerinutteringagivensentenceonaparticularoccasionintendstoconveytoahearer.•Griceoffersananalysisofspeaker-meaningintermsofspeakers’intentions,beliefs,andotherpsychologicalstates,andhastenablyrefinedthatanalysisinthelightofmanyobjections.•Gricehasalsoofferedananalysisofasentence’sownmeaningintermsofspeaker-meaning.•Thatanalysisovercomessomesevereobstacles,butseeminglyonlybyconcedingtoomuchtocompetingtheoriesofsentencemeaning.Questions1CanyouhelpGriceavoidoneormoreofobjections1–4?2CanyouthinkoffurtherobjectionstoGrice’stheoryofspeaker-meaning?3DiscussGrice’s“firststage”;willhiselaboratemethodofreducingsen-tencemeaningtospeaker-meaningwork?\nPsychologicaltheories:Grice’sprogram97Furtherreading•Schiffer(1972)istheclassicworking-outofGrice’sview.SeealsoGilbertHarman’sreview(1974a),andAvramides(1989).RelatedworksofGrice’sownarecollectedinGrice(1989).•Bennett(1976)isavaluabledefenseoftheGriceanprojectbyonewhowasnotaninsider.MacKay(1972),Black(1973),Rosenberg(1974:ch.2),andBiro(1979)arecriticalofGrice.\n8VerificationismOverviewAccordingtotheVerificationTheory,asentenceismeaningfulifandonlyifitsbeingtruewouldmakesomedifferencetothecourseofourfutureexperi-ence;anexperientiallyunverifiablesentenceor“sentence”ismeaningless.Morespecifically,asentence’sparticularmeaningisitsverificationcondition,thesetofpossibleexperiencesonsomeone’spartthatwouldtendtoshowthatthesentencewastrue.Thetheoryfacesanumberofobjections:ithasruledanumberofclearlymeaningfulsentencesmeaningless,andviceversa;ithasassignedthewrongmeaningstosentencesthatitdoescountasmeaningful;andithassomedubi-ouspresuppositions.Buttheworstobjectionisthat,asDuhemandQuinehaveargued,individualsentencesdonothavedistinctiveverificationcondi-tionsoftheirown.Quinewentontobitethatbulletandinferthatindividualsentencesdonothavemeanings;accordingtohimthereisnosuchthingassentencemeaning.Quinealsoattackedtheformerlywidespreadviewthatsomesentencesare“analytic”inthesenseofbeingtruebydefinitionorsolelyinvirtueofthemeaningsoftheircomponentterms.ThetheoryanditsmotivationTheVerificationTheoryofmeaning,whichflourishedinthe1930sand1940s,wasahighlypoliticaltheoryofmeaning.Itwasmotivatedby,andreciprocallyhelpedtomotivate,agrowingempiricismandscientisminphilosophyandinotherdisciplines.Inparticular,itwastheenginethatdrovethephilosophi-calmovementoflogicalpositivism,whichwascorrectlyperceivedbymoralphilosophers,poets,theologians,andmanyothersasdirectlyattackingthefoundationsoftheirrespectiveenterprises.Unlikemostphilosophicaltheo-ries,italsohadnumerouspowerfuleffectsontheactualpracticeofscience,bothverygoodeffectsandverybad.Buthereweshallexamineverification-ismsimplyasanothertheoryoflinguisticmeaning.Asonepopularpositivistsloganhadit,adifferencemustmakeadiffer-ence.Thatistosay,ifsomebitoflanguageissupposedtobemeaningfulat\nVerificationism99all,thenithasgottomakesomekindofdifferencetothoughtandtoaction.Andthepositivistshadaveryspecificideaofwhatkindofdifferenceitoughttomake:thebitoflanguageoughttomatter,specifically,tothecourseofourfutureexperience.Ifsomeoneutterswhatsoundslikeasentence,butyouhavenoideahowthetruthofthatsentencewouldaffectthefutureinadetect-ableway,theninwhatsensecanyousaythatitisneverthelessameaningfulsentenceforyou?Thepositiviststhrewoutthatrhetoricalquestionasachallenge.SupposeIputalineofsomethingthatlookslikegibberishontheblackboardandIassertthatthescribbleisameaningfulsentenceinsomeone’slanguage.Youaskmewhatwillhappendependingonwhetherthescribbleistrueorfalse.Isay“Nothing;theworldwillgoonjustasitotherwisewould,whetherthissentenceistrueorfalse.”Thenyoushouldbecomedeeplysuspiciousofmycontentionthatthisapparentgibberishactuallymeanssomething.Lessdras-tically,ifyouhearsomeoneuttersomethinginanalientongue,youpresumethatitdoesmeansomething,butyouhavenoideawhatitmeans;thatisbecauseyoudonotknowwhatwouldshowwhetheritistrueorfalse.ThepositivistswereconcernedaboutthebasicpropertyofmeaningfulnessbecausetheysuspectedthatmanyofwhatpassedformeaningfulutterancesintheworksoftheGreatDeadPhilosopherswerenotinfact(even)mean-ingfulatall,muchlesstrue.So,theirverificationprinciplewasmostnotablyusedasacriterionofmeaningfulnessasopposedtomeaninglessness:asen-tencewascountedasmeaningfulifandonlyiftherewassomesetofpossibleexperiencesonsomeone’spartthatwouldtendtoshowthatthesentencewastrue;callthissetthesentence’sverificationcondition.(Asentencealsohasafalsificationcondition,thesetofpossibleexperiencesthatwouldtendtoshowthatitwasfalse.)If,inexaminingaproposedsentence,onecouldnotcomeupwithsuchasetofexperiences,thesentencewouldfailthetestandwouldberevealedasbeingmeaningless,howeverproperitssurfacegrammar.(Classicexamplesofallegedfailuresinclude:“Everything[includingallyard-sticksandothermeasuringdevices]hasjustdoubledinsize.”Eleventh-hourcreation:“Theentirephysicaluniversecameintoexistencejustfiveminutesago,completewithostensiblememoriesandrecords.”Demonskepticism:“Weareconstantlyandsystematicallybeingdeceivedbyapowerfulevildemonwhofeedsusspeciousexperiences.”)1Buttheverificationistsdidnotconfinetheirconcerntomeaningfulnessitself.Thetheoryalsotookamorespecificform,anticipatedbyC.S.Peirce(1878/1934).Itaddressedtheindividualmeaningsofparticularsentences,andidentifiedeachsentence’smeaningwiththatsentence’sverificationcondition.Thus,thetheoryhadapracticaluse,asanactualtestforwhatanindividualsentencedoesmean;itpredictsthesentence’sparticularpropositionalcon-tent.Thisisanimportantvirtue,notsharedbyallitscompetitors.(ThenaivePropositionTheorysaysnothingofhowtoassociateaparticularpropositionwithagivensentence.)TheVerificationTheorywasmeanttobeused,and\n100Theoriesofmeaninghasbeenused—evenbypeoplewhodonotacceptitinfull—asaclarificatorytool.Ifyouareconfrontedbyasentencethatyoupresumetobemeaningfulbutyoudonotentirelyunderstand,askyourselfwhatwouldtendtoshowthatthesentencewastrueorthatitwasfalse.TheVerificationTheoryisthusanepistemicaccountofmeaning;thatis,itlocatesmeaninginourwaysofcomingtoknoworfindingoutthings.Toaverificationist,asentence’smeaningisitsepistemology,amatterofwhatitsproperevidencebasewouldbe.(Ononeinterpretation,theSellarsianfunctionalorInferentialTheoryofMeaningmentionedinchapter6isverifi-cationist,asSellars’inferencerulesareepistemicdevices.)Thepositivistsallowedthatthereisaspecialclassofsentencesthatdonothaveempiricalcontentbutarenonethelessmeaningfulinaway:thesearesentencesthatare,sotospeak,truebydefinition,truesolelyinvirtueofthemeaningsofthetermsthatcomposethem.“Nobachelorismarried”;“Ifit’ssnowing,thenit’ssnowing”;“Fivepencilsaremorepencilsthantwopencils.”Suchsentencesmakenoempiricalpredictions,accordingtothepositivists,becausetheyaretruenomatterwhathappensintheworld.Buttheyhavemeaningofasortbecausetheyaretrue;theirtruth,howevertrivial,isguaranteedbythecollectivemeaningsofthewordsthatoccurinthem.Suchsentencesarecalledanalytic.Verificationismisanattractiveviewthathasbeenheldferventlybymany.Butlikeeveryothertheoryofmeaning,ithasitsproblems.SomeobjectionsThepositivistsneverachievedaformulationoftheVerificationPrinciplethatsatisfiedeventhemselves;theycouldnevergetittofitjustthestringsofwordstheywantedittofit.Everypreciseformulationprovedtobetoostrongortooweakinonerespectoranother(seeHempel1950).Thereisamethodologicalproblemaswell:totestproposedformulations,thepositiv-istshadtoappealtoclearcasesofbothkinds;thatis,ofmeaningfulstringsofwordsandmeaninglessstrings.Butthisassumesalreadythattherearestringsofwordsthatareliterallymeaninglesseventhoughtheyaregrammaticallywell-formedandcomposedofperfectlymeaningfulwords;andthatis,whenyouthinkaboutit,averyboldclaim.Theseproblemsdonotconstituteprincipledobjectionstoverificationism,buttheysuggesttwomorethatdo.OBJECTION1WittgensteinwouldanddidcomplainthattheVerificationTheoryisyetanothermonolithicattempttogetatthe“essence”oflanguage,andallsuchattemptsaredoomedtofailure.Butinparticularandlessdogmatically,thetheoryappliesonlytowhatthepositivistscalleddescriptive,fact-statinglan-guage.Butdescriptiveorfact-statinglanguageisonlyonekindoflanguage;\nVerificationism101wealsoaskquestions,giveorders,writepoems,telljokes,performceremo-niesofvariouskinds,andsoon.Presumablyanadequatetheoryofmeaningshouldapplytoalltheseusesoflanguage,sincetheyareallmeaningfulusesoflanguageinanyordinarysenseoftheterm;butitishardtoseehowtheVerificationTheorycouldbeextendedtocoverthem.REPLYThepositivistsacknowledgedthattheywereaddressingmeaningonlyinarestrictedsense;theycalledit“cognitive”meaning.Tobe“cognitively”meaningfulisroughlytobeastatementoffact.Questions,commands,andlinesofpoetryarenotfact-statingordescriptiveinthatsense,eventhoughtheyhaveimportantlinguisticfunctionsandare“meaningful”intheordi-narysenseasopposedtogibberish.Therestrictionto“cognitive”meaningwasfineforthepositivists’largermetaphysicalandanti-metaphysicalpurposes,butfromourpointofview,theelucidationoflinguisticmeaninggenerally,itisdamaging.Atheoryofmeaninginoursenseischargedwithexplainingallthemeaningfacts,notjustthosepertainingtofact-statinglanguage.Further,theretreatto“cogni-tive”meaningdoesnothelpwithobjection2.OBJECTION2Aswenoted,thepositivistswereworkingwithadmittedlypreconceivedideasofwhichstringsofwordsaremeaningfulandwhicharenot,tryingtoruleouttheintuitivelymeaninglessonesandtoruleintheobviouslymeaning-fulones.Butitisnotonlythepositiviststhathadpreconceivedideasaboutwhichstringsofwordsaremeaningful.Supposewelookatagivenstringofwords,andaskwhetherornotitisverifiable,andifsowhatwouldverifyit.Inordertodothat,wealreadyhavetoknowwhatthesentencesays;howcouldweknowwhetheritwasverifiableunlessweknewwhatitsays?Todeterminehowtoverifythepresenceofavirus,say,wemustknowwhatvirusesareandwhere,ingeneral,theyaretobefound;thusitseemswemustunderstandtalkofvirusesinordertoverifystatementsaboutviruses,ratherthanviceversa.But,ifwealreadyknowwhatoursentencesays,thenthereissomethingthatitsays.Andtothatextent,italreadyismeaningful.Thus,thequestionofverifiabilityandverificationconditionsisconceptuallyposteriortoknowingwhatthesentencemeans;itseemswehavetoknowwhatasentencemeansinordertoknowhowtoverifyit.2ButthatisjusttheoppositeofwhattheVerificationTheorysays.Arelatedpointisthatthereisaglaringdifferencebetweenthesentencesthatthepositivistswantedtoruleoutasmeaningless(“Everythinghasjustdoubledinsize,”“Theentirephysicaluniversecameintoexistencejustfiveminutesago”)andparadigmcasesofmeaninglessstrings,gibberish,orwordsaladofthesortillustratedinchapter1(“wgfjsdkhjjiobfglglfud,”“Goodof\n102Theoriesofmeaningoffprimlytheathethewhy”).Surelytheformerstringsarenotmeaninglessinthesamedrasticandobviouswayasthelatter.Whatevermaybewrongwiththemfromanepistemologicalpointofview,theyarenotmeregibberish.REPLYTheverificationistmustcomeupwithsomedifferencebetweenthetwotypesofstring,withoutadmittingthatstringsofthefirsttypearemeaningfulafterall.Hereisapossiblemove.StringsofthefirsttypearemadeofregularEnglishwordsand,becausetheyaregrammaticalfromasuperficiallysyn-tacticpointofview,thereisakindofillusionofunderstanding.Sincethesearethekindsofstringsofwordsthatoftendosayandmeansomething,theyproduceinusafeelingoffamiliarity.Wehavethefeelingthatweknowwhattheysay.Andinaweaksensewedo:Wecanparsethemgrammatically,andweunderstandeachofthewordsthatoccurinthem.Butitdoesnotfollowthatthesestringsofwordsdo,infact,meananythingaswholes.OBJECTION3TheVerificationTheoryleadstobadoratleasthighlycontroversialmeta-physics.Recallthataverificationconditionisasetofexperiences.Thepositivistsmeantsuchverifyingexperiencestobedescribedinauniformkindoflanguagecalledan“observationlanguage.”Supposeour“observationlanguage”restrictsitselftothevocabularyofsubjectivesenseimpressions,asin“Inowseemtoseeapinkrabbit-shapedthinginfrontofme.”ThenitfollowsfromverificationismthatanymeaningfulstatementIsucceedinmakingcanultimatelyonlybeaboutmyownsenseimpressions;ifsolipsismisfalse,Icannotmeaningfullysaythatitis.Andneithercananyoneelse.Evenifinsteadweloosenournotionof“observation”andincludewhatHempel(1950)calledthe“directlyobservablecharacteristics”ofordinaryobjects,itremainstruethatverificationismcollapsesasentence’smeaningintothetypeofobservationalevidencewecanhaveforthatsentence,withoutremainder.Forexample,wearedriventoagrotesquelyrevisionistviewaboutscientificobjects—theinstrumentalistviewthatscientificstatementsaboutelectrons,memorytraces,othergalaxies,andthelikearemerelyabbrevia-tionsofcomplexsetsofstatementsaboutourownlaboratorydata.Whatistheverificationconditionofasentenceaboutanelectron?Ofcourseitissomethingmacroscopic,somethingaboutmeterreadingsorvaportrailsinacloudchamberorscatteringpatternsonacathoderaytubeorsomethingofthesort.Itisobservablewiththenakedeyeinthehereandnow.Arewereallytobelievethatwhenwetalkaboutsubatomicparticleswearenotreallytalkingaboutlittleparticles—particlessosmallthattheycannotbeobserved—butinsteadaboutmeterreadings,vaportrails,andthelike?(Thepositiviststhemselvesdidnotconsiderthisinstrumentalismgrotesque,butthoughtitimportantlytrue;Ithinkitisgrotesque.)\nVerificationism103Andwhenweturntoquestionsaboutthehumanmind,wefindthataverystrongversionofbehaviorismfallsrightout:statementsaboutpeople’smindsaremerelyabbreviationsofstatementsaboutthosepeople’sovertbehavior.FortheonlysortofobservationalevidenceIeverhaveregardingyourinner-mostthoughtsandfeelingsisthebehaviorIseeandhearyouengagingin.Ifoneisaverificationist,philosophyofmindisoveranddonewith.Possiblyoneormoreoftheforegoingandtomeunappetizingtheoriesaretrue.Perhapstheyarealltrue.Mypointhereisjustthatourtheoryoflinguisticmeaningshouldnotshowinonestepthattheyare.Metaphysicsshouldnotbesettledbyatheoryoflanguage,forlanguageisjustalateadap-tationfoundinoneprimatespecies.(Perhapsitisnotevenanadaptation,butapleiotropism;thatis,amerebyproductofothertraitsthatarethemselvesadaptive.)OBJECTION4HowdoestheVerificationPrincipleapplytoitself?Eitheritisempiricallyverifiableoritisnot.Supposeitisnotverifiable.Theneitheritisjustmeaninglessoritisanempty“analytic”ordefinitionaltruth.Atleastonepositivist(Ihaveforgot-tenwhich)gallantlyembracedtheideathattheprincipleisjustmeaningless,aladdertobekickedawayonceonehasclimbedit.Somepositiviststookthelinethattheprinciplewasausefulstipulativedefinitionoftheword“mean-ing,”fortechnicalpurposes.Hempel(1950)calledtheprinciplea“proposal,”henceneithertruenorfalse,butsubjecttoeachofseveralrationaldemandsandconstraints,hencenotsimplyarbitrary.Ofcourse,anyphilosophercanstipulateanythingatanytime;buthowdoesthathelpthoseofuswhoarelookingaroundforacredible,indeedcorrecttheoryofmeaning(asis)?Stipulationshavetheirusesbut,whenwearetryingtocometoanadequatephilosophicaltheoryofapre-existingphenomenon,astipulationisnotofmuchhelp.Isupposesomepositiviststhoughtoftheprincipleasafaithful,correctdefinitionthatcapturestheantecedentmeaningof“meaning.”Thetroublewiththatideaisthatwedonotknowwhatspecificallysemanticevidencewouldbearoutthedefinitionascorrect.Certainlythepositivistshadnotsubjectedtheterm“meaning”tothesortofanalysisthatRussellhadlavishedontheword“the”;andneitherordinarypeoplenornonpositivistphiloso-pherssharedmanyintuitivejudgmentsinlinewiththeVerificationPrinciple.Itdoesnotseemtobeanalytic,like“Nobachelorismarried”;Idoubtthatanyonewhounderstandswhattheword“meaning”meansandwhat“verify”meansknowsthattobemeaningfulisjusttobeverifiableandthatasen-tence’smeaningisitsverificationcondition.Supposetheprincipleistakentobeempiricallyverifiable.Thatis,assumeitissupposedtobeconfirmedbyourexperiencesofsentences,theirmean-ings,andtheirverificationconditions,andmeaninghasbeenfoundtotrack\n104Theoriesofmeaningverificationcondition.But(asinobjection1)thatpresupposesthatwecanrecognizesentencemeaningsindependentlyofassigningthemverificationconditions.Anditisnotclearjustwhatweshouldcountasthe“empiri-cal”dataonwhichtheprincipleisbased.Surveyresultsfromstreetcorners?Dictionarydefinitions?(Neverthat.)One’sownlinguistic“intuitions”?(Also,theVerificationPrinciple’sownmeaningwouldthen,bytheprincipleitself,coincidewithitsownverificationcondition,thesetofexperiencesasofmeaningscoincidingwithverificationconditions;thatisanastytangle,thoughIamunsurewhetheritisultimatelyvicious.)Atanyrate,theself-applicationproblemisarealone,notjustasuperficialtrickquestion.3OBJECTION5Erwin(1970)offersanargumenttoshowthateverystatementisverifiable,triviallyandinmuchthesameway.Supposewearepresentedwithafunny-lookingmachinethatturnsouttobeamarvelouspredictor.Namely,whenonecodesadeclarativesentenceontoapunchcardandinsertsitintoaslotinthemachine,themachinewhirrsandclunksandlightsupeither“TRUE”or“FALSE”;moreover,sofarasweareabletocheck,themachineismiracu-louslyalwaysright.Nowconsideranarbitrarilychosenstringofwords,S.ThefollowingsetofexperienceswouldsufficetoraiseS’sprobabilitytoadrasticdegree:1WecodeSontoapunchcard.2Wefeedthecardintoourmachine.3Themachinelightsup“TRUE.”(Andrememberthatthemachinehasneveroncebeenwrong.)Thus,thereexistsapossiblesetofexperiencesthatwouldconfirmS,evenifSisintui-tivelygibberish.AndS’sownparticularverificationconditionwouldbethat,whenitiscodedandputtothemachine,themachinelightsup“TRUE.”ThustheVerificationTheoryistrivialized,sinceeverystringofwordsisverifiable,anditassignsthewrongmeaningstoparticularsentences(becauseveryfewsentencesmeananythingaboutpunchcardsbeingfedintoinfernalmachines).Somethingiswrongwiththatargument.ButIhavefounditveryhardtosayexactlywhat.OBJECTION6AnyversionoftheVerificationPrinciplemustpresupposean“observationlanguage”inwhichexperiencesaredescribed;henceitmustcountenanceafirmdistinctionbetween“observational”and(correlatively)“theo-retical”terms.AsIhavementioned,someofthepositivistsrestrictedtheir\nVerificationism105observationlanguagetostatementsaboutpeople’sprivate,subjectivesenseimpressions.Butthatdidnotserveforpurposesofintersubjectivelycheck-ablescience,somostpositivistsjoinedHempel(1950)inappealingtothe“directlyobservablecharacteristics”ofordinaryobjects.Therearetwoprob-lemshere.First,thenotionof“directobservation”isavexedone,andseemstotallytechnology-relativeandinterest-orproject-relativeaswell.Isavisualobservation“direct”whenyouarewearingeyeglasses?Howaboutifyouareusingamagnifyingglass?Howaboutthroughamicroscope,atthisorthatdegreeofmagnification?Howaboutthroughanelectronmicroscope?Second,“observations,”andstatementscouchedin“observationlanguage,”aretheory-ladenatleasttoadegree;whatcountsasanobservationandwhatcountsasobservedandhowa“datum”isdescribedarealldeterminedinpartbytheverytheoriesthatareinquestion.Boththeseproblemsareknottyissuesinthephilosophyofscience;Imerelymentionthemhere.4Buttheyhelptosetupamuchdeeperobjectiontoverificationism.ThebigoneOBJECTION7FollowingPierreDuhem(1906/1954),W.V.Quine(1953,1960)arguesthatnoindividualsentencehasadistinctiveverificationcondition,exceptrelativetoamassofbackgroundtheoryagainstwhich“observational”testingtakesplace.Thiswilltakesomeexplaining.Thereisanaiveideathatmanypeoplehaveaboutscience.Itisthatoneputsforwardascientifichypothesisandthenteststhehypothesisbydoinganexperiment,andtheexperimentshows,allbyitself,whetherthehypothesisiscorrect.Duhempointedoutthatinthehistoryoftheuniversetherehasneverbeenanexperimentthatcouldsinglehandedlyverifyorfalsifyahypothesis.Thereasonisthattherearealwaystoomanyauxiliaryassumptionsthathavetobemadetobringthehypothesisintocontactwiththeexperimentalapparatus.Hypothesesdosometimesgetdisconfirmed,outrightrefutedifyoulike,butonlybecausethescientistsinvolvedareholdingcertainotherassumptionsfixed,assumptionsthataredisputableandmayevenbequitewrong.Supposewearedoinganastronomicalstudy,andweareverifyingandrefutingthingsbymakingobservationsthroughcomplicatedtelescopes.Inusingsuchtelescopes,theastronomersareassumingvirtuallyallofopticaltheory,andcountlessotherthingsbesides.Surprisingly,Duhem’spointholdsineverydaylifeaswell.Takeanygoodordinarysentenceaboutaphysicalobject,suchas“Thereisachairattheheadofthetable.”Whatisitsverificationcondition?Afirstthingtonoticeisthat“the”setofexperiencesthatwouldconfirmthatsentenceisinawayconditional,onone’shypotheticalvantagepoint.Wemighttrysomethinglikethis:Ifyouwalkintotheroomfromthedirectionofthisdoorhere,you\n106Theoriesofmeaningwillhaveanexperienceasofachairattheheadofthetable.Buteventhatdepends.Itdependsonwhetheryouhaveyoureyesopen,anditdependsonwhetheryoursensoryapparatusisfunctioningproperly,anditdependsonwhetherthelightsareon,and....Thesequalificationsdonotforeseeablycometoanend.Ifwetrytobuildintheappropriatehedges(“Ifyouwalkintotheroom,andyouhaveyoureyesopen,andyoursensoryapparatusisfunctioning,...”),morequalificationscropup:Areyouwalkingforwardratherthanbackingintotheroom?Hassomethingbeeninterposedbetweenyouandthechair?Hasthechairbeencamouflaged?HasitbeenrenderedinvisiblebyMartians?HasyourbrainbeenalteredbyafreakishburstofQ-radiationfromthesky?Wecangoonlikethisfordays.Themoralisthatwhatwetaketobe“the”verificationconditionforagivenempiricalstatementpresupposesamassivebackgroundofdefaultauxiliaryassumptions.Thoseassumptionsareusuallyperfectlyreasonable,anditisnoaccidentthatwemakethem.Butaparticular“verificationcondi-tion”isassociatedwithagivensentenceonlyifwechoosetorelyonsuchassumptions,almostanyofwhichmayfail.Intrinsically,thesentencehasnodeterminateverificationcondition.Thatis(tosaytheleast)anembarrassmentforatheorythatidentifiesasentence’smeaningwiththatsentence’sverificationcondition.Butasweshallnowsee,thematterdoesnotquiteendhere.TWOQuINEANISSuESInthe1950sand1960s,W.V.Quineposedtwochallengestothepositivists’philosophyoflanguage.First,heattackedthenotionofanalyticity(Quine1953,1960);thatis,heattackedtheclaimthatsomesentencesaretrueentirelyinvirtueofwhattheymeanandnotbecauseofanycontributionfromtheextralinguisticworld.Quinegivesanumberofdifferentargumentsagainstanalyticity.Someofthoseareunconvincing.Othersarebetter,andhavekept“analytic”afairlydirtywordeversince,oratleasttillarecentresurgence.Iwillnotitemizethem,butonlygiveageneralideaofwhatIthinkisatthebottomofQuine’srepudiationofanalyticity.Quinesharesandmaintainsthepositivists’epistemologicalbent,andbelievesthatiflinguisticmeaningisanythingitisafunctionofevidentialsupport.Buthisownepistemologydiffersfromthepositivists’inbeingholistic.Thereareindividualsentencesyouholdtrueandsentencesyourejectasfalse,butineachcasethesupportforyourbeliefisacomplexmat-teroftheevidentialrelationsyoursentencebearstomanyothersentences.Wheneveritseemsthatbeliefrevisionisrequired,youhaveawidechoiceofwhichbeliefstogiveupinordertomaintainasuitablycoherentsystem(recallDuhem’spoint).Andthereisnobeliefthatiscompletelyimmunetorevision,nosentencethatmightnotberejectedunderpressurefromempiri-calevidenceplusaconcernforoverallcoherence.Evenapparenttruthsoflogic,suchastruthsoftheform“EitherPornotP,”mightbeabandonedin\nVerificationism107lightofsuitablyweirdphenomenainquantummechanics.Butananalyticsentencewouldbydefinitionbeentirelyunresponsivetotheworld’sinput,andsoimmunetorevision.Therefore,therearenoanalyticsentences.Itmayseemoflittlepracticalconsequencewhetherthereareanysentencesthatoccupythequaintphilosophers’categoryof“analytic.”ButQuine’srejectionofanalyticitydoeshaveoneinterestinglittlerepercussion.SupposetwoEnglishsentences,S1andS2,arepreciselysynonymous.Thentheconditionalsentence“IfS1,thenS2”shouldbeanalytic,havingthecontent“If[thisstateofaffairs],then[thisverysamestateofaffairs],”whichcouldhardlybefalsifiedbyanyempiricaldevelopment.So,iftherearenoanalyticsentences,notwoEnglishsentencesarepreciselysynonymous,noteven“Bambi’smotherwasadoe”and“Bambi’smotherwasafemaledeer.”5Itgetsworse.HereisQuine’ssecondchallengetothepositivists,andindeedtopracticallyeveryone.Itisnotjustthattherearenoanalyticsen-tences,andnotjustthatnotwosentencesaresynonymous.Itisthatthereisnosuchthingasmeaning.Quinedeniesour“meaningfacts”inthefirstplace,andurgesaneliminativismornihilismaboutmeaning,intheformofhisdoctrineofthe“indeterminacyoftranslation.”HeretooQuinehasgivenanumberofarguments,somemoreconvincingthanothers.One(fromQuine1969)canbestatedverysimply:Individualsen-tencesdonothaveverificationconditions.But,ifasentencehadanymeaningatall,itwouldbeaverificationcondition.Therefore,individualsentencesdonothavemeaningsatall.ThusdoesQuinesaveverificationismfromobjec-tion5.Butitisadesperatelunge,sinceitsavesthevillagebydestroyingit,simplyeliminatingmeaningandthemeaning-factsthemselves.Theproblemwiththeargument,ofcourse,isinjustifyingthesecondpremise;ifsentencesdonothaveverificationconditions,whycontinuetoacceptverificationismwhentherearesomanyothertheoriesofmeaningonoffer?Abetter-knownargumentstartswiththehypothesisofafieldlinguistinvestigatinganaliennativelanguagefromscratchandtryingtoconstructa“translationmanual”orNative–Englishdictionary.Quinearguesthatthetotalevidenceavailabletothelinguistfailstodetermineanyonetranslationmanual;manymutuallyincompatibleonesareentirelyconsistentwiththatevidence.Moreovertheunderdeterminationhereisnotmerelythestandardunderdeterminationofscientifictheoriesbytheevidenceonwhichtheyarebased.Itisradical:noteventheworld’stotalityofphysicalfactsufficestovin-dicateoneoftherivaltranslationmanualsasagainsttheothers.Therefore,notranslationiscorrecttotheexclusionofitsrivaltranslations.Butifsentenceshadmeaningsthentherewouldbecorrecttranslationsofthem,namelythetranslationsthatdidpreservetheiractualmeanings.Therefore,sentencesdonothavemeanings.Theproblemhereistojustifythepremisethatnoteventheworld’stotal-ityofphysicalfactrulesinoneoftherivaltranslationmanualsascorrect.Thedefenseofthatpremiseremainsobscure.\n108TheoriesofmeaningSummary•AccordingtotheVerificationTheory,asentenceismeaningfulifandonlyifitsbeingtruewouldmakesomedifferencetothecourseofourfutureexperience;andasentence’sparticularmeaningisitsverifica-tioncondition,thesetofpossibleexperiencesthatwouldtendtoshowthatthesentencewastrue.•Thetheoryfacesanumberofmedium-sizedobjections.•Buttheworstobjectionisthat,asDuhemandQuinehaveargued,individualsentencesdonothavedistinctiveverificationconditionsoftheirown.•Quineattackedtheviewthatthereare“analytic”sentences,sentencestruesolelyinvirtueoftheirmeanings.•FromDuhem’spoint,Quineinferredtheradicalclaimthatindividualsentencesdonothavemeanings;thereisnosuchthingassentencemeaning.Questions1Respondontheverificationist’sbehalftooneofobjections1–6.2Trytotackleobjection7.3HaveyouanyfurthercriticismtomakeoftheVerificationTheory?4DiscussQuine’sattackonanalyticity,orhisdefenseofmeaningindeterminacy.(Someoutsidereadingwouldberequiredforeitherofthese.)Furtherreading•Ayer(1946)isaclassicand/butveryaccessibleexpositionanddefenseofverificationism.•Someinfluentialanti-verificationistpapersbesidesQuine’swereWaismann(1965b)andvariouscollectedessaysbyHilaryPutnam(1975b),especially“Dreamingand‘DepthGrammar’.”•Quine’sdoctrineoftheindeterminacyoftranslationspawnedavastandtoxicliterature.Foroneviewofthedoctrineandtheearlylitera-ture,seeLycan(1984:ch.9)(youwereexpectingmetorecommendsomeoneelse’sview?);also,seeBar-On(1992).•The1970sand1980ssawanoutbreakofneoverificationism,duelargelytowritingsofMichaelDummettcollectedinhis(1978)book.ForanoversimplifyingbutveryclearattackonDummett,seeDevitt(1983).\n9Truth-ConditionTheories:Davidson’sprogramOverviewAccordingtoDonaldDavidson,wewillobtainabettertheoryofmeaningifwereplacethenotionofasentence’sverificationconditionwiththatofthesentence’struthcondition:theconditionunderwhichthesentenceactuallyisorwouldbetrue,ratherthanastateofaffairsthatwouldmerelyserveasevidenceoftruth.Davidsonoffersseveralarguments,chiefamongwhichisthatcompositionalityisneededtoaccountforourunderstandingoflong,novelsentencesandasentence’struthconditionisitsmostobviouslycom-positionalfeature.AsamodelofthewayinwhichtruthconditionscanbeassignedtosentencesofnaturallanguagessuchasEnglish,Davidsontakesthewayinwhichtruthisdefinedforanartificialsystemofformallogic.But,sinceEnglishsentences’surfacegrammardivergesfromtheirlogicalforms,atheoryofgrammaranditsrelationtologichastobebroughttobear;suchatheoryexistsandissupportedindependently.Davidson’stheoryfacesmanyobjections.Oneisthatmanyperfectlymeaningfulsentencesdonothavetruth-values.Someothersarethathispro-gramcannothandleexpressions(suchaspronouns)whosereferentsdependoncontext,predicatesthatarenotsynonymousbuthappentoapplytojustthesamethings,andsentenceswhosetruth-valuesarenotdeterminedbythoseoftheircomponentclauses.TruthconditionsSofar,onlyoneofourtheorieshasmanagedtoshedmuchlightonwhatactuallydeterminesthemeaningsofparticularsentences.ThePropositionTheorytooksentencemeaningsandjustreifiedthem(madethemintoobjectsofacertainkind),withoutmuchfurthercommentandwithoutconnectingtheobjectthusreifiedwithanyone’slinguisticpracticesorbehavior.Griceattemptedtofoboffthequestionintothephilosophyofmindbytryingtoconnectsentenceswiththecontentsofpeople’sactualintentionsandbeliefs,whichwasnotverysuccessfuland,moretothepoint,simplytooktheintentions’andbeliefs’contentsthemselvesforgranted.Aswehaveseen,\n110Theoriesofmeaningtheverificationistsdidbetter;theyofferedusatestforthepropositionalcontentofanygivensentence,thatcontentbeing(precisely)thesentence’sverificationcondition.Thetroubleisthat,evenifweignoretheDuhem–Quineproblem(objection7inthepreviouschapter),theverificationtestoftenseemstopredictthewrongcontent(objection3).DonaldDavidson(1967a,1970/1975)arguedthatwewillgetwherewewanttobeifwereplacethepositivists’notionofasentence’sverificationconditionwiththatofthesentence’struthcondition.Onthisview,toknowasentence’smeaningistoknowtheconditionsunderwhichthatsentencewouldbetrue,ratherthantoknowhowtotellwhetherthesentenceisactuallytrue.(Nevermindepiste-mology.)Fortwosentencestobesynonymousisforthemtobetrueunderjustthesameconditions;forasentencetobeambiguousisforittobebothtrueandfalseinthesamecircumstanceyetwithoutself-contradiction;foronesentencetoentailanotherisforittobeimpossiblethatthefirstbetruewithoutthesecondbeingtruealso.Wearealreadyfamiliarwiththetruth-conditionalapproachtomeaning,thoughnotbyname,fromourdiscussionofRussell’sTheoryofDescriptions:Russellproceedspreciselybysketchingthetruthconditionsofsentencescontainingdescriptionsandarguingonvariousgroundsthattheyarethecorrecttruthconditions.ButmoreofRussellinthenextsection.Davidsonbeginswithtwoideasthatprovetoberelated.Oneisthatatheoryofmeaningshouldaffordguidanceonwhatdeterminesthemean-ingofaparticularsentence.Theotheristhatofgivingcentralimportancetothewondrousphenomenonwithwhichthisbookbegan:ourabilitytounderstandlongnovelsentencesinaflash.Focusingonthefirstidea,heaskshowonemightgivea“theoryofmeaningfor”aparticularlanguage—notageneraltheoryofmeaninginourphilosophicalsense,butatheoryofEnglishorofChineseorofKwakiutl—thatspecifiedtheparticularmeaningsofthatlanguage’ssentencestakenonebyone.Whatformmightsuchatheorytake?Davidsonoffersandmotivatessev-eralguidelinesandconstraints.Thefirstisthis:Sincethereseemstobenoclearlimittothenumberofmeaningfulexpressions,aworkabletheorymustaccountforthemeaningofeachexpressiononthebasisofthepatternedexhibitionofafinitenumberoffeatures.Buteveniftherewereapracticalconstraintonthelengthofthesentencesapersoncansendandreceivewithunderstanding,asatis-factorysemanticswouldneedtoexplainthecontributionofrepeatablefeaturestothemeaningsofsentencesinwhichtheyoccur.(Davidson1970/1975:18)Hereheisappealingtoourabilitytounderstandlongnovelsentences,andsuggestinganexplanationofthatability.HowdoweunderstandapotentialinfinityofEnglishsentencesonthebasisofourfinitevocabularyandlimited\nTruth-ConditionTheories:Davidson’sprogram111experienceoflanguage?Theanswermustbethatwehavemastered“afinitenumberoffeatures,”arelativelysmallandmanageablesetofmeaningfulexpressionsthatserveasmeaning“atoms,”andalsosomerulesofcomposi-tion,“patterned”waysofcombiningthoseatomsorsemanticprimitivesthatgeneratethemeaningsofmorecomplexexpressions.1Verycrudely,themeaningatomsareindividualwords,andtherulesofcompositionarerulesofgrammarorsyntaxthatspecifyhowwordscanbecombinedinordertoprojecttheirindividualmeaningsintomorecomplexmeanings.Davidsoncontendsthatthemeaningofasentenceisafunctionofthemeaningsofitsconstituentwords.2Thisisthethesisofcompositionality,aswecalleditinchapter6.Compositionalityistheobvioushypothesistoexplainourunderstandingoflongnovelsentences:weunderstandcomplexmeaningsbydecomposingsentencessyntacticallyintosmallermeaningfulelements,andcomputingthecomplexmeaningsassyntacticfunctionsofthesentences’smallestmeaningfulparts.Soanadequatetheoryofmeaninginthegeneralphilosophicalsenseshouldguideusinconstructingasystematic“theoryofmeaningfor”anygivenlanguagethatwouldspecifythemeaningofeachgrammaticalsentenceofthatlanguagebychroniclingthesentence’scompositionoutofitscon-stituentwords.Thus,itshouldhavethemeanstogeneratealist:“Snowiswhite”meansthatsnowiswhite.“Grassisgreen”meansthatgrassisgreen.“Poltergeistsmakeuptheprincipaltypeofmaterialmanifestation”meansthatpoltergeistsmakeuptheprincipaltypeofmaterialmanifestation.“In1931,AdolfHitlermadeavisittotheUnitedStates,inthecourseofwhich...”[Yougettheidea.]Andthislistisinfiniteorpotentiallyso.Ofcourse,thisexamplespecifiesthemeaningsofEnglishsentencesinEnglish(andsoitsoundsabituninterest-ing),butwemustalsobeabletodothesameforotherlanguages:“DerSchneeistweiss”means[inGerman]thatsnowiswhite.“DasGrasistgrün”meansthatgrassisgreen.“DiePoltergeistenrepresentieren...”