【continuum哲学指南系列】语言哲学
“unlikemanyanthologies,thisbookisnotamerecollectionofseminalworksintherecenthistoryofthefield,butratheracollectionofcontemporaryessayswrittenwiththeprofessedpurposeofintroducingstudentsandscholarstothefield...Eachoftheoriginalessaysprovidesbothacriticalhistoryofthefundamentalconceptsthecentraltothephilosophyoflanguageandapointedanalysisoftheissuesinvolvedfromthecontributor’sownpointofview.TheresultisavaluableresourceforhyoflanguagethephilosopstudentsandscholarsinterestedineitherthehistoryofthedisciplineorthestatebloomsburyCompaniontoofthecontemporarydebate.”A.L.Morton,SaintXavierUniversity,USA,CHOICEthe“García-CarpinteroandKölbelhavesucceededincompilingasetofelevenwonderfullyclearandhighlyaccessibleoverviewarticlesonthephilosophyoflanguage.Thisisaterrificcollectionwithacoupleofrealgems.Thearticlesarebloomsburygenerallyinformative,concise,andapleasuretoread.Iamconvincedthatevenexpertsinphilosophyoflanguagewillbenefitfromworkingthroughthevolume.IamsureIwillrevisittheCompanionmanytimesinthefuture.”AndersJ.Schoubye,UniversityofEdinburgh,UK,NotreDamePhilosophicalReviewCompaniontoBeginningwithFrege,Russell,andWittgenstein,thiscomprehensivecompanionprovidesacriticalhistoryofthecoreconceptsinthearea.Fromgenerativesyntaxandformalsemanticsthephilosophytobroaderphilosophicalissuessuchastheoriesofmeaning,andcontextdependence,awell-knownteamofexpertsofferinsightfulanalysisintosomeofthefundamentalquestionsaskedbythephilosophyoflanguage.Featuringaseriesofstudytools,includinganAtoZofkeytermsandconcepts,adetailedlistofresourcesandafullyannotatedbibliography,TheBloomsburyoflanguageCompaniontothePhilosophyofLanguageisidealforstudentsandscholarslookingtobetterunderstandthequestionsanddebatesinformingthesubjecttoday.andmaxkÖlbelmanuelgareditedbymanuelgarCía-CarpinteroisProfessorintheDepartmentofLogic,HistoryandthePhilosophyofScienceattheUniversityofBarcelona,Spain.maxkÖlbelisICREAResearchProfessorintheDepartmentofLogic,HistoryandthePhilosophyofScienceattheUniversityofBarcelona,Spain.Cía-arpinteroeditedbymanuelgarCía-CarpinteroandmaxkÖlbelphilosophywww.bloomsbury.comCoverdesignbyEleanorRoseCoverimage©JupiterImagesAlsoavailablefromBloomsbury\nTheBloomsburyCompaniontothePhilosophyofLanguage9781472578235_txt_print.indd108/08/201415:45\nTheBloomsburyCompanionsseriesisamajorseriesofsinglevolumecompanionstokeyresearchfieldsinthehumanitiesaimedatpostgraduatestudents,scholarsandlibraries.Eachcompanionoffersacomprehensivereferenceresourcegivinganoverviewofkeytopics,researchareas,newdirectionsandamanageableguidetobeginningordevelopingresearchinthefield.Adistinctivefeatureoftheseriesisthateachcompanionprovidespracticalguidanceonadvancedstudyandresearchinthefield,includingresearchmethodsandsubject-specificresources.Titlescurrentlyavailableintheseries:Aesthetics,editedbyAnnaChristinaRibeiroAnalyticPhilosophy,editedbyBarryDaintonandHowardRobinsonAristotle,editedbyClaudiaBaracchiContinentalPhilosophy,editedbyJohnÓMaoilearcaandBethLordEpistemology,editedbyAndrewCullisonEthics,editedbyChristianMillerExistentialism,editedbyJackReynolds,FelicityJosephandAshleyWoodwardHegel,editedbyAllegradeLaurentiisandJeffreyEdwardsHeidegger,editedbyFrancoisRaffoulandEricSeanNelsonHobbes,editedbyS.A.LloydHume,editedbyAlanBaileyandDanO’BrienKant,editedbyGaryBanham,DennisSchultingandNigelHemsLeibniz,editedbyBrandonC.LookLocke,editedbyS.-J.Savonius-Wroth,PaulSchuurmanandJonathanWalmsleyMetaphysics,editedbyRobertW.BarnardandNeilA.MansonPhilosophyofMind,editedbyJamesGarveyPhilosophyofScience,editedbyStevenFrenchandJuhaSaatsiPlato,editedbyGeraldA.PressPragmatism,editedbySamiPihlströmSocrates,editedbyJohnBussanichandNicholasD.SmithSpinoza,editedbyWiepvanBunge,HenriKrop,PietSteenbakkers,JeroenM.M.vandeVen9781472578235_txt_print.indd208/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageEditedbyManuelGarcía-CarpinteroandMaxKölbel9781472578235_txt_print.indd308/08/201415:45\nBloomsburyAcademicAnimprintofBloomsburyPublishingPlc50BedfordSquare1385BroadwayLondonNewYorkWC1B3DPNY10018UKUSAwww.bloomsbury.comBloomsburyisaregisteredtrademarkofBloomsburyPublishingPlcFirstpublishedinpaperback2014FirstpublishedasTheContinuumCompaniontothePhilosophyofLanguage2012©ManuelGarcía-Carpintero,MaxKölbelandContributors2014ManuelGarcía-CarpinteroandMaxKölbelhaveassertedtheirrightundertheCopyright,DesignsandPatentsAct,1988,tobeidentifiedastheEditorsofthiswork.Allrightsreserved.Nopartofthispublicationmaybereproducedortransmittedinanyformorbyanymeans,electronicormechanical,includingphotocopying,recording,oranyinformationstorageorretrievalsystem,withoutpriorpermissioninwritingfromthepublishers.NoresponsibilityforlosscausedtoanyindividualororganizationactingonorrefrainingfromactionasaresultofthematerialinthispublicationcanbeacceptedbyBloomsburyortheauthor.BritishLibraryCataloguing-in-PublicationDataAcataloguerecordforthisbookisavailablefromtheBritishLibrary.ISBN:ePDF:978-1-4725-7821-1ISBN:ISBN:LibraryofCongressCataloging-in-PublicationDataAcatalogrecordforthisbookisavaliablefromtheLibraryofCongress.TypesetbyFakenhamPrepressSolutions,Fakenham,NorfolkNR218NN9781472578235_txt_print.indd408/08/201415:45\nContentsContributorsviPrefacevii1EditorialIntroduction:HistoryofthePhilosophyofLanguage1ManuelGarcía-Carpintero2OntheNatureofLanguage:ABasicExposition26JamesHigginbotham3FormalSemantics47JoshDever4TheoriesofMeaningandTruthConditions84KathrinGlüer5Reference106GenovevaMartí6IntensionalContexts125MichaelNelson7ContextDependence153KentBach8Pragmatics185FrançoisRecanati9SemanticNormativityandNaturalism203JoséL.Zalabardo10Analyticity,Apriority,Modality228AlbertCasullo11NewDirectionsinthePhilosophyofLanguage251MaxKölbelA–ZofKeyTerms266SelectBibliography279Index286v9781472578235_txt_print.indd508/08/201415:45\nContributorsKentBachisProfessorofPhilosophyatSanFranciscoStateUniversity.AlbertCasulloisProfessorofPhilosophyattheUniversityofNebraska-Lincoln.JoshDeverisAssociateProfessorofPhilosophyattheUniversityofTexasatAustin.ManuelGarcía-CarpinteroisprofesorcatedráticintheDepartmentofLogic,HistoryandPhilosophyofScienceattheUniversitatdeBarcelona.KathrinGlüerisProfessorofTheoreticalPhilosophyatStockholmUniversity.JamesHigginbothamisDistinguishedProfessorofPhilosophyandLinguisticsandLindaMacDonaldHilfChairinPhilosophyattheUniversityofSouthernCalifornia.MaxKölbelisICREAResearchProfessorintheDepartmentofLogic,HistoryandPhilosophyofScienceattheUniversitatdeBarcelona.GenovevaMartíisICREAResearchProfessorintheDepartmentofLogic,HistoryandPhilosophyofScienceattheUniversitatdeBarcelona.MichaelNelsonisAssistantProfessorofPhilosophyattheUniversityofCaliforniaatRiverside.FrançoisRecanatiisResearchFellowattheCentreNationaldelaRechercheScientifiqueandattheInstitutJeanNicod,Paris.JoséZalabardoisReaderattheUniversityCollegeLondonPhilosophyDepartment.vi9781472578235_txt_print.indd608/08/201415:45\nPrefaceThisbookaimstoprovideacomprehensiveguideforthosewhowanttoembarkuponresearchinthephilosophyoflanguage.Wehaveselectedninecentralareasofthephilosophyoflanguage,andwereabletoenlistthehelpofnineleadingexpertstowriteaccessible,yethigh-levelandup-to-dateintroductionstoeachoftheseareas.Theircontributionsmakeupthemainbodyofthisbook.Wecomplementedthiswithanintroductionrecountingthehistoryofthefield,anessayonnewdirectionsofresearch,anA–Zofkeytermsandabibliographycontainingsuggestedfurtherreadingsineachoftheareas.Theresultwill,wehope,beausefultoolforadvancedundergraduates,beginningresearchersandanyonewishingtogainanoverviewaboutwherethephilosophyoflanguage,orsomeofitssub-disciplines,standstoday.TheContinuumCompaniontothePhilosophyofLanguage*hasbeenalongtimeinthemaking,andweareverygratefulforallthosewhohavehelpedbringittogether.Firstandforemost,wethankthecontributors,someofwhomhavebeenpatientinwaitingfortheircontributiontoappearandsomeofwhomhavemanagedtofindtimetowritetheircontributionwhenmanymoreurgentdemandswereeatinguptheirtime.WewouldequallyliketothanktheeditorsatContinuumfortheirsupportandpatience,inparticularSarahCampbellandTomCrick.Finally,wewouldliketothankJohnHordenforpreparingtheindexandhelpingwiththeproofreading.ManuelGarcía-CarpinteroMaxKölbel*Theoriginalhardbackeditionpublishedin2012asTheContinuumCompaniontothePhilosophyofLanguagevii9781472578235_txt_print.indd708/08/201415:45\n9781472578235_txt_print.indd808/08/201415:45\nEditorialIntroduction:1HistoryofthePhilosophyofLanguage1ManuelGarcía-Carpintero2ThePhilosophyofLanguagehasahistoryalmostaslongasthehistoryofPhilosophyitself.Plato’sCratylusandSophist,andAristotle’sDeInterpretationeandPriorAnalytics,containimportantreflectionsontopicssuchastheconven-tionalityoflanguage,thesubject–predicatestructure,validinferenceanditsrelationswiththestructureoflanguageandthought,truth,ortheontologicalimplicationsoflinguisticcategories.Medievalphilosopherscarriedoutstudiesofreference(“suppositio”)andgeneralizationassophisticatedasany.ThePort-Royallogicians,HobbesandLocketookthosediscussionsforward,and,inthelattercase,anticipatedcurrentconcernsaboutthewaynaturalkindtermswork.Inthefollowingfewpages,however,Iwilllimitmyselftodrawingaveryrough(andratheridiosyncratic)mapoftheterrainofthecontemporaryscene,asitwassetoutintheworkofFrege,RussellandtheearlyWittgenstein–thepresupposedcommonbackground,taughttobeginnersinthediscipline,forthethemestobefurtherexploredfromapresent-dayperspectiveintheensuingchapters.Inthefirstpartofthechapter,IwilloutlinesomecoreissuesastheyarepresentedinwhatinmyviewistheinsightfulsystematicarticulationofFrege’sandRussell’sthemesinWittgenstein’sTractatusLogico-Philosophicus.Inthesecondpart,Iwillsumupthemainissues,describesomecontributionstotheminFrege,Russell,Wittgensteinandotherhistoricallandmarks,andindicatehowtheyareapproachedtoday,aspresentedintheensuingchapters.Theintroductionconcludeswithabriefdiscussionofresearchmethodsandproblemsinthefield.MeaningandModalityintheTractatusThecoreissuesinthephilosophyoflanguagearefirstputforthwithcompellingself-consciousdepthinWittgenstein’sTractatusLogico-Philosophicus,hisappraisalofthepresuppositionsofFrege’sandRussell’sLogicistProgram19781472578235_txt_print.indd108/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguage–evenifthebookwouldnothavebeenpossiblewithoutFrege’sandRussell’sground-breakingresearch.Itistruethat,incontrastwithFrege’sandRussell’sworks,theTractatusisanopaquepiece,whoseclaims(andevenmore,thereasons,argumentsoratleastmotivationsforthem)aredifficulttomakeout,inthisrespectareflectionoftheratherdogmaticmethodologicalattitudeofitsauthor.ItisalsotruethatsuchdogmatismappearstohaveprecludedWittgensteinfromseeingthe,insomecasesglaringlymanifest,difficultiesfortheviewshehadputforward,andtheextenttowhichthealternativeviewsofhistwopredecessors,whichhehadhaughtilydismissed,weremuchmoresensible.However,inmyviewitwasintheTractatusthattheproperdimen-sionsandinterconnectionsofthemainproblemsconfrontedafterwardsinthedisciplineareclearlyenvisagedforthefirsttime.NeitherFregenorRussellappearstohavepaidmuchthoughttowhathasbecome,sincetheTractatus,acoreissueinthephilosophyoflanguage–thelinkbetweengraspingtherepresentationalcontentsofthoughtsandsentences,andknowledgeofmodality;orsoIwilltrytosuggestinthenextfewparagraphs.Forthemostparttheyaimnotmainlytoestablishthisperhapsidiosyncratichistoricalpoint,buttosketchoutthesecoreproblems,sothatlaterwecantracetherelationswithhowtheyareapproachedtoday,aspresentedinthechapterstofollow.ThosecoreproblemsinthephilosophyoflanguageonlyperspicuouslyadumbratedintheearlyhistoryofanalyticphilosophyintheTractatusconcerntherelationsbetweenmeaning,modalityandourknowledgeofthem.Frege’sproject,whichhepursuedrelentlesslyformostofhisintellectuallifeandwhose(fromhisperspective)tragicfailureRussellspotted,wastheLogicistProgram,aimedatprovingthatarithmeticreducestopurelogic.Frege’sworkwashardlyafully-fledgedfailure:hehadcomeveryclosetoatleastreducingarithmetictologicandsettheory,alongthelinesusedlaterinRussell’sandWhitehead’sPrincipiaorintheindependentlypursuedCantorianprogram.Intheprocess,hecameupwithoutstandinglysignificantconceptualinnova-tions,frommodernlogicandsemanticstoanoriginalandinfluentialviewinthephilosophyofmathematicsthatmanystillthinkfundamentallycorrect.However,afullappraisaloftheepistemologicalandontologicalyieldsoftheprojectrequiredanexaminationoftheepistemologicalandontologicalstatusoflogicandlogicalvaliditythemselves;andthatinitsturnleadstoathoroughexaminationofthenatureoftherepresentationaldevicesthroughwhichwecarryoutlogicallyvalidinferences:naturallanguageandthethoughtsitconveys(whatwemaycallatheoryofintentionality).FregeandRussellsomehowsawthis,andinfactmadesuggestionsaboutthematter(outlinedbelow)attimesmoresensiblethanthoseintheTractatus,attimessimplyincorporatedintoit.Butitisonlyinthatwork,Ithink,thatthenatureoftheproblemsandtheirinterconnectionsissystematicallyrealized,throughthe29781472578235_txt_print.indd208/08/201415:45\nEditorialIntroduction:HistoryofthePhilosophyofLanguagerealizationthatrepresentationinnaturallanguagesandinthoughtisinextri-cablytiedupwithdiscriminationbetweenpossibilities.Notoriously,theTractatuscontainsaflawedtheoryofintentionality,theso-called“picturetheory”;but,morethanitsfailures,whatisinterestingforourpresentpurposesistoappreciatewhatitsetouttoachieve–especiallyhowWittgensteinhopedthatitwoulddeliverwhatinhisviewFregeandRussellhadfailedtoprovide:aphilosophicallyadequateaccountoflogicalvalidityandhenceofthefoundationsoftheirlogicistproject.3ToputitinthemetaphorhelaterusedintheInvestigations,criticisinghisearlierviews,Wittgenstein’sobjectionintheTractatustotheviewonthenatureoflogicalvaliditythatFregeandRussellhaddefendedisthatitdoesnotaccountforthe“sublimity”oflogic:theydidnotaccountforthecharacteristicmodalpropertiesoflogicaltruthsandvalidities,andourknowledgethereof,asresultingfromessentialpropertiesoftherepresentationalmeansinwhichtheyarecashedout.Itissuchanaccount,accordingtohim,thatthepicturetheoryprovides.AccordingtoFregeandRussell,logicallyvalidpropositions,andinfer-entialtransitionsamongthem,aredistinguishedbytheirmaximalgenerality;forinstance:giventhataequalsb,andbequalsc,wecaninferthataequalsc,nomatterwhata,bandcare.AccordingtotheTractatus,however,thisiswrong(Tractatus6.1231).Ontheonehand,somelogicaltruthsarenotliterallyspeakinggeneral(ifHesperusisPhosphorus,andPhosphorusisVenus,thenHesperusisVenusisitselfalogicaltruth);ontheother,ageneraltruthmaywellbeonlyaccidentallytrue(wecanexpressinpurelygeneraltermstheclaimthatthereareinfinitelymanythings,whichaccordingtoWittgensteinisnotalogicaltruth).Logicalvaliditiesarenecessary;andtheyareapriori.4Frege’sandRussell’sproposalsdonotaccountforthiscrucialfact:whyshouldmaximalgeneralityentailnecessityandapriority?Itwasthefactthat,inhisview,thepicturetheoryaccountedforitthatmainlyrecommendeditinhiseyes.ThepicturetheoryisrelevanttosolvetheproblembecauseforWittgensteinlogicalvaliditiesareexpressedinnaturallanguages(Tractatus5.5563)–orthethoughtstheyconvey–whoseessentialrepresentationalpropertiesthepicturetheorycharacterizes.Artificiallanguages,farfrombeing“ideallanguages”worthstudyingintheirownrightasmoreadequatetocarryingoutvalidinferences–asFregeandRussellthought–aremere“frictionlessplanes”;theyareusefulfictionswhosestudyisaconvenientmeanstoexhibitinasimplerwaythelogicalpropertiesofourordinaryassertionsandthoughts.Asidefromitsmotivationasawayofaccountingforthemodalpropertiesoflogicaltruthandvalidity,Wittgensteinsupportedhispicturetheoryofinten-tionalityarguingthatonlysuchatheoryaccountsfortwofundamentalfactsaboutrepresentationinlanguageandthought.First,weunderstandlinguisticrepresentationsandgraspthoughts(atleastinparadigmcases,letussay,soasnottoprejudgeanyrelevantissue)withoutknowingwhetherornottheyare39781472578235_txt_print.indd308/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguagecorrect,whetherornottherepresentedrealityisinfactasrepresented;Iwillsummarizethishenceforthwiththeslogan“representationsmayfail”.Second(“representationsmaybenew”),wecanunderstandorgraspimmediately,withoutfurtherexplanation,representationsthatwehavenotencounteredbefore.5Howisthepicturetheorysupposedtodealeffectivelywiththeseexplanatoryissues?(Therewillbenopointinconsideringthefurtherissueofwhetheritreallyistheonlytheorythataccountsforthem.)Thepicturetheory,asIunderstandit,ascribestoanyintentionalsystem,i.e.,anysystemexhib-itingthetwopropertiestobeexplained,twocrucialsemanticfeatures,whichwemaydescribeasanexternalandaninternalone.Theexternalingredientcomprisesalexiconandthecorrelationsoftheitemsinitwithindependentobjects,correlationswhichWittgensteinthoughtofasconsistingofimplicitostensivedefinitions.Theinternalingredientisanabstractsyntaxapplyingtotheitemsinthelexiconwhichsignifies,bywayofwhatGoodman(1976,52)callsexemplification,6identicalrelationsbetweentheitemscorrelatedwiththembytheexternalingredient.Itisthelatterfeaturethatmakessentencesandthoughtsintopictures:thedistinguishingfeatureofpicturesisthattheyrepresentpropertiesthattheythemselvesexemplify;theyrepresentthankstothefactthatthereisarangeofpropertiestheyliterallysharewiththerepre-sentedsituations.Letusseehowthisissupposedtosolvethefirstproblem,thatrepresen-tationsmayfail.Thesyntaxdeterminesaclassofwell-formedelementarysentences;notjustanyconcatenationofitemsinthelexiconisacceptable,onlysomearepermitted.Eachofthemisinthatrespectapossibility:itispossibletosayit,asopposedtoabstainingfromsayingit,independentlyoftheothers.Sayingisherethelowestcommonfactorofdifferentspeechacts–asserting,ordering,conjecturing,requesting,andsoon–whosedistinguishingdifferencesWittgensteinthoughtirrelevantforhisconcerns.Thesyntaxthusdeterminesaclassofmaximal“discourses”–allowedcombinationsofthetwodesignatedpossibilitiesforeachelementarysentence.Correspondingly,giventhatthesyntaxissharedbythelexiconandcorrelateditems,itdeterminesthepossibilitythatthecombinationofitemscorrespondingtothenamesinanygivenelementarysentence(astateofaffairs)obtains,andthepossibilitythatitdoesnotobtain.Itdeterminestherebyacorrespondinglogicalspaceofmaximalcombinationsofthesetwopossibilitiesforeachstateofaffairs;onlyoneofthemcanbeactualized,constitutingtheactualworld.Whatisrequiredtounderstandasentenceistoknowtheinterpretedlexiconfromwhichitisbuilt,anditslogicalsyntax;whatistherebyknownisapossiblestateofaffairs,theclassofmaximalcombinationsconstitutingthelogicalspacecompatiblewithitsobtaining,whatWittgensteincalls(followingFrege)thesentence’struth-condition;itisnotrequiredtoknowwhetherornotthisclassincludestheactualworld.749781472578235_txt_print.indd408/08/201415:45\nEditorialIntroduction:HistoryofthePhilosophyofLanguageAccordingtothis,all(andonly)truth-conditionsare(contentsof)possiblesayings,notonlythoseexpressedbyelementarysentences.Someappro-priatesetofexpressions(the“logicalconstants”,ontheTractarianaccount)isneeded,togaintheadditionalexpressivepotentialneededtoexpressalltruth-conditions.Buttheclaimmadeabouttheexplanatoryvirtueofthepicturetheoryforthecaseofelementarysentencesisintendedtoapplyalsotocomplexsentencesincludingtheseexpressions.Understandingthemrequires,accordingtothepicturetheory,knowingtheinterpretedlexicon,theirlogicalsyntaxandtheidentical“syntax”intheworldsignifiedbyexemplification,plusthesetoflogicalconstantsneededinordertoexpresseverypossibletruth-conditiontherebydetermined.Thisassignstoanynon-defective(neithertautologousnorcontradictory)sentenceatruth-condition,withouttherebyestablishingwhetherornotitactuallyobtains.Wittgenstein(Tractatus,2.1511;cf.Investigations,§§95,194)particularlylikedthefactthatthislittletheoryaccountsforthefirstproblemofintentionality,thatrepresentationsmayfail,whilepreservinganessentialconnectionbetweenlinguisticrepresentationsandtheworld–andthusrepresentationsareofrealitems,notsomeinterme-diateghosts,asinrepresentationalistaccountsofperceptualexperience.Thisisachievedinthattherepresentedpossiblestatesofaffairsaremadeofrealobjects,constitutingtheactualworld(allpossibleworlds,giventhatalllexicalitemsareontheTractarianviewKripkean“rigiddesignators”,designatingthesameentitywithrespecttoallpossibleworlds)andofequallyreal,possibility-determining,“syntactical”relationsbetweenthem.Accountingforthesecondexplanatoryissue(thatrepresentationsmaybenew),assumingthepicturetheoryaspresented,isstraightforward.Knowingthelexicon,thelogicalsyntaxthataswehaveseensignifiesbyexemplification,andtherelevantsetoflogicalconstantssufficesforunderstandingsentencesbeyondthosethatonehasinfactencountered;incontrast,themeaningofanynewlexicalitemmustbeexplainedtous.Finally,thisishowthepicturetheoryissupposedtoaccountforthe“sublimity”oflogic,thefactthatweknowapriorinecessarytruthsandrelationsofnecessarytruthpreservation,toconcludethissketchyoutline:“Itisthepeculiarmarkoflogicalpropositionsthatonecanrecognizethattheyaretruefromthesymbolalone,andthisfactcontainsinitselfthewholephilosophyoflogic”(Tractatus,6.113).Iftherelationsthatdeterminewhichstatesofaffairsarepossiblearereflectedbyidenticalrelationsdeterminingwhichcombina-tionsoflexicalitemsarelogico-syntacticallywellformed,wehaveattheveryleasttheimpressionisticbeginningsofanexplanation.Knowingthefactsthatdeterminewhichpossibilitiesthereare,whichonescorrespondtoagivensaying,andwhichones,expressedbyagivensaying,areincludedintheonesexpressedbyothersisalreadyapresuppositionofunderstandingthose(orany)sayings.Logicaltruthisjusttruthwithrespecttoallpossibilities,and59781472578235_txt_print.indd508/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguagelogicalvaliditythecontainmentofallthepossibilitiesforthepremisesinthepossibilitiesfortheconclusion.Allthesemattersaredeterminedbythelogico-syntacticalrelationsdeterminingwell-formedness,signifiedbyexemplification(whatIcalledthe“internal”semanticrelations).Noparticularsetof“external”semanticrelations(nospecificlexicon,setofcorrelationswithexternalobjects)mustbeknownforthat,althoughsomemust;inthatrespect,theknowledgemightbeconsideredapriori.IhavesummarilysketchedthepicturetheoryofrepresentationthatappearstobepropoundedintheTractatus,theevidenceallegedlysupportingit,andhowitissupposedtodealwithwhatappearstobeitsmainmotivation,providinganaccountofthemodalpropertiesoflogicaltruthsandvaliditiesandouraprioriknowledgethereof.Buttherearegoodreasonstoremainscepticalaboutthisaccount,tosaytheleast.Forstarters,whenoneleavesbehindthetoyexamplesthatWittgensteinconsideredearlyon(suchasthree-dimensionalmodelsofcaraccidents)andmovestotheparadigmcasestowhichthetheoryissupposedtoapply–linguisticrepresentationsinnaturallanguagesandthethoughtstheyexpress–itseemsunbelievablethatthereareanypropertiessharedbytherepresentationandtheobjectstheyareabout.Howcouldidenticalrelationalproperties,nomatterhowabstract,relatelexicalitemstodeterminelogico-syntacticalwell-formedness,ontheonehand,andtheitemstheystandfortodeterminepossiblesituations,ontheother?Agreed,thisisnotobviouslywrong.Wittgensteinmentions,tojustifyhisview,thecaseoftransitiverelationsandthesentencesrepresentingthem(Tractatus,3.1432).Atfirstsight,thesyntacticresourcesthat“accusative”languagesand“ergative”languagesusetorepresenttransitiveeventualitiesareindeedverydifferent.However,somegrammariansarguethat,atasuffi-cientlyabstractlevel,alllanguagesusethesamesyntacticalrelations(Baker,1997).Grantingthis,however,doesnotyettakeustotheclaimthattheverysameabstractsyntacticrelationsareinstantiatedintherepresentedtransitiveeventualities.Asidefromthis,thetheoryappearstobeplainlyfalse,andthereforeactuallyunabletoprovidetheexplanationspredicatedofit.Ifthepicturetheoryweretrue,atmostelementarylogicalvaliditieswouldbenecessary,andknownapriori.Butmodalintuitionsasstrongasthoseestablishingthenecessityandapriorityofelementarylogicalvaliditiescreditthesamemodalstatustoredisacolourornothingcanbeentirelyredandentirelygreen,andthesuggestionsbyWittgensteintodealwiththesecasesonbehalfofhistheoryleadnowhere;nottomentionhissuggestionsofhowtodealwithmathematicaltruths,orallegedphilosophicaltruths,likethepicturetheoryitself.AndtherealsoareKripkeanexamplessuchasthenecessity,givenitstruth,thatwatercontainsoxygen,alsoestablishedbycompellingmodallyrelevantintuitions(moreonthembelow).Nonetheless,eventhoughthepicturetheorystandsasrefuted69781472578235_txt_print.indd608/08/201415:45\nEditorialIntroduction:HistoryofthePhilosophyofLanguageasanyphilosophicalviewmightbe,onecanseehowitissupposedtoaccountforsomephilosophicallyrelevantdata;and,insodoing,itdrawsattentiontothedata:theremustbeaphilosophicalaccountoflogicalvalidity,whichshouldexplain,oratleastexplainaway,the“sublimity”oflogic–ouraprioriacquaintancewithmodalrealitymanifestinthiscase;suchanaccountshouldrelyonaphilosophicalaccountofintentionality;aphilosophicalaccountofintentionalityshouldexplainourcapacitytounderstandnewthoughts,andourcapacitytounderstandfalsethoughts.Somepsychologistsareprouderofdiscovering“effects”(unexpecteddataforanytheorytoaccountfor)thanofthetheoriestheyputforwardtoaccountforthem:thetheorieswillprobablybesuperseded,whiletheeffectswillprobablyremain.Asimilarattitudemightwellprevailinphilosophy.ThepicturetheoryhighlightswhatinmyviewmakestheTractatusimportant,whichistheconglomerateofphilosophical“effects”justmentioned.Inthesecondsection,Iwillindicatehowthey(andrelatedsuggestionsbyWittgenstein’spredecessors)havebeendevelopedinthecurrentliterature,asdiscussedintheensuingcontributions.Iwillreferthereadertothechaptersinwhichfurtherelaborationcanbefound,expandingonlyonafewissuesnottakenupbyourcontributors.ContemporaryThemesfromFrege,RussellandtheTractatus(i)Reference.GenovevaMartí’schapter,“Reference”,presentsthedebatesthathaveoccupiedcentrestageincontemporaryphilosophyoflanguagebetweenthedescriptivistaccountsrootedintheworkofFregeandRussellandtheNewTheoryofReferenceputforwardsincethe1970sbyphilosopherssuchasBurge,Donnellan,Kaplan,Kripke,PerryandPutnam.HereIwillpresentsomedifferencesbetweenFrege’sandRussell’sformsofdescriptivism–intheTractatus,WittgensteinhailsRussell’sTheoryofDescriptionsasaphilosophicalturningpoint,adoptingtheRussellianview.ThecoreclaimoftheTheoryofDescriptions(cf.Neale’s(1990)excellentdiscussion)isthat,inatleastoneoftheirsemanticfunctions,definitedescrip-tionssuchas“theKingofSpain”or“myfather”makecontributionstothecontentsexpressedbysentencesinwhichtheyoccuranalogoustothoseofquantifierssuchas“every”or“some”,andcontrastingwiththoseofgenuinelyreferentialexpressions,suchassomepropernamesandindexicals.Russellhimselfmadethepointbycontendingthatdescriptionsare“incompletesymbols”which,havingmerely“contextualdefinitions”,lackameaningoftheirown,anddisappearonanalysis;butthiswasjustaproductofthetheoreticaltools–theformalsystem–bymeansofwhichhepresentedtheview.Thus,considerasentencesuchas(1):79781472578235_txt_print.indd708/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguage(1)TheKingofSpainistall.ThewayRussellputit,themainclaimofthetheoryisthat,inatleastoneofitssemanticinterpretations,thissentenceexpressesacontentequivalenttotheoneexpressedby(2)–amoreorlessstrainednaturallanguageequivalentof(3),(1)’sformalizationinthesortofformalsystemRussellwasusing,assumingtheobvioustranslationkey.(2)SomeoneissuchthatheisKingofSpain,thereisnoKingofSpainotherthanhim,andheistall.