[etc.]HowmightatheoryofEnglishoratheoryofGermangeneratesuchalist?Noticefirstthat,correspondingtoourabilitytounderstandlongnovelsentences,wehavetheabilitytodeterminethosesentences’truth-valuesifweknowenoughfacts.Forexample,ifIhappentoknowthatKatherineDienes’“AveMaria”settingemployschantsegments,drones,overlapping“orapronobis”figures,andotherdevicestosuggestthesonorityofmedievalconventmusic,andIencounterthesentence\n112Theoriesofmeaning(1)KatherineDienes’“AveMaria”settingemployschantsegments,drones,overlapping“orapronobis”figures,andotherdevicestosuggestthesonorityofmedievalconventmusic(asentencethatIamquitesureisasnewtoyouasitoriginallywastome);Ialsoknowthatthatsentenceisinfacttrue.AndifIhadencounteredasentencejustlikeitexceptthat“medievalconventmusic”hadbeenreplacedby“Ice-T’sbrandofrapmusic”andaclausehadbeenadded,“...andDieneshasrecentlymovedtoNewark,NewJersey,”Iwouldinstantlyhaveknownitwasfalse.Thus,itseemswegraspthetruthconditionsoflongnovelsentencesonsight,aswellasunderstandingthem,andthesamequestionarises:Howisthatpossible?Davidsonthinksthiscoincidenceisnocoincidence.Theques-tionhasthesameanswer:compositionality.Thetruthconditionsoflongsentencesaredeterminedbythetruthconditionsoftheshortersentencesofwhichtheyarecomposed,andthesyntacticprocessesthatgeneratethelongersentencescarrytruth-relatedsemanticpropertiesalongwiththem,thuscompoundingsimpletruthpropertiesintomorecomplexones.3Wehaveanelegantmodelforthiscompositionalityoftruthconditions,anditservesalsoastheonlymodelwehaveforthecompositionalityofmean-ing.Itisthesemanticsofaformallanguagesuchasthepredicatecalculus,asformulatedbylogicians.Ifyouhavetakenacourseinsymboliclogic,youwillalreadyhaveseenthiscomingandwillbeaheadofme.Ifyouhavenot,Iwilltrytoexplaintheideainformally,withoutrelyingontechnicalnotation.Ishalldescribeaverysimplelittlelanguage,nearlyassimpleasWittgenstein’sbuilders’languagebutwithacrucialdistinguishingfeature.Ithastwotermsorpredicates,FandG,whichcorrespondtotheEnglishwords“fat”and“greedy”;FdenotesorappliestoallandonlythefatthingsintheworldandGappliestoallandonlythegreedythings.Thelittlelanguage(whichIshallcall“Oafish”)alsohastwopropernames:a,whichdenotesAlbert,andb,whichdenotesBetty.Andithasasemanticruleforformingsubject–predicatesentences:asentencemadebyprefixingapredicatePtoapropernamenistrueiffwhatndenotesisincludedamongthethingstowhichPapplies.Andfinally,Oafishcontainstwofurtherexpressionscalled“sentenceconnectives”:“not,”whichcanbestuckontoanygivensentence,and“and,”whichcanbeinsertedbetweentwowholesentencestomakealongersentence.Eachoftheconnectivesisgovernedbyitsdistinguishingsemanticrule.The“not”ruleisthatasentencemadebysticking“not”ontoanothersentenceAwillbetrueifandonlyifAitselfisnottrue.The“and”ruleisthatasentenceofthecompositeform“AandB”willbetrueifandonlyifAistrueandBistruealso.Thus:\nTruth-ConditionTheories:Davidson’sprogram113TruThdefiniTionforoafish“F”appliestofatthings.“G”appliestogreedythings.“a”denotesAlbert.“b”denotesBetty.Asubject–predicatesentence“Pn”istrueifandonlyifwhat“n”denotesisamemberoftheclassofthingsthat“P”appliesto.Asentenceoftheform“NotA”istrueifandonlyifthesentence“A”isnottrue.Asentenceoftheform“AandB”istrueifandonlyifitscomponentsentences“A”and“B”arebothtrue.Thisisthewholelanguage—allofitsvocabulary,allofitsmeaningrulesofanykind.Itisoflimitedinterest,andencouragestediousrepetitiveness.Butitstruthdefinition,eveninitsbrutesimplicity,hasthetwinfeaturesthatweneed:ItallowsforindefinitelylongandindefinitelymanygrammaticalsentencesofOafish,and(nonetheless)itmanagestospecifythetruthcondi-tionofeveryoneofthem.Forexample,ifanOafishspeakerutters“Fa,”welearnfromoursubject–predicateclausethatthatsentenceistrueifandonlyifthedenotationofa,thatis,Albert,isincludedintheclassofthingstowhichFapplies,thatis,theclassoffatthings,whichisjusttosaythatAlbertisfat.(Theclassofthingstowhichatermappliesiscalledtheterm’sextension.)OronecansaythatAlbertisgreedy.Oronecansaythatheisfatandheisgreedy,becauseourtruthrulefor“and”tellsusthat“FaandGa”isgoingtobetrueonlywhenAlbertisfatandAlbertisgreedy.(Checkthatforyourself.)Andtheword“and”canbeiterated,thatis,appliedoverandoveragain,tomakelongerandlongersentenceswithoutletup:“FaandnotFb”;“FaandnotGaandFbandnotGb”;“FaandGaandnotFbandGbandFaandnotFb”;andsoonforever.(Ofcoursethelatersentenceswillberepeti-tious,sinceOafishhassuchasmalllexicon,buteventhemostrepetitioussentencesarestillgrammaticalandhaveperfectlycleartruthconditions.)So,justfromthistriflinglittletruthdefinitionwehavealreadygotinfi-nitelymanygrammaticalsentences,andwehaveprojectionrulesthattellus,nomatterhowlongasentenceis,theconditionunderwhichthatsentenceistrue.Armedwiththis,wecouldencounteranynovelsentenceofOafish,evenifitwerefivemileslong,andcomputeitstruthcondition.Wehaveexplainedapotentiallyinfinitecapacitybyfinite,indeedminuscule,means.\n114TheoriesofmeaningSupposewehavederivedatruthconditionstepbystepfromourtruthdefinitionandmadeitexplicit:“FaandnotGaandFbandnotGb”istrueifandonlyifAlbertisfatandAlbertisnotgreedyandBettyisfatandBettyisnotgreedy.WehavetakenasentenceofOafishandspecifieditstruthcondition.Buthavewenotalsospecifieditsmeaning?SurelywhatthechosensentencemeansisjustthatAlbertisfatandAlbertisnotgreedyandBettyisfatandBettyisnotgreedy.Anditmeansthatcompositionally,invirtueofwhata,b,F,andGdenoteplusthesemanticrulesfordeterminingcomplextruthconditionsfromsimplerones.SupposewecoulddothesameforEnglish,thatis,constructatruthdefini-tionthatspitsoutsomethingoftheform,“‘——’istrueifandonlyif——”foreachEnglishsentence.(Suchproductsarecalled“Tarskibiconditionals”or“T-sentences,”sincetheywereinspiredbytheformofTarski’s(1956)theoryoftruth.)AndsupposeeachT-sentenceisseentogetitstargetsentence’struthconditionright.Then,Davidsonasks,whatmorecouldreasonablybeaskedofatheoryofmeaningforEnglish?Consider:Acorrectassignmentofmeaningtoasentenceshoulddeterminethatsentence’struthcondition;soweknowthatanadequatetheoryofmean-ingforalanguageshouldyieldatleastatruthdefinitionforthatlanguage.Soifthetruthdefinitionalsodoeseverythingwewouldexpectatheoryofmeaningtodo,itwouldbereasonablesimplytoidentifyasentence’smean-ingwithitstruthcondition.Whataboutthemeaningfacts,then?IhavealreadymentionedthewaysinwhichtheTruth-ConditionTheoryaccountsforsynonymyandambiguity.Itaccountsformeaninginclusionandespeciallyforentailmentaswell.“FaandnotFb”entails“Fa”because,accordingtoourtruthdefinition,“FaandnotFb”couldnotbetrueunless“Fa”were.Atruthdefinitionforalanguagepredictsthefeltsynonymies,entailments,andothersemanticrelationsbyreferencetothesemanticcompoundingrulesitcodifies.And,inpart,thecontemporarytruth-conditiontheoriststudieslinguisticconstructionsinjustthesamewaythatRussellworkedondescriptions.S/hemarshalsawholebunchofmeaningfactsaboutaparticularkindorgroupofsentencesinwhichs/heisinterested—factsaboutsynonymyrelations,ambiguities,entailmentrelations,andsoon—andtriestoexplainthosefactsintermsoftruthconditions.Russellnotedthesemanticpropertiesofsentencesofthiskindorthat,especiallyinterestingpropertiesthatcreatelogicalpuzzles,andthenasked,howcanweputtogetheratheoryofsuchsentencesthatexplainswhythesentencesexhibitthosepuzzlingsemanticfeatures?Hisanswer,asintheTheoryofDescriptions,wouldbeaputativetruthcondition.\nTruth-ConditionTheories:Davidson’sprogram115TheTruth-ConditionTheoryseesmeaningasrepresentation.IneffectitrevertstotheReferentialTheory’sideaofmeaningasmirroringorcorre-spondencebetweensentencesandactualorpossiblestatesofaffairs;Russellemphasizedthisidea(andindeedmadeitacornerstoneofhismetaphysics).Thetruthdefinitionisfoundedonthereferentialrelationsbetweentermsandtheirworldlydenotataorextensions.Wesawinchapter1thatthecrudeReferentialTheorywasfartoosimpleanideaofthecorrespondencebetweenwordsandtheworld;thetruth-conditiontheoristdoesnotpositsostrongorsimple-mindedacorrespondence,sinces/hedoesnotcontendthatallwordsarenames.Butthetruth-conditiontheoristisbackinthebusinessofmir-roringnature,ofaskingwhatactualorpossiblestatesofaffairsdoesagiventargetsentencedepictorrepresent.Truth-definingnaturallanguagesOafishisexplicitlytruth-defined.Itssentencesweartheirtruthconditionsontheirsleeves,inthesensethatthereisnodisparitybetweenasentence’ssurfacegrammaticalformandwhatRussellcalleditslogicalform(chapter2).AndonecanjustlookatanOafishsentenceand,truthdefinitioninhand,workone’swaybackthroughthesentence’scompositionalstructureandcalculatethesentence’struthcondition.ThatisDavidson’sparadigm.Thereisahuge“but”(indeeda“But...!!”),whichhasprobablyalreadyoccurredtoyou.Itisonethingtoprovideatruthdefinitionforamade-upformallanguage,evenforamuchricheronethanOafish;itisquiteanothertorevealtruthrulesallegedlyunderpinninganalreadylivingnaturallan-guagelikeEnglish.Thenaturallanguagewasherefirst.And,muchmoretothepoint,sentencesofEnglishdonotweartheirtruthconditionsontheirsleeves.Notoriously,aswesawinchapter2,theirsuperficialgrammaticalformsdifferunpredictablyfromtheirlogicalforms.Well,saysthetruth-conditiontheorist,notquiteunpredictably.Thatiswheresyntaxentersthepicture.(Indeed,thetheoristmaysay,thatiswhatsyntaxisfor.)Iwouldliketogiveyouanentirecourseinsyntax;failingthat,Iwouldliketogiveyoujustthebasics.Butspaceallowsneither.Ishallmerelygesturetowardthefundamentalideaandhopethatyouwillpickupsomeoftherestelsewhere.Forsimplicity,Iwillusejargonthatrecallstheearlydaysoftheoreticalsyntax(roughly,the1960s)oncethatdisciplinehadbeenfoundedbyZelligHarrisandNoamChomsky.Asyntaxorgrammarforalanguage,naturalorartificial,isadeviceforsortingwell-formedorgrammaticalsentencesfromamongallthestringsmadeupofwordsfromthatlanguage.Andagain(aswithsemantics),themodelisthatofformationrulesforalogicalsystem.RecallOafish.SentencesofOafishcanbeparsed,diagrammedbywhatarecalled“phrasemarkers,”\n116Theoriesofmeaninginawaythatdirectlydepictshowtheyarecompoundedsyntacticallyoutofindividualterms.Hereis“FaandnotFb.”SentenceSentenceConnectiveSentencePredicateNameConnectiveSentencePredicateNameFaandnotFbAsentencecanbeformedbyplacinganameafterapredicate,so“Fa”and“Fb”aresentences.Asentencecanbeformedbyprefixingasentencewith“not,”so“notFb”isasentence.Finally,asentencecanbeformedbyplacing“and”betweentwosentences,sothewholethingisasentence.SimpleEnglishsentencescanbediagrammedsimilarly.Hereisaclassic:“Theboyhitthecolorfulball.”SentenceNounphraseVerbphraseDeterminerNounVerbNounphraseDeterminerAdjectiveNounTheboyhitthecolorfulballThenodesinsuchaphrasemarkerarelabeledaccordingtogrammaticalcat-egory,andthelowestonesbegintolooklikeEnglish“partsofspeech”:noun,adjective,andsoon.Highernodescorrespondtomorecomplexgrammaticalstructuressuchasnounphrases.\nTruth-ConditionTheories:Davidson’sprogram117ButfewEnglishsentencesaresosimple.Mosthavestructuresthatcannotbeentirelyrenderedbyphrasemarkersofthisstraightforwardtype(called“context-free”markers),becausetherearegrammaticalrelationsthatarerobustandunmistakablethatcannotberepresentedinthisform.Chomsky(1957,1965)arguedthatthephrase-markergrammarneedstobeaugmentedbyadevice,specificallyasetofrules,thatcantakeonephrasemarkerandturnitintooneofadifferentanddependentkind;hecalledsuchrules“trans-formations.”Forexample,apassivetransformationmighttaketheforegoingphrasemarkerandrearrangeitspartsintoaphrasemarkerfor“Thecolorfulballwashitbytheboy.”Transformationsareconceivedasdynamic,asagentsthatchopupphrasemarkersandrebuildtheirpartsintomorecomplicatedtreediagrams.Withanyluck,then,everygrammaticalstringofEnglishhaseitheracontext-freephrasemarkeroronethathasbeenderivedbyaseriesofoneormoretransformationsfromacontext-freemarker.Nootherstringisgrammatical.(Grammarsnolongerhavethissimplearchitecture,nordopresent-daylinguistsusemyantiquatedterminology.Buttolearnmoreyouwillhavetoreaduponyourown.)AsIsaid,linguistsoriginallyconceivedagrammarsimplyasamachinethatseparatedwell-formedstringsfromgibberish.Somelinguistsleaveitatthat,anddonotseethatthatenterprisehasmuchtodowithsemanticsorsentencemeaningproper.ButasDavidsonsays,somethingtakesthemean-ingsofindividualwordsandcomposesorprojectsthemintowholesentencemeanings.Whatisitthatdoesthat?Presumably,rulesforstickingthewordstogetherinsomerationalorder,anorderthatgivesthewholecompositeameaning.Butnoticethatoneandthesamesetofwordscanbearrangedindifferentorders,andtwooftheresultingstrings,evenifeachiswell-formed,canhavedifferentmeanings:Tragically,“JohnlovesMarsha”doesnotmeanthesamethingas“MarshalovesJohn,”eventhoughthesamethreewordscomposebothsentences.So,inordertogeneratedifferentmeaningsforthosesentences,theprojectionrulesmustalsodosomefinertuning;theyhavegottolook,notjustatthewordsthemselves,butatsomefinerdistinc-tions.Buttheverysyntacticrulesthatcompoundgrammaticallyacceptablestringsoutofindividualwordsalsoseemideallysuitedtoserveasthosemeaning-projectionrulesalso.Inthelate1960smanylinguistscametotakethatview,andheldthattransformationspreservemeaning(thoughthelatterthesiswasqualifiedandpartlyabandonedbytheExtendedStandardTheoryofthe1970sandbyGovernmentandBindingTheoryofthe1980s).Supposewehaveaphrase-structuregrammarforanexplicitlytruth-definedformallanguage.Andsupposewehavegrammaticaltransformationsthatarecapableofconvertingformulasofthatlanguageintowell-formedstringsofEnglish.Thenwehaveagrammarwhosephrase-structurecom-ponentspitsoutunderlyingstructures(thelogic-likeformulas)andwhosetransformationalcomponentproducesEnglishvariationsonthoseunder-lyingstructures.Giventhattransformationspreservemeaningor,more\n118Theoriesofmeaningnarrowly,thattransformationspreservetruthproperties,wecanthenseehowEnglishsentenceshavetheirmeanings.Namely,theyhavemeaningsinvirtueofhavingtruthconditions,andtheyhavetruthconditionsinvirtueofbeingtransformationallyderivedfromexplicitlytruth-definedformulasofalogic-likenotationalsystem.Synonymoussentencesaretransformationalvariationsofeachother;ambiguoussentencesaretheproductsofmorethanonepossibletransformationalprocess,andsoforth.Ideally,thetruth-conditiontheoristwantstobeempiricallymorerespon-siblethanRussellwas.Russellapproachedtruthconditionsapriori;hewouldwriteanEnglishsentenceontheblackboard,writealogicalformulanexttoit,eyeballthetwo,andjudgethatthelatterseemstogettheformer’struthconditionright.Better,hedidappealtohishypotheses’puzzle-solvingabili-tiesaswell.Butacontemporarytruth-conditiontheoristshouldwanther/hissemanticalhypothesestobe,inaddition,atleastsomewhatresponsibletoplausiblesyntactictheories.ObjectionstotheDavidsonianversionOBJECTION1LiketheVerificationTheory,theTruth-ConditionTheoryseemstoapplyonlytodescriptive,fact-statinglanguage;questionsandcommandsandsoonarenottrueorfalseatall.AWEAKREPLYAlthoughwedonotordinarilycallquestionsorcommands“true”and“false,”theydohavebipolar,truth-likesemanticvalues.Aquestioniscorrectlyanswered“yes”or“no”;acommandisobeyedordisobeyed.Intuitively,anondeclarativesentencecorrespondstoastateofaffairsthatmayormaynotobtain,eventhoughitsfunctionisnottodescribeorreportthatstateofaffairs.Andforsemanticalpurposeswemayaswelltreatthosesemanticvaluesastruth-values.Forexample,acommandis“true”ifitdoesinfactgoontobeobeyed,“false”ifitdoesnot.Ofcoursethisisanonstandarduseof“true”and“false”;wearewideningtheirapplicationtoallsemanticbipolarity.(Perhapsweshouldmakeupapairofmoregeneralsemanticalterms,suchas“positive”and“negative.”)AFIRSTREjOINDERNotallnondeclarativesarethusbipolar.Consider“wh-”questions,suchas“Whorobbedthediaperservice?,”“Whattimeisit?,”and“Whydidyoublowupmyboat?”Noneofthesehasa“yes”or“no”answer;indeed,eachadmitsaverylargerangeofpossiblecorrectanswers.\nTruth-ConditionTheories:Davidson’sprogram119ASECONDREjOINDERThedifficultyaboutlackoftruth-valueisnotconfinedtonondeclarativesentences.Foronething,ithasbeenarguedthatcertaingrammaticallydeclarativesentenceslacktruthconditionsandhaveonlyepistemic“assert-ibility”conditions.Mostnotably,Adams(1965)andothershavedefendedtheviewthatindicativeconditionalslacktruthconditionsandtruth-value.Moreover,somephilosophershold(followingthePositivists)thatcertaingrammaticallydeclarativesentencesarenotfact-statingeventhoughtheymightbetakenbythenaivetobeso.Accordingtotheemotivistsinmoralphilosophy,moraljudgmentsareonlyevincingsorventings,semanticallyjustlikegroans,gruntsofprotest,cheers,andthelike.Ifso,thensuch“factuallydefective”sentencesdonothavetruth-values.SoaT-sentencedirecteduponone(“‘Murderiswrong’istrueiffmurderiswrong”)shouldcomeoutfalseoranomalous.4REPLyTOTHESECONDREjOINDERItiseasyenoughforthetruth-conditiontheoristwhoisalsoanemotivist(orwhatever)torestricther/histruththeoryagainstnonfactualsentencesinthefirstplace.Butcontrariwise,onemayarguefromthegeneralplausibilityoftruth-conditionalsemantics(ifonebelievesinit)totheimplausibilityofemotivismandotherviewsthatdenytruth-valuetoperfectlygrammaticaldeclaratives.OBJECTION2Davidsontalksasiftheright-handsidesofhisT-sentenceswillbewritteninEnglish,orinthetheorist’sownnaturallanguagewhateveritmaybe,sothattheycanbereadilyseentobecorrectorincorrect.Indeed,DavidsontoutstheT-sentencesasempiricallytestableconsequencesofaproposedtruthdefinitionforalanguage.ButnoactualDavidsoniantruthdefinitioncoulddeliversuchT-sentences(Stich1976;Blackburn1984).ForsuchatheorytoyieldT-sentences—oranythingelse—astheorems,itmustbeformulatedinafairlyformalandregimentedlanguage,somethinglogic-like.(LookagainatthetruthdefinitionforOafish.)Moreover,oncethetheoristgetsaroundtonatural-languageconstructionsthatdonotoccurinstandardsymboliclogics,suchasadverbs,beliefoperators,andthelike,theright-handsidesofT-sentencesinvolvingthosemaycontainsomeradicallyunfamiliarnotation.ArecentversionofDavidson’sown(1967b)theoryofactionsentencesgener-atessuchT-sentencesas:“Jonesbutteredthetoastatmidnight”istrueiff(∃e)(BUTTERING(e)&PROTAG(Jones,e)&VICTIM(thetoast,e)&OCCURRED-AT(e,midnight)).\n120TheoriesofmeaningTheright-handsidehereistobereadmuchasfollows:“Thereoccurredanevent,whichwasabutteringofthetoast,performedbyJonesatmid-night.”(BymakingtheunderlyingsubjectthewholeeventratherthantheagentJones,Davidsonisabletoexplainwhythetargetsentenceentailssuchsimplersentencesas“Jonesbutteredthetoast,”“Jonesbutteredsomething,”“Somethinghappenedtothetoast,”and“Somethinghappenedatmidnight,”whichentailmentsareotherwisehardtocapture.)Butwhat,then,becomesofDavidson’sclaimoftestability?HowarewesupposedtoknowwhethersucharcaneT-sentencesarecorrectorincorrect?REPLYTestabilityweakens,butnottothevanishingpoint.Forwecanstilltestcon-volutedT-sentencessuchastheforegoingagainstourlogicalintuitions,andwecanstillevaluateDavidson’sclaimstohaveilluminatedstrikingsemanti-calfeaturesofthetargetsentence.OBJECTION3Ferocioustechnicalproblemsariseonceonestartsexaminingsentenceswithdeicticelementsinthem(anticipatedbyDavidson1967a).(A“deictic”or“indexical”elementisonewhosesemanticinterpretationvarieswithcontextofutterance,likeatensemarkerorademonstrativepronoun.)Forexample,howwouldonestateatruthconditionforthesentence“Iamsicknow”?“‘Iamsicknow’istrueifandonlyifIamsicknow”wouldneverdo,sinceitstruth-valuedependsonwhouttersitandwhenandisnotingeneraldeterminedbymy(yourhumblenarrator’s)stateofhealth.Deicticsentencesdonotevenhavetruth-values,exceptonactualorhypotheticaloccasionsoftheiruse(apointthatwouldgratifyStrawson).Davidson’sownresponseistorelativizetruthtoaspeakerandatime.TherelevantT-sentencewouldbeformulatedas,“‘Iamsicknow’istrueaspotentiallyspokenbypattifandonlyifpissickatt.”Butthisisunsatisfac-toryineachofseveralways,5notleastinthatspeakerandtimeofutterancearenottheonlycontextualfactorsthataffecttruth-value.(Recall“Thisisafineredone.”)Weshallreturntothisissueinchapter11.6OBJECTION4ADavidsoniantruthdefinitionhasahardtimedistinguishingexpressionsthathappentocoextend(thatis,toapplytoexactlythesamerangeofref-erents)butwithoutbeingmutuallysynonymous(Reeves1974;Blackburn1984).Considertwosinglevocabularyitemsthatdifferinmeaningbutthathappentohaveexactlythesameextensions.Thestandardexampleofthisisthewords“renate”and“cordate,”meaningrespectively“creaturewithkid-neys”and“creaturewithaheart.”7ADavidsoniantruththeorywillnotbe\nTruth-ConditionTheories:Davidson’sprogram121abletodistinguishthemeaningofasentencecontaining“renate”fromthatofonecontaining“cordate,”foreachtermwillhavebeenassignedjustthesameclassofobjectsasitsextension.FIRSTREPLYInatruththeoryofthesortdescribedhere,thewordsusedintheright-handsidesoftheT-sentencesaresupposedtocorrespondascloselyaspossibletotheexpressionscomposingthetargetsentence.(Lookbackatthetruthdefi-nitionforOafish.)Thus,onewillwritetheclausefor“renate”as“‘Renate’denotesrenates”ratherthanas“‘Renate’denotescordates.”Toderivethelat-ter(albeittrue)statementfromourtruththeory,onewouldneedtoaddthecontingentandnonlinguisticpremise,“Allandonlyrenatesarecordates.”And,accordingtoDavidson,themeaningofatargetsentenceisgiven,notbyjusttheT-sentencedirecteduponthattargetsentence,butbytheT-sentencetogetherwithitsderivationfromtheaxiomsofthetruththeory.Toavoidthesuggestionthatbeingcordateispartofthemeaningof“renate,”wecanrequirethattheT-sentence’sderivationcontainnononlinguisticpremises.SECONDREPLY“Renate”and“cordate”willbedistinguishedinsentencescontainingcertainsortsofconstruction,notablyinmodalsentencesandinbeliefsentences.WhateversemanticsDavidsongivesforsentenceslike“Therecouldbearenatethatwasnotcordate”and“Geoffbelievesthathispetturtleisarenate”wouldhavetoaccommodate(indeedpredict)thenoninterchangeabilityof“cordate”for“renate”inthosesentences.REjOINDERTOTHESECONDREPLySuchsentences—inwhichcoextensivetermscannotbesubstitutedwithoutpossiblychangingthetruth-valueofthesentencesthemselves—arepuzzlingintheirownright.(Theyarecalledintensionalsentences;thisisageneraliza-tionofthephenomenonthatinchapter2wascalled“referentialopacity.”)Onewouldexpectthesubstitutiontomakenodifference;afterall,evenifweareusingadifferentword,wearecontinuingtotalkaboutexactlythesamethingorclassofthings.Wehavealreadyencounteredaspecialcaseofthisprobleminchapters2and3,theProblemofSubstitutivityfordefinitedescriptionsandforpropernames.Anytheoryofmeaningmustoffersomeexplanationofsubstitutivityfailures.Sothephrase“WhateversemanticsDavidsongivesforsentenceslike...”isnotinnocent.ItwillbehardworkforDavidsontosolvethatproblemgivenhisformatforatheoryofmean-ing.(Davidsondoesaddresstheproblemofintensionalityhereandthere,principallyinDavidson(1968).Hissolutionis,roughly,totreatintensionalsentencesasmakingtacitreferencetotheverywordsthatoccurinthem.Weshalllookataquitedifferentapproachinthenextchapter.)\n122TheoriesofmeaningOBJECTION5Itisallverysimpletowriteatruthruleforasentence-compoundingwordlike“and”.Afterall,“and”iswhatlogicianscallatruth-functionalconnective:thetruth-valueof“AandB”isstrictlydeterminedbythetruth-valuesofitscomponentsentencesAandB.Butmanysentence-compoundingexpressionsdonotsimplypasstruthoninthatway.Taketheword“because”:Thetruthof“AbecauseB”isnotdeterminedbythetruth-valuesofthecomponentsentencesAandB,for,evenifbothAandBaretrue,“AbecauseB”maybefalse,dependingonotherfeaturesoftheworld.How,then,mightonewriteatruthrulefor“because,”paralleltoOafish’srulefor“and”?Ortakeadverbs.Howmightonewriteatruthrulefor“slowly,”orfor“very”?Stillworse,consider“believesthat”again,asin“JoebelievesthatMarybelievesthatIrmabelievesthatthehouseisburningdown.”Howwouldwewriteatruthruleforit?“nbelievesthatA”istrueifandonlyif...”what?Oneobviousstrategyistoinvokeadomainofhelpfulentities,suchaspropositions(!),andwritetruthrulesfornon-truth-functionalexpressionsintermsofquantificationoverthatdomain.(Aswehaveseen,todealwithsomeadverbs,Davidsonintroducedadomainof“events,”andmadeadverbsintoadjectivalpredicatesofevents.)Themainproblemwiththatstrategyisthatitstrainssyntax,sincethetransformationshavetoworkhardertotransformthenew,zanylogicalformsintofamiliarEnglish;asBlackburnpointsout(1984:289),aDavidsoniantreatmentofanon-truth-functional(andintensional)constructionsuchas“because”or“believesthat”requiresatleasta“heavycommitmenttoconcealedlogicalforms.”(But,asbefore,beliefsentencesinparticulararealreadyanuglyproblemforanytheoryofmeaning.)OBJECTION6Truth-conditionalsemanticsmustfullydisclosethegeneralnotionoftruthitispresupposing.Buttheonlyplausiblegeneralanalysisoftruthisintermsofstatingorassertingthings:“Onewhomakesastatementorassertionmakesatruestatementifandonlyifthingsareas,inmakingthatstatement,hestatesthemtobe”(Strawson1970:15).Whichistosaythattheanalysismustbemadeintermsofcommunication,whichmustbecashedàlaGrice.Thus,althoughtheletteroftruth-conditionalsemanticsmaybeunobjectionable,itsspiritisbroken,foritcollapsesintoGriceanismratherthanstandingasasuperioralternative.FIRSTREPLYWhetherornotwehaveanacceptedgeneralphilosophicalanalysisoftruth,wehavetheconceptexpressedbytheordinaryEnglishword“true,”andwealsohavetheformalstructureintroducedbyTarski,whichgeneratesT-sentences.Ifthenotionofmeaningcanbereducedtothatoftruth,thatis\nTruth-ConditionTheories:Davidson’sprogram123atheoreticaleconomyevenifweprovidenofurtherphilosophicalexplicationoftruth;itisnotacaseof“theconceptsofmeaningandtrutheachpointingblanklyandunhelpfullyattheother”(Strawson1970:16).REjOINDERAllright,butwhatifthecorrectanalysisof“true”is(indeed)intermsofcommunicating?SECONDREPLYIfso,thentherewould(indeed)beGriceancollapse.Butwhyshouldweacceptthestating/assertingtheoryoftruthinthefirstplace?Whataboutalltheothergeneraltheoriesthatphilosophershaveofferedoverthepasttwothousandyears?Notably,therearetheclassicalCorrespondence,CoherenceandPragmatictheories.Morerecently,thereistheProsententialTheorydevisedbyGrover,CampandBelnap(1975).MybestguessastowhyStrawsonsimplyignorestheseisthathemustbeassumingthateverysuchtheorywouldsomehowhavetobuyintotheGriceanideaatsomeearlystage:Forexample,sincebeliefsareprimarilywhatcohereorfailtocohere,theCoherenceTheoryofTruthwouldhavetotreatsentencesonlyinsofarastheyexpressbeliefs,andsoon.ButIdonotseewhyweshouldgrantthatassumption(ifitisStrawson’s)justonhissay-so.Onthisinterpretation,Strawsonisinthepositionofinsisting,atbottom,“ButsurelysomeversionofGriceanismiscorrect.”Onthatnote,letmepickupthesuggestionGricemadeaboutunutteredandnovelsentences,hisappealtoabstract“resultantprocedures”(seetheendofthepenultimatesectionofchapter7).Itnowseemsthatwhathehadinmindtherewascompositionalityachievedbysyntax.SupposeGricecoulddeliveranotionofpublicmeaninganalogoustosentencemeaningbutapply-ingtosubsententialexpressionssuchaswords;callit“expressionmeaning.”Thenhecouldinvokesyntaxandabstractlyconstructsentencemeaningsoutofexpressionmeanings(thoughhereagain,hewouldhavetoworktodistinguishtheabstract“sentencemeanings”frompropositions).Howtogoaboutexplicatingexpressionmeaning?Recallthatinchapter2wedefinedanotionof“speaker-reference”forsingularterms,intendedinpreciselyGriceancontrasttothe“semanticreference”oftheterm,under-stoodintermsofspeakers’intentionstocallhearers’attentiontothings.Perhapswecoulddefineananalogousconceptof“speaker-extension”forpredicatesintermsofspeakers’intentionsstandingsomehowbehindusesofthosepredicates,andsoon.ThenwecouldborrowGrice’slanguageaboutrepertoiresandproceduresfromhisdiscussionofunstructuredutterances,anduseittofashioncorrespondingtypesofexpressionmeaning.Theresult-ingtwo-tieredreductionofsentencemeaningtospeaker-meaningwouldstillfaceproblems,butnolongerobjections2–4.\n124TheoriesofmeaningAlso,thepresentideasuggestsaninterestingresearchprogram,forittossesusbackintothetheoryofreferencefromanewdirection.Forexample,canthesemanticdenotationofapropernamereallybeanalyzedintermsofspeaker-reference?Onthefaceofit,thatideacompeteswiththeDescriptionTheoryofnamesandwiththeCausal–HistoricalTheory.Butthiscompositeview,theexplicationofsentencemeaningintermsofprimitiveexpressionmeaningplusaGriceantheoryoftheindividualprimi-tiveexpressionmeanings,istoconcedethattheTruth-ConditionTheoryiscorrect,andtoaddonly,thoughverysignificantly,anewtypeoftheoryofreferringthatwouldcompetewiththoseconsideredinchapters2–4.Summary•DavidsonoffersseveralargumentsindefenseoftheTruth-ConditionTheory.Themainargumentisthatcompositionalityisneededtoaccountforourunderstandingoflong,novelsentences,andasen-tence’struthconditionisitsmostobviouslycompositionalfeature.•Tarski’sstyleofdefiningtruthforasystemofformallogicisamodelforthewayinwhichtruthconditionscanbeassignedtosentencesofnaturallanguages.•ButsinceEnglishsentences’surfacegrammardivergesfromtheirlogicalforms,atheoryofgrammarandsyntactictransformationisneeded.•Suchatheoryexistsandhasindependentsupport.•Davidson’stheoryfacesmanyobjections.Perhapsthemostdamagingisthatmanyperfectlymeaningfulsentencesdonothavetruth-values.SomeothersarethathisTarskianprogramcannothandleexpressions(suchaspronouns)whosereferentsdependoncontext,predicatesthatarenotsynonymousbuthappentoapplytojustthesamethings,andsentenceswhosetruth-valuesarenotdeterminedbythoseoftheircomponentclauses.•ItmaybepossibletofuseDavidsonwithGricebyprovidingaGriceantheoryoftermextensions.Questions1EvaluateDavidson’smainargumentforhisTruth-ConditionTheory;thatis,hisappealtocompositionality,andTarskiantruthdefinitions.2Discussthefurtherargument,givenverybrieflyabove,whosepremisesarethat(a)asentence’smeaningshoulddeterminethatsentence’struthcondition,and(b)atruthdefinitionforalanguagealsodoeseverythingwewouldexpectatheoryofmeaningtodo.3Ifyouhappenalreadytoknowsomethingabouttheoreticalsyntax,assesstheprospectsofitsbeingusedasavehicleforconnectingEnglishsentencestotheright-handsidesofDavidsonianT-sentences.\nTruth-ConditionTheories:Davidson’sprogram1254AdjudicateoneormoreoftheobjectionsthatwehaveraisedagainsttheTruth-ConditionTheory.5IfyouarefamiliarwiththeLiarParadox,exploretheproblemitposesfortheTruth-ConditionTheory.(Davidson(1967a)addressesthisbriefly.)6Raiseanew,furtherobjectiontotheTruth-ConditionTheory.7Developthecompositional,“two-tiered”Griceantheorysuggestedattheendofthischapter.Or,pursuethe“first-stage”Griceanreductionofsemanticdenotation(fornamesorforpredicates).Furtherreading•OtherthanLycan(1984),thebestgeneralintroductiontotheDavidsonianprogramisHarman(1972).Thatpaperandmanyothergoodonesinandabouttruth-theoreticsemanticsarereprintedinDavidsonandHarman(1975);seealsotheanthologiesEvansandMcDowell(1976)andPlatts(1980).Platts(1979)isagoodcriticaldiscussionoftheDavidsonianprogram.•Harman(1974b,1982)brokewithDavidsontofound“ConceptualRole”semantics.Forasurveyoftheensuingliterature,seeLycan(1984:ch.10).•Davidson(1986)isanimportantcriticismofDavidson’sownposi-tion,basedonthephenomenonofmalapropism.•Animportantspinofffromandcompetitoroftruth-theoreticseman-ticsis“Game-Theoretic”semantics,developedbyJaakkoHintikka(1976,1979).IamunsurehowfarHintikka’sprogramcompeteswithtruth-theoreticsemanticsratherthanbeingavariantofit.ThebasicpapersonGTSarecollectedinSaarinen(1979).•Radford(1997),Culicover(1997),SagandWasow(1999),Carnie(2001),andLasnikandUriagereka(2005)areexcellentintroductionstocontemporarysyntactictheory;seealsoHornstein(1995).LarsonandSegal(1995)expoundtheconvergenceofsemanticswithcontem-porarysyntaxfromtheviewpointoftheoreticallinguistics.\n10Truth-ConditionTheories:possibleworldsandintensionalsemanticsOverviewKripkeanpossibleworlds(aspresentedinchapter4)affordanalternativenotionofatruthcondition:Wesawthatacontingentsentenceistrueinsomeworldsbutnotinothers.Soasentence’struthconditioncanbetakentobethesetofpossibleworldsinwhichthesentenceistrue.Moreover,possibleworldscanbeusedtoconstruct“intensions”ormeaningsforsubsententialphrases,andparticularlyforindividualwordsormeaningatoms,thatarelikeFrege’s“senses”inbeingindependentofactualreferents.Forexample,apredicatehasdifferentextensionsindifferentworlds,anditsintensioncanbetakentobethefunctionthatassociatesanygivenworldwiththepredicate’sparticularextensioninthatworld.Thenagrammarcanshowhowthosesub-sententialintensionscombinetomakeatruthcondition,henceameaning,foracompletesentenceofwhichtheyarecomponents.TheresultingviewneatlyavoidsseveraloftheobjectionsthatbesetDavidson’stheory,mostnotably4,theproblemofcoextendingbutnonsynonymousterms,and5,theproblemofnon-truth-functionalconnec-tives.ItalsolendsahandwiththeProblemofSubstitutivity.ButitinheritstherestofDavidson’sdifficultiesandincursoneortwomore.TruthconditionsreconceivedAswesawinthepreviouschapter,theTruth-ConditionTheoryunderstandsmeaningasrepresentation,asmirroringorcorrespondencebetweensen-tencesandactualorpossiblestatesofaffairs.ButwecantakethenotionofahypotheticalstateofaffairsmoreseriouslythanDavidsoniswillingto,andconsider“possiblestatesofaffairs/circumstances/conditions”asKripkeanpossibleworlds(chapter4).Recallthatapossibleworld(otherthantheactualworld,ourown)isanalternativeuniverse,inwhichthingsgootherwisethanthewaytheygohere.And,becauseworldsdifferamongthemselvesinrespectoftheircomponentfacts,ofcoursethetruthofagivensentencedependsonwhichworldweareconsidering.Thisaffordsanewversionoftheideaofasentence’struthcondition.Thesentenceistrueinsomepossiblecircumstancesandnotinothers.Which,in\nTruth-ConditionTheories:possibleworldsandintensionalsemantics127thevernacularofpossibleworlds,istosaythatthesentenceistrueinsomeworldsandnotinothers.Whentwosentenceshavethesametruthcondition,theywillbetrueinjustthesamecircumstances,injustthesameworlds.Whentheydifferintruthcondition,thatmeanstherewillbesomeworldsinwhichoneistruebuttheotherisfalse,sotheywillnotbetrueinjustthesameworlds.Asafirstapproximation,then,letustakeasentence’struthconditionsimplytobethesetofworldsinwhichitistrue.Forthetruth-conditiontheorist,ofcourse,thatsetofworldswillalsobethesentence’smeaning.Itwouldfollowthatsynonymoussentencesaretrueinjustthesameworlds,whereasforanytwononsynonymoussentencestherewillbeatleastoneworldinwhichoneofthesentencesistruebuttheotherfalse.Thisideageneralizestothemeaningsofsubsententialexpressions.But,toshowhowthatworks,Imustbacktrackforaparagraphortwo.Wesawinchapter2that,unlikeRussell,Frege(1892/1952b)hadrejectedthesisJ3/K3(“Ameaningfulsubject–predicatesentenceismeaningful(only)invirtueofitspickingoutsomeindividualthingandascribingsomepropertytothatthing”),bypositingabstractentitiesthathecalled“senses”andarguingthatasingulartermhasoneoftheseoverandaboveitsputativereferent.AndFregedefendedcompositionality:Accordingtohim,thesubject–predicatesentencehasacompositesensemadeupoftheindividualsensesofitsparts,andismeaningfulinvirtueofhavingthatcompositesense,whetherornotitssubjectevenhasareferentatall.(ThusdidFregeattacktheProblemofApparentReferencetoNonexistents.)Assketchedsofar,Frege’sviewsoundslikeaversionofthePropositionTheory.Andsoitis;itispreytothevariousobjectionsraisedagainstthattheoryinchapter5.ButRudolfCarnap(1947/1956),RichardMontague(1960,1970),andJaakkoHintikka(1961)developedintensionallogic,givingapossible-worldsinterpretationandexplicationofFregeansenses.Here,roughly,istheidea.Asingulartermorapredicateissaidtohavebothanextension(inthesenseintroducedinthepreviouschapter)andaFregeansenseor“intension.”Thetrickistoconstrueaterm’sintensionasafunctionfrompossibleworldstoextensions.Thus,theintensionofapredicateisafunctionfromworldstosetsofthingsexistinginthoseworldsthatareinthepredicate’sextensionsinthoseworlds.Forexample,theintensionof“fat”looksfromworldtoworldandineachworldpicksouttheclassoffatthingsthere.“Fat”meansnotjusttheactualfatthings,butwhateverwouldbefatinotherpossiblecircum-stances.(Toputtheideainmorehumanterms,ifyouknowthemeaningof“fat,”youknowwhatvarioushypotheticalthingswouldcountasfataswellasjustthelistofwhichthingsactuallyarefat.)“Individualsenses,”theintensionsofsingularterms,arefunctionsfromworldstoindividualdenizensofthoseworlds.Thatshouldsoundabitfamil-iarfromchapter4;arigiddesignatorexpressesaconstantfunctioninthatitpicksoutthesameindividualineveryworld.Butaflacciddesignatorchangesitsreferentfromworldtoworld:aswesaw,“theBritishPrimeMinisterin\n128Theoriesofmeaning(thesecondhalfof)2007”designatesGordonBrownintheactualworld,butvariousotherpeople(orothercreatures)inotherworldsandnooneatallinstillothers.Thesenseorintensionof“theBritishPrimeMinister”looks(orhops)fromworldtoworldandpicksoutwhoeveriscurrentlyPrimeMinisterthere.Aswithpredicates,ifyouknowthemeaningofthephrase“theBritishPrimeMinister,”youknowwhowouldbethePrimeMinisterundervari-oushypotheticalconditions,evenifyoudonotknowwhoisactuallyPrimeMinisternow.Functionsofthissortcombinetomakesensesorintensionsforwholesentences.Take:(1)ThepresentBritishPrimeMinisterisfat.Inanotherpossibleworld,(1)’ssubjectdenoteswhoeverisPrimeMinisterthere,and“fat”hasanextensiontherethatprobablydiffersfromtheactualclassoffatthings.So,compositionally,weknowhowtotellwhether(1)istrueinthatworld:(1)willbetrueinthatworldifandonlyifthePrimeMinistertherebelongstothatlocalextension.Therefore,ifweknowtheintensionof“thepresentBritishPrimeMinister”andtheintensionof“fat,”weknowwhetheragivenworldisonethatmakes(1)true,whichistosaythatweknowhowtotellinwhichworlds(1)istrue;forwehaveineffectacompositefunctionfromworldstotruth-values.Thereforeweknowwhatsetofworldsis(1)’struthset.(Strictlyspeaking,thesentence’sintensionisthefunctionratherthantheresultingtruthsetitself,butIshallignorethistechnicaldistinctionhereafter.)Andthatistosaythatweknowthepropositionexpressedby(1),whichistosaythatweknow(1)’smeaning.(Donotbemisled:allthistalkofour“knowing”thingsdoesnotmeanweareslippingintoverificationism.Iamspeakingmetaphoricallyofhowonecomputesacomplexintensiongivensomeprimitive,simpleintensionsandsubject–predicategrammar.)Ifapropositionisinthiswayconstruedasasetofpossibleworlds,thenwedo,afterall,obtainnontrivialexplanationsofthemeaningfacts.Twosentenceswillbesynonymousifandonlyiftheyaretrueinjustthesameworlds.Asentencewillbeambiguousifthereisaworldinwhichitisbothtrueandfalsebutwithoutcontradiction.Andthepossible-worldsconstrualaffordsanelegantalgebraofmeaningbywayofsettheory:Forexample,entailmentbetweensentencesisjustthesubsetrelation.S1entailsS2ifandonlyifS2istrueinanyworldinwhichS1is;thatis,thesetofworldsthatisS2’smeaningisasubsetofS1’smeaning.Thus,theimplementationoftruthconditionsintermsofpossibleworldssavesthissophisticatedversionofthePropositionTheoryfromHarman’sobjection3(chapter5),forittellsuswhata“proposition”is,intermsthatwecanworkwithindependently:Apropositionisasetofworlds.(Onemayhavemetaphysicalqualmsabouttheideaofa“nonactualpossibleworld,”butatleastonealreadyknowswhataworldissupposedtobe.)Thepresentview\nTruth-ConditionTheories:possibleworldsandintensionalsemantics129alsoavoidsoursecondobjectiontoIdeationalTheories,whichcarriedovertothePropositionTheory,forittellsuswhatanabstract“concept”is:afunc-tionfromworldstoextensions.(ShortlyIshallintroduceacomplication.)Finally,thereisadirectargumentforthepossible-worldsversionoftheTruth-ConditionTheory,givenverybrieflyinLewis(1970):Inordertosaywhatameaningis,wemayfirstaskwhatameaningdoes,andthenfindsomethingthatdoesthat.Ameaningforasentenceissomethingthatdeterminestheconditionsunderwhichthesentenceistrueorfalse.Itdeterminesthetruth-valueofthesentenceinvariouspossiblestatesofaffairs,atvarioustimes,atvariousplaces,forvariousspeakers,andsoon.(p.22)Ibelievetheideaisthis:IfyouunderstandacertainsentenceS,andyouareshownapossibleworldatrandom—weflyyouthereanddumpyoudowninthatworld,miraculouslymakingyouomniscientasregardsitsfacts—thenrightawayyouknowwhetherSistrueorfalse.(IfyouknoweverysinglefactofthatworldandyoustillcannottellwhetherSistruethere,thenyoucannotunderstandS.)Soonethingthatameaningdoesistospitoutatruthvalueforanyworlditisgiven.Whichistosaythatameaningisatleastatruthconditioninthesenseofaparticularsetofworlds.(Thisleavesitopenthatameaningmayincludemorethanjustatruthcondition.)AdvantagesoverDavidson’sviewThepossible-worldsviewhassomeimportantadvantagesoverDavidson’sversionoftheTruth-ConditionTheory.Specifically,itavoidsobjections4and5thatwemadeagainstDavidson.Objection4wastheproblemofcoextensivebutnonsynonymousterms.Onthepossible-worldsview,thatisnoproblematall.“Renate”and“cordate”differinmeaningbecause,althoughtheyapplytojustthesamethingsintheactualworld,theirextensionsdivergeinotherpossibleworlds;countlessworldscontainrenatesthatarenotcordatesandviceversa.Endofstory(thoughshortlyweshallresurrectFrege’ssolutiontotheProblemofSubstitutivity).Objection5wastheproblemofnon-truth-functionalsentenceconnec-tives.Herethepossible-worldsviewdisplaysauniquestrength.Foritenablesustostatetruthconditionsforcertainconnectivesdirectlyintermsofworlds.Takethesimplemodaloperator“Itispossiblethat,”asin“ItispossiblethatthepresentU.S.Presidentisfat.”ThelattersentencewillbecountedastrueifandonlyifthereisaworldinwhichthepresentU.S.Presidentisfat.Andifwewantedtosay“Necessarily,ifthereisaU.S.President,theUnitedStatesexists,”intensionalsemanticswouldcountthatastrueifandonlyifineveryworld,ifthereisaU.S.President,theUnitedStatesexists.\n130TheoriesofmeaningFromthiswecanseethatouroriginalformulaneedsqualification:noteverysimpleexpression’ssenseorintensioncanbecastasafunctionfromworldstoanextensionorreferent.Somearefunctionsfromintensionstootherintensions;“Itispossiblethat”takestheintensionofthesentencetowhichitisappliedandturnsitintoadifferentintension.Another,sub-sententialexamplewouldbeadverbs,suchas“slowly.”“Janeswims”istrueinaworldifandonlyifthereferentof“Jane”inthatworldisamongthethingsthatswimthere,becausetheextensionof“swims”isjusttheclassofthatworld’sdenizensthatswim.Butwhatabout“Janeswimsslowly”?Grammatically,“slowly”modifiesthepredicate“swims,”makingitintothecomplexpredicate“swimsslowly.”Andtheintensionalsemanticistmaintainsthatthesemanticsfollowsinjustthesameway:Theintensionof“slowly”isafunctionfromintensionstointensions;itpicksuptheintensionof“swims”andturnsitintoamodifiedintension,namelythefunctionthatlooksataworldandpicksouttheclassofthingsthatswimslowlyinthatworld.1Thepossible-worldstheoryhasadeftwaywithbeliefsentencesalso.LetusreturnforamomenttoFrege.AsasolutiontotheProblemofSubstitutivity,Fregeproposedthatabeliefsentencecanchangeitstruth-valueasaresultofsubstitutionofcoreferringtermsbecause,eventhoughthetwotermshavethesamereferent,theymayhavedifferentsenses,andsoadifferentcompositesensemayresultfromthesubstitution.(Andbelief,acognitivestate,hasa“thought”orcompositesenseasitsobject,notareferent.)AsalwayswithunexplicatedversionsofthePropositionTheory,thatsoundsrightbutdoesnotreallyexplainanythingsolongas“sense”ismerelytakenforgranted.Butthepossible-worldstheoristcangivetheexplanationmorecontent:Althoughthetwotermscoreferintheactualworld,theydivergeinotherworlds,sotheirintensionsdiffer.Thereforethecompositeintensionsofotherwisesimilarsentencesinwhichtheyappearwilldifferalso.Ifbeliev-ingisarelationbetweenthebelieverandaproposition—thatis,asentenceintension—thenofcoursethebelievermaybelievetheoneintensionwithoutbelievingtheother.Atthispointanadjustmentisneeded.AsInotedabove,thepresentversionofthepossible-worldstheorycountstwosentencesasbeingsynonymouswhenandonlywhenthetwoaretrueinjustthesameworlds.Butwhatofnecessarytruthsthatholdineveryworld?Itwouldfollowthateverysuchtruthissynonymouswitheveryother;forexample,“Eitherpigshavewingsortheydon’t”and“Ifthereareediblemice,thensomemiceareedible”wouldmeanexactlythesame,whichtheyobviouslydonot.Moreover,anysentencewouldbecountedasbeingsynonymouswithanyothersentencenecessarilyequivalenttoit:“Snowiswhite”wouldbesaidtomeanjustthesameas“Eithersnowiswhiteorpigshavewingsandpigsaremammalsandnomam-malshavewings”;andwhoeverbelievedtheformerwouldbeautomaticallycountedasbelievingthelatter.Somethinghastogive.Thesourceoftheproblemseemstobethatcomplexintensionscanbenecessarilycoextensiveevenwhentheyaremadeupoutofquitedifferent\nTruth-ConditionTheories:possibleworldsandintensionalsemantics131concepts.Thecure,then,asCarnap(1947/1956)saw,wastorequirethat,forsynonymy,sentencesshouldnotonlyhavethesameintensionbuthavethatintensioncomposedinthesameway(ormuchthesameway)outofthesameatomicintensions.Thisiswhathecalledintensionalisomorphism,anditrulesoutalltheforegoingproblemcases.Forexample,“Eitherpigshavewingsortheydon’t”and“Ifthereareediblemice,thensomemiceareedible”arecomposedoutofentirelydifferentintensions(thoseof“pig”and“wing”inthefirstcaseandthoseof“mouse”and“edible”or“eat”inthesecond).RemainingobjectionsThepossible-worldstheoryinheritsseveraloftheobjectionsraisedagainstDavidson’sversion:1(nondeclarativesandnon-fact-statingsentences),2(testability),and6(takingtruthforgranted);anintensionaltheoristwouldmakemuchthesamerangeofrepliesaswedidonDavidson’sbehalf.Objection3(deixis)arisesinadifferentway,sincethepossible-worldsapproachdoesnotinvolveT-sentences,butitdoesarise,sincenoprovisionhasasyetbeenmadefordeixisintheintensionalapparatus.Objection3willbethemainbusinessofthenextchapter.Thepossible-worldsviewalsoinheritsthefirsttwoobjectionsmadeagainstthePropositionTheoryinchapter5:weirdnessandalienness.AsInotedinchapter4,itisonethingtotake“possibleworlds”asametaphororheuristicforexplainingawayoflookingatthings,asIdidinexplainingKripke’sviewofpropernames.Itisanothertoappealtothemdirectlyinserioustheorizing,astheintensionalsemanticistsdo.Inwhatsensearetherereallyalternativeworldsthatdonotreallyexist?ButthisisalargesubjectandIcannotgointoithere.2Thepossible-worldsviewisalsosubjecttoobjection4againstthePropositionTheory(neglectofmeaning’s“dynamicfeature”).Atthetime,werepliedsimplythat,evenifpropositionsdonothelpintheexplanationofhumanbehavior,behaviorisnottheprimarythingthatneedsexplain-ing;rather,themeaningfactsare.ButtheobjectionhasbeenpushedfurtheragainstbothversionsoftheTruth-ConditionTheory.OBJECTION7Thereisstillaproblemofsubstitutivity.Forthereseemtobecontextsinwhichsynonymous(notjustcoextensive)termscannotbeintersubstitutedwithoutpossiblechangeoftruth-value.“Ophthalmologist”and“eyedoctor”aresynonymous(orsowemaysupposeforconvenience).But,ifSheiladoesnotknowthat,“Sheilabelievesthateveryeyedoctordoctorseyes”maybetruewhile“Sheilabelievesthateveryophthalmologistdoctorseyes”isfalse;likewise“Irvingwenttoanophthalmologistbecauseanophthalmologistisaneyedoctor”istruewhile“Irvingwenttoanophthalmologistbecauseaneyedoctorisaneyedoctor”isfalse.\n132TheoriesofmeaningOBJECTION8SomeDavidsonians(forexample,Lycan1984)andsomeintensiontheoriststhinkofthekindofsemanticallychargedsyntaxIhavebeendescribingasamachineprogramforcomputinglargemeaningsfromsmallerones,aprogramthatisinsomesensebeingruninthebrainsofspeakersandhearers.Butthatideaisproblematic.Hereisamorespecificworryaboutthe“dynamicfea-ture,”pointedoutbyMichaelDummett(1975)andbyHilaryPutnam(1978).Dummett’sandPutnam’sownwritingsaredenseandsomewhatobscure,buthereisasimplewayofputtingoneoftheirconcerns:Asentencemeaningiswhatoneknowswhenoneknowswhatasentencemeans.Buttoknowwhatasentencemeansisjusttounderstandthatsentence.Andunderstandingisapsychologicalstate,onethatinheresinaflesh-and-bloodhumanorganismandaffectsthatorganism’sbehavior.Now,ifwhatasentencemeansisjustitstruthcondition,howcanknowledgeofatruthconditionperseaffectanyone’sbehavior,when(asiseasilyshownbyTwin-Earthexamples)truthconditionsareoften“wide”propertiesofsentencesinthesensethatthey“ain’tinthehead”andknowledgeoftruthconditionsisaconspicuouslywidepropertyofpeople?Thetruthconditionof“Dogsdrinkwater,”here,differsfromthatof“Dogsdrinkwater”onTwinEarth,butthedifferenceisirrelevanttobehaviorandcannotaffectit.Butunderstanding(=knowingmeaning)mustanddoesaffectbehavior.Thereforeunderstandingisnot,ornotmerely,knowledgeoftruthcondition,andsomeaningisnot,ornotsimply,truthcondition.FIRSTREPLYPutinthisway,theargumentassumesthat“understanding”mustitselfbea“narrow”or“inthehead”concept.Thatis,tosaytheleast,notobvious.(IleavetoyoutheexerciseofconstructingaTwin-Earthcounterexample.)Realizingthattheargumentneedsanarrowconceptofunderstandingalsoshouldmakeusreconsiderthesimpleequatingof“knowingmeaning”withunderstandingandviceversa,truisticasthatequatingmayhavesoundedatfirst.SECONDREPLYFurther,theargumentassumesthatwideconceptscannotpersefigureintheetiologyofbehavior.Asismadeclearbythe“intentionalcausation”litera-tureofsomeyearsago,3“figuringin”canbedoneinmanyways.Thereisnodoubtthatbehaviordependscounterfactuallyonwidestatesofpeople:HadIwantedwater(H2O),Iwouldhavegoneintothekitchentogetsome.AndIthinkthatisthestrongestetiologicalnotionguaranteedbycommonsense.Ifanyonethinksthatunderstandingaffectsbehaviorinastrongersenseof“affect”thanjustthatthebehaviordependscounterfactuallyontheunder-standing,wewouldhavetohearsomedefense.\nTruth-ConditionTheories:possibleworldsandintensionalsemantics133The“use”theoristisnotquitefinishedwiththetruth-conditionview.Weshallbeginchapter12byconsideringafurtherobjection.Summary•Asentence’struthconditioncanbetakentobethesetofpossibleworldsinwhichthesentenceistrue.•Moregenerally,possibleworldscanbeusedtoconstruct“intensions”forsubsententialexpressions,whichwillcombinecompositionallytodeterminethecontainingsentence’struthcondition.•Theresultingviewavoidsboththeproblemofcoextendingbutnonsynonymoustermsandtheproblemofnon-truth-functionalconnectives.•Thepossible-worldstheoryalsodeepensFrege’ssolutiontotheProblemofSubstitutivity.•ButthetheoryinheritsanumberofDavidson’soriginaldifficultiesandincursoneortwomore.Questions1EvaluateLewis’directargumentforthepossible-worldsversionoftheTruth-ConditionTheory.2Discussthepossible-worldstheoryfurther,pro,con,orboth.(Ifyoudonotalreadyknowsomepossible-worldssemantics,youwillwanttodoatleastabitofoutsidereading;IrecommendLewis(1970).)3Adjudicateobjection7orobjection8.Furtherreading•ThesimplestandmostnaturalintroductionIknowtothePossible-Worldsversionoftruth-conditionalsemanticsisLewis(1970).ThenworkuptoCresswell(1973).(Toughstuff,requiringknowledgeofformallogicandsettheory;butitallcamefromsomethingmuchtougher,collectedposthumouslyinMontague(1974).)•TwogoodtextbookintroductionstoMontagueGrammarareChierchiaandMcConnell-Ginet(1990)andWeisler(1991).\n\nPartIIIPragmaticsandspeechacts\n\n11SemanticpragmaticsOverviewLinguisticpragmaticsischaracterizedasstudyinglinguisticexpressions’usesinsocialcontexts.Buttherearetwoimportantlydifferentwaysinwhichanexpression’susedependsoncontext.First,owingtothepresenceofsuchdeicticelementsaspersonalpronounsandtense,asentence’spropositionalcontentvariesfromcontexttocontext(recallthat“Iamsicknow”saysdif-ferentthingsdependingonwhenitisutteredandbywhom).Second,evenoncethesentence’spropositionalcontenthasbeenfixed,thereareseveralotherimportantaspectsofitsusethatwillstillvarywithcontext.Semanticpragmaticsstudiestheformerphenomenon,thedeterminationofproposi-tionalcontentbycontext;pragmaticpragmaticsstudiesthelatter.DavidsondealswiththeproblemofdeicticelementsbycomplicatingthestandardformofhisT-sentences.Thepossible-worldstheoristsdealwithitbyrelativizingtruthtoasetofcontent-affectingcontextualfactorssuchasspeakerandtime.Butbothapproachesneedtobeliberatedfromhavingtolistafixedsetofcontextualfeatures.Semanticpragmaticshasacomplicatedrangeofdatatodealwith.Itmustnotonlychartthecomplicatedusesofpronouns,tense,andthelike,butsolvethegeneralproblemofdisambiguation:GiventhatnearlyeveryEnglishsentencehasmorethanonemeaning,howdoesaheareridentifythecorrectoneuponhearingthesentenceuttered?CharlesMorris(1938)dividedlinguisticstudyintosyntax,semantics,andpragmatics.Impressionisticallyput,thedistinctionwassupposedtobethis:Syntaxisthestudyofgrammar,thestudyofwhichstringsofwordsarewell-formedsentencesofagivenlanguageandwhy.Semanticsisthestudyofmeaning,construedprimarily(thoughasweknownotuncontroversially)asamatteroftherelationsthatlinguisticexpressionsbeartotheworldinvirtueofwhichtheyaremeaningful.Incontrast,pragmaticsstudiestheusesoflinguisticexpressionsinvarioussocialpracticesincluding,ofcourse,everydayconversationandcommunication,butnotonlythose.Onthisusage,Wittgenstein’sview(chapter6above)canbeputbysayingthateither“semantics”isentirelymisguidedoritcollapsesintopragmatics.\n138PragmaticsandspeechactsSemanticpragmaticsvs.pragmaticpragmaticsThesinglewordwehearmostofteninthestudyandpracticeofpragmaticsis“context,”meaningcontextofutterance.Pragmaticsisspecificallyaboutthefunctioningoflanguageincontext.Thismarksasignificantcontrast,becausesyntaxandsemanticshavegenerallyaspiredtobecontextless.Syntaxisaboutwhetherasentenceisgrammaticalorwhetherastringofwordsconstitutesagrammaticalsentence,period.Semanticshasalwaysfocusedonsentencemeaning,themeaningofasentencetypeinabstractionfromanyparticularusetowhichthesentencemightbeput.ButtherearealwayspestslikeWittgenstein,Strawson,andJ.L.Austinremindingusthattheveryideaofa“sentencetype”isaviolentabstractionfromlinguisticreality.Whenasentenceisuttered,itisinvariablyutteredinaparticularcontextbyapar-ticularspeakerforaparticularpurpose.Andthisissomethingthatcannotbeignored,forsolidreasonsthatIshalltrytomakeclearintheremainingchaptersofthisbook.Isaidthatthedistinctionbetweensemanticsandpragmaticswassupposedtobethattheformerdealswiththeacontextualmeaningsofsentencetypes,whereasthelatteraddressesthesocialusesoflinguisticexpressionsincon-text.Buttherearetworeasonswhythatcharacterizationistoosimple.Thefirstreasonisthatthereisanimportantsenseinwhichmostsentencetypessimplydonothaveacontextualmeanings.Thesecondisthat,asweshallseelateron,social-usefactorsinterpenetrateincertainspecialwayswithwhatwewouldotherwisethinkofaspropositionalmeaning.Hereisthesenseinwhichmostsentencetypeslackacontextualmean-ings.Recallthephenomenonofdeixis,broughtupinobjection3againsttheTruth-ConditionTheory,andconsideraheavilydeicticsentence.SupposeyouandIcomeintoanemptyclassroomandfindthefollowingwordswrit-tenontheblackboard:(1)Iamillandwillnotholdclasstoday.Unlesswecanfindoutwhohadwrittenthosewordswhenandtowhom,wedonotknowwhatexactlyhasbeensaid(eventhoughweknowsome-thingaboutwhathasbeensaid);wedonotknowwhatpropositionhasbeenexpressed.Intermsofthepossible-worldstheory,wedonotknowthesen-tence’sintension.Infact,ifthesentencehadbeenscribbledontheboardmerelyasalinguisticexampleandnoreferentshadeventacitlybeenassignedtoitsdeicticelements,itwouldnotexpressanypropositionatallandwouldnotevenhaveanintension.Thecommonmoraloftheoriginalobjection3andofthislastargumentisthatasentence’scompletetruthconditiondependsoncontextualfactors.And,evenifonedoesnotaccepttheTruth-ConditionTheoryofmeaning,onecanseethatasentence’smeaning,inthesenseofitspropositionalcon-tent,dependsoncontextinjustthesameway.\nSemanticpragmatics139Cresswell(1973)distinguishedbetweentwokindsofpragmatics:semanticpragmaticsandpragmaticpragmatics.1Semanticpragmaticsdealswiththoseelementsofmeaninginthesenseofpropositionalcontentthatsimplydodependoncontext.Itisthedisciplinethattellsushowpropositionalcontentisdeterminedbycontextualfeatures.Butbeforewesaymoreaboutitandexplainthecontrastingnotion,letusdealwithobjection3.TheproblemofdeixisReturningtoDavidson’sproblem:HeneedstofindawayofformulatingT-sentencesthataccommodatesdeicticorindexicalelementswithoutgettingtruthconditionswrong.ImentionedDavidson’sownproposalfordoingthis.OthernotableattemptshavebeenmadebyWeinstein(1974)andespeciallyBurge(1974),buthereIshallpresentasimpleideasuggestedbyHarman(1972).2WesawthatonedrawbackofDavidson’sproposalwasitslimitingthepotentiallyrelevantcontextualfactorstospeakerandtime.Therearemanyothers.Anobviousexampleisobjectsindicatedbythespeaker’spointinggesture,aswhenthespeakersays,“Thisoneismoreexpensivethanthatone,”successivelypointingtotwodifferentobjectsondisplay.Totakeamoreexoticexample,hemisphere.3“Itisautumn”istrueasIwritethisinNorthCarolina,USA,butitwouldnotbetruewereitutteredsimultaneouslyinSydneyorinBuenosAires.(Noristherelevanthemispherenecessarilydeterminedbythespeaker’slocation;itdependsontheaudienceandonconversationalpurposesaswell.IfIamconversingwithanAustralianaboutAustralianaffairs—evenifwearebothcurrentlyinNorthCarolinainNovember—Imaysay,“Becauseit’sspring,thestudentswillbejuststartingtothinkabouttheirfinalexams.”)Soweneedanapproachtodeictictargetsentencesthatdoesnotpresupposeafixednumberofcontextualvariables.Letusgetthewholejobdoneinonestroke.Wecanrelativize“true”tocontexts—sincewealreadyknowthatthetruthofasentencetypedoesvarywithcontext—andintroduceafunction,α,thatwilllookatadeicticele-mentoccurringinacontextandtelluswhatthatelementcontributesinthatcontexttopropositionalcontent.4Forexample,if(asisgenerallythought)thefirst-personpronoun“I”alwaysdenotesthespeaker,αwilllookatanoccurrenceof“I”inaparticularutteranceandmapthatexpressionontothepersonwhodidtheuttering.Forshort,α(“I”,C)—readas“whatαassignsto‘I’inthecontextC”—isthespeakerinC.Likewise,if“now”denotesroughlythetimeatwhichanutteranceisitselfuttered,thenα(“now”,C)isthattime.Andα(“tomorrow”,C)wouldbethedayimmediatelyfollowingtheutteringinC.Thenwecanwritetheright-handsidesofDavidson’sT-sentencesintermsofwhatαassignsinthecontextCtoeachdeicticelementinthetargetsen-tence.Thus:\n140Pragmaticsandspeechacts“Iamsicknow”istrueinCifandonlyifα(“I”,C)issickatα(“now”,C).“Iamillandwillnotholdclasstoday”istrueinCifandonlyifα(“I”,C)isillduringα(presenttense,C)5andα(“I”,C)doesnotholdclassduringα(futuretense,C)onα(“today”,C)“Shehasneverbeentoakaraokebar,butyouandshewillbevisitingonetomorrowmorning”istrueinCifandonlyifα(“She”,C)doesnotgotoakaraokebarduringα(perfecttense,C)butα(“you”,C)andα(“she,C)visitakaraokebarduringthemorningofα(“tomorrow”,C).Problemsolved.Davidson’stechnicalproblemofformulatingT-sentences,thatis;furtherphilosophicalquestionscanandwillberaisedaboutα.Theintensionallogiciansdealtwithdeixisbyrelativizingtruthtoan“index,”whichwasafixedsetofcontextualvariables.Montague(1968/1974)andScott(1970)tookanindextobeasetofeightstandardelementsconsist-ingofapossibleworldw,atimet,aplacel,aspeakerp,anaudiencea,asequenceofindicatedordemonstratedobjectsi,a“discourse-segment”d,anda“sequenceofassignmentstofreevariables”s(nevermindwhatthoselasttwomean).Inthissystem,atruthconditionassignmentwouldlooklikethis:“Iamsicknow”istrueatifandonlyifinw,pissickatt.ButthissharesthedrawbackofDavidson’smethod,thoughnotasseverely,inthatitarbitrarilyrestrictsthenumberofcontextualfeaturesthatcanbecited.6Wehavenowayofforeseeingwhatfurthersuchfeaturesmightbecomerelevanttothetruthofanutterance.Forexample,wehavealreadyintroducedoneunexpectedvariable,hemisphere(asbetweensouthernandnorthern).Thereareplentyofothers,seeminglywithoutlimit.Thetruthof“Itis5:00p.m.”dependsontimezone,anentirelyconventionalconstruct.(AsWittgensteinoncenoted,timezonesareboundtoourplanet;“Itis5:00p.m.onthesun”hasnotruth-value.)Andsomelocutionspresupposeakindofvantagepoint,oftendistinctfromtheplaceofactualutterance,thatcanshiftevenwithinasinglesentence(Fillmore1975;Taylor1988).Take(2a)ZonkerwenttoUncleDuke’sparty.(2b)ZonkercametoUncleDuke’sparty.(2c)I’mgoingouttocleanthepasturespring;...—Youcometoo.(RobertFrost,ThePasture)\nSemanticpragmatics141(2a)and(2b)canhavethesametruthcondition,but(2b)canbesaidproperlyonlybyaspeakerwhoseassumedvantagepointisthepartylocationitself.(Notetoothatwhatcountsisthevantagepointatthetimeofthepartyunderdiscussion,notatthetimeofutterance;thisisyetanothercontextvariable,usuallycalledthereferencetime.)In(2c),thevantagepointshiftsfluidlyfromtheplaceofutterancetothepasturespring,oratleasttosomewherealongthejourneyatwhichthespeakerispicturedasaheadofthehearer.ArrivingatPrincetontogiveatalk,IammetbyaformercolleaguewhomIlastsawteachingatWellesley.Iask,“Areyouherenow?,”askingnotwhethersheisphysicallylocatedinPrinceton(duhh)butwhethersheisnowemployedinthePrincetonphilosophydepartment(Nunberg1993:28);thustruth-valuecanvarywithemployinginstitution.Ortake(3)TomorrowisalwaysthebiggestpartynightoftheyearutteredontheFridaybeforeclassesbegin(Nunberg1993:29;NunbergcreditsDickOehrlewiththeexample).“Tomorrow”in(3)cannotrefer,asitusuallywould,tothedayornightfollowingthedateofutterance;itreferstoatypeofdateonthestudents’academiccalendar,namelytotheannualSaturdaybeforeclassesbegin.Icouldgoonandon.Themoralisthatwecannoteverbesurewehaveanticipatedallthecontextvariablesthatcanaffecttruth-value.SoIwouldadvisetheintensiontheoriststoavailthemselves,instead,ofHarman’smightyassignmentfunctionα.TheworkofsemanticpragmaticsThetrickistofindouthowαiscomputed;thatis,whatrulesweuseinparticularcontextstofillinthemissingchunksofpropositionalcontentcorrespondingtodeicticelements.Presumablyeachsuchelementinthelan-guageisgovernedbyanappropriaterule.Forexample,wemightlookatthepronoun“I”andsuggestthat,inagivencontext,“I”alwaysdenotesthespeaker.Turningto“now,”itseemsreason-abletosaythat“now”alwaysrefersincontexttothetimeoftheutterance.Infact,thesefirstattemptsaretoosimple.“I”canbeusedasadeviceofdeferredreferencetoapositionorrole,aswhenthecondemnedprisonersays“IamtraditionallyallowedtoorderwhateverIlikeformylastmeal”(Nunberg1993:20).Sometimes“I”isusedinformulatingageneralization,asin“IfI’mamusicdepartment,I’masnakepit.”Thetemporalreferenceof“now”canbedeferredalso,aswhenwearelookingatatime-linerepresentationoftheevolutionoflifeand,pointing,Isay,“Nowthedinosaursappear,”orwhenyouleaveamessageonyouransweringmachinethatsays“Iamnothomenow.”“Now”issometimesspatialratherthaninanywaytemporal—“NowHillsboroughRoadcrossesAirportRoadandbecomesUmsteadDrive”—andsometimesnotevenspatiotemporal—“Nowcomesthefirstprimenumber\n142Pragmaticsandspeechactswhosesquareisgreaterthan1,000.”Butonejobofsemanticpragmaticsistorefinesuchrulesuntiltheyareadequatetothedata.TheintensionallogicianDavidKaplan(1978)thinksofsuchrulesasfunctions.Asanintensionisafunctionfromworldstoextensions,asemantic–pragmaticruleisafunctionfromcontextstointensions.Atthelevelofthesentence,theintensionisafunctionfromworldstotruth-values.Kaplancallsthatthesentence’s“content”and,asbefore,itcorrespondstothetraditionalnotionofaproposition.Thecompositesemantic–pragmaticruleisafunctionfromcontextstocontents;Kaplancallsthat“character.”Contentiswhatisleftundeterminedbythedeicticsentencesinourexamples;characteriswhatdoesdeterminecontentgivenalltherelevantcontextualfeaturesofacontextofutterance.Thus,whenwecomeupontheblackboardsentence,charactertellsustolookforthespeaker(tofindα(“I”,C)),andthehearerandthedateofutterance;onceweknowthose,wewillknowwhathastoobtaininapossibleworldinorderforthesentencetobetrueinthatworld.Isaidthat,whenweencountertheblackboardsentenceunprepared,wedonotknow(infull)whatitsays.AndIwasright.Butthereisanotherperfectlygoodsenseinwhichweunderstandthesentenceitself,andvirtuallyanyEnglishspeakerunderstands“Iamsicknow”entirelyoutofcontext.Kaplanarguesthatthe“m”-wordshouldbereservedforcharacterratherthanforcontent,ontheentirelyreasonablegroundthatordinaryEnglishspeakerssurelyknowthemeaningsofeverydaydeicticsentencesevenwhentheydonotknowthecontextualparametersthatwouldfixthosesentences’contents.Yetcontentinhissenseisalsostillaperfectlygoodthingtomeanby“meaning.”Itishardlyamatterforheateddispute.Computingαand/orcharacterizingcharacterisnottheonlytaskofsemanticpragmatics.Anotherandhorriblyvexedoneisdisambiguation.Manysentences,suchas“Visitingphilosopherscanbeboring,”“Tedislyingaboutmeditating,”and(PaulZiff’sexample)“Themousetoreupthestreet,”areobviouslyambiguous.Andinfact,almosteverysentenceweeverencounterinlifeistechnicallyambiguous,inthesensethatithasoneormorepossibleiffarfetchedmeaningsinadditiontotheonethatwouldnormallybeintendedbyanutterer.Yetwerarelypausetothink,orevennoticethatwearechoosingfromamongarangeofpossiblemeanings(notmerelyfillinggapsinanotherwiseuniquepropositionalcontent).Howwedothisisadeepquestion,muchdeeperthanthatofhowwecomputeα.Certainlytoodeepforthisbook,thoughsomehintswillbefurnishedinchapter13.Thatissemanticpragmatics.Incontrast,pragmaticpragmaticstakespropositionalcontentforgranted,andaskswiderquestionsabouttheuseofsentencesincontexts.Oneandthesamesentencewithanalreadyfixedpropositionalcontentcanstillbeusedtodointerestinglydifferentthingsindifferentcontexts.Asweshallseeinourremainingchapters,thereiscon-siderablymoretoproducingandunderstandinglanguagethanjustgraspingpropositionalmeaning,hardasthelatteristoexplain.\nSemanticpragmatics143Summary•Linguisticpragmaticsischaracterizedasstudyinglinguisticexpres-sions’usesinsocialcontexts.•Semanticpragmaticsstudies,inparticular,thedeterminationofpropositionalcontentbycontext.•DavidsondealswiththeproblemofdeicticelementsbycomplicatingthestandardformofhisT-sentences.•Thepossible-worldstheoristsdealwithitbyrelativizingtruthtoasetofcontent-affectingcontextualfactorssuchasspeakerandtime.•Butbothapproachescanbefreedfromhavingtolistafixedsetofcon-textualfeatures,ifweavailourselvesoftheassignmentfunctionα.•Inadditiontotrackingthecomplicatedusesofparticulardeicticexpressions,semanticpragmaticsischargedwithsolvingtheterribleproblemofdisambiguation.Questions1IsthereabetterwayforDavidsonorforthepossible-worldstheoristtosolvetheproblemofdeixisthanbyimportingHarman’sassignmentfunctionα?Inparticular,doesαcreatenewdifficultiesofitsown?2Takeanexpressionsuchas“I”or“now”(or“tomorrow”or“recently”or“west”...)andtrytostatetheexactruleaccordingtowhichitaffectsthepropositionalcontentofasentenceinwhichitoccurs.3Makeatleastafeeblestartontheproblemofdisambiguation.(Donotexpectimpressiveresults.)Furtherreading•Forrecentworkonthesemantics–pragmaticsdistinction,seeSzabó(2005).•ForasomewhatlesstechnicaldiscussionofindexicalsthanKaplan(1978),seeKaplan(1989).Recanati(1993)takesuptheDRapproachtoindexicals.•Yourgrau(1990)isagoodcollectionondemonstratives.•Taylor(1988)andNunberg(1993)containexcellentexamplesofunusualindexicalconstructions.\n12SpeechactsandillocutionaryforceOverviewJ.L.Austincalledourattentiontowhathecalleda“performative”utteranceofadeclarativesentence,wherebyoneperformsaconventionalsocialactbutdoesnotstateordescribeanything—forexample,“Iapologize”or(inagameofbridge)“Idouble.”Thekindsofactsthatcanbeperformedinthiswayarecalledspeechacts.Eachtypeofspeechactisgovernedbyrulesoftwosorts:constitutiverules,whichmustbeobeyediftheactistohavebeenaccom-plishedatall,andregulativerules,violationofwhichmerelyrenderstheactdefectiveor,inAustin’sword,infelicitous.Therearemanyandsurprisinglyvariouswaysinwhichagivenspeechactcanbeinfelicitous.ButAustincametoseethatthereisnoprincipleddistinctionbetweenperformativeutterancesandthoseofordinarydeclaratives.Rather,everyutterancehasaperformativeaspectorillocutionaryforce,whichdeterminesthetypeofspeechactperformed,andvirtuallyeveryutterancehasadescrip-tiveorpropositionalcontentaswell.Further,manyutteranceshavefeaturesthatincorporatetheirdistinctiveeffectsonhearers’mentalstates;thesefeaturesarecalledperlocutionary.JonathanCohenhasraisedanastyproblemaboutthetruthconditionsofsentencesthatcontainexplicitperformativeprefacesthatspecifythetypeofspeechacttobeperformed,forexample,“IadmitthatIhadseveralprivateconversationswiththedefendant.”Nosatisfactorysolutiontothatproblemhasbeenfound.WilliamAlstonandStephenBarkerhaveofferedadistinctivetypeof“use”theoryofmeaning,basedontheillocutionarynotionofaspeechact.PerformativesConsiderthefollowingsentences.(1)Ipromisetopayyouforthediapers.(2)Ipronounceyouhusbandandwife.(3)IchristenthisshiptheLudwigWittgenstein.\nSpeechactsandillocutionaryforce145(4)Iapologize.(5)Idouble.[Abidinbridge.](6)Raiseyoufive.[Abetinpoker.](7)Nay.[Avoteonaformalmotion.]Exceptpossiblyforthelasttwo,thesearedeclarativesentences,so(inparticular)theverificationistmustaddressthem;whataretheirrespectiveverificationconditions?Perhapsthatquestionistoohard,orunfairinviewofQuine’sDuhemianobjection.Butwhataretheirtruthconditions?WecoulddirectT-sentencesuponthem.Forexample,“Ipromisetopayyouforthediapers”istrueifandonlyifIpromisetopayyouforthediapers.Really?(No,notreally.)“Idouble”istrueifandonlyifIdouble.Possibly;perhaps“Idouble,”assaidbymeontheappropriateoccasion,istrueifandonlyifIdodoubleonthatoccasion.Butitseemsthatweareleav-ingsomethingout,somethingthatismoreimportantthantheutterance’sslightlydegeneratetruthcondition.AsJ.L.Austin(1961,1962)mighthaveputit,whenIsay“Idouble,”Iamnotdescribingmyselfasdoingsomedou-bling;Iamactuallydoubling,andnothingmore.(Doublingisamoveinthegameofbridge.Itispartofareallanguage-game,intheliteralsense.)Andnoonecouldacceptablyrejoin,“That’sfalse,youdon’tdouble.”Ifsomeoneelsethensaysofme,“Hedoubled,”thatisatruereportofwhatIdid.ButwhenIoriginallysayit,simplyasamoveinthegame,itdoesnotseemlikeacandidatefortruthorfalsity.