(3)$x(Kx∧¬$y(Ky∧y≠x)∧Tx)Indeed,in(2)thedefinitedescriptionhasvanishedasaspecificconstituent,distributedintoquantifiers,negation,andtheidentityrelation.However,asNealeexplains,thisaspectofRussell’sviewcanbeshowntobeidiosyncraticbypresentingthecoreofRussell’stheorybymeansofadifferentformalsystem.Incontemporarysemantics,quantifiersareanalysedintheframeworkofthetheoryofGeneralizedQuantifiers.IreferthereaderagaintoNeale(1990)foradditionaldetailsandreferences;JoshDever’schapterinthisbook,“FormalSemantics”,hasausefulintroductiontotheuseofformalframe-worksinsemantictheorizing,and,inSection7,furtherinformationabouttheGeneralizedQuantifiersframework.Inanintuitiveversionofthisframework,quantifierssuchas“every”and“some”contributetoexpressquantityrelationsbetweentheclassesofobjectstowhichtwopredicatesapply.Forinstance,“somewritersmokes”expressestheclaimthattheclassofwritersandtheclassofsmokersshareatleastoneobject;and“everywritersmokes”,theclaimthatthedifferencebetweentheclassofwritersandtheclassofsmokershasnomembers.Oneadvantageofthisframework,relativetotheoneRussellwasusing,isthatitallowsustoaccountforothersimilarexpressions,suchas“few”,“most”,“many”,etc.Inthisframework,whatItaketobethecoreofRussell’stheorycanbeputlikethis:inatleastoneofitssemanticfunctions,“the”isanexpressioninthegeneralcategoryofdeterminers,includingalso“every”,“some”,“most”,“few”,“many”etc;whenitoccursinsentencesoftheformthePQ,ithelpstomaketheclaimthattheclasstowhichPapplieshasjustonemember,anditisfullyincludedintheclasstowhichQapplies.Putinthisway,descriptionsdonotdisappearafteranalysis:inthesemanticanalysis,“theP”isasmuchaspecificconstituentas“everyP”in“everyPQ”.WhatremainsiswhatItaketobethecoreclaimofaRussellianTheoryofDescriptions;torepeat:inatleastoneofitssemanticfunctions,definitedescriptionscontributetomakinggeneral,quantificationalclaims,exactlylikequantificationalexpressionssuch89781472578235_txt_print.indd808/08/201415:45\nEditorialIntroduction:HistoryofthePhilosophyofLanguageas“everychild”do,incontrastwiththesingularclaimsmadewiththehelpofgenuinereferentialexpressionssuchassomepropernamesandindexicals.Beforemovingontoexplainwhatthiscontrastmightbebetweenmakingsingularandgeneralclaims,IneedtoelaborateonafewissuesIhavepassedbyquicklyinthepreviousparagraphs.Inthefirstplace,Ihavebeenspeakingofatleastoneofthesemanticfunctionsofdefinitedescriptionsbecause,asweareabouttoseeoncewehavesaidmoreaboutthedifferencebetweensingularandgeneralterms,theRussellianshouldallowforthepossibilitythatdefinitedescriptionsalsohaveareferentialfunction.Russellhimself,andmanyRussellians,rejectthatview;butthecoreRussellianclaim,Itakeit,isonlythatdescriptionsbehavelikequantifiersinatleastoneoftheirsemanticuses.ThesecondwarningIneedtomakeatthispointisthatIhavebeenignoringissuesofcontext-dependence.Thus,“tall”in(1)isacontext-dependentexpression:whatcountsasbeingtallinacontextdiffersfromwhatcountsassuchinothercontext.Also,forthepredicate“KingofSpain”withwhich“the”formsthedefinitedescriptionin(1)toapplytojustoneobject,somehiddencontext-dependencemustbepresumed;itmightbethatthepredicateissomehow“presentKingofSpain”,orthatquantificationalexpressionssomehowpresupposeacontextuallygiven“domainofdiscourse”.TheotherexampleofdefinitedescriptionImentioned,“myfather”,ismoreobviouslycontext-dependent.KentBach’schapter,“ContextDependence”,discussesthisissueingeneral,andDever’schapter,“FormalSemantics”,describeswaysforformaltheoriestoencompassthephenomenon.Letusgobacknowtothecontrastbetweengeneralandsingularclaims.FollowingKripke(1980,14),byrelyingontheTractarianviewthatacrucialcomponent(ifnotthewhole)ofthecontentsofsentencesandthoughtsthatwegrasparetheirtruth-conditions(thewaytheydiscriminatebetweenpossibilities,thoserelativetowhichtherelevantcontentwouldobtainfromthoserelativetowhichitwouldnot)wegetthefollowingcharacterization.Whenweconsiderdifferentpossibilitiesforageneralclaimsuchas“everywritersmokes”tobetrue,thesmokingwritersinsomeofthemmightwelldifferfromthoseinothers;allthatmattersisthatallwritersineachpossiblestateofaffairssmoke.Thesameappliestodefinitedescriptionssuchas“thefirstSpaniardtowintheTourdeFrance”,inthesensethatRussell’sTheoryofDescriptionscaptures.Thefalsesentence“thefirstSpaniardtowintheTourdeFrancewasborninCuenca”iseasilyintuitivelyunderstoodinsuchawaythatitselectspossibleworldswhereF.M.Bahamontes,theactualfirstSpaniardtowintheTourdeFrance,wasborninCuencaratherthanbeingborninToledoasinfacthewas,butitalsoselectspossibleworldswheretheactualsecondSpaniardinwinningtheTour,L.Ocaña,whowasactuallyborninCuenca,isinfactthefirstSpaniardtowintheTour.Inotherwords,thepersonsatisfyingthedescriptionmightdifferfrompossibilitytopossibility,amongthosewherethecontentobtains.99781472578235_txt_print.indd908/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageHowever,onlyworldsofthefirstkindareselectedby“F.M.BahamonteswasborninCuenca”.Inthissense,definitedescriptionsarenot“rigiddesignators”:theypickoutdifferentindividualswithrespecttodifferentworlds;butpropernamessuchas“F.M.Bahamontes”(andindexicalssuchas“thisman”,utteredpointingtothesameperson)are.8These“intuitionsofrigidity”,thefactthatwhenweconsiderpossiblestatesofaffairscompatiblewiththetruthofagivenutterancewekeepfixedthedenotation,ifany,ofthereferentialexpressionintheactualstateofaffairs,isthemostimportantmarkdistinguishingsingularfromgeneralclaims.Kripkepointedoutthat,aswejustconfirmed,“wehaveadirectintuitionoftherigidityofnames,exhibitedinourunderstandingofthetruthconditionsofparticularsentences”(Kripke1980,14;cf.6,62).Russellmighthavebeensensitivetothisintuition.Inthefamouschapter“KnowledgebyAcquaintanceandKnowledgebyDescription”inTheProblemsofPhilosophy(p.30),afteraskingustoconsidertheusethatBismarckhimselfmakesof“Bismarck”torefertohimself,hesays:“Herethepropernamehasthedirectusewhichitalwayswishestohave,assimplystandingforacertainobject,andnotforadescriptionoftheobject.”Russelliscontrastingherea“direct”usethatnames“wish”tohave,incasesinwhichweunderstandthembybeing“acquainted”withtheircontents(asBismarckiswithhimself),withthedescriptiveonetheymostofthetimehave,accordingtohim,forreasonsweareabouttosee;this“directuse”mightwellbethatrigidusethatourintuitionsaboutthetruthconditionsofparticularsentencesreveal,accordingtoKripke.InRussell’sdirectuse,thenamesimplystandsforthebearer;thebeareristhename’scontent:nowonderthat,whenweconsiderpossiblesitua-tionsrelativetowhichsentencesincludingthenamearetrue,allthatwehavetoexamineishowthingsstandwiththebearerineachsituation.Mostofthetime,however,propernamessuchas“Bismarck”(forinstance,whentheyareusedbypeopleotherthanBismarckhimself)expressaccordingtoRussellthecontentsofdefinitedescriptions;understandingtheminvolvesthatdescriptiveknowledge,andnotanacquaintancewiththeirreferentsthat,forRussell,wouldbeimpossibletohaveinthatcase.9Whyisthis?AmainepistemologicalconsiderationforRussellandWittgenstein–letusrefertoitaspotentialwreck–isthat,unlikewhenusedbyBismarckhimself,auseof“Bismarck”cannotguaranteetheexistenceofareferent.Inthe“direct”use,however,thereferentisthemeaning;withoutreferenttherewouldbenomeaning,forthenameorforthesentencesincludingit.Intuitively,however,evenifBismarckwereamassivehoaxandinfacttherewasnoBismarck,“BismarckwasPrussian”ismeaningfulinourmouths.Anotherepistemo-logicalconsideration(aspectualbias)isthatnameswiththesamereferent,suchas“Hesperus”and“Phosphorus”(orasinglename,fordifferentusers),mightbeassociatedwithdifferentpurportedlyidentifyingaspectsoftheintendedreferent,sothatreplacingonewiththeotherinasentencemightintuitively109781472578235_txt_print.indd1008/08/201415:45\nEditorialIntroduction:HistoryofthePhilosophyofLanguagealteritssignificance.Aspectualbiassuggeststhatintherelevantcasesthereferentcannotbeallthatthereistothemeaningofareferentialexpression;potentialwreck,thatthereferentisnotevenaconstitutivepartofit.ThereaderwillfindinGenovevaMartí’schapterconsiderationsspeakinginfavourofRussell’sformofdescriptivism(fundamentally,thewayitaccountsforaspectualbiasandpotentialwreck),andalsothealmostdecisivereasonsofferedbyKripkeandothersagainstit.ItisworthnoticingatthispointthatRussell’s(andWittgenstein’s)descriptivismwasnotexposedtotheproblempresentedinMichaelNelson’schapter,“IntensionalContexts”,Strawson’sreduplicationargument–that,intuitively,referencewouldbeunacceptablyindeterminategivendescriptivism,forwecannotexcludethattherearequali-tativeduplicatesofourintendedreferents.For,aswehaveseen,bothRussellandWittgensteinacceptedthethesisofdirectreference(thatthereferentexhauststheterm’smeaning-contribution;seealsoMartí’sandNelson’schaptersforfurtherclarification)forsomeexpressions,whichtheythoughtwerenotsubjecttothetwoconcernsofaspectualbiasandpotentialwreck.EntitiesinthiscategorythatRussellmentionsasobjectsofacquaintanceanddirectreferenceincludetheself(untilWittgensteintalkedhimoutofit),sensedata,andtheiruniversalqualities.10Wittgensteinhimselfiscagier,speakingmerelyofsimpleswithoutgivinganyexample;but,onthebasisofsomeofthelatterremarksintheTractatus,hiscriticaldiscussionofhispreviousviewsontheearlysectionsofthePhilosophicalInvestigations,andotherindirectmaterial,Ithinkitisclearthathealsohadinmindsensedataandtheirattributes.Inanycase,ifthedescriptionsofentitiessuchasBismarckareallowedtoincludedirectlyreferentialexpressionstoaself,ortotheparticularsensedataaselfisawareof,reduplicationwillnotposeaproblem;referencewouldbedeterminate,foritwouldnotbesolelybasedonqualitativeidentification.TheKripkeanargumentsthatMartípresents,however,stillshowhowimplausiblethisRussellian–Wittgensteiniandescriptivistconceptionofreferenceis.Structurally,thealternativeviewsproposedbypartisansofNewTheoriesofReferencedonotdiffermuchfromtheonewehavejustdescribed;itismostlytheepistemologythatchanges.Someexpressions(mostpropernames,including“Bismarck”inallofitsuses,indexicals)referdirectly;theirreferentsaretheircontents,andforspeakerstounderstandthemtheymustbeacquaintedwiththeircontents–butacquaintanceisnowconceivedonthebasisofamorelenientepistemology,allowingforacquaintancewiththespatiotemporallyremote.Notallpropernamesarelikethat,however;onmostofthoseviews,“descriptive”names(suchasGarethEvans’s“Julius”,whichbydefinitionreferstowhoeverinventedthezipfastener,ifanybodyuniquelydid)canbeunderstoodwithoutanyacquaintancewiththeirreferents,andarethereforeexcludedfromthepicture,evenifpartisansoftheseviewsareunclearaboutwhattheirsemanticsis.ThereaderisreferredtoMartí’sandNelson’s119781472578235_txt_print.indd1108/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguagechapterstofindouthowaspectualbiasandpotentialwreckaresupposedtobehandledbyNewTheorists.AlthoughFrege’spictureisalsomotivatedbythesetwoproblems,andcanbenaturallyclassifiedasdescriptivist,itinfactdiffersimportantlyfromRussell’s.Thedifferenceoriginates,Ithink,inthefactthatFregefocusedontheotheruseofdefinitedescriptionsIhadinmindwhen,inpresentingabovetheTheoryofDescriptions,Ispokeof“atleastoneoftheuses”ofdescriptions.Supposethat,intellinganepisodeinthebiographyofF.M.Bahamontes,Isay“andso,thefirstSpaniardtowintheTourdeFrancemighthavebeenborninCuenca”.ItisclearinthecontextthatIamusing“thefirstSpaniardtowintheTourdeFrance”asamerelyrhetoricalalternativeto“F.M.Bahamontes”or“he”,toavoidboringrepetitionsofthatname,presupposingthatmyaudienceisfullyawareofthefactthatBahamonteswasinfactthefirstSpaniardtowintheTourdeFrance.ThisisacaseofwhatDonnellan(1966)calls“referentialuse”ofdescriptions.TherehasbeenadebateconfrontingstrictRussellians,forwhichtheseusesaremerely“pragmatic”,perhapstobeaccountedforinthemodelofGriceanconversationalimplicatures(cf.Kripke1977),thoseforwhomdefinitedescriptionsarejustsemanticallyambiguousbetweentherefer-entialandquantificationaluses,and“contextualists”whorejectthedichotomyasbothpartiesinthedebateunderstandit.FrançoisRecanati’schapter,“Pragmatics”,introducesthereadertothesedebates.Forourpurposes,however,weonlyneedtokeepinmindboththatthequantificationalusesthatRussell’stheoryaccountsforundoubtedlyexist,andthatreferentialusesalsoexist,bethey“semantic”or“pragmatic”.Now,thereadermightcheckhisorherintuitionstoestablishthatinthereferentialcasetheexpressionworks“rigidly”:allthepossibilitieswithrespecttowhichmyclaimwouldbetrueconcernF.M.Bahamontes,andnotanybodyelsewhohappenstobeaccordingtothosepossibilitiesthefirstSpaniardtowintheTourdeFrance.Fregehadasemanticcategoryofpropernames(“Eigenname”)inhissystem,includingordinarypropernames,descriptionsandindexicals;itseemscleartomethathewasthinkingofreferentialusesofdescriptionsastheparadigmcase.ThisleadstoaviewratherdifferentfromtheRusselliandivisionbetweenexpressionsunderstoodbyacquaintancewiththeircontents,andexpressionsunderstoodbydefinitionalsynonymywithgeneralexpressions,evenifitissimilarlydescriptivist,andmotivatedbythesameproblemsofaspectualbiasandpotentialwreck.11Themaindifferenceliesinthat,insteadofadichotomyoftypesofreferringexpressions,theFregeanproposalhasadichotomyofsemanticfeaturesforreferentialexpressions(infactforallexpressions).Theproblemofaspectualbiasisdealtwithbyascribingtoreferentialexpressionsadescriptivesense,inadditiontothereferent(“Bedeutung”intheoriginalGerman).Theproblemofpotentialwreckisdealtwithbyclassifyingcasesofreferencefailureas129781472578235_txt_print.indd1208/08/201415:45\nEditorialIntroduction:HistoryofthePhilosophyofLanguagesomehowderivativefromcasesofreferencesuccess.Themoststraightforwardversionofthisproposal,duetoEvans(1982)andMcDowell(1977),wouldbea“disjunctivist”account,onwhichbothcasesareessentiallydifferent,thereferentbeinganessentialcomponentofthemeaningofasuccessfulrefer-entialexpression;butthereareother,lessradicalvariants(cf.Sainsbury2005).ProposalsalongtheselinesmaystillbesubjecttoKripkeancriticisms,thistimenotofviewsonwhichreferentialexpressionsaresynonymouswithdescrip-tions,butratherofviewsonwhichdescriptionsaresupposedto“fixtheirreferents”.The“two-dimensional”semanticsmentionedinthenextepigraphprovidesanotherframeworkforalternativeneo-descriptivist,neo-Fregeancontemporaryperspectivesonreference.IreferthereaderagaintoMartí’sandNelson’schaptersforafullerappraisalthancanbeundertakenhere.(ii)Meaningandmodality.Aswehaveseen,intheTractatusWittgensteinwascentrallyconcernedwithaccountingforthemodalpropertiesoflogicaltruths:howitisthattheyarenecessary,andhowitisthat,consistentlywithourpoten-tiallyknowingtheirmodalstatus,wecancometoknowthem.Accordingtohisownremarkslater,thepicturetheoryisintendedtoprovidetherequiredexpla-nationsinawaythatallowsfortheobjectivityofthemodalstatusofnecessarytruths,i.e.,itavoidscharacterizingthemas“mind-dependent”inanyway.Inspiteofthefailureofthepicturetheory,hewassuccessfulinconvincingphilosophersoftheimportanceofthetopic;andhislogicalpositivistfollowers,suchasCarnaporAyer,wereinfluencedbyhissuggestionthatthewaytoapproachtheissuewasthroughatheoryofintentionality.Theygaveitananti-realist,conventionalisttwist,however.Theviewnowwas,inanutshell,thatthesemanticrulesofagivenlanguage,whichhaveaconventionalstatus,determinethespaceofpossibilitiesandwithitthenecessarytruths,whichcanbeknownbyknowingthoserulesandaretothatextentknowableaprioriandanalytic.AlbertCasullo’schapter,“Analyticity,Apriority,Modality”,carefullyexplainstheseconcepts,providingimportantdistinctionsandclarifications.HepresentsAyer’sandCarnap’sview,themainchallengestoitintheworkofQuineandKripke,andcriticallyevaluatesthesechallenges.Toalargeextent,recentdebatesaboutthesemattershavefocusedonthescepticismaboutourmodalknowledgethatKripke’sviewsinparticularmightengender,andtodevelopmentsofthe“two-dimensional”suggestionsthatKripke’sownworkalreadyintimates,whichCasullo’schapteralsohelpfullyintroduces.Aswehaveseen,Kripke(1980)arguedthatreferentialexpressionssuchasindexicalsanddemonstratives,propernamesandnaturalkindtermsaredejurerigiddesignators–expressionsthatdesignatethesamethingwithrespecttoeverypossibleworld.Thisfeaturedistinguishesthemfromothersingulartermssuchasdefinitedescriptions,which(puttingasidereferentialuses)mightalsobehavedefactoasrigiddesignators,butdejurearenotso.139781472578235_txt_print.indd1308/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageKripkewaswellawarethathisproposalscreatedaphilosophicalpuzzle.Hisviewaboutreferentialexpressionsandalethicmodalitiesentailstheexistenceofmodalillusions:truthsthatareinfactnecessaryappeartobecontingent.Paradigmcasesareinstancesoftheschemaifnexists,nisF,witharigiddesig-natorintheplaceof“n”andapredicatesignifyingahiddenessentialpropertyofitsreferentintheplaceof“F”.Forthesakeofillustration,letusreplace“F”intheschemawith“is-identical-to-Hesperus”and“n”with“Phosphorus”:(1)IfPhosphorusexists,Phosphorusis-identical-to-HesperusTheexistenceofthosemodalillusionselicitedbyKripke’sviewsaboutrefer-entialexpressionsandalethicmodalitiesispuzzlinginthelightofanothercompellingviewabouttheepistemologyofmodality:thatwehaveareasonablyreliableaccesstopossibleworlds.Kripkesuggeststhis(tome,atleast)whenhestatestheintuitionthatapossibleworld“isn’tadistantcountrythatweare…viewingthroughatelescope…‘Possibleworlds’arestipulated,notdiscoveredbypowerfultelescopes”(Kripke1980,44);“thingsaren’t‘foundout’aboutacounterfactualsituation,theyarestipulated”(op.cit.,49).Ofcourse,accordingtoKripkehimselfwearenotfreetostipulateanypossibleworldwewantintoexistence;otherwise,itwouldmakelittlesensetospeakofmodalillusions,suchasthosepreviouslydescribed.Whatthedichotomyofstipulationvs.discoveryrathersuggestsisthatwehaveaprimafaciereliableaccesstomodalreality–that,primafacie,whatweconceiveaspossibleispossible.Thispuzzleisnotanoutrightparadoxconstitutedbycontradictoryclaims;thatonehasingeneralreliableaccesstothemodalrealmallowsformistakenmodalimpressions.However,Kripke’sviewssuggestthatmodalillusionsdonotariseonlyinafew,systematicallyunrelatedcases;onthecontrary,asystematicandfar-reachingpatternispredicted.Tosustainmodalreliabilismrequiresthusaphilosophicalaccountoftheillusionsconsistentwithit.Kripkeissensitivetothis,and,inhischaracteristicallynuanced,cautionarymood,heprovidesone:“Anynecessarytruth,whetherapriorioraposteriori,couldnothaveturnedoutotherwise.Inthecaseofsomenecessaryaposterioritruths,however,wecansaythatunderappropriatequalitativelyidenticalevidentialsituations,anappropriatecorrespondingqualitativestatementmighthavebeenfalse”(Kripke1980,142).Incasessuchas(1),somethingmorespecificcanbesaid:Inthecaseofidentities,usingtworigiddesignators,suchastheHesperus-Phosphoruscaseabove,thereisasimplerparadigmwhichisoftenusabletoatleastapproximatelythesameeffect.Let“R”and“R”bethetworigid12designatorswhichflanktheidentitysign.Then“R=R”isnecessaryiftrue.12Thereferencesof“R”and“R”,respectively,maywellbefixedbynonrigid12149781472578235_txt_print.indd1408/08/201415:45\nEditorialIntroduction:HistoryofthePhilosophyofLanguagedesignators“D”and“D”,intheHesperus-Phosphoruscasethesehavethe12form“theheavenlybodyinsuch-and-suchpositionintheskyintheevening(morning)”.Thenalthough“R=R”isnecessary,“D=D”maywellbe1212contingent,andthisisoftenwhatleadstotheerroneousviewthat“R=R”12mighthaveturnedoutotherwise.(Kripke1980,143–4)WhatKripkeproposeshere,cautiously,onlyasapossiblemodelapplyinginsomecases,istheblueprintfortwo-dimensionalaccounts;thecentralideaisthat“anappropriatecorrespondingqualitativestatement”,differentfromtheoriginal,necessaryone,which,unlikethis“mighthavebeenfalse”,issomehowmixedupwithit,thusengenderingtheillusionofitscontingency.Kripkerefrainsfrommakinggeneralclaimsabouttheapplicabilityofthismodel.Nevertheless,hisinfluentialargumentsagainstmind–bodyidentitylaterintheNamingandNecessitylecturesdependessentiallyonthepremisethatthemodelistheonlyavailableonethatproperlyexplainsthefactsatstake.Thiscoretwo-dimensionalistideacanalsobeinvokedtodealwiththeotherpuzzlingKripkeancategoryofthecontingentapriori,althoughKripke’sindicationsaboutthisapplicationarelessclear.Ashealsofamouslynoted,ifonestipulatesthatadesignatorNistobeusedtorefertoanobjectintroducedbyadescriptionDthatthusfixesitsreference,onecanbesaidtoknowtherebyapriori“insomesense”(op.cit.,63)thetruthofthecorrespondingstatement“NisDifNexists”;(2)providesanexample,correspondingto(1):(2)Phosphorusiswhateverappearsasshiningbrightlyintheeastjustbeforesunrise,ifitexists.Toapplythemodelhereweshouldhavethat,althoughwhat(2)saysisacontingentproposition,thereis“anappropriatecorrespondingqualitativestatement”whichexpressesanecessaryone.ThiswouldprovideforthepartialrescuethatKripke(1980,63fn.)envisagesforthetraditionalviewthatevery-thingaprioriisnecessary.Kaplan(1989)hadsuggestedrelatedideas,forspecificexamplesofthecontingentaprioriinvolvingindexicals,like“Iamherenow”or“Iamtheutterer”;KentBach’schapter,“Context-Dependence”,discussesthem.Kaplaninvokedhisdistinctionoftwodifferentsemanticfeaturesofcontext-dependentexpressions,indexicalssuchas“I”,“here”and“now”inparticular,acharacterthatcapturesthestandingmeaningoftheexpression,andacontentthatconsistsoftheirtruth-conditionalcontributioninparticularcontexts.Givenaparticularcontext,sentencessuchas“Iamherenow”expressacontingentcontent;however,theyare“character-valid”inthatexpressionsinthemhavecharacterssuchthattheywillalwaysexpresstruthswhenutteredinanycontext.159781472578235_txt_print.indd1508/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageFinally,Kripkesuggestedthattheavailabilityof(whatIampresentingashisblueprintfor)thecoretwo-dimensionalistexplanationofthenecessaryaposterioriandthecontingentapriorisuppliesanimportantroleforconceptualanalysis,compatiblewiththeAristotelian-essentialistviewthattherearederenecessitieswhichcanonlybeknownthroughempiricalresearch:Certainstatements–andtheidentitystatementisaparadigmofsuchastatementonmyview–iftrueatallmustbenecessarilytrue.Onedoesknowapriori,byphilosophicalanalysis,thatifsuchanidentitystatementistrueitisnecessarilytrue…Allthecasesofthenecessaryaposterioriadvocatedinthetexthavethespecialcharacterattributedtomathematicalstatements:philosophicalanalysistellsusthattheycannotbecontin-gentlytrue,soanyempiricalknowledgeoftheirtruthisautomaticallyempiricalknowledgethattheyarenecessary.Thischaracterizationapplies,inparticular,tothecasesofidentitystatementsandofessence.Itmaygiveacluetoageneralcharacterizationofaposterioriknowledgeofnecessarytruths.(Kripke1980,159)Kripke’sandKaplan’ssuggestionsweretakenupanddevelopedintechni-callysystematicwaysinthetwomostinfluentialarticlesoriginatingthetwo-dimensionaltraditionafterKripke’sinauguratingconsiderations,Stalnaker’s(1978)“Assertion”andDaviesandHumberstone’s(1980)“TwoNotionsofNecessity”.Otherwriters,includingChalmers(1996),Jackson(1998,chs1–3)andPeacocke(1999,ch.4)inparticular,havesubsequentlyelaboratedontheidea.(iii)CompositionalityandSemanticTheorizing.Aswesaw,followingsuggestionsinFregeandRussellaswellashisowninsight,Wittgensteinhighlightedtwopiecesofdatathatanyphilosophicalaccountofrepresentationalsystemssuchasnaturallanguageshouldcapture:thefactthatwecanunderstandsentenceswehaveneverencounteredbefore,andthefactthatinunderstandingdeclar-ativesentences(thosesusceptibleofevaluationastrueorotherwise)wetypicallygrasppossibilitiesthatneednotbeactual(truth-conditions).Bothissuesinformtwoofarguablythemostsuccessfulandinfluentialprogramsforunderstandingnaturallanguages,initiallypropoundedinthe1960sandleadingtoworkthatisstillflourishingtoday.OneistheChomskianprograminlinguistics;theotheristhetraditionofformaltruth-conditionalsemanticsstartedinslightlydifferentdirectionsbyresearcherssuchasRichardMontagueandDonaldDavidson.Recentmanualsintroducingtheverysubstantialexplan-atoryachievementsoftheMontagovianandDavidsoniantraditionssuchas,respectively,HeimandKratzer(1998)andLarsonandSegal(1996)showtheextenttowhichtheChomskianprograminlinguisticsandtruth-conditional169781472578235_txt_print.indd1608/08/201415:45\nEditorialIntroduction:HistoryofthePhilosophyofLanguagesemanticsineachofthoseversionsareconvergingnowadaysintoafruitfulresearchprogram.JamesHigginbotham’schapter,“TheNatureofLanguage”,togetherwiththeonesalreadymentionedbyJoshDever,“FormalSemantics”,andKathrinGlüer,“TheoriesofMeaningandTruthConditions”,willprovidethereaderwithfurtherelaborationandreferencesonthesematters.(iv)SemanticsandPragmatics.Asalreadymentioned,Wittgensteinconcen-tratedintheTractatusontherepresentationalpropertiesofwhathecalledsayings,whosecorehetooktobetruth-conditions.Materialfromhislecturesandconversationsinthelate1920sandearly1930sshowsthathewaswellawarethat,aspartofourmasteryoflanguage,wedonotmerelydeploydeclarativesentencessusceptibleoftruthandfalsity,butalso,say,inter-rogativeorimperativesentencesnotsubjecttotruth-evaluation;wedonotmerelyassertbymeansoflanguage,orperformothersimilarlytruth-evaluablespeechacts,butwealsoask,request,promise,andsoonandsoforth.TheearlyWittgensteinmighthavedismissedthispointwithamovemadeexplicitmuchlaterbyspeechacttheoristssuchasJohnAustinorJohnSearle–workpresentedinFrançoisRecanati’schapter,“Pragmatics”.Thoseotheractsmayalsobeevaluatedintermsof,say,theirfulfilmentorsatis-factionorotherwise,ifnottruthorfalsity.Theircontents,whichwegraspascompetentspeakers,thenencodethesefulfilmentconditions,whichmightbeentirelycoincidentwiththetruth-conditionsofassertions,andwhichposethesametwoproblemsthatWittgensteinhighlighted:wecanunderstandthefulfilmentconditionsofordersweconsiderforthefirsttime,andwegraspthemindependentlyofwhetherornottheyareactuallyfulfilled.Thesemanticundertakingmaythusbecharacterizedaspurportingtosystematicallycharac-terizethesefulfilmentconditions,using“truth”insteadof“fulfilment”inanextendedway.ContentsinthisgenericsenseappeartobewhatWittgensteinmeantbysayings,somethingveryclosetowhatAustinlatercalledlocutionaryacts.Characterizingthenatureofwhat,inadditiontothepotentiallycommoncontentsorfulfilmentconditions,distinguishesthedifferentspeechacts(whatAustincalledillocutionaryacts)isoneofthetaskslefttopragmatics;theearlyWittgenstein,inthisway,didnotoverlookthistaskbutsimplydismisseditasunimportantforwhathetooktobehismainconcern–accountingfortruth-conditions.Somethingsimilarmightbesaidaboutanothertaskusuallylefttopragmatics,context-dependenceingeneralandtheworkingofindexicalsinparticular–butthereadershouldexaminebothRecanati’sandBach’schaptersfordevel-opmentsandimportantreservationsaboutthepurelypragmaticnatureofthemeaningofcontext-dependentexpressions.Fregehadoccupiedhimselfwiththetopic,and,ofcourse,theauthoroftheTractatuswasgreatlyconcernedattheveryleastwiththeway“I”functions.12179781472578235_txt_print.indd1708/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageNotoriously,inhislaterwritingsWittgensteinrejectedhisearlierdismissiveattitudetowardssuch“pragmatic”matters.Inamuchdiscussedprogram-maticpaper,discussedagainhereinbothGlüer’s“TheoriesofMeaningandTruthConditions”andRecanati’s“Pragmatics”,Strawsonspokeofa“Homericstruggle”confronting,accordingtohim,thetruth-conditionalformalsemanti-cistswithanopponentemphasizinginsteadtheconstitutivecharacterinthephilosophicalaccountofmeaningofnotionssuchasintentionalaction,normsorconventions.Thelinchpin,accordingtoStrawson,concernsthepossibilityofasufficientlysubstantiveexplanationofthenatureoftruthandthetruth-conditionsthatthe“formalsemanticist”appealsto.AccordingtoStrawson,suchanaccountcanonlycomefromtherolethenotionhasintheappraisalofspeechactssuchasassertion,whichaccordingtohimgivesthevictorytotheopponent.Inrecentdiscussions,Kripke’s(1982)relatedreconstructionofthelaterWittgenstein’sremarksonthenormativityofmeaninghasbeenveryinfluential;manyresearchershavefocusedinparticularonthebleakconsequencesforprojectssuchasChomskianlinguisticsortruth-conditionalsemanticsthatappeartofollowfromKripke’saccount.JoséZalabardo’schapter,“SemanticNormativityandNaturalism”,takesupthesematters.Onarelatednote,differentwriters–someinfluencedbythelaterWittgensteinandotherphilosophersinthe“Meaning–intention–action”Strawsoniancamp,suchasAustinorGrice,otherssimplyasaresultoftheirpayingcloseheedtotheimplicationsofthemanyformsofcontext-dependencepresentinouruseofnaturallanguages(asdiscussedinKentBach’schapteronthetopic)–haveemphasizedthatthecontributiontowhatsentencessignifyfromasystematic,compositionalsemanticcomponentappearstobeveryabstractandremotefromordinaryintuitions(ifthereisanysuchcontributionatall,whichsomeofthese“contextualist”writerssuchasCharlesTravisdoubt).Recanati’sandBach’schapterswillfurtherpresentthesematterstothereader.ResearchMethodsinthePhilosophyofLanguageInthissummaryofthemesandtopicsfromtheearlyhistoryofanalyticphilosophyoflanguage,Ihaveatseveralpointsmadeappealtointuitions.Forinstance,wesawthatKripkeclaimsthatourintuitiveunderstandingofthedistinctivetruth-conditionsofsentencesincludingpropernamesgivesusadirectintuitionoftheirrigidity;andImentionedthatwehavesimilarintui-tionsabouttherigidityofdefinitedescriptionsinreferentialuses.Similarly,inoneofhismostcelebratedargumentsagainstdescriptivisttheories–discussedinMartí’schapter–Kripkefamouslyelicitsourintuitionsaboutathought-experimentconcerningafictitioussituation,inwhichthepersonusuallyaddressedbythename“KurtGödel”,whosepassportregisteredthatname,189781472578235_txt_print.indd1808/08/201415:45\nEditorialIntroduction:HistoryofthePhilosophyofLanguageetc.,infactdidnotprovetheincompletenessofarithmetic,butstoletheprooffromsomeonecalled“Schmidt”.Underthe(probablytrue)assumptionthatallthedescriptiveinformationweassociatewiththename“Gödel”isthediscovereroftheincompletenessofarithmetic,iftheimaginedcircumstanceswerereal,by“Gödel”wewouldbereferringtoSchmidt.However,intuitivelythisdoesnotseemso,Kripkecontends.Intuitionsaregenerallysupposedtoplayanimportantevidentialroleincontemporaryanalyticphilosophy;particularlyintuitionsconcerningcircum-stancesimaginedinthoughtexperimentssuchastheonejustmentioned,whichactas“crucialexperiments”withrespecttocontrastingtheoriesaboutthenatureofsomeconceptsinwhichphilosophyisinterested(suchasthedescriptivistanddirecttheoriesofreference).Thus,Gettier’s(1963)three-pagearticlefamouslyrefutedtheclaimthatknowledgeisjustifiedtruebeliefbymeansofonesuchthoughtexperiment,describingasituationinwhich(wewouldintuitivelyconcur)someonehasjustifiedtruebeliefwithouthavingknowledge.Similarly,intuitionsalsoplayafundamentalevidentialroleincontemporarylinguistics:intuitionsabouttheacceptabilityorotherwiseofsentencesplaysucharoleinsyntax,andthesortofintuitionaboutthetruth-conditionsofspecificsentenceswesawKripkeappealingtoabove,toestablishtherigidityofpropernames,similarlyplayafundamentalevidentialroleinsemantics.Whatareintuitions?Tosecureforthemacentralevidentialroleinphilosophy,Bealer(1998)takesintuitionstobe“intellectualseemings”–specificmentalstatessuchasperceptualexperiencesareusuallythoughttobe,playinginphilosophyasimilarroletotheoneplayedbyexperiencesinempiricalknowledge:likeperceptualexperiences,theyare“given”,theycannotbejustifiedorunjustified;nonethelesstheyhavejustificatorypower,makingbeliefsbasedonthemimmediatelyjustified(justifiednotbyotherbeliefs),eveniftheycanbeillusory(keepingtheir“pull”orattractionwhileweresistit,asintheMüller-Lyerillusion);theyprovidebasicinformationabouttheintensionofconcepts–theconditionsunderwhichtheyapplyinpossiblecircumstances–confirmingordisconfirmingapriorigeneraltheoriesabouttheirnature(thewayKripke’sthoughtexperimentdisconfirmsthedescriptivisttheoryofreference,andGettier’sthetraditionaltheoryofknowledge),andtherebyestablishingconceptualnecessities,whichiswhatphilosophyisabout.OtherssuchasSosa(2007,ch.3)arguethatthereareimportantdifferencesbetweenexperiencesandintuitions;Sosasuggestsunderstandingtheminsteadasconsciousentertainingsofcontentwhichattractourassent;underthatguise,healsogivesthemanequalepistemo-logicallysalientroleinphilosophy,understoodasafundamentallyaprioridisciplineprovidingtheoriesaboutthenatureoffundamentalconceptssuchasknowledgeorreference.199781472578235_txt_print.indd1908/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageFinally,otherssuchasWilliamson(2007,ch.7)–scepticalofthetradi-tionalconceptionofthephilosophicalundertakingasconceptualanalysisconductedaprioriandseeingphilosophy,inaQuineanlight,ascontinuouswithscienceandordinaryempiricalbelief–refusetounderstandintuitionsasanythingotherthanconsciousopinionsorbeliefsthatwefindappealingforsomereasonorother.AlthoughWilliamsondisqualifiesthemainQuineanargumentsforsuchscepticism–contendingthatnotionssuchassynonymy,meaning,belief,necessityandpossibility,areinasufficientlygoodstanding,evenifwehavetounderstandthemintermsofeachother–heendsuppropoundingasimilarview.Meaning-determiningfactorsare,accordingtohim,factsaboutourlinguisticpracticesanddispositions,atmostempiricallyaccessible(2007,121–30);andthought-experimentsshouldbeunderstoodasaformofreasoningaboutcounterfactualcircumstances,essentiallydependentonpremisesonlyknownaposteriori.“Intuitions”(i.e.,justconsciousjudgmentswefindappealing)doconstituteevidence,buttheevidenceconsistsofthecontentswethusintuitivelyaccept,notthepsychologicalfactthatwehavethoseintuitions.HereWilliamson’sviewisclosetoSoames’s(1984,174)“Platonistic”viewofwhatlinguisticsisabout,andhiscorrespondingviewabouttheroleofintuitionsinthatdiscipline:“evenintuitionsofgrammaticalityarenotdatafortheoriesinlinguistics;whereasfactsaboutgrammaticalityare”;Soamesisassumingherethat“data”is“whattheoriesmakeclaimsorpredictionsabout.”Intuitionsofgrammaticality,orsemanticintuitions,arelikemathematicalorgeometricalintuitions:indications,whichwemusttaketobereliableifwearetohavesomestartingpointatall,ofsomeofthefacts–theonlyrealdata–aboutnumbersorspacethatmathematiciansaimthentocollectunderanencompassingsystemcharacterizingthestructureofnumbersorspace.Numbertheoryisabouttwoplusthreebeingfiveandrelatedintuitivefacts,notaboutourintuitionthatthisisso;similarly,linguistics,onthisview,isaboutabstractlanguages,say,languagesthathave“thecatisonthemat”asagrammaticalsentence,and“somecatisonthemat”aslogicallyfollowingfromit–notaboutourintuitionthatanyofthisisso:theseintuitionsmerelyprovidethefactstobecapturedandsystematizedbythelinguist.Inthecaseoflinguistics,thereis,Ithink,adecisivereasontorejectSoames’sview,andtotakeintuitionsthemselves,notjusttheircontents,asevidencetobeaccountedforbylinguistictheories.ThereasonistheguidingrolethatthePrincipleofCompositionalityplaysinlinguistictheorizing,whichwehavealreadymentionedatseveralpointsabove.ThesyntacticandsemanticstructureassignedbythosetheoriestonaturallanguagesistakentoexplainthefactsthatleadustoacceptthePrincipleofCompositionality.Butthesearefactssuchasourcapacitytounderstandnewsentences,manifestedbytheintuitionsexpressingunderstandingofthose“new”sentences,andtheir209781472578235_txt_print.indd2008/08/201415:45\nEditorialIntroduction:HistoryofthePhilosophyofLanguagegrammaticalacceptability.(ThereaderwillfindusefulHigginbotham’schapteron“TheNatureofLanguage”inthisregard.)Inanutshell:ifthePrincipleofCompositionalityistohaveanytheoreticalbite,itisbecauseonlytheoriesthataccommodateitarecapableofexplainingourhavingspecificsyntacticandsemanticintuitions.Inthecaseofphilosophy,inordertoconfrontWilliamson’sscepticalchallenge,thefriendofconceptualanalysisandtheconceptionofphilosophyasafundamentallyaprioridisciplineprovidingknowledgeofconceptualnecessitiesshouldtrytopointoutthatthehavingofintuitionsthemselves,andnotjusttheircontents,providescrucialdataforphilosophicaltheoriestoaccountfor.Inthiscase,theideatobearticulatedisthatintuitionsaboutcases(inparticular,theinterestingintuitionselicitedbywell-designed“crucial”thought-experiments,suchasKripke’sorGettier’s)arejustmanifestationsofthepossessionoftherelevantconcepts,andconstituteaccesstotheirinten-sions.IreferthereadertoIchikawaandJarvis(2009)forinterestingrecentsuggestionsalongtheselines;butitmustbesaidthatthedifficultiesoftheseattemptshighlighttheimportanceofWilliamson’schallenge.Amoreradicalchallenge–inthat,ifcompelling,itquestionsboththetradi-tionalconceptionofphilosophyandWilliamson’saposteriorimethodologicalalternative–hasbeenrecentlyposedbysomeresultsoftheso-called“experi-mentalphilosophy”.Thisconsistsinthedesignofempiricalexperimentsofthekindpsychologistsregularlyconduct,addressedtoexaminewhetherornotordinarypeoplehavetheintuitionswhicharesupposedtobeelicitedbyphilosophicalthought-experimentssuchasGettier’sorKripke’s.Someoftheresearchersconductingthemarguethat,surprisingly,whetherornotpeoplesharethoseintuitionsappearstobeinfluencedbyfactorssuchasculture,raceorsocialclass.Theseresultswouldquestionthetraditionalapproachtophilosophy(astraditionallyunderstood,themethodwouldatmostprovideinformationaboutculturallyidiosyncraticconcepts)butalsoWilliamson’s,becausetheywouldsuggestthatthecontentstowhichwehaveintuitiveaccessareverydoubtfullyfacts.ThelastsectioninGenovevaMartí’schapterdiscussessomeoftheseexperiments–thoserelatingtotheconceptofreferenceandKripke’sthought-experiment–andprovidesfurtherreferences.Assumingwecanhaveaconvincingreplytotheexperimentalphilosophychallenge,andnomatterwhetherweendupsupportingaformofapriorimethodologyorratherthinkwemustmakesenseofthemethodologyweemployinphilosophyalongthelinesenvisagedbyWilliamson,itseemsplausiblethatweneedtoadopta“widereflectiveequilibrium”viewofthekinddescribedbyDaniels(2011)thus:“workingbackandforthamongourconsideredjudgments(somesayour‘intuitions’)aboutparticularinstancesorcases,theprinciplesorrulesthatwebelievegovernthem,andthetheoreticalconsidera-tionsthatwebelievebearonacceptingtheseconsideredjudgments,principles,219781472578235_txt_print.indd2108/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageorrules,revisinganyoftheseelementswherevernecessaryinordertoachieveanacceptablecoherenceamongthem.”Inparticular,weneedtomakesurethatphilosophicaltheoriesareconsistentwiththeresultsofempiricalscience(thewayaphilosophicalaccountofpersistenceintimeshouldbeconsistentwithSpecialRelativity),andweshouldadmitthatphilosophicaltheoriesaredefeasibleonthebasisofempiricalevidence;Jeshion(2000)providesgoodreasonswhythisshouldbesoevenonthemostaprioristicconceptionofthediscipline.Thus,inthecaseoftheconceptofreferencewehavebeendiscussing,writershaveappealedtofactsabouttheunderstandingbyautisticpeopleorlittlebabiesofreferentialexpres-sions,contendingforinstancethattheyareincompatiblewithdescriptivistviews(cf.García-RamírezandShatz(2011)forarecentexampleofthatlineofargument,andreferencestherein).Reflectiveequilibrium,however–asWilliamson(2007,244–6)pointsout–isnotwhatdistinguishesphilosophy.Asimilarmethodologyisemployedinscience;butwhatischaracteristicofscienceisthewayitdependsonempiricalevidence,andthecriteriaforselectingadequateempiricalevidence.Likewise,inthecaseofphilosophythecrucialissueistherolethatintuitionsplay,whetheritisdistinctiveandwhetheritunderwritesanaprioriknowledgeofnecessarytruths.Onthatmatter,aswehaveseen,thejuryisstillout.Notes1.IamgratefultoMaxKölbelforcommentsonapreviousversionthatledtoimprovements.FinancialsupportformyworkwasprovidedbytheDGI,SpanishGovernment,researchprojectFFI2010-16049andConsolider-IngenioprojectCSD2009-00056;throughtheawardICREAAcademiaforexcellenceinresearch,2008,fundedbytheGeneralitatdeCatalunya;andbytheEuropeanCommunity'sSeventhFrameworkProgrammeFP7/2007–2013undergrantagreementno.238128.2.DepartamentdeLògica,HistòriaiFilosofiadelaCiència,UniversitatdeBarcelona,email:m.garciacarpintero@ub.es3.TherearemanyindicationsofthecentralityofthisissueamongtheproblemsthatWittgensteinwasconcernedwithatthetime,beginningwiththeamountofdiscussiondevotedtoitintheTractatusitself.TheearlylettersandnotebooksreflecthowhisfocusonintentionalityevolvedfromhisprimaryinterestingivinganaccountoflogicalvalidityimprovingonthoseputforwardbyFregeandRussell.AnotherpieceofevidencecomesfromtheInvestigations.Theearlyhundred-oddsectionsofthatworkreadlikeacriticismoftheTractarianphilosophy.Whileproviding,toserveasafoil,someglimpsesofhisnewviewsontheissues,WittgensteincriticisesthereseveralaspectsoftheTractatus,inanorderthatappearstoretraceinreversetheintellectualpathleadingtothem:thedisregardforthedifferencesbetweenillocutionaryforces(towhichwewillcomebacklater)anditsfocusingonlyonwhatIwillcallsayings;thenotionofalogicalname,andthecorrelativenotionofasimple(towhichwewillalso229781472578235_txt_print.indd2208/08/201415:45\nEditorialIntroduction:HistoryofthePhilosophyofLanguagecomeback);theassumptionofauniqueanalysisofallcontents,andsoonandsoforth.Afterallthis,in§65wearetoldthat“thegreatquestionwhichliesbehindalltheseconsiderations”istheproblemofgiving“thegeneralformofpropositions”,forwhichthepicturetheoryprovidedtheintendedanswer.Therefollowsatherapeuticbashingoftheassumptionssettingthisasanissue,andfinallywearetold,in§89,thattheproblemtowhichtheprecedingconsid-erationsleadis“Inwhatsenseislogicsomethingsublime?”4.ForWittgenstein,asforKantbeforehim,thesepropertiesaremanifestlycoextensive;AlbertCasullo’schapter,“Analyticity,Apriority,Modality”,discussesthesenotions.5.ThecentralelementofthepicturetheorythatWittgensteinmarshalstoaccountforthesefeaturesistheclaimthatthepictureanditsrepresentedrealityshareacertain“form”.Thus,regardingthefirstexplanatoryissue,2.17saysthat,inorderforapicturetobeabletodepictrealitycorrectlyorincorrectly,it“musthaveincommon”withrealityitsform;regardingthesecond,4.02saysthatwecanseethatapropositionisapictureofreality“fromthefactthatweunderstandthesenseofapropositionalsignwithoutitshavingbeenexplainedtous”.(Iassumethatthedemonstrative“this”occurringin4.02referstomaterialin4.01,theparagraphimmediatelyprecedingitinthe“alphabetic”orderindicatedbytheirnumbers.)6.Thewayasamplesignifiesasetofpropertiesbyitselfinstantiatingthoseveryproperties.7.KathrinGlüer’schapter,“TheoriesofMeaningandTruth-Conditions”,developsthenotionfurther.8.Infact,theyaredejureso,Kripkecontends,unlikedescriptionssuchas“theevennumber”,whichmerelydefactopickoutthesameentitywithrespecttoeverypossibility.Kripkeneverexplainswhathemeanswiththedejure–defactodichotomy;whatitsuggestsisthatpropernamesarerigidasamatterofthesemanticnormsorrulesgoverningthem.9.SeeSainsbury(1993/2002)foranuancedexaminationofRussell’sactualviews,incontrastwithwhatitisattributedtohimincontemporarydiscussions.10.Wehavebeendiscussingsingulartermssofar,butourconsiderationsapplytoexpressionsinothercategories.Ifwethoughtthatcommonnounssuchas“water”and“tiger”,orpredicatessuchas“electricallycharged”,“yellow”or“circular”,signifyobjectivekindsorproperties,whosenatureistobediscerned(totheextentthatitis)onlythroughscientificinvestigation,thenwewouldhaveequivalentsoftheaspectualbiasandpotentialwreckproblems;accordingtodescriptivism,theywouldhavesimilarsolutions.11.ForRussell’sactualviews,thereadershouldconsultSainsbury(1993/2002).12.TheproposalintheTractatus,asexplainedlaterbytheauthorinlecturesandconversationsinthelate1920sandearly1930s,appearstobeclosetoDavidLewis’s(1979)accountofso-calleddesecontents–thecontentswhose“irreduciblyindexical”characterispointedoutbyPerry(1979)withforcefulexamples.Lewis’sproposalistotakeawaythesubjectfromthecontentitself,andthinkingoftruth-conditionsnotasfunctionsfromworldstotruth-values(i.e.,notasdiscriminatingamongpossibleworlds),butratherasfunctionsfromsubjects,worldsandtimestotruth-values(i.e.,asdiscriminatingamongpossiblesubjects,astheyareataparticulartimeintheirlivesatagivenpossibleworld).239781472578235_txt_print.indd2308/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageReferencesBaker,Mark(1997),“ThematicRolesandSyntacticStructure”,inL.Haegemaned.ElementsofGrammar,Dordrecht:Kluwer,73–137.Bealer,George(1998),“IntuitionandtheAutonomyofPhilosophy”,inM.R.DePaulandW.Ramsey(eds),RethinkingIntuition,Oxford:RowmanandLittlefield,201–39.Chalmers,David(1996),TheConsciousMind,Oxford:OxfordUniversityPress.Daniels,Norman(2011),“ReflectiveEquilibrium”,TheStanfordEncyclopediaofPhilosophy(Spring2011Edition),EdwardN.Zaltaed.forthcoming.Availableonlineathttp://plato.stanford.edu/archives/spr2011/entries/reflective-equilibrium/(accessed12September2011).Davies,MartinandHumberstone,Lloyd(1980),“TwoNotionsofNecessity”,PhilosophicalStudies38,1–30.Donnellan,Keith(1966),“ReferenceandDefiniteDescriptions”,PhilosophicalReview75,281–304.Evans,Gareth(1982),TheVarietiesofReference,Oxford:ClarendonPress.García-Ramírez,E.andShatz,M.(2011),“OnProblemswithDescriptivism:PsychologicalAssumptionsandEmpiricalEvidence”,MindandLanguage26,53–77.Gettier,E.(1963),“Isjustifiedtruebeliefknowledge?”Analysis23,121–3.Goodman,Nelson(1976),LanguagesofArt,Indianapolis:Hackett.Heim,I.andKratzer,A.(1998),SemanticsinGenerativeGrammar,Oxford:Blackwell.Ichikawa,J.andJarvis,B.(2009),“Though-ExperimentsandTruthinFiction”,PhilosophicalStudies142,221–46.Jackson,Frank(1998),FromMetaphysicstoEthics.Oxford:OxfordUniversityPress.Jeshion,Robin(2000),“OntheObvious”,PhilosophyandPhenomenologicalResearch60,333–55.Kaplan,David(1989),“Demonstratives”,inJ.Almog,J.PerryandH.Wettstein(eds),ThemesfromKaplan,Oxford:OxfordUniversityPress,481–563.Kripke,Saul(1977),“Speaker’sReferenceandSemanticReference”,inFrench,P.,Uehling,T.andWettstein,H.,ContemporaryPerspectivesinthePhilosophyofLanguage,Minneapolis:UniversityofMinnesotaPress,255–76.—(1980),NamingandNecessity,Cambridge,Mass.:HarvardUniversityPress.—(1982),WittgensteinonRulesandPrivateLanguages,Cambridge,Mass.:HarvardUniversityPress.Larson,RichardandSegal,Gabriel(1996),Knowledgeofmeaning:SemanticvalueandLogicalForm,Cambridge,Mass.:MITPress.Lewis,David(1979),“AttitudesDeDictoandDeSe”,PhilosophicalReview88,513–43.McDowell,John(1977),“OntheSenseandReferenceofaProperName,”Mindlxxxvi,159–85.Neale,Stephen(1990),Descriptions,Cambridge,Mass.:MITPress.Peacocke,Christopher(1999),BeingKnown,Oxford:OxfordUniversityPress.Perry,John(1979),“TheProblemoftheEssentialIndexical”,Noûs13,3–21.Russell,Bertrand(1912/1980),TheProblemsofPhilosophy.Oxford:OxfordUniversityPress.249781472578235_txt_print.indd2408/08/201415:45\nEditorialIntroduction:HistoryofthePhilosophyofLanguageSainsbury,Mark(1993/2002),“RussellonMeaningandCommunication”,inDepartingfromFrege,London:Routledge,85–101.—(2005),ReferencewithoutReferents,Oxford:ClarendonPress.Soames,Scott(1984),“LinguisticsandPsychology”,LinguisticsandPhilosophy7,155–79.Sosa,Ernest(2007),AVirtueEpistemology,Oxford:ClarendonPress.Stalnaker,Robert(1978),“Assertion”,inP.Coleed.SyntaxandSemantics9,NewYork:AcademicPress,315–32.Strawson,Peter(1969),MeaningandTruth.AnInauguralLecturedeliveredbeforetheUniversityofOxfordon5November1969,Oxford:ClarendonPress.Williamson,Timothy(2007),ThePhilosophyofPhilosophy,Oxford:Blackwell.259781472578235_txt_print.indd2508/08/201415:45\nOntheNatureofLanguage:2ABasicExpositionJamesHigginbothamIntroductionLanguage,likeothernaturalphenomena,canbestudiedfromdifferentpointsofview.Inthischapterweaimtoconveyatleastpartofthesourceofinterestintwoofthese,namely:(i)theabstractcharacteroflanguage,orthefeaturesoflinguisticsystemsthathavebeenanobjectofreflectivestudysinceancienttimes,followingespeciallyGreek,Indian,andOrientalscholars;and(ii)theissuesthatarisewhenwetakeupthefactthatthesesystemsareexemplifiedinhumancognition;i.e.,thattheybelongtousinsomesense,evenapartfromanyreflectiveorscholarlyunderstandingonourpart.Thelatterissues,ifnottheformer,areaboveallassociatedwiththeworkofNoamChomsky,andhavebeencarriedforwardbymany,withthevarietyofdisagreementsovermattersofprincipleaswellasdetailthatmaybeexpectedinsuchanenterprise.Thereareseveralexcellent,andpopularlywritten,introductionstothesubject,includingPinker(2007),GuéronandHaegeman(1998),andothers.Butwewillbeginfurtherback,withsomeobservationsaboutEnglishthatshouldexemplifysomeoftheattainments,andsomeofthetensions,inthesubject.Myexpositionwillbeelementary,butIhopethatthissimplicitywillifanythinghighlightthemoralofthestory.Wetakeitasevidentthatwordsfallintodifferentcategories,or“partsofspeech”,asmanyofusweretaughtinschool(infactthedistinctionsweweretaughttomakeamongstcategoriesofwordsemergedastheresultofscholarlystudyandreflectionovermanycenturies).Wealsoassumemeaningfulelementsthatcanonlyoccuraspartsofwords,suchasthe“-ed”oftheregularEnglishpasttense,orthe“-er”thatconverts“view”into“viewer”,“walk”into“walker”,andsoforth.DistinguishingaswedoamongstthemajorcategoriesofNoun,Verb,Adjective,andPreposition,andallowingtherough-and-readynotionsofsubjectandobject,apointthatmarksadistinctionbetweeninventoriesofwordsandtheinflectionssuchas“-ed”and“-er”thatgowiththem,ontheonehand,andtheunderstandingoftheconstructionofwholesentencesontheother,istherealizationthatcomplexesofwordsthemselvesbelongtocategories,and269781472578235_txt_print.indd2608/08/201415:45\nOntheNatureofLanguage:ABasicExpositionthatexpressionsofthesecategoriescanoccurasconstituentsofothers.Thusacompletenominal,suchas“theredhouse”,canoccuraspartofasentence;butsententialelements,suchastherelativeclausein“thehousewhich/thatIsaw”,canoccurinsidenominals.Theprocesscanofcoursebeiterated,yieldinganestedinfinityofsyntacticcomplexes.Thestudyofthestructuresthatgowiththecombinatorialdevicesoflanguageisthestudyofsyntax,asopposedtothestudyofmorphology,themakeupofwords,orlexicography,theenumerationanddescriptionofprimitivewords,ortheprimitivepartsofwords.WithinrecenthistoryanimportantstepwastakenbyHarris(1951),whoproposeddescriptivesystemsforsyntax.Asanillustration,weconsiderhistreatmentofthenominalsystemofEnglish(roughly,thoseexpressionsthatcanconstitutethesubjectofacompletesentence).Wehaveataminimumthefollowinginventory:(i)Pronouns(e.g.“I”,“he”,“she”,or“they”,standingalone);(ii)Demonstrativesthatmuststandalone(e.g.,“now”or“then”);(iii)Demonstrativesthatcanbefollowedby(possiblymodified)Nouns(e.g.,“these(old)books”,or“thosebooks(ontheshelf)”);(iv)ArticlesandexpressionsofgeneralitythatmustbefollowedbyNouns(e.g.,“the”,“every”).Modificationraisesinterestingquestions,inpartbecausetheorderofadjec-tivesisrestricted(soitisfarmorenaturalinEnglishtosay“theoldredhouse”than“theredoldhouse”).Further,inEnglish,relativeclausemodifiersfollowtheNoun,asin“thebookthatIread”.Abstractingfrommanydetails,wemayhavethefollowingpicture:theEnglishNominalisaNounPhrase(hereafter:NP);theelementsin(i)and(ii)aboveareNPbythemselves;thoseillustratedin(iii)and(iv)consistofNouns,possiblyaccompaniedbymodifiersincertainpositions,andprecededbyademonstrativeorexpressionofgenerality(whichinsimplepluralssuchas“bighouses”neednotbeexplicitlyrealized).Modifierscanbeiterated;butthereisauniquelicensing(orhead)Noun(soinnominalcompoundsthefinalNounwillbethehead,as“thesoapbar”referstoabarofsoap,whereas“thebarsoap”referstoanyquantityofsoapinabarshape);moreover,theinsertionofanarticleorexpressionsuchas“every”closesofftheconstruction,sowecannothave*”theeverybook”,wheretheasterisk“*”indicates(theauthor’sjudgment)thattheelementinquestionissyntacti-callydeviant,inwhateverlanguageisinquestion.Now,Harrisobservedthattheabovesimpledescriptioncanbesystema-tizedinthefollowingway:wehavethreelevelsofNominal:thesimpleNounN(aword,generally,thoughperhapsacomplexsuchas“SovietUnion”);theclutchofmodifiers,whichcanbeaugmentedwithoutlimit,constitutinganN1,say,andafinalpartofspeech,theDeterminer,whichclosesitoffas,sayN2.279781472578235_txt_print.indd2708/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageThesuperscriptsthusindicatethenominallevel,andthesystemisabstractlydescribedthroughthefollowingtwosyntacticrules:(Ri)IfXisanNorN1,andYisasuitablemodifier,thenX^Y(orY^X,depending)isanN1;(Rii)IfXisanNorN2,andYisadeterminer,thenY^XisanN2.wherethearch“^”linkingXandYjustindicatestheresultofwritingXfollowedbyY(perhapswithaspacebetween;wesuppressthismarkinwhatfollows).Harris’smethodcanbeextendedtoVerbalcomplexes,andothers,aswell.Fromthegeneralpointofviewtakenhere,whatiscriticalaboutthesystemisthatitenvisagesthelayoutofsyntaxintermsofanexplicitinductivedefinitionofmembershipofexpressionsincategories.Ofcoursethisstephadlongbeentakeninthestudyofformallanguages,goingbacktoGottlobFrege(1879),andwascommoncoininmathematicallogic.WhatwasnovelinHarris’sapproach,andtakenupinmuchgreaterexplicitdetailinChomsky(1957;itselfexpoundinghisearlierwork)wastheapplicationofthemethodtohumanfirstlanguages(conceivingofthese,asitwere,asformallanguagesforwhichtheformalizationwasbeingsought).Thesystemofrules,togetherwithlexicalstipulations(e.g.,that“book”isaNoun)couldtogetherconstituteagrammar(orthesyntacticpartthereof)ofalanguageinuse.Ageneralmathematicalpictureoflanguagesadmittinggrammarsofthesortillustratedin(Ri)and(Rii)aboveisstraightforwardlygiven:agrammarconsistsofasetofcategories,andrulesthatallowthereplacementofacategoryCbyasequenceXY…Zofelementseachofwhichisacategoryorprimitiveexpression;andasequenceofreplacementsthatresultsinafinalsequencesofprimitiveexpressionsisaderivationofsfromC,oraproofthatsεC.Weindicatethereplacementbythearrow“→”.So,foroneoftheNominalexamplesabove,wemighthave,whereC=N2:N2→DN1N1→AN1N1→NN→house;A→big,A→red;D→thedeployingthesesoastoderive“thebigredhouse”(wheretherule“N1→DN1”mustbeusedtwice,onceforeachmodifier).Thefullsequencewillbe:289781472578235_txt_print.indd2808/08/201415:45\nOntheNatureofLanguage:ABasicExpositionN2DN1DAN1DAAN1DAANandfinally,afterappropriatelexicalreplacement:Tree(1)thebigredhouseEquivalently,wecanrepresentthederivationbyatreeasin(1)N2(1)DN1theAN1bigAN1redNhousewhichitselfreflectstheconversionoftherulesfollowedintheabovederivationbytheclausesofaninductivedefinitionalongthelinesof(Ri)and(Rii),togetherwiththestipulationthatcertainwordsaregivenasbelongingtocertaincategories.Thereisnodoubtthatasystemsuchasthatillustrated,technicallyacontext-freephrasestructuregrammar,isadequatenotonlyforformallanguages(andmost()Higginbotham:Trees17deoctubrede20112/6computerlanguages,thoughsomedeviateinvariousrespects)butalsoforlargetractsofEnglishandotherfirsthumanlanguages.Eventheelementarydiffer-encesareinteresting,asshowingthatthehierarchicaldimensionoflanguageiscriticalforlinguisticdescription:thusforinstancetheNominalinCambodianisverynearlythereverseofEnglish,sothatthenaturalphrasefor“thebigredhouse”comesoutas“houseredbigthe”,whichiswhatyouwouldgetifyouflipped(1)aroundonthepageasifitwereamobile.Weturnnow,however,totwoofthequestionstakenupbyChomsky(1957,1965andlaterwork),namely:(I)Aregrammarsassigningphrasestructurestosentencesandother299781472578235_txt_print.indd2908/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguagecategoriesinthemannerdiscussedaboveinprincipleadequatetohumanlanguages;and(II)Whatcanbesaidabouttherelationsofanadequategrammarforaspeakerandthelinguisticpowersofthespeakerwhosegrammaritis?For(I):Chomskyandothersarguedatlengththat,inadditiontorulesgener-atingphrasestructures,therewererelationsbetweensuchabstractstructuresthatwerenotinanyreasonablewayexpressedintermsofrelationsbetweenwholesentences.Anearlyexample,discussedextensivelyinChomsky(1957),istheverbalauxiliaryofsimpletensedsentencesofEnglish.EnglishsimplesentencesmustcarryTense,whichfiguresasanaffixontheVerb,eitherPresentorPast(asin“is”versus“was”).BesidesTense,thereareoptionalelements,namely:theModals(“will”,“can”,“may”,“must”,etc.),thePerfective“have”,andtheProgressive“be”.Allarepresentin(2):(2)Thechildmayhavebeenreadingthebook.Theorderoftheseelementsisfixed,namelyTense–Modal–Perfective–Progressive(thoughallbutthefirstcanbeomitted).WenotenowthatthePerfectiveconditionsthesuffixontheProgressive,soiftheformerisomittedwehave,not*“thechildmaybeenreadingthebook”(theasterisk“*”indicatingthatthefollowingisnotaproperlyformedorgrammaticalexpression,inthiscaseofthecategorySentence)butrather(2):(2)Thechildmaybereadingthebook.Similarly,iftheProgressiveisomitted,thenthesuffix“-ing”doesnotappearwiththeVerb.Wehave(3):(3)Thechildmayhavereadthebook.Andfinally,iftheTensedModal“may”isomitted,thentheTensemustappearonwhateverfollows,asin(4)and(5):(4)Thechildhasreadthebook.(5)Thechildisreadingthebook.Theseobservations(andotherslikethemintheGermanicandRomancelanguages)suggestaruleofthefollowingsort:(6)ThefundamentalsequenceisTense-Modal-“have”+“-en”-”be”+“-309781472578235_txt_print.indd3008/08/201415:45\nOntheNatureofLanguage:ABasicExpositioning”-Verb,whereeachoftheaffixes(Tense,“-en”,“-ing”)movesinthecourseofthederivationtotheelementthatfollowsit.Arulesuchas(6),whichmustapplyinthecourseofaderivation,isobviouslynotaphrasestructurerule:Chomsky(1957)usedthetermtransformationtodescribetheoperationof(6)andsimilarrules.However,itexpressesasimplegeneralizationthatreducesthespacethatwouldotherwisebetakenupingivingtheformsoftheauxiliary.Theexampleof(6)(astatementof“affix-hopping”,tousethenamesometimesgiventoit)istypicalofprinciplesthatsuggesttheneedfordescriptivepowerthatgoesbeyondphrasestructure.Besidestheverbalauxiliary,thereareotherexamplesthatsuggesttheneedfortransformations;seeGuéronandHaegeman(1998),orothertexts,forexamples.Thepointofthisrehearsalofacentralargumentthatphrasestructurecanbeonlypartofthestoryaboutgrammarsforhumanlanguagesisthatitputsus,evenwithonlythismuch,inapositiontoraisequestion(II)above.Obviously,nativespeakersofEnglish“follow”(6),inthesensethattheyproduce,andperceiveasnormal,potentialsentencesthatareinaccordwithit,anddonotproduce,andperceiveasstrange,sentencesthatarenot.What,beyondthis,canbesaidabouttherelationoftheruletothenativespeakerofEnglish?Theanswerto(II)suggestedbyChomsky(1965)isthatspeakersstandinanepistemicrelationtotherulesofgrammar,whichhedubbedtheircompe-tence.Thetermisperhapsunfortunate,sinceitsuggeststhatcompetenceisapropertypossessedbythecompetent;whatisintended,however,isthatthespeaker’scompetenceisjustwhatthespeakerknows,asystemofrulesand/orprinciplesfromwhichtheconsequencesforthedataofgrammaticaltheoryaredeductiveconsequences.Thereferencetoknowledgeismorethanametaphor,becauseitinvolvesunderstandingthegraspofafirstlanguageasanintellectualaccomplishment,arrivedatonthebasisoflinguisticandperceptualexperience,thusgivingrisetothesubjectofDevelopmentalPsycholinguistics.1Seeninthislight,andbearinginmindthatthereisconsiderablevarietyinhumanlanguages,thereisatensionbetweenthepositedformalpropertiesofgrammarandtheknowncourseoflinguisticmaturation:therelevantpropertiesofgrammarmustbeinferablefromprinciplesonthebasisoftheambientevidence;butthoseprinciplescannotbesostrongastoruleoutpossiblehumanlanguages.2ToadoptChomsky’sterminology,wepositalanguagefaculty(perhapsaspecializationofmoregeneralmentalfaculties)that,throughprincipleswehopetodiscover,quicklyzeroesin(bytheageof6,say)onthecriticalpropertiesoftheambientlanguage,onthebasisofsuchevidenceashumanchildrenactuallyreceive.Thequestionofthe“NatureofLanguage”,fromthispointofview,isfirstofallconcernedwiththenatureofthatfaculty.319781472578235_txt_print.indd3108/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageWehaveconcentratedaboveonthematterofsyntax,ortheorderandarrangementofwords,partsofwords,andwholephrasesandsentences.Butexactlysimilarconsiderationsapplytophonology,thetheoryofthesoundsofspeech,andtosemantics,orthatpartofageneralaccountofmeaningthatisconveyedthroughlinguisticform.Itwasremarkedabovethatthereisasimplemappingbetweenphrasestruc-turesandtheclausesofaninductivedefinition(thelatterbeingtheoperation“Merge”ofChomsky(1995,andlaterworkbyhim)andothers).Instandardformalizedlanguages,suchasthelanguageoffirst-orderlogic,theclausesthatbuildformulaecomeeachwithitscharacteristicsemantics.Theresultisagrammarandsemantics(anaccountofreferenceandtruth,eitherinamodelTreesEndSection1orabsolutely)thatruninparallel,andareequallyunambiguous,representablebytreesinthesenseusedabove(theneedforexplicitmentionofthetreesissecuredbymeansofparentheses,whichdistinguish,e.g.,“A&(BvC)”from“(A&B)vC”,wherethetreesFFA&FF∨CB∨CA&Bvividlyandfaithfullyreflectthescopedistinction).Aquicksurveyshowsthathumanlanguagesarenotsoneatlyorganized,atleastonthesurface:thereisanambiguityofquantification,forinstance,in(7)(didonemangointoeverystore,orwaseverystorevisitedbysomemanoranother?),and(8)canmeaneitherthatJohnwasreluctanttobeinstructed,orthatMarywasreluctanttoinstructhim:()Higginbotham:Trees17deoctubrede20113/6(7)Amanwentintoeverystore.(8)JohnwasreluctantlyinstructedbyMary.(Crucially,apersonasserting(7)or(8)mustintendoneoftheirtwointerpre-tations:forthatreason,weknowthatwehavegenuineambiguityandnotmerevagueness.)Bothofthesearecasesofstructuralambiguity,inthesensethattheyareambiguouseventhoughnowordtheycontainisambiguous.Thequestionthereforeariseswhetherthereisalevelofanalysis–hence,byhypothesis,aleveloflinguisticknowledge–wheretheseandsimilarambiguitiesaresortedout,eitherthroughsyntacticrepresentationorperhaps329781472578235_txt_print.indd3208/08/201415:45\nOntheNatureofLanguage:ABasicExpositionthroughtheoperationofsemanticrulesthatarenotstrictlydeterminedbylinguisticstructure.3Similarquestionsarisefortheinteractionoflinguisticformwithfeaturesofcontextthatgointointerpretation.4IfMarysaysaftertheparty,“Everyonehadagoodtime”,ofcourseshemeans,andintendstobeunderstoodasmeaning,thateveryonewhocametothepartyhadagoodtime;butshedoesnotactuallysaythat.Inthenextsection,weconsidersomeelementsofsyntacticvariation,turningafterwardstothearticulationofaframeworkforsemantictheory.Cross-LinguisticFeaturesAgeneralfeatureofhumanlanguagesisdisplacement;i.e.,theoccurrenceofalinguisticelementinasyntacticpositionotherthanthepositionwhereitwouldbelicensedinbasicphrasestructure.Onecanonicalexampleisthedisplacement(inEnglishandelsewhere,asdiscussedbelow)ofthequestion-wordsandlongerphrases“who”,“whichgame”,“howmanycats”,andsoforthinquestions,asin(9)and(10):(9)Whodidyousee?(10)Johnlearnedhowmanycatstheykeptinthecages.(thequestionin(9)issaidtobedirect;thatin(10),whichfiguresastheobjectof“learn”,indirect).Clearly,the“who”of(9)isunderstoodasquestioningtheobjectoftheVerb“see”,and“howmanycats”in(10)istheobjectof“kept”.Thisfactsuggestsatransformation,calledWH-movement,whichdisplacestheobjecttothefrontofitsclause.(ForitsgeneralfeaturesinEnglish,seeGuéronandHaegeman(1998).)Suchatransformationisfoundinmanylanguages,buthardlyinall:inChinese,forexample,thequestionwordsarefoundinexactlythepositiontowhichthedisplacedelementisrelated,asin(11),translating(9):(11)Nikan-jianshei?Youseewho?andnodisplacementispossible.Moreover,thereareanumberofintermediatecasesattestedinotherlanguages.InEnglish,theWH-expressionsdoubleasintroducingrelativeclauses,asin(12):(12)book[whichIsaw]339781472578235_txt_print.indd3308/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageandtheWH-expressionmaybereplacedby“that”,or(inEnglishinparticular,thoughnot,e.g.,inGerman)omittedentirely,asin(13):(13)book[that/∅Isaw]Therelativeclauseisofcourseamodifierconstructedoutofaclause;anditoccursfollowingtheNounmodified;andithasitslinguisticmarkers,theWH-expressionortheword“that”.ThereisastraightforwardcontrastherewithChinese,where,althoughthedeterminerprecedestheNoun,asinEnglish,(a)theWH-expressionsonlyhaveauseinquestions;(b)thereisnowordcomparableto“that”;and(c)theclauseprecedesratherthanfollowstheNounwithwhichitisinconstruction.Thisdifferencemayillustrateinasimplefashionhowmuchmaybegainedbyabstractingfromlinearordertoconsiderinsteadthehierarchicalarrangementofphrases.ThustheChinesefor(12)or(13)is(14):(14)Wokan-guo-deshu5Isawbook“bookthatIsaw”Thedifferenceoforderdoesnotaffectthehierarchicalstructure,whichisthesameinbothlanguages;andthefullstructure,withlinearorder,isjustthemirrorimageoftheEnglish,asindicated,withdetailsomitted,in(15),theTree(15)resultof“flipping”theEnglishsoastomatchtheChinesewordorder:(15)N1RCNNPV1bookIV_sawEvidently,thismaterialraisesthequestionwhat,ifanything,tosayaboutthepositionofthemissingobject,markedabovebytheblank“__”.Althoughit34()Higginbotham:Trees17deoctubrede20114/69781472578235_txt_print.indd3408/08/201415:45\nOntheNatureofLanguage:ABasicExpositionwouldtakeustoofarafieldtogodeeplyintothismatterhere,Inotethatithasbeenwidelyassumedthatthepositionisinfactoccupiedbyan“empty”elementt(for“trace”)6,syntacticallypresentbutnotpronounced.Toclarifysomewhatintermsofcustomarylogicalnotation:obviouslyin(12)–(14)what(sotospeak)gluestheelementstogether,theNoun“book”andtherelativemodifier“Isaw”orChineseequivalents,isthesharingofavariable,sothatin“thebookIsaw”somesinglethingissaidtobeabookandseenbyme,ornotationallyasin(16):(16)book(x)&IsawxWeconsidered,above,thecaseofdisplacementofquestion(inEnglish,WH-)expressions.Thereareanumberofotherexamplesthathavebeenarguedtofallundersomethinglikethesyntacticconditionsthatapplytotheseexpres-sions,astopicalization,illustratedby(17),andnegativeinversion,illustratedby(18):(17)JohnIthoughtnoonewouldpayattentiontot.(18)Notonebookdidtheteacheraskustoreadt.whereineachcasethepositiontowhichthedisplacedelementisinrelationismarkedbythetracet.Besidesthesecases,wherethedisplacedelementmovesoutsidethecoreclausealtogether,thereareothersinwhichdisplacementmovesfromoneclausalpositiontoanother.Weillustratewithtwoofthese.ThePassiveismarkedbythealterationoftheVerbtoitsperfectiveform(e.g.,“see”to“seen”,“fly”to“flown”,etc.)andconstructionwith“be”asin(19):(19)TheplanewasflowntoParis.Herethedisplacementisfromtheobjectpositionof“fly”tothesubjectpositioninthesentence.Theprocesscaniterate,sowehaveforinstance(20):(20)TheplanewasthoughttohavebeenflowntoParis.Thissuggestsadoubledisplacement,startingfromastructureasin(21),andendingwithdoubly-markedtraces,asin(22):(21)__wasthought[__tohavebeenflowntheplanetoParis](22)Theplanewasthought[ttohavebeenflownttoParis]359781472578235_txt_print.indd3508/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageInasimilarvein,theoperationusuallycalledraisingdisplacesasubjecttothesubjectpositionofahigherclause.Thusthetensedcomplementin(23)contrastswiththeuntensed(infinitival)complementin(24):(23)Itseems[(that)Johnisanicefellow](24)Johnseems[tobeanicefellow]In(23)thesubjectpositionistakenbypleonastic“it”;butin(24)itisfilledbythesubjectofthecomplementoftheVerb.Sothelatterisacaseofdisplacement;andifwefollowthethesisthatdisplacementleavesatraceofthevacatedpositionthesubjectoftheinfinitiveist,appropriatelyrelatedtothesurfacesubject“John”.Moreover,displacementinacomplementasin(23)isimpossible,andthatintheinfinitival(24)obligatory,asshownbytheungram-maticalities(25)–(26):(25)*Johnseems(that)isanicefellow.(26)*ItseemsJohntobeanicefellow.Again,displacementmayiterate,asin(27):(27)Maryseems[ttobecertain[ttobeelectedt]]where“Mary”,theobjectof“elected”,hasmovedfirsttothePassivesubjectposition,then(becausethePassiveisalsoaninfinitive)tothesubjectpositionof“certain”,andfinallytothesubjectpositioninconstructionwiththetensedVerb“seems”.Anaccountoftheselatterdisplacementphenomena,suggestedoriginallybyJean-RogerVergnaud,isthattheyreflectanabstractcondition,thateveryNominalconstructionmustbearaCase,eithernominativeorsomeother.InEnglish,Caseisvisibleonlyonpronouns,suchas“I”Nominative(i.e.,togetherwithTense,inthesubjectposition)versus“me”foreverythingelse,forexample.If,however,wesupposethatCaseisgiveninthesyntaxevenifnotovertlyinspeech,anditisassumedthatneitherthePassiveformofaVerbnortheinfinitiveallowCaseonthefollowingNominal,itfollowsthattheNominalmustbedisplacedtoapositionwhereCaseispossible.Thatrulesout(26),andforcestheiteratedmovementin(22)(ontheassumptionthatdisplacementmustpassthroughtheintermediateposition),butallows(24)and(27).IfweaddfurtherthatonceCaseismadeavailablethendisplacementisnotpossible,weruleout(15)aswell.369781472578235_txt_print.indd3608/08/201415:45\nOntheNatureofLanguage:ABasicExpositionKnowledgeofMeaningWehavebeenconsideringsomefeaturesofEnglishsyntax,withsidelonglooksatcross-linguisticdata.Meaninghasnotbeenneglected:onthecontrary,wehaveappealedregularlytosamenessanddifferenceinmeaninginjudgingvariousexamples.Butthatisnottogiveanexplicitaccountofmeaning,towhichquestionwenowturn.Thetargetofsemantics,likethatofsyntax,willbetakentobethe(native)speaker’scompetence:shecaninterpretanyofanunboundedrangeofsentences(andexpressionsofothercategories)insuchawayastoactappropriatelygivenherunderstanding.Thuswhatmediatesbetweentheuseoflanguageandtheactionthenundertakenwillcomprisetheconceptionofmeaningthatweaimtocapture.AtdinnerIaskthepersononmyleftifshecouldpleasepassthebroccoli;thatis,Isay,“Couldyoupassthebroccoli?”Noproblem:acertainphysicalobjectcontainingbroccoligetsmovedtowardsme.Herunderstandingofmywords,andthepointofmyremark,ledhertoeffectaphysicalchangeintheworld.Howdidthathappen?Asatisfactoryaccountofmeaningmustprovideaconceptionofunderstandingthatfitsappropriatelywithageneralaccountofinteractionsbetweenintentionalagents.Wedonotrequireanythingaselaborateasadevelopedhumanlanguagetoillustratemeaning.Ithaslongbeenknownthatavarietyofanimalshavemoreorlesselaboratesignalsthathavemeaningatleastinthesenseofbeingemittedonparticularstimulusoccasions,andeffectingreactionsonthepartoftherecipientsofthosesignals.SomesimpleinterjectionsofEnglish–“ouch!”