“Nay”istrueifandonlyifnay.Forgetit;that“T-sentence”isnotevengrammatical.WehaveherethebasisforafurtherobjectiontoverificationismandtotheTruth-ConditionTheory,acompoundofaWittgensteinianobjectionandourfirstobjectiontotheTruth-ConditionTheory.AWittgensteinianmightlookat(4),(5),and(7),especially,assimilatingthemtothebuilders’primitive(“Slab!”)language,andremindusagainofthemanydevicessuchas“Hello”and“Oh,dear”thathaveconventionalsocialusesandareperfectlymeaningfulwithouthavinganythingtodowithverificationorwithtruthitself.Evenwhenweturntothemorehighlystructured(1)–(3)and(6),itseemsthat,althougheachisdeclarativeinmood,noneisinthebusinessof\n146Pragmaticsandspeechactsstatingafactordisclosingatruth.Theyareindifferentlinesofwork;sotheyseemtocountas“factuallydefective.”InAustin’soriginalarticle(1961),hecalledsentenceslike(1)–(7)“perfor-matives,”todistinguishthemfrom“constatives”(constativesbeingjusttheusualsortsofdescriptive,fact-stating,true-or-falsesentencesthatphiloso-pherslike).Inutteringaperformativeoneisnot,oratleastnotostensibly,describinganythingorstatingafact,butperformingasocialact.WhenIutter(1),Iamactuallymakingthepromise.WhenIutter(4)Iamsimplyapologizing.WhenIutter(6)Iaddtomybet,makingafinancialcommit-ment.WhenIutter(3),inthepropercontextwithabottleoftheappropriatekindofchampagne,Iactuallyperformthechristening.Austincalledsuchsocialactions“speechacts,”andsogavebirthtothebranchoflinguisticsandphilosophyoflanguagethateversincehasbeencalled“speech-acttheory.”Whatevertheoutcomeforanyone’stheoryofmeaning,wemuststudythephenomenonof(inAustin’stitlephrase)“doingthingswithwords,”onpainofleavingoutaveryimportantrangeoflinguisticphenomena.(Therearetwofurtherreasonsaswell.Oneisthatspeech-acttheoryisthemosteffectivecureforphilosophers’otherwiseovermasteringtendency,vividlyexemplifiedinthisverybooksofar,tothinkthatdeclarativesentencesaretheonlyonesthatmatter.Theotheristhatmanymistakeshavebeenmadeandfallaciescommittedinareasofphilosophyotherthanphilosophyoflan-guage,throughignoranceofspeech-acttheory;butspacedoesnotpermit.)ILLOCuTION,LOCuTION,ANDPERLOCuTIONNaturally,Austinbeganseekingaworkable,fairlyprecisetestforperfor-mativity.Hetriedtocharacterizethenotionsyntactically,andranintovariouskindsoftroublethatneednotdetainus.Butinhis1961articlehesettledfairlycomfortablyontheso-called“hereby”criterion:Anutteranceiscountedasperformativeifonecouldfairlyinterpolatetheword“hereby”beforethemainverb.Thus,(1)isperformativebecausethespeakercouldaswellhavesaid,“Iherebypromisetopayyou...”The“hereby”emphasizesthattheactinquestion,herethegivingofapromise,isconstitutedbythespeaker’sveryutteranceitself.Thecriterionworkswellfor(2)–(6)also:“Iherebypronounce...,”“Iherebychristen...,”andsoon.“Iherebydouble”wouldbestilted,butitsmeaningwouldbeperfectlyinorder.Certainlythecriterionmarksoffperformativesfromconstatives.IfIutteraparadigmaticconstative,suchas“Thecatisonthemat,”Icouldnothaveinserted“hereby.”“Thecatisherebyonthemat”isnonsensicaloratleastfalse,becausethecatis(orisnot)onthematregardlessofmysayingthatitis.Mysayingitdoesnothingtomakeitso.Austinnoticedapeskyclassofclearnonconstatives,apparentperforma-tives,thataretoosimpletopassthe“hereby”test.Actually(7)mightbetakenasanexample,since“Herebynay”isungrammatical.Butitisplausibletosaythat“Nay”isjustalaconicformof“Ivotenay,”whichdoesmeetthe“hereby”condition.\nSpeechactsandillocutionaryforce147However,whatabout“Hooray!,”“Shame!,”and“Damn!”?Noneoftheseadmits“hereby,”andtheyareharderthan“Nay”wastohearasmerelyshortfordeclarativescontainingperformativeverbs.Onemighttryarguingthat“Hooray!”reallymeans,“Iherebycheer”;Lewis(1970:57–8)proposedtounderstand“HoorayforPorky”as“IcheerPorky.”Perhaps“Shame!”means“Iherebycastigateyou”and“Damn!”means“Iherebycurse.”Butthesehypothesesarenotobviouslycorrect.Austingrewfarmoredeeplydissatisfiedwiththeperformative/constativedistinctionwhenhenoticedanothertypeofsentence.Consider:(8)IstatethatIhavenevertraveledtoaCommunistcountry.(8)passesthe“hereby”test,andsoshouldbecountedasperformative.WhenIsayit,Itherebydoperformacertainspeechact:anactofstating.Butitisalsoindeliblyfact-stating,descriptive.Indeed—whetherornotthespeakerhasinfactevertraveledtoaCommunistcountry—thatisitswholepoint;theoperativeverbis“state.”Thespeaker’sstatementistrueorfalse.If(8)isutteredunderoathandthespeakerhastraveledtoaCommunistcountry,thespeakercanbeindictedforperjury.So(8)seemstobeeitherbothperforma-tiveandconstative,orneither.Andtherearemore:(9)Ijudgethatweareoverextendedintheareaofsealskinfutures.(10)IreportthattheCommitteehasvotedunanimouslytoexpelGrannie.(11)IadviseyouthatitwouldbeverystupidtobuymoreAmalgamatedAmalgamstock.(12)IwarnyouthatthatRottweilerhasbeenstarvedforthreedaysandispeevish.Even(1)hasaparaphrasewithsimilarlyconstativefeatures:“IpromisethatIwillpayyouforthediapers,”whichatleastassertsthatIwillpayyou.SuchexamplesmadeAustinrealizethatasingleutterancecanhavebothaperformativepartoraspectandaconstativepartoraspect.Infact,virtuallyeveryutterancedoes,evenifitlacksanexplicitperformativeprefacelikethosethatbegin(8)–(12).Ifinsteadof(8),Itestifymerely“IhavenevertraveledtoaCommunistcountry,”Istillperformanactofstating,inaddi-tiontomerelyexpressingthepropositionalcontentthatIhavenevertraveledtoaCommunistcountry.WheneverImakeanassertion—thatis,wheneverImakeanutterancehavingtheforceofanassertion—Iperformanactofasserting.Declarativescanbeutteredwithotherforcesaswell.IfIweretodeletetheperformativeprefacesfrom(9)–(12)andsayonly“Weareoverextended...,”\n148Pragmaticsandspeechacts“TheCommitteehasvoted...,”andsoon,inthesamecontexts,thoseutter-anceswouldhavetheforcesrespectivelyofajudgment,areport,anadvising,andawarning.Austincalledthistypeoffeature“illocutionaryforce,”andhecontrasteditwith“locutionary”orpropositionalcontent.1Indifferentcontexts,thesamedeclarativemayhavedifferentillocution-aryforces.“ThatRottweilerhasbeenstarvedforthreedaysandispeevish”couldhavetheforceofathreatratherthanawarning;oritcouldbemerelyanobservation;or(notice)itcouldbeasoothingreassurance.Evenchildrenrecognizedifferencesinpotentialforce:Acomplaintsuchas“Ifyoudon’tquititI’mgoinghome”ismetbythegibe,“Isthatathreatorapromise?”Turningtonondeclaratives,itisconsiderablymoreobviousthattheyhavedistinctivevarietiesofforces.Infact,thepointofmoodssuchasinterroga-tiveandimperativeseemstobetoindicaterangesofillocutionaryforce.(13)AreyouamemberoftheSalvationArmy?couldbeparaphrasedas“I(hereby)askyouwhetheryouareamemberoftheSalvationArmy,”andlikewisefor“wh”-questionssuchas“WholetFluffyoutofhercage?”(14)GototheMusicLibraryandfindacopyofLanaWalter’sPetiteMasscouldhavetheforceofacommand,anorder,amererequest,orjustasug-gestion,dependingontheintentionsandpurposesofspeakerandhearerandonwhateverpowerrelationorinstitutionalauthoritymayobtainbetweenthem.2ThusdidAustin’soriginaldistinctionbetweenperformativeandconstat-iveutterancesturnintoadistinctionbetweenforceandcontentasaspectsofasingleutterance.Austin(1962)elaboratedamassivecatalogueofdif-ferentillocutionaryforcesandthefactorsthatdistinguishthem.Hereareafewfurtherexamplesofdistinctillocutionaryacts:admitting(ineitheroftwosenses);announcing;assuring;authorizing;censuring;committing;complimenting;conceding;confessing;congratulating;defining;denying;granting;hypothesizing;inquiring;insisting;pardoning;pleading;pledging;predicting;proposing;reprehending;thanking;urging;vowing.Austinintroducedathirdfeatureofutterances,inadditiontoillocution-aryforceandlocutionarycontent:perlocutionaryeffect.Someverbsarelikeperformativeverbsinmeaningakindofsocialactperformedbylinguisticmeans,butfailthe“hereby”testbecausetheydescribetheactintermsofitsactualeffectsonthehearerratherthanintermsofthespeaker’sinten-tion.Take“frighten,”and“convince.”Icannotrightlysay“Iherebyfrightenyou”or“IherebyconvinceyouthatGranniedidit,”becausewhetheryouarerespectivelyfrightenedorconvinceddependsinpartonyouandisinnowayguaranteed(muchlessconstituted)bymyutteranceitself.Actsoffrightening\nSpeechactsandillocutionaryforce149andconvincingarewhatAustincallsperlocutionaryacts;theyarethingswedowithwords,butnotinthesameintimatesenseasareillocutionaryacts.Hereareafewmoreexamplesofperlocutionaryacts:alarming;amazing;amusing;annoying;boring;embarrassing;encouraging;deceiving;distract-ing;impressing;informing;inspiring;insulting;irritating;persuading.TheVerificationandTruth-ConditionTheoriesofmeaningidentifyasentence’smeaningwiththesentence’spropositionalorlocutionarycontentalone.Butisnotillocutionaryforceakindofmeaning?Certainly,ifyoudonotunderstanddistinctionsofforce,thenthereisanimportantaspectoflanguagethatyouhavenotyetmastered.Soitseemsthattheverificationistsandtruth-conditiontheoristshaveleftsomethingout.Theymayreply:“Important,ofcourse;pragmaticpropertiesareimportantinreallife.Butthey’renotpartofmeaning.”Ibelievethisisjustaschoolyardscuffleoverthe“m”-word,whichisoftenusedmoregenerallyasanumbrellatermforwhateveraspectsoflinguisticactivityareconsideredimportant.Wealreadyknowthattherearekindsofmeaningbesideslocutionarysentencemeaning—speaker-meaning,forexample.Nowwecanaddthathereisnowanillocutionarykindofmeaning,force,whichisnotthesamethingaslocu-tionarymeaningeither.Eachofthesekindsofmeaningisperfectlyrealandindispensabletolanguageuse.3InfelicitiesandconstitutiverulesSpeechactsareconventionalacts;justasany“use”theoristwouldhaveit,theyareembeddedinanddefinedbysocialcustoms,practices,andinstitu-tions.Theirperformingsaregovernedbyrulesofmanykinds.Therulesareusuallyunwritten,merelyimplicitinnormativesocialbehavior.Searle(1965,1969)dividesspeech-actrulesintoconstitutiverulesandregulativerules.(Merely)regulativerules“regulateantecedentlyorinde-pendentlyexistingformsofbehavior,”whereasconstitutiverules“createordefinenewformsofbehavior”(1969,p.33).Thus,forexample,rulesofeti-quetteregulateactivitiesorpracticesthatexistquiteindependentlyofthoserules:“Officersmustweartiesatdinner”;“Neverchewwithyourmouthopen.”ButtherulesofchessorofAmericanfootballactuallydefinethegameinquestion,andthegameassuchwouldnotexistwithoutthem:“Bishopsmoveonlydiagonally”;“Atouchdownisscoredwhenaplayerhaspossessionoftheballintheopponents’endzonewhileaplayisinprogress.”Wemayintroduceamoredemandingandmoreinterestingnotion:Astronglyconstitutiveruleisarulewhoseviolationabortsthepurportedspeechact.SupposeIutterasentencewiththeintentionofperformingacertaintypeofspeechact,A.IfIviolateastronglyconstitutiverule,itfollowsthatIhavesimplyfailedtoperformanactoftypeA.Forexample,iftomorrowIutter(3)andbreakabottleovertheprowoftheUSSNorthCarolina,Idonotsucceedinchristeningher,forIhavenotthestandingorauthoritytodoso.(TheU.S.Navyhasexplicitrulesforappointingthedignitarieswhochristen\n150Pragmaticsandspeechactsbattleships.ThereisalsothefactthattheNorthCarolinahasbeenchristenedalready,on12June1940.)4Ifaclergymanutters(2)toayoungcouplestand-ingbeforehiminaChicagochapel,butisnotlicensedtoperformmarriagesinthestateofIllinois,orifoneofthecoupleisnotoflegallymarriageableage,theweddingdoesnotsucceed(indeedisnotaweddingatall,despitetheorganmusic,rings,andrice).Toraisesomeonefivebyuttering(6),Ihavetobeplayingpokeratthetime,andfivehastobewithintheagreed-uponbettinglimit.Violationofamerelyregulativeruleislessgrave.IfIutterasentenceintendingtoperformaspeechactoftypeA,andviolatenoconstitutiverulesbutdoviolatesomeregulativerule,theresultisthatIdoperformanactoftypeA,butdefectivelyor,inAustin’sofficialvocabulary,“infelicitously.”Iftheweddingdidsucceed,buttheresultingmarriageisonlyoneofcrassconvenienceandthecouplewerelyingtheirheadsoffwhentheyspoketheirvows,theweddingwasdefective;itisaregulativeruleofmarriagethatthecoupleloveeachotherandsincerelyintendtoremainmarried.Promisingisacloselyrelatedexample:IfIutter(1)insincerely,havingnointentionofpayingyouyourmoney,itisaninfelicitouspromise.Forthatmatter,ifIshout(1)toyouacrossacrowdedroom,andyoucannothearme,thatisaninfelicityofadifferentsort.Thereareborderlinecasesasbetweenstronglyconstitutiveandregulativerules.WhatifIutter(4),butinaflauntedlyunrepentant,jeering,sneeringtone?Isthatagrievouslyinfelicitousapology,ornoapologyatall?Austin(1962)wasgreatlyconcernedtoemphasizethemultifariousnessofinfelicity.Anutterancecangowronginanyoneofanynumberofquitedifferentways.Itcanbeanill-advisedmoveinagame,aswhenoneutters(6)becauseonehasmiscalculatedtheodds.Oritmaybeinsincere.Oronemaylackthestandingorauthoritytoperformanactofthekindintended.Oritmaybeveryrude.Oritmaybemadetoosoftlyandgounheard.Oritmaybemade,tactlessly,infrontofthewrongpeople.Oritmaybeverboseandpompousandblatheronandon.Oritmaypresupposesomethingfalse,asifIweretoapologizefordoingsomethingthatmyhearerhadwanteddone,orthatwasnotinanywayabadthingtohavedone,orthatIdidnotevendoatall.Thisgreatvarietyofdefectswillbecomephilosophicallyimportantlateron.Inparticular,nowthatwehaverecognizedthatsomespeechactsareactsofstating,asserting,andthelike,weseethatfalsityisonecommonbesettingdefectofsuchacts;aregulativeruleregardingactsofthatclassisthatwhatissaidshouldbetrue.Austincomplainsatlengththatphilosophersareobsessedwith“thetrue–falsefetish,”themisguidedideathattruth-valueisallthatmattersinspeech.Inparticular,weoftenmistakeotherkindsofinfelicitiesforfalsity;whenwehearasentencethatissomehowdefective,wetendtoassume,fallaciously,thatthesentenceisnottrue.(Inchapter13weshallscouttwoinstancesofthisfallacy.)Therearemanywaysinwhichutterancescangowrong—badly\nSpeechactsandillocutionaryforce151wrong—withoutbeingfalse.Falsityisjustoneformofinfelicityamongmanyothers.5Cohen’sproblemJonathanCohen(1964)raisedanastyproblemaboutsentenceslike(8)–(12).Itisaproblemabouttruthconditions.Take(8)(“IstatethatIhavenevertraveledtoaCommunistcountry”).Whatis(8)’struthcondition?Cohensays(p.121)“ItistemptingatfirsttosupposethatinAustin’sviewthemeaningofourutteranceisfoundtotallyintheclausethatfollowstheperformativepreface.”Substituting“truthcondition”for“meaning,”itisindeedtemptingtoreadthetruthconditionoutfromundertheperformativepreface.Forwhatanuttererof(8)statesisthats/hehasnevertraveledtoaCommunistcountry,notthats/heisstatingsomething.Onecouldhardlyevadeaperjurychargebyresponding“ThesentenceIutteredwastrue,notfalse:IdidindeedstatethatIhadnevertraveledtoaCommunistcountry;thefactthatIactuallyhadtraveledtoaCommunistcountryisirrelevant.”Likewise,surely(9)–(12)arenotautomaticallytruesimplybecauseIdidrespectivelysojudge,report,advise,andwarn.(ThoughLewis(1970)tookexactlythatboldposition.)Thelocutionarycontent,oratleastthetruthcondition,isjustthatIhavenevertraveledtoaCommunistcountry,andthe“Istatethat”isjusttheperformativeprefacethatmakestheforceexplicit.Afurtherargumentforthis“tempting”viewisthatformal,explicitperfor-mativessuchas(8)–(12)and(15)seemtobejustverbose,inflatedequivalentsofthesimplerstatements,warnings,orders,andsoononecouldhaveissuedwithouttheperformativeprefaces.ButCohenraisesaseriousobjectiontothetemptingview.Consideranyof(8)–(12).SupposeEleanorutters(12)toFranklin,andLucy,overhearing,says“She’swarnedhimthatthatRottweilerhasbeenstarved...,”or“EleanorhaswarnedFranklinthatthatRottweilerhasbeenstarved....”Ineachcase,Lucyreferstojustthesameindividualsandpredicatesjustthesamerelationbetweenthem,andonlythetensehaschanged.Surely,inparticular,“warn”in(12)meanswarn.Thewordsthatoccurin(12)’sperformativeprefacehavetheirstandardsensesandreferents.So,theprefaceisnotjustatagorflagmeanttosignalforce.(Therearesuchtagsandflags;grammaticalmoodisbasicallythat,asimplerange-of-forceindicator.But“Iwarnyouthat”andtheotherprefacesin(8)–(11)arenotjustforcelabels;theyhaveinternalgrammaticalstructureandtheirpartshavetheirownmeaningsandreferentialproperties.)Sowhy,then,wouldwegettopretendthatthosepartsofthesentencesdonotexist,andreadthelocutionarymeaningoutfromunderthem?Itgetsworse.Asitturnsout,theideathatperformativeprefacesaremerelyforcelabelsissimplyuntenable.Forsuchprefacescanhavealotofstructure.Forexample,theycancontainadverbialmodifiers.Longadverbialmodifiers.\n152Pragmaticsandspeechacts(15a)IadmitfreelythatIhadseveralprivateconversationswiththedefendant.(15b)IadmitwithreluctancethatIhadseveral....[Noticethat“withreluctance”modifies“admit,”not“hadseveral....”](15c)IadmitgladlyandwiththegreatestpleasurethatIhad....(15d)BecauseIamconcernedtotellthewholetruth,IadmitthatI....(15e)MindfulthatthereisajustandmightyGodinHeavenwhopunishesthosewhowithholdinformationincourtsoflaw,andinmortalfearofthewormthatdiethnotandthefirethatisnotquenched,Iadmit....Accordingtothetemptingview,theonlylocutionarycontentin(15a–e)isthatoftheircommoncomplementclause(“Ihadseveralprivateconversationswiththedefendant”).Butthatclaimgrowslessplausibleasweworkourwaydownthelist.(15d)’sperformativeprefacecontainsanentireclausethatthespeakerasserts,thoughinpassing,asfact.(15e)’scontainsseveralsomewhatcontroversialassertions;ifIweretoutter(15e)seriously,youcertainlycoulddescribemeafterwardsashavingexpressedahighlycontentfultheologicalview.Andnotjustexpressed;thetheologycertainlyseemstobepartofwhatissaid.Itseemsthetemptingviewcannotbesustained.Whatbecomestemptingatthispointisinsteadtogobackandadmitthatthesentences’locutionarycontentsincludetheirperformativeprefaces.(Callthisthe“liberal”view.)Whatissobadaboutthat?Hereiswhatissobad,incasewehaveforgotten.Iftheliberalviewiscorrect,then(8)–(12)aresimplyandautomaticallytruewhenevertheyareutteredandtherelevantconstitutiverulesarenotviolated.Noperjurychargecouldbemadetostick,ifthewitnesswerecarefultotestifyonlyinexplicitperformativessuchas(8).Noticethatsemantically,(8)–(12)wouldnotevenentailtheircomplementclauses(becauseonecanstate,report...thingsthatarenotso).Myutteringof(10)wouldnotcommitmesemanticallytotheclaimthattheCommitteehasvotedunanimouslytoexpelGrannie.Now,Cresswell(1973)andBachandHarnish(1979)havejoinedthelib-eralviewinquestioningAustin’svehementdenialthattheagentsofspeechactsaredescribingthemselvesasperformingthoseacts;butthesephiloso-phershavesuggestedthat,inadditiontothemainacts,theutterersarealsodescribingthemselvesasperformingthem.Thus,ifIutterIorderyoutoattackandcapturetheUniversityofChicago(16)\nSpeechactsandillocutionaryforce153mymainspeechactisthatofgivinganorder,andassuchdoesnotinvolvetruth-value,butinadditionIdodescribemyselfasgivingtheorder,andsomysentenceistrueinthatdegeneratesense.Onthishypothesis,suchsentencesas(8)–(12),whichdifferfrom(16)inthattheirassociatedmainspeechactsaretruth-liable,wouldhavetwolocutionarycontentsandtruth-valueseach:aprimarycontentattachingtowhatwasstated,ordered,orwhatever(in(8),thatIhavenevertraveledtoaCommunistcountry),andaself-descriptivetruth-valuethatwouldnearlyalwaysbeautomatically“true”(thatIdosostate).Thistwo-truth-valuehypothesisisattractive,forinlightofexampleslike(15a–e),neitherthetemptingtruth-valuenortheliberaltruth-valueseemsexpendable.Andwecanmakethetwo-truth-valuehypothesismorepalatablebyarguingthatthetwotruth-valuesattachtoslightlydifferentsortsofthing.Noticethatinuttering(8)Imakeastatement.Whatstatement?ThestatementthatIhavenevertraveledtoaCommunistcountry.So,althoughImadethatstatementbyutteringasentencethat,takenliberally,doesnotentailitspropositionalcontent,Inonethelessmadeit.AndifthefactisthatIhavetraveledtoaCommunistcountry,mystatementisfalseeventhoughthesentenceIuttered,takenliberally,isatruesentence.Icouldbeconvictedofperjury,notforhavingutteredafalsesentence,butforhavingmadeafalsestatement.(15d)and(15e)wouldrequiresomeelaboration.Onefeelsthattheuttererof(15e),inparticular,hasmadetwoorthreeassertionsinadditiontothatwhichisexpressedbythecomplementclause.Yettheearlierexamplesinthelistareborderlinecases;wouldanuttererof(15a)assertthather/hisadmis-sionwasgivenfreely?Afulltheoryofspeechactswouldhavetosortoutsuchsubtletiesatlength.IllocutionarytheoriesofmeaningWilliamAlston(1963)triedseriouslytoworkAustin’sspeech-actpragmaticsintoatheoryoflocutionarymeaningitself,identifyingasentence’smeaningwiththesentence’s“illocutionaryact-potential,”therangeofillocutionaryactsthesentencecouldbeusedtoperform.Ifyouareabletouseasentenceineveryillocutionarywaythatitaffords,youknowitsmeaning,andthatisallthereistosentencemeaning.(Thiswouldcertainlyqualifyasa“use”theoryofmeaning,thoughsuperficiallyfarfromwhatWittgensteinhadinmind.)ButinfactAlston’sviewdidnothingtoilluminatelocutionarymeaning,sincepotential-speech-actdescriptionssuchas“assertthatgorillasareveg-etarians”alreadypresupposeanotionofpropositionalcontentandexploitthemeaningsoftheircomplementclauses.Also,asMaureenCoyleonceobservedtome,sentencesthatsharetheirlocutionarycontentscandifferviolentlyintheirillocutionaryact-potentials:“Motherwilleattheoyster”;“Willmothereattheoyster?”;“Mother,eattheoyster!”\n154PragmaticsandspeechactsBarker(2004)ineffectheadsofftheseobjections.HeavoidsthefirstinGriceanfashion,byunderstandingillocutionaryactsintermsofspeakers’intentionsandbeliefs;forexample(oversimplifyingviciouslyofcourse),toassertthatPistoutterasentencewiththeintentionofgettingone’shearertobelievethatP.Theacts’descriptionsinherittherelevantpropositionalcontents,notfromthemeaningsoftheircomplementclauses,butfromthecontentsofthementalattitudeswhichpartlyconstitutethoseacts.Barkeravoidsthesecondobjectionpartlyinthesameway(tracingthecommon-alitiestosamenessofsomeunderlyingpropositionalattitudes),andpartlybyarguingonsophisticatedsyntacticgroundsthat“locutionarycontent”cannotbeseparatedfromillocutionaryforceinthefirstplace.Noticethatthelatterpointisalsoafurtherobjectiontothestandardtruth-conditionaltheoryofmeaning,becausethattheoryassumesthelocu-tionarycontentisdeterminedindependentlyofforce.Summary•Austincalledourattentionto“performative”utterancesandspeechactsmoregenerally.•Eachtypeofspeechactisgovernedbyrulesoftwosorts:constitutiveandregulative.•Violationofaregulativerulerendersaspeechactdefectiveorinfelici-tous.Therearemanyandvariedwaysinwhichagivenspeechactcanbeinfelicitous.•Thereisnoprincipleddistinctionbetweenperformativeutterancesandthoseofordinarydeclaratives;rather,everyutterancehasanillocutionaryforce,andvirtuallyeveryutterancehasapropositionalcontentaswell.•Inaddition,manyutteranceshaveperlocutionaryfeatures.•Cohen’sproblemaboutthetruthconditionsofsentencesthatcontainexplicitperformativeprefaceshasnotbeensolved.•Theillocutionarynotionofaspeechactaffordsanewtypeof“use”theoryofmeaning.Questions1Areallspeechactslike“Idouble”inthattheyareconstitutedentirelybyconvention?(SeeStrawson(1964).)2Caneveryspeechactbeassignedalocutionarycontent?Discussourapparentcounterexamplesandargueproorcon.3Chooseaparticulartypeofspeechactandtrytoenumerateitsconsti-tutiverulesanditscharacteristicregulativeconditions.(Searle(1969)doesthisfortheactofpromising.)\nSpeechactsandillocutionaryforce1554SpotsomedifficultiesforAustin’sdistinctionbetweenlocutionary,illocutionary,andperlocutionaryfeatures.Findsometroublesomeborderlinecases.5Lewis(1970)defendstheanti-Austinianideathat,whenonetokens(even)a“pure”performative,oneatthesametimestatesthatoneisperformingtheactinquestion—oratleastthesentenceoneuttersistrueifandonlyifoneisperformingthatact.Examinethisview.6GomoredeeplyintoCohen’sproblem.7Ifyouhavereadit,pursueAlston’sorBarker’stheoryofmeaning.Furtherreading•Theacknowledgedclassiconspeech-acttheoryfollowingAustinisSearle(1969).ButconsiderablybetterisSearle(1979a),acollectionofessays.SeealsoTravis(1975)andHoldcroft(1978).•Twoexcellentworks(inadditiontoSchiffer(1972))thatconnectspeech-acttheorytootherissuesinpragmaticsandtocurrentresearchinlinguisticsandpsychologyareBachandHarnish(1979)andGazdar(1979).SeealsoColeandMorgan(1975),Levinson(1983),Green(1989),andSadock(2004).•Ginet(1979)isanexcellentpaper,andilluminatesCohen’sproblem.Waysoutoftheproblem(noneofthementirelysatisfactory)havebeenofferedbyCresswell(1973),BachandHarnish(1979),andLycan(1984:ch.6).•Alston(2000)furtherdevelopshisillocutionarytheoryofmeaning.\n13ImplicativerelationsOverviewSentencesentailothersentences,andinthatstrongsenseimplythem.Butthereareseveralwaysinwhichsentencesorutterancesalsolinguisticallyimplythingstheydonotstrictlyentail.First,veryoftenaspeakerusesasentencetoconveysomethingotherthanwhatthatsentenceliterallymeans,asforexampleinsarcasmorinbroadhinting.AccordingtoGrice’stheoryof“conversationalimplicature,”suchimplicationsaregeneratedbyasetofprinciplesthatgoverncooperativeconversation.Hearerspickuptheimplica-tionseitherbyassuming(contrarytoappearances)thatspeakersarebeingcooperativeanddrawinginferencesfromthatassumption,orbynotingthatspeakersarebeingdeliberatelyuncooperativeanddrawinginferencesfromthatassumption.However,itisnotclearhowwearesupposedtodothisasrapidlyandasaccuratelyaswedo.Second,Strawson’scriticismofRussell’sTheoryofDescriptionssuggestsanotionof“presupposition”distinctfromentailmentinthat,whenasen-tence’spresuppositionfails,thesentenceisnotfalsebutlackstruth-valueentirely.Butitishardtofindclearexamplesofthisrelation.Third,someimplicationsarecarriedbythechoiceofaspecialword,suchas“but”asopposedto“and,”inthat“but”meansjustthesameas“and”exceptforcarryingacontrastiveconnotation.Gricecallsthisphenomenon“conventionalimplicature.”RelevancetheoristshavechallengedGrice’smodelandofferedcompetingexplanationsofsomeimplicaturephenomena.Fourth,therearesomesentencesthatwouldstandardlybeusedtoper-formspeechactsotherthantheactsindicatedbytheirgrammaticalmoodsandsemanticcontents.Toexplainthisanomalyof“indirectforce,”SearletriesanextensionofGrice’stheoryofconversationalimplicature.Butthatstrategyfallsshortofaccountingforallthedata,andthereisnoverysatisfac-toryalternative.Davidsontalksofasemantics’capturingthe“feltimplications”oftargetsentences,bywhichhemeansthesentences’entailmentrelations.ButGrice(1975)hastaughtusthatimplicationcomesindifferentkinds.Thereareanumberofphenomenathatfallnaturallyunderthelabelof“implication”butarenot,ornotobviously,casesofentailment.InthischapterIshallsurveyfourofthem.\nImplicativerelations157ConveyedmeaningsandinvitedinferencesFirst,therearewhatwemightcall“conveyedmeanings”ofutterances.Itisnatural(thoughnotobligatory)todescribethisphenomenonintermsofspeaker-meaning:Inmanycases—thisisquiteprevalentinordinaryconversation—aspeakeruttersasentencethatmeansthatPbutitisobvioustoallthatthespeaker’smaincommunicativeintentistoconveysomethingdifferent,thatQ.Forexample,Isaytoanobstreperousvisitor,“There’sthedoor,”meaningthatthevisitoristoleavenow.Butthesentence“There’sthedoor”doesnotmean“Youaretoleavenow,”norcouldIbedescribedashavingcomeoutandsaidthatthevisitoristoleave.Isayonething,Imeananother;andthisisperfectlycleartobothpartieswithouteitherofthemhavingtothinkaboutitforamoment.Inchapter7,ofcourse,wehavediscussedmismatchesbetweenspeaker-meaningandsentencemeaning.Buttherewetendedtofocusonpathologicalcasesinwhich,forexample,aspeakerhasabizarrebeliefaboutthemeaningofthewordoraboutsomeoneelse’sunderstandingoftheword(orareason-ablebeliefaboutsomebodyelse’sbizarreunderstandingoftheword).ButinthecaseofwhatIamcallingconveyedmeaning,thereisnopathology;itisaperfectlynormalconversationalphenomenon.SupposeyouaskmewhetherSmedleyisagoodphilosopher,andIsay:(1a)Smedleysummarizestextsprettyaccuratelyandhasverynicehandwriting,or,lesssubtly:(1b)Smedleyisverygoodatpingpong.Clearly,whatIamconveyingtoyouisnotwhatmysentenceliterallymeans.Whatmysentencemeansmayormaynotbetrue,butthatisimmaterial.WhatIconveyissomethingdifferent,thatSmedleyisverybadoratleastnotverygoodatphilosophy.Myhearershouldgraspthatimmediately;andindeed,competenthearersdograspsuchconveyedmeaningswithouteverrealizingthatthatiswhattheyaredoing.Here,then,wehavecomeuponanotherlinguisticphenomenonthat(likeillocutionaryforce)ispartofwhatanyonewouldhavetounderstandinordertobecountedasafullycompetentspeakerofthelanguage.Ifyouwereaforeignerwell-versedinEnglishoratleasthadlearnedthelexicalmeaningsofthewordsandenoughgrammartounderstandtheliteralmeaningsofsen-tences,butyoutookutterancesliketheforegoingexamplesliterally,therewouldstillbesomethingimportantthatyouweremissing.Anotherkindof“implication”thathasexercisedlinguistsiswhatGeisandZwicky(1971)originallycalled“invitedinference.”Oneexampleistheperfectionofconditionalsintobiconditionals:SupposeIsay,\n158Pragmaticsandspeechacts(2)Ifyoumowmylawn,I’llgiveyoutendollars.Takenliterally,(2)isonlyaone-wayconditional;withoutlogicalimproprietyIcouldhaveadded,“Cometothinkofit,ifyoudon’tmowmylawnI’llgiveyoutendollarsanyway.”Butuponhearing(2)alone,youwouldimmediatelyfillinthatifyoudonotmowmylawnthenIwillnotgiveyouthetendollars.Youhearthemere“if”asan“ifandonlyif.”Anotherexamplewouldbetheelevationofwhatareonlyconjunctionsintocausalclaims.Thus:(3)MarthawatchedtheEducationSchoolburningandsmiledwithpleasure.Anyonewouldhear(3)asimplyingthatMarthawascausedpleasurebysee-ingtheEducationSchoolburning;somepeoplewouldhear(3)asactuallysayingthat.But(3)doesnotsaythat.(3)saysonlythatonethinghappenedandanotherdid(compare“MarthawatchedtheEducationSchoolburningandscratchedhernose”).Likewise,“and”isoftenheardascarryingatem-poralimplication.Thereisadifferencethatmostpeoplewouldhearinthemeaningsof(4a)JohnandMarshafellinloveandtheygotmarriedand(4b)JohnandMarshagotmarriedandtheyfellinlove.Eventhough(4a)doesnotentailthatJohnandMarshafellinloveandgotmarriedinthatorder,thetemporalinferenceisinvited.ConversationalimplicatureGrice(1975)addressedhimselftophenomenaoftheforegoingkinds.Seeingspeaker-meaningascommunicationofthecontentsofone’smentalstates,hebeganthinkingaboutmechanismsofconversationandthesocialnormsthatgoverncooperativeconversation.Hewentontodevelopthetheoryofwhathecalledconversationalimplicature.AccordingtoGrice,theconversationalnorm-in-chiefistheCooperativePrinciple:(CP)Makeyourconversationalcontributionsuchasisrequired,atthestageatwhichitoccurs,bytheacceptedpurposeordirectionofthetalk-exchangeinwhichyouareengaged.\nImplicativerelations159(CP)maysoundvacuous,butitsummarizesasetofcorollariesthatareany-thingbut.Gricecallsthecorollaries“conversationalmaxims.”Hereareafew(thenumberingismine).(M1)Makeyourcontributiontoaconversationasinformativeasisrequired(forthecurrentpurposesoftheexchange).[Callthisthe“MaximofStrength.”](M2)Donotmakeyourcontributionmoreinformativethanisrequired.(M3)Donotsaywhatyoubelievetobefalse.[TheMaximofTruthfulness.](M4)Donotsaythatforwhichyoulackadequateevidence.[TheMaximofEvidence.](M5)Berelevant.[TheMaximofRelevance.](M6)Avoidambiguity.(M7)Bebrief(avoidunnecessaryprolixity).Themaxims’functionistoexpeditethegivingandreceivingofinformation,infairlyobviousways.Themaximsaffordexplanationsofhowaspeakercansayonethingandbecorrectlyperceivedasmeaninganother.Griceoffersatemplateforsuchexplanations,intheformofastandardpatternofreasoningmeanttobeengagedinbythehearer:He[thespeaker]hassaidthatp;thereisnoreasontosupposethatheisnotobservingthemaxims,oratleast...[CP];hecouldnotbedoingthisunlesshethoughtthatq;heknows(andknowsthatIknowthatheknows)thatIcanseethatthesuppositionthathethinksthatqisrequired;hehasdonenothingtostopmethinkingthatq;thereforeheintendsmetothink,orisatleastwillingtoallowmetothink,thatq;andsohehasimplicatedthatq.(p.50)(“Implicate”isGrice’stechnicaltermforthisindirectmeansofcommunication.)WhenIutter“There’sthedoor,”IintendmyvisitortoreasoninGrice’sway.Roughly:“Thedoor?ThedoorhasnothingtodowithanythingIcur-rentlyhaveinmind.So,bytheMaximofRelevance(M5),thedoormustberelevanttosomethinghehasinmind.Andheknows(andknows...)that\n160PragmaticsandspeechactsImusthaveworkedthatout.Sohehasdeliberatelyshownmethathewantsmetoknowwherethedooris.Whymightthatbe?Egad,hemustwantmetogooutthedoor.”Ofcourse,allthisreasoningtakesplacesubconsciouslyandatveryhighspeed.(Gricehimselfdidnotofferhissystemasaliteralpsychologicalmodel,butweneedoneandatthispointwehavenoother.)Contextualinformationmayhelpwiththereasoningtoo.Inthiscase,thevisitormayrealizethathehasbeenbeingfairlyoffensive,thatIhavenotofferedhimadrink,thatIamnotsmiling,andthatitis6:45p.m.Inanyconversation,backgroundinformationismutuallyassumed,andvastfundsofassumptionsareassumedbybothpartiestobesharedbybothparties;Stalnaker(1978)callssuchmaterial“commonground.”WhenIutter(1a)or(1b),IimplybywayoftheMaximofStrengthandtheMaximofEvidencethatIamnotinapositiontosayanythingstrongeraboutSmedley’sability.Butsince(wemaysuppose)thereasonIwasaskedisthatIamthepersonwhoisinanidealoratleastgoodpositiontoassessSmedley’sability,thisdeliberatelyinvitesmyhearertoconcludethatthereisnothinggoodtosayaboutit.1WhenIutter(2),Iintendmyhearertoreflectthat,ifIweregoingtogivehertendollarsinanycase,uttering(2)wouldviolateboththeMaximofRelevance(whymentionthelawninparticular?)andtheruleagainstprolix-ity.(Thereisalsothebackgroundinformationthatpeopledonotoftengoaroundgivingoutmoneywhennoservicehasbeenrenderedandnochari-tablepurposeisevident.)(3)and(4a)arealittlehardertoexplain.Whatpromptsustoinferfrom(3)thatMarthasmiledbecauseshesawtheEducationSchoolburningisprobablysomecombinationoftheMaximofRelevancewithourknowledgeoftheeffectsofburning,ofMartha’slikelyattitudetowardschoolsofeducation,andoftheconnectionbetweendesire-satisfactionandfacialmusculature.(4a)mayhavetodowithsomedeepnarratologicalassumption.Suchmat-ters,andtheotherwisedangerouslyvaguenotionof“relevance”generally,havebeeninvestigatedinsomedepthbySperberandWilson(1986),aworkthatdepartsfromGriceinsomewaysandhasspawnedanewapproachtoimplicativerelations;seebelow.Gricementionsthatonecanalsogenerateanimplicaturebyfloutingaconversationalmaxim,thatis,byviolatingitblatantly.MyfavoriteGriceanexample(paraphrasedfrompp.55–6):(5)MsXproducedaseriesofsoundsthatcorrespondedquitecloselytothescoreofHandel’s“IKnowThatMyRedeemerLiveth.”[Saidbyaconcertreviewer.]Whyhasthereviewerdribbledoutallthisprolixstuff,insteadofsayingsimplythatMsXsang“IKnowThatMyRedeemerLiveth”?