–say,mayhavethispropertyaswell.Butourinterestmustturntomoreelaboratematters,wheresofarasweknowonlyhumanshavethesystemsthatwetakeforgrantedeveninsmallchildren.Whether,inlightofthemoredevelopedviewwetakeofhumanlanguage,weshouldregardtheelementsoftheprimitivesystemsastrulymeaningfulwouldappeartobeatbestatermi-nologicalquestion.WhenIspoketotheneighbouronmyleftandaskedifshecouldpassthebroccoli,IusedasyntaxwhosefinalNounadmittedanynumberofpossi-bilities:“Couldyoupassthe____.”TheindividualNounswillhavemeanings,whichwillcombinefreelywiththemeaningsoftheotheritemstogeneratethemeaningofthewhole.Itiscustomary,andcorrectasfarasisknown,todistin-guishsharplybetweenlexicalmeaning,orthemeaningofwordsinisolation,andthecombinatoricsofmeaning,ortheprinciplesthatdeterminethemeaningsofwholesentencesandothercomplexphrasesfromthemeaningsoftheirparts.Foranelementarybutcanonicalexample,consider(28):(28)Fidobarks.379781472578235_txt_print.indd3708/08/201415:45\nTree(29)TheBloomsburyCompaniontothePhilosophyofLanguageThesyntaxisunproblematic,andmaybegivenasin(29):(29)SNPVP+TenseFidobark+nonpastSupposethat“Fido”answersto(or,asweshallsayshortly,refersto)acertaindog,and“barks”signifiesbeingpronetomakebarkingnoises(howeverthesearedescribed).Wenowhavethesyntaxand(therelevantfragmentof)adictionary.Butwhatwewantasaconclusionissomethingtotheeffectthatsomeonewho,say,makesanassertionusing(28)assertsthatFidobarks;andthisweareasyetinnopositiontoderive.Theabovepointmaynotbeobvious,justbecausealanguagethatweuseissotransparenttous.Anobservationthatmayunderscorethepointisthe()Higginbotham:Trees17deoctubrede20115/6differencebetween(30)and(31):(30)IsawJohnhappy.(31)IsawhappyJohn.Thelexicalitemsareexactlythesamein(30)and(31).Buttheirmodesofcompositionaredifferent.(30)meansthatIsawJohnatatimewhenheappearedevidentlyhappytome(generallyrelaxedandsmiling,forinstance);but(31)onlymeansthatIsawsomepersonwhoisknownbythenickname“happyJohn”.Whatisthesourceofthedistinctionhere?Onreflectionitisclearthatin(30)theword“happy”isactuallypredi-catedofJohn,asinmyinformaldescriptionofthemeaning,whereasin(31)itisamodifier,andthereisnopredication(IcouldseehappyJohnsad,forinstance).7IthappensthatinEnglishthesemanticdistinctioniscorrelatedwithconstituentorder.Assemantics,thestudyofmeaninginlanguage,iscustomarilypractised,thecentralnotionsthatarewantedforcombinatoricsarethoseofreferenceandtruth-value(perhapsrelativetovariouscontextualparameters,asnotedbelow).Inkeepingwiththisconception,referencewillcomeindistinctbutrelatedforms,dependinguponcategoryofexpression;butalsoreferencewillbe,toputitsimplybutappropriately,tothepubliclyavailablethingswetalkabout(seeChapter7onContext-Dependence).Thus:thereferenceof“John”(inanappropriatecontext)willbetoacertainperson(eveniftherearelotsofpeople389781472578235_txt_print.indd3808/08/201415:45\nOntheNatureofLanguage:ABasicExpositionnamed“John”);thereferenceofapredicatewillbetothethingstowhichitproperlyortrulyapplies–“happy”appliestothethingsthatarehappy,“barks”tothethingsthatare(atthetimeofspeaking)pronetobark.And,finally,weconcludethatthereferenceof(28)istruthifindeedthethingFidoisamongstthethingstowhich“barks”trulyapplies,andfalsehoodotherwise.Here,then,ishowwefillinthegapbetweenthesyntax(29)for(28)andthereferentialunderstandingofitsNPandVP+nonpastsoastoderivetheconclusionthatwewant:consider(32):(32)ForalltreesSwithrootSandbranchesNPandVP+nonpast,andtanytime,andforallx,ifNPreferstox,andforanyφ,VP+nonpastappliestoanobjectyattjustincaseφappliestrulytoyatt,then:Sistruejustincaseφappliestrulytoxatt.Suchisthewantedcombinatorialstatement,whichtiesthedictionaryentriesfortheingredientsofthesentencetoitssyntaxsoastoproducesomethingthatcanbethecontentofanassertion.Thestatement(32)isofcourseperfectlygeneral:itsaysnothingaboutanyparticularitemsinthelanguage.SolongasNPiscomplete,whetheranamesuchas“Fido”oranelaboratesubjectsuchas“themanItoldyouaboutyesterday”,andsolongasVP+nonpast,whetherVPconsistsofasimpleVerborincludesmodifiersandcomplexobjects,asin“falselybelievesthattheYankeeswillwinthepennant”,itcranksoutwhathappenstoSwithrootS,providedonlythatweknowhowtofillintherelevantvaluesxforNPandconditionφforVP+nonpast.Theprojectofcombinatorialsemantics,onthisview,istoelucidatethegeneralrules,asin(32),fortheinterpretationofwholesyntacticstructures.Foranyparticularexampletowhich(32)applies,astatementofwhatistobeknownofitstruthconditionswillfollowoncetheparametersoftime,NP-reference,andconditionφonthepredicatearefilledin:thusoncewehavesaidthattheNP“Fido”refersto(thedog)Fido,thatthetimetis1April2011,andthattheconditionontheVP+nonpast“barks”isthatitappliestrulytojustthethingsthatarepronetobark,wewoulddeductivelyobtain(33):(33)The(potential)utteranceuof(29)on1April2011istruejustincaseFidoispronetobark.Conditionsonreferenceandtruthareboundtobesensitivetonon-linguisticparameters:obviously,thereferenceofdemonstrativessuchas“this”or“that”,orlongerphrasescontainingthem,suchas“thisbook”,arenotdeterminedbylinguisticform;similarlyformanyusesofthepersonalpronouns,andforthefirstpersoninparticular.Linguisticformdoescontributetoreference:thus399781472578235_txt_print.indd3908/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguage“I”and“me”refertotheirownspeakers;“this”isinoppositionto“that”asreferringtothemoreproximateoftwoobjects;andifyourefertoathingas“thatashtray”youwillretractuponlearningthatitisacandydish;andanywaywehavealreadyrelativizedtruthandfalsehoodtotimeofspeechingivingthecombinatorialrule(32).Ingeneral,theuseofalanguagewithsuchcontext-sensitivedeviceswouldnotbepossiblewithoutknowledgeofthecontext,includingtheknowledgeandassumptionsthatwemakeaboutothers’purposesinspeaking(again,seeChapter7onContext-Dependence).Butsemanticsitselftracksthespecificallylinguisticcontributionstoourspeechacts,generalizingovercontextualdomains,andthereforeleavingspecificvaluesonanoccasiontobedeterminedjointlybylanguageandcontext.Wehavethusfaremphasizedwhatinphilosophyareoftencalledthecognitiveaspectsofmeaning;i.e.,thoseaspectsthatgotodeterminingtheobjectsspokenof,andwhatpropertiestheyaresaidtohave.Lackingsuchinformation,oneissaidnottoknow,ornotfullytoknow,themeaningofanexpression.Simpleexamplesabound:I(andmanyofus)knowthatfuelpumpsandgasketsarefoundinautomobiles,butwedon’t(ordidn’talways)knowwhattheyare,andeithernoworthencouldnottellonefromanother.Still,thewordswereinourlexicons,andwecouldevenusetheminindirectdiscourse,aswhenweexplainedthatthemechanichastoldusthatthere’saleakinthegasket.Likewise,honestmistakesarepossible,whereweregardourselvesascorrectedwhenthemeaningisexplainedtous.Inthisregard,itisworthemphasizingthatignoranceofmeaningisnotonlylexical,butmaybecombinatorialaswell;andthattheinformationrequiredforknowledgeofmeaningisbynomeansequivalenttoanyphilo-sophicaldistinctionbetween“truthbyvirtueofmeaning”(analytictruth,ifthereissuch)and“truthbyvirtueofthewaytheworldis”(synthetictruth),orbetweennecessarytruthandcontingenttruth(seealsoChapter10,Analyticity,Apriority,Modality).Toillustratethefirstpoint:Bowerman(1982)documentstheusebychildrenofresultativecomplementsthatovergeneralizethepatternsavailabletoadults.Aresultativethatisacceptableforbothadultandchildis(34),wherethedirectobjectoftheverbisthesubjectoftheadjectivalcomplement:(34)Iwipedthetableclean.Themeaningof(34)isthatIwipedthetable,withtheresultthatitbecameclean.ItiseasytoimaginecircumstancesinwhichIcouldwipethetable,withtheresultthatitbecamedirty.However,theresultativecomplementin(35)isdeviant:(35)Iwipedthetabledirty.409781472578235_txt_print.indd4008/08/201415:45\nOntheNatureofLanguage:ABasicExpositionThesentence(35),althoughinterpretableasaresultative,issomehow“strange”.ChildrenlearningEnglish,accordingtoBowerman’sresearch,arenotsoinhibitedasweare.Shecollectedexamplessuchas(36):(36)Ipulleditunstapled.(37)I’mpattingherwet.8(Inthelattercase,onechildwaspattinganotherwithawethand.)Evidently,pullingcancauseadocumenttobecomeunstapled,andpattingwithawetinstrumentcancausethepatteetobewet;sothecausalnexusin(37),saidtoholdbetweenactionandresult,isfine.Whathasgonewrong?Anobvioussuggestionisthattheresultativeconstructionimposesacloserlinkbetweenactionandresultthanmerecausation:itmustbethatthepurposeofsomefamilyofactionsasgivenbytheVerb,thereasonthatsuchactionsareintentionallyundertaken,istobringabouttheresultsindicated.Wipingisundertakensoastocleanthings,notsoastomakethemdirty;pattingtogetthemdry,notwet;andpullingmaybetogetthem(inaliteralormetaphoricalsense)down,up,closed,orover,butnottobringabouttheunstapledstateofadocument.Insum,thereisateleologybuiltintotheconstructionthatthechildren(aged3–6)failtoappreciate,orsoitwouldappear.9Forthesecondpoint,thatthedistinctiontobedrawninsemantictheory,viewedasarticulatingnormalhumanlinguisticcompetence,doesnotlineupneatlywithanyproposeddistinctionbetweenanalyticandsynthetic,ornecessaryandcontingent,truth,itissufficient,itwouldseem,todrawattentiontotheinformationthatacompetentspeakerwouldgivetosomeonewhodidn’tknowawordorotherexpression.Often,thatinformationisperceptuallyimparted,andthusgivesahandleonmeaningthroughmeansofrecognition,whichmayormaynotbenecessary.Further,local,contingentinformationisintuitivelycritical.Onewouldnotbesaidtoknowthemeaningoftheword“dog”whodidnotknowthatdogsarekeptaspets;orsoIwouldsay.10Thereareotherelementsofmeaningbesidesreference,includingbutnotlimitedtotheexpressionofattitudestowardtheobjectsofreference,andcollateralinformationthatiswantedtointerpretfiguresofspeech.Forthefirst,wehaveepithetssuchas“Chink”(Chineseperson),andeuphemismslike“passaway”(die).Thereferenceoftheepithetoreuphemismisjustthereferenceoftheexpressionforwhichitisanepithetoreuphemism;buttheformerconveyderogation,andthelatteravoidanceofunpleasantness.Forthesecond,consider,“He’saregularNapoleon”(adisciplinarian),“Thenightislikeanun”(comforting),andmanyothers.11Evidently,speakersexpectotherstoknowthepointofusingtheseexpressions(eveniftheydonotusethemthemselves).Atanextremeend,sometermsaretaboo–sostrongthattheyarenottobeused419781472578235_txt_print.indd4108/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageormentionedunderanycircumstances–buteventheirtaboostatusis,sotospeak,earnedthroughbeingfullyunderstood.Thereisnoreasonforsemanticstosweepasidetheseelementsofmeaninginfavourofreference;notealsothattheirunderstandingcannotreplacereference,butactuallydependsuponit,aswiththereferentialequivalenceof“passaway”and“die”.SemanticComputationWehaveseenthatthelexiconandgrammarconspiretogethertogeneratetheinterpretationsofcomplexphrases,ofthesententialorothercategories.Thefundamentalobjectsthathavemeaningsaresyntactictrees,orutterancescorre-spondingtothem,decoratedasitwerewithlinguisticinformation:identityoflexicalelements,categoryorwordorphrase,linguisticrelationsofvarioussorts,andsoon.Interestthereforeattachestothequestionofthenatureoftheprinciplesthroughwhichinterpretationsofcomplexstructuresaregeneratedfromtheirparts.Inthisrespect,aworkingassumptionhasbeenthatinterpre-tationproceedsbylocalcomposition,asfollows.LetSbeasyntacticstructure,andletXbeapointinS.Sitself,beingalinguistictree,willhavearoot;andanypointinTreeSection4SmaybeviewedastherootofthesyntacticstructureconsistingofthepointsdominatedbyX.ThatsubstructureofSwillhaveXimmediatelydominatingasequenceof(simpleorcomplex)elementsY,…,Yn,andtheyinturnmaybetherootsoffurther1complexsubstructures.Forsimplicity,takingn=2,wemighthave:XY1Y2____withmaterialbothaboveXandbelowYandY,asindicatedbytheslashes.12Now,thesepointswillcomewithcertainformalinformation(oftentakentobegivenintheformofbinaryfeatures),anditisevidentfromthesimplestexamplesthattheinterpretationofXwilldependupontheinterpretationsofYandY,as“Fidobarks”doesupontheinterpretationsof“Fido”and“barks”.12AnaturalconjectureisthenthatXdependsonlyontheinterpretationsofY1andY2,andnothingelseinthestructure;nothing,thatis,aboveortotheside()Higginbotham:Trees17deoctubrede20116/6ofX,andnothingbelowYandYthemselves.Wemaycallthisprinciplelocal12compositionality.12429781472578235_txt_print.indd4208/08/201415:45\nOntheNatureofLanguage:ABasicExpositionItiseasytodeviselanguagesforwhichlocalcompositionalityfails;where,forexample,informationnotonlyaboutthemeaningsofYandY,butalso12abouthowtheywerearrivedat,iscrucialtodeterminingthemeaningXistohave,orwherethecomputationofthemeaningofXmustlookupthetreetoseewhatitisembeddedwithin.Atthesametime,thestandardformallanguagesofmathematics,e.g.,thelanguagesofPeanoArithmeticorfirst-orderlogic,arelocallycompositional.13Theassumptionoflocalcompositionalityhasledtoanumberofinter-estingquestionsaboutsemanticcomputation,particularlyasregardsstructuralambiguity:shouldwesupposethatexampleslike(7)or(8),repeatedbelow,aredisambiguatedatsomelevelofstructure,orthatthedisambiguationiseffectedbyhavingmultiplepossibilitiesforsemanticcomputation(thatis,asemanticsthatisnondeterministicintheformalsense)?(7)Amanwentintoeverystore.(8)JohnwasreluctantlyinstructedbyMary.Wedonottakeapositiononthisquestionhere.Butitshouldbenotedthat,howeverthematterispursued,thehypothesisoflocalcompositionalityisnotonlyadvancedasformallynarrowingtheabstractspaceofpossiblehumanlanguages,butalsomayprovideawindowintotheacquisitionoflanguage.Thelearnermustgraspmanywordsonlythroughthecontextofwholesentencesinwhichtheyareused,andisinthissensepresentedwithaprobleminmanyunknowns,asetofsimultaneousequationsasitwere.Then,justassemanticinformationaboutwordsmaybecriticalfordeterminingthemeaningsofsentences,soconverselyinformation,evenpartialinformation,aboutsentencemeaningmayprovideevidenceforthemeaningsofindividualwords.14ConclusionIntheabovebriefremarksIhavetriedtogivethereadernewtothesubjectsomeorientationtowards,andsomeoftheissuesin,contemporarygenerativegrammarconceivedasthestudyoflinguisticcompetence.Ofcourse,Ihaveconsideredonlyafractionalcrumbofwhathasbecomeovertimeaconsid-erableenterprise,involvingmanyhundredsofresearchersgatheringandanalysingoftennewmaterial.InsubsequentchaptersinthisvolumeyouwillfindfarmoredetaileddiscussionthanIhavebeenabletopresenthere.Atthesametime,Ihopetohaveimpartedonesenseinwhich“figuringouthowlanguageworks”isacomplexundertaking,andformanyanexcitingjourneyaimedatmappingonefeatureoftheanatomyofthehumanmind.439781472578235_txt_print.indd4308/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageNotes1.Forarecentbasictext,seeGuasti(2004).2.Chomskyhasrepeatedlyemphasizedthistension;forarecentstatementseeChomsky(1995).3.Higginbotham(1985)surveyssomeofthepossibilitieshere.4.SeeMerchant(2001)forarecentdiscussionofellipsisandsimilarphenomena.5.Inthisexample,“guo”is(roughly)amarkerofthepast,and“de”aparticleseparatingmodifierfromNoun.Tonemarkshereandelsewhereareomitted.6.Alternatively,byacopyofthemissingelement,butonethathasnovocalreali-zation.Weabstractfromthisdetailinwhatfollows.7.Themodifieris,inclassicalgrammaticalterms,appositive,inthattheinterpre-tationofthenameisfixedindependently,andtheAdjectivesuppliesafurtherdetermination.ManyNominalsareambiguousasbetweentheappositiveandthedescriptiveinterpretations.Thus“theunfortunateconclusion”canbeunder-stooddescriptively(theconclusionthatistheunfortunateone),butperhapsoccursmorefrequentlyinapposition(theconclusionisso-and-so,andthatso-and-soisunfortunate).Beforenames,theappositiveinterpretationisoblig-atory,asintheonlynaturalinterpretationof“themisguidedIndependenceParty”andthelike.Ofcourse,thecombinatoricsoftheappositive–descriptivedistinctionmustbeatargetofsemanticinquiryaswell.8.Bowerman(1982),113.9.Thereareotherindicationsofteleologicalconceptionsbeing“builtinto”complexcausativeconstructions.Itwouldtakeustoofarafieldtoexamineanyoftheseinanydetail,butInotethatlanguages(Chinese,forexample)havingso-calledSerialVerbconstructionsoftheformV-V^NP,asin(toparaphrase)“cook-eatthechicken”,or“bring-sellcarrots”,thecompoundingamountstomorethanconjunction:theexamplesdonotmeanmerelythatthesubjectcookedthechickenandateit,orbroughtthecarrotsandsoldthem,butimplicatealsothatthechickenwascookedinorderthatitbeeaten,thecarrotswerebroughtwiththeintentionofsellingthem,andthelike.SomethingofthesameorderisobservedinEnglishbyomittingtheDeterminerin“takethecowto(the)market”,“goto(the)College”,andmanyothers:onemaytakethecowtothemarketforanypurpose,buttotakethecowtomarketistotakeitinordertosellit;similarly,togotoCollege(not“theCollege”)isonlytoattenditasastudent.StillanothercasefromEnglishisfoundinoneclassofdenominalVerbsas“saddle(thehorse)”,“shelve(thebook)”,etc.,wherethepurposeoftheobjectsreferredtobytheNounfromwhichtheVerbisderived,here“saddle”and“shelf”,isfulfilledbytheaction:saddlesaremadeforhorsesandotheranimalstoberidden,shelvesforbooksandotherthingstobestored,andsoforth.HencethemeaningoftheVerb“saddle”,forinstance,isnottobeparaphrasedbyanythingsosimpleas“putasaddleon”–Icanputasaddleonanarmchair,butIcan’tsaddleanarmchair.SeeHigginbotham(2009)andreferencescitedthereforfurtherdiscussion.10.HilaryPutnam(1978)andelsewherespeaksofthe“stereotype”associatedwiththeterm;e.g.,stripeswithtigers.TheviewIsketchhereisdefendedatsomelengthinDummett(1974).449781472578235_txt_print.indd4408/08/201415:45\nOntheNatureofLanguage:ABasicExposition11.Ofcourse,thereisdebateovertheexactcharacterizationoffiguresofspeechandtherest.ButthepointofCohenandMargalit(1972)iswell-taken:metaphorandsimile,andtheinterpretationofepithetsandeuphemisms,arenotachieve-mentssetontopoflinguisticcompetence,butdevelopandaredeployedinconjunctionwithit,andformpartofit.ForfurtherdiscussionofmetaphorseeStern(2000)amongothers.12.Forsomeelaborationuponthistheme,seeHigginbotham(2007).13.Andevenwhentheyaretwo-dimensional,theyareexplainedintermsoflocallycompositionallanguage,asforinstanceb∫f(x)dxaisexplainedasreferringto(g(x)forx=b)−(g(x)forx=a),wheregistheantideriv-ativeoff.14.Inthisregard,seeGleitman(1990)andotherson“syntacticbootstrapping”.ReferencesBowerman,Melissa(1982),“StartingtoTalkWorse”.InS.Straussed.U-ShapedBehavioralGrowth.NewYork:AcademicPress,101–45.Chomsky,Noam(1957),SyntacticStructures.TheHague:Mouton.—(1965),AspectsoftheTheoryofSyntax.Cambridge,MA:TheMITPress.—(1995),TheMinimalistProgram.Cambridge,MA:TheMITPress.Cohen,L.JonathanandMargalit,Avishai(1972),“TheRoleofInductiveReasoningintheInterpretationofMetaphor.”InD.DavidsonandG.Harman(eds),SemanticsofNaturalLanguage.Dordrecht:Reidel,721–62.Dummett,Michael(1974),“WhatisaTheoryofMeaning?”InS.Guttenplaned.MindandLanguage.Oxford:OxfordUniversityPress,97–138.Gleitman,Lila(1990),“TheStructuralSourcesofVerbMeanings”.LanguageAcquisition1(1),3–55.Guasti,MariaTeresa(2004),LanguageAcquisition:TheGrowthofGrammar.Cambridge,MA:TheMITPress.Guéron,JacquelineandHaegeman,Liliane(1998),EnglishGrammar:AGenerativePerspective.Oxford:Blackwell.Harris,Zellig(1951),MethodsinStructuralLinguistics.Chicago:TheUniversityofChicagoPress.Higginbotham,James(1985),“OnSemantics”.LinguisticInquiry16(4),547–93.—(2007),“RemarksonCompositionality”.InG.RamchandandC.Reiss(eds),TheOxfordHandbookofLinguisticInterfaces.Oxford:OxfordUniversityPress,425–44.—(2009),Tense,Aspect,andIndexicality.Oxford:OxfordUniversityPress.Merchant,Jason(2001),TheSyntaxofSilence:Sluicing,Islands,andtheTheoryofEllipsis.Oxford:OxfordUniversityPress.Pinker,Steven(2007),TheLanguageInstinct:HowtheMindCreatesLanguage.NewYork:HarperCollins.459781472578235_txt_print.indd4508/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguagePutnam,Hilary(1978),“Meaning,ReferenceandStereotypes”.InF.GuenthnerandM.Guenthner-Reutter(eds),MeaningandTranslation:PhilosophicalandLinguisticApproaches,NewYork:NewYorkUniversityPress,61–81.Stern,Josef(2000),MetaphorinContext.Cambridge,MA:TheMITPress.469781472578235_txt_print.indd4608/08/201415:45\n3FormalSemanticsJoshDeverHumanlanguagesaresophisticatedtoolsfortransferringmentalstatesfromonepersontoanother.Consideranexample:JonesbelievesthatSmithwouldbeboredatthepartytonight,becausebooleanalgebraswillbethedominanttopicofconversation.Sheassemblesacollectionofwords,withappropriatemeaningsappropriatelyrelated,intoanorganizedstructure.Onthebasisofthatstructure,sheproducescertainsoundsbymovinghermouth,throat,anddiaphragminparticularways.Smithhearsthosesoundsandreconstructsbothwordsandstructurefromthem,receivingthesentence“Everyonewillbetalkingaboutbooleanalgebras.”Heworksoutthemeaningofthatsentencebycombiningknowledgeaboutthemeaningsofwordsandtheircombinationswithcontextualinformationsuggestingthat“everyone”herepicksoutthoseattheparty.InferringthatJoneswouldonlygivehimthisinformationaboutthepartyasawarning,andtakinghertobesincereandwell-informed,hecomestoshareJones’sbeliefthathewouldbeboredattheparty.Anexchangeofthissortcanseemsimpleandpedestrian,becausewearesoadeptattheskillsdeployedincarryingitout.Butskilfulnesscanoftenmaskenormouscomplexityintheunderlyingtask.Ananalogouscase:walkingacrossastreetmayseemasimpletask,butdecadesofworkinreproducingthistaskinroboticshasshownthatbipedalwalkingrequiresacomputationallyandconceptuallydemandingintegrationofvisualdataabouttheenvironment,proprioceptivedataaboutthebodilypositioningofthewalkingagent,andinformationaboutadynamiccentreofgravityandproperproductionanddistributionofforces.TheusetowhichJonesandSmithputlanguageintheaboveexchangeiscomplicated,andwearefarfromhavingafulltheoryofhowthisandothersimilarexchangesoccur.Itiscommon,however,tothinkthatthefollowingfourstagesareinvolvedinlanguageuse:1.Syntax:Atheoryofsyntaxdeterminesrulesaccordingtowhichlexicalitemsenterintovalidstructuralarrangements,suchasphrases,clauses,andcompletesentences.2.PhoneticsandPhonology:Theoriesofphoneticsandphonologydeterminetherelationbetweenastructuredsequenceoflexicalitems479781472578235_txt_print.indd4708/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageandasequenceofsounds(phonology)andthephysicalproductionandauditoryreceptionofthosesoundsequences(phonetics).3.Semantics:Atheoryofsemanticsassignsameaningtoeachstructureprovidedbythesyntactictheory(or,ameaningtoeachcontextualizeduseofeachstructure).4.Pragmatics:Atheoryofpragmaticsdetermineshowspeakersoughttoreact(updatetheirbeliefsandplansofaction)uponlinguisticreceptionofaparticularpieceofinformation.Veryroughly:JoneswantsSmithtoknowthathewouldbeboredattheparty.JonesknowsthatSmithisuninterestedinbooleanalgebras,andthatSmithisawarethatthepartyisthecurrenttopicofconversation,andthatwhatshe,Jones,sayswillberelevanttothattopic.Jonesisthus,viahergraspofatheoryofpragmatics,abletoworkoutthatSmith,uponreceivingtheinformationthatallofsomeunspecifiedgroupwillbetalkingaboutbooleanalgebras,will(a)concludethatfortheclaimtoberelevant,theunspecifiedgroupmustbethegroupofparty-goers,and(b)realizethat,givenhisinterests,hewouldbeboredattheparty.Jonesthusdecidesthataspeechactwhosesemanticcontentisthatallofsomeunspecifiedgroupwillbetalkingaboutbooleanalgebraswillmeethercommunicativegoals.Bycombininghergraspofatheoryofsyntaxandatheoryofsemantics,Jonesisabletodeterminethatthewords“everyone”,“will”,“be”,“talking”,“about”,“boolean”,and“algebras”canbecombined(inaparticularstructuralarrangement)toformasentencewhosesemanticcontentisthenthedesiredthatallofsomeunspecifiedgroupwillbetalkingaboutbooleanalgebras.Jonesthenuseshergraspofatheoryofphonologytomapthatsentencetoaparticularsequenceofsounds,andhergraspofatheoryofphoneticstodeterminehowheshouldmovehismouthtoproducethosesounds.(Ofcourse,Jones’sgraspofthetheorieswilltypicallybeinsomesensetacitorimplicit,aswithhergraspofthetheoryofwalking.Shecandowhatneedstobedone,butisnotusuallyawareofhowshedoesit.)Smiththenmoreorlessgoesthroughthisprocessinreverseorder.Heusesatheoryofphoneticstocharacterizehisauditoryperceptionasasequenceofsounds,andatheoryofphonologytomapthosesoundstothesentence“Everyonewillbetalkingaboutbooleanalgebras.”Withatheoryofsyntax,heassignsastructuretothatsentence.Thatstructureisthenfedintoasemantictheory,whichoutputsthesemanticcontentthatallofsomeunspecifiedgroupwillbetalkingaboutbooleanalgebras.Usingapragmatictheory,Smiththendeter-minesfirstthattheunspecifiedgroupmustbethegroupofparty-goers,andthendeterminesthatJoneswantshimtoknowthathewillbeboredattheparty.(AswithJones,Smith’sgraspofthetheoriesistacitorimplicit.)Thefocusofthischapterwillbeonthesemanticcomponentofthismulti-stageprocess.Inparticular,wewillfocusontheprojectofformalsemantics.489781472578235_txt_print.indd4808/08/201415:45\nFormalSemanticsThereisnoprecisedelineationofwhatcountsasformalsemantics.Roughly,though,formalsemanticsistheattempttogivepreciseaccountsoftherelationbetweensyntacticstructuresandsemanticvalues,typicallywhilemakinguseoftoolsfrommathematicsandlogic.Manyofthecentralideasofformalsemanticsoriginatedingivingtruthdefinitionsandmodeltheoryforformal,artificiallanguages,suchasfirst-orderlogicorvariousprogramminglanguages.Themodeltheoryofalanguageofformallogiccomeswithawell-definednotionofwhatcountsasameaning,orsemanticvalue–forexample,predicateextensions(conceivedassubsetsofamodel-theoreticdomain)orsetsofsequences(usedasassignmentstofreevariables),inthecaseoffirst-orderlogic.Thesemantictoolsdevelopedfortheseartificiallanguageshavethenbeenadapted,andextended,fornaturallanguagesaswell.Thegoalistoproducetheoriesthatrigorouslyandsystematicallyassignappropriatesemanticvaluestonatural-languageconstructions.Theprojectisoftencoupledwiththedualthoughtsthat(a)theattempttoproducesuchrigoroustheorieswilluncovercomplexitiesinthefunctionofnaturallanguagethatmightotherwisehavebeenoverlooked,andthat(b)itwilloftenthenbepossibletosystematizethosecomplexitiesundergeneraltheoriesofconsiderableeleganceandexplanatorypower.ASemanticPuzzleCaseManydifferentstylesofformalsemantictheorieshavebeendevelopedandpresented.Wewillnotattemptasystematicsurveyhere,butwillinsteadsamplebrieflyfromvariousstylesinordertogiveasenseofthekindsofresourcesandmethodsthatformalsemantictheoriescanuse.Oureventualgoalinthisdiscussionwillbetoseehowformalsemanticscanaidourunder-standingofcertainaspectsofpsychologicalattitudereports.Considerabeliefreportclaim,suchas:(1)JohnbelievesthatthetallestspyisinParaguay.Itisoftenthoughtthatsuchclaimsareambiguous.1Considertwoscenarios:1.S1:John’sjobattheCIAistokeeptrackofspymovementsaroundtheworld.HehasmanagedtohavedevicesthatcombineGPStrackingwithsomebasicbiometrictelemetryhiddenonthebodiesofspiesaroundtheworld.RightnowhistrackingboardshowsoneofthespiesinParaguay.Hehasnoideawhothespyis,buthecantell,fromthetelemetry,thatitisthetallestofallthespies.2.S2:John’snext-doorneighbourKarlis7’2”tall.UnbeknownsttoJohn,499781472578235_txt_print.indd4908/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageKarlisinfactaspy.Infact,heistheworld’stallestspy.KarlrecentlytoldJohnthathewasheadedofftoParaguayforvacation(infact,heisgoingtoParaguaytoassistwithan“extraordinaryrendition”).BothS1andS2looklikescenariosinwhichJohnbelievesthatthetallestspyisinParaguay.ButJohnbelievesthisindifferentwaysinthetwoscenarios–oneindicationofthisisthatineachscenario,whilethereisasenseinwhichJohnbelievesthatthetallestspyisinParaguay,thereisalsoasenseinwhichJohndoesnotbelievethatthetallestspyisinParaguay.InS1,Johnhasadedictobelief(or,John’sbeliefisbeingreportedinadedictomanner–wewillnotconcernourselvesherewiththequestionofwhethertherelevantdistinctionisbetweenkindsofbeliefsorbetweenkindsofbeliefreports).Hisbeliefcharacterizestheobjectofbeliefinaparticularway(asthetallestspy),buthedoesnotknowwhoitisthatissocharacterized.(Inatleastsomesenseofknowingwho.)(Infact,dedictobeliefsarecompatiblewiththereinfactbeingnoobjectsocharacterized.Theremight,forexample,benospies–perhapsJohnisthevictimofamassiveschemeofinternationaldeception–andhencenotallestspy.)InS2,Johnhasaderebelief.Hisbeliefisaboutacertainobject,andinreportingonthebeliefwecharacterizethatobjectinacertainway,butwithoutrequiringthatJohnthinkoftheobjectinthatway,orrealizethatitcanbesocharacterized.(Thederesensecanbeemphasizedbytheconstruction:(2)JohnbelievesofthetallestspythatheisinParaguay.butisavailablealsointheconstruction(1).)Whatiswanted,then,isatheorythatcaninasystematicwayassigntwodistinctmeaningsto(1).Craftingsuchatheoryinvolvesbothaconceptualcomponent–workingoutwhatsortofthingsmeaningsare,andhowtheyarerelatedtodatasuchastruthinascenario–andanengineeringcomponent–workingouthowtoassignmeaningstolinguisticexpressionsonthebasisoftheirinternalsyntacticstructure.PossibleWorldsSemanticsThisisatalltask.Wewillbeginwithsomethingmuchmorehumble:givingaverysimplefirst-draftmeaningtheoryforafragmentofthelanguage.Fromthathumblebeginning,wewillworkupastepatatimetowardsthemoreambitiousgoal.Here,then,isastartingthought.Wewantmeaningstodeterminetruthconditionsforutterances.Suppose,then,thatwehaveanotionofapossibleworld.2Weassociateeachsentenceinthelanguagewitha509781472578235_txt_print.indd5008/08/201415:45\nFormalSemanticssetofpossibleworlds–namely,theworldsatwhichthatsentenceistrue.Suchasetwouldprovideatleastadecentfirstapproximationtotruthconditions.Forthisideatoberealizedinatheory,wewantasystematicmethodofmakingtherequiredassociation.Inparticular,itisstandardtoassumethatwewantafinitesetofbasicprinciplesfromwhichwecan,foranygivensentence,calculatetheassociatedsetofpossibleworlds.Theformalsemantictheoryismeanttobeanarticulationofsomethingboth(a)known(albeittacitly)tocompetentspeakers,and(b)acquiredbyspeakersviacontactwithafinitesampleofthelanguage.Bothofthesefeatures–andespeciallythesecond–seemtopointtowardafinitetheory.3Sincenaturallanguageshaveaninfinitenumberofsentences,thismeans,minimally,thatourformaltheorycannotsimplyconsistofalistpairingsentenceswithsetsofpossibleworlds.Thisisthefirstofacollectionofideasderivingfromformalsemanticalconsiderationsthatwewillsetoutasweproceed:•Idea#1:Humanlinguisticunderstandingisaninfinitecapacity,butmustbeacapacitywithafiniteorigin.Thuswewantasemantictheorythat,onthebasisofafinitecollectionofbasicprinciples,assignssemanticvaluestoeachofaninfinitenumberofsentencesinthelanguage.Suppose,then,thatwehavealanguageLthathasafinitelists,…,sofsimple1n(atomic)sentences.LetLcontainthetruth-functionalconnectives¬,∧,∨,and→,andallowcreationofcomplexsentencesfromtheseconnectivesusingthefollowingsyntax:1.(R1)Everysimplesentencesisasentence.i2.(R2)IfAisasentence,then¬Aisalsoasentence.3.(R3)IfAandBaresentences,then(A∧B),(A∨B),and(A→B)arealsosentences.Thissyntaxproducesinfinitelymanysentencesfromtheinitialstockofsimplesentences.Apossibleworldcanthenbetakentobeanassignmentoftruthvaluetoeachsimplesentence.Wecanwritew(i)=Ttoshowthatpossibleworldwassignstruthtosimplesentences,andw(i)=Ftoshowthatwassignsifalsitytos.LetAsymbolizethesetofpossibleworldsassignedbythetheoryitoagivensentenceA.Thenthebehaviorof⋅isfullydeterminedbythefollowingbasicprinciples:1.(P1)Forasimplesentences,s={w:w(i)=T}.ii2.(P2)ForanysentenceA,¬A=W−A.(WhereWisthesetofallpossibleworlds.)3.(P3)ForanysentencesAandB,A∧B=A∩B.519781472578235_txt_print.indd5108/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguage4.(P4)ForanysentencesAandB,A∨B=A∪B.5.(P5)ForanysentencesAandB,A→B=(W−A)∪B.Thislittletheoryisabare-bonesversionofpossibleworldssemantics.Itincor-poratestwoideasthateachplayanimportantroleinsomeversionsofformalsemantics:•Idea#2:Themeaningofacomplexexpressionisasystematicfunctionofthemeaningsofitscomponentparts.Thus,syntaxandsemanticsgohand-in-hand:thesyntacticstructureofanexpressionisaroadmapforthecalculationofthesemanticvalueofthatexpression.•Idea#3:Meaningscanbeconstructedfromtruthconditions,conceivedasfunctionsfrompossiblesituationstotruthvalues.Thislittletheoryallowsustousesomebasicsettheorytoextractsomesimplesemanticpredictions.Forexample,wehave:¬A∨¬B=¬A∪¬B=(W−A)∪(W−B)=W−(A∩B)=W−A∧B=¬(A∧B).Thesemantictheorythuspredictsthesemanticequivalenceofanegatedconjunctionwiththedisjunctionofthenegatedconjuncts.Insimilarfashion,itcanshownthatanytwologicallyequivalentsentenceshavethesamesemanticvalue.Thispredictionisn’twithoutsomeproblems.Whilelogicallyequivalentsentencesclearlyhavesomethingincommon,fullsynonymymayseemliketootightaconnection.Arobustliteratureontheso-calledproblemoflogicalomnis-ciencehasariseninresponse.4Butevenwithsomeproblems,it’snotbadforafirstattempt.Anotherprediction:define@,theactualworld,tobetheworldthatassignstruthtoasentencesjustincasesis,infact,true.Thenwecandefinetruthforiisentences:•AsentenceAistrueifandonlyif@∈A.Considerasententialconnective*–eitheraone-placeconnective,combiningwithasinglesentenceAtoformacomplexsentence*A,oratwo-placeconnective,combiningwithtwosentencesAandBtoformacomplexsentence(A*B).*isanextensionalconnectiveif,wheneverAandChavethesametruthvalue,*Aand*Chavethesametruthvalue,and(A*B)and(C*B)havethe529781472578235_txt_print.indd5208/08/201415:45\nFormalSemanticssametruthvalue(andsimilarlywhenBandChavethesametruthvalue).Extensionalconnectives,thatis,allowsubstitutionofmateriallyequivalentsentencesintheirimmediatescopesalvaveritate.Ourtoytheoryhastheeasyconsequencethat¬,∧,∨,and→areallextensionalconnectives.ModalsTheaboveformalsemantictheoryisabare-bonesversionofpossibleworldssemantics.5Fromthesehumblebeginnings,additionalelementscanbeaddedtothetheory.Wecaneasily,forexample,addmodaloperatorstothelanguage.Themoststraightforwardwaytodothisaddsamight,orpossibly,sententialoperator◊,andamust,ornecessarilysententialoperator□,byaddingthebasicprinciples:1.(P6)ForanysentenceA,◊A={w:A∩W≠∅}.2.(P7)ForanysentenceA,□A={w:W⊆A}.ThesesimpleprincipleshavetheconsequencethatasentenceoftheformMightAistrueifandonlyifAistrueatsomepossibleworld,andasentenceoftheformMustAistrueifandonlyifAistrueateverypossibleworld.However,attentiontothebehaviourofmodalvocabularyinnaturallanguagesquicklyshowsthatthissemanticanalysisistoosimple.Mustandmightintroducemodalitiesofavarietyofflavours:61.Deontic:Youmustcrossthestreetatthecrosswalk.2.Circumstantial:Youmustwateryourtomatoplantsforthemtoremainhealthy.3.Epistemic:Theremightbelifeonotherplanets.4.Metaphysical:McCainmighthavebeenpresident.5.Preferential:WemightgoseeaGodardmovietonight.6.Dispositional:Ifyoumustsneeze,pleasecoveryourmouth.Sincethesamesentencewilltypicallyhavedifferenttruthconditionswhenthemodalisgiven(forexample)anepistemicratherthanadeonticreading,thesimpleprinciples(P6)and(P7)mustnotbethewholestory.Evenonceaflavourofmodalityhasbeensettled,truthconditionscanstillvary.Thesameepistemic-modal-containingsentencecanbetruewhenutteredbyonespeakerandfalsewhenutteredbyanother,iftheevidencebaseofthetwospeakersdiffers.Thesamecircumstantial-modal-containingsentencecanbetrueinonecontextandfalseinanothercontext,ifstandardsofstrictnesschangeacrosscontexts.(Mustyouwateryourtomatoplantsforthemto539781472578235_txt_print.indd5308/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageremainhealthy?Theanswermaydepend,forexample,onhowsalientthepossibilityofdroughtis.)Finally,mustandmightarefarfromtheonlymodalsinthelanguage.Othermodalssuchasnormally,can,ought,orprobablywillrequiresemanticanalysesgoingbeyondthesimpleprinciples(P6)and(P7).7Hereisaslightly(butonlyslightly)moresophisticatedtheoryofmodals.Takeeachmodaltobeassociatedwithamodalbase.Roughly,amodalbaseisasetofpossibleworlds:thoseworldsthatarerelevanttotheparticularflavour(andcontextualizeduse)ofthemodalinquestion.Morecarefully,amodalbaseisafunctionfrompossibleworldstosetsofpossibleworlds,givingineachpossiblesituationthoseworldsthatare,inthatsituation,relevanttotheparticularmodalflavourinquestion.8Sentencescannowbeassignedsetsofpossibleworldsrelativetoamodalbase.InsteadofthesimpleA–thesetofpossibleworldsassignedtosentenceA–wehaveAg–thesetofworldsassignedtosentenceArelativetomodalbaseg.Forsentenceswithoutmodals,therelativizationtomodalbaseswillhavenoeffect.Thusourearlierprinciple(P1)isgivenanidlingmodification:1.(P1g):Forasimplesentences,sg={w:w(i)=T}iiBecausegdoesnotappearontheright-handsideoftheidentity,thechoiceofmodalbasemakesnodifferencetothesemanticvalueassignedtosi.Thesituationisthesamefor(P2)–(P5).Theprinciplesformodals,however,makenon-trivialuseofthemodalbase:1.(P6g)ForanysentenceA,◊Ag={w:g(w)∩Ag≠∅}.2.(P7g)ForanysentenceA,□Ag={w:g(w)⊆Ag}.Forexample,givenamodalbasethatmapseachworldtothepossibleworldsthataremoralexemplarsforthatworld,□AwillhaveasitscontentallofthoseworldsinwhichAistrueatallmoralexemplars,andthuswillexpressthedeonticnecessityofA.Amodalbasedeterminesasetofrelevantworldsforamodal.Somemodalsmayrequire,inadditiontoasetofrelevantworlds,alsoanorganizationofworldsintocloserandfarther,ormoreandlesssignificant.Normallyandprobably,forexample,canbeanalysedusinganorderingofworlds,withthesemodalsplacingconstraintsonthecloserworlds.Wewillsetasideissuesarisingfromorderingsourcesintheremainderofthisdiscussion.Themodalbasesemanticsassignssemanticvaluesrelativetoamodalbase.Howarewetounderstandthisextraparameter?Forexample,howdowenowdefinetruthforsentences?Thepreviousdefinitionwillnowmaketruth,too,relativetoamodalbase:549781472578235_txt_print.indd5408/08/201415:45\nFormalSemantics•AsentenceAistruerelativetogifandonlyif@∈Ag.Butsurelywewantanotionoftruthsimpliciter–inpartsothatoursemanticscanproperlyinteractwithourpragmatics,whichmaycontainprinciplessuchasassertonlythatwhichyouknowtobetrue.Oneobvioussuggestionisthatwehavehereaspecificinstanceofaverygeneralphenomenonofcontextsensi-tivityinnaturallanguages.Thecontextsensitivityofmodalsisrathersubtle;muchcontextsensitivityinlanguageisconsiderablymoreovert,suchasthatofindexicalexpressionssuchas“I”,“you”and“here”.9Tomodelcontextsensitivitywewillwantarepresentationofcontext.Somecontext-sensitiveaspectsoflanguage,suchasovertindexicals,willdemandarepresentationofcontextthatspringsprimarilyfromtheenvironmentalsituationoftheutterance(factsabouttheproducerof,thelocationof,andthetimeoftheutterance,forexample).Othercontext-sensitivedevicesinlanguagewillplacedifferentdemandsontherepresentationofcontext–modals,onthecurrentapproach,willimposetherequirementthatthecontextinwhichasentenceisusedcontaininformationsufficienttodetermineamodalbase.Thesemanticcontentofacontextualizedutteranceofasentenceisthenthesemanticcontentofthatsentencerelativetothecontextually-determinedmodalbase,andthetruthvalueofacontextualizedutteranceisthetruthvalueoftheutteredsentencerelativetothecontextually-determinedmodalbase.Herewehaveanotheridea:•Idea#4:Context-sensitivityoflanguagecanbemodelledinaformalsemantictheorybyassigningsemanticvaluesrelativetoacontext,orbyassigningsemanticvaluesthatarefunctionsfromcontextstomorestandardsemanticvalues.Asemantictheoryofthissortwillthenengagewithatheoryofpragmaticsthatdetermines,foraparticularspeechsituation,whatthecontextisthatfillsthefunction’sargumentplace.BeliefContextsThesemantictechniquesthatareusedformodalscanbeusedforotherexpressionsthatarenotobviouslymodalinnature.Inparticular,beliefcontextscanbetreatedasakindofintensionalcontext.10GivenapersonP,letWbeafunctionfromPworldstosetsofworldssuchthatW(w)isthesetofworldsthatareconsistentPwitheverythingPbelievesinw–calltheseP’sbeliefworlds.Thenwecansay:1.(P8)PbelievesthatA={w:W(w)⊆A}.PPbelievesthatAisthentrue(inthesensethat@∈PbelievesthatA)justincaseeveryoneofP’sbeliefworlds(intheactualworld)isanAworld–thatis,559781472578235_txt_print.indd5508/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguagejustincaseeveryworldconsistentwithwhatP(actually)believesisaworldthatmakesAtrue.Thissemantictheoryforbeliefcontexthassomeimmediateconsequences:1.Beliefcontextsarenon-extensional:SupposeAandBhavethesametruthvalue.Thuseither@∈Aand@∈B,or@∉Aand@∉B.ButsupposeA≠B,soAandBarenotintensionallyidentical,orsynon-ymous.Inparticular,supposethereisaworldwsuchthatw∈A,butw∉B.IfwehaveW(@)=A,thenwewilltriviallyhavePbelievesthatAPtrue.ButPbelievesthatBwillbefalse,becauseW(@)isnotasubsetofPB(sincewisinthefirstbutnotinthesecond),andhence@∉PbelievesthatB).Thussamenessoftruthvaluedoesnotguaranteeintersubstitut-abilityinabeliefcontext.Thisisagoodresult–justbecausePbelievesonetruesentence,itdoesnotfollowthathebelieveseverytruesentence.2.Beliefcontextsaresensitivetothebelievingagent:LetPandQbetwopersons.SinceWPandWQarejusttwoindependentfunctionsfromworldstosetsofworlds,itisstraightforward,foragivensentenceA,tosetthingsupsuchthatPbelievesthatAistrueandQbelievesthatAisfalse.Wecould,forexample,letW(@)=A,andletW(@)=¬A.Again,thisPQisagoodresult.Differentpeoplehavedifferentbeliefs,andanadequateformalsemanticsshouldmakeiteasytovarythetruthvalueofbeliefreportsasthebelievingagentvaries.3.Beliefsareclosedunderlogicalconsequence:IfPbelievesAandA,12andAandAjointlyentailsomesentenceB,thenthesemantictheoryis12committedtotheclaimthatPalsobelievesB.IfAandAjointlyentailB,12thenanyworldinwhichAandAaretrueisaworldinwhichBistrue.12ThusA∩A⊆B.SincePbelievesA,W(@)⊆A.SincePbelieves121P1A,W(@)⊆A.ThusWP(@)⊆A∩A.ItthenfollowsthatWP(@)⊆2P212B,andhencethatPbelievesthatB.Isthisagoodresult?Itentails,forexample,thatanyonewhobelievesaconjunctionbelievesbothconjuncts,andthatanyonewhobelievesbothconjunctsofaconjunctionalsobelievestheconjunction.Thoseseemlikegoodresults.Ontheotherhand,itentailsthatanyonewhobelievesalloftheaxiomsofsettheoryalsobelievesallofthetheoremsofmathematics–aratherlessimpressiveresult.Thequestionoftheextenttowhichbeliefsareclosedunderanyformoflogicalentailmentisadifficultone,andthecurrenttheorycertainlygivesarathernaiveanswertothatquestion.Wenowhaveabasictheoryofbeliefreports,butthetheorywon’tyethelpinexplainingourguidingpuzzleofJohn’sbeliefaboutthetallestspy.Withourcurrenttools,allwecandoisaskwhetherthesemanticvalueassignedto569781472578235_txt_print.indd5608/08/201415:45\nFormalSemanticsThetallestspyisinParaguayhasJohn’sbeliefworldsasasubset.Wehavenomethodavailableforassigningtwodistinctreadingstothebeliefreport.Type-TheoreticSemanticsToremedythisproblem,weneedtoaddtechniquesfordealingwithsubsen-tentialstructure.Theformalsemantictheoriessketchedsofartakeastheirstartingpointacollectionofatomicsentences,andthenconstructandanalysecomplexsentencesbuiltfromthoseatoms.Butamorethoroughtheorywilltakeindividualwordsasstartingpoints,andthenbuildupmeaningsoflargerexpressions,includingthesimplestsentences,fromthemeaningsoftheircomponentwords.Whatwouldsuchatheorylooklike?Let’sbeginwitheasycases,suchasthesentence:(3)Marysnores.Herearetwoassumptionstoguidetheconstructionofasemantictheoryintendedtocoverthissentence,andotherslikeit:1.Thesemanticvalueofwholesentencesshould(still)besetsofpossibleworlds.2.Thesemanticvalueofapropername,like“Mary”,istheobject/persontowhichthenamerefers.Bothoftheseassumptionsarecontroversial,buttheyserveasplausibleandhelpfulstartingpoints.Importantly,togethertheytightlyconstrainthesemanticvalueoftheothersententialcomponent,“snores”.Thatsemanticvaluemustcombinewiththesemanticvalueof“Mary”(thepersonMary)todeterminethesemanticvalueof“Marysnores”(thesetofpossibleworldsinwhichMarysnores).“Snore”couldthusbeassigned:•Asetoforderedpairs
ofobjectsandworlds,where∈snoresiffxsnoresinw.•Afunctionfromobjectstosetsofworlds,wherew∈snores(x)iffxsnoresinw.•Afunctionfromobjectstoafunctionfromworldstotruthvalues,wheresnores(x)(w)=Tiffxsnoresinw.Thereis,obviously,littletochoosefromamongtheseformallyinterchangeableoptions.Asaformalconvenience,wewilladoptthethird.11Thisformulationhastheadvantageofbeginningasystematicprocedureofcalculatingsemantic579781472578235_txt_print.indd5708/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguagevaluesofcomplexexpressionsviafunctionalapplicationofthesemanticvalueofonecomponentofthecomplexonthesemanticvalueofanothercomponent.Thus:•Marysnores=snores(Mary)(Thisthenrequiresthatthesemanticvalueofasentencebeafunctionfromworldstotruthvalues,ratherthanasetofworlds.However,functionsfromworldstotruthvaluesandsetsofworldsareimmediatelyinterchangeable,bytakingthefunctiontobethecharacteristicfunctionoftheset.)Generalizing,propernameshaveobjectsassemanticvalues,sentenceshavefunctionsfromworldstotruthvaluesassemanticvalues,andintransitiveverbshavefunctionsfromobjectstofunctionsfromworldstotruthvaluesassemanticvalues.Buthowdoweproceedassentencesbecomemorecomplex?Considertwoslightincreasesincomplexity:(4)Marysnoresloudly.(5)MaryadmiresSusan.Howshouldweworkoutthesemanticcontributionoftheadverb“loudly”,orofthetransitiveverb“admires”?Itwouldbenicetohaveasystematicwayofdeterminingatleastthetypeofsemanticvalueneeded.Hereisamethod.Beginwiththetwoassumptionsabove.Propernameshaveobjectsassemanticvalue;sentenceshavefunctionsfromworldstotruthvalues.Combinetheseassumptionswiththeprinciplethatsemanticvaluesofcomplexesaretheresultoffunctionalapplicationofonecompo-nent’ssemanticvaluetoanothercomponent’ssemanticvalue.Thenworkbackwardsthroughthesyntacticstructureofthesentenceinquestion.Assume,forexample,that“MaryadmiresSusan”hasthesyntacticstructure:•1.[Mary[admires[Susan]]]Noverysophisticatedsyntactictheoryisneeded(orused!)here–allweneedisthattheimmediatecomponentsof“MaryadmiresSusan”are“Mary”and“admiresSusan”,andthatthecomponentsof“admiresSusan”are“admires”and“Susan”.Thenwereasonasfollows:1.MaryadmiresSusanisafunctionfromworldstotruthvalues.Thatsemanticvaluemustbeeither(a)theresultofapplyingMarytoadmiresSusan,or(b)theresultofapplyingadmiresSusantoMary.589781472578235_txt_print.indd5808/08/201415:45\nFormalSemantics2.Maryisanobject,andhencenotafunction.Itthuscannotbeappliedtoanything,andoption(a)isruledout.3.SowemusthaveMaryadmiresSusan=admiresSusan(Mary).4.SinceMaryisanobjectandMaryadmiresSusanisafunctionfromworldstotruthvalues,admiresSusanmustbeafunctionfromobjectstofunctionsfromworldstotruthvalues.5.Thesemanticvalueof“admiresSusan”mustbeeither(a)theresultofapplyingSusantoadmires,or(b)theresultofapplyingadmirestoSusan.6.Susanisanobject,andhencenotafunction.Itthuscannotbeappliedtoanything,andoption(a)isruledout.7.SowemusthaveadmiresSusan=admires(Susan).8.SinceSusanisanobjectandadmiresSusanisafunctionfromobjectstofunctionsfromworldstotruthvalues,admiresmustbeafunctionfromobjectstofunctionsfromobjectstofunctionsfromworldstotruthvalues.“Afunctionfromobjectstofunctionsfromobjectstofunctionsfromworldstotruthvalues”isabitofamouthful.Let’saddsomenotationtomakeiteasiertokeeptrackofcomplicatedsemanticvaluessuchasthese.Lete(forentity)bethesetofobjects(thereferentsofpropernames).Leti(forintension)bethesetoffunctionsfromworldstotruthvalues(themeaningsofwholesentences).Anypropernamethenhasasemanticvalueoftypee;anysentencehasasemanticvalueoftypei.Wecanthenconstructothersemanticvaluesfromthestartingpointofeandi.Let(e,i),forexample,bethesetoffunctionsfromitemsoftypeetoitemsoftypei–thatis,functionsfromobjectstofunctionsfromworldstotruthvalues.Thenanyintransitiveverbhasasemanticvalueoftype(e,i).Moregenerally,wehavethefollowingidea:•Idea#5:Thecollectionofallsemanticvaluesweneedtoanalysearbitrarylexical,complexsubsentential,andsententialexpressionscanbesystem-aticallygeneratedfromthestartingpointofthetwobasictypeseandi.Letαandβbeanytwotypes.Then(α,β)isanadditionaltype:thetypeoffunctionsfromthingsoftypeαtothingsoftypeβ.Functionsfromobjectstofunctionsfromobjectstofunctionsfromworldstotruthvaluesarethenitemsoftype(e,(e,i)).Typesofincreasingelaboratenesscantheneasilybeconstructedbyiteratingtheparenthesisnotation.12TypesandtheSyntax–SemanticsInterfaceThetypeofthesemanticvalueofthewholesentencemustbei.NowconsideranycomplexcomponentexpressionEofthewholesentence(includingthe599781472578235_txt_print.indd5908/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguagesentenceitself).SupposethatcomplexexpressionisthesyntacticresultofcombiningtwocomponentexpressionsEandE,andthatthecomplex12expressionhasasemanticvalueoftypeα,foranyα.Sincesemanticvaluescombineviafunctionalapplication,EandEmusthavesemanticvaluesof12type(β,α)andβ,forsomeβ.Workingourwaydownthesyntactictreeforthesentence,wecanworkoutpossiblesemanticvaluetypesforallofthepartsofthesentence.Soconsider“Marysnoresloudly”.Supposethesyntacticstructureofthatsentenceis:•1.[Mary[snores[loudly]]Thewholesentenceisoftypei,and“Mary”isoftypee,so“snoresloudly”isoftype(e,i).Thusoneof“snores”and“loudly”mustbeoftype(β,(e,i))andtheotheroftypeβ,forsomeβ.Wedon’tyetknowwhichisofwhichtype,andwhattypeβis,sothesemantictypingisn’tyetfullydetermined.Butwehaveindependentreason,fromconsiderationofthesentence“Marysnores”,tothinkthat“snores”isoftype(e,i).Ifthisisright,then“snores”isnotoftype(β,(e,i))foranyβ,soitmustbetheexpressionoftypeβ,andtheparticularβinquestionmustbe(e,i).Thus“loudly”mustbeoftype((e,i),(e,i)).Toknowthetypesofthesemanticvaluesisnotyettoknowthespecificsemanticvalueswithinthosetypes.Maryisoftypee,andhenceisanobject.Butthatisnotyettosaywhichobjectitis.Toassignspecificsemanticvalues,wemustalsoconsiderthetruthconditionsofthesentence“Marysnoresloudly”.Thesetruthconditionsthenleadtothefollowingassignment:•Mary=Mary•snores=thefunctionthatmapseachobjectxtoafunctionthatassignstoaworldwthetrueifandonlyifxsnoresinw.Suchafunctionisequiv-alenttoafunctionthatmapseachobjecttothesetofworldsinwhichthatobjectsnores.Roughlyspeaking,wecanthinkofsemanticvaluesoftype(e,i)asbeingproperties–inthiscase,thepropertyofsnoring.–Somenotationtomakethiskindofspecificationoffunctionseasier:takeanarbitrarysentence,suchas“Johnownsacar”.Thentakeanamethatappearsinthatsentence,suchas“John”,andreplaceitwithavariable,suchasx.Finally,prefixtothemodifiedsentencealambda-extractoperatorλx.Theresultisλx(xownsacar).Thisexpressionrepresentsafunctionthattakesasinputanobjecto,andproducesasoutputthetruthvalueofthesentence“xownsacar”,whenxistakentorefertoo.Thesemanticvalueofof“snores”canthenbegiven,usingthisλ-notation,asλxλw(xsnoresinw).13•loudlyisthefollowingrathercomplicatedfunctionf.ftakesasinputa609781472578235_txt_print.indd6008/08/201415:45\nFormalSemanticsfunctiongthatmapsanobjecttoasetofworlds(i.e.,toafunctionfromworldstotruthvalues).fthenproducesasoutputafunctionhthattakeseachobjectotothoseworldswinwhichohaswhateverfeaturecausedftomapotoasetofworldsincludingw,andinadditionhasthatfeatureloudly.–Itshouldbeobviousthatthisspecificationofthesemanticvalueof“loudly”isfarfromunproblematic.Theappealtoafeaturecausingafunctiontomapaninputtoanoutputisnot,intheend,legitimate,sincewehavedonenothingelsewhereinthesemanticstogroundsuchtalkoffeaturesandcausation.Furthermore,itisnotatallobviousthat,forexample,snoringloudlyisthesameashavingloudlythefeatureofsnoring(or,indeed,thatthelattermakessenseatall).Thisleadstoanotheridea:•Idea#6:Bypursuingathoroughandrigorousformalsemantictheory,wecandiscoverunexpectedcomplicationsinthelanguage.Atafirstglance,onemighthavethoughtthatadjectivesandadverbswouldfunctionsemanticallyinmuchthesameway.Butwhenweattempttodescribebothadjectivesandadverbsinthetype-theoreticalframework,wediscoverthatadverbscreatecomplicationsthatdon’tarisewithadjectives,andareforcedtobetterunderstandthelanguagetodealwiththosecomplications.Thatbetterunderstandingcanthenhaveinterestingfurtherconsequences.Themostprominentformalsemanticmethodforanalyzingadverbsistotakeverbstointroduceevents,andthentotakeadverbsas,ineffect,adjectivesdescribingevents.14“Marysnoresloudly”,onthisapproach,becomesequivalentto“Thereisasnoringeventthatisloud,andthathasMaryasitsagent”.Carefulattentiontotheformalsemanticdetails,then,allowsustouncoveraubiquitoustacitontologyofeventsinnaturallanguages.•snoresloudlyisthendeterminedbyfunctionalapplicationofloudlyonsnores,andisthusthefunctionthatmapseachobjectotothoseworldswinwhichohasthefeaturethatcausedsnorestomapotoasetofworldsincludingw,andmoreoverhasthatfeatureloudly.Since(inaworldw)thefeaturethatcausessnorestomapotoasetofworldsincludingwisthefeatureofsnoringinw,thisthensimplifiesto:thefunctionthatmapseachobjectotothoseworldsinwhichosnores,anddoessoloudly.•Marysnoresloudlyis,finally,determinedbyfunctionalapplicationofsnoresloudlyonMary,andisthusthesetofworldsinwhichMarysnoresloudly.619781472578235_txt_print.indd6108/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageQuantifiedNounPhrasesHow,inthistype-theoreticalsetting,arewetounderstandquantifiednounphrasessuchas“everyphilosopher”and“somelinguist”?Thesenounphraseshavethesamesyntacticdistributionaspropernameslike“Mary”:(6)Marysnores.(7)Everyphilosophersnores.(8)Somelinguistsnores.Oneoption,then,istogivethemthesamesemantictypeas“Mary”–typee.Butif“somelinguist”isoftypee,thensomelinguistisanobject.Whichobjectcoulditbe?Noparticularlinguistwillyieldtherighttruthconditions,since,foreachparticularlinguist,“Somelinguistsnores”canbetrueevenifthatlinguistdoesnotsnore.Sosomelinguistmustbesomestrangenewobject–thearbitrarylinguist,perhaps,distinctfromeveryparticularlinguist.Butthisisaverystrangeobject.Since“Somelinguistsnores”istrue,thearbitrarylinguistmustsnore.Butsince“Somelinguistdoesnotsnore”isalsotrue,thearbitrarylinguistmustnotsnore,andhencemusthavecontradictoryproperties.Reasoningalongthesamelines,wewillfindthateveryphilosopherisadifferentkindofstrangeobject–onewhichformanypropertiesP,hasneitherPnornot-P.15Thecarefulformalsemanticanalysisoflanguagemayuncoverunexpectedmetaphysicalcommitmentslurkinginourlinguisticpractice,butitwouldbeniceifthecommitmentswerenotquitethisoutrageous.Fortunately,thereisanotheroption.Ouranalysisof“Marysnores”commitsustosnoresbeingoftype(e,i).Somelinguistsnoresisoftypei,andmustbeeither(a)theresultoffunctionalapplicationofsnorestosomelinguist,or(b)functionalappli-cationofsomelinguisttosnores.Thefirstoptionrequiressomelinguisttobeoftypee,whichthenleadstothestrangeobjectsdescribedabove.Butthesecondoptionrequiressomelinguisttobeoftype((e,i),i),andhencedoesnotrequire“somelinguist”torefertoanobject.Let’sexplorethepossibilityofgivingquantifiednounphrasessemanticvaluesoftype((e,i),i).Recallthatasemanticvalueoftype(e,i)is,roughly,aproperty(somethingthat,givenaworld,determinesasetofobjectsthathavethatpropertyatthatworld).Asemanticvalueoftype((e,i),i)canthusbethoughtofasafunctionfrompropertiestofunctionsfromworldstotruthvalues,orafunctionfromproperty-worldpairstotruthvalues,or,roughly,apropertyofproperties.Butwhichpropertiesofproperties?Suppose,plausibly,that“Somelinguistsnores”istrue.Thenweneed@∈somelinguistsnores,629781472578235_txt_print.indd6208/08/201415:45\nFormalSemanticswhichmeansweneedsomelinguisttobeapropertyofpropertiessuchthatthepropertysnores(thatis,thepropertyofsnoring)has,intheactualworld,that(second-order)property.AndthesameholdsforanypropertyQ.Ifthesentence“‘SomephilosopherQ’s”istrue,thensomephilosophermustbeasecond-orderpropertythatisinstantiated,intheactualworld,byQ.Thuswehave:•somephilosopheristhesecond-orderpropertyofbeingapropertythatsomephilosopherhas.Let’ssupposethatcommonnouns,suchas“linguist”,havesemanticvaluesoftype(e,i)(theyexpressproperties,suchasthepropertyofbeingalinguist.Since“somelinguist”isoftype((e,i),i),thedeterminer“some”mustbeoftype((e,i),((e,i),i)).“Some”,thatis,isafunctionthattakesapropertytoapropertyofproperties.Wecanspecifytheproperfunctionfor“some”asfollows:•sometakesasinputapropertyPandproducesasoutputthesecond-orderpropertyofbeingapropertythathasanon-emptyintersectionwithP.Abitofthoughtshowsthatthepropersemanticvaluefor“every”is:•everytakesasinputapropertyPandproducesasoutputthesecondorderpropertyofbeingapropertythathasPasasubset.Therearemanypropertiesofproperties,sowenowhavearobustspaceofpossiblesemanticvaluesforquantifiednounphrases,andhencefordeter-miners,totake.16Wecanhave,forexample:1.mosttakesapropertyPtothesecond-orderpropertyofbeingapropertywhoseoverlapwithPislargerthanitsoverlapwithnot-P.2.notakesapropertyPtothesecond-orderpropertyofbeingapropertythathasanemptyintersectionwithP.3.atmostthreetakesapropertyPtothesecond-orderpropertyofbeingapropertywhoseoverlapwithPcontainsatmostthreeobjects.4.thetakesapropertyPtothesecond-orderpropertyofbeingapropertysuchthatpropertyPholdsofexactlyoneobject,andholdingofthatobject.Thediscoveryofthisspaceofpossiblequantifiednounphrasespointstoanotheridea:•Idea#7:Aformalsemantictheorycanprovidenotonlyspecificsemanticvaluesforspecificlexicalitems,butalsocategoriesofsemanticvalues.639781472578235_txt_print.indd6308/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageIdentifyingthesecategoriescanthenaidinidentifyingdiverseconstruc-tionsorlexicalitemsassharingasemanticstructure,oringuidingcross-linguisticinvestigationintoappearancesofothermembersofthesamecategory.Wecanalsocharacterizegeneralfeaturesofquantifiednounphrases.Forexample,thereisaclassofmonotoneincreasingquantifiednounphrases:•Monotoneincreasing:AquantifiednounphraseQNPismonotoneincreasingif,givenanypropertyPthatinstantiatesQNPandanypropertyQsuchthateveryinstanceofPisaninstanceofQ,itfollowsthatQalsoinstantiatesQNP.17“Some”,“every”and“most”allproducemonotoneincreasingquantifiednounphrases.Monotoneincreasingphrasessupportinferencesofthegeneralform:1.Somelinguistsownredcars.Therefore,somelinguistsowncars.2.Everyphilosopherownsaredcar.Therefore,everyphilosopherownsacar.3.Mostmathematiciansownaredcar.Therefore,mostmathematiciansownacar.“No”,“atmostthree”and“exactlytwo”,ontheotherhand,arenotmonotoneincreasing,anddonotsupportthisinferenceform:1.Nolinguistownsaredcar.Butitdoesnotfollowthatnolinguistownsacar.2.Atmostthreephilosophersownaredcar.Butitdoesnotfollowsthatatmostthreephilosophersownacar.3.Exactlytwomathematiciansownaredcar.Butitdoesnotfollowthatexactlytwomathematiciansownacar.Theflipsideofmonotoneincreasingphrasesaremonotonedecreasingphrases:•AquantifiednounphraseQNPismonotonedecreasingif,givenanypropertyPthatinstantiatesQNPandanypropertyQsuchthateveryinstanceofQisaninstanceofP,itfollowsthatQalsoinstantiatesQNP.“No”and“atmostthree”bothproducemonotonedecreasingphrases,andsuchphrasessupporttheoppositedirectionoftheinferencessupportedbymonotoneincreasingphrases:649781472578235_txt_print.indd6408/08/201415:45\nFormalSemantics1.Nolinguistownsacar.Therefore,nolinguistownsaredcar.2.Atmostthreephilosophersownacar.Therefore,atmostthreephiloso-phersownaredcar.Thecategoriesofmonotonedecreasingandmonotoneincreasingquantifiednounphraseshaveanumberofpredictiveapplications.Forexample,manynaturallanguagescontainnegativepolarityitems,whichcanbeusedonlyinanegativeenvironment.18“Ever”isanegativepolarityiteminEnglish.Comparetheacceptable:(9)AlbertdoesnoteverreadHegel.totheunacceptable:(10)∗AlbertdoeseverreadHegel.Butaprecisecharacterizationof“negativeenvironments”isdifficult.Ahelpfulfirststepcanbemadebyobservingthatmonotonedecreasingquantifiednounphrases,butnotmonotoneincreasingandnotnon-monotonicphrases,licensenegativepolarityitems.Socomparetheacceptable:(11)AtmostthreephilosopherseverreadHegel.withtheunacceptable:(12)∗AtleastthreephilosopherseverreadHegel.and:(13)∗ExactlythreephilosopherseverreadHegel.Theformalsemanticframeworkthuspredictsthatadeterminernewlyintro-ducedtoEnglish,oradeterminerinanewlydiscoveredlanguage,willlicensenegativepolarityitemsifitproducesmonotonedecreasingquantifiednounphrases.Herewesee:•Idea#8:Carefulformalsemantictheorizingcanbearfruitbybringingintoengagementphenomenafromdistinctpartsofthelanguage.Westartwithtwopuzzles:howtoprovidesemanticvaluesforquantifiednounphrases,andhowtocharacterizetheenvironmentsinwhichnegativepolarityitemscanappear.Butthesetwopuzzlesinteract,sincethesyntacticflexibilityofthelanguageallowsthepossibilityofquantified659781472578235_txt_print.indd6508/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguagenounphrasesbeingoneofthelicensingenvironments.Thatinteractionthenallowsformaltoolsfromonephenomenontoinformtheorizingabouttheotherphenomenon,andprovidesindirectconfirmationfortheoriesofbothareas,byshowingthatthosetwotheoriesinteractwell.ScopeAmbiguitiesAmongthedeterminerssetoutabovewasthedefinitearticle“the”.Giventhecentralroleofthedefinitedescription“thetallestspy”inourproblematicbeliefreport,thisisapromisingsteptowardaformalsemanticanalysisoftheambiguity.Butitisonlyafirststep,sinceourtoolsstillassignasinglesemanticvaluetoeverysentenceandthusfailtocaptureambiguity.Asafirststeptowardthedesiredambiguity,weconsiderscopeambiguitiesarisingfrommultiplequantifiednounphrases.Considerthesentence:(14)Everyphilosopheradmiressomelinguist.Thissentencehastworeadings.Onwhatwewillcallthe“∀$”reading,itrequiresthateachphilosopherhavealinguistheadmires,butallowsthatdifferentphilos-ophermayadmiredifferentlinguists.Onthe$∀reading,itrequiresthattherebeasinglelinguistwhoisadmiredbyeveryphilosopher.Howcanweusethetype-theoreticformalsemanticsgivenabovetocapturethesetworeadings?Webeginbynotingthatthereisatypingproblemcreatedbyquantifiednounphrasesindirectobject,ratherthansubject,position.Recallthatwehavetypedquantifiednounphrases((e,i),i),andtransitiveverbs(e,(e,i)).Thecrudesyntacticstructureof(14)is[[Everyphilosopher][admires[somelinguist]].Thus“admires”and“somelinguist”mustsemanticallycombine.Butthetwotypes((e,i),i)and(e,(e,i))cannotcombineineitherorder.Thefirstrequiresanargumentoftype(e,i),andthesecondoftypee.Thusthecurrentframeworksimplyfailstoassignsemanticvaluestoexpressionswithquantifiednounphrasesindirectobjectposition.Oneapproach(amongmany)tothisproblembeginswiththesyntax.Theambiguitybetweenthe∀$and$∀readingwillformanypeoplealreadybefamiliarasascopeambiguity–roughly,anambiguityresultingfortwopossibleprocessingordersforthetwoquantifiers.Inaformalsystemlikefirst-orderlogic,theambiguityiseliminatedthroughasyntaxinwhichquantifiersequencingismandatorilydeterminedbythelinearorderingofthequantifiersinthesentence.