“Presumably,toemphasizeastrikingdifferencebetween[Ms]X’sperformanceandthosetowhichtheword‘singing’isusuallyapplied.”Amorecommontypeof\nImplicativerelations161exampleiswhenthespeaker’ssentenceistooobviouslyfalse;Gricecitessarcasmthere.Gricesuggeststhathistheorywillaccountformetaphor,sincemetaphori-calutterancestypicallyflout(M3):Exampleslike“Youarethecreaminmycoffee”characteristicallyinvolvecategoricalfalsity,sothecontradictoryofwhatthespeakerhasmadeasiftosaywill,strictlyspeaking,beatruism;soitcannotbethatthatsuchaspeakeristryingtogetacross.Themostlikelysuppositionisthatthespeakerisattributingtohisaudiencesomefeatureorfeaturesinrespectofwhichtheaudienceresembles(moreorlessfancifully)thementionedsubstance.(p.53)Weshallassessthissuggestioninchapter14.Twofeaturesarecharacteristicofconversationalimplicature.First,animplicaturemustbesomethingoneworksout,orcouldworkout,usingreasoningofthekindillustratedabove.Ifnosuchreasoningisavailable,thentheimplicationmustbeofsomeotherkind.Second,animplicatureiscancel-able,inthesensethataspeakerwhowantedtocouldforestalltheinferencethatwouldotherwisebereasonable:“Smedleyisverygoodatpingpong.Butdon’tgetmewrong—he’saterrificphilosophertoo.Imentionedthepingpongfirstbecausewe’vejustbeenplayingandI’mexhausted.”2,3InthepreviouschapterInotedAustin’scomplaintthat,whenaphiloso-pherperceivesinfelicityinanutterance,thephilosophertendstooquicklytorejectthesentenceutteredasfalse.Duringthe1950sand1960s,thiswasevenafashionablestyleofargument:from“Thissentencewouldsoundfunnyifuttered”to“Thissentenceisfalse/incoherent/meaningless.”Griceisinpartconcernedtoeradicatethatargumentform.Andwearenowinapositiontoappreciateanexampleofit(aslightlyconvolutedexample,becauseitisitselfaboutfalsity).Recall,fromchapter2,Strawson’sfirstobjectiontoRussell’sTheoryofDescriptions.Hesaysthatnoonewouldeverrespondtoanutter-anceof“ThepresentKingofFranceisbald”bysaying,“That’sfalse.”Andheisrightaboutthat.Butfromitheinfersthatthesentenceutteredwasnotfalse;thatis,that“That’sfalse”woulditselfbefalse.Andthatdoesnotfollow.Theobviousreasonwewouldnotsay“That’sfalse”isthattodosowouldbemisleadingbywayoftheMaximofStrength:Youareinapositiontosaysomethingmuchstrongerandmoreinformativeandabettercontribu-tiontotheconversation,namely,“Holdon;thereisnokingofFrance.”So,evenifStrawson’sowncompetingthesis(thatstatementswithnonreferringsingulartermsinthemshouldbetakenasnonstatementsratherthanasfalsestatements)isactuallycorrect,hisargumentdoesnotshowthat.Thebasicideaofconversationalimplicatureisalmostuniversallyaccepted,asaremostofitsstandardusesinphilosophy.ButGrice’stheoryofitisnownotsowidelygranted.Therearetwomaindirectcomplaints.First,some\n162Pragmaticsandspeechactsphilosophersaresuspiciousoftheamountofcomplexbutnearlyinstanta-neousandalmostentirelyunconsciousreasoningpositedbyGrice’stheory.(ReadthroughGrice’stemplateagain,andseehowlongittakesyou.)4Butthen,inmanywalksoflifewedoagreatdealofreasoningveryquicklyandsubconsciously.ThesecondandmoreseriouscomplaintisduetoHarnish(1976),SperberandWilson(1986)andespeciallyDavis(1998):MostcasesofGriceanreason-ingdivideintotwostages,aninitialnegativestageandasubsequentpositivestage.Inthenegativestage,thehearerdetectsthatthespeaker’smeaningdivergesfromsentencemeaning.Inthepositivestage,thehearercomestoaconclusionregardingwhatthespeakerdoesmeaninstead.AppealstotheMaximofRelevancecertainlyworkthatway.SodoesanyGriceanreasoningthatstartswith“[Thespeaker]couldn’tmeanthat[becauseitistooobvi-ouslyfalseandweallknowthat].”Weknowthatsomethingisup;butthenthereisthepositivepartoffiguringoutjustwhatitisthatisup.DavisobjectsthatGricegivesusverylittlehelpwiththepositivepart.Take(3).Isuggestedsomebackgroundknowledgeofcausalrelationsthatwouldhelpahearercalculatetheimplicature.Butwhyisitobviousthattheneededrelevanceiscausalrelevanceinthefirstplace?Itdoesseemthatcausalrelevanceistheobviouscandidate,butnothinginGrice’stheorypredictsorevenhintsatthat.OrconsiderGrice’ssuggestionaboutmetaphor.Itisindeedobviousthatthespeakermeanssomethingotherthanwhat“Youarethecreaminmycoffee”(literally)means,butwhatsignalsthat“thespeakerisattributingtohisaudiencesomefeatureorfeaturesinrespectofwhichtheaudienceresembles[creamincoffee]”?Whyisthat“themostlikelysupposition”?DavispointsoutthatphilosophersoflanguagehavemissedthisimportantlacunainGrice’stheorybecause,wheneverwelookatanexample,wealreadyknowwhatwouldnormallybeimplicatedbyanutteranceofthesentenceinquestion,andsowearenotmovedtoaskourselveshowthepositivecalcula-tionisworkedout.Thecureforthisistopretendthatwedonotalreadyknow,andjustlookattheutteranceincontextandtrytohituponcluesthatwouldshowanentirelyuninitiatedhearerwhatthespeakermeanttoconvey.Itisnoteasy.Practitionersofthe“relevance”literature(seebelow)havediscoveredwhattheyargueisanewkindofimplication,called“explicature,”intermediatebetweenconversationalimplicatureandentailment,inthattheexplicatumiscancelablebut,ifleftuncanceled,iscountedassaidratherthanmerelyimplied—seeCarston(1988)andRecanati(1989).Anallegedexamplewouldbe:(6)Sheputdowntheletter,shedasingletear,andwalkedslowlybutsteadilytothecliff’sedge;thenshejumped.Thatsentencedoesnotstrictlyentailthatitssubjectjumpedoffthecliff,becauseonecouldcanceltheimplicationwithoutcontradiction,forexample\nImplicativerelations163adding“—notoffthecliff,mindyou,justupanddownneartheedge.”But,CarstonandRecanatimaintain,ifthespeakerdoesnotcancelthefatalimpli-cationwithinaconversationallyreasonabletime,thespeakerwillbecountedashavingsaid,andnotmerelyimplicated,thatthesubjectjumpedoffthecliff.Buttherearetenableargumentsonbothsidesofthisissue.PresuppositionandconventionalimplicatureAsecondspeciesofimplicationsthatarenotentailmentswassuggestedbyStrawson’spositionondefinitedescriptions.Recallthat,inresponsetoRussell,Strawson(1950)saidthat“ThepresentKingofFranceisbald”doesnotentailtheexistenceofapresentking,butmerelypresupposesit.Themarkofthis,accordingtoStrawson,wasthat,whenthereisnoking,“ThepresentKingofFranceisbald”isnotfalse,butlackstruth-valueentirely.Thesamegoesfor“ThepresentKingofFranceisnotbald.”AfewphilosophersandmanymorelinguiststookupStrawson’sidea,andhavemadeitalittlemoreformal:WhenasentenceS1entailsasentenceS2,andS2isfalse,thennecessarilyS1isfalseandS1’snegationistrue.ButwhenS1presupposesS2andS2isfalse,thenS1doesnotgofalse,butlackstruth-value,andsodoesS1’snegation.5Noticethatpresuppositioninthissense(calledsemanticpresupposition)islikeentailmentandunlikeconversationalimplicatureinbeingnoncancelable.BothS1andS1’snegationnecessitateS2intheabsolutewaycharacteristicofentailment.Thereareactuallynouncontroversialexamplesofsemanticpresupposi-tion.Buthereareafewcandidatesentencepairs.(7a)ItwasGranniewhorobbedthediaperservice.(7b)Someonerobbedthediaperservice.(8a)Haveyoustoppedbeatingyourspouse?(8b)Youhavebeatenyourspouse.(9a)Rockyrealizedthathisflywasopen.(9b)Rocky’sflywasopen.(10a)Fred,whowasfat,couldnotrun.(10b)Fredwasfat.(11a)Shewaspoorbutshewashonest.(11b)Beingpoorinhibits[orsomehowcontrastswith]beinghonest.\n164PragmaticsandspeechactsIneachcase,ithasbeenclaimed,theasentencenecessitatesthebsentence,andsodoestheasentence’snegation;ifthebsentenceisfalse,theasentenceisnotfalsebutgoestozip.Andindeed,ineachcasetheasentence’snegationdoesintuitivelyseemtocarrythesameimplicationasdoestheasentenceitself.Butinsomecases,althoughtheasentencedoesnecessitatethebsentence,theasentence’snegationdoesnot.Thisistrueof(7).For(7′)ItwasnotGranniewhorobbedthediaperservice;noonerobbedit,thoughperhapspeculiar,isnotself-contradictory.If(7′)isnotself-contra-dictory,thenthenegationof(7a)doesnotnecessitate(7b)inthestrongsenserequiredforsemanticpresupposition.(7a)’snegationdoesconversationallyimplicate(7b),bywayoftheMaximofStrength;someonewhoutters“ItwasnotGranniewhorobbedthediaperservice”isinapositiontomakeastrongerandmoreusefullyinformativedenial,namelythatof(7b)itself.Butconversationalimplicatureiscancelable,asnecessitationisnot.Nonecessita-tion,nosemanticpresupposition.(8),thoughinterrogative,meetsasimilarfate.Ifyouaremarriedandyouareasked(8)(andyouhaveneverbeatenyourspouse),hereisthecorrectanswer:“No.”6Becauseonecanstopdoingathingonlyifonehasatsometimebeendoingit.(Ofcoursetheanswer“No”ismisleadingbecause,viatheMaximofStrength,itimplicatesthatonehasbeatenone’sspouseandcontinuestodoso.Thecorrectandnonmisleadinganswerwouldbe,“No,becauseIneverhavebeatenher/himinthefirstplace.”)(9)ispossibletodismissinthisway,butharder.(9′)Rockydidnotrealizethathisflywasopen;hecouldhardlyhaverealizedthat,becausehisflywasn’topendoesnotseemcontradictoryeither;butthereisnotsoobviousaGriceanexplanationof(9a)’snegationimplicating(9b).(10)isperhapsthebestallegedexampleofsemanticpresuppositiononourlist.(10′)It’sfalsethatFred,whowasfat,couldnotrun,becauseFredwasn’tfatdoessoundcontradictoryoratleastsemanticallyanomalous.Yettomyear,ifFredwasnotfat,(10)doesnotgotruth-valueless.Ihear(10)asstrictlyspeakingfalse,becausethespeakerhas(thoughunemphatically)calledFredfat.Butthatdoesnotexplainwhatiswrongwith(10′).Ourremainingcase,(11),ismoredistinctive,andIshallpostponeitbriefly.\nImplicativerelations165IftheDirectReferenceTheoryofpropernamesiscorrect,thenperhapsStrawsonisrightandsentencescontainingnonreferringnameslacktruth-value.Ofcourse,thatiswhatgaverisetotheproblemsofApparentReferencetoNonexistentsandNegativeExistentials.But,ifso,thensentencescontain-ingpropernamesatalldosemanticallypresupposetheexistenceofreferentsforthosenames.Somelinguistshavedistinguishedaloosernotionof“pragmaticpresup-position”fromthatofsemanticpresupposition.Butthattermhasnotbeenclearlydefined,andnoonetypeofpragmaticimplicationhasbeenmeantbythistotheexclusionofothertypes.Toreturnto(11):(11′)It’sfalsethatshewaspoorbutshewashonest;beingpoordoesnotinhibitbeinghonestisnotcontradictory,butisawkward.And(11a)’simplicationof(11b)isnon-cancelable.Supposethepersonspokenofwasbothpoorandhonest.Then,intuitively,thespeakerhassaidtwotruethingsabouther.Butifbeingpoordoesnotatallinhibit(orsomehowcontrastwith)beinghonest,somethingisstillverywrongwith(11a)(compare“Wiltis7'2"buthe’stall”).Thespeakerhaschosenthewrongword:“but”islike“and,”exceptthat“but”carriesaspecialconnotationthat“and”doesnot;verylikelythatis“but”’sraisond’être.Grice(1975)hadathirdclassificationfor(11),neitherconversationalimplicaturenorsemanticpresupposition.Hecalleditconventionalimplica-ture.Conventionalimplicatureisimplicature,inthataspeakerimplicatessomethingratherthanactuallysayingit,butitdiffersfromconversationalimplicatureintwoways.First,conventionalimplicaturesarenotworkedout;theyaregraspedimmediately,notonthebasisofreasoning.Second,theyarenotcancelable(Icannotsay“Georgeisalinguistbuthe’ssmart—don’tgetthewrongidea,though;Ididn’tmeananythingaboutlinguistsnotbeingsmart”).Conventionalimplicaturesarenormallycarriedbytendentiouschoicesofparticularwords,asof“but”over“and.”Grice’soriginalexamplewas“HeisanEnglishman;heisthereforebrave.”WhileIhavesaidthatheisanEnglishman,andsaidthatheisbrave,IdonotwanttosaythatIhavesaid...thatitfollowsfromhisbeinganEnglishmanthatheisbrave,thoughIhavecertainlyindicated,andsoimplicated,thatthisisso.Idonotwanttosaythatmyutteranceofthissentencewouldbe,strictlyspeaking,falseshouldtheconsequenceinquestionfailtohold.(pp.44–5)7\n166PragmaticsandspeechactsFurtherexamplesinvolvethewords“too”and“either”:“Jonnieisalinguistandherhusbandisverysmarttoo”;“Paulwasaphilosopherandhiswifewasn’tverysmarteither.”Here,asinmanycases,agoodwaytoinvestigatethenatureofthesedif-ferentkindsofimplicationsistoaskaboutthepenaltyorsanctionthatensueswhenanimplicatumisfalse.WhenS1entailsS2andS2isfalse,thepenaltyisthatS1isfalse.WhenS1semanticallypresupposesS2andS2isfalse,thenS1issentignominiouslytozip.WhensomeoneuttersS1,therebyconversationallyimplicatingS2,andtheconveyedmeaningorinvitedinferenceS2isfalse,thenthepenaltyisthat,evenifS1istrue,thespeaker’sutteranceismisleading.IfS1conventionallyimplicatesS2andS2isfalse,thenS1ismisworded,evenifnotfalse.Afurthertypeof“pragmaticpresupposition”notalreadymentionedheremightbecalled“illocutionaryimplication”:Theperformingofaspeechactinsomesenseimpliesthesatisfactionofitsdistinctivefelicityconditions.Forexample,mypromisingtoreturnyourchampagneglassesimpliesthatIintendtoreturnthem,anddoessononcancelably(Icannotadd,“...butdon’tgetmewrong;Ihavenointentionofreturningthem”).Toourpenaltycatalogue,wemightaddthatifsomeoneuttersS1,therebyperformingaspeechacthavingS2asadistinctivefelicitycondition,andS2isfalse,thenthespeechactisinfelicitousinadistinctivelyillocutionaryway.RelevanceTheoryRelevancetheorists(originallySperberandWilson1986;seeespeciallyalsoCarston2002)beganbydevelopingGrice’smodel,buttheirprogramisnowavigorouscompetitor.Theyrejecttheideathatthereareparticularconver-sationalmaximsofGrice’ssort.Rather,theymaintain,implicaturesaretheproductofall-purposecognitiveprocessingthataimsatefficiencyofinfor-mationtransfermoregenerally.“UtterancesraiseexpectationsofrelevancenotbecausespeakersareexpectedtoobeyaCo-operativePrincipleandmaximsorsomeothercommunicativeconvention,butbecausethesearchforrelevanceisabasicfeatureofhumancognition,whichcommunicatorsmayexploit”(WilsonandSperber2004:608).“Relevance”isunderstoodasafavorablebalanceof“positivecognitiveeffects”overprocessingtimeandeffort;apositivecognitiveeffectis(roughly)animprovementinthehearer’srepresentationoftheworld,suchasacquisitionofatrueandusefulbelief.Ingeneral,wearealwaysseekingtoimproveourcognitiveconditioninacost-andtime-efficientway.Inthematterofcommunicationinparticular,speakersandhearersatleasttacitlyknowthatweareallthusdrivenbyrelevance.So,whenaspeakerproducesanutterance,s/heintendshearerstoassumethatitiscognitivelybeneficialenoughtobeworthprocessing.Andthereforeeveryutterancecarries“apresumptionofitsownoptimalrelevance”—relevantenoughtobeworththehearer’seffort,and“themostrelevantonecompatiblewith\nImplicativerelations167communicator’sabilitiesandpreferences”(p.612).Thisistherelevancetheo-rists’rivaltoGrice’sCooperativePrinciple(CP).Andtheyarguethatthetwotheoriesmakedifferentpredictions.Forexample,weconstantlysaythingsthatarenotstrictlytrueandareknownbybothspeakerandhearertobenottrue,butwithoutbeingperceivedasfloutingtheMaximofTruthfulness.Wespeakloosely;forexample,usingthenoun“record”toincludemusicaltapesandcompactdiscs;andexaggerationisubiquitous.8BututterancesofthosekindsdonotnormallyinitiateGriceanreasoning,asin“Hmmm,shehasfloutedtheMaximofTruthfulness.Isitsarcasm?Ormaybeshe’sindicat-ingthatshe’snotallowedtotalkaboutthis....”Wedonoteventacitlynoticesuchviolations.WilsonandSperbersuggestthathearers’expectationsoftruthfulnessareonlyarough-and-readybyproductoftheirmorebasicexpectationsofrelevance(p.619).Further:AccordingtoGrice’spicture,ahearerrecoverstheutterance’sfullliterallocutionaryorpropositionalcontent(“Hehassaidthatp,”),andthenproceedstoreasonfromthatcombinedwiththeCooperativePrinciple.Butrelevancetheoristspointoutthat,psychologicallyatleast,thatcannotberight.Thesentenceutteredneveritselfexpressesor“encodes”acompleteproposition.Aswehaveseen,itmustalwaysbedisambiguated.Anditsdeicticelementsmustbeassignedreferents.Thesedecisions,includingthecomputationofHarman’sfunctionα,arealsoguidedbythegeneralcogni-tivedriveforrelevance.Sothereissimultaneousdisambiguation,treatmentofdeicticelements,andworkingoutofimplicatures,withconstanttradeoffsbetweenthem.Tothelatterthreeprocesses,relevancetheoristshaveaddedfurtheronesthatmakefor“explicatures”asdescribedabove.ThereiswhatRecanati(1993,2001)calls“saturation,”theplugginginofanappropriatevaluetoanunderlyingpositioninlogicalform:“Thiswayisshorter”(thanwhatway?);“Iamtooold”(forwhat?);“Areyoustrongenoughinthelegs?”(todowhat?).Thereisalso“freeenrichment”(Recanati1993;Carston2002),whenthereseemstobean“unarticulated”(notjustunvoicedorunpronounced)constituentofwhatissaideventhoughthereisnoevencovertholeinlogicalformorotherspecificallylinguisticcontrol:“I’veeaten”(themostrecentmeal);“It’sraining”(insuch-and-suchlocation);“Ourbuildingissomedistanceaway”(farenoughthatitwouldtakecontextuallyconsiderabletimetogetthere).Sentence(6)above(“Sheputdowntheletter,shedasingletear,andwalkedslowlybutsteadilytothecliff’sedge;thenshejumped”)mightbethoughttobeacaseofsaturation,ifitisarguedthat“jumped”logicallyexpressesarelationbetweenanagentandalocation-plus-direction.Morelikely,itisacaseoffreeenrichment.Congenitallyliteral-mindedphilosopherswillbeskepticalof“freeenrich-ment,”andwanttodenythatspeakersoftheforegoingsentencesactuallysaidthattheyhadeatenlunchinparticular,orthatitisraininghereinChapelHill,orthatitwouldtakeagooddealoftimetogettothebuilding.Suchphilosopherswouldhearthose“enrichments”asordinaryimplicatures.9Butlinguistsaremorelikelytoberightonthissortofissue.\n168PragmaticsandspeechactsIndirectforceAsImentionedinthepreviouschapter,thethreeprincipalEnglishmoodscorrespondtothreebroadillocutionarygeneraofwhichindividualtypesofspeechactarespecies:Thenormaluseofadeclarativeistomakeastatement,thatofaninterrogativeistoseekinformation,andthatofanimperativeistoissueadirectiveofsomekind.Butthecorrespondenceisfarfromperfect:(12)IwantyoutogototheBroccoliFestivalwithme.(13)Canyoupassthesalt?(14)BelievemewhenIsayI’llneveragainmixGlenfiddichandparegoric.(15)TellmehowyousavedKateWinslettfromthegianttreefrogthatatePittsburgh.(16)Iwantyoutotellmewhathasbecomeofmychildrenthree.(12)isgrammaticallydeclarativebutwouldnormallybeusedtoissuearequestorevenanorder.(13)isinterrogativebutwouldnormallybeheard,notliterallyasaninformation-seekingquestionaboutthehearer’sabilities,butasarequestalso.Theimperatives(14)and(15)wouldnormallybeusedrespectivelytomakeastatementandtoaskaquestion.(16),thoughdeclara-tive,wouldalsobeusedtoaskaquestion.Grammaticalmoodinitselfishardlysacred.Therealproblemisdeeper:Eachof(12)–(16)alsohasamoreliteralreadingthatdoescorrespondtoitsgrammaticalmood.Forexample,(12)couldbeutteredasapurelyfactualanswerto“Pleasetellmewhatdesireisuppermostinyourmindrightnow,sothatIcanbeginyourpsychoanalysis.”Thesemoreliteralreadingsexist,butareuncommonandhardtohear.Whatneedsexplainingiswhy(andhow)thenonliteralusesarethenormaluses.Thissortofillocutionarydisplacementiscalled“indirectforce.”Searle(1975)advocateswhatIshallcalltheconservativeapproachtoindirectforce.Namely,hemaintainsthatanutterance’sindirectforcecanbepredictedusingjustgeneralprinciplesofspeech-acttheorythatwealreadyknowtogetherwithGriceanmechanismsthatwealreadyknow.Nonewapparatusneedbeintroduced.Searlebeginsbyputtingforwardsomegeneralizationsaboutwaysofperformingparticularspeechactsindirectly.Forexample:“Scanmakeanindirectrequest(orotherdirective)byeitheraskingwhetherorstatingthatapreparatoryconditionconcerningH’sabilitytodoAobtains”;“Scanmakeanindirectdirectivebystatingthatthesincerityconditionobtains,butnotbyaskingwhetheritobtains.”Searleillustratesandexplainssuchgeneralizations\nImplicativerelations169byderivinginstancesofthemusingonlyprinciplesofspeech-acttheoryandGriceanconversationalreasoning.Take(13),literallyaquestionaboutthehearer’sabilitybutnormallyusedtomakearequest.AccordingtoSearle,thehearerfirstinfersinGriceanfashionthatthespeakerintendstoconveysomethingotherthan(13)’sliteralmeaning(thespeakerobviouslylackstheoreticalinterestinthehearer’sfinemusclecoordination,forexample).Thenthehearercleverlynotestwothings:thatthespeakerhasalludedtothesatisfactionofapreparatoryconditionforarequest,andthattherequestinquestionisonewhoseobediencethespeakerveryprobablywantsthehearertobringabout.Thatishowtheheareridenti-fiesthespeaker’sutteranceasarequesttopassthesalt.TheconservativeviewinheritsthetwomainobjectionsincurredbyGrice’stheoryofconversationalimplicature.Thereisskepticismregardingthepositedinstantaneoussubconsciousreasoning,andDavis’s(1998)com-plaintariseswithavengeance:Thenegativestage,thatofrecognizingthatthereissomeulteriorforce,iseasy,butthepositivestageismuchharder.Davis’sproblemisaggravatedforthecaseofindirectforce,becausethehearerfacesagreaterchallenge—notjusttoidentifyanimplicatedcontent,butinadditiontopickouttheunexpectedforce.(Hereagain,thedifficultyismaskedbythefactthat,whenwelookatexamples,wealreadyknowwhatindirectforcetheywouldhave.)Inthecaseof(13),whatcuesleadthehearertoidentifyitasarequest?Searleacknowledgestheproblem.Thehearerhastonoticethatthespeakerhasalludedtoapreparatoryconditionforrequesting.Butwhatcuewouldtipthehearerofftothat?Moreover,theabilityconditionisalsoafelicitycondi-tionforspeechactsofmanydifferentsorts;sohowwouldthehearersingleoutrequestinginparticular?Perhapss/hecouldnarrowitdowntodirectives.Withintheclassofdirectiveswecouldperhapsexcludeorders,commands,suggestions,andothersongroundsofpowerrelationsandtoneofvoice.Wealsohavecorroboratinginformationabouttheobedienceconditionsbeingonesinwhichthespeakermaywellhaveaninterest.Butateachsubstageofthisreasoning,thereisplentyofroomforerror.AtonepointSearlesuggeststhatthereisaconventionatworkinadditiontothepurelyGriceanreasoning.Thewords,“Canyou...?”dohaveasortofconventionalring.Butifso,itcannotbesuchabruteconventionthatitmakesthelocutioninquestionamounttoanidiomlike“kickthebucket”or“burythehatchet.”Requestsbeginningwith“Canyou...”are,ifyoulike,idiomatic,buttheyarenotidioms,fortheyadmitofliteralanswers.Thehearercouldsay,“Ican,yes,butareyousureyoushouldbeputtinganymoresaltonyourfood?”Thiscouldbeasmart-aleckanswer(“Doyouhavethetime?”—“Yes,Ihaveit”),butitneednotbe;perhapsthehearerknowsthespeakertohavehighbloodpressure.Oneisatleastabletoreplytothesentence’sliteral,compositionalcontent,evenifthatisnotentirelyfelicitousgiventhatthespeaker’sutterancewasanindirectrequest.If“Canyou...”hasconventionalforceofsomekind,thekindbadlyneedsexplaining.\n170PragmaticsandspeechactsMorgan(1978)makesanimportantattemptatexplainingtheonlyquasi-conventionalmechanisminvolvedhere.BachandHarnish(1979)argueforalessconvention-likedeviceof“standardization.”Boththinkofindirectforceas“short-circuited”implicature;thatis,asanimplicaturesocommonastohavebecomeinsomewayautomatic.Thereisafurtherproblemaboutindirectforce,calledtoourattentionbyGordonandLakoff(1975):Therearesyntacticmarksofindirectforce.Thatis,therearesurface-grammaticalfeaturesthatdemandindirectinterpreta-tionofthesentencesinwhichtheyoccur.(17)Whypaintyourhousepurple?(18)Whydon’tyoubenicetoyourbrotherforachange?/Benicetoyourbrotherforachange,whydon’tyou?(19)Wouldyougetmeaglassofwater?(20)Iwouldlikeadrymartini,please.(21)Here,Ineedthatwrench.Noneofthesesentencescanhavetheforceassociatedwithitsgrammaticalmood.Unlike“Whyareyoupaintingyourhousepurple?,”(17)cannotbeaninnocentquestion,butmustbeadiscouragement.(18)mustbeareproach;(19)and(20)mustberequests;unlikethesimpler“Ineedthatwrench,”(21)mustbearequestorsomethingstronger.Searle’sconservativeapproachisbalkedhere.Ithasnowayofpredictingthesedata.Worse,itcannotevenapplytothembecause,àlaGrice,itpositscalculation;indirectforceistobeworkedout.Butwith(17)–(21)thereisnothingtoworkout.Theyweartheirindirectforcesontheirsleeves.Othertheoriesofindirectforcehavetriedtocopewiththisproblem;nonehasbeenuncontroversiallysuccessful.Thedatathemselvesarenotentirelyuncontroversial:BachandHarnish(1979:ch.9)disputesomeofthem,mostexplicitlythat(19)isgrammatical.Summary•Oftenaspeakerusesasentencetoconveysomethingotherthanwhatthatsentenceliterallymeans.•AccordingtoGrice’stheoryofconversationalimplicature,suchimplicationsaregeneratedbyasetofprinciplesthatgoverncoopera-tiveconversation.ButDavishasofferedasignificantchallengetothisview.•Relevancetheoristsrejecttheideathatimplicaturesaregeneratedbyasetofconversationalmaxims.Theyholdinsteadthatimplicaturesare\nImplicativerelations171theproductofall-purposecognitiveprocessingthataimsatefficiencyofinformationtransfermoregenerally.•Strawson’scriticismofRussell’sTheoryofDescriptionssuggestsanotionof“presupposition”distinctfromentailment.Butitishardtofindclearexamplesofthisrelation.•Athirdtypeofimplication,conventionalimplicature,iscarriedbythechoiceofaspecialword.•Therearesentencesthatarenormallyusedwithindirectforce.Toexplainthis,SearletriesanextensionofGrice’stheoryofconversa-tionalimplicature.Butthatstrategyfallsshortofaccountingforallthedata,andthereisnoverysatisfactoryalternative.Questions1Thinkofsomemoreexamplesofconveyedmeaningsandinvitedinfer-ences,andtrytoexplainthemusingGrice’sprinciples.2Statesometraditionalphilosophicalissueandshowhowthenotionofconversationalimplicatureilluminatesit.3AdjudicateoneofourtwoobjectionstoGrice’stheoryofconversa-tionalimplicature,orcomeupwithafurtherobjectionofyourown.4Isthereanythingtothenotionof“presupposition”inanaturallan-guagethatcannotbeaccountedforintermsofimplicature?Lesstendentiously,istheremoretothenotionof“presupposition”innaturallanguagethanIhaveadmittedinthischapter?5Discussthenotionofconventionalimplicature,andtrytothinkofmoreexamples.IsGricerightincontendingthatitdiffersbothfromordinaryentailmentandfromconversationalimplicature?6AdjudicateadisputebetweenarelevancetheoristandGrice,orsimplydiscusssomeaspectoftheRelevanceTheoryliterature.7Saysomethinghelpfulaboutthepuzzleofindirectforce.Furtherreading•Grice(1978)isasequeldealingwithstressandirony.Grice’sposthu-mouscollection(1989)containsthatandotherimportantpapersonrelatedtopics.•Davis(1998)isacomprehensivecritiqueofGrice’stheoryofconver-sationalimplicature.Italsodiscussesindirectforce.•Seedsofthe“explicature”literatureareinCohen(1971).SeealsoBach(1994b).•Agoodexpositorypro-presuppositionpaperisKarttunen(1973).Twoverygoodbooksonthe“presupposition”literatureareKempson(1975)andWilson(1975);forascorched-earthcritique,seeLycan(1984:ch.4).\n172Pragmaticsandspeechacts•Sadock(1975)exploresthedistinctionbetweenconversationalandconventionalimplicature.Thecontainingvolume,ColeandMorgan(1975),isasplendidone,andcontainsseveralotherniceworksonimplicature;seealsoCole(1978).KarttunenandPeters(1979)isgoodonconventionalimplicature,asisWarner(1982).ButthebestgeneraltreatmentofconventionalimplicatureisLycan(1984:ch.5).•Bach(1999)iconoclasticallybuttenablydisputestheveryexistenceofconventionalimplicature.•TheRelevanceTheoryliteratureisnowverylarge.SeeBlakemore(1992)andCarston(2002).•TheclassicpapersonindirectforceareinColeandMorgan(1975);seeparticularlythepapersbyGordonandLakoff,GeorgiaGreen,andAliceDavison.•Morgan(1978)’stheoryofindirectforceisdevelopedfurtherinLycan(1984:ch.7).•BachandHarnish(1979)offeranimposingmastertheoryofcom-munication,incorporatingallthephenomenawehavesurveyedinthischapterandthelast.•Levinson(1983)isagoodgeneralbookonpragmatics.Davis(1991)isanexcellentanthology,andHornandWard(2004)isanoutstandingreferencework.\nPartIVThedarkside\n\n14MetaphorOverviewThephenomenonofmetaphorisfarmoreprevalentthanisgenerallyadmit-tedbyphilosophers,anditraisestwomainquestions:Whatis“metaphoricalmeaning”?Andhowdohearersgraspmetaphoricalmeaningasreadilyastheydo?Mosttheoristshavethoughtthatmetaphorissomehowamatterofbringingoutsimilaritiesbetweenthingsorstatesofaffairs.DonaldDavidsonarguesthatthis“bringingout”ispurelycausal,andinnowaylinguistic;hearingthemetaphorjustsomehowhastheeffectofmakingusseeasimilarity.TheNaiveSimileTheorygoestotheoppositeextreme,havingitthatmetaphorssimplyabbreviateexplicitliteralcomparisons.Bothviewsareeasilyseentobeinadequate.AccordingtotheFigurativeSimileTheory,ontheotherhand,metaphorsareshortforsimilesthemselvestakenfiguratively.ThisviewavoidsthethreemostobviousobjectionstotheNaiveSimileTheory,butnotallthetoughones.Searletreatsmetaphoricalmeaningasspeaker-meaningthatisalsocon-veyedmeaning,andinvokesGriceanapparatustoexplainitinmuchthewayheexplainedindirectforce.ThishassomeplausibilityandovercomesDavidson’sleadingobjectionstometaphoricalmeaning,butincursotherobjections.Afurthertheoryofmetaphorisbasedonthephenomenon,importantinitsownright,ofsinglewords’analogicaldifferentiationintohostsofdistinctthoughrelatedmeanings.AphilosophicalbiasPhilosopherslikelanguagetobeliteral.Thepreviousthirteenchaptershaveconcentratedontheoriesofliteralreferenceandliteralmeaning,andevenourdiscussionsofindirectforceandconversationalimplicaturehaveviewedthosephenomenaasmerelyspeaker-meaningsderivedbysomediscretemechanismsfromliteralsentencemeaning.Ihavebarelymentionedmeta-phorandotherfigurativeusesoflanguage.\n176ThedarksideThatbiasreflectsstandardphilosophicalpractice.Philosopherstendtothinkthatliteralspeechisthedefaultandmetaphoricalutterancesareoccasionalaberrations,mademainlybypoetsandpoetsmanqué.Butthebiasisonlyabias;sentencesareveryoftenusedinperfectlyordinarycon-textswithotherthantheirliteralmeanings.Indeed,virtuallyeverysentenceproducedbyanyhumanbeingcontainsimportantlymetaphoricalorotherfigurativeelements.Myusejustnowoftheword“element”wasatleastinpartmetaphorical.Orconsiderthenumberoftimesinadaythatsomeoneutterstheword“level.”“Level”isalmostinvariablymetaphor,unlessthespeakerisactuallytalkingaboutahorizontallayeringofsomephysicalthing.Nonliteralusageistherule,nottheexception.Theletteroftheclaimthatalmosteverysentencecontainsfigurativeele-mentsiswidelyconceded,becauseeveryonegrantsthatamongtheliteralexpressionsaremany“dead”metaphors;thatis,phrasesthatevolvedfromwhatwereoriginallynovelmetaphorsbuthaveturnedintoidiomsorclichésandnowmeanliterallywhattheyusedtomeanmetaphorically.Wespeakofariver’s“mouth,”butnooneinthepresentcenturythinksofthisasametaphoricalallusiontohumanoranimalmouths.Likewise“inclinedto[dosuch-and-such],”“richdessert,”“deadmicrophone,”and,forthatmatter,“deadmetaphor.”Perhaps“level”asin“higher/lowerlevel”isnowliteraltoo.“Level”in“carpenter’slevel,”meaningthetool,iscertainlydead;thereisnoothertermforthattool,andinadictionaryitwouldbelistedasaseparatemeaningoftheword.However,ashasbeenemphasizedbyLakoffandJohnson(1980),thedistinctionbetweennovelorfreshmetaphorand“dead”metaphorisoneofsmoothdegree,notofkind.Freshmetaphorsgetpickedupandbecomecurrent,andthenonlyverygradually—sometimesovercenturies—sicken,harden,anddie.(Exactlyhowmanyexpressionsinthepreviousparagraphwereusedmetaphoricallyratherthanliterally,assumingthedistinctionisnotoneofdegree?)Soitseemsintellectualhonestyrequiresustoconfrontmetaphor.Theissues,andtwosimpletheoriesThereissomevariationintaxonomyastohowmetaphorisclassifiedwithrespecttootherfiguresofspeech.Sometheoristsusetheterm“metaphor”verybroadly,asalmostsynonymouswith“figurative.”Othersuseitverynarrowly,asnamingoneveryspecificfigurealongsidemanyotherones.Iwillnottrytomakefinedistinctionshere.Themainphilosophicalquestionsconcerningmetaphoraretwo:whatis“metaphoricalmeaning,”broadlyconstrued?Andbywhatmechanismisitconveyed?Thatis,howdohearersgraspthatmeaning,giventhatwhattheyhearisonlyasentencewhoseliteralmeaningissomethingdifferent?Metaphorraisesmanyfurtherimportantphilosophicalquestions,about\nMetaphor177therationaleforexpressingoneselfmetaphoricallyinsteadofdirectly,thedistinctiveeffectivenessandpowerofmetaphorasafigureofspeech,andthecentralityofmetaphorineachofseveralwalksoflife,butinthischapterIshallconfinemyattentiontothemorespecificallylinguisticquestions.Afewexamplestoworkwith:(1)Simonisarock.1(2)Julietisthesun.2Amorecomplexsolarmetaphor:(3)NowisthewinterofourdiscontentMadeglorioussummerbythissunofYork.3(4)Whenthebloodburns,howprodigalthesoulLendsthetonguevows.4Whatseemstocharacterize(1)–(4)andothersentencescalledmetaphori-cal?Beardsley(1967)identifiestwofeaturesworkingintandem:Withinsuchasentencethereisaconceptual“tension”(humanbeingsdiffercategoriallyfromrocksorsuns,andsoulsandtonguesarenotthekindsofthingsthatcouldinteractcommercially);yetthesentenceisnotonlyintelligiblebutperhapsevenexceptionallyinformativeorilluminating,andmayexpressanimportanttruth.Othertheoristshaveexpressedthefirstofthesetwofea-turesmorestrongly,sayingthatametaphoricalsentenceinterpretedliterallyisincoherent,absurd,oratbesttransparentlyandwildlyfalse—thoughweshallseebelowthatthatisnotalwaysso.Davidson’scausaltheoryFigurativelanguagewasthoroughlydisdainedduringthelogicalpositivistperiod,presumablyonaccountofthepositivists’verificationism.5Sincesuchsentencesas(1)–(4)—atleastastheyareintended—arenotverifiableintheordinaryempiricalway,theywerejudgednottobecognitivelymeaning-ful.Onthisview,thereisnosuchthingas“metaphoricalmeaning”ifby“meaning”onemeanslinguisticmeaning;thereisonlyemotiveoraffectivesignificance.(Ofcourse,manysentencesusedmetaphoricallyhaveliteralmeaningsaswell:“Harold’souttolunch”;“We’reupagainstit.”)DonaldDavidson(1978)toorejects“metaphoricalmeaning”anddeniestheexistenceoflinguisticmechanismsbywhichmetaphoricalsignificanceisconveyed.Unlikethepositivists,hethinkssentenceslike(1)–(4)dohavemeanings;buthecontendsthatthemeaningstheyhavearejusttheirliteralmeanings(howeverstrange).“Metaphorsmeanwhatthewords,intheirmostliteralinterpretation,mean,andnothingmore”(p.30).WhenRomeouttered(2),\n178ThedarksidehewassayingonlythatJulietwas,literally,thesun,thoughdoubtlesshewasdoingmorethanjustexpressingthatludicrousfalsehood.Davidson’sarticleislargelydevotedtohisnegativecaseagainst“meta-phoricalmeaning”;hegivesseveralcriticalarguments,twoofwhichweshallconsiderlateron.Buthedoessketchapositiveaccountofthesignificanceofmetaphor.Itisbrutelycausal:Ametaphormakesusattendtosomelikeness,oftenanovelorsurprisinglikeness,betweentwoormorethings.(p.31,italicsadded)[A]similetellsus,inpart,whatametaphormerelynudgesusintonoting.