∀x$yRxyreceivesonlythe∀$reading,and$y∀xRxyreceivesonlythe$∀reading.InEnglish,theprocessingorderofquantifiersneednotmatchtheirlinearordering(hencetheavailabilityofa$∀readingof(14).ButwecanreproduceinEnglishtheformal-systemsolutionofmatchinglinearorderingandprocessingorderingofquantifiersifweadoptabolder669781472578235_txt_print.indd6608/08/201415:45\nFormalSemanticssyntactictheory.Thesurfaceformof(14)hasthetwoquantifiersinaspecificlinearordering(namely,universalfollowedbyexistential).Butmanysyntactictheoriesposithiddenlevelsofsyntacticrepresentation,inwhichthecomponentsofasentencemaynotappearinthesamepositionsastheydoinsurfaceform.Forexample,asyntactictheorymayincorporatearuleofQuantifierRaising,orQR:•QR:ThereisasyntacticlevelofrepresentationcalledLogicalForm,orLF.LFisderivedfromsurfaceformviatheapplicationofaruleQR.QRrequireseachquantifiednounphrasetobemovedfromitspositioninsurfaceformandadjoinedtoaS(entence)structureofwhichitis,insurfaceform,apart.19Considerfirstasimpleexample.Thesentence:(15)Albertseesmanylogicians.containsaquantifiednounphraseindirectobjectposition.Supposethesurfaceformsyntacticorganizationof(15)is:•(15s)[[Albert][[sees][manylogicians]]]SNPVPTVNP“Manylogicians”mustthen,atthelevelofLF,moveoutofitssurfacepositionandadjointotheSnodeattherootofthesyntactictree.AdjoininganexpressiontoanSnodeinvolves(a)creatinganewSnode,and(b)placingboththeadjoinedexpressionandtheoriginalSnodeimmediatelybelowthennewSnode.TheresultofapplyingQRto(15)isthen:•(15l)[[manylogicianst][[Albert][[sees][t]]]]SNPSNPVPTVNPWhen“manylogicians”movesfromitssurfaceposition,itleavesbehindatrace–an“invisible”lexicalitemwithoutanyphonologicalcontent.20Whyshouldweacceptaquantifierraisingstoryaboutsyntax?Herearesomebasicconsiderations:1.Considerthebehaviourofwh-phrasesinEnglish,suchas“whichlinguist”and“whatphilosopher”.Thesephrasesseemtoundergomovementalreadyatsurfaceform.Thusin:(16)WhichlinguistdoesJohnadmire?thephrase“whichlinguist”appearstobedisplacedfromitsnaturallocationafter“admires”(sinceitisthedirectobjectofthatverb)toaninitialpositionadjoinedtothemainS.Similarlywithimbeddedquestions,asin:679781472578235_txt_print.indd6708/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguage(17)JohnknowswhatphilosopherMarymet.inwhich“whatphilosopher”againappearsdisplacedfromitsnaturallocationafter“met”.Thisinitialimpressionofmovementcanbebolsteredbyconsideringcaseassignment.Thuscompare:(18)JohnknowswhomMarymet.(19)∗JohnknowswhoMarymet.Themandatoryaccusativecaseon“whom”isexplainedif“whom”beginslifeasthedirectobjectof“met”,andthenundergoesmovementtotheadjoinedposition.Anotherconsideration:comparethepair:(20)WhichphilosopherdoesJohnwannainvitetotheparty?(21)∗WhichphilosopherdoesJohnwannainviteSusantotheparty?In(20),butnotin(21),theinformalcontractionof“wantto”to“wanna”isacceptable.Thedifferenceinacceptabilityisexplicableifweassumethat,inbothsentences,thewh-phrase“whichphilosopher”isdisplacedfromanoriginalposition:(22)Johnwantstoinvitewhichphilosophertotheparty?(23)JohnwantswhichphilosophertoinviteSusantotheparty?Ifweassume,asabove,thatmovementleavesatraceintheoriginalposition,thenin(21),butnotin(20),atraceintervenesinvisiblybetween“want”and“to”,blockingthe“wanna”contraction.212.Wh-phrasesshareasyntacticcategorywithquantifiednounphrases.Botharetheresultofcombiningadeterminerwithanounphrase.Thisisaprimafaciereasontothinkthatthesamemovementpossibilitiesapplytoboth.Buttheprimafaciecasecanbestrengthened.Movementofwh-phrasesisunavailableinsomecontexts.Forexample,wh-phrasescannotmoveoutofcoordinatedconjunctions.Wecanhave:(24)JohnsawChomskyandwhichphilosopher?butnot:(25)∗WhichphilosopherdidJohnseeChomskyand?Butwealsodiscoverthatcoordinatedconjunctionsmakeunavailablecertainscopereadingsofsentenceswithquantifiednounphrases.Thusin:(26)SomelinguistsawChomskyandeveryphilosopher.thereisno∀$reading,requiringthat,foreveryphilosopher,thereis689781472578235_txt_print.indd6808/08/201415:45\nFormalSemanticssomelinguistwhosawbothChomskyandthatphilosopher.Ingeneral,invertedscopereadingsareunavailableinexactlythesamecontextsinwhichwh-phrasescannotbemoved.IftheapplicationofQRcanexplainscopeambiguities,assuggestedbelow,thenthisconvergenceisfurtherevidencethatquantifiednounphrasesshare,albeitatlogicalform,themovementbehaviourofwh-phrases.Thesesyntacticconsiderationsbringout:•Idea#9:Becauseaformalsemantictheoryseeksaclosealignmentbetweensyntacticstructureofexpressionsandthecalculationofsemanticvaluesofthoseexpressions,itbecomespossibleforevidencethatbearsdirectlyonsyntactictheoriestobearindirectlyonsemantictheories,andimprovemethodologicaltoolsforconfirmingordisconfirmingthosetheories.22NowconsidertheapplicationofQRto(14).Wecan,forexample,movethequantifiednounphrase“somelinguist”,tocreatethestructure:•(14)l[[somelinguist][[everyphilosopher][[admires][t]]]]1SNPtSNPVPTVNPQRmustthenbeappliedagaintothequantifiednounphrase“everyphilos-opher”.IfweassumethatagivenSnodecanhaveonlyasinglephraseadjoinedtoit,wearethenforcedtoadjoin“everyphilosopher”tothenew,higher,S:•(14)l[[everyphilosopher][[somelinguist][[u][[2SNPuSNPtSNPVPTVadmires][t]]]]]NPHowever,wecouldinsteadhavemade“everyphilosopher”thetargetofthefirstapplicationofQR,creating:•(14)l[[everyphilosopher][[u][[admires][some3SNPuSNPVPTVNPlinguist]]]]andthenappliedQRto“somelinguist”,tocreatetheLF:•(14)l[[somelinguist][[everyphilosopher][[u][[4SNPtSNPuSNPVPTVadmires][t]]]]]NPWethusareabletoderivetwodistinctlogicalformsfor(14).Inone,thequantifiersarelinearlyorderedina∀$manner;intheother,thequantifiersarelinearlyorderedinan$∀manner.Theexistenceoftwologicalformsforthesamesurfaceformissignificantbecauseof:699781472578235_txt_print.indd6908/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguage•Idea#10:Aformalsemantictheorycanexplainambiguousexpressionsbycombiningwithasyntactictheorythatassignstwodistinctsyntacticformstotheambiguousexpression,andthencalculatingsemanticvaluesfromthosetwoforms.Therelationbetweensurfaceformandlogicalformisone-many,whiletherelationbetweenlogicalformandsemanticvalueisone-one.Tree1,p.70Thebracketnotationusedabovecanbereplaced,foreasiervisualization,withtreenotation.Onthetreenotation,thesurfaceformof(14)is:SNPVPEveryphilosopherTVNPadmiressomelinguistTree(14)l2,p.70Thetwopossiblelogicalformsare:1.(14)lS2()Dever:Trees27deoctubrede20112/8NPSeveryphilosopheruNPSsomelinguisttNPVPuTVNPTree(14)l4,p.70admirest2.(14)lS4()Dever:Trees27deoctubrede20113/8NPSsomelinguisttNPSeveryphilosopheruNPVPuTVNPadmirest70()Dever:Trees27deoctubrede20114/89781472578235_txt_print.indd7008/08/201415:45\nFormalSemanticsThemoresophisticatedsyntaxthusallowsustoassigntwodistinctsyntacticformsto(14),onewitheachofthedesiredquantifierscopings.RecallthegoalthatpromptedinvestigationofasyntaxwithaQRrule.Wewanttogiveaformalsemanticexplanationofthetwotruth-conditionalreadingsof(14).WithQRinhand,wenowhavetwodifferentlogicalformsfor(14).Thisisapromisingstart,butthetwologicalformswillbehelpfulonlyifwecancouplethemwithsemanticrulesthatprocessthemintothedesired∀$and$∀readings.Butherewequicklydiscoverthatourearliertypingproblemspersist.Considerfirstasimplequantifiedsentencesuchas(15).Wehavetakenonthefollowingcommitments:•Thesemantictypeof“Albert”(apropername)ise.•Thesemantictypeof“sees”(atransitiveverb)is(e,(e,i)).•Thesemantictypeof“manylogicians”(aquantifiednounphrase)is((e,i),i).•ThesemantictypeofanySnodeisi.Ontheplausibleassumptionthatatraceisoftypee,thelowerSnode[[SNPAlbert][[sees][t]]]worksouttobeoftypei,asdesired.ButnowweVPTVNPmustcombineanexpressionoftypeiwithoneoftype((e,i),i)toproduceaterminalsemantictypeofi,andneitherdirectionofcombinationworks.Againthesemanticprocessingcrashes.GiventheQRsyntax,quantifiednounphraseswillappearinLFasadjunctstoSnodes,andhencewillreliablybeitemsthatcombinewithanitemoftypeitoproduceanitemoftypei.Thisthensuggeststhatquantifiednounphrasesneed,inthissyntacticsetting,tobeoftype(i,i),ratherthanoftype((e,i),i).Butasemantictypeof(i,i)ispuzzlingforquantifiednounphrases.Recallthatthetypeiconsistsoffunctionsfromworldstotruthvalues.Truthvaluesaloneareclearlyinsufficientforgivingthesemanticsofquantifiednounphrases.Ifquantifiednounphrasesareoftype(t,t),wheretisthetypeoftruthvalues,thenquantifiednounphrasesareservingsemanticallyasmonadicfunctionsfromtruthvaluestotruthvalues(justlikenegation).Butthiscannotyieldtherightresults–thereareonlyfoursuchfunctions,andtherearemorethanfourlogicallydistinctquantifiednounphrases.Anditisnotallobvioushowaddingworldsandintensionswillhelp,sincequantifiednounphrasesappeartobeextensional,andcanbeusedinpurelyextensionalcontexts.ComplicatingtheBaseWhathasgonewronghere?OneanswercanbereachedbyconsideringthestandardTarskiansemanticsforquantifiedlogic.23Onthissemantics,the719781472578235_txt_print.indd7108/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguagefundamentalsemanticconceptsaretruth,referenceandsatisfaction(inamodel)relativetoanassignmentfunction.Assignmentfunctionsdirectlydeterminethereferenceofvariablesandderivatively,viaarecursiveprocedure,thereferenceofallterms.Sentences(bothopenandclosed)thenaresatisfiedornotbyassignmentfunctions,andtherecursivesatisfactionclausesdefine,forexample,satisfactionofquantifiedsentencesintermsofthesatisfactionoftheircomponentparts,asinthefamiliar:•$xAissatisfiedbyassignmentfunctionσiffthereisanassignmentfunctionσ',differingfromσatmostinitsassignmenttox,suchthatσsatisfiesA.WecanintroducetheseTarskianideasintoatype-theoreticsemanticsbyreplacingthebasictypeseandiwithrelativizedtypese∗,t∗,andi∗.e∗isthetypeofassignment-function-relativeobjects,orfunctionsfromassignmentfunctionstoobjects.t∗isthetypeofassignment-function-relativetruthvalues.,orfunctionsfromassignmentfunctionstotruthvalues.i∗,finally,isthetypeofmappingsfromworldstoitemsoftypet∗.Typetcontainedonlytwoentities–theTrueandtheFalse–andthusprovidedtoofewoptionsforquantifiers.Typet∗ismuchlarger–ifassignmentfunctionshaveinfinitedomains,thent∗isuncountableinsize.Theseassignment-function-relativesemanticvaluesrepresentanothersignificantidea:•Idea#11:Thetechnicalflexibilityofatheorycanbesubstantiallyincreasedbymakinguseofvariousformsofindexedtruth.Truthrelativetoanassignmentfunctionisoneexample,asaretruthrelativetoaworld,atime,orastanceofassessment.Versionsofindexedtruthallowforawiderangefornon-extensionaloperatorsthatshiftthevalueofindicesinvariousmanners.Indexedtruthcanbeatheoryinternalsemanticnotion,appearingonlyintheinternalcalculationofsemanticvaluesofcomponentpartsofexpressions,butgivingwaytotruthsimpliciterforentiresentences.Suchanapproachcanbethoughtofashavingtwotypesofsemanticvalue–ingredientsemanticvalue,exploitingtheindexedtruth,andassertoricsemanticvalue,usingtruthsimpliciteranddeterminingthewayinwhichlinguisticagentsengagewithutterances.Orindexedtruthcanpersistthroughoutthetheory,sothatterminalsemanticvaluesarealsogivenintermsofindexedtruth–inthiscase,thepragmatictheorywillneedtobeconstructedtoreceiveandengagewithindexedtruth,asinrecentversionsoftruth-relativism.24Propernameswillreceiveassemanticvaluesfunctionsthatareindifferenttotheinputassignmentfunction,asin:729781472578235_txt_print.indd7208/08/201415:45\nFormalSemantics•Albertisafunctionfsuchthatforallσ,f(σ)=Albert.Traces,ontheotherhand,willreceiveassemanticvaluesfunctionsthataresensitivetotheinputassignmentfunction,asin:•tisafunctionfsuchthatforallσ,f(σ)=σ(t).Manyofthesemanticdetailscanbeeasilytransposedfromthepreviousframework.Theintransitiveverb“snores”,forexample,willbeofsemantictype(e∗,i∗),andwilltakeasitsspecificsemanticvalueinthistypethefollowing:•snoresisafunctionfthattakesasinputafunctiongfromassignmentfunctionstoobjects,andoutputsafunctionthatmapsagivenworldwtoafunctionhthatmapsanassignmentfunctionσtotheTruejustincaseg(σ)snoresinw.Quantifiednounphrasesnowcompositionallyreceiveanaturalandeleganttreatmentasexpressionsofsemantictype(i∗,i∗).Forexample,“somelinguist”receivesthefollowingsemanticvalue:•somelinguistisafunctionfthattakesasinputafunctionthatmapsteachworldwtoafunctiongfromassignmentfunctionstotruthvalues,andoutputsafunctionthatmapswtoafunctionthatmapsanassignmentfunctionσtotheTruejustincasethereisanobjectothatisalinguistinw,andanassignmentfunctionσtthatdiffersfromσatmostinitsassignmenttot,andgmapsσttotheTrue.–(Alternatively,let[σ]bethesetofassignmentfunctionsthatdifferfromtσatmostinthetposition.Thensomelinguisttakesasinputafunctiontthatmapseachworldwtoasetofassignmentfunctions,andproducesasoutputafunctionthatmapswtothoseassignmentfunctionsσsuchthatσthasanon-emptyintersectionwiththesetofassignmentfunctionswhoset-valueisalinguistinw.)Inordercompositionallytoachievethissemanticvaluefor“somelinguist”,“some”mustreceiveasemanticvalueinthetype((e∗,i∗),(i∗,i∗)).Inparticular,itmustreceiveavalueinthattypethattakesasinputasemanticvalueofthetype(e∗,i∗)(thetypeofcommonnouns),andproducesasoutputasemanticvalueofthetype(i∗,i∗),wheretherelevantoutput(i∗,i∗)itemisofthesortdescribedinthesemanticvalueof“somelinguist”,butwiththespecificationthatoisalinguistinwweakenedtotherequirementthatobeanobjectsuchthatanyfunctionmappingeveryassignmentfunctiontoitcombinewiththe739781472578235_txt_print.indd7308/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageinput(e∗,i∗)semanticvaluetoproduceafunctionthatmapseveryassignmentfunctiontruthinw.Whentheappropriatesemanticvaluesarecombinedwiththesyntacticforms(14l)and(14l),thedesired∀$and$∀truthconditionsfallout.24ASemanticPuzzleSolvedWearenow(finally!)inapositiontoofferapreliminaryanalysisoftheproblematicbeliefreportsentence:(1)JohnbelievesthatthetallestspyisinParaguay.Webeginwithasimplesurfaceformsyntacticanalysisof(1):(1)sf[[John][[believes[[that][[thetallestspy][isinSNPVPSVCPCSNPVPParaguay]]]]]BecausetherearetwoSnodesin(1sf),therearetwolandingspotsforthequantifiednounphrase“thetallestspy”whenQRisapplied.Thetworesultinglogicalformsare:(1)l[[John][[believes[[that][[thetallestspy][[t][1SNPVPSVCPCSNPtSNPVPTreep.74isinParaguay]]]]]]•SNPVPJohnSVCPbelievesCSthatNPSthetallestspytNPVPtisinParaguay()Dever:Trees27deoctubrede20115/8749781472578235_txt_print.indd7408/08/201415:45\nFormalSemantics(1)l[[thetallestspyt][[John][[believes[[that][[t][2SNPSNPVPSVCPCSNPVPisinParaguay]]]]]]Treep.75•SNPSthetallestspytNPVPJohnSVCPbelievesCSthatNPVPtisinParaguay()Dever:Trees27deoctubrede20116/8Thesemantictypingworksoutproperly.Letthesententialverb“believes”beoftype(i∗,(e∗,i∗)),andthecomplementizer“that”beoftype(i∗,i∗).Then:1.In(1)l,thetypee∗tracetcombineswiththetype(e∗,i∗)verbphrase“is1inParaguay”toformanexpressionoftypei∗.Thisthencombineswiththetype(i∗,i∗)quantifiednounphrase“thetallestspy”toformanotherexpressionoftypei∗.Thiscombineswiththetype(i∗,i∗)complementizer“that”toformyetanotherexpressionoftypei∗.Thiscombineswiththetype(i∗,(e∗,i∗))“believes”toformanexpressionoftype(e∗,i∗).Finally,thiscombineswiththetypee∗“John”toformthefinalsemantictypeofi∗.2.In(1)l,thetypee∗tracetcombineswiththetype(e∗,i∗)verbphrase“is2inParaguay”toformanexpressionoftypei∗.Thisthencombineswiththetype(i∗,(e∗,i∗))“believes”toformanexpressionoftype(e∗,i∗).Thiscombineswiththetypee∗“John”toformanexpressionoftypei∗.Thisthencombineswiththetype(i∗,i∗)quantifiednounphrase“thetallestspy”toformthefinalsemantictypei∗.“Thetallestspyt”willthentakeasitsspecificsemanticvalueinthetype(i∗,i∗)afunctionfthattakesasinputafunctionthatmapseachworldwtoafunctiongfromassignmentfunctionstotruthvalues,andoutputsafunctionthatmapswtoafunctionthatmapsanassignmentfunctionσtotheTruejustincasethereisauniqueobjectothatisthetallestspyinw,andanassignmentfunctionσ'thatdiffersfromσatmostinitsassigmenttot,andgmapsσ'totheTrue.759781472578235_txt_print.indd7508/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageWhenthepiecesareassembled,wegetthefollowingresults:1.(1)listruewithrespecttoaworldwjustincaseineveryworlduthatis1oneofJohn’sbeliefworldsinw,thereisauniqueobjectthatisthetallestspy(inu),andthatobjectisinParaguay(inu).2.(1)listruewithrespecttoaworldwjustincasethereisauniqueobject2thatisthetallestspy(inw),andineveryworlduthatisoneofJohn’sbelief-worldsinw,thatobjectisinParaguay(inu).Inthefirstreading,thequantifiednounphraseisprocessedinsidethebeliefcontext,andhenceitselectsforassignmentfunctionswhosetvalueisthetallestspyinthebeliefworld,ratherthanintheactualworld.ThisreadingisthuscompatiblewiththerebeingnodeterminateobjectaboutwhichJohnthinks(becausethesatisfierof“tallestspy”variesfrombelief-worldtobelief-world)andalsowiththerebeingnoactualtallestspy.Itstruthconditionsthusmatchthoseofthededictoreadingof(1).Inthesecondreading,thequantifiednounphraseisprocessedoutsidethebeliefcontext,andhenceitselectsforassignmentfunctionswhosetvalueistheactualtallestspy.ThequestionofwhetherthatobjectisinParaguayineachofJohn’sbeliefworlds–butnotthequestionofwhetherthatobjectisthetallestspyineachofJohn’sbeliefworlds–isthenconsidered.ThisreadingisthuscompatiblewithJohnbelievingthatnooneinParaguayisaspy.Itstruthconditionsthusmatchthoseofthederereadingof(1).Wenowhaveaformalsemantictheorythatinasystematicwayassignstwodistinctsemanticvaluesto(1),semanticvaluesthatfitnaturallywiththetworeadingsweinitiallyidentified.TwoPredictionsTheanalysisofdereanddedictoreadingsofbeliefreportsdevelopedabovecorrelatesthedere/dedictodistinctionwiththedistinctionbetweenaquantifiednounphrasetakingwidescope(outsidethebeliefcontext)andtakingnarrowscope(insidethebeliefcontext).Inthissection,weextracttwoquickconse-quencesofthismechanismforexplainingdereanddedictoreadings,andconsidertheramificationsfortheformalsemanticaccountofthosetwoconsequences.Supposeabeliefreportcontainstwoquantifiednounphrasesinthecomplementsentencetotheverb“believes”.Thenforeachnounphrasethereisaquestionaboutwhetheritistobeinterpretedinadereoradedictomanner.Thusconsider:(27)Johnbelievesthetallestspyfollowedtheshortestdiplomat.769781472578235_txt_print.indd7608/08/201415:45\nFormalSemanticsAsbefore,wecanread(27)suchthatthereisinfactnotallestspy,orsuchthatthereisnodeterminatespythatJohnthinksabout–producingareadingdedictowithrespectto“thetallestspy”.Wecanalsoread(27)suchthatJohnbelievesofanindividualwho,unbeknownsttoJohn,isthetallestspy–areadingderewithrespectto“thetallestspy”.Thesameoptionsareavailablefor“theshortestdiplomat”.Perhapsthereisnoshortestdiplomat,orperhapsthereisnodeterminatediplomatthatJohnthinksabout(readingsdedictowithrespectto“theshortestdiplomat”).PerhapsJohnhasabeliefaboutanindividualwho,unbeknownsttohim,istheshortestdiplomat(readingsderewithrespectto“theshortestdiplomat”).25Therearethuspotentiallyfourreadingsof(27):1.Dedictowithrespecttoboth“thetallestspy”and“theshortestdiplomat”.2.Dedictowithrespectto“thetallestspy”;derewithrespectto“theshortestdiplomat”.3.Derewithrespectto“thetallestspy”;dedictowithrespectto“theshortestdiplomat”.4.Derewithrespecttoboth“thetallestspy”and“theshortestdiplomat”.And,conveniently,thereareforeachofthesereadingsappropriatewaysofscopingthetwoquantifiednounphrases:1.[[John][[believes][[that][[thetallestspy][[theSNPVPSVCPCSNPtSNPshortestdiplomat][[t][[followed][u]]]]]]]]uSNPVPTVNP2.[[theshortestdiplomat][[John][[believes][[that][[SNPuSNPVPSVCPCSNPthetallestspy][[t][[followed][u]]]]]]]]tSNPVPTVNP3.[[thetallestspy][[John][[believes][[that][[theSNPtSNPVPSVCPCSNPshortestdiplomat][[t][[followed][u]]]]]]]]uSNPVPTVNP4.[[thetallestspy][[theshortestdiplomat][[John][[SNPtSNPuSNPVPSVbelieves][[that][[t][[followed][u]]]]]]]]CPCSNPVPTVNPNotethatfor“thetallestspy”tobedereand“theshortestdiplomat”tobededicto,thefirstmusttakewidescopeoverthebeliefcontextandthesecondmusttakenarrowscopeinsidethebeliefcontext.Itthenfollowsthatthefirsttakesscopeoverthesecond.Inthisparticularexample,thisdoesnotmatter,becausetherelativescopingofthetwodefinitedescrip-tionsdoesnotaffectthetruthconditions.Butwithdifferentdeterminers,itwillmatter.779781472578235_txt_print.indd7708/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageConsiderthesentence:(28)Johnbelievesthatsomelinguistfollowedeveryphilosopher.Inthissentence,thereisbothanambiguitybetween∀$and$∀readingsofthetwoquantifiednounphrases,andalsoanambiguityforeachnounphrasebetweenbeingreaddedictoandreaddere.Therearethusinprincipleeightdistinctreadingsofthesentence(pickascopingandaItemsp.78,Idedicto/derereadingforeachquantifiednounphrase).However,thesyntaxonlyproducessixpossiblescopings:1.[S[NPJohn][VP[SVbelieves][CP[Cthat][S[NPsomelinguistt][S[NPeveryphilosopheru][S[NPt][VP[TVfollowed][NPu]]]]]]]]•∃∀,dedicto,dedicto2.[S[NPJohn][VP[SVbelieves][CP[Cthat][S[NPeveryphilosopheru][S[NPsomelinguistt][S[NPt][VP[TVfollowed][NPu]]]]]]]]•∀∃,dedicto,dedictoItemsp.78,II3.[S[NPeveryphilosopheru][S[NPJohn][VP[SVbelieves][CP[Cthat][S[NPsomelinguistt][S[NPt][VP[TVfollowed][NPu]]]]]]]]•∀∃,dere,dedicto4.[S[NPsomelinguistt][S[NPJohn][VP[SVbelieves][CP[Cthat][S[NPeveryphilosopheru][S[NPt][VP[TVfollowed][NPu]]]]]]]]•∃∀,dedicto,dere()Dever:Trees27deoctubrede20117/85.[S[NPsomelinguistt][S[NPeveryphilosopheru][S[NPJohn][VP[SVbelieves][CP[Cthat][S[NPt][VP[TVfollowed][NPu]]]]]]]]•∃∀,dere,dere6.[S[NPeveryphilosopheru][S[NPsomelinguistt][S[NPJohn][VP[SVbelieves][CP[Cthat][S[NPt][VP[TVfollowed][NPu]]]]]]]]•∀∃,dere,dereThemissingcombinationsare:•Areadingwitha$∀scoping,thatisdedictowithrespectto“somelinguist”andderewithrespectto“everyphilosopher”.()Dever:Trees27deoctubrede20118/8•Areadingwitha∀$scoping,thatisdedictowithrespectto“everyphilos-opher”andderewithrespectto“somelinguist”.Thecurrentformalsemantictheorythuspredictsthatthesetwopotentialreadingscannot,infact,beobtainedfromtheEnglishsentence(28).Thecrucial789781472578235_txt_print.indd7808/08/201415:45\nFormalSemanticsquestion,then,iswhetherthispredictionisborneout.Considerthefollowingscenario:S3:Susan,Albert,andMaryarelinguists.Johnknowseachofthem,butisunawarethatanyarelinguists.Johnisundertheincorrectimpressionthatthereareexactlythreephilosophers(althoughhetakeshimselfnottoknowwhotheyare,andalthoughinfacttherearenophilosophers).HethinksthatSusanfollowedoneofthephilosophers,Albertfollowedanother,andMaryfollowedathird.IsthisascenarioinwhichJohniscorrectlydescribedasbelievingthatsomelinguistfollowedeveryphilosopher(onanydisambiguationofthatbeliefreport)?Judgmentsaboutsuchintricatecasesaredelicateandunreliable(andoneofthepointsofinterestofformalsemantictheoriesisthattheybringtolight,byhighlightingthesignificanceofcasessuchasthis,importantmethodologicalquestionsaboutthesourceandstabilityofdataforlinguistictheorizing).Butitatleastappearsthatthescenariocannotbesoreported.Ifthisiscorrect,itisasignificantvalidationforthegivenformalsemanticapproach.Wenowconsiderasecondconsequenceofthescopingexplanationofthededicto/deredistinction.WenotedearlierthatthesyntacticruleofQRcannotbeappliedtoquantifiednounphrasesincertaincontexts,suchasincoordinatedconjunctions.Considerbeliefreportscontainingquantifiednounphrasesinsuchcoordinatedpositions,asin:(29)JohnbelievesthatAlbertandthetallestspyareinParaguay.(30)JohnbelievesthatAlbertandeveryphilosopherareinParaguay.IfderereadingsofquantifiednounphrasesaregeneratedbytheuseofQRtoliftthequantifiednounphraseoutofthebeliefcontext,andiftheapplicationofQRisblockedbythecoordinationconstruction,thenthequantifiednounphrasesin(29)and(30)shouldhaveonlydedictoreadings.Isthiscorrect?Considerthefollowingtwoscenarios:S4:Susanisthetallestspy.AlthoughJohnisacquaintedwithSusan,heisunawarethatsheisthetallestspy.JohnthinksthatAlbertandSusanarebothinParaguay.S5:Susan,Albert,andMaryareallthephilosophers.Johnknowseachofthem,butisunawarethatanyarephilosophers.Hebelievesthatthethreeofthem,togetherwithAlbert,areallinParaguay.If(29)istrueinS4and(30)istrueinS5,then–contrarytothepredictionsof799781472578235_txt_print.indd7908/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguagethetheorywehavedeveloped–thosetwobeliefreportsdohavederereadings.Infact(althoughthedataisagainmurkyanddelicate),itseemsthat(29)canbereadastrueinS4,but(30)cannotbereadastrueinS5.Thetheoryisthushalfwayverified–itproducesonecorrectandoneincorrectprediction.Halfwayisn’tgoodenough,ofcourse,sosomethingmustbechanged.Perhapssomenewmechanismneedstobeadded;perhapssomequantifiednounphrases,suchasdefinitedescriptions,needadifferenttreatmentfromthatdevelopedhere;perhapstheentirescope-basedapproachisfundamentallymistaken.26Butwehavenowfollowedthislineasthoughtasfaraspossibleinthecurrentcontext.ConclusionInthischapterwehaveexploredonepartofoneapproachtodevelopingaformalsemanticanalysisofonepuzzlingconstruction.Thereismoretosaytofinishdevelopingthisapproach;thereareotherradicallydifferentapproachestothesamepuzzlingconstruction;therearevastlymanyotherinterestingcasestotheorizeabout.Wehavesetoutsomeoftheimportanttechnicaltools,someoftheexplanatorygoals,andsomeofthemethodologicalconstraintsofformalsemantics.Butmuchofthefullstrengthoftheapproachofformalsemanticsliesinattemptingtosatisfysimultaneouslyconstraintsimposedbyproblematicconstructionsfromalldifferentpartsofthelanguage,andbytheinteractionsamongtheseconstructionsinmorecomplexsentences.Constructingatheorythatdescribesinapreciseandrigorousmannerthewayinwhichthemeaningsofdiversepartsofthelanguageinteractisanextraordinarilydifficulttask,butonewhichisbothessentialtoafullunderstandingofhumanlinguisticexchangesandwhichpromisestorevealmanyunexpectedfeaturesofourintricatelinguistictools.Notes1.Theobservationoftheambiguityisintroducedintothecontemporaryphilo-sophicaldiscussionbyQuine,1956.2.Fordetailedexaminationofthenatureandphilosophicalapplicationsofpossibleworlds,seeLewis,1986,andDivers,2002.3.Thefinitudeconstraintasbeenemphasizedby,e.g.,Davidson,1984.4.See,forexample,Parikh,1987,Stalnaker,1991,andchapter9ofFaginetal.,1995.5.PossibleworldssemanticsisendorsedasasemanticframeworkthroughoutStalnaker’swork,asinStalnaker,1979.TheideaofpossibleworldssemanticsreachesbackatleastasfarasKripke,1959.809781472578235_txt_print.indd8008/08/201415:45\nFormalSemantics6.SeeKratzer,1977,fordevelopmentofthisobservation.7.See,forexample,Kratzer,1981,foraproposalfortreating“probably”withinthisframework,Yalcin,2010,forsomecriticismsofthatproposal,andvonFintelandIatridou,2008,forsomediscussionoftreating“ought”inthisframework.8.ThisisaslightsimplificationoftheframeworkformodalsgiveninKratzer,1981.9.ThecanonicalworkoncontextsensitivityofthisoverttypeisinKaplan,1989.10.Theuseofapossibleworldssemanticstomodelbeliefcontexts,andotherpropositionalattitudecontexts,goesbacktoHintikka,1969.11.Indoingso,wefollowthemodelofMontague,2002.12.Forthoroughdevelopmentofthisidea,seeMontague,2002,andLewis,1970.TheunderlyingideaofacollectionoftypesofthissortgoesbacktoAjdukiewicz,1967.13.Thelambda-calculus,whichprovidesaformaltheoryforthistypeofnotation,wasdevelopedbyChurch,1932.14.TheeventsemanticsforadverbsisintroducedinDavidson,1967.15.SeeLewis,1970,forthisobservationonthesortofobjectsneededtotreatquantifiednounphrasesasofsemantictypee.16.SeeBarwiseandCooper,1980,andKeenanandWesterstahl,1997,foroverviewsofgeneralizedquantifiertheory.Russell’s(1905)theoryofdefinitedescriptionsisacrucialearlypointintherecognitionthatthereisacommonsemanticframeworkforquantifiednounphrasesinnaturallanguage.17.SeeBarwiseandCooper,1980,forfurtherdiscussionofmonotoneincreasingandmonotonedecreasingaspropertiesofquantifiers.18.SeeLadusaw,1979,foraclassicdiscussionofnegativepolarityitems.19.TheruleofquantifierraisingisintroducedinMay,1977,andnowfeaturesinmanystandardsyntactictheories.20.TheideaofatraceisintroducedinChomsky,1973.21.Theconnectionbetweentracesand“wanna”contractionismadeinChomsky,1976.22.Thisconnectionbetweensyntacticandsemantictheoriesisacentralthoughtbehindtheprincipleofcompositionality.SeeDever,2006,foranoverviewofcompositionality.23.SeeTarski,1983,fortheclassichistoricalwork,andstandardtextbooksonfirst-orderlogic,suchasEnderton,2001,formoreconcisepresentationofthecentralideas.24.See,forexample,Kölbel,2002,MacFarlane,2003and2005,Eganetal.,2005,andEgan,2007,forrecentdevelopmentsoftruth-relativism.25.Kripke,1989,considersasimilarcaseinwhichthescopetheoryofthededicto/deredistinctionistestedusingaconstructioninwhichaquantifiednounphraseisembeddedundertwopropositionalattitudeverbs.26.See,forexample,Kamp,1981,andJacobson,2000,fortheoriesthattreatsomequantifiednounphrasesquitedifferentlyfromthetreatmentdevelopedhere,andprovidetoolsforadifferentstyleofanalysisofambiguitiesinattitudecontexts.819781472578235_txt_print.indd8108/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageReferencesAjdukiewicz,K.(1967),“Syntacticconnexion”.InS.McCalled.PolishLogic.Oxford:ClarendonPress.Barwise,J.andCooper,R.(1980),“Generalizedquantifiersandnaturallanguage”.LinguisticsandPhilosophy,4:159–219.Chomsky,N.(1973),“Conditionsontransformations”.InS.AndersonandP.Kiparsky(eds),AFestschriftforMorrisHalle.NewYork:Holt,ReinhartandWinston.—(1976),“Conditionsonrulesofgrammar”.LinguisticAnalysis,2:303–51.Church,A.(1932),“Asetofpostulatesforthefoundationoflogic”.AnnalsofMathematics,33:346–66.Davidson,D.(1967),“Thelogicalformofactionsentences”.InN.Reschered.TheLogicofDecisionandAction.Pittsburgh:UniversityofPittsburghPress.—(1984),“Theoriesofmeaningandlearnablelanguages”.InInquiriesIntoTruthandInterpretation.Oxford:OxfordUniversityPress.Dever,J.(2006),“Compositionality”.InE.LeporeandB.Smith(eds),TheOxfordHandbookofPhilosophyofLanguage.Oxford:OxfordUniversityPress.Divers,J.(2002),PossibleWorlds.London:Routledge.Egan,A.(2007),“Epistemicmodals,relativism,andassertion”.PhilosophicalStudies,133:1–22.Egan,A.,Hawthorne,J.andWeatherson,B.(2005),“Epistemicmodalsincontext”.InG.PreyerandG.Peter(eds),ContextualisminPhilosophy.Oxford:OxfordUniversityPress.Enderton,H.(2001),AMathematicalIntroductiontoLogic.NewYork:AcademicPress.Fagin,R.,Halpern,J.,Moses,Y.andVardi,M.(1995),ReasoningaboutKnowledge.Cambridge,MA:TheMITPress.Hintikka,J.(1969),Semanticsforpropositionalattitudes.InJ.D.Daviset.al.(eds),PhilosophicalLogic.Dordrecht:Reidel.Jacobson,P.(2000),“Paycheckpronouns,Bach-Peterssentences,andvariable-freesemantics”.NaturalLanguageSemantics,8:77–155.Kamp,H.(1981),“Atheoryoftruthandsemanticrepresentation”.InGroenendijk,J.,Janssen,T.andStokhof,M.(eds),Truth,Interpretation,andInformation:SelectedPapersFromtheThirdAmsterdamColloquium.Dordrecht:Foris.Kaplan,D.(1989),“Demonstratives”.InJ.Almog,J.PerryandH.Wettstein(eds),ThemesFromKaplan.Oxford:OxfordUniversityPress.Keenan,E.andWesterstahl,D.(1997),“Generalizedquantifiersinlinguisticsandlogic”.InJ.vanBenthamandA.terMeulen(eds),HandbookofLogicandLanguage.NewYork:Elsevier.Kölbel,M.(2002),TruthWithoutObjectivity.London:Routledge.Kratzer,A.(1977),“What‘must’and‘can’mustandcanmean”.LinguisticsandPhilosophy,1:337–55.—(1981),“Thenotionalcategoryofmodality”.InP.PortnerandB.Partee(eds),FormalSemantics:TheEssentialReadings.Oxford:Blackwell.Kripke,S.(1959),“Acompletenesstheoreminmodallogic”.JournalofSymbolicLogic,24:1–14.829781472578235_txt_print.indd8208/08/201415:45\nFormalSemantics—(1989),“Speaker’sreferenceandsemanticreference”.InP.French,Jr.etal.(eds),ContemporaryPerspectivesinthePhilosophyofLanguage.Minneapolis:UniversityofMinnesotaPress.Ladusaw,W.(1979),“PolaritySensitivityasInherentScopeRelations”.PhDthesis,Austin,TX:UniversityofTexas.Lewis,D.(1970),“Generalsemantics”.Synthese,22:18–67.—(1986),OnthePluralityofWorlds.Oxford:Blackwell.MacFarlane,J.(2003),“Futurecontingentsandrelativetruth”.ThePhilosophicalQuarterly,53:321–36.—(2005),“Makingsenseofrelativetruth”.ProceedingsoftheAristotelianSociety,105:321–39.May,R.(1977),TheGrammarofQuantification.NewYork:GarlandPublishing.Montague,R.(2002),“ThepropertreatmentofquantificationinordinaryEnglish”.InPortner,P.andPartee,B.(eds),FormalSemantics:TheEssentialReadings.Oxford:Blackwell.Parikh,R.(1987),“Knowledgeandtheproblemoflogicalomniscience”.InZ.RasandM.Zemankova(eds),InternationalSymposiumonMethodologyforIntelligentSystems1987.TheHague:Elsevier.Quine,W.V.(1956),“Quantifiersandpropositionalattitudes”.TheJournalofPhilosophy,53(5),177–87.Russell,B.(1905),“Ondenoting”.Mind,14:479–93.Stalnaker,R.(1979),“Assertion”.InP.Coleed.SyntaxandSemantics9.NewYork:AcademicPress.—(1991),“Theproblemoflogicalomniscience,I”.Synthese,89:425–40.Tarski,A.(1983),“Theconceptoftruthinformalizedlanguages”.InLogic,Semantics,Metamathematics.Indianapolis:Hackett.vonFintel,K.andIatridou,S.(2008),“HowtosayOughtinforeign:Thecompo-sitionofweaknecessitymodals”.InJ.GuéronandJ.Lecarme(eds),TimeandModality.Dordrecht:Springer.Yalcin,S.(2010),“Probabilityoperators”.PhilosophyCompass,5:916–37.839781472578235_txt_print.indd8308/08/201415:45\nTheoriesofMeaningand4TruthConditionsKathrinGlüerPhilosophicalMeaningTheoryInthephilosophicaltheoryoflinguisticmeaning,adistinctioncanbedrawnbetweentheoriesaimingatcorrectsemanticdescriptionofindividuallanguagesandtheoriesthesubjectofwhichislinguisticmeaningassuch.Theoriesofthefirstkindcanconcernformallanguagessuchasthelanguageoffirst-orderpredicatelogic,or(fragmentsof)naturallanguagessuchasEnglish.Theyareusuallyformallyworkedoutand,therefore,thedisciplineofdevelopingthemiscalled“formalsemantics”.Whileformalsemanticsisasubjectbelongingtologicandlinguisticsasmuchastophilosophy,philosophicalmeaningtheoryisamore“foundational”enterprise.Philosophicalmeaningtheoryaimsatansweringthemostbasicquestionsconcerninglinguisticmeaning,questionsaboutitsverynature.1Themostgeneralandbasicsuchquestionisofcoursethefollowing:Whatismeaning?Or,inDonaldDavidson’smuchquotedwords:“Whatisitforwordstomeanwhattheydo?”(Davidson1984,xiii).Davidsonhimselfsuggestedapproachingthismatterbyaskingtwodifferentquestions:Whatformshouldaformalsemanticstake?And:Whatisitthatmakesasemantictheorycorrectforaparticularlanguage,i.e.whatdeterminesmeaning?2Thesecondquestionconcernstheplaceofsemanticfactsinawidermetaphysicalspace:Howdothesefactsrelatetonon-semanticfacts?Cantheybereducedtonon-semanticfacts,dotheymerelysuperveneonnon-semanticfacts,oraretheysomethinglikemetaphysicalprimitives?Inthesecondhalfofthetwentiethcentury,philosophersoflanguagehavebeenespeciallyinterestedintherelationbetweensemanticfactsandfactsthatcanbedescribedinnatural-isticterms,anddifferentversionsofreductiveandnon-reductivenaturalismhavebeendiscussed.Another,thoughrelated,debateconcernsthequestionwhetherthefactsdeterminingmeaning(andthoughtcontent)arefactsinsomesenseinternal,orexternal,tothesubjectsayingorthinkingsomething.Here,weshallfocusonDavidson’sfirstquestion,however:Whatformshouldaformalsemantictheorytake?Tobeatallrelevantforthebasic849781472578235_txt_print.indd8408/08/201415:45\nTheoriesofMeaningandTruthConditionsconcernsofphilosophicalmeaningtheory,aformalsemanticsneedstobeapplicabletonaturallanguage.Notonlywouldwenothaveunderstoodthegeneralnatureofmeaningifourtheoriesconcernedonlyartificiallanguages;theexpressionsofartificiallanguageswouldnothaveanymeaningwithoutthepriormeaningfuluseofnaturallanguage.Linguisticmeaning,thus,ultimatelyarisesinnaturallanguagecommunication;ifwewanttounderstanditsnaturesuchcommunicationiswhatweshouldlookat.Butlinguistsandphilosophersdoingformalsemanticsfornaturallanguagehavedevelopedanumberofdifferentframeworks.DavidsonhimselfsuggestedthatTarski-styletruth-theories(“T-theories”)canbeusedasformalsemantictheoriesfornaturallanguage.OthertheoriesonthemarketincludePossible-worldsandProof-TheoreticSemantics,SituationSemantics,Game-TheoreticSemantics,andDiscourseRepresentationTheory,tonamebutafew.Thisofcoursepromptsthequestion,whichframework,whichformofsemantictheoryisthebestortherightone.Butwhyisthisaquestionforthephilosophicaltheoryofmeaning?Takeaparticularformalsemantictheoryfor(afragmentof)English.ThistheorydescribesEnglishbymeansofacentral,orbasic,semanticconcept.ForDavidson,thisconceptistruth.ADavidsoniansemanticsforEnglishascribestruthconditionstothesentencesofEnglish.IfDavidsonisrightabouttheformasemantictheoryshouldtake,theconceptoftruthcanbeusedtoascribemeaningstolinguisticexpressions.Byansweringtheform-questionwethuslearnsomethingaboutthenatureofmeaning,aboutitsessentialrelationtotruth.Differentsemanticframeworks,however,workwithdifferentbasicsemanticconcepts.Thebasicsemanticconceptofpossible-worldssemantics,forinstance,isthatoftruthatapossibleworld,andthatofproof-theoreticsemanticsisthatofprooforruleofinference.Consequently,ifweknowwhatformthesemantictheoryshouldtake,weknowwhatthebasicsemanticconceptis,theconceptbymeansofwhichwecancharacterize,explainorelucidatethatofmeaning.Thereis,thus,anaturalconnectionbetweenformalsemanticsandphilo-sophicalmeaningtheory.Thisconnectionis,however,hostagetoanumberofconditions.Forone,thebasicsemanticconceptusedbyasemantictheorymustbesufficientlydifferentfromthatofmeaning(andverycloselyrelatedconceptssuchasreference).Otherwise,therewouldatleastbesomedangeroftheoreticalcircularity.3Foranother,naturallanguagemustactuallybesuchthatitcanbeadequatelydescribedbytheformalmethodsoriginallydevelopedfor“artificial”languagessuchasthelanguageoffirst-orderpredicatelogic.Forinstance,naturallanguagemustbesuchthatwhatcanbeexpressedbymeansofithasasufficientlylarge,andsufficientlystable,“core”ofliteral,orsemantic,meaning.Withoutsuchacore,semantictheorywouldhavenoroletoplayinexplainingcommunicationbymeansofnaturallanguage.859781472578235_txt_print.indd8508/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageSomephilosophersandlinguists,suchasradicalcontextualistsandothersinfluencedbycertainreadingsofthelaterWittgenstein,havedisputedtheexistenceofaliteralorsemanticcoreinnaturallanguage.4This,however,isnotaquestiontobedecidedonprinciple.Howmuchofnaturallanguage,andnaturallanguagecommunication,canbeaccountedforbymeansofsystematicsemantictheory,orbymeansofsystematicsemantictheorycomplementedbysystematicpragmatictheory,hastobefoundoutbycarefulandcreativetheoryconstructionandtesting–muchasforanyothercomplexandprimafaciedisor-derlyempiricalphenomenon.Whatweshallconcentrateonhere,however,arequestionsconcerningthefundamentalconcept(s)ofphilosophicalmeaningtheory.Moreprecisely,weshallusethesuggestionthattruthisthebasicsemanticconcept,andsomeofthemostseriousproblemsfortruth-conditionalsemantics,asthefocusofourdiscussion.Truth-ConditionalSemanticsandtheDavidsonianProgrammeAlineofthoughtgoingbackatleasttoFregeandtheearlyWittgensteinconnectssentencemeaningwithtruthconditions.ForWittgenstein,thisconnectionessentiallyinvolvestheideathatthemeaningofasentenceiswhatacompetentspeakerofthelanguageinquestionunderstands,orknows,whensheunderstandsthesentence,oranutteranceofit.Whatsuchaspeakerknows,Wittgenstein’sideais,isunderwhatconditionsthesentenceistrue:“EinenSatzverstehen,heisst,wissenwasderFallist,wennerwahrist”(T4.024).5Inthisway,meaningisessentiallyboundupnotonlywithtruth,butalsowiththeactivitiesandthepsychologyofcompetentspeakers.Linguisticcompetence,theabilitytounderstandthesentencesofanaturallanguage,hasacertainunboundedcharacter:Thereisanenormousvariety,aseeminglylimitlesssupplyofsentenceswithdifferentmeaningsthatnaturallanguagehasonofferforitsspeakers.Andcompetentspeakerspossessanastoundingcapacitytoefficientlyandspeedilyproduceandunderstandthesesentences,eveniftheyhaveneverheardthembefore.Anaccountofmeaningasanobjectofknowledge,theproductofacompetencepossessedbyordinaryspeakers,hassomehowtocometotermswiththelimitlesscharacterofthiscompetence.Sentenceshaveaconstituentstructure;theyarecomposedof“smaller”parts,andtheirmeaningseemstodepend,inasystematicmanner,onthepartstheyarebuiltof,andthewayinwhichthesepartsareputtogether.6Theideathat(PC)themeaningofasentence,orcomplexexpressioningeneral,isdeter-minedby,orafunctionof,themeaningsofitsconstituentpartsandmodeofcomposition869781472578235_txt_print.indd8608/08/201415:45\nTheoriesofMeaningandTruthConditionsiswellknownsinceFrege(cf.Frege1892):Itiscalledthe“principleofcompositionality”.7Ifnaturallanguageisinfactcompositional,accountsoflinguisticcompetencecansystematicallyexploitthisfeature.Compositionalaccountsstartwithsimpleexpressions,specifytheirmeaningsorsemanticvalues,andthenspecifythemeaningsofcomplexexpressionsbymeansofrecursiverulesfor“buildingthemup”onthebasisofthoseofthesimpleones.DavidsonthensuggestedthattheformalapparatususedbyTarskitodefinetruthforcertainformallanguagescouldbeusedtoputthesetwoFregeanideas–compositionalityandtruthconditions–together;heproposedtouseTarski-styletruthdefinitions(T-theories)toascribetruthconditionstonaturallanguagesentencesinacompositionalway.Tarskihadshownhowtodefineapredicate‘true-in-L’foralanguageLofacertainkindbymeansofarecursivedefinitionofthenotionofsatisfactionforL.Thatis,heprovidedaformalmachinerythatallowsforconstructingafinitelyaxioma-tizedtheoryoftruthforalanguageLonthebasisofL’ssyntacticstructure.FromaT-theoryforL,wecanderive,foreverysentencesofL,atheorem(aso-called“T-sentence”)specifyings’struthconditions.Thetheoryworks,Davidsonexplains,bygivingnecessaryandsufficientconditionsforthetruthofeverysentence,andtogivetruthconditionsisawayofgivingthemeaningofasentence.Toknowthesemanticconceptoftruthforalanguageistoknowwhatitisforasentence–anysentence–tobetrue,andthisamounts,inonegoodsensewecangivetothephrase,tounderstandingthelanguage(Davidson1967,24).ThetheoremsofaT-theoryhavethefollowingbasicform:(T)‘s’istrueiffp,where‘s’isreplacedbyasentenceofthelanguagewearespecifyingtruthconditionsfor,the“object-language”,and‘p’byasentenceofthelanguagethetheoryisformulatedin,the“meta-language”.Tarskihadshownthatatruth-definitionforalanguageLis“materiallyadequate”,i.e.definesapredicatetrue-in-Lthatisco-extensionalwiththetruthpredicateasintuitivelyunderstood,if,andonlyif,foreverysentencesofLitimpliesaT-sentencesuchthatwhatreplacespisatranslationofs(andobject-andmeta-languagesatisfycertainformalconditions).Translation,however,isidentityofmeaningacrosslanguages;Tarskithusconnectshisotherwiseuninterpretedtrue-in-Lpredicateswithtruthsimpliciterbymeansoftheconceptofmeaning.879781472578235_txt_print.indd8708/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageTounderstandtruthintermsofmeaningis,accordingtoDavidson,philo-sophicallymisguided.Meaningisamuchmoreobscureconceptthantruth,hemaintains;moreover,“truthisoneoftheclearestandmostbasicconceptswehave”(2005,55).HethereforesuggeststoturnTarskionhisphilosophicalfeet,andtousetheformalmachineryofTarskiantruththeoriestounderstandmeaning(cf.Davidson1984,xiv).8IfwecanconstructcorrectT-theoriesfornaturallanguage,theideais,wecanunderstandthemeaningofsentencesintermsoftheirtruthconditions,andthemeaningofsub-sententiallinguisticexpressionsasthesystematiccontributiontheymaketothetruthconditionsofsentences.Sinceheisconvincedthatthiscanbedone–atleastforfragmentsofnaturallanguagelargeenoughtovindicatetheidea–Davidsonalsoclaimsthatthetheoryofmeaningcandowithoutmeanings:Thereisnoneedtoidentifymeaningswithsomekindofabstractobjectsuchaspropositions.Ifweunder-standmeaningintermsofsystematiccontributionstotruthconditions,theassociationofsuchentitieswithexpressionsissimplyredundant(cf.Davidson1967,20f).Intriguingasthisideais,itisfacedwithaformidablerangeofproblems.Lessseriousareobservationssuchasthatnotallutterancesofnaturallanguageareofwholesentences,orthatevenofthosethatare,manydonotseemtoexpresstruth-evaluablecontents.Theobservationthatquestions,requestsorcommands,forinstance,donotseemtohavetruthconditionsrequiresdrawingadistinctionbetweenthespeechactperformedbymeansofutteringasentenceandtheproposi-tional,ortruth-conditional,contentofthatspeechact.Itseemsnaturaltothinkthatthepropositionalcontentofthequestion“Istherebutterinthefridge?”andtheassertion“Thereisbutterinthefridge”isthesame.Thiscontentcanbespecifiedbyaso-called“that-clause”:thatthereisbutterinthefridge.Ascribingtruthconditionstoitssentencesthuscanbeseenascharacterizingthesemanticcoreofalanguage.Phenomenasuchasironyormetaphorshowthatthesemantictheoryisrestrictedtoliterallanguageuse;itdescribestheliteralcoreofalanguage.9Onewayofconceivingoftheconnectionbetweenthiscoreandtheactualuseofexpressionsincompletespeechactswouldthenbethefollowing:Whatasemanticsdeliversisthecontentofliteralassertion,i.e.foreverysentencesitspecifieswhatwouldbe(literally)assertedwerestobeutteredassertori-cally.10Standardaccountsofindexicals,however,requirefurtherrestrictionoftheseclaims.AccordingtoKaplaniansemanticsforindexicals,forinstance,thepropositionassertedbymeansofanutteranceofasentencelike‘Iamhere’isdeterminedbyitsmeaning(whatKaplancalls“character”)togetherwiththeextra-linguisticcontextoftheutterance.Recentdebatesaroundphenomenasuchas“modulation”–using,forinstance,thesentence‘Thehamsandwichleftwithoutpaying’toclaimthatthepersonwhoorderedthehamsandwichleftwithoutpaying–suggestthatthe(systematic)influenceofcontextonthecon-tentassertedmightbemuchmorepervasivethantraditionallyexpected.11889781472578235_txt_print.indd8808/08/201415:45\nTheoriesofMeaningandTruthConditionsMoreseriousisthequestionwhethernaturallanguageindeedissuchthatitcan,toasufficientlylargedegree,bedescribedbymeansoftheformalapparatusofaT-theory,oranyotherversionoftruth-conditionalsemanticssuchaspossible-worldssemantics.Forasentencestobe“fedinto”themachineryofaT-theory,forinstance,ithastobepossibletoassignsa“logicalform”expressibleinthelanguageoffirst-orderpredicatelogic.Alotcanbehandledthatway,butnoteverything.Problemsarise,forinstance,forcondi-tionalsandforintensionalcontextssuchasthosecreatedbypropositionalattitudeoperatorssuchas‘believesthat’ormodaloperatorssuchas‘itisnecessarythat’.Problematicalsoareattributiveadjectives(‘good’,asin‘goodactress’),indexicals(‘I’,‘this’),massterms(‘snow’,‘water’),tenseoperators,andmanymore(cf.Davidson1967,35f).Inmanyoftheseareas,considerableprogresshasbeenmadesincethedayswhenDavidsonsuggestedthataformalsemanticsforanaturallanguageshouldtaketheformofaT-theory.Someofthese,notablythepossible-worldstreatmentsofalethicmodaloperatorssuchas‘itisnecessarythat’or‘itispossiblethat’,however,requireleaving“pure”truth-theoreticsemanticsbehindandadoptingsomethingstronger,aversionofpossible-worldssemantics.Evenifsufficientlymuchofnaturallanguagecouldbe“tamed”bytruth-conditionalsemantictheory,however,morefundamentalquestionswouldremain.Mostimportantlyinourcontext,wecouldstillaskwhethertruth(ortruthatapossibleworld)reallyisthefundamentalconceptofmeaningtheory.Thishasbeenchallengedfromseveraldirections.Let’slookatsomeofthemostimportantobjectionsinturn.Strawson’sHomericStruggleOneearlyimportantchallengetotruth-conditionalsemantics,issuedbyStrawsoninthelate1960s,tooktheformofageneralchallengetoformalsemantics–eventhoughStrawsonineffectwasconcernedonlywithtruth-conditionalsemantics.Strawsonarguedthattruthcannotbethefundamentalconceptofmeaningtheorybecausetheattempttounderstandmeaningintermsoftruthleadstotheoreticalcircularity.Sincetruthitselfneedstobespelledoutintermsof“communicationintention”,StrawsoncontendedagainstDavidson,thefundamentalconceptsofmeaningtheoryultimatelyarenotthoseofformalsemantics.Inhis1969inaugurallectureinOxford,Strawsonfamouslydescribedthe“conflictbetweenthetheoristsofcommunication-intentionandthetheoristsofformalsemantics”asa“Homericstruggle”(Strawson1969,5).12InourcontextthisstruggleisinterestingpreciselybecauseStrawsonsetsitupasconcerning“thefundamentalconceptinthetheoryofmeaning”(Strawson1969,6).899781472578235_txt_print.indd8908/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageOnwhatweshallcallan“intentionalaccount”ofmeaning,thefundamentalconceptisthatof“anutterer’smeaningsomethingbyanaudience-directedutteranceonaparticularoccasion”(ibid.).Meaning,thatis,istobeunderstoodintermsofthecommunicativeintentionswithwhichspeakersutterlinguisticexpressions.Togetabasicideaofhowthismightwork,letustakeGrice’saccountasourmodel.13Griceworkswithabasicdistinctionbetweennaturalandnon-naturalmeaning,wherenaturalmeaningismeaninginthesenseinwhichsmokemeansfireorreddishspotsmeanmeasles.Linguisticmeaningisaprimeexampleofnon-naturalmeaning,andGricesuggeststhefollowinganalysis(cf.Grice1957):(M)AnuttererUmeansthatpbyanutterancexiffthereisanaudienceANNsuchthat(i)U,bymeansofx,intendstoinduceAtoformanattitudejtowardsp(forinstance,tobelievethatp),and(ii)UintendsAtorecognizeU’sintention(i),and(iii)UintendsthatthisrecognitionbepartofA’sreasonforformingjtowardsp.Theserathercomplex,andinsomesenseself-referential,intentionsinvolvedinnon-naturallymeaningsomethingareoftencalled“M-intentions”.Gricenotonlysuggestedanalysingtheutterer’smeaningintermsofsuchM-intentions,butcontendedthatthefurthernotionsofanexpression’smeaningsomething,eitheronparticularoccasionsofutteranceorinatimelesssense,canbedefinedonthebasisofM-intentions.Inhisratherprogrammatic1969paper,Strawsonpredictedthattheinten-tionaltheoristwouldwinthestruggleagainstthetruth-conditionalsemanticist.Whathewasconcernedaboutwastheverynotionofatruthcondition.Strawson’sfirstquestionwassimply:What,exactly,isatruthcondition?Unlesswesaysomethinginformativeaboutthat,ourwholefoundationalenterprisemightbebuiltonsand.Andthisisabiggerenterprisethanitmightprimafacieseem,fortounderstandthenotionofatruthconditionweneedtounderstandthenotionoftruth.Sofar,thechallengeisfairenough;clearly,thetruth-conditionalsemanticistoughttobeabletotellussomethinginterestingabouttruthandtruthcondi-tions.Beforewelookintopossiblereplies,however,wehavetoaskwhatthischallengemighthavetodowithintentionalaccountsofmeaning.Strawsonboldlysurmisedthatanyanswertothequestionofthenatureoftruthitselfwouldhavetomakeuseofspeechacttheoreticconceptssuchasassertoriccontent.Thenotionoftruth,heclaims,canonlybeunderstoodinapplicationtointentionalhumanactssuchastheutteranceofsentencesinassertoricspeechacts.Consequently,thetruthofanassertion,andtherewith909781472578235_txt_print.indd9008/08/201415:45\nTheoriesofMeaningandTruthConditionsthetruthofthecontentofsuchaspeechact,wouldbeanalyticallypriortothatofthesentenceuttered.Strawsonconcludedthatthenotionsofcommunicativeintentionweremorefundamentalinthetheoryofmeaningthanthatoftruth.However,oneofthemostseriousproblemsoftheGriceananalysisoflinguisticmeaningisthefollowing.SupposeUuttersthesentence‘SwedenisintheNorthofEurope’withtheM-intentiontogetheraudienceAtobelievethatSwedenisintheNorthofEurope.Suppose,moreover,thatUsucceeds.EvenifwegrantthatGrice’sconditionsaresatisfiedineverysuchcaseofsuccessfulcommunication,wedonotgetananswertothefollowingquestion:Whatisitabouttheutteredsentencethatmakesitusableforinducingpreciselythethatbelief?Whatconnectionistherebetween‘SwedenisintheNorthofEurope’andthebeliefthatSwedenisintheNorthofEurope–asopposedtoanyotherbelief,forinstancethebeliefthatitiscoldattheSouthPole?Intuitively,itisthefactthatthesentencemeansthatSwedenisintheNorthofEuropethatmakesitpossibletoutteritwiththeintentiontoinducethatbelief.Intuitively,thatis,meaninghereseemstoexplainthecontentoftheintention,ratherthantheotherwayaround,thusreversingtheexplanatoryrelation.Whatweneed,butdonotgetfromGrice–atleastnotimmediately–isasystematicaccountofwhatitisthatmakesthesentencesofalanguagesuchaptvehiclesforconveyingthoughtswithparticularcontents.Andagain,whatweneedisanaccountabletocopewiththeenormousvarietyofsentencesthatnaturallanguagehasonofferforitsspeakers.Intuitively,theredoesnotseemanylimittothesentencesthusstandingreadyconveyingparticularcontents.WhatemergesfromtheseconsiderationsistheneedforsupplementingtheGriceanaccountwithasystematicaccountofsentencemeaning.Moreprecisely,supplementingitwithanaccountthatcanhandlespeakers’abilitytoquicklyandefficientlyproduceandunderstandanenormousvarietyofnaturallanguagesentences,includingsentencestheyhaveneverheardbefore.Compositionalityarguablyisthebestwayofhandlingthesefeaturesofnaturallanguagecommunication.14Ifso,thedemandisthatforacompositionalaccountofsentencemeaning.ButcansuchanaccountbeincorporatedintotheGriceanframework?TomakegoodontheclaimthatthefundamentalconceptofmeaningtheoryisthatofcommunicativeorM-intention,thecompositionalaccountofsentencemeaningitselfwouldneedtobeformulatedinGriceanterms.Thatis,wewouldneedtobeabletoexplainatleastreferenceandpredicationintermsofM-intentions.15Thereisverygoodreasontodoubtthatthisisfeasible,however.Likeanyothercompositionaltheory,suchanaccountwouldhavetocontainaxiomsspecifyingreferenceforsimplesingularterms,andextensionsforsimplepredicates.ButtheuseoftheverynotionsofcommunicativeorM-intentionsintheseaxiomswillcreateintensionalcontexts.16Consequently,insuchatheory,validtheoremderivationdependsonpriorrelationsofsynonymy:Intensional919781472578235_txt_print.indd9108/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageaxiomslicensesubstitutionsonlyiftheypreservemeaning.Sinceitpresup-posesunderstandingoftheveryconceptitissupposedtoilluminate,itappearsdoubtfulthatsuchatheorycantellusverymuchaboutmeaning.17,18ThereisthusreasontothinkthatStrawsonwaswrong:Itdoesnotlookasifintentionalaccountscouldprovidemeaning-theoreticalfoundations.19Nevertheless,itremainsofcoursetruethatthemeretruth-conditionalformofasemantictheorydoesnot,byitself,tellusanythingabouthowthetruthpredicateistobeinterpreted.Moreover,itisnotclearthattruthultimatelycanbeunderstoodintermsnotinvolvingmeaning.AndthatbringsustothefirstpartoftheStrawsonianchallenge:Thechallengetosaysomethinginterestingabouttruthwithoutbegginganymeaning-theoreticalquestions.Contemporarytheoriesoftruthcanroughlybecategorizedbymeansoftheiranswerstothefollowingtwoquestions:First,istruthasubstantiveproperty?Andsecond,ifyes,isitanepistemicoranon-epistemicproperty?Redundancytheoristsanddeflationistsdenythattruthisasubstantiveproperty.Epistemictheoriesholdthatitisasubstantive,butinprincipleepistemicallyaccessibleproperty.Epistemicismabouttruthisoftenalsocalled“anti-realism”.Realistsabouttruthholdthattruthisasubstantivepropertyindependentofourbeliefsandcognitiveabilities.Epistemicconceptionsoftruthhavebeenthoughtofasmotivatedbymeaning-theoreticalconcerns.MichaelDummettsuggeststoaccountformeaningintermsofverificationinsteadoftruth(seebelow,pp.97ff),andarguesthat,onsuchanaccount,thenotionoftruthmustbeexplained“intermsofourcapacitytorecognizestatementsastrue”(Dummett1976,75).InspiredbythelaterWittgenstein,philosopherssuchasCrispinWrighthavearguedthat,onpainofthekindofPlatonismattackedbyWittgensteinintherule-followingconsiderations(esp.PhilosophicalInvestigations138–242),meaninghastobeunderstoodintermsofanepistemicconceptoftruth(cf.Wright1980).Anti-realismabouttruthstrikesmanyashavingverycounterintuitiveconse-quences,however;Dummett,forinstance,arguesthatsentencesaboutthepastaremadetruenotbypastfactsinthemselvesbutbywhatispresentlyknownorknowable.Heisoneofthefewanti-realistswhoembracethisconsequence;others,suchasWright,argueagainstthisactuallybeingaconsequenceofanti-realism(cf.Wright1986).Dummettalsoarguedthatredundancytheoristsanddisquotationalistscannotunderstandmeaningintermsoftruthconditions(cf.Dummett1959,7).Bothholdthat‘‘p’istrue’meansthesameas‘p’;disquotationalistsclaimthatthereisnothingmoretothetruthpredicatethanitsfunctionasadeviceofdisquotation20or“semanticascent”,adeviceof“replacingtalkabouttheworldwithlogicallyequivalenttalkaboutwords”(Williams1999,547).21Suchadeviceaddsexpressiveresourcestoalanguagesinceitallowsspeakerstoendorsethingstheycannotdirectlyassert.22Nevertheless,“thefunctionof929781472578235_txt_print.indd9208/08/201415:45\nTheoriesofMeaningandTruthConditionstruthtalkiswhollyexpressive,thusneverexplanatory”(ibid.).MostdeflationistsagreewithDummettandtakethistoimplythatmeaningcannotbeexplainedintermsoftruth(cf.forinstanceHorwich1998,71ff).Moreprecisely,problemsariseassoonasthedeflationisttriestoaccountforcertainusesofthetruth-predicate,usesexemplifiedbyT-sentences:metalinguisticascriptionsoftruthtosentencesofanobjectlanguagenotidenticalwiththemetalanguage.Sincethecorrectapplicationofthetruthpredicateheredependsontranslation,thedisquotationalistcannotuseT-sentencessimultaneouslytoexplainwhatitisforobjectlanguagesentencestohavemeaningandtoexplainhowthe(metalinguistic)truthpredicatefunctions(cf.Patterson2005).DavidsonoriginallythoughtthatTarski-theoriesinfactprovidedsomesortofcorrespondencetheoriesoftruth(cf.Davidson1969),butlaterarguedthatthiswasamistake(cf.Davidson2005,38ff,esp.fn4).Hecametothinkthattruthcannotbedefinedatall,noteveninthe“minimalist”waythedeflationistsfavour,andthattheonlywaytosaysomethingrevealingabouttheconceptoftruthisbytracingitsrelationstoconceptsequallyfundamentalandbeyonddefinition(cf.Davidson1996,20f).23Hesuggestedthat,eventhoughoftenbeyondrecognition,truthisessentiallyrelatedtothepropositionalattitudes:Thetruthpredicateisinterpretedonlythroughthe“pattern”truthmakesamongsttheattitudes,includingspeechandaction,andtheircauses.IthasempiricalcontentpreciselybecauseT-theoriescanbeappliedtointentionalcreatures,canbecorrectorincorrectforaspeaker,orgroupofspeakers:“Ifweknewingeneralwhatmakesatheoryoftruthcorrectlyapplytoaspeakerorgroupofspeakers,wecouldplausiblybesaidtounderstandtheconceptoftruth”(Davidson2005,37).Givenouroverallmeaning-theoreticalproject,wecannottakemeaningsforgrantedincharacterizingtruth,however.Wemust,thatis,findawayofrelatingtruthtotheverysamenon-semanticdataaboutspeakers’behaviourinobservablecircumstancesthat,accordingtoDavidson,providethedetermi-nationbaseformeaning,orcontentingeneral:“Ithereforeseetheproblemofconnectingtruthwithobservablebehaviourasinseparablefromtheproblemofassigningcontentstoalltheattitudes”(Davidson1996,37).Ultimately,then,beliefandtrutharepartofasetofbasic,irreducibleandinterdependentconceptscapturingwhatisessentialtointentionalminds.Theirempiricalcontentderivesfromthemetaphysicsofcontentdetermination.AccordingtoDavidson,contentisdeterminedonthebasisofnon-semanticfacts;moreprecisely,factsaboutobservablebehaviourinobservablecircumstances,bymeansofthe“principleofcharity”.Itrequires“assigningtruthconditionstoaliensentencesthatmakenativespeakersrightwhenplausiblypossible”(Davidson1973,137).Asaprincipleofcontentdeterminationingeneral,charityrequiresthebeliefsofintentionalcreaturestobeminimallycoherent,rationallyconnectedwithdesiresandactions,and,inthemostbasiccases,“about”theobjectswhichtypicallycausethem(cf.Davidson1991).24Because939781472578235_txt_print.indd9308/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageofitsnon-semanticdeterminationbase,contentdeterminationbycharitythusallowsustosaysomethingrevealingalsoabouttruth:“byrelatingittoconceptslikebelief,desire,causeandaction”(Davidson1996,21).OnaDavidsonianpicture,Strawsonwasrightinsofarastruthcannotbecharac-terizedincompleteindependencefromhumanintentionality.Thisdoesnotmean,however,thatitcannotbecharacterizedwithoutanymeaning-theoreti-callyquestion-begginguseofthenotionof(assertoric)content.25IntensionalityandIndeterminacyAnotherimportantchallengetotruth-conditionalsemanticsquestionstheveryideathattruthconditionsevercandothejobofmeanings.Meaningsmightdeterminetruthconditions,butnottheotherwayaround;thereare,theargumentgoes,simplynotenoughtruthconditionsaroundtododutyasmeanings.AmaybeslightlydifferentworrycanbeputintermsofT-theories:Allwecanrequireoftheirtheoremsisthattheybetrue.ButwhatreasonistheretoexpecttrueT-sentencestocapturemeanings?AsDavidsonhimselfpointsout,aT-sentencesuchas(G)‘Snowiswhite’istrueiffgrassisgreen,whiletrue,certainlydoesnotspecifythemeaningof‘snowiswhite’inEnglish(cf.Davidson1967,25).Let’scallT-sentencesthatspecifymeanings“inter-pretive”,andT-sentencesthatdonot,“non-interpretive”.T-theoriesimplyingnon-interpretiveT-sentencessuchas(G)canbeexcludedbecauseaT-theoryiscorrectforalanguageLonlyifitimpliesatrueT-sentenceforeverysentenceofL.AT-theoryimplying(G)willmostprobablyhaveotherconsequencessuchasthat‘thatiswhite’istrueiffthedemonstratedobjectisgreen,orthat‘thatissnow’istrueiffthedemonstratedobjectisgrass.Therefore,noreasonablysimpletheoryimplying(G)issuchthatitimpliesatrueT-sentenceforeverysentenceofthelanguage,Davidsonclaims(Davidson1967,26,fn.10).Theproblemismoreseriousthanthat,however:Intuitively,therearenumerousnon-interpretiveT-sentencesthetruthofwhichdependsonnothingbuttheco-extensionalityofnon-synonymouspredicatessuchas(C)‘Pigsarerenate’istrueiffpigsarecordate.Inresponse,Davidsonstressedtheempiricalnatureofsemantictheories.ItisanempiricalquestionwhetheraT-theoryiscorrectforaparticularnaturallanguageornot.Thatmeans,Davidsonargues,thatitstheoremsarelaw-likestatements:Theyformulatenaturallaws,andthereforemustnotonlybetrue,949781472578235_txt_print.indd9408/08/201415:45\nTheoriesofMeaningandTruthConditionsbutalsocounterfactualsupporting.Thus,forinstance,(C)wouldhavetobetrueevenundercounterfactualcircumstanceswhereitisnotthecasethatcreatureswithaliveralsopossesskidneys.This,heargues,goesatleastsomewaytowardsdistinguishingbetweennon-synonymous,butco-extensionalexpressions(Davidson2005,54).Still,itisfarfromclearthatsufficientlyfine-graineddistinctionsinmeaningcanbeachievedthisway;necessarilyco-extensionalbutarguablynon-synonymouspredicatessuchas‘triangular’and‘trilateral’,forinstance,cannotbedistinguishedthisway.26Problematicalsoareallthosenon-interpretivetrueT-sentencesthatresultiftheright-handsideofaninterpretiveT-sentenceisreplacedbysomethingnecessarilyequivalentwithit,asillustratedbythefollowingpair:(S)‘Snowiswhite’istrueiffsnowiswhite.(S’)‘Snowiswhite’istrueiffsnowiswhiteandtwoplustwoequalsfour.ForanysuchpairofT-sentences,bothareimpliedbyexactlythesameT-theories.Here,Davidsoninvokestheideaofacanonicalproof:OnlyT-sentencesderivedbymeansofacanonicalproofareinterpretive,heclaims;i.e.,onlyT-sentencesthatcanbedirectlyderivedfromtherelevantaxiomsalone.27Besidesthoseofaformalnature,thereare,accordingtoDavidson,empiricalconstraintsonacceptableT-theories.Aswesaw,theseareempiricaltheories,andassuchtobeholisticallyconfirmedbytheavailableevidence;thecorrectT-theoryisthatwhichachievestheoverall“bestfit”withthedata(Davidson1973,136).T-theoriescanthusberankedbyhowwelltheyfitthedata,andtheprinciplegoverningthisrankingis,again,theprincipleofcharity:Themostcharitabletheoryisalsothebest.Charitythusplacesmassiveempiricalrestric-tionsonacceptableT-theories.Takentogether,Davidsonclaims,theformalandempiricalconstraintsonacceptableT-theoriesarestrongenoughtoletonlythosepassthatactuallycanbeusedtointerpretthespeakersofL.28Nevertheless,itisunlikelythattheywillnarrowthenumberofacceptableT-theoriesdowntoone.Theremightwellbemorethanonetheoryfittingtheevidencetoanequaldegree.Davidsonconsiderssuchtheoriesasempiricallyequivalent,andheholdsthatanysuchtheorycanbeconsideredasacorrectmeaningtheoryforL.FromhisQuine-inspiredperspectiveonmeaningdetermination,whatcannotbedeterminedaboutmeaningfromtheevidenceavailableinradicalinterpre-tationsimplyisindeterminate.29Thisindeterminacy,Davidsonargues,isratherharmless,however;itisanalogoustothatbetweenmeasuringtemperatureindegreesFahrenheitorCelsius.Nothingsignificantislostbyswitchingbetweenscaleslikethat(cf.Davidson1977,225).Boththeextenttowhichthisanalogysucceedsandtheextenttowhichindeterminacycaneverbeanacceptable959781472578235_txt_print.indd9508/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageconsequence,ratherthanareductioadabsurdum,ofameaningtheoryremainmattersofdispute,however.30MeaningandUnderstandingFrege,WittgensteinandDavidsonallinitiallythinkofmeaningintermsoftruthconditionsbecauseofthelinkbetweenunderstanding,orknowing,themeaningofasentenceanditstruthcondition.Thephilosopheroflanguagewhoprobablyhasmademostofthislink,however,isMichaelDummett.LikeDavidson,hewasconvincedofthecrucialmeaning-theoreticsignificanceoftheformofoursemantictheory,buthealsowasconvincedthattruthcannotbethefundamentalsemanticconcept.Rather,thisrolehastobegiventoanotionsuchasproof,warrantedassertibilityorfalsifiability(cf.Dummett1976)–preciselybecausetheultimategoalofatheoryofmeaningisatheoryofunderstanding:Tograspthemeaningofanexpressionistounderstanditsroleinthelanguage:acompletetheoryofmeaningforalanguageis,therefore,acompletetheoryofhowthelanguagefunctionsasalanguage.