(p.36,italicsadded)Thereisnologictoit,Davidsonseemstobesaying,muchlessanylinguis-ticmechanismthatindicatesthelikenesstobe“noted.”Apillor“abumponthehead”(p.44)coulddoaswellandasproperly.Obviouslytheeffectofmetaphorisfarfromrandom,orpoetryandotherliteraturewouldnotmakethesensetheydo,muchlesssucceedbrilliantly;butthepsychologicalmeansbywhichtheydosucceedarenotinthelinguist’sdomain.Davidson’sviewimpliesthattheonlyrelevantdifferencebetween(1)–(4)andnonsensestringssuchaschapter1’s“Goodofoffprimlytheathethewhy”isthat,forwhateverreason,(1)–(4)havepsychologicaleffectsthatthewordsaladdoesnot.Butsurelythereisahugecognitivedifferencebetween(1)–(4)andthewordsalad:Weoftennotonlyunderstandthembutcanparaphrasethemmoreliterally;wedrawinferencesfromthem;wesome-timestakeourselvestohavelearnednewempiricalfactsfromhavingheardmetaphoricalutterances.Thatcognitivevaluemanifestlydoesnotderivefromtheirusuallybizarreliteralmeanings.Moran(1997)addstheexampleofembeddinginconditionalantecedents(“Ifmusicbethefoodoflove,playon,”oreven“Ifmusicisthefoodoflove,I’mgoingtobuysomerecords”).Acognatepointisthat,ifDavidsonisright,onecannevermisinterpretametaphor.6IfinresponsetoRomeo’sutteranceof(2),someeavesdropperhadchirped,“Igetit!—Julietdepresseshimbecauseshe’ssostupidandshesmellshorrible,”ontheCausalTheorythiswouldnothavebeenanincor-rectaccountofRomeo’smetaphoricalutterance,butonlyevidencethattheeavesdropper’smentalarchitecturewascausallydifferentfromRomeo’sandfromours.Moreover,asGoodman(1981)urges,Davidsoncannotallowformeta-phoricaltruth.Ifmetaphoricalutteranceshaveonlyliteralmeaning,therebeingnoothercandidateforabeareroftruth-value,theywillnormallybefalseandonlyoccasionallyandaccidentallytrue.Butremember(howeverreluctantly)theprevalenceofnonliteralusage.Evenifwediscountuncon-troversially“dead”metaphor,fewhumanutterancesareentirelyfreeof\nMetaphor179metaphoricalelements.Ifmetaphoricalutterancesarerarelytrue,thenutter-ancesarerarelytrue.Finally,Moran(1997:263)notesthat,whenametaphordies,therelevantexpressionacquiresanewliteralmeaningandaccordinglygetsanadditionaldictionaryentry.Thiswouldbeinexplicable,oratleastarbitraryandodd,ifthemetaphorhadpreviouslyhadnosortofmeaningatall.Andtherearecontemporaryviewsthatrejectmetaphoricalsentencemean-ingbutgivemoreplausibleaccountsofmetaphoricalcommunication.Giventheavailabilityofsuchaccounts,thereisnoreasontoacceptDavidson’spurelycausaltheory.TheNaiveSimileTheoryPhilosophersbeginningwithAristotlehavenoticedastrikingsimilaritybetweenmetaphorsandsimiles:Itseemsthatbothmetaphorsandsimilesexpressorinvitecomparisonsoftheirtopicstosomethingabitunexpected.Simonwaslikearock,Julietislikethesuninoneormorerespects,andEdwardIVresembledthesuninperhapsadifferentway.Thissuggestsanevencloserkinship:theideathatametaphorisjustanabbreviatedsimile.AccordingtotheNaiveSimileTheoryinparticular,ametaphorderivesfromthecorrespondingsimilebyellipsis.Thus,(1)isshortfor“Simonislikearock,”and(2)isshortfor“Julietresemblesthesun.”(3)isconsiderablytrickiersince,althoughitsgrammaticalsubjectcanbetranslatedas“Ourdiscontent,whichislikeawinter,”theconcludingreferent(EdwardIV)isnotmentionedinanyliteralway;onewouldhavetorenderthatlastpartof(3)assomethinglike“byx,aYorkistwhoresemblesthesun,”wherethereferenceof“x”toEdwardissomehowdeterminedbycontext.“Madeglorioussummer”wouldhavetobeinterpretedassomethinglike“alleviatedinamannersimilartoawinter’sgivingwaytoaglorioussum-mer.”Butprobablyallthiscouldbeworkedout.((4)isstilltougher;Ishallreturntoit.)ThissimileviewreconcilesBeardsley’stwofeatures:Itaccommodatesthe“conceptualtension”characterizingametaphor,whileexplainingthemeta-phor’sintelligibility.Theintelligibilityisstraightforward,sincestatementsoflikenessorresemblanceareobviouslyintelligible.Thetensionarisesfromthemovefromlikenesstoactualascription(“Julietisthesun”).TheNaiveSimileTheoryhasseemedplausibleto,andeventakenforgrantedby,manyliterarytheoristsandphilosophersalike.Butitfacesobjec-tions,ofwhichherearethree.First,Beardsley(1967)complainsthat,althoughthetheorydoesexplainthedistinctivetensioninthewayIhavenoted,thatexplanationisveryshallow.Ifametaphorisonlyshortforthecorrespondingsimile,thenitissimplysynonymouswiththesimileandshouldnotbeheardasanomalousorpuzzlinginthefirstplace.Onthisview,thetensionisthemerestsurfaceappearance.Butthatseemswrong.Thereisnoparticulartensionin“Julietis\n180Thedarksidelikethesun,”evenifonewantstobetoldmoreabouttherespectsinwhichJulietresemblesthesun.Onefeelsthatametaphorworksbycontaininganinherenttensionthatismoresubstantive.(Davidson(1978)andSearle(1979b)willgoontoarguethat,inparticular,themetaphorworksbyhavingtheanomalousliteralmeaningthatitdoes.)Second,Searlecomplainsthatasimiletakenbyitselfisalmostentirelyuninformative.“Similarityisavacuouspredicate:anytwothingsaresimilarinsomerespectorother”(1979b:106;seealsoGoodman1970).InwhatwayisJulietsupposedlylikethesun?Notbybeingagiganticballofgas,orbyconsistinginlargepartofnuclearfusion,orbybeing93millionmilesfromtheearth.AsSearlepointsout,thosepropertiesaresalientandwell-knownfeaturesofthesun;yettheNaiveSimileTheorygivesnohintastowhyRomeo’smetaphorimputesdifferentpropertiestoJulietratherthanthose.Thus,thetheoryfailstoofferanymechanismbywhichmetaphoricalsignifi-cancemightbeconveyed.Third,evenwhenwehaveidentifiedtherelevantrespectsofsimilarity,theyoftenprovetobethemselvesmetaphorical.Searlegivestheexample,“Sallyisablockofice.”How,accordingtothenaivesimiletheorist,isSallylikeablockofice?Perhapssheishardandverycold.Butnot,ofcourse,literallyhardorcold;“hard”and“cold”arethemselvesusedmetaphoricallyhere.SoSallyisonlylikesomethingthatishardandcold.Inwhatways?Perhapssheisunyielding,unemotional,andunresponsive.But,Searlepointsout(p.107),thereisnosenseinwhichblocksoficeareunyielding,unemo-tional,andunresponsivebutmanyotherinanimatethingsarenot.Bonfirestooareunyielding,unemotional,andunresponsive;butneither“Sallyislikeabonfire”nor“Sallyisabonfire”ismetaphoricallycompatiblewiththeoriginalsentence.Thenaivesimiletheoristwouldhavetoinsistthatthereisafurtherunderlyingliteralsimilaritybetweencoldthingsandunemotionalthings.Butwearegivennoevidenceforthatclaim.Searleconjecturesthat,onthegroundsofheavenknowswhatpsychologicalfactors,“people[justdo]findthenotionofcoldnessassociatedintheirmindswithlackofemo-tion”(p.108).ThislastobjectionsuggestsasimplebutradicalmodificationoftheNaiveTheory,whichpreservesthecentralclaimthatmetaphorsarecompressedsimilesbutavoidsmostofoursixobjections.ItisarticulatedanddefendedatlengthbyFogelin(1988):thatmetaphorsabbreviate,notsimilestakenliter-ally,butsimilesthemselvestakenfiguratively.TheFigurativeSimileTheorySimilesareoften,perhapsusually,figuresofspeech.Sallyisonlyfigurativelylikeablockofice,forsheisonlyfigurativelyhardandcold.Simonisonlyfigurativelylikearock,andJulietisonlyfigurativelylikethesun.Onewaytoseethis(notFogelin’sownway)istonotethatliteralsimilarityissymmetric:IfAisliterallysimilartoB,thennecessarilyBisliterallysimilartoA.But\nMetaphor181ablockoficeisnotliterallylikeSally,norarockliterallylikeSimon,northesunliterallylikeJuliet.Andnoonewouldproposesuchcomparisonsassimiles,asin“Thesun?—Oh,thesunislikeJuliet.”Itiswhensimilesarethemselvesnonliteralthattheybestparaphrasemetaphors.Thissuggeststhehypothesisthatametaphorisjustanabbreviatedfigurativesimile,derivingfromthecorrespondingsimiletakenfiguratively.ThisFigurativeTheoryeasilysidestepsourthreeobjectionstotheNaiveTheory.Firstobjection:Sincethefigurativetheoristdoesnotreducemeta-phorstoliteralandnear-trivialassertionsofsimilarity,itcannotbesaidthattheFigurativeTheorytreatsthemetaphors’conceptualtensionassuper-ficial.Thereisalreadyconceptualtensionintheunderlyingsimile.Secondobjection:Takenfiguratively,thesimilealreadycarriesoneormoreparticularrespectsofsimilarity.Soitdoesnotfailtoexplainhowthemetaphorbringsoutthosesamerespects.Thirdobjection:OfcoursethefigurativetheoristisnotcommittedtoliteralsimilaritiesbetweenJulietandthesun,Sallyandablockofice,orSimonandarock.Thesethreeadvantagescomeatanobviousprice.Ineachcase,theFigurativeTheoryremediesadeficiencyoftheNaiveTheorybylodgingtheneededmaterialinthenowfigurativelyinterpretedcorrespondingsimilesandlettingtherespectivelyderivedmetaphorsinheritit.Butthedangerhereisthatofonlyputtingofftheproblem.Fornowtheexplanatoryworkisbeingdonebythefigurativenatureoftheunderlyingsimiles,andsotheirfigurativeinterpretationsneedexplaininginturn.Indeed,ouroriginaltwomainquestionsariseforfigurativesimiles:Whatisitforsuchsentencestohavefigurativemeanings,andhowarethosemeaningsconveyedtohearers?Fogelin(1988)exploitsthenotionofasalientfeatureofathing.7(Thosearethefeaturesthatjumpoutatyou,orloomlargeforyou.)Inthatwayheisabletomobilizeanonsymmetricalsimilarityrelation(p.78):“AissimilartoBifandonlyifAhasasufficientnumberofB’ssalientfeatures.”AmayshareasufficientnumberofB’ssalientfeatureswithoutB’ssharingasufficientnumberofA’ssalientfeatures,sincetheparticularfeaturesofBthatAsharesneednotbesalientinA.Forexample,achipmunkisverylikearat,exceptforbeingcuteorperceivedassuchbyhumans;ithasmostoftherat’ssalientfeatures,beingasmallscavengingrodentofloosemorals.Butonewouldnotsaythataratislikeachipmunk,becausethecutenessofchipmunksishighlysalienttohumansandratsarenotcute.AccordingtoFogelin,thedifferencebetweenafigurativecomparisonandaliteraloneisinthestandardofsalience,whichinawayreverses.Itis,Fogelinsays(p.90),literallytruethatWinstonChurchilllookedlikeabulldog,butliterallyfalsethatChurchillwaslikeabulldog(hehavingbeenhumanratherthancanine,two-legged,lackinginfur,giventotalkingratherthanbarking,andfartoobigtocrawlintoadoghouse).Yetitisfigurativelytruethathewaslikeabulldog.Incallinghimone,Fogelinsays,“wecomparehimtoabulldog(asopposed,say,toaFrenchpoodle),whileatthesametimetrimmingthefeaturespaceintermsofthesubject’s[Churchill’s]salient\n182Thedarksidefeatures”(p.91).UnfortunatelyFogelindoesnotgointodetailabout“trim-mingthefeaturespace.”Ibelievetheideaisthat,havingrejectedthesimileasliteral,thehearernonethelesscharitablyassumesthattheallegedsimilaritydoesobtain,andnowignoresthesalientfeaturesofbulldogsthatmostobvi-ouslymaketheliteralcomparisonfalseandlooksforfeaturesthatmatchthesalientfeaturesofChurchill.(Iamnotsurewhatthesewouldbe;toughness,tenacity,earthiness,andlookinglikeabulldog?)Onthisview,sentenceshavemetaphoricalmeaningsincontextthatdifferfromtheirliteralmeanings;yetitdoesnotfollowthatanyexpressioninthesentencehaschangeditsmeaningfromliteraltofigurativeuse,orthatthemetaphoricalmeaningsarespookyormagical.Rather,resemblanceisalwaysandeverywhererelativetoastandardofsimilarity,a“featurespace”thatdetermineswhichpropertiesaretobematchedwithwhich.Thestandardofsimilarityislikeanindexicalinbeingdeterminedbycontextualfactors,butalsocantakemorethanonevaluewithinasinglecontext.Thatiswhythesentencecanbebothtrue(metaphorically)andfalse(literally)ononeandthesameoccasionofutterance:becausetwodifferentstandardsofsimilarityareinplay—muchas“Muffieissmall”canbebothtrueandfalseifMuffieisanundersizedmoose.ThisisaniceadvantageofFogelin’stheory.However,Fogelinfacesatleastthreemoredifficulties.First,astatementmaycontinuetobeacceptedasmetaphoricallytrueevenwhenthecor-respondingsimilehasprovedtobefalse.Searle(1979b:102–3)offerstheexample,“Richardisagorilla,”whichtheNaiveSimileTheorywouldparseas“Richardislikeagorilla.”LetussupposethatwhatismeantisthatRichardislikeagorillainbeingfierce,nasty,pronetoviolence,andperhapsnotverybright.Butprimatologiststellusthat,infact,gorillasarenotnastyorpronetoviolence;theyareshy,rathersensitive,andveryintelligentanimals.Likewisepigs,whichfigureinmanymetaphorsimputingmessiness,filth,greed,obesity,crassness,orsomecombinationofthose:Imyselfknowofnoevidencethatpigsareparticularlygreedy,orthattheyarefatterrelativetotheirskeletalsizethanotheranimalsare.8OnemightthinkthatFogelinhaseasilyavoidedthisnewobjection,forwhenasimileisfigurativeitdoesnotrequiretheactualcorrectnessoftherelevantstereotype.“Samactslikeagorilla”and“Merleeatslikeapig”arecorrectlyexpressedandunderstooddespitethefactthatthetwostereotypesarerespectivelysimianandporcineslanders,because,inthesimiles,“gorilla”and“pig”arethemselvesbeingusedfigurativelyratherthanliterally.ButFogelin’spictureof“trimmingthefeaturespace”presupposesoratleaststronglysuggeststhatthefeaturesrelevantlysharedby,say,Churchillandabulldogarepossessedliterallybyeachofthetwo.Andinthatsense,onFogelin’stheoryametaphormuststillbottomoutinaliteralsharingofgenu-ineproperties.InexamplessuchasSearle’s(inwhichthestereotypeisjustwrong)itisfarfromobviouswhatthepropertieswouldbe.9Second,considerthatmanysentencesindividuallyadmitofeitherliteralormetaphoricalinterpretation.(“Adolfisabutcher”;“Thewormhasturned.”)\nMetaphor183Evenwhenasentenceseemsanomalous,oftenwecanimaginecircumstancesinwhichitwouldbeliterallytrue;asDavidsonnotes(1978:41),“Youarepigs”wouldhavebeenliterallytrueifaddressedbyOdysseustohismeninCirce’spalace.Verylikelytherehasneverbeenasentencethatdoesnotadmitofsomemetaphoricalunderstanding.Foranysentencethatdoesbearmeta-phoricalinterpretation,evenonethatwouldalmostalwaysbeheardliterallyratherthanmetaphorically(say,“Ernestislost”),anysimiletheoristwillhavetocallitsemanticallyambiguous,asbetweenitsliteralmeaninganditssimile-abbreviatingmeaning(thatErnestresemblesalostperson).Butsuchaproliferationofsupposedlygenuinesemanticambiguitiesisimplausible.Thethirdnewobjectionisthatsomemetaphoricalstatementsaretooconvolutedtobeparsedassimiles.(4)aboveissuchanexample.Itisnotliterallyaboutanyone’sblood,andbloodcannotliterallyburn(whilestillwithinthebodyunderevenfaintlynormalconditions);“thesoul”isprob-ablyitselfbeingusedmetaphorically,and,evenifnot,soulscannotliterallylendanythingtotongues;but“tongues”isnotbeingusedtomeantongues,either,andvowsarenotthesortsofthingsthatcanbelent.Soanysimiletheoristfacesthedauntingtaskoftranslatingallofthosethingsatonceintoresemblancetalk.OnewouldhavetomakefreeuseofthesortofcontextualplaceholderthatIusedinexplicating(3).Afirstpassmightbe:“Whenx,whichislikeaperson’sblood,doessomethingthatresemblesburning,howprodigallyy,whichislikeaperson’ssoul,doessomethingsimilartolendingsomethingsthatarevowliketoz,whichresemblesaperson’stongue.”Wearenotmuchthewiser.Andrefinementisneeded,becausefor“theblood”metaphoricallytoburnisprobablysomethingdistinctivetoabloodlikesub-stance,notforittodosomethingthatresemblestheliteralburningof,say,apieceofwood.Itisnowonderthatsimiletheoristshaveinthemainstucktosimplesubject–predicateexampleslike(1)and(2).ThePragmaticTheoryUnlikethecausalview,theNaiveSimileTheorysuppliedanotionof“meta-phoricalmeaning”;sentenceshadmetaphoricalmeaningsinadditiontotheirliteralones,eventhoughtheformermeaningsprovedtobeshallowandunsatisfactory.And,aswehavejustseen,Fogelin’sversionofthefigu-rativeviewendorsesanevenmorerobustnotionofmetaphoricalmeaning,inthathismetaphoricalmeaningsare(evenifineffable)moresubstantiveandilluminating.AsImentionedearlier,Davidsonarguedgloballyagainstmetaphoricalmeaning,indeedseemstoenactascorched-earthpolicy;soweshouldaddresshisarguments.Hegivesfiveorsix,butIhavespacetodiscussonlythetwothatItaketobemostpertinentasdirectedagainsttheothertheoriesdiscussedhere.First:“Therearenoinstructionsfordevisingmetaphors;thereisnomanualfordeterminingwhatametaphor‘means’or‘says’;thereisnotestformetaphorthatdoesnotcallfortaste”(Davidson1978:29).\n184ThedarksideSecond:Itisalmostuniversallyconcededthatalthoughsomemetaphorscanbeparaphrasedinliteraltermswithoutgreatloss,manyareopen-endedinthattherelevantsetofsimilaritiesisvagueandindefinite,andsome(asinthepoetryofe.e.cummings)10cannotbeparaphrasedatall.Thesestrikingfactsareneatlyexplainedbytheclaimthatthereisnometaphoricalmean-ing,foronthatviewthereisnothingtoparaphraseorcircumscribe(p.30).FogelinarguesthattheFigurativeSimileTheoryexplainsthosefactsaswell.ButDavidsonaddsthatouruptakeofametaphor,“whatwenoticeorsee,”“isnot,ingeneral,propositionalincharacter[atall]...Seeingasisnotseeingthat”(p.45).Moreover,ifagivensentencedidhaveametaphoricalmeaning,wewouldexpectthatthatcontentcouldbefairlyaccuratelyexpressedbysomeparaphrase,eveniftheparaphrasewerecumbersome,prolix,flat,bor-ing,orallofthose.Now,Davidson’sattackonmetaphoricalmeaningcanbeexaggerated,asitisinspotsbyhisownrhetoric.AsIsaid,itispresentedasascorched-earthorzero-tolerancepolicy.ButinfactDavidsonconcentrateshiscriticalargumentsontheideathatlinguisticexpressionschangetheirmeaningsinmetaphoricalusage;hisbêtenoireisthepositingoflinguisticambiguity.Andatoneimportantpointheiscareful“nottodenythatthereissuchathingasmetaphoricaltruth,onlytodenyitofsentences”(p.39).Thisleavesopenthepossibilitythatthereisamiddlewayorcompromiseposition.Searle(1979b)proposesanaccountofmetaphorthatjoinsDavidsonindeflating“metaphoricalmeaning”evenfurtherthandidtheNaiveTheory,andinrejectingalinguisticambiguityview.ButasagainstDavidsonittakesseriouslytheideathatmetaphoricalutteranceisgenuinelylinguisticcom-municationratherthanmerecausation,anditpositsacognitivemechanismthatcomputessomethingwellworthcallingmetaphoricalmeaning.IshallcallSearle’sviewthePragmaticTheory,forheseesmetaphorassimplyaspeciesofGricean(orrelevance-theoretic)communicationinthesenseofchapter13.11RecallfromthatchapterthatSearle(1975)hadoffereda“conservative”accountofhowindirectspeechactsareperformedandunder-stood.Thespeakeruttersasentencegrammaticallymarkedforonerangeofillocutionaryforcebutprimarilymeanssomethingbyitthathasadifferentforceoratleastacharacteristicallydifferentlocutionarycontent.ThehearerfirstusesGriceanreasoningtodeterminethatthespeakeristryingtoconveysomethingotherthanwhather/hissentenceliterallymeans;thenthehearerusesfurtherGriceanreasoningaugmentedbyprinciplesofspeech-acttheoryandbymutuallyobviouscontextualassumptionstoworkouttheintendedforceandcontentoftheutterance.AccordingtoSearle:Theproblemofexplaininghowmetaphorsworkisaspecialcaseofthegeneralproblemofexplaininghowspeakermeaningandsentenceorwordmeaningcomeapart....Ourtaskinconstructingatheoryof\nMetaphor185metaphoristotrytostatetheprincipleswhichrelateliteralsentencemeaningtometaphorical[speaker’s]utterancemeaning.(pp.92–3)Searlebreaksdowntheinterpretiveprocessintothreesteps(paralleltotheprocesshehadpositedfortheinterpretationofindirectspeechacts).Firstthehearermustdeterminewhethertolookforanonliteralinterpretationinthefirstplace.Second,ifthehearerhasdecidedtoseekametaphoricalinterpretation,s/hemustthenmobilizesomesetofprinciplesorstrategiesforgeneratingarangeofpossiblespeakermeanings.Third,s/hemustemployafurthersetofprinciplesorstrategiesforidentifyingwhichmeaningormeaningsfromamongthatrangearemostlikelytobeinplayonthepres-entoccasion.(Notethat,ifthisfurthersetcannotparethelikelymeaningsdowntooneortwo,thatwouldexplainthefrequentopen-endednessofmetaphor.)TheobviousstrategyunderlyingthefirststepisGricean:Whenanutterancewouldbeobviouslydefectiveiftakenliterally,lookforadifferentspeakermeaning.Our(1)–(4)allfitthismodelsince,consideredliterally,eachisfalsetothepointofconceptualconfusion.(AsSearlesays,however,notallmetaphoricalsentencesareoutrageousfalsehoodsorevenfalseatall.Thedefectinliterallyuttering“Rockyisarealman,”“Thequalityofmercyisnotstrained,”orMáoZédōng’s“Arevolutionisnotadinnerparty,”istheirutterpointlessnessduetotheirexcessivelyplaintruth.)TheGriceanstrategyisnottheonlyfirst-stepoption.Somemetaphoricalutterancesarenotinanywaydefective;thereareothercontextualcues,suchasthekindofdiscoursethatistakingplace.Searleobservesthat,“whenread-ingRomanticpoets,weareonthelookoutformetaphors”(p.114).Kittay(1987:76)notesthatmetaphorsaresometimesexplicitlyflaggedassuch:Avictimistiedagainstawallbythieves.“Herealizedthatbothliterallyandmetaphorically...hewasupagainstthewall,and...hishandsweretied.”Theprincipalgeneralstrategyforthesecondstep,Searleunsurprisinglysays,istolookforsimilaritiesorcomparisons.Searleofferseightprinciplesaccordingtowhichtheutteredphrasecancalltomindadifferentmeaning“inwaysthatarepeculiartometaphor.”Forexample(principle2),thediffer-entmeaningcanbea“salientorwellknownproperty”ofthethingorstateofaffairsmentioned.Or(principle3),asinour“gorilla”and“pig”examples,theintendedpropertycanbeonethatisonlyoftenimputedtothething.Searlementionsjustonestrategyforthethirdstep:toconsiderwhichofthemeaningcandidatesarelikelyorevenpossiblefeaturesofthesubjectunderdiscussion.Julietcouldnotbeagiantballofgas,orconsistlargelyofnuclearfusion,orbe93millionmilesfromtheearth.Ofcourse,hear-ersalsoknowthingsaboutwhatideasparticularspeakersarelikelytobeexpressing.Thereremainsthetaskofdistinguishingmetaphorasaspeciesofindirectcommunicationfromotherssuchasordinaryimplicature,irony,andwhat\n186ThedarksideSearlecalls“indirectspeechacts”proper.Searlecontrastsmetaphorwithindirectspeechactsbycontending(p.121)thatinthelattercasethespeakermeanswhats/hesaysinadditiontomeaningsomethingmoreaswell.(Hedoesnotaddressordinaryimplicature,buthemightwellsaythesameinitsregard.)Thedifferencebetweenmetaphorandironyseemstobejustthatwithmetaphorthesecondandthirdstagesofinterpretationworkbysimilarityorcomparison,whereaswithironythereisasimplersortofreflex:Theutterancetakenliterallyisdefectiveinthateithertheoppositeisclearlytrueorthespeakermaybeexpectedtobelievetheopposite,sothe“natural”choiceofindirectmeaningisjusttheopposite.DavidsonandSearleagreemorethantheydisagree.Bothdenythatlin-guisticexpressionshavespecialmetaphoricalmeanings,andbothholdthatmetaphorcanbeunderstoodusingapparatusalreadyonhandinmainstreamphilosophyoflanguage.(FogelinaptlyclassifiestheCausalandPragmaticTheoriestogetheras“fecundfalsehood”theories.)Moreover,IdonotseewhyDavidsonshould,orhowhecould,disputeSearle’sviewthatthereismetaphoricalspeaker-meaning.Hedoesargue,contrarytoSearle’sview,thatwhatsomemetaphorsconveyisnotpropositionalatall.Butthebiggestdisagreementisoverrules,principlesandcognitivemechanisms,DavidsonstoutlydenyinganyandSearleeagerlyproposingquiteafew.SoletusseehowSearlemightrebutDavidson’stwoargumentsagainst“metaphoricalmeaning.”Davidsonfirstarguedthattherearenoinstructionsorrulesforgeneratingorforinterpretingmetaphors.Asifdirectlyinspiredbythatpassage,Searleproducedquiteanumberofsuchrules,andsofarastheygotheyareplausible.Davidsonaddedthequalification,“notestformetaphorthatdoesnotcallfortaste”;verylikelySearlewouldconcedethatpoint,sincehemakesnoclaimtocompletenessanddoesnotpredictthatevenafinalsetofprincipleswillgiveperfectlydeterminateresults.Buthewinsthisroundonpoints.Davidson’ssecondappealwastoopen-endedness,unparaphrasability,anddownrightnonpropositionality.Searle’saccountpredictsopen-endedness,sincewemayexpectthathissecondandthirdstageswilloftenfailtotampdownthepossiblespeaker-meaningstojustoneortwo.Asforunparaphras-ability,Searlegrantsthatoftenweusemetaphorpreciselybecausethereisnohandyandaccessibleliteralexpressionthatmeansthesamething,buthearguesthat,ifsomethingisalinguisticmeaningatall,inprincipleitcouldbeformulated(howevercumbersomely)insomelanguageorother.IthinkSearlewinsthatroundalso,butthereisadeeperissueaboutnonpropositionality.Searle’saccountispropositionaltothecore,sinceallspeaker-meaningismeaningthatso-and-so.IfDavidsonisrightthatwhatwenoticeorseeinmetaphor“isnot,ingeneral,propositionalincharacter,”thenbySearle’sownprincipleaforementioneditisnotalinguisticmeaningofanykind,notevenaspeaker-meaning.Davidson’s“ingeneral”makeshisclaimfairlyambitious,indeedfalse.Perhapsmanypoeticandotherliterarymetaphorsaresorichastobe\nMetaphor187nonpropositionalintheirpurport,buteverydaymetaphorsusedcasuallybyordinarypeopleareoftenperfectlyparaphrasableincontext.Quiteoften,justasSearlesays,thespeakercertainlydoesmeansomething,possiblysomethingquitespecific.Hanscomesintohisapartmentandfindsitadis-gustingmess12—dirtyunderwearonthefloor,fourdays’worthofdishesfesteringinthesink,otheritemsnottobementionedinafamilypublicationsuchasthisbook—andHansaccostshisroom-mate:“Youpig!”Hemeansfairlypreciselythathisroom-mateisafilthyslob.(Hadheinsteadfoundtheapartmentneatandcleanbutallthegoodfoodgonebecausehisroom-matehadscarfedit,hemighthavesaid“Youpig!”meaningthathisroom-mateisaglutton.)SoIthinkDavidsonhasoverstatedhiscasebyoverlookingfactsofspeaker-meaning.Ontheotherhand,justasDavidsonsays,writerswhostrewfreshliterarymetaphors,farfromalwayshavingdeterminatespeaker-meanings,mayhavenospeaker-meaningsorotherpropositionalintentatall.Thatdoesnotmakethemetaphorsanylessgoodoruseful,becausemetaphordoessometimeshavethequasi-perceptualcharacternotedbyDavidson;insomecasesmeta-phoraffectsone’sliterallyperceptualset.(Inother,intermediatecases,themetaphorjustputsoneinadifferentintellectualframeofmindforthinkingaboutthetopicathand.)AndthatisatellingpointagainstSearle.Thus,eachviewhasatleastoneadvantageovertheother.Ibelievethatarapprochementispossible,ahybridviewthatcombinestheadvantagesoftheCausalTheoryandthePragmaticTheory.ButIleavethattoyouasanexercise,andmerelynotetwofurtherobjectionsmadeagainstSearle.First,Cooper(1986)andMoran(1997)pointoutthat,ifmetaphoricalmeaningissimplyspeaker-meaning,thenitisdeterminedbyandconfinedtothespeaker’sintentions.Yetincasesoffreshmetaphor,asCoopersays(p.73),“evenaquitedefinitespeaker-intentiondoesnotfinallydeterminethemeaningofametaphor.”Moranaddsthat“theinterpretationofthelight[themetaphor]shedsonitssubjectmayoutrunanythingthespeakeristhoughtexplicitlytohavehadinmind”(p.264).Second,Ross(1981)andKittay(1987)callourattentiontoaclassofmetaphorphenomena,sometimescalledanalogical,thatindisputablyinvolvemeaningandmeaningshiftbutareaddressedneitherbyDavidson’sviewnorbySearle’s.Theyarepervasive;theyoccurinnearlyeverysentencethatcomesoutofourmouths.Iwilltrytoacquaintyouwiththem;unfortunately,thetheoriesofmetaphorthatexploitthemareverydenseandcomplicated,andIwillnothavethespacetoexpoundthose.MetaphorasanalogicalAsawayintotheareainquestion,IintroducetheinfinitepolysemythesisdefendedbyWeinreich(1966),Lyons(1977),Cohen(1985),andDavidson(1986),aswellasbyRossandKittayandbyrelevancetheorists.Thatdoctrineconcernslexicalmeaning,themeaningsofwordsandshortphrasesrather\n188Thedarksidethanthoseofwholesentences.Itisthatvirtuallyanyword,evenapronoun,maytakeonanynumberofnovelanddistinctlexicalmeaningswithoutlimit,givenasuitablevarietyofenvironmentswithinsentencesinwhichtheyoccur.Indeed,oneandthesameword,dependingonthesubsententialcontextandunderstrangeenoughexternalcircumstances,canmeanalmostanything.Moreover—whatismostsurprising—wordsdothisinsuchawaythatthenovelmeaningscanbegraspedonthespotbynormalhearers.Allthisisbecausenovelwordmeaningsaregeneratedincontextfromexistingonesbyintricatebutfairlytractablemechanismsofanalogythataremobilizedautomaticallybyeverynormalspeaker.13Forthesamereason,veryfewsuchdifferencesofwordmeaningareutter,bruteambiguitiessuchasthatof“bank”(financialvs.flyingtechnique)or“die”(toperishvs.asusedincrapsandinboardgames);thepolysemousmeaningsaresystemati-callyinterrelated.Considerthefollowingsetsofexamples:a“Shedroppedastitch”;“Shedroppedherhem-line;“Shedroppedherbook”;“Shedroppedafriend”;“Shedroppedhercourses”(Ross1981:33);“Shedroppedhereyes”(Kittay1987:154).Eachoccurrenceof“dropped”inthislistmeanssomethingatleastslightlydifferent(andwemayaddthefurthernounforms“letterdrop,”“parachutedrop,”and“dropofblood”).Moreover,Rosssays,“themeanings...areappropri-ate,fittedtothecompletionwords”.b“Hepickedadate”;“Heappointedadate”;“Hefixedadate”;“Hewantedadate”;“Heborrowedadate”(Ross1981:80–1).Rossnotesthateachofthesesentencesisstillambiguous,andtheambiguitycouldbereducedonlybytheadditionofwidercontexts.c“Hechargedthegun”;“Hechargedthejury”;“Hechargedherwithmurder”;“Hechargedhimwithresponsibility”;“Hechargedmorethanthelawallowed”;“Hechargedtheboytoomuch”;“Hechargedthebattery”(paraphrasedfromRoss1981:100).dMyownexample:“Deadman”;“deadduck”;“deadsilence”;“deadringer”;“deadmarch”;“deadeye”;“deadend”;“deadhead”;“deadassets”;“deadheat”;“deadbolt”;“deadlanguage”;“deadwrong”;“deaddrunk”;“deadtired”;“deadboring”;“deadset(on)”;“thedeadofwinter.”ePrepositionssuchas“in,”“on”and“of”notoriouslyhavenoconstantmeaningfromcontexttocontext.(LakoffandJohnson(1980)makethispointtrenchantly.)fItoccurstomethatevenaffixesandcasemarkersarethuspolysemous.Thepossessiveinparticulardenotesseeminglycountlessdifferentrela-tionships,onlysomeofthemexpressibleas“ownership”inanysenseatall.\nMetaphor189Itistruethatwearemakingveryfinedistinctionshere.Someonemightwelldenythatalltheforegoingwordusesactuallydifferinmeaning,anditmaybesuggestedthatsomeofthedifferencesareonlyoftoneorofconnota-tion.Butwhenweask,asadiagnostic,whetherasentenceofthesortlistedcansimultaneouslyhavemorethanonetruth-valuedependingondisam-biguation,theanswerisobviouslyyes.KittayremindsusofPeggyParish’schildren’sbooks,whosemaincharacter,thehousemaidAmeliaBedelia,isdeaftosuchvariations.Whenaskedto‘dustthefurniture’,sheusesapowder-pufftospreadfacepowderonthefurniture;whenrequiredto‘drawthecurtains’,sheproducesasketchofthem;andwhenaskedto‘dressthechicken’,sheputsaminiaturepairoftrousersandshirtonabirdintendedforthatnight’sdinner.(1987:111)“Dust,”“draw”and“dress”eachhavedifferentsenses,howevercloselythosesensesmayberelatedtoeachother.Unfortunatelyforourpurposes,eachofthetwoleadingtheoriesofana-logicalmeaningdifferentiation,Ross’sandKittay’s—andevenmoretheirextensiontoatheoryofmetaphorinparticular—arefartoocomplicatedeventobesketchedhere.Icanonlyreferyoutotheirworks,andhopeyouhaveenjoyedthislimitedlookatthedarksideofphilosophyoflanguage.Summary•Thephenomenonofmetaphorisfarmoreprevalentthanisgenerallyadmittedbyphilosophers,anditraisestwomainquestions:Whatis“metaphoricalmeaning”?Andhowdohearersgraspmetaphoricalmeaningasreadilyastheydo?•Mosttheoristshavethoughtthatmetaphorissomehowamatterofbringingoutsimilaritiesbetweenthingsorstatesofaffairs.•Davidsonarguesthatthestimulationofcomparisonsispurelycausal,notlinguistic.Attheoppositeextreme,theNaiveSimileTheoryhasitthatmetaphorssimplyabbreviateliteralcomparisons.Bothviewsareeasilyrefuted.•AccordingtotheFigurativeSimileTheory,rather,metaphorsareshortforsimilesthemselvestakenfiguratively.ThisviewavoidsafewoftheobjectionstotheNaiveSimileTheory,butnotothers.•SearlemobilizesGriceanapparatustoexplainmetaphoricalmean-ingasspeaker-meaning.ThishassomeplausibilityandovercomesDavidson’sleadingobjectionstometaphoricalmeaning,butincursotherobjections.\n190Thedarkside•Afurthertheoryofmetaphorisbasedonthephenomenon,importantinitsownright,ofsinglewords’analogicaldifferentiationintohostsofdistinctthoughrelatedmeanings.Questions1IstheremoretobesaidforDavidson’scausaltheory,orfortheNaiveSimileTheory?2PursueFogelin’sFigurativeSimileTheory,attackingordefending.3TrytocomeupwithacompromiseviewbetweenDavidsonandSearle.4DefendSearleagainstoneormoreofourobjections,ormakeafurtherobjection.5Ifyouarewillingtodosomeoutsidereading,discussRoss’sandKittay’s“analogy”phenomena.(Notforthefaintofheart.)Furtherreading•Black(1954/1962)wasaseminalpaperandattractedwidecommentary.•Reimer(2001)rehabilitatesDavidson’scausaltheory.•Johnson(1981)isausefulanthology.•Cohen(1975),ElginandScheffler(1987),Tirrell(1989)andStern(2000)offerfurtherapproachestometaphor.White’s(1996)workisparticularlysubtleanddetailed.\nNotesChapter11Alice’sAdventuresinWonderlandandThroughTheLookingGlass(London:Methuen,1978),p.180.2CompletePoems,1913–1962(NewYork:Harcourt,Brace,Jovanovich,1972).Chapter21TheAdventuresofSherlockHolmes,Vol.I,ed.E.W.Smith,NewYork:HeritagePress,1950.AcuriousfactaboutMoriartyisthathehasabrother,anarmycolonel,whoisalsonamedJames.(IfyouareaHolmesbuffandyoudidnotalreadyknowthat,youwillenjoyverifyingityourself.)2Itwasbyarguingthatexistenceisaperfection,inparticular,thatAnselmpurportedtoprovetheexistenceofGod.3No,sorry,youwillhavetogetthatstoryfromRussell’sbiographies.4Hemeant“Scottish.”(Inlatertwentieth-centuryusageandsince,Scotchisatypeofwhisky,indeedtheonlytypethatisallowedtobespelled“whisky”asopposedto“whiskey.”)ButsincetheexampleisafamousoneofRussell’sIshallstickwithitasitis.5Asyoushouldexpect,thereisasecondwayofapplyingtheanalysisto(9),justasthereweretwowaysofapplyingitto(7)onaccountofourhavingachoiceastowheretoputthe“not.”Theotherwayistogivethedescriptionprimaryoccurrence,orwidescopewithrespectto“Albertbelievesthat.”TheRussellianparaphrasewouldthenbe:“AtleastonepersonauthoredNothingandBeingness,andatmostonepersonauthoredNothingandBeingness,andwhoeverauthoredNothingandBeingnessisbelievedbyAlberttobeaprofoundthinker.”Onthisreading,(9)assertsarelationofbeliefholdingbetweenAlbertandourmoonlightingauthor—thepersonhimself,regardlessofhowheisdescribed—butthisreadingisexceptionallyhardtohear,especiallysincecoreferringdescriptionsdosubstituteintoitwithoutchangingtruth-value.The“secondary”understandingof(9)ismuchmorecommonandnatural.6Russelladdedafifthpuzzleaswell,whichwemaycalltheProblemoftheExcludedMiddle:Neither(1),“ThepresentKingofFranceisbald,”noritsapparentnegation,“ThepresentKingofFranceisnotbald,”istrue.Yetalawoflogicsaysthat,ofasentenceanditsnegation,onemustbetrue.(Russelladdedthat,sinceitseemstheKingisneitherbaldnornotbald,“Hegelians,wholoveasynthesis,willprobablyconcludethathewearsawig”(1905/1956:48).)Ileaveittoyouasanexercisetosolvethisfifthpuzzle,inlightofRussell’streatmentsoftheotherfour.\n192Notes7Strawsonnotesthatthereareexceptions;occasionallyasentencecontaininganonreferringdescriptionisout-and-outfalse.SeeNeale(1990),Lasersohn(1993),andYablo(2006).8G.K.ChestertonbasesoneofhisFatherBrownmysterystories,“TheInvisibleMan,”entirelyonthisphenomenon.9ActuallyDonnellan’scharacterizationsdonotperfectlylineupwitheachother.Forexample,eveninthereferentialcaseasheintendsitwedonotalways“expectandintendouraudiencetorealizewhomwehaveinmind,andmostimportantly,toknowthatitisthispersonaboutwhomwearegoingtosaysomething,”forImaybreathe“Smith’smurdererisinsane”solelytomyself,withoutexpectingorintendinganyonetorealizeanything.“Donnellan’sdistinction”seemstobeafamilyofrelatedbutdistinctdistinctions;commentatorshavegoneontotrytosortthisout(forexample,Searle1979b;Bertolet1980;Devitt1981b).10AdeterminedRussellianmighttrytoexplicatetheambiguitiesalongthelinesofthatof(7)and(9),asdependingonwhetherweapplyRussell’sanalysisinsideoroutside“because”and“wish”respectively.Tryit.Chapter31Meinong(asmentionedinChapter2)wouldhaveinsistedthatthereisawingedhorse,named“Pegasus,”andwhatsentence(2)doesispredicatenonexistenceofthatparticularhorse.Onthisview,(2)isjustlike“Pegasusneveratealfalfa”;existingissomethingthatyouandIdobecausewegotlucky,butthatPegasusdidnotmanagetodo,whetherornotanyofushasanychoiceinthematter.NeitherFregenorRussellcouldacceptthatview(eventhoughRussellhadoncetakenitveryseriously);itismuchmoreplausibletothinkthat(2)means,rather,justthatthemythwasonlyamyth,thattherewasnosuchwingedhorsethatwasriddenbyBellerophon.2However,donotthinkthatRussell’stheoryisonlyavariantofFrege’s.Thereareseveralimportantdifferencesbetweenthetwo,andtheymakefordifferentsetsofstrengthsandweaknesses.3InChapter2wedefinedthenotionofadescription’ssemanticdenotatum/referent,andsomeonemightsuggestthatthisprovidestheneededpointofcontactbetweenourtwodisputants.ButthatwouldignorethefactthatthereisnoincompatibilitywhateverbetweenRussell’s(7)and(8).ImprovingonRussell,alaterdescriptivistmightputtheNameClaimintermsofdescriptionsusedreferentially,andappealtothefactnotedinourdiscussionofDonnellanthatcommunicationgoesbyspeaker-referentratherthansemanticreferent.Thatwouldintroduceanotionof“disagreement”betweenspeakersthatisweakerthanclashofsemanticcontent.4Thispointneedsfurtherinvestigation,tosaytheleast,sinceonSearle’sview,eventhoughtwosuchspeakersmaysucceedinpickingoutthesameindividual,thesentencestheyusewillstillhavedifferentmeaningsand,forallthathasbeenshown,wemaystillgetthenondisagreementproblem.5ThisisanexampleoftheRussellian’sdeparturefromFrege.Fregedidnotallowthatpropernameshavescopeatall.6InintroducingthisexampleinlectureformatPrincetonUniversityin1970,Kripkeinterjected,“IhopeProfessorGödelisnotpresent”(p.83).Chapter41SeeLewis(1986)andLycan(1994).\nNotes1932Thisisanimportantqualification.Ifatermdesignatedthesameitemineverypossibleworldthereis,thatwouldmeanthattheitemexistedineverypossibleworld,andthatinturnwouldmeanthattheitemcouldnotbuthaveexisted.Noordinarythingorpersonhasthatkindofinevitability.ThoughyouandIandtheBrooklynBridgedoreallyexist,wemightnothave,andsothereareworldsinwhichwedonotexist.Whatsortofitemexistsineverypossibleworld?God,perhaps.Kripkeisinclinedtothinkthatthenumbers—atleastthenaturalnumbersstarting0,1,2,...—existineverypossibleworld.Ifso,thenthenumeralsthatrefertothempresumablydodesignatethesamethingsineverypossibleworld,period.Butthatishardlythenormalcase.3Forfurtherexamplesofnamesusedflaccidly,seeBoër(1978).4InKripke(1979b)hecomesbackandheusesavariationontheSubstitutivitypuzzleaboutreferringexpressionstorefutetheMillianview.Hisargumenttherealsoseemstoembarrasshisownrigiditythesis,buthedoesnotofferanyalternatepositiveview.Kaplan(1975)fashionsamade-upword“dthat”(pronounced“that”),whichtakesanordinarydescriptionlike“themaninthecorner”andmakesitdenoteitssatisfierrigidlyratherthanflaccidlyorattributively.Thus,“dthatmaninthecorner”referstoagivenpossibleworld,nottowhatevermanisinthecornerinthatworld,buttothesamemanwhoisinthecornerinthisworld.IfIuse“dthatmaninthecorner,”youshouldunderstanditastalkingsimplyaboutthatperson,andmyhavingputintheconceptualcontent,alludingtomanhoodandin-the-cornerness,isjustawayofcallingyourattentiontothatman,asifIwerefixingthereferenceofmyowndescriptionwithoutfixingitssense.So“dthat”functionsasarigidifier.Plantinga(1978)andAckerman(1979)enlistadifferentversionoftherigidifyingideaindefendingpositivetheoriesaccordingtowhichpropernamesarerigidbutnotMillian;seebelow.5Ofcourse,if“Tully”isalsoaMillianname,thatwouldamounttodoubtingthatthepersonreferredtoisthatveryperson.Butthistooisapossibleunderstandingof(4).Incidentally,thepointabouttransparentreadingscanalsobemaderegardingpronouns.AddressingJacqueshimself,wecouldsay“Thepolicethinkyoudroppedtheanchoviesasyouran”(Sosa1970;Schiffer1979).6EventhepaperbackeditionofLycan(1994)isexpensive,Iamafraid,butwellwortheverypenny.7AndinsufficientlyemphasizedbyKripke.ItwasfirstreallyprosecutedbyDevitt(1989).8KripkecitesH.L.Ginsberg,TheFiveMegillothandjonah(JewishPublicationSocietyofAmerica,1969),asseriouslydefendingthisview.NoticeaswellthatJonah’snamecouldnothavebeen“Jonah”;thereisno“j”soundinHebrew.DavidKaplanoncemaintained(ina1971lecture)thatthereisatleastonereal-worldexampleofthistypethatfavorstheCausal–HistoricalTheoryoverSearle’saccountofreferring:thename“RobinHood.”Itseemshistorianshaddiscoveredthattherereallywasapersonwho(causally)gaverisetotheRobinHoodlegend.Itturnsout,though,thatthatpersonwasnotpoor,helivednowherenearSherwoodForest,hewasnotanoutlaw(infact,heandtheSheriffofNottinghamwerefairlyclose),andhisnamewasnoteven“RobinHood.”Onthecausal–historicalaccount,thismakesperfectsense.9Thismovewouldalsohelpwithtwosimilarproblems:thenamesoffutureindividuals(“Let’strytohaveababy,andifwesucceeditsnamewillbe‘Kim’”);andthenamesofabstractobjects,suchasindividualnumbers,whichhavenocausalpowers.Youshouldwonderwhy,iftherelevantcausal–historicalchainoriginatedinanamingevent,thenamingeventitselfisnottheactualreferentofthename.(Thus,“Pegasustookonly30secondsandwaslittlestrainfortheauthor”might\n194Notesbeatruesentenceaboutanactualitem,notafictionalsentenceatall.)Onemightfairlystipulatethatnamingeventsarenotreferentsunlesstheyarethemselvestheobjectsoffurthernamingevents;orseethereplytoobjection4below.10HecitesIsaacTaylor’s1898book,NamesandTheirHistory:AHandbookofHistoricalGeographyandTopographicalNomenclature(Detroit,MI:GaleResearchCo.,1969).11London:Sidgwick&Jackson,1927.12ItiscontestedbySearle(1983),Rosenberg(1994),andSegal(2000).13ThealertreaderwillhavenoticedaninfelicityinPutnam’sexample:sinceaveryhighproportionofthehumanbodyisconstitutedbywater,theTwin-Earthlingscanhardlybemolecularduplicatesofus.Ignorethis,orifitreallybothersyouchangetheexampletoanaturalkindthatisunrepresentedinthehumanbody.14Burge(1979)arguesonthebasisofaTwin-Earth-styleexamplethatwhataspeakermeansbyalinguistictermdependsinpartonusageinthesurroundingcommunity,andsoisnotdeterminedbythespeaker’sheadcontents.ThiswouldbelesssurprisingthanPutnam’spoint,thoughitdoessustainhisclaimthatmeaning“ain’tinthehead.”(Burge’srealconcerninthearticleisformind,notlanguage:toshowthatevenbeliefcontentsarenotinthehead.)Chapter51“ThefactisthatabsolutelyallthecontentsoftheUniverse,absolutelyeverythingthatisatall,maybedividedintotwoclasses—namelyintopropositions,ontheonehand,andintothingswhicharenotpropositionsontheotherhand”(Moore1953/1962:71).Moorereportedinanautobiographicalnotethathehadoncehadanightmareinwhichhedreamedthatpropositionsweretables.2Though,like“idea,”“concept”hasalsobeenusedtomeanakindofparticularmentalentity.Thisequivocationhascausedsomeconfusionincontemporarycognitivepsychology.3“Whydoesopiumputpeopletosleep?”—“Becauseithasadormitivevirtue.”ThatmaysoundprofounduntilonerealizesthatthephraseisjusttransliteratedLatinfor“powerofproducingsleep.”Thephysician(ArganinLeMaladeImaginaire)mightaswellhavespokeninPigLatin:“Itputspeopletosleepbecauseitayutspayeoplepayotayeepslay.”Thatishardlyanexplanation.Chapter61Herearethreeinfrequentlynoticedwaysinwhichthenotionofa“sentence”isquiteaconsiderableabstractionawayfromreal-worldlinguisticactivity.First(youmaybesurprisedtolearn),humanutterancesdonotcomebrokenupintoseparatewords.Anacousticalanalysisoforalspeechproductionshowsacontinuousthoughofcoursevariegatedstreamofsound.(When...we...talk...we...do...not...pause...even...briefly...between...words.)Whenwehearastreamofsoundthatconstitutessomebodyspeaking,weputthewordbreaksinourselves,automaticallyandwithoutevereventhinkingaboutit.Thatisalreadyanabstraction,atheoreticaloranalyticalmovethatwemake.Second,tothinkofsomethingasa“sentence”presupposesthenotionofgrammaticalwell-formedness.Notjustanystringofwordsconstitutesasentence;onlythegrammaticalstringsdo.Andtheideaofgrammaticalityisasophisticatedone,eventhoughitisgraspedhoweverdimlybyfour-year-olds.Third,considerthecategoryofwhatlinguistsusedtocallsemi-grammaticalutterance.Someoftheutterancesthatpeopleproduceareonlysemi-grammatical,\nNotes195inthatiftheirwordswerewrittendownonpaper,theresultwouldnotcountasanentirelygrammaticalsentencebysomeruleofgrammar(ithassomegrammaticalinfelicityinit),butitiscoherentenoughtobeunderstood.Infact,Isuspectmostpeopletalkthatwaymostofthetime.Attheveryleast,wealldothingslikemakingfalsestarts,andweallengagevigorouslyinmid-courseediting.Yetnotonlydowegetunderstoodbyourhearers;nobodyevennotices.Wecorrectforsemi-grammaticalityquiteautomatically.Thatcorrectionisatheoreticalmovemadebyourbrains,andyetanotherabstractionawayfromspeecheventsintherealworld.2Paragraph43ofWittgenstein’s(1953)PhilosophicalInvestigationsisfamouslymisquoted.Itreads,“Foralargeclassofcases—thoughnotforall—inwhichweemploytheword‘meaning’itcanbedefinedthus:themeaningofawordisitsuseinthelanguage.”Wittgensteinwasveryseriousabout“thoughnotforall”;hedidnotholdthat“meaningisuse,”period.Indeed,hewasallergictouniversalgeneralizations.Hethoughtitadeepdefectofphilosophythatphilosophyseeksuniversalgeneralizations;therealworld,hecontended,isalwaysmorecomplicatedthanthat.AsGeorgHenrikvonWrightputit,Wittgenstein“livedontheborderofmentalillness...throughouthislife”(“BiographicalSketch,”inNormanMalcolm,LudwigWittgenstein:AMemoir,Oxford:OxfordUniversityPress,1958).Wittgensteinalsodistinguishedhimselffrommosttwentieth-centuryEnglish-speakingphilosophersbyhavinghadquiteaninterestinglife;seeRayMonk’swonderfulbiography,LudwigWittgenstein:TheDutyofGenius(NewYork:FreePress,MaxwellMacmillanInternational,1990).3ReportedbyNormanMalcolm(1958:65).“Acentralideaofhisphilosophy,thenotionofa‘language-game’,apparentlyhaditsgenesisinthisincident.”4Horwich(1998)offersasimilariflesswellworked-outpicture.UnlikeBrandom,heemphasizesthatindividualexpressionshavemeanings:agivenexpression’s“meaningproperty”is“itsusebeinggovernedbysuch-and-suchregularity—or,morespecifically,thepropertythateveryuseofthewordisexplainedintermsofthefactthatweacceptcertainspecifiedsentencescontainingit”(p.6,italicsoriginal).Foreachword,thereisa“basicuseregularity.”Examples:Wetendtoaccept“That’sred”(ifitisactuallyuttered)inthepresenceofaredthing;weaccept“pandq”iffweacceptpandweacceptq.(“Accepting”asentenceissupposedtobeapsychologicalnotion(pp.94–6),ratherthanarecognizableformofactualsocialbehavior;thisisanotherdeparturefromWittgensteinandfromBrandom.)Compositionalityisbrieflyaddressedinchapter7:Acomplexexpression’smeaningpropertyconsistsina“constructionproperty,”e.g.:xmeansDOGSBARKinthat“xresultsfromputtingtermswhosemeaningsareDOGandBARK,inthatorder,intoaschemawhosemeaningisNSV”(p.156).But,unlessIhavemissedit,nothingissaidastohowthe“schema”issupposedtohavea“useregularity”despitenotbeinganexpressionofEnglish.Chapter71DonotmissKingsleyAmis’taleofthiswordinTheKing’sEnglish(London:HarperCollins,1998:118–19).Amisswearshehasseenthewordmisspelledas“jejeune”andevenpronouncedinpseudo-Frenchas“zherzherne.”Cometothinkofit,donotmisstherestofAmis’bookeither.2ThereisatendencyintheGriceanliteraturetoassumethatspeaker-meaningisunique,thatagivenutterancehasbutasinglespeaker-meaning.Thatassumptionisfalse;wearecomplexcommunicators,andwesometimesmeanmorethanonethingatatimebyutteringthesentencethatwedo.PerhapsImeanwhatthe\n196Notessentencemeansandalsosomefurtherconveyedmeaning.Orifyouaregoodatpuns,yoursentencemaybeitselfambiguousandyouintendbothmeaningsatonce.Shakespearecouldmeanasmanyasfivedifferentthingsinasingleutterance.3Itdidnotcomechronologicallyfirst,butwaspresentedinGrice(1968).4Inparticular,letusconfinethediscussiontodeclarativesentences,thoughGricewascarefultoaddressimperativesandothersaswell.5Yes,ithas.Canyoubelievethat?6Onepossiblefix,suggestedtomebythelateWendyNankas,istotalknotjustofactivationbutofreinforcement.7Ziff’scaseisstronglysimilartoanexampleofJ.O.Urmson’sregardingthumbscrews,discussedbyGrice(1969:152–3).Inresponse,Griceofferedwhathelabels“RedefinitionI”;butIhaveneverseenexactlyhowthatredefinitionwassupposedtoruleoutthiskindofcounterexample.ThereisasetofexamplesbeguninconversationbyDennisStampe,StephenSchiffer,andP.F.Strawson,involvingdeceitandsecond-guessingofacertainkind.Stampe’sversionwasthefirsttobeaddressedbyGrice(1969).Thecounterexamplesandresponsesleadtoanindefiniteregressofparticularlyconvolutedcasesandrevisionsoftheanalysis.Idoubtthatyouwouldthankmefordraggingyouthrougheventhesecondexampleintheregress.(Youmighteventrytoreturnthisbookandgetafractionofyourmoneyback.)SoIshallnotevenexpoundthefirst.8ThisistheopeninglineofasonglyricthatappearsinGoethe’snovelWilhelmMeistersLehrjahre(1795–6),book3,chapter1.9Schiffer(1972:chapters5–6)pursuedadifferentmethod,employingLewis’theoryofconventions.Chapter81Theforegoingexamplesareskepticalhypothesesofakindthateveryphilosophicaltraditionhastakenseriously;thepositivistshadtoworkhardtoarguethatthose“hypotheses”aremeaninglesseventhoughthesentenceslookperfectlymeaningfulatfirstglance.ThepositivistshadlesspatienceandlesstroublewiththeHegelianidealismofthelatenineteenthcentury,asin“TheAbsoluteisperfect,”andwithHeideggerianexistentialism,asin“TheNothingnoths”(“DasNichtsnichtet”).Ioncereceivedabrochure,advertisinganewlypublishedphilosophybook.Thebrochurecontainedabulletedlistofthebook’sspecialfeatures.Andoneofthebulleteditemswas:“Elevennewwaysinwhichnegationnegatesitself.”IswearIamnotmakingthisup.2Ofcourse,therearedegreesofunderstanding.Wemaynotunderstandatermcompletely.(Doyouknowexactlywhatacamshaftis?Howaboutalinearaccelerator?)Buttounderstandasentenceeveninpart,wehavetohavesomeideaofwhatitsays.Butagain,thatimpliesthattherealreadyissomethingthatitsayspriortoanythingbeingdeterminedaboutitsverificationconditions.3VerificationismflirtswithwhatthelateDavidStove(1991)calledthe“IshmaelEffect,”thephenomenonofaphilosophicaltheory’smakingasoleexceptionofitself.(ThereferenceistoMobyDick:“AndIonlyamescapedalonetotellthee”;actuallythisisitselfaquotationfromJob1:15.)Forexample:“Allwecanknowisthatwecanknownothing.”“Theonlymoralsinisintolerance.”“Absolutelyeverythingisrelative.”4SeeAchinstein(1965)andChurchland(1988).Butonthesecondpointtherehasbeensomedissent,suchasFodor(1988).5ActuallyagoodthoroughgoingQuineanshouldnotacceptthisargument.Whynot?(Hint:seethepreviousparagraph.)\nNotes197Chapter91HerehefollowsZiff(1960).2“Words”isnotquiteright.Somemeaningatomsaresmallerthanwords:affixessuchas“un-”(prefix)and“-able”(suffix).Somewordsareonlypleonasticpartsofmeaningatoms,asintheFrench“ne...pas.”Linguistscalltruemeaningatomsmorphemes.ButforconvenienceandfamiliarityIshallcontinuetospeakof“words.”3Thecompositionalitythesisissimplyassumedbymosttheorists,butitishardtoformulateprecisely,andithasbeenseriouslyquestioned,asbyPelletier(1994);seealsoSzabó(2007).4Moresophisticatedpresent-daydescendantsoftheemotivistsincludeBlackburn(1984,1993)andGibbard(1990);buttheytrytofindwaysofgrantingthatmoraljudgmentscanbecalled“true”or“false”andfigureinT-sentences,withoutgrantingthatmoraljudgmentsstatefactsabouttheworld.5SeeLycan(1984:chapter3).IshouldconfessthatthatworkisaglobaldefenseoftheTruth-ConditionTheory.Ibelievethetheoryiscorrectandworthpayinghighpricestohearliveinconcert.6Thereisalsoanastyproblemaboutambiguoussentences;seeParsons(1973)andLycan(1984:chapter3).7Ihavebeentoldbyatleastonebiologistthatthetwodonotapplytothesamethings;thereareanimalswithheartsbutnokidneys,ortheotherwayaround.Butignorethissqualidfactandpretendthat“renate”and“cordate”doapplytoexactlythesameanimals.Chapter101Montague(1960)builtupastructureofsuchhigher-(andhigher-)orderintensionscorrespondingtomoreandmoreabstractpartsofspeech.Infact,outofadesiretoone-upQuine,Montagueexplicitlyassignedveryrarefiedindividualintensionsto“sake,”“behalf,”and“dint.”AsImentionedinchapter1,inthiswayhemeantalsotostrikeablowonbehalfoftheReferentialTheory.(Butitisatbestaglancingblow:thewordsarenottakenasdenotingtheirintensionsasiftheywerepropernames.)2SeeagainLewis(1986)andLycan(1994).3See,forexample,HeilandMele(1993).Chapter111Thedistinctionsdevelopedherearefairlytraditional.Butrecentlytherehasbeensomedisputeoverthebestway(s)inwhichtodrawthesemantics–pragmaticsdistinctioninparticular.See,forexample,Bach(2002)andBezuidenhout(2002).2Itisfurtherdevelopedinchapter3ofLycan(1984).3ThiswasoncepointedouttomebyPetervanInwagen.4Thistreatmentoftenseisafudge,forconvenience;forafullertreatmentoftense,seeLycan(1984:55–62).5Thispresupposesthatdeicticelementsaresomehowmarkedassuchinlogicalform.6Thereisamoreseriousobjectiontoitaswell,pointedoutbyBurge(1974).\n198NotesChapter121Austintookpropositionalcontentnearlyforgranted.Hestronglyopposedentitytheories,soby“locutionarycontent”hedidnotmeananythingaboutpropositionsasthings.Hemerelygesturedtowards“senseandreference,”alludingtoFregebutevidentlynotusing“sense”tomeanakindoftheoreticalentity.Austinwasoffhandaboutpropositionalcontentbecausehisfocuswasontheotherthing,illocutionaryforce,whichvariesindependently.2InarecentKudzucomicstrip,thepreacherWillB.DunnresiststheurgingofaliberalparishionerthattheTenCommandmentsberenamed“TheTenSuggestions.”Strawson(1964),Schiffer(1972),andBachandHarnish(1979)argueconvincinglythatnotallillocutionaryforceisaspurelyconventionalasthatof(1)–(7),thecomparatively“pure”performativeswithwhichwebegan.Someforce,thatofbeinganadvisingorbeingaquestion-asking,say,ismoreamatterofGriceanspeaker-intentions.3Moreover,thereisevidencethatsomesemanticalphenomenacannotbeexplainedexceptbyreferencetoillocutionaryfactors(seeBarker1995,2004).4Justtosaveyoulooking:ShewaschristenedbyIsabelHoey,daughterofthethenGovernorofNorthCarolina.IamtoldthatHoeydidusethetraditionalbottleofchampagne,whileabandplayed“AnchorsAweigh.”5LinguistshaveoftenmissedthefactthatAustinused“infelicitous”asthebroadestpossibleumbrellaterm.Theysometimesusethewordasapplyingtosentences,andneologisticallymeansomethinglike“nonsyntacticallyandnonsemanticallybutpragmaticallydefective[inawaythatissupposedtobefairlyspecificbutisneverspecified].”Chapter131Notice,bytheway,thatifyouaskmeaquestionandIthenproduceanutterance,youautomaticallyassumethatmyutteranceisintendedasananswertoyourquestion.Supposeyouask,“Whyareyoulateforclass?,”andIsay,“Wehadspaghettiatourhouselastnight.”Youwouldthink:What?Howdoeshavingspaghettiimpedegettingtoclassonthefollowingday?Wasittainted?Ifyoudidstartthinkingalongthoselines,noticethatyouwouldsimplyhaveassumedthatIwascooperatingbyproducingananswertothequestionyouasked.Onethingyouwillgraduallylearn,themoreyouthinkaboutconversationalcooperation,ishowpatheticallyeasyitisforatrainedlinguistorphilosopheroflanguagetomislead,deceive,cheat,andhoodwinkotherpeoplewithouteversayinganythingfalse.Advertisingcopywritersandpoliticiansaremastersofconversationalimplicature,havingworkeditoutinstinctively,becauseitallowsthemtoconveyfalsehoodswithoutbreakingthelawbyutteringfalsesentences.2Griceaddsathirdfeature,“detachability”.Sincehisformofconversationalreasoningproceedsfromthepropositionalcontentofthesentenceuttered,anylogicalequivalentofthesentenceshouldgeneratethesameimplicaturesinthesamecontext.Butthereareobviousobjectionstothis,aswhentheruleagainstprolixityisexploited.3Gricedistinguishes(p.73)between“particularized”implicature,onessuchas(1a)or(1b),whichmustbeworkedoutonthespotfromthecontextualfacts,and“generalized”implicature,suchas(2),(3)or(4a),which“isnormallycarriedbysayingthatp.”Noonedisputestheexistenceofsuchadistinction,butitsbasisishighlycontroversial(Bach1994a;Levinson2000;Carston2002;Horn2004;Recanati2004).\nNotes1994PhilosophersimpressedbytheConnectionistapproachinArtificialIntelligencewillbeespeciallysuspicious,nottosayderisive.Butsuchphilosophersarealsoskepticalofsyntax,andofsemanticsconceivedasexplaininganythingabouthumanbeings.5OfcoursethisformulationisunfaithfultoStrawson’soriginalintent,becausehedidnotwantustoreplaceourtwo-valuedlogicforsentenceswithathree-valuedlogicforsentences;itwasnotthatinsteadofhavingonlythesetwopossibletruthvalues,“true”and“false,”asentencenowcanhaveathirdone,“zip”or“neuter.”Strawson’spointwasthatitisnotsentencesthathavetruth-valuesatall.6There;wasthatworththepriceofthisbook,orwhat?7Ironically,ImyselfdoubtthatGrice’sownsentenceisanexampleofconventionalimplicature,becauseIbelievetheimplicationiscarriedbythesemanticmeaningoftheword“therefore.”“Therefore”means“forthatreason.”AndsoGrice’ssentenceissynonymouswith“HeisanEnglishman,andforthatreason,heisbrave,”whichinmyspeechentailsthathisbeinganEnglishmanisareasonorgroundofhisbeingbrave.Fortunately,thereareplentyofbetterexamples.8Everyoneisalwaysexaggerating!9Stanley(2000)arguesinsteadthatmanyofthemarecasesofsaturation.Chapter141“ThouartPeter,anduponthisrockIwillbuildmychurch”(Matthew16:18).2SaidbyRomeo,afterasking(tendentiously)whatlightthroughyonderwindowbroke.Julietalso“hangsuponthecheekofnight/LikearichjewelinanEthiop’sear,”butthatwasbackinActIandIhopehethoughtbetterofit.3RichardIII,I.i.1.4Hamlet,I.iii.116–17.ButthemostelaboratelymixedofShakespeare’smetaphorsknowntomeispatriotic:Thisroyalthroneofkings,thissceptredisle,Thisearthofmajesty,thisseatofMars,ThisotherEden,demi-paradise,ThisfortressbuiltbyNatureforherselfAgainstinfectionandthehandofwar,Thishappybreedofmen,thislittleworld,Thispreciousstonesetinthesilversea,Whichservesitintheofficeofawall,Orasamoatdefensivetoahouse,Againsttheenvyoflesshappierlands,Thisblessedplot,thisearth,thisrealm,thisEngland,Thisnurse,thisteemingwombofroyalkings,Fearedbytheirbreedandfamousbytheirbirth...(RichardII,II.i.40–52)Hegotawaywithit.5Andbyearlierempiricistsaswell.Blackburn(1984:172)offersapungentquotefromHobbes’Leviathan.6IowethisobservationtoFranklinGoldsmith.7HereandelsewherehedrawsonTversky(1977).8Ifyouwantgreed,trycats.Butnooneevercallssomeoneacatasametaphoricalwayofsayingthatthatpersonisgreedy.\n200NotesAfurtherexampleis“bastard.”Iknowofnoevidencethatamalepersonwhoseparentswerenotmarriedwhenhewasbornisanymorelikelytobecallousorunscrupulousthanisanyoneelse.9Fogelinaddressesthisobjection(pp.44–5),butIthinkweakly.Hecomplainsthat“gorilla”isnotametaphorbutadeadmetaphor;ifso,thatseemsinessentialtotheexample.Thenhesuggeststhateithertheellipsisislargerthanusual,including“whatmostpeoplethink—sarelike,”orthespeaker“speaksfromtheperspectiveofcommonbeliefwhichheandhislistenerknowcontainsfalsebeliefstheydonotshare.”Thefirstofthelattermovesissemanticallydesperate;thesecond,intheabsenceofsomeindependentmotivation,isadhoc.10Accordingtocummings’poem“anyonelivedinaprettyhowtown”(citedinchapter1),itsprotagonist[A]nyone“sanghisdidn’thedancedhisdid”(line4).11Searlehimselfreservestheterm“indirect”fortypesofcommunication,suchasindirectforceandsomeconversationalimplicature,inwhichoneconveysasecondmeaninginadditiontomeaningwhatone’ssentencesays.12Areal-lifeexample,Iamsorrytosay,broughtupbyHansduringaseminar.13Thisideaisactuallyveryold;Aristotleexploredit,anditwasvigorouslyelaboratedbythemedievalphilosophers.\nGlossaryAnalyticAsentenceisanalyticifitstruth,howevertrivial,isguaranteedbythecollectivemeaningsofthewordsthatoccurinit.AnaphoricexpressionAnanaphoricexpressioninheritsitsmeaningfromanotherexpression,itsantecedent,whichusuallyoccursearlierinthesentenceorinaprevioussentence.Antecedent(ofananaphoricTheantecedentofananaphoricexpression)expressionistheexpressionfromwhichtheanaphoricexpressioninheritsitsmeaning.Compositionally,Acharacteristicofhowweunderstandcompositionalitynovelsentences,namelythatweunderstandsuchsentencesinvirtueofunderstandingtheindividualwordsandhowtheyarestrungtogether;presupposesthatthemeaningofasentenceisdeterminedbythemeaningsofitscomponentwordstogetherwiththeirsyntacticrelationstoeachother.ContextofutteranceThesettinginwhichapieceoflanguageisusedbyaspeaker.ContextualdefinitionAtypeofdefinition,tobecontrastedwithexplicitdefinition,wherebyoneexhibitstheroleplayedbythewordtobedefinedbyshowinghowonecanparaphrasewholesentencesinwhichthewordoccurs.\n202GlossaryConventionalimplicatureConventionalimplicatureisimplicature,inthataspeakerimplicatessomethingratherthanactuallysayingit,butitdiffersfromconversationalimplicatureinthatconventionalimplicaturesaregraspedimmediately,notonthebasisofreasoning.Normallytheyarecarriedbytendentiouschoicesofparticularwords.ConversationalimplicatureConversationalimplicatureisimplicaturethatoneworksout,orcouldworkout,usingreasoningbasedonsomethinglikeGrice’sconversationalmaxims.DeicticAdeicticelementisonewhosesemanticinterpretationvarieswiththecontextofutterance,suchasatensemarkerorademonstrativepronoun.Deicticelementsarealsocalledindexicals.DescriptiontheoryofproperThethesisthatnamesareequivalentinnamesmeaningtodescriptions.DomainTheclassofthingsoverwhichaquantifierranges.ExtensionAterm’sextensionistheclassofthingstowhichthetermapplies;theextensionof“red”istheclassofredthings.FlacciddesignatorAsingulartermthatdesignatesdifferentthingsindifferentpossibleworlds.FreeenrichmentThecontextualfillinginofaconstituentofwhatissaideventhoughthereisnoevencovertholeinlogicalformorotherspecificallylinguisticcontrolforthatconstituent.GeneraltermsTerms,suchas“dog”and“brown”,thataremeanttoapplytomorethanonething.IdeationaltheoriesTheoriesthatholdthatmeaningsarementalentities.IdentitystatementAnidentitystatementcontainstwosingularterms.Ifthestatementistrue,bothitstermspickoutordenotethesamepersonorthing.IdiolectThepersonalanddistinctivespeechofaparticularindividual.\nGlossary203IntensionalisomorphismThereisanintensionalisomorphismbetweentwosentenceswhenthesentenceshavethesameintensionandhaveitinvirtueofbeingcomposedinthesameway(ormuchthesameway)outofthesameatomicintensions.IntensionallogicAformalsystemdelineatingthelogicofFregeansenses.IntensionalsentencesSentencesinwhichcoextensivetermscannotbesubstitutedwithoutpossiblychangingthetruth-valueofthesentencesthemselves.LexicalmeaningThemeaningofawordorshortphrase,ascontrastedwithsentencemeaning.ModalConcerningpossibilityandnecessity.NaturalkindtermsCommonnouns,suchas“gold”and“tiger,”thatrefertonaturalsubstancesororganisms.PossibleworldAworld,oruniverse,whichmighthavebeentheactualworld.QuantifiersWords,suchas“all”or“some,”thatquantifygeneralterms.ReferentialTheoryofLinguisticThistheoryattemptstoexplaintheMeaningsignificanceormeaningofalllinguisticexpressionsintermsoftheirhavingbeenconventionallyassociatedwiththingsintheworld,andattemptstoexplainahumanbeing’sunderstandingofasentenceintermsofthatperson’sknowingwhatthesentence’scomponentwordsreferto.RestrictedquantificationAfeatureofmostquantifiedstatementswherebythedomainoverwhichthequantifiersrangeisnottheentireuniverse.Thedomainisrestrictedinsomewaytypicallyindicatedbycontext.RigiddesignatorAsingulartermthatdenotesthesamethingineverypossibleworld(morestrictly,ineverypossibleworldinwhichthatthingexists).\n204GlossarySaturationThecontextualplugginginofanappropriatevaluetoanunderlyingpositioninlogicalform.SemanticpresuppositionAsentenceS1semanticallypresupposesasentenceS2onlyinthecasethat,ifS2isfalse,thennecessarilyS1lackstruth-value.“Sense”(Frege)Thesortofmeaningalinguisticexpressionhasoverandaboveits(putative)referent.SemanticreferentTheindividual,ifoneexists,thatadescriptionpurportstopickoutinvirtueoftheindividual’sfittingthedescription;alsocalledthesemanticdenotatum.Speaker-meaningWhataspeakermeansinutteringasentence;alsocalled,byGrice,“utterer’smeaning.”Speaker-referenceTheobject,ifany,towhichthespeakerwhousesadescriptionintendstocalltotheattentionofher/hisaudience.TruthconditionTheconditionunderwhichasentencewouldbetrue.Truth-functionalAconnectiveistruth-functionalifthetruthvalueofcompoundsentencescontainingitisstrictlydeterminedbythetruthvaluesofthecomponentsentences.Forexample,“and”isatruth-functionalconnectivebecausethetruthvalueofsentencesoftheform“AandB”isstrictlydeterminedbythetruthvaluesof“A”and“B”respectively.\nBibliographyNote:Datesgivenintheform“1954/1962”referfirsttothedateoforiginalpublicationfollowedbythemoreaccessiblereprintingactuallycited.Abbott,B.(2003)“AReplytoSzabó’s‘DescriptionsandUniqueness’.”PhilosophicalStudies103:221–9.Achinstein,P.(1965)“TheProblemofTheoreticalTerms.”AmericanPhilosophicalQuarterly2:193–203.Ackerman,D.(1979)“ProperNames,PrepositionalAttitudesandNon-DescriptiveConnotations.”PhilosophicalStudies35:55–69.Adams,E.(1965)“TheLogicofConditionals.”Inquiry8:166–89.Almog,J.,Perry,J.andWettstein,H.(eds.)(1989)ThemesfromKaplan.NewYork:OxfordUniversityPress.Alston,W.(1963)“MeaningandUse.”PhilosophicalQuarterly51:107–24.——(2000)IllocutionaryActsandSentenceMeaning.Ithaca,NY:CornellUniversityPress.Austin,J.L.(1961)“PerformativeUtterances.”InJ.L.Austin,PhilosophicalPapers.Oxford:OxfordUniversityPress.(1962)HowToDoThingsWithWords.Oxford:ClarendonPress.Avramides,A.(1989)MeaningandMind.Cambridge,MA:MITPress.Ayer,A.J.(1946)Language,TruthandLogic,2ndedn.London:VictorGollancz.Bach,K.(1987)ThoughtandReference,Oxford:OxfordUniversityPress.——(1994a)“SemanticSlack:WhatisSaidandMore.”InS.L.Tsohatzidis(ed.),FoundationsofSpeechActTheory:PhilosophicalandLinguisticPerspectives,London:Routledge.——(1994b)“ConversationalImpliciture.”MindandLanguage9:124–61.——(1999)“TheMythofConventionalImplicature.”LinguisticsandPhilosophy22:327–66.——(2002)“Semantic,Pragmatic.”InJ.K.Campbell,M.O’Rourke,andD.Shier(eds.),MeaningandTruth:InvestigationsinPhilosophicalSemantics,NewYork:SevenBridgesPress.Bach,K.andHarnish,R.M.(1979)LinguisticCommunicationandSpeechActs.Cambridge,MA:MITPress.Barker,S.(1995)“TowardsaPragmaticTheoryof‘If’.”PhilosophicalStudies78:185–211.——(2004)RenewingMeaning.Oxford:OxfordUniversityPress.Bar-On,D.(1992)“SemanticVerificationism,LinguisticBehaviorism,andTranslation.”PhilosophicalStudies66:235–59.\n206BibliographyBeardsley,M.(1962)“TheMetaphoricalTwist.”PhilosophyandPhenomenologicalResearch22:293–307.——(1967)“Metaphor.”InP.Edwards(ed.),TheEncyclopediaofPhilosophy,vol.5.NewYork:Macmillan.Bennett,J.(1971)Locke,Berkeley,Hume:CentralThemes.Oxford:ClarendonPress.——(1976)LinguisticBehaviour.Cambridge:CambridgeUniversityPress.Bertolet,R.(1980)“TheSemanticSignificanceofDonnellan’sDistinction.”PhilosophicalStudies37:281–8.Bezuidenhout,A.(2002)“RadicalPragmatics.”InJ.K.Campbell,M.O’Rourke,andD.Shier(eds.),MeaningandTruth:InvestigationsinPhilosophicalSemantics.NewYork:SevenBridgesPress.Bezuidenhout,A.,andReimer,M.(eds.)(2004)DescriptionsandBeyond.Oxford:OxfordUniversityPress.Biro,J.(1979)“IntentionalismintheTheoryofMeaning.”Monist62:238–58.Black,M.(1954/1962)“Metaphor.”InM.Black,ModelsandMetaphors.Ithaca,NY:CornellUniversityPress.——(1973)“MeaningandIntention:AnExaminationofGrice’sViews.”NewLiteraryHistory4:257–79.Black,M.andGeach,P.(eds.)(1952)TranslationsfromthePhilosophicalWritingsofGottlobFrege.Oxford:BasilBlackwell.Blackburn,S.(1984)SpreadingtheWord.Oxford:ClarendonPress.——(1993)EssaysinQuasi-Realism.Oxford:ClarendonPress.Blakemore,D.(1987)SemanticConstraintsonRelevance.Oxford:BasilBlackwell.——(1992)UnderstandingUtterances.Oxford:BasilBlackwell.Boër,S.(1978)“AttributiveNames.”NotreDamejournalofFormalLogic19:177–85.——(1985)“SubstanceandKind:ReflectionsontheNewTheoryofReference.”InB.K.MatilalandJ.L.Shaw(eds.),AnalyticalPhilosophyinComparativePerspective.Dordrecht:D.Reidel.Bradley,F.H.(1930)AppearanceandReality.Oxford:ClarendonPress.Brandom,R.(1994)MakingItExplicit.Cambridge,MA:HarvardUniversityPress.——(2000)ArticulatingReasons.Cambridge,MA:HarvardUniversityPress.Braun,D.(2005)“EmptyNames,FictionalNames,MythicalNames.”Noûs39:596–631.Brogaard,B.(2007)“ThebutnotAll:ANewAccountofPluralDefiniteDescriptions.”MindandLanguage22:402–26.Burge,T.(1973)“ReferenceandProperNames.”journalofPhilosophy70:425–39.ReprintedinD.DavidsonandG.Harman(eds.)(1975)SemanticsofNaturalLanguage.Dordrecht:D.Reidel.——(1974)“DemonstrativeConstructions,ReferenceandTruth.”journalofPhilosophy71:205–23.——(1979)“IndividualismandtheMental.”InP.French,T.E.Uehling,andH.Wettstein(eds.),MidwestStudiesinPhilosophyIV:StudiesinMetaphysics.Minneapolis:UniversityofMinnesotaPress.Campbell,J.K.,O’Rourke,M.,andShier,D.(eds.)(2002)MeaningandTruth:InvestigationsinPhilosophicalSemantics.NewYork:SevenBridgesPress.Carnap,R.(1947/1956)MeaningandNecessity,2ndedn.Chicago:UniversityofChicagoPress.\nBibliography207Carnie,A.(2001)Syntax.Oxford:BasilBlackwell.Carston,R.(1988)“Implicature,Explicature,andTruth-TheoreticSemantics.”InR.Kempson(ed.),MentalRepresentation.Cambridge:CambridgeUniversityPress.ReprintedinS.Davis(ed.)(1991)Pragmatics:AReader.Oxford:OxfordUniversityPress.——(2002)ThoughtsandUtterances:ThePragmaticsofExplicitCommunication.Oxford:BasilBlackwell.Cartwright,R.(1962)“Propositions.”InR.J.Butler(ed.),AnalyticPhilosophy,vol.1.Oxford:BasilBlackwell.——(1987)“OntheOriginsofRussell’sTheoryofDescriptions.”InR.Cartwright,PhilosophicalEssays.Cambridge,MA:MITPress.Chierchia,G.andMcConnell-Ginet,S.(1990)MeaningandGrammar:AnIntroductiontoSemantics.Cambridge,MA:MITPress.Chomsky,N.(1957)SyntacticStructures.TheHague:Mouton&Co.——(1965)AspectsoftheTheoryofSyntax.Cambridge,MA:MITPress.Churchland,P.M.(1988)“PerceptualPlasticityandTheoreticalNeutrality.”PhilosophyofScience55:167–87.Cohen,L.J.(1964)“DoIllocutionaryForcesExist?”PhilosophicalQuarterly14:118–37.——(1971)“SomeRemarksaboutGrice’sViewsabouttheLogicalParticlesofNaturalLanguage.”InY.Bar-Hillel(ed.),PragmaticsofNaturalLanguages.Dordrecht:D.Reidel.——(1985)“AProblemaboutAmbiguityinTruth-TheoreticSemantics.”Analysis45:129–34.Cohen,T.(1975)“FigurativeSpeechandFigurativeActs.”journalofPhilosophy71:669–84.Cole,P.(ed.)(1978)SyntaxandSemantics,Vol.9:Pragmatics.NewYork:AcademicPress.Cole,P.andMorgan,J.L.(eds)(1975)SyntaxandSemantics,Vol.3:SpeechActs.NewYork:AcademicPress.Cooper,D.E.(1986)Metaphor.Oxford:BasilBlackwell.Cresswell,M.J.(1973)LogicsandLanguages.London:Methuen.Culicover,P.(1997)PrinciplesandParameters:AnIntroductiontoSyntacticTheory.Oxford:OxfordUniversityPress.Davidson,D.(1967a)“TruthandMeaning.”Synthese17:304–23.ReprintedinD.Davidson(1984)InquiriesintoTruthandInterpretation.Oxford:ClarendonPress.——(1967b)“TheLogicalFormofActionSentences.”InN.Rescher(ed.),TheLogicofDecisionandAction.Pittsburgh:UniversityofPittsburghPress.——(1968)“OnSayingThat.”Synthese19:130–46.ReprintedinD.DavidsonandG.Harman(eds.)(1975)TheLogicofGrammar.Encino,CA:Dickenson.AlsoreprintedinD.Davidson(1984)InquiriesintoTruthandInterpretation.Oxford:ClarendonPress.——(1970/1975)“SemanticsforNaturalLanguages.”InD.DavidsonandG.Harman(eds.)(1975)TheLogicofGrammar.Encino,CA:Dickenson.ReprintedinD.Davidson(1984)InquiriesintoTruthandInterpretation.Oxford:ClarendonPress.——(1978)“WhatMetaphorsMean.”InS.Sacks(ed)OnMetaphor.Chicago:UniversityofChicagoPress.ReprintedinD.Davidson(1984)InquiriesintoTruthandInterpretation.Oxford:ClarendonPress.\n208Bibliography——(1984)InquiriesintoTruthandInterpretation.Oxford:ClarendonPress.——(1986)“ANiceDerangementofEpitaphs.”InE.LePore,TruthandInterpretation:PerspectivesofthePhilosophyofDonaldDavidson.Oxford:BasilBlackwell.Davidson,D.andHarman,G.(eds.)(1972)SemanticsofNaturalLanguage.Dordrecht:D.Reidel.——(eds.)(1975)TheLogicofGrammar.Encino,CA:Dickenson.Davis,S.(1991)Pragmatics:AReader.Oxford:OxfordUniversityPress.Davis,W.(1998)Implicature.Cambridge:CambridgeUniversityPress.Devitt,M.(1981a)Designation.NewYork:ColumbiaUniversityPress.——(1981b)“Donnellan’sDistinction.”InP.French,T.Uehling,andH.Wettstein(eds.),MidwestStudiesinPhilosophyVI:TheFoundationsofAnalyticPhilosophy.Minneapolis:UniversityofMinnesotaPress.——(1983)“Dummett’sAnti-Realism.”journalofPhilosophy80:73–99.——(1989)“AgainstDirectReference.”MidwestStudiesinPhilosophy14:206–40.——(1996)ComingtoOurSenses.Cambridge:CambridgeUniversityPress.Devitt,M.andSterelny,K.(1987)LanguageandReality:AnIntroductiontothePhilosophyofLanguage.Cambridge:MITPress.Donnellan,K.(1966)“ReferenceandDefiniteDescriptions.”PhilosophicalReview75:281–304.——(1968)“PuttingHumptyDumptyTogetherAgain.”PhilosophicalReview77:203–15.——(1970)“ProperNamesandIdentifyingDescriptions.”Synthese21:335–58.ReprintedinD.DavidsonandG.Harman(eds.)(1972)SemanticsofNaturalLanguage.Dordrecht:D.Reidel.——(1974)“SpeakingofNothing.”PhilosophicalReview83:3–31.——(1979)“SpeakerReference,Descriptions,andAnaphora.”InP.French,T.UehlingandH.Wettstein(eds.),ContemporaryPerspectivesinthePhilosophyofLanguage.Minneapolis:UniversityofMinnesotaPress.Duhem,P.(1906/1954)TheAimandStructureofPhysicalTheory,trans.P.Wiener.Princeton,NJ:PrincetonUniversityPress.Dummett,M.(1973)Frege:PhilosophyofLanguage.NewYork:Harper&Row.——(1975)“WhatIsaTheoryofMeaning?”InS.Guttenplan(ed.),MindandLanguage.Oxford:OxfordUniversityPress.——(1978)TruthandOtherEnigmas,Cambridge,MA:HarvardUniversityPress.Dupré,J.(1981)“NaturalKindsandBiologicalTaxa.”PhilosophicalReview90:66–90.Elgin,C.andScheffler,I.(1987)“MainspringsofMetaphor.”journalofPhilosophy84:331–5.Erwin,E.(1970)TheConceptofMeaninglessness.Baltimore,MD:JohnsHopkinsUniversityPress.Erwin,E.,Kleiman,L.,andZemach,E.(1976)“TheHistoricalTheoryofReference.”AustralasianjournalofPhilosophy54:50–7.Evans,G.(1973)“TheCausalTheoryofNames.”AristotelianSocietySupplementaryVolume47:187–208.——(1977)“Pronouns,Quantifiers,andRelativeClauses(I).”CanadianjournalofPhilosophy7:467–536.——(1982)TheVarietiesofReference.Oxford:OxfordUniversityPress.Evans,G.andMcDowell,J.(eds)(1976)TruthandMeaning.Oxford:OxfordUniversityPress.\nBibliography209Everett,A.,andHofweber,T.(2000)EmptyNames,Fiction,andthePuzzlesofNon-Existence.Stanford,CA:CSLIPublications.Fillmore,C.(1975)TheSantaCruzLecturesonDeixis.Bloomington,IN:IndianaUniversityLinguisticsClubPublications.Fine,A.(1975)“HowtoCompareTheories:ReferenceandChange.”Noûs9:17–32.Fodor,J.A.(1988)“AReplytoChurchland’s‘PerceptualPlasticityandTheoreticalNeutrality’.”PhilosophyofScience55:188–98.Fogelin,R.(1988)FigurativelySpeaking.NewHaven,CT:YaleUniversityPress.Frege,G.(1892/1952a)“OnConceptandObject.”InM.BlackandP.Geach(eds.)(1952)TranslationsfromthePhilosophicalWritingsofGottlobFrege.Oxford:BasilBlackwell.——(1892/1952b)“OnSenseandReference.”InM.BlackandP.Geach(eds.)(1952)TranslationsfromthePhilosophicalWritingsofGottlobFrege.Oxford:BasilBlackwell.ReprintedinD.DavidsonandG.Harman(eds.)(1975)TheLogicofGrammar.Encino,CA:Dickenson.——(1918/1956)“TheThought.”Mind65:289–311.French,P.,Uehling,T.,andWettstein,H.(eds.)(1979)ContemporaryPerspectivesinthePhilosophyofLanguage.Minneapolis:UniversityofMinnesotaPress.Gazdar,G.(1979)Pragmatics:Implicature,Presupposition,andLogicalForm.NewYork:AcademicPress.Geach,P.(1962)ReferenceandGenerality.Ithaca,NY:CornellUniversityPress.Geis,M.andZwicky,A.(1971)“OnInvitedInferences.”LinguisticInquiry2:561–6.Gibbard,A.(1990)WiseChoices,AptFeelings.Cambridge,MA:HarvardUniversityPress.Ginet,C.(1979)“Performativity.”LinguisticsandPhilosophy3:245–65.Goodman,N.(1970)“SevenStricturesonSimilarity.”InL.FosterandJ.W.Swanson(eds.),ExperienceandTheory.Amherst,MA:UniversityofMassachusettsPress.——(1981)“TwistedTales:OrStory,Study,andSymphony.”Synthese46:331–50.Gordon,D.andLakoff,G.(1975)“ConversationalPostulates.”InP.ColeandJ.L.Morgan(eds.),SyntaxandSemantics,Vol.3:SpeechActs.NewYork:AcademicPress.Green,G.M.(1989)PragmaticsandNaturalLanguageUnderstanding.Hillsdale,NJ:LawrenceErlbaum.Grice,H.P.(1957)“Meaning.”PhilosophicalReview66:377–88.ReprintedinH.P.Grice(1989)StudiesintheWayofWords.Cambridge,MA:HarvardUniversityPress.——(1968)“Utterer’sMeaning,Sentence-Meaning,andWord-Meaning.”FoundationsofLanguage4:225–42.ReprintedinH.P.Grice(1989)StudiesintheWayofWords.Cambridge,MA:HarvardUniversityPress.——(1969)“Utterer’sMeaningandIntentions.”PhilosophicalReview78:147–77.ReprintedinH.P.Grice(1989)StudiesintheWayofWords.Cambridge,MA:HarvardUniversityPress.——(1975)“LogicandConversation.”InD.DavidsonandG.Harman(eds.),TheLogicofGrammar.Encino,CA:Dickenson.ReprintedinP.ColeandJ.L.Morgan(eds.)(1975)SyntaxandSemantics,Vol.3:SpeechActs.NewYork:AcademicPress(pagereferencesaretothelatter).AlsoreprintedinH.P.Grice(1989)StudiesintheWayofWords.Cambridge,MA:HarvardUniversityPress.\n210Bibliography——(1978)“FurtherNotesonLogicandConversation.”InP.Cole(ed.)(1978)SyntaxandSemantics,Vol.8:Pragmatics.NewYork:AcademicPress.ReprintedinH.P.Grice(1989)StudiesintheWayofWords.Cambridge,MA:HarvardUniversityPress.——(1989)StudiesintheWayofWords.Cambridge,MA:HarvardUniversityPress.Grover,D.,Camp,J.,andBelnap,N.(1975)“AProsententialTheoryofTruth.”PhilosophicalStudies27:73–125.Gunderson,K.(ed.)(1975)MinnesotaStudiesinthePhilosophyofScience,Vol.8:Language,Mind,andKnowledge.Minneapolis:UniversityofMinnesotaPress.Hallett,G.(1967)Wittgenstein’sDefinitionofMeaningasUse.NewYork:FordhamUniversityPress.Harman,G.(1967–8)“QuineonMeaningandExistence,I.”ReviewofMetaphysics31:124–51.——(1972)“LogicalForm.”FoundationsofLanguage9:38–65.ReprintedinD.DavidsonandG.Harman(eds.)(1975)TheLogicofGrammar.Encino,CA:Dickenson.——(1974a)“ReviewofStephenSchiffer’sMeaning.”journalofPhilosophy71:224–9.——(1974b)“MeaningandSemantics.”InM.MunitzandP.Unger(eds.),SemanticsandPhilosophy.NewYork:NewYorkUniversityPress.——(1975)“Language,Thought,andCommunication.”InK.Gunderson(ed.),MinnesotaStudiesinthePhilosophyofScience,Vol.8:Language,Mind,andKnowledge.Minneapolis:UniversityofMinnesotaPress.——(1982)“ConceptualRoleSemantics.”NotreDamejournalofFormalLogic23:242–56.Harnish,R.M.(1976)“LogicalFormandImplicature.”InT.Bever,J.Katz,andT.Langendoen(eds.),AnIntegratedTheoryofLinguisticAbility.NewYork:Crowell.ReprintedinS.Davis(1991)Pragmatics:AReader.Oxford:OxfordUniversityPress.Heil,J.(1998)ThePhilosophyofMind:AContemporaryIntroduction,London:Routledge.Heil,J.andMele,A.(eds.)(1993)MentalCausation.NewYork:ClarendonPress.Heim,I.(1990)“E-TypePronounsandDonkeyAnaphora.”LinguisticsandPhilosophy13:137–77.Hempel,C.G.(1950)“ProblemsandChangesintheEmpiricistCriterionofMeaning.”RevueInternationaledePhilosophie4:41–63.Hintikka,K.J.J.(1961)“ModalityandQuantification.”Theoria27:119–28.——(1976)“QuantifiersinLogicandQuantifiersinNaturalLanguages.”InS.Körner(ed.),PhilosophyofLogic.Oxford:BasilBlackwell.ReprintedinE.Saarinen(ed.)(1979)Game-TheoreticalSemantics.Dordrecht:D.Reidel.——(1979)“QuantifiersinNaturalLanguages:SomeLogicalProblems.”InE.Saarinen(ed.)(1979)Game-TheoreticalSemantics.Dordrecht:D.Reidel.Holdcroft,D.(1978)WordsandDeeds.Oxford:OxfordUniversityPress.Horn,L.R.(2004)“Implicature.”InL.R.HornandG.Ward(eds.),TheHandbookofPragmatics.Oxford:BlackwellPublishing.Horn,L.R.,andWard,G.(eds)(2004)TheHandbookofPragmatics.Oxford:BlackwellPublishing.\nBibliography211Hornstein,N.(1995)LogicalForm:FromGBtoMinimalism.Oxford:BasilBlackwell.Horwich,P.(1998)Meaning.Oxford:OxfordUniversityPress.Jackson,F.(1998)“ReferenceandDescriptionRevisited.”InJ.Tomberlin(ed.),PhilosophicalPerspectives,Vol.12:Language,MindandOntology.Atascadero,CA:RidgeviewPublishing.Johnson,M.(ed.)(1981)PhilosophicalPerspectivesonMetaphor.Minneapolis:UniversityofMinnesotaPress.Kamp,H.andReyle,U.(1993)FromDiscoursetoLogic.Dordrecht:KluwerAcademicPublishing.Kaplan,D.(1972)“WhatisRussell’sTheoryofDescriptions?”InD.F.Pears(ed.),BertrandRussell.GardenCity,NY:AnchorBooks.ReprintedinD.DavidsonandG.Harman(eds.)(1975)TheLogicofGrammar.Encino,CA:Dickenson.——(1975)“HowtoRussellaFrege-Church.”journalofPhilosophy72:716–29.——(1978)“Dthat.”InP.Cole(ed.),SyntaxandSemantics,Vol.9:Pragmatics.NewYork:AcademicPress.ReprintedinP.French,T.Uehling,andH.Wettstein(eds.)(1979)ContemporaryPerspectivesinthePhilosophyofLanguage.Minneapolis:UniversityofMinnesotaPress.——(1989)“Afterthoughts.”InJ.Almog,J.Perry,andH.Wettstein(eds.),ThemesfromKaplan.NewYork:OxfordUniversityPress.Karttunen,L.(1973)“PresuppositionsofCompoundSentences.”LinguisticInquiry4:169–93.Karttunen,L.andPeters,S.(1979)“ConventionalImplicature.”InC.OhandD.A.Dineen(eds.),SyntaxandSemantics,Vol.11:Presupposition.NewYork:AcademicPress.Kempson,R.(1975)PresuppositionandtheDelimitationofSemantics.Cambridge:CambridgeUniversityPress.Kenny,A.(1973)Wittgenstein.Cambridge,MA:HarvardUniversityPress.Kittay,E.(1987)Metaphor.Oxford:ClarendonPress.Kripke,S.(1972/1980)NamingandNecessity.Cambridge,MA:HarvardUniversityPress.AnearlierversionappearedinD.DavidsonandG.Harman(eds.)(1972)SemanticsofNaturalLanguage.Dordrecht:D.Reidel.——(1979a)“Speaker’sReferenceandSemanticReference.”InP.French,T.Uehling,andH.Wettstein(eds.),ContemporaryPerspectivesinthePhilosophyofLanguage.Minneapolis:UniversityofMinnesotaPress.——(1979b)“APuzzleaboutBelief.”InA.Margalit(ed.),MeaningandUse.Dordrecht:D.Reidel.——(1982)WittgensteinonRulesandPrivateLanguage.Cambridge,MA:HarvardUniversityPress.Kroon,F.(1987)“CausalDescriptivism.”AustralasianjournalofPhilosophy651–17.Kvart,I.(1993)“MediatedReferenceandProperNames.”Mind102:611–28.Lakoff,G.(1975)“PragmaticsinNaturalLanguage.”InE.L.Keenan(ed.),FormalSemanticsofNaturalLanguage.Cambridge:CambridgeUniversityPress.Lakoff,G.andJohnson,M.(1980)MetaphorsWeLiveBy.Chicago:UniversityofChicagoPress.Larson,R.andSegal,G.(1995)KnowledgeofMeaning.Cambridge,MA:BradfordBooks/MITPress.\n212BibliographyLasersohn,P.(1993)“ExistencePresuppositionsandBackgroundKnowledge.”journalofSemantics10:112–22.Lasnik,H.,andUriagereka,J.(withC.Boeckx)(2005)ACourseinMinimalistSyntax.Oxford:BlackwellPublishing.Lemmon,E.J.(1966)“Sentences,Statements,andPropositions.”InB.WilliamsandA.Montefiore(eds.),BritishAnalyticalPhilosophy.London:Routledge&KeganPaul.Lepore,E.(2004)“AnAbuseofContextinSemantics:TheCaseofIncompleteDefiniteDescriptions.”InA.BezuidenhoutandM.Reimer(eds.),DescriptionsandBeyond.Oxford:OxfordUniversityPress.Levinson,S.(1983)Pragmatics.Cambridge:CambridgeUniversityPress.——(2000)PresumptiveMeanings:TheTheoryofGeneralizedConversationalImplicature.Cambridge,MA:MITPress.Lewis,D.(1969)Convention:APhilosophicalStudy.Cambridge,MA:HarvardUniversityPress.——(1970)“GeneralSemantics.”Synthese22:18–67.ReprintedinD.DavidsonandG.Harman(eds)(1972)SemanticsofNaturalLanguage.Dordrecht:D.Reidel.——(1986)OnthePluralityofWorlds.Oxford:Blackwell.Linsky,L.(1963)“ReferenceandReferents.”InC.Caton(ed.),PhilosophyandOrdinaryLanguage.Urbana,IL:UniversityofIllinoisPress.——(1967)Referring.London:Routledge&KeganPaul.——(1977)NamesandDescriptions.Chicago:UniversityofChicagoPress.Loar,B.(1976)“TheSemanticsofSingularTerms.”PhilosophicalStudies30:353–77.Locke,J.(1690/1955)EssayConcerningHumanUnderstanding.Chicago:EncyclopaediaBritannica.Loux,M.(1998)Metaphysics:AContemporaryIntroduction.London:Routledge.Ludlow,P.(2007)“Descriptions.”InE.N.Zalta(ed.),TheStanfordEncyclopediaofPhilosophy.Availableathttp://plato.stanford.edu/entries/descriptions/(lastaccessed15November2007).Ludlow,P.,andG.Segal(2004)“OnaUnitarySemanticalAnalysisforDefiniteandIndefiniteDescriptions.”InA.BezuidenhoutandM.Reimer(eds.),DescriptionsandBeyond.Oxford:OxfordUniversityPress.Lycan,W.G.(1974)“CouldPropositionsExplainAnything?”CanadianjournalofPhilosophy3:427–34.——(1984)LogicalforminNaturalLanguage.Cambridge,MA:BradfordBooks/MITPress.——(1994)ModalityandMeaning.DordrechtandBoston,MA:KluwerAcademicPublishing.Lyons,J.(1977)Semantics,vol.1.,Cambridge:CambridgeUniversityPress.MacKay,A.F.(1968)“Mr.DonnellanandHumptyDumptyonReferring.”PhilosophicalReview77:197–202.——(1972)“ProfessorGrice’sTheoryofMeaning.”Mind81:57–66.McKinsey,M.(1976)“DividedReferenceinCausalTheoriesofNames.”PhilosophicalStudies30:235–42.——(1978)“NamesandIntentionality.”PhilosophicalReview87:171–200.Malcolm,N.(1958)LudwigWittgenstein:AMemoir.Oxford:OxfordUniversityPress.Marcus,R.B.(1960)“Extensionality.”Mind69:55–62.\nBibliography213——(1961)“ModalitiesandIntensionalLanguages.”Synthese13:303–22.——(1981)“AProposedSolutiontoaPuzzleaboutBelief.”MidwestStudiesinPhilosophy6:501–10.Marsh,R.(ed.)(1956)LogicandKnowledge.London:Allen&Unwin.Meinong,A.(1904/1960)“TheTheoryofObjects.”InR.M.Chisholm(ed.),RealismandtheBackgroundofPhenomenology.Glencoe,IL:FreePress.Mill,J.S.(1843/1973)ASystemofLogic.London:Longmans.Montague,R.(1960)“LogicalNecessity,PhysicalNecessity,EthicsandQuanitifiers.”Inquiry3:259–69.——(1968/1974)“Pragmatics.”InR.Montague(1974)FormalPhilosophy.NewHaven,CT:YaleUniversityPress.——(1970)“EnglishasaFormalLanguage.”InB.Visentinietal.(eds.),LinguagginellaSocietàenellaTecnica.Milan:EdizionidiComunità.——(1974)FormalPhilosophy,NewHaven,CT:YaleUniversityPress.Moore,G.E.(1953/1962)SomeMainProblemsofPhilosophy.NewYork:CollierBooks.Moran,R.(1997)“Metaphor.”InC.WrightandR.Hale(eds.),ACompaniontothePhilosophyofLanguage.Oxford:BasilBlackwell.Morgan,J.L.(1978)“TwoTypesofConventioninIndirectSpeechActs.”InP.Cole(ed.),SyntaxandSemantics,Vol.9:Pragmatics.NewYork:AcademicPress.Morris,C.(1938)FoundationsoftheTheoryofSigns.Chicago:UniversityofChicagoPress.Neale,S.(1990)Descriptions.Cambridge,MA:MITPress.Nunberg,G.(1993)“IndexicalityandDeixis.”LinguisticsandPhilosophy16:1–43.Ostertag,G.(1998)DefiniteDescriptions:AReader.Cambridge,MA:BradfordBooks/MITPress.Parsons,K.(1973)“AmbiguityandtheTruth-Definition.”Noûs7:379–93.Parsons,T.(1980)NonexistentObjects.NewHaven,CT:YaleUniversityPress.Peirce,C.S.(1878/1934)“HowtoMakeOurIdeasClear.”InC.HartshorneandP.Weiss(eds.),CollectedPapersofCharlesSandersPeirce,vol.5.Cambridge,MA:HarvardUniversityPress.Pelletier,F.J.(1994)“ThePrincipleofSemanticCompositionality.”Topoi13:11–24.Pitcher,G.(1964)ThePhilosophyofWittgenstein.EnglewoodCliffs,NJ:Prentice-Hall.Plantinga,A.(1978)“TheBoethianCompromise.”AmericanPhilosophicalQuarterly15:129–38.Platts,M.(1979)WaysofMeaning.London:Routledge&KeganPaul.——(ed.)(1980)Reference,TruthandReality.London:Routledge&KeganPaul.Putnam,H.(1975a)“TheMeaningof‘Meaning’.”InK.Gunderson(ed.),MinnesotaStudiesinthePhilosophyofScience,Vol.8:Language,Mind,andKnowledge.Minneapolis:UniversityofMinnesotaPress.——(1975b)Mind,LanguageandReality:PhilosophicalPapers,vol.2.Cambridge:CambridgeUniversityPress.——(1978)MeaningandtheMoralSciences,London:Routledge&KeganPaul.Quine,W.V.(1951)“TwoDogmasofEmpiricism.”PhilosophicalReview60:20–43.——(1953)FromaLogicalPointofView.Cambridge,MA:HarvardUniversityPress.——(1960)WordandObject.Cambridge,MA:MITPress.\n214Bibliography——(1969)OntologicalRelativityandOtherEssays.NewYork:ColumbiaUniversityPress.Radford,A.(1997)SyntacticTheoryandtheStructureofEnglish:AMinimalistApproach.Cambridge:CambridgeUniversityPress.Recanati,F.(1989)“ThePragmaticsofWhatisSaid.”MindandLanguage4:295–329.——(1993)DirectReference.Oxford:Blackwell.——(2001)“WhatIsSaid.”Synthese125:75–91.——(2004)LiteralMeaning,Cambridge:CambridgeUniversityPress.Reeves,A.(1974)“OnTruthandMeaning.”Noûs8:343–59.Reimer,M.(1992)“IncompleteDescriptions.”Erkenntnis37:347–63.——(2001)“DavidsononMetaphor.”MidwestStudiesinPhilosophy25:142–55.Rhees,R.(1959–60)“Wittgenstein’sBuilders.”ProceedingsoftheAristotelianSociety60:171–86.Rosenberg,J.F.(1974)LinguisticRepresentation.Dordrecht:D.Reidel.——(1994)BeyondFormalism.Philadelphia:TempleUniversityPress.Ross,J.F.(1981)PortrayingAnalogy.NewYork:CambridgeUniversityPress.Routley,R.(1980)ExploringMeinong’sjungleandBeyond.Canberra:DepartmentalMonograph3,PhilosophyDepartment,ResearchSchoolofSocialSciences,AustralianNationalUniversity.Russell,B.(1905/1956)“OnDenoting.”Mind14:479–93.ReprintedinR.Marsh(ed.)(1956)LogicandKnowledge.London:Allen&Unwin.AlsoreprintedinD.DavidsonandG.Harman(eds.)(1975)TheLogicofGrammar.Encino,CA:Dickenson.——(1918/1956)“ThePhilosophyofLogicalAtomism.”InR.Marsh(ed.)(1956)LogicandKnowledge.London:Allen&Unwin.——(1919/1956)“OnPropositions:WhatTheyAreandHowTheyMean.”InR.Marsh(ed.),LogicandKnowledge.London:Allen&Unwin.——(1919/1971)IntroductiontoMathematicalPhilosophy.NewYork:ClarionBooks/Simon&Schuster.——(1957)“Mr.StrawsononReferring.”Mind66:385–9.Saarinen,E.(ed.)(1979)Game-TheoreticalSemantics.Dordrecht:D.Reidel.Sadock,J.(1975)“TheSoft,InterpretiveUnderbellyofGenerativeSemantics.”InP.ColeandJ.L.Morgan(eds.),SyntaxandSemantics,Vol.3:SpeechActs.NewYork:AcademicPress.——(2004)“SpeechActs.”InL.R.HornandG.Ward(eds.),TheHandbookofPragmatics.Oxford:BlackwellPublishing.Sag,I.A.andWasow,T.(1999)SyntacticTheory:AFormalIntroduction.Stanford,CA:CSLIPublications.Salmon,N.(1981)ReferenceandEssence.Princeton,NJ:PrincetonUniversityPress.——(1986)Frege’sPuzzle.Cambridge,MA:BradfordBooks/MITPress.——(1998)“Nonexistence.”Noûs32:277–319.Schiffer,S.(1972)Meaning.Oxford:ClarendonPress.——(1979)“NamingandKnowing.”InP.French,T.Uehling,andH.Wettstein(eds.),ContemporaryPerspectivesinthePhilosophyofLanguage.Minneapolis:UniversityofMinnesotaPress.Schwartz,S.(ed.)(1977)Naming,Necessity,andNaturalKinds.Ithaca,NY:CornellUniversityPress.\nBibliography215Scott,D.(1970)“AdviceonModalLogic.”InK.Lambert(ed.),PhilosophicalProblemsinLogic.Dordrecht:D.Reidel.Searle,J.R.(1958)“ProperNames.”Mind67:166–73.——(1965)“WhatIsaSpeechAct?”InM.Black(ed.),PhilosophyinAmerica.Ithaca,NY:CornellUniversityPress.——(1969)SpeechActs.London:CambridgeUniversityPress.——(1975)“IndirectSpeechActs.”InP.ColeandJ.L.Morgan(eds.),SyntaxandSemantics,Vol.3:SpeechActs.NewYork:AcademicPress.ReprintedinJ.R.Searle(1979a)ExpressionandMeaning.Cambridge:CambridgeUniversityPress.——(1979a)ExpressionandMeaning.Cambridge:CambridgeUniversityPress.——(1979b)“Metaphor.”InA.Ortony(ed.),MetaphorandThought.Cambridge:CambridgeUniversityPress.ReprintedinJ.R.Searle(1979a)ExpressionandMeaning.Cambridge:CambridgeUniversityPress.——(1979c)“ReferentialandAttributive.”Monist62:190–208.——(1983)Intentionality:AnEssayinthePhilosophyofMind.NewYork:CambridgeUniversityPress.Segal,G.(2000)ASlimBookaboutNarrowContent.Cambridge,MA:MITPress.Sellars,W.(1963)“SomeReflectionsonLanguageGames.”InN.Sellars,Science,Perception,andReality.London:Routledge&KeganPaul.——(1974)“MeaningasFunctionalClassification(aPerspectiveontheRelationofSyntaxtoSemantics).”Synthese27:417–37.Sharvy,R.(1980)“AMoreGeneralTheoryofDefiniteDescriptions.”PhilosophicalReview89:607–23.Soames,S.(1987)“DirectReference,PropositionalAttitudes,andSemanticContent.”PhilosophicalTopics15:47–87.——(2002)BeyondRigidity:TheUnfinishedSemanticAgendaof“NamingandNecessity”.Oxford:OxfordUniversityPress.Sosa,E.(1970)“PropositionalAttitudesDeDictoandDeRe.”journalofPhilosophy67:883–96.Sperber,D.andWilson,D.(1986)Relevance:CommunicationandCognition.Cambridge,MA:HarvardUniversityPress.Stalnaker,R.(1970)“Pragmatics.”Synthese22:272–89.ReprintedinD.DavidsonandG.Harman(eds)(1972)SemanticsofNaturalLanguage.Dordrecht:D.Reidel.——(1978)“Assertion.”InP.Cole(ed.),SyntaxandSemantics,Vol.9:Pragmatics.NewYork:AcademicPress.Stanley,J.(2000)“ContextandLogicalForm.”LinguisticsandPhilosophy23:391–434.Stanley,J.,andSzabó,Z.(2000)“OnQuantifierDomainRestriction.”MindandLanguage15:219–61.Stern,J.(2000)MetaphorinContext.Cambridge,MA:MITPress/BradfordBooks.Stich,S.(1976)“Davidson’sSemanticProgram.”CanadianjournalofPhilosophy4:201–27.Stove,D.(1991)ThePlatoCultandOtherPhilosophicalFollies.Oxford:BasilBlackwell.Strawson,P.F.(1950)“OnReferring.”Mind59:320–44.——(1964)“IntentionandConventioninSpeechActs.”PhilosophicalReview73:439–60.——(1970)MeaningandTruth.Oxford:ClarendonPress.Szabó,Z.(2000)“DescriptionsandUniqueness.”PhilosophicalStudies101:29–57.\n216Bibliography——(2003)“DefiniteDescriptionswithoutUniqueness:AReplytoAbbott.”PhilosophicalStudies114:279–91.——(ed.)(2005)SemanticsversusPragmatics.Oxford:OxfordUniversityPress.——(2007)“Compositionality.”InE.N.Zalta(ed.),TheStanfordEncyclopediaofPhilosophy.Availableathttp://plato.stanford.edu/entries/compositionality/(lastaccessed15November2007).Tarski,A.(1956)“TheConceptofTruthinFormalizedLanguages.”InJ.H.Woodger(ed.andtrans.),Logic,Semantics,Metamathematics.Oxford:ClarendonPress.Taylor,K.A.(1988)“We’veGotYouComingandGoing.”LinguisticsandPhilosophy11:493–513.——(1998)TruthandMeaning.Oxford:BasilBlackwell.Tirrell,L.(1989)“Extending:TheStructureofMetaphor.”Noûs23:17–34.Travis,C.(1975)SayingandUnderstanding.NewYork:NewYorkUniversityPress.Tversky,A.(1977)“FeaturesofSimilarity.”PsychologicalReview84:327–52.Unger,P.(1983)“TheCausalTheoryofReference.”PhilosophicalStudies43:1–45.Waismann,F.(1965a)ThePrinciplesofLinguisticPhilosophy,ed.R.Harré.NewYork:St.Martin’sPress.——(1965b)“Verifiability.”AristotelianSocietySupplementaryVolume19:119–50.Warner,R.(1982)“DiscourseLogicandConventionalImplicature.”StudiaAnglicaPosnaniensia14:91–102.Weinreich,U.(1966)“ExplorationsinSemanticTheory.”InT.A.Sebeok(ed.),TheoreticalFoundations.TheHague:Mouton.Weinstein,S.(1974)“TruthandDemonstratives.”Noûs8:179–84.ReprintedinD.DavidsonandG.Harman(eds.)(1975)TheLogicofGrammar.Encino,CA:Dickenson.Weisler,S.(1991)“AnOverviewofMontagueSemantics.”InJ.GarfieldandM.Kiteley(eds.),MeaningandTruth.NewYork:Paragon.Wettstein,H.(1991)HasSemanticsRestedonaMistake?Stanford,CA:StanfordUniversityPress.White,R.M.(1996)TheStructureofMetaphor.Oxford:BasilBlackwell.Wilson,D.(1975)PresuppositionsandNon-Truth-ConditionalSemantics.NewYork:AcademicPress.Wilson,D.,andSperber,D.(2004)“RelevanceTheory.”InL.R.HornandG.Ward(eds.),TheHandbookofPragmatics.Oxford:BlackwellPublishing.Wittgenstein,L.(1953)PhilosophicalInvestigations,trans.G.E.M.Anscombe.Oxford:BasilBlackwell.Wolterstorff,N.(1970)OnUniversals:AnEssayinOntology.Chicago:ChicagoUniversityPress.Yablo,S.(2006)“Non-CatastrophicPresuppositionFailure.”InJ.J.ThomsonandA.Byrne(eds.),ContentandModality:ThemesfromthePhilosophyofRobertStalnaker.Oxford:OxfordUniversityPress.Yourgrau,P.(ed.)(1990)Demonstratives.Oxford:OxfordUniversityPress.Ziff,P.(1960)SemanticAnalysis.Ithaca,NY:CornellUniversityPress.——(1967)“OnH.P.Grice’sAccountofMeaning.”Analysis28:1–8.\nIndexα(assignmentfunction)139–43,167Biro,J.97Black,M.97,190Abbot,B.29Blackburn,S.94,119–20,122,197,199abstractentities11,71,73,76,78,127Blakemore,D.172Ackerman,D.F.52,193Boër,S.62actionsentences119Bradley,F.H.4–6actualreferent26–7,30,126,193Bradley’sregress4–6Adams,E.119Brandom,R.76,83–5Almog,J.61Braun,D.44,52Alston,W.84,153Brogaard,B.29ambiguity24,26,69,79,114,159,184,Burge,T.44,139188ambiguousnames54–5Camp,J.123Amis,K.195cancelabilityofimplicature161–6analogicalmeaningshift187–9Carnap,R.131analyticsentences98,100,103,106–8Carnie,A.125anaphora9,27–30,84Carston,R.161,166,167antecedent(ofanaphor)27–8Cartwright,R.30,75arithmeticaltruths49causaltheoryofmetaphor177–9,186–7,asserting21,50–1,58,79,83,122–3,190147,150Causal–HistoricalTheoryofreferringassignmentfunctionseeα53–61,124attributiveuseofdefinitedescriptionsChambers,E.K.5823–4,29;seealsoreferentialuseofcharacterofadeicticexpression142definitedescriptionsChesterton,G.K.192Austin,J.L.76–7,84,138,144–55,161,Chierchia,G.133198Chomsky,N.115–17Avramides,A.92,97clustertheoryofreference38–43Ayer,A.J.108coextendingterms120,126–7,129,133;seealsoextensionBach,K.44,152,155,156,170–2,197,“cognitive”meaning101198Cohen,L.J.151–3,171,187Bar-On,D.108Cohen,T.190Barker,S.84,154Cohen’sproblem151–3Beardsley,M.177,179Cole,P.155,172behaviorism103communication25,27,53–4,60,84,beliefsentences12,17–19,33–4,50–1,90–1,122,137,158–9,166,172,179,121–2,130–1,191184–5,192,200Belnap,N.123communicativeintention88Bennett,J.87,74compositionality84,109,111–12,Bezuidenhout,A.139123–4,127,195,197\n218Indexconcepts22,34,67,70–1,89–90,122,entailment66,71,79,114,120,128,156,129,132,194162–3,171constatives146–8entitytheoriesofmeaning66,71constitutiverules144,149–50,152,154Erwin,E.62,104contextofutterance21,130,138,142,Evans,G.28,56,58,61,125contextualdefinition13Everett,A.44contingenttruth12,17,33,121,126,explicature195conventionalimplicature156,163,expression:ofproposition20,70,73,78,165–6,171–2,19986,147;ofthought66–68conversationalimplicature95,156,158,extension113,115,120–1,123–4,161–5,169–71,175,198,200126–30,142;seealsocoextendingconversationalmaxims159–60,166,170termsconveyedmeanings157,166,171,175,196fictionalnames42–4,61Cooper,D.E.187FigurativeSimiletheoryofmetaphorCooperativePrinciple(CP)158–9,167175,180–4,189–90Coyle,M.153figurativeusage94,175–7,182CPseeCooperativePrincipleFillmore,C.140Cresswell,M.J.133,139,152,155Fine,A.62Culicover,P.125flacciddesignator47–8,52,127,193cummings,e.e.4,184,200Fodor,J.A.196Fogelin,R.180–84,186,190,200Davidson,D.109–25,129,131,133,137,formallogic109,124,133139–40,143,156,175,177–9,180,freeenrichment167183–4,186–7,189–90Frege,G.6,9–11,17,19,31–5,43,47,Davis,W.162,169–7152,67–8,74,126–7,129–30,133,192,Davison,A.172198deicticexpressionseedeixisFrege’spuzzle9,19,52deixis131,138–40,143demonstrativepronouns10,22,50,120Gazdar,G.155descriptions9–29,31–43,47–9,52–8,Geach,P.2877,84,110,114,121,153,156,163,Geis,M.157171,191–2;indefinite28–9,42,53;Ginet,C.133,155plural13,27,29Gödel,K.40,192Devitt,M.52,56–8,61,108,192,193Goldsmith,F.199DirectReference(DR)theoryofGoodman,N.178,180singularterms45,49–61,91,165Gordon,D.170,172disambiguation40,55,137,142–3,167,grammar34,99,108,109–11,115,189117,124,126,128,137,195;seealsodomainofaquantifier22,122syntaxDonnellan,K.9,23–7,29,30,42–4,47,Green,G.M.155,17254,56,192Grice,H.P.25,27,67,84,86–97,109,“dormitivevirtue”73,194122–5,154,156,158–71,175,184,DRseeDirectReference(DR)theoryof185,189,195,196,198–9singulartermsGrover,D.123“dthat”193Duhem,P.98,105–6,108,110,145Hallett,G.85Dummett,M.39,108,132Harman,G.62,73–5,97,125,128,Dupré,J.62139–41,143,167“dynamicfeature”ofmeaning131–2Harnish,R.M.152,155,162,170,172,198Elgin,C.190Heim,I.22,29ellipsis21–2,179,200Hempel,C.G.100,102–5,emptynames56“hereby”criterionforperformatives146–8\nIndex219Hintikka,J.125,127Lasnik,H.125Hofweber,T.44Lemmon,E.J.75Holdcroft,D.155Lepore,E.22Horn,L.R.172,198Levinson,S.155,172,198Hornstein,N.125lexicalmeaning157,187–8Horwich,P.195Linsky,L.24–5,30,62literalmeaning25,94–6,157,169,ideationaltheoriesofmeaning66–70,175–8374,79,86,87,129Loar,B.44identitystatements11,17Locke,J.66,74idiolect95locutionarycontent148–54,184,198illocutionaryforce84,144,148–9,154,logicalform13–21,34,109,115,118,157,184,198122,124,167,197implication22,156–66,170–1,199LogicalPositivism79,98–110,119,177,“inthehead”,meaningisorisnot45,19659,61,132,194Loux,M.6,75indeterminacyoftranslation107,108Ludlow,P.22,29,30indexicals84,120,139,143,182Lyons,J.187indirectforce156,168–72,175,200infelicitousspeechacts144,149–51,MacKay,A.F.27,30,97154,161,166,198Marcus,R.B.39,49–50,52InferentialTheoryofMeaning79,83,MaximofEvidence159–60100MaximofRelevance159–62inferring76,79,83,84MaximofStrength159–64infinitepolysemy187McConnell-Ginet,S.133instrumentalism102McDowell,J.85,125intensions126–33,142,197;intensionalMcKinsey,M.61isomorphism131;intensionallogic“meaningfacts”65–74,79,101,107,114,127,140;intensionalsemantics126,128,131129,130;intensionality121Meinong,A.11,30,35,192intentionalcausation132metaphoricalmeaning175–8,182–9invitedinference157,166,171metaphysics46,102–3,115Inwagen,P.van197Mill,J.S31,49irony171,185–6Milliannames31–4,39,49–52,60,193modality,modalproperties39,47–8,Jackson,F.52121,129Johnson,M.176,188,190Montague,R.5,127,133,140,197mood145,148,151,156,168–70Kamp,J.29Moore,G.E.68–9,71–2,194Kaplan,D.30,50,142–3,193Moran,R.178–9,187Karttunen,L.171–2Morgan,J.L.155,170,172Kempson,R.171Moriarty,J.10–11,32,191Kenny,A.85Morris,C.137Kittay,E.185,187–90Kleiman,L.62NaiveSimileTheoryofmetaphor175,Kripke,S.25–7,31,39–44,45–62,80,179–83,189–90126,131,192–3NameClaimthesis34–44,49,53,61,Kroon,F.58192Kvart,I.61Nankas,W.196narrowproperties132;seealso“intheLakoff,G.170,172,176,188head”;widepropertieslanguagegames78–82,145,195natural-kindterms45,58–9,61,194;seeLarson,R.125alsoTwinEarthLasersohn,P.192Neale,S.22,29,30,192\n220Index“near-miss”cases25–6Radford,A.125necessarytruths49,130Recanati,F.61,143,161,162,163,167,negativeexistentials9,11,15–6,31–2,19835,52,165Reeves,A.120nonexistents3–4,9–11,14,19–20,31–2,ReferentialTheoryofmeaning1,3–6,35,52,56,127,165,1929–12,31,65,67,71,83,115,197novelsentences2,6,76,80,84,94–6,referentialuseofdefinitedescriptions109–13,123–423–4,29;seealsoattributiveuseofNunberg,G.141,143definitedescriptionsreferring4–5,10,13,19,20,23,26–7,Oafish112–15,119,121–238,42,45,48,51–5,58–9,61,78,82,observationlanguage102,104–5124,193Occam’sRazor72regulativerules144,149–50,154opacity12,18,33,51,121Reimer,M.22,30,190Ostertag,G.30RelevanceTheory166,171representation115,126,141Parsons,T.30restrictedquantification22Pelletier,F.J.197resultantprocedures95–6,123performatives175–6Reyle,U.29perlocutionaryeffect144,148–9,154–5Rhees,R.85personalpronouns10,137rigiddesignator47–9,52–3,59–62,127,Peters,S.172193;seealsoMilliannamesphrasemarkers115–17rigidifieddescriptions52;seealsoPitcher,G.85“dthat”Plantinga,A.52,193Rosenberg,J.F.85,97,194Platts,M.93,125Ross,J.F.187–90possibleworlds45–9,59–60,71,Routley,R.30126–33,137–8,140,142–3,193Russell,B.5,9–30,31–44,45,47–50,53,pragmaticpragmatics137–9,142;see58,68–70,74,76–7,103,110,114–15,alsosemanticpragmatics118,127,156,161,163,171,191–2PragmaticTheoryofmetaphor183–4,187Saarinen,E.125predicativeexpressions13Sadock,J.155,172presupposition20,156,163–6,171;seeSag,I.125alsosemanticpresuppositionSalmon,N.50,52,62primaryoccurrence16,42,191;seealsosaturation167,199scope;secondaryoccurrenceSBVAUN38–9;seealsoclustertheorypropernames1,9–10,12,19,23,31–43,Scheffler,I.19045–9,52,54,60,80,84,91,121,124,Schiffer,S.90,92,97,155,193,196,198165,192–3,197Schwartz,S.62propositionalcontent82,99,110,137–9,scope16,18,40,42,50,191–2;seealso141–4,147–8,153–4,167,198primaryoccurrence;secondarypropositions20,65,68–75,76,78,84,occurrence86,122–3,131,194,198Scott,D.140PropositionTheoryofmeaning65,Searle,J.31,36,38–44,48,52–3,58,91,68–75,76,82,99,109,127–31149,154–6,168–71,175,180,182,Putnam,H.45,58–62,80,108,132,194184–90,192–4,200secondaryoccurrence16,18,40;seealso“qua-problem”58primaryoccurrence;scopequalities4Segal,G.22,29,125,194quantifiers13,16,22Sellars,W.35–7,48,76,79–80,83–5,Quine,W.V.4,12,41,75,98,105–8,100110,145,197semanticpragmatics137–9,141–3;seealsopragmaticpragmatics\nIndex221semanticpresupposition163–5truth,theoriesof119–21semanticreferent19,23,25–7,192truthcondition109–15,118–20,122,semi-grammaticalutterances194–5124,126–9,131–3,138–41,144–5,“senses”11,31–5,37,43,52,54,126–30,149,151,154,197148,151,189truth-functionalconnectives122,126,sentencemeaning53,84,86–7,92–6,13398,104,108–9,117,123–4,132,138,truth-value9,12,18,26,29,33–4,43,149,153,157,162,175,179,18546,50,70,109,111,118–22,124,Sharvy,R.29128–31,140–2,150,153,156,163,similarity175,179–82,186178,189,191,199simile175,178–84,189T-sentences114,119–24,131,137,similetheoriesofmetaphor175–90139–40,143,197singularterms9–14,18–19,29,31,34,TwinEarth45,58–62,79–80,132,19450–1,58,123,127,161Soames,S.52understanding1,3,15,50,68,77,80–1,speaker-meaning25–6,86–9,92–3,96,102,109,111,132,142,157,196123,149,157–8,175,184–7,189,195Unger,P.62speaker-reference25,123–4Uriagereka,J.125speaker-referent25–7,192utterer’smeaningseespeaker-meaningSperber,D.160,162,166–7“spot-checktest”35–7,48–9verificationcondition98–9,101–8,110,Stalnaker,R.160145,196Stampe,D.196verificationism98,100,102,105,107,Stanley,J.22,199128,145,177,196Sterelny,K.58Stern,J.190Waismann,F.6,74,82,85,108Stich,S.119Ward,G.172Strawson,P.F.9,19–25,29,30,46,77,Warner,R.17287,120,122–3,138,154,156,161,Wasow,T.125163,165,171,192,196,198–9Weinreich,U.187subject–predicateform4–5,9–10,13,Weinstein,S.13915,112–13,127–8,183Weisler,S.133substitutivity9,19,50,121,126,129–31,Wettstein,H.52,61133,193“wh-”questions118,148synonymy79,114,131wideproperties132;seealso“inthesyntax111,115,122–5,132,137–8,199;head”;narrowpropertiesseealsogrammarWilson,D.160,162,166,171Szabó,Z.22,29,112,197Wittgenstein,L.6,74,76–8,80,82–5,87,93,100,112,137–8,140,145,153,Tarski,A.114,122,124195Taylor,K.A.30,140,143,194Wolterstorff,N.6theory-ladenness105;seealsoDuhem,P.Yablo,S.20Tirrell,L.190Yourgrau,P.143transformations117–18,122,124transparentoccurrenceofaterm12,33,Zemach,E.6250–1,193Ziff,P.89–93,142,196–7Travis,C.155Zwicky,A.157