[…][A]theoryofmeaningisatheoryofunderstanding;thatis,whatatheoryofmeaninghastogiveanaccountofiswhatitisthatsomeoneknowswhenheknowsthelanguage,thatis,whenheknowsthemeaningsoftheexpressionsandsentencesofthelanguage(Dummett1974,2f).Dummett,thatis,conceivesoflinguisticcompetenceasaformofknowledge,andofthetheoryofmeaningastheobjectofthisknowledge.31Languageuse,heemphasizes,isanessentiallyrationalactivity.Itisrationallymotivatedbyourknowledgeofmeaning.Justlikeknowledgeofanyothersubjectmatter,knowledgeofmeaningjoinsforceswithourwantsanddesirestoprovidereasonsforaction,linguisticorotherwise.Thisessentiallyrationalcharacteroflanguageuse,Dummettclaims,requiresthetheoryofmeaningtobeatheoryofunderstanding.Moreover,onlya“full-blooded”–asopposedto“modest”–theoryofmeaningcanhopetobeatheoryofunderstanding(Dummett1974,4ff).Afull-bloodedtheoryofmeaningmust,“inexplainingwhatonemustknowinordertoknowthemeaningofeachexpressioninthelanguage,simultaneouslyexplainwhatitistohavetheconceptsexpressiblebymeansofthatlanguage”(ibid.,4).Amodesttheory,ontheotherhand,onlygives“theinterpretationofthelanguagetosomeonewhoalreadyhastheconceptsrequired”(ibid.,5).Theforcebehindthedemandforfull-bloodednessderivesfromtheclaimthatthereareanynumberofconceptsthatonlylanguage-userscanhave.Amongstthemostimportantoftheseareconceptssuchasthoseofexpressingathoughtor969781472578235_txt_print.indd9608/08/201415:45\nTheoriesofMeaningandTruthConditionsassertingthatp.Inordertofullyunderstandmeaningandlanguageuse,then,wemust,inourmeaningtheory,notmakeuseofanysuchconcepts.Andthat,inturn,meansthatwemustexplainwhatitistomeaningfullyuselinguisticexpressionswithoutpresupposingtheideasofexpressingthoughtcontentsandpossessingconcepts.AsMcDowellexplains,[t]hereisindeedaconnectionwiththequestionhowconceptsmightbeimparted;butitisnotthatasubjectoughttobeabletoacquiretheconceptsexpressibleinthelanguagebybeingtoldwhatthetheorystates,butthat,ifthedescriptionoflinguisticpracticedoeswhatisrequiredofit,asubjectcanacquirethoseconceptsbyachievingmasteryofthepracticethatthetheorydescribes(McDowell1997,109).Onthebasisofthesedemands,Dummettpresentsthetruth-conditionalseman-ticistwithabasicdilemma:IfaT-theoryisusedtospecifytruthconditions,theaccountofmeaningtherebygiveneitherremainsmodest,orithastogoholistic.Neitheroptionisanygood(cf.Dummett1974,20).Takenindividually,Dummettargues,thetheoremsofaT-theoryforalanguageLdonottellusanymorethanatranslationmanualforLwould;tounderstandLyoueitherneedtoknowthemeta-language,orthelanguagethatListranslatedinto,inadvance.Consequently,aT-theorydoesnotprovidemorethanamodesttheoryofmeaning.Buttryingtoescapethisconsequencebytakingthetheoryasawholetomodellinguisticcompetenceistojumpfromthefryingpanintothefire,accordingtoDummett.“Goingholistic”ineffectamountstogivinguponasystematicaccountoflinguisticcompetencealtogether.“Onsuchanaccount”,Dummettargues,therecanbenoanswertothequestionswhatconstitutesaspeaker’sunder-standingofanyonewordorsentence:onecansayonlythattheknowledgeoftheentiretheoryoftruthissuesinanabilitytospeakthelanguage(ibid.,16).Butthisispreciselywhatweneedtohaveatheoryofunderstanding:Foreachparticularwordorsentence,weneedtobeabletosaywhatconstitutesunderstandingthatverywordorsentence.ThatholismcannotbeananswerherebecomesmanifestintheconsequencesDummetttakessemanticholismtohaveforlanguagelearning.Itbasicallymakeslanguagelearningcompletelymysterious,ifnotimpossible:“[O]naholisticview,itisimpossiblefullytounderstandanysentencewithoutknowingtheentirelanguage”(Dummett1976,44).Atheoryofunderstanding,Dummettargues,thereforecanonlybegiveniflanguageis“molecular”:Ifthereis,foreverysentencesofalanguageL,alimitedfragmentofLunderstandingofwhichsufficesforunderstandings.Onlythenisthereanyhopeofprovidingafull-bloodedtheoryofmeaning.979781472578235_txt_print.indd9708/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageButthiswayoutisnotopentothetruth-conditionalsemanticist.For,eventhoughthemeaningofmanysentencescanbelearnedthroughverbalexplana-tions,thiscannotbethemostbasicformoflanguagelearning.Therefore,eventhoughknowingthemeaningofmanysentencescanbeunderstoodintermsoftheabilitytounderstandtheirverbalexplanations,aspeaker’sunderstandingofthemostprimitivepartofthelanguage[…]cannotbeexplainedthisway:ifthatunderstandingconsistsinaknowledgeofthetruthconditionsofsentences,suchknowledgemustbeimplicitknowledge,andthetheoryofmeaningmustsupplyuswithanaccountofhowthatknowledgeismanifested(ibid.,45,emph.mine).Itisthisrequirement,oftencalledthe“manifestationrequirement”,onwhichthetruth-conditionalaccountultimatelyfalters,accordingtoDummett.For“naturallanguageisfullofsentenceswhicharenoteffectivelydecidable,onesforwhichthereexistsnoeffectiveprocedurefordeterminingwhetherornottheirtruthconditionsarefulfilled”(ibid.,46).Examplesofsuchsentencesaresentencesquantifyingoverinfiniteorunsurveyabledomains,subjunctiveconditionals,orcertainsentencesaboutthefutureorthepast.Forsuch“verifi-cationtranscendent”sentences,Dummettsubmits,“thereisnocontenttoanascriptionofimplicitknowledgeofwhat[theirtruthcondition]is,sincethereisnopracticalabilitybymeansofwhichsuchknowledgecanbemanifested”(ibid.,46).Dummettthereforecomestotheconclusionthatthebasicsemanticnotioncannotbetruth.32Heconsiderstwoalternatives:verificationismorfalsifica-tionism(cf.Dummett1976,62ff).Accordingtoverificationism,understandingconsistsintheabilitytorecognizetheconditionsunderwhichasentenceisverified,andaccordingtofalsificationismitconsistsintheabilitytorecognizetheconditionsunderwhichitisfalsified.Dummettarguesthattheverifica-tionistconceptionofthemeaningofmathematicalsentencesthathasbeendevelopedwithintheintuitionistictraditionprovidesuswithagoodmodelforageneralverificationistsemantics(cf.ibid.,70f).Thebasicsemanticnotionthusbecomesthatofproof.Knowledgeofthemeaningofasentencethenconsistsinbeingabletotellforanygivenobjecta,whetherthepropertyaisaproofofydoes,ordoesnot,applytoit.Foraproof-theoreticsemanticstoapplytonaturallanguageingeneral,weneedanunderstandingofwhat“proofs”ofnon-mathematicalstatementsconsistin.Thebasicideawouldbetouserelated,moregeneralnotionssuchasthatofwarrantedassertibility,andtothinkofmeaningintermsofthecondi-tionsunderwhichinferencesfromgivenpremisesarevalidinabroadersense.Nevertheless,itremainsfarfromclearthatproof-theoreticsemanticscanbeextendedsoastocoversufficientlylargefragmentsofnaturallanguage.A989781472578235_txt_print.indd9808/08/201415:45\nTheoriesofMeaningandTruthConditionsrelated,butnolongerverificationist,ideawouldbetoconceiveofmeaningentirelyintermsofinferentialconnections,andtouseinferentialroleasthebasicsemanticconcept.33Mostimportantly,however,Dummett’schargethattruth-conditionalsemanticsisdoomedtoremainmodestcanbedisputed.McDowell,forinstance,arguesthatmodestyisavirtue,notavice.Accordingtohim,theholisticnatureofmeaningandmindprecludesanythingbutamodesttheoryofmeaning.This,however,doesnotpreventusfromgainingphilosophicalinsightsintounderstandingmeaningfuluseoflanguage;suchunderstanding,hesubmits,needstobeconceivedofasnotrequiringanykindof“interpre-tation”,butasanacquiredabilitytoliterallyhearanother’smeaningsintheirwords(cf.McDowell1987).Moresubstantively,theconsequencesDummettchargessemanticholismwithcanbedisputed.AccordingtoDummett,suchholismmakeslanguageacquisitionimpossible,butthischargeisjustifiedonlyforcertainversionsofsemanticholism.Davidsoniansemanticholism,forinstance,doesnothavethisconsequence:AccordingtoDavidson,meaningisdeterminedbytheprincipleofcharityonthebasisofnon-semanticfactsaboutsentencesheldtrueunderobservablecircumstances.Theprincipleofcharityinducesamany–onedeterminationrelationbetweenthesenon-semanticfactsandmeanings,thuspreventingtheholismfromimplyingthateverynewacquisitionofawordchangesthemeaningofeveryotherexpression(cf.Pagin1997;2006a).Davidsonhimselfandothershavearguedthatmeaningscanbepartiallyacquired(cf.Davidson1965).Theidea,then,thatatruth-conditionalformalsemantictheory,beitoftheT-theoreticorpossible-worldskind,canholisticallymodellinguisticcompetencemightafterallturnouttobeaviableanswertooneofthefoundationalquestionsofphilosophicalmeaningtheory.34Notes1.ThisdistinctionbetweenformalsemanticsandphilosophicalmeaningtheoryresemblesthatdrawnbyRobertStalnakerbetween“descriptivesemantics”and“foundationalsemantics”:“First,therearequestionsofwhatIwillcall‘descriptivesemantics’.Adescriptivesemantictheoryisatheorythatsayswhatthesemanticsforthelanguageis,withoutsayingwhatitisaboutthepracticeofusingthatlanguagethatexplainswhythatsemanticsistherightone.[…]Second,therearequestions,whichIwillcallquestionsof‘foundationalsemantics’,aboutwhatthefactarethatgiveexpressionstheirsemanticvalues,ormoregenerally,aboutwhatmakesitthecasethatthelanguagespokenbyaparticularindividualorcommunityhasaparticulardescriptivesemantics”(Stalnaker1997,535).Butphilosophicalmeaningtheoryisnotrestrictedtothequestionofmeaningdetermination;rather,itcomprisesanykindof999781472578235_txt_print.indd9908/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguagephilosophicalinquiryintothenatureoflinguisticmeaning.Onesuchquestionthathasbeencruciallyimportantduringthesecondhalfofthetwentiethcenturyispreciselyoneconcerningthephilosophicalsignificanceofthechoiceofformalsemantics:Thequestionofwhatformsuchasemanticsistotake.2.Cf.Davidson1973,125;1984,xiii.ForDavidson,meaningisessentiallypublic,thatis,accessibleto,orknowableby,thespeakersofanaturallanguage.Accordingtohim,thismeansthatmeaningisdeterminedbyobservablebehaviourinobservablecircumstances.Suchbehaviouratthesametimeprovidestheevidenceforasemantictheory.ForDavidson,thequestionofmeaningdeterminationthereforecoincideswiththequestionofwhattheevidenceforasemantictheoryis.Itissometimesobjectedthatthisamountsto“verificationism”(see,forinstance,Williamson2004,137),butaslongastheobjectiondoesnottakeonthespecificsoftheDavidsonianaccountofmeaningdetermination,itishardtoseewhatiswrongwiththeclaimthatnaturallanguagemeaningis,inprinciple,knowablebyitsspeakers.3.Thisisnotthesamethingasrequiringthatmeaningmustbeanalysableintermsof,orreducibleto,somethingcompletelydifferent.Onemightwellhold–asforinstanceDavidsondoes–thatnosuchanalysis,orreduction,willbeforth-coming,andyetinsistonsomeminimalconditionofinformativeness.4.PhilosophicalexamplesincludeSearle(1978)andTravis(1989).LessradicalcontextualistssuchasRecanati(2004)andPaginandPelletier(2007)arguethat,eventhoughpragmaticinfluencesonunderstandinglinguisticutterancesusuallystartbeforeatruth-evaluablecontentisoutput,thisneitherpreventssemantictheoriesfrombeingindispensableforexplaininglinguisticcommu-nicationnordoesitpreventanaccountofsuchunderstandingfrombeingsystematic.Semanticminimalists,suchasBorg(2004)andCappelenandLepore(2004),holdthateveryutteranceofa(non-indexical)sentenceexpressesoneandthesamesemanticcontent,the“minimalproposition”(accordingtoCappelenandLepore,manyotherpropositionsmightbeexpressedatthesametime,however).5.TheOgdentranslation(1922;seeWittgenstein1921)has:“Tounderstandapropositionmeanstoknowwhatisthecaseifitistrue.”6.Thatthewayinwhichthepartsofasentenceareputtogether,i.e.itssyntacticmodeofcomposition,playsaroleherecanbeseenfromexamplessuchas‘BobkicksMary’and‘MarykicksBob’.Thesesentencesarecomposedofthesameparts,butdifferinmeaning.Thedifferencedependsonwhichsyntacticrolethepartsplay.7.Formoreoncompositionality,seePaginandWesterståhl2010a,2010b.8.Thisstrategyconsequentlyinvolvesconceivingofthesemanticconceptsused“inside”thetheory,suchastheconceptsofreferenceandsatisfaction,aspurelytheoreticalconcepts.Nopre-theoreticunderstandingoftheseconceptsispresupposed;theyareinterpreted(totheextentthattheyare)bymeansofbeingpartofanempiricaltheorythatgetsitsempiricalcontentexclusivelythroughitstheoremsandtheirconnectionwiththedata,theevidencesupportingit.9.Onetraditionalwayofdrawingthesemantics–pragmaticsdistinctionispreciselythis:Semanticsiswhatevercanbedescribedbymeansofasystematictruth-condi-tionalsemanticsforalanguage.Everythingbeyondthiscorebelongstopragmatics:1009781472578235_txt_print.indd10008/08/201415:45\nTheoriesofMeaningandTruthConditionsSpeechacttheory,thetheoryofindirectdiscourseincludingthetheoryofnon-literallanguageuse,etc.Onthispicture,pragmaticmechanismsarebasicallyGriceaninthesensethattheyoperateoncompletepropositions,i.e.truth-evaluablecontents.Recently,thispicturehascomeunderconsiderablepressure;today,thereseemstobesomeconsensusthatpragmaticprocessessuchassaturationandmodulation(thetermsareRecanati’s(2004))alreadyoperateonsimplerconcepts.10.Thisistrue(fornon-indexicalsentences)onatruth-conditional,butnotonapossible-worldssemantics.Possible-worldspropositionshavetruthvaluesatpossibleworlds,butanassertionconcernsonlytheactualworld.Lepore(1982)arguesthatthisisadefectofpossible-worldssemanticspreciselybecauseitisthejobofasemantictheorytodeliverthecontentsofassertions.However,whataccountsfortheconnectionwiththeactualworldmightbetheforceoftheutterance,notitscontent;cf.Recanati(2007,37).11.Seenote9above.12.Amongstphilosophers,helistsFregeandtheearlyWittgenstein.RegardingFrege,thisismostprobablyhistoricallymisleading;hewasnotprimarilyinter-estedinsemantics,butinthereductionofmathematicstologic.However,Fregediddevelopthebasicformalmethodsthatallowtheconstructionofaformalsemanticsforfirst-orderquantifiedlogic,andformulatedaversionofcomposition-ality(moreprecisely,Fregeformulatedwhatiscalledthe“substitutionversion”forwhathecalled“Bedeutung”(reference)).WhatStrawsonineffectalmostexclu-sivelydiscussesisDavidson’sseminalarticleTruthandMeaning(1967).Somewhatcomically,StrawsonalsolistsChomskyasa“theoristofformalsemantics”(cf.Strawson1969,5).–Ontheoppositeside,thesideofthe“theoristsofcommuni-cation-intention”,StrawsonnamesGrice,AustinandthelaterWittgenstein(ibid.).13.Cf.Grice1957,1968.FordiscussionanddefenceofamodifiedversionoftheGriceanaccount,seeSchiffer1972.14.Cf.Pagin2009.15.Grice’sownattempt,intermsofresultantprocedures,simplyusesthesemanticconceptsofreferenceandsatisfaction;cf.Grice1968.Itcanthusnotformthesoughtmeaning-theoreticalcompletionofintentionalmeaningtheory.Schiffer(1972)andBennett(1976),bothinspiredbyLewis’saccountintermsofconvention(1969,modified1975),provideGriceanalternatives.16.Cf.Taylor1982.TaylorusesSchiffer’s(1972)versionofaGriceanaccountashisexampleaccount.17.Evenanti-reductivistsaboutmeaning(seenote3above)shouldnotbehappywithsoblatant,or“narrow”,atheoreticalcircularity.18.Davidson1967,22f,arguesthatthetheoremsdeliveredbyacompositionalmeaningtheoryforpreciselythisreasoncannotbeoftheform‘Smeansthatp’.Seealsonote28below.19.Itcouldbearguedthatthereissomesortofmutualdependencebetweentruth-conditionalandintentionalaccountsofmeaning:eachisinneedofsupplementation–moreprecisely,inneedofthekindofsupplementationtheothercouldprovide.Foranargumenttothiseffect,seeRumfitt1995.20.“Disquotation”becausethetruthpredicateworkslikethis:predicatedofaquotedsentence,theresulthasthesamemeaningasthe“disquoted”sentence,i.e.thesentencewithoutthequotationmarks.1019781472578235_txt_print.indd10108/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguage21.Cf.Patterson2005,however,foramorecarefulcharacterizationofthenatureoftheequivalence.22.Thus,Imightforinstanceuse‘Fermat’slasttheoremistrue’toassertthatFermat’slasttheoremistrueeventhoughIdon’trememberthetheorem.23.SomedisquotationalistsarguethatDavidsonismistakenintakinghimselftoexplainmeaningintermsoftruth(cf.Williams1999;Kölbel2001):InaDavidsonianframework,onlyconsiderationsofradicalinterpretation,i.e.meaningdeterminationonthebasisoftheultimateevidence,gointothisexpla-nation,theyclaim.Radicalinterpretation,however,doesnot(Williams),orneednot(Kölbel),involve(asubstantivenotionof)truth.Thismightbepartlyaterminologicaldisputeconcerningtheuseoftheterm“explanation”.Whatseemsrightisthatthemethodofradicalinterpretation,asusedforconstructingaT-theory,isneutralontheinterpretationofthetruthpredicate.Itwouldnotberight,however,toconcludefromthisthattheverychoiceofformalsemanticsdoesnothaveanymeaning-theoreticalsignificance.Radicalinterpretation,ifpossibleatall,wouldlookverydifferentifwhattheradicalinterpreterhadtoconstructwasaproof-theoreticsemantics.ThattruthisthebasicsemanticconceptthusremainsanimportantpartoftheDavidsonian“explanation”ofmeaning.24.Formoreontheprincipleofcharityasthegeneralprincipleofcontentdeter-minationinDavidson,anditsepistemicandmodalstatus,cf.Glüer2006,2007;Pagin2006b.25.Thus,Davidsonalwaysinsistedthat“truthisoneoftheclearestandmostbasicconceptswehave”(2005,55)andthat“meaningnotonlyisamoreobscureconceptthanthatoftruth;itclearlyinvolvesit:ifyouknowwhatanutterancemeans,youknowitstruthconditions”(1996,37).Thisgoestogetherwiththeclaimthattruthisoneofasetofinterdependentconceptsendowedwithempiricalcontentbycontentdetermination:Theoverall“theory”beinginter-pretedbyitsrelationtothenon-semanticevidenceusestheconceptsofbeliefandtruth,butnotthoseofmeaningandcontent.Therelevantconceptofbelief,forinstance,couldbecharacterizedasintentional,butnotintensional–itisthatofapropositionalattitudeindividuatedasanattitudeinabstractionfromanyparticularpropositionalcontent(cf.Davidson2005,67).Andthesameholdsfortheotherelementsofthebasic“intentionalset”–belief,desire,speech,andaction.26.Evenifweconclude,however,thatonlysemantictheoriesworkingwithmorefine-grainedintensionswillcomesufficientlyclosetoourintuitivesynonymyjudgments,thisisnotaproblemsuchtheorieswon’thave.Possibleworldssemantics,forinstance,cannotbyitselfcapturedifferencesbetweennecessarilyequivalentpredicates,either.Nordoesitachievethefinenessofgrainintui-tivelyrequiredbyattitudeascriptions.27.AclassicsourcefortheseproblemsisFoster1976.SeealsoDavidson1976,Soames1992.28.Ithasbeenobservedthat,oncewehaveaninterpretiveT-theory,wecangoonestepfurtherandfromeachofitsT-sentencesderiveasentenceexplicitlystatingwhatobjectlanguagesentencesmean(see,forinstance,Kölbel2001).Kölbelarguesthatwecouldevenincorporatethisstepintoourformalsemantictheory1029781472578235_txt_print.indd10208/08/201415:45\nTheoriesofMeaningandTruthConditions(2001,618ff).Thatismistaken.IncorporatinginferencerulesintotheT-theorythatwouldlicensethederivationof“meaningtheorems”wouldbepossibleonlyifalltheconstraintsoninterpretiveT-theorieswerepurelyformal.Davidsonisveryclear,however,thatbothformalandempiricalconstraintsarerequired.Thepossibilityofderiving“meaningtheorems”fromaninterpretiveT-theorydoesnotimplythatusingittoilluminatethenatureofmeaningwouldbequestion-begging,ortheoreticallycircular,either.Whetherornotwederive“meaningtheorems”fromit,alltheexplanatoryworkisdonebytheinnerworkingsoftheT-theorytogetherwiththeformalandempiricalrestrictionsplacedonit.29.Cf.Quine1960,chapter2.30.Forfurtherdiscussion,seeLeporeandLudwig2005,II.15.31.Mostofthisknowledgehastobeconstruedasimplicitknowledge.Dummettrepeatedlystruggleswiththeprecisecharacterizationofsuchknowledge–ontheonehand,itcannotbepropositionalincharacter,butontheotheritcannotbeamerelypracticalabilitysuchastheabilitytorideabicycle,either.Cf.esp.Dummett1978.32.ItissometimesassumedthatDummett’sargumentsonlyestablishthatthebasicsemanticconceptcannotbe“bivalent”truth,ortruthas“realistically”,ornon-epistemically,understood.Thisismistaken;evenifwedefinebeingtrueas,forinstance,havingaproof,meaningcouldnotconsistintruthconditions:For,eventhoughthepropertyofbeingaproofisadecidableone,thepropertyofhavingaproofisnot.Cf.Pagin1998.33.Formoreonthis,seeGreenbergandHarman2006,Peregrin2008.34.IwouldliketothankPeterPaginandtheeditorsofthisvolumeforhelpfulcomments.Thelatteralsoprovidedthebasicoutlineofthischapter.ReferencesBennett,J.(1976),LinguisticBehaviour.Cambridge:CambridgeUniversityPress.Borg,E.(2004),MinimalSemantics.Oxford:OxfordUniversityPress.Cappelen,H.andLepore,E.(2004),InsentitiveSemantics.ADefenseofSemanticMinimalismandSpeechActPluralism.Oxford:Blackwell.Davidson,D.(1965/1984),“Theoriesofmeaningandlearnablelanguages”,inInquiriesintoTruthandInterpretation,Oxford:Clarendon,pp.3–15.—(1967/1984),“Truthandmeaning”,inInquiriesintoTruthandInterpretation,Oxford:Clarendon,pp.17–36.—(1969/1984),“Truetothefacts”,inInquiriesintoTruthandInterpretation,Oxford:Clarendon,pp.37–54.—(1973/1984),“Radicalinterpretation”,inInquiriesintoTruthandInterpretation,Oxford:Clarendon,pp.125–39.—(1976/1984),“ReplytoFoster”,inInquiriesintoTruthandInterpretation,Oxford:Clarendon,pp.171–9.—(1977/1984),“RealitywithoutReference”,inInquiriesintoTruthandInterpretation,Oxford:Clarendon,pp.215–25.—(1984),InquiriesintoTruthandInterpretation.Oxford:Clarendon.1039781472578235_txt_print.indd10308/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguage—(1991/2001),“ThreeVarietiesofKnowledge”,inSubjective,Intersubjective,Objective,Oxford:ClarendonPress,pp.205–20.—(1996/2005),“Thefollyoftryingtodefinetruth”,inTruth,Language,andHistory,Oxford:Clarendon,pp.19–37.—(2005),TruthandPredication.CambridgeMA:TheBelknapPressofHarvardUniversityPress.Dummett,M.(1959/1978),“Truth”,inTruthandotherEnigmas,Cambridge,MA:HarvardUniversityPress,pp.1–24.—(1974/1993),“Whatisatheoryofmeaning(I)”,inTheSeasofLanguage,Oxford:OxfordUniversityPress,pp.1–33.—(1976/1993),“Whatisatheoryofmeaning(II)”,inTheSeasofLanguage,Oxford:OxfordUniversityPress,pp.34–93.—(1978/1993),“WhatdoIknowwhenIknowalanguage?”inTheSeasofLanguage,Oxford:OxfordUniversityPress,pp.94–105.Foster,J.(1976),“Meaningandtruththeory”,inG.EvansandJ.McDowell(eds),TruthandMeaning.Oxford:OxfordUniversityPress.Frege,G.(1892/1952),“ÜberSinnundBedeutung”.EnglishtranslationinP.GeachandM.Black(eds),TranslationsfromthePhilosophicalWritingsofGottlobFrege,Oxford:Blackwell,56–78.Glüer,K.(2006),“ThestatusofcharityI:conceptualtruthoraposteriorinecessity?”InternationalJournalofPhilosophicalStudies14,337–59—(2007),“Criticalnotice:DonaldDavidson’sCollectedEssays”.Dialectica61,275–84.Greenberg,M.andHarman,G.(2006),“ConceptualRoleSemantics”,inE.LeporeandB.Smith(eds),OxfordHandbookofthePhilosophyofLanguage,Oxford:OxfordUniversityPress,pp.295–322.Grice,P.(1957/1987),“Meaning”,inStudiesintheWaysofWords,Cambridge,MA:HarvardUniversityPress,pp.213–23.—(1968/1987),“Utterer’sMeaning,Sentence-Meaning,andWord-Meaning”,inStudiesintheWaysofWords,Cambridge,MA:HarvardUniversityPress,pp.117–37.Horwich,P.(1998),meaning,Oxford:ClarendonPress.Kölbel,M.(2001),“TwoDogmasofDavidsonianSemantics”.JournalofPhilosophy98,613–35.Lepore,E.(1982),“WhatModelTheoreticSemanticsCannotDo”.Synthese54,167–87.Lepore,E.andLudwig,K.(2005),DonaldDavidson.Meaning,Truth,Language,andReality.Oxford:OxfordUniversityPress.Lewis,D.(1969),Convention:APhilosophicalStudy.Cambridge,MA:HarvardUniversityPress.—(1975),“LanguagesandLanguage”,inK.Gundersoned.MinnesotaStudiesinthePhilosophyofScience,VolumeVII,Minneapolis:UniversityofMinnesotaPress,pp.3–35.McDowell,J.(1987),“Apleaformodesty”,inB.Taylored.MichaelDummett.ContributionstoPhilosophy.Dordrecht:Nijhoff,pp.59–80.—(1997),“Anotherpleaformodesty”,inR.G.Hecked.Language,Thought,andLogic.Oxford:OxfordUniversityPress,pp.105–29.1049781472578235_txt_print.indd10408/08/201415:45\nTheoriesofMeaningandTruthConditionsPagin,P.(1997),“Iscompositionalitycompatiblewithholism?”Mind&Language12,11–33.—(1998),“Bivalence:meaningtheoryvs.metaphysics”.Theoria64,37–66.—(2006a),“Meaningholism”,inE.LeporeandB.Smith(eds),HandbookofPhilosophyofLanguage.Oxford:OxfordUniversityPress,pp.213–32.—(2006b),“ThestatusofcharityII.Charity,probabilityandsimplicity”.InternationalJournalofPhilosophicalStudies14,361–83.—(forthcoming),“Communicationandthecomplexityofsemantics”,inW.Hinzen,E.MacheryandM.Werning(eds),OxfordHandbookofCompositionality.Oxford:OxfordUniversityPress.Pagin,P.andPelletier,J.(2007),“Content,contextandcomposition”,inG.PeterandG.Preyer(eds),ContentandContext.EssaysonSemanticsandPragmatics.Oxford:OxfordUniversityPress,pp.25–62.Pagin,P.andWesterståhl,D.(2010a),“CompositionalityI:DefinitionsandVariants”.PhilosophyCompass5,250–64.—(2010b),“CompositionalityII:ArgumentsandProblems”.PhilosophyCompass5,265–82.Patterson,D.(2005),“Deflationismandthetruthconditionaltheoryofmeaning”.PhilosophicalStudies124,271–94.Peregrin,J.(2008),“Aninferentialistapproachtosemantics:timeforanewkindofstructuralism?”PhilosophyCompass3,1208–23.Quine,W.V.O.(1960),WordandObject.Cambridge,MA:MITPress.Recanati,F.(2004),LiteralMeaning.Oxford:OxfordUniversityPress.—(2007),PerspectivalThought.Oxford:OxfordUniversityPress.Rumfitt,I.(1995),“Truthconditionsandcommunication”.Mind104:827–62.Schiffer,S.(1972),Meaning.Oxford:OxfordUniversityPress.Searle,J.(1978),“Literalmeaning”.Erkenntnis13,207–24.Soames,S.(1992),“Truth,meaning,andunderstaning”.PhilosophicalStudies65,17–35.Stalnaker,R.(1997),“Referenceandnecessity”,inB.HaleandC.Wright(eds),ACompaniontothePhilosophyofLanguage.Oxford:Blackwell,pp.534–54.Strawson,P.F.(1970),MeaningandTruth.AnInauguralLecturedeliveredbeforetheUniversityofOxfordon5November1969.Oxford:ClarendonPress.Taylor,B.(1982),“Ontheneedforameaningtheoryinatheoryofmeaning”.MindXCI,183–200.Travis,C.(1989),TheUsesofSense.Oxford:OxfordUniversityPress.Williams,M.(1999),“MeaningandDeflationaryTruth”.JournalofPhilosophy96,545–64.Williamson,T.(2004),“Philosophical‘Intuitions’andScepticismaboutJudgement”.Dialectica58,109–53.Wittgenstein,L.(1921/1961),TractatusLogico-Philosophicus(T).EnglishtranslationC.K.Ogden,London:Routledge&KeganPaul1922;transl.D.F.PearsandB.F.McGuinness,NewYork:HumanitiesPress.—(1953),PhilosophischeUntersuchungen.Englishtranslation:PhilosophicalInvestigations,transl.G.E.M.Anscombe,Oxford:Blackwell.Wright,C.(1980),WittgensteinontheFoundationsofMathematics.London:Duckworth.—(1986),Realism,MeaningandTruth.Oxford:Blackwell.1059781472578235_txt_print.indd10508/08/201415:45\n5ReferenceGenovevaMartíIntroductionWetalkaboutthings.Whenwesay“NewYorkisalargecity”or“CicerowasaGreekgeneral”wesaysomethingaboutatownoraboutaman,somethingthatiseithertrueorfalse.Whatconnectsourwordswiththethingstheyareabout?Whatmakesitpossibleforawordtorepresentanobject,sothatsomething,trueorfalse,canbesaidaboutit?Thisisafundamentalquestionforthetheoryofreference,perhapsthefunda-mentalquestion,soitisnowonderthatphilosophersoflanguagehavedevotedalotofattentiontosingularterms,expressionsthatseemtobedesignedtoallowustotalkaboutthings:propernames–“NewYork”,“Cicero”–definitedescriptions–“thetutorofAlexandertheGreat”,“theBritishPrimeMinisterduringWorldWarII”–andindexicalsordemonstratives–termssuchas“I”or“thatman”.Foralongtimephilosopherstookdefinitedescriptionstobetheparadigmcaseofhowreferenceoccurs,andtheywouldexplainthelinkbetweennamesandtheirreferentsindescriptivistterms.“Theinventorofbifocals”referstoBenFranklinbecauseheistheindividualwhouniquelysatisfiestheattributesexpressedbythedescription–beingthepersonwhoinventedbifocallenses.Fromadescriptivistperspective,apropernamefunctionsroughlyinthesameway:anameissemanticallytiedtoadefinitedescription,andthereferentofthenameistheindividualwhofitsthedescriptiveprofile.Butstartinginthe1970swefindanapproachthatseesfundamentaldiffer-encesbetweenthewayinwhichdescriptionsandpropernamesmanagetorefer.Fromthisperspective,theconnectionbetweenanameanditsreferentisnottobeexplainedintermsofthereferent’ssatisfactionofaprofilesemanti-callyconnectedtothename.Whenitcomestobasicsemanticquestions,suchas“whatdeterminesthereferenceofaname?”and“whatdonamescontributetothetruthconditionsofsentences?”,thisapproachholdsthatnamesareentirelydevoidofdescriptivecontent.Mypurposehereistopresentanddiscussthechangeofperspectivethatledasubstantialnumberofphilosophersoflanguageawayfromthetraditionaldescriptivistapproach,andtoexaminethenewtheorythatemergesfromthatchangeofperspectiveaswellassomeofthechallengesthenewtheoryfaces.1069781472578235_txt_print.indd10608/08/201415:45\nReferenceDescriptivismPropernameslookverydifferentfromdefinitedescriptions.Foronething,theyaresimple,seeminglystructurelessexpressionsthatdonothavepartsthatarethemselvesmeaningfulexpressions.“Theinventorofbifocals”hasameaningthatdependsonthemeaningsof“the”,“inventor”,“of”and“bifocals”.Itisthosemeaningsthatcontributetothemeaningofthewholedescription.Ifthatcomplexmeaningappliestoanobject,thenthat’sthereferentof“theinventorofbifocals”.“Franklin”,ontheotherhand,isnotmadeoutofothermeaningfulexpres-sions,say“frank”and“lin”,soitisprimafaciehardtoseewhyonewouldbetemptedtothinkthatthetwokindsoftermsaresemanticallysimilar.AsBertrandRussellputit:Anameisasimplesymbol(i.e.,asymbolwhichdoesnothaveanypartsthataresymbols),asimplesymbolusedtodesignateacertainparticular[…].“TheauthorofWaverley”,isnotanamebecauseitisacomplexsymbol.Itcontainspartswhicharesymbols.Itcontainsfourwords,andthemeaningsofthosefourwordsarealreadyfixedandtheyhavefixedthemeaningof“TheauthorofWaverley”[…]Inthatrespect,itdiffersfrom“Scott”.1Nevertheless,descriptivismisanattractiveposition.BertrandRussellhimself,inspiteoftheremarksjustquoted,heldexplicitlytheviewthatordinarypropernames,propernamessuchas“Scott”,“Cicero”and“BenFranklin”,areinfactabbreviationsofdefinitedescriptions.2AndGottlobFrege,arguablythefatherofthephilosophyoflanguage,canbeinterpretedassuggestingasimilarview.3AccordingtoFrege,propernamesanddefinitedescriptions,likeallmeaningfulterms,expressasenseandthatsensedeter-minesareferent.FollowersofFregetraditionallyequatedthesenseofapropernametothesenseofsomeassociateddefinitedescription,orsomefamilyofdescriptions.4Therearepowerfulmotivationsfordescriptivism.Forinstance,thinkabouthowwelearnnames,howweincorporatethemtoourvocabulary.Oftentimesweareintroducedtothepersonnamed.Inmanyothercaseswearetoldsomethingaboutthereferent:whenwefirstencounterthename“Aristotle”wearetoldthathewrotetheNichomacheanEthicsorthathedefendedhylemor-phism.Inallthesecases,eitherbyperceptionorbytestimony,weacquiresomeinformationaboutthereferent,somepieceofinformationthatidentifiesthereferentforus.Itisverynaturalthentothinkthatthisinformationisstoredasadescription,anditispreciselybecauseoftheassociationwiththedescriptionthatthenamebecomessignificanttous,thatwefeelcapableofusingthenametosaysomethingaboutthereferent,thepersonthattheinformation1079781472578235_txt_print.indd10708/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageencapsulatedinthedescriptionappliesto.Inotherwords,wedonotjustlearnnamesinavacuum;whenwelearnaname,wealwaysattachdescriptiveinformationtoit.Certainly,whenwethinkaboutthenamesofourfriendsorpeopleinourfamilyitmightseemthatourusesoftheirnamesarealmostlikeactsofpointing.Thereisprobablysomeevolutionaryreasonthatexplainswhy,ifIpointatadressinawindowandutter“beautiful,buttooexpensive!”,myaudiencewillknowwhatIamtalkingabout.Theremaybeasimilarexplanationofwhywehavenames.AsJohnPerryhasclaimed,cavedwellersprobablyinventednamesonlysothattheycouldcalltheirchildrentodinner.5Butwedonotonlyrefertothepeoplearoundus.Wealsocanrefertopeoplelonggone,orverydistantfromus,peoplewe’veneverhadanycontactwith.Howdowedothat?Descriptivismprovidesanaturalanswer:thedescriptionweassociatewiththenamedoestheworkforus.Assoonasthewords“theinventorofbifocals”leaveourlipsthey,allbythemselvesandbecauseofwhattheymean,findtheirreferent,sotospeak.Ifnamesaresemanticallyassociatedwithdescriptionstheyalso,allbythemselvesandbecauseofwhattheymean,findtheirreferent.WetalkaboutBenFranklinwhenweuse“BenFranklin”becausetheassociateddescription,“theinventorofbifocals”orsomeotherdescription,appliestohim.Inotherwords,werefertothingswedonotseemtohaveanyconnectionto;theviewthatanameoperatesviaanassociateddescriptiongivesusanexplanationofhowwemanagetodothat;itgivesusanaccountofthemechanismofreference.Descriptivismalsoallowsustoexplainthecognitivedifferencesbetweendifferentnamesofoneandthesamething.Inancienttimespeoplesawabrightheavenlybodyintheeveningskyandcalledit“Hesperus”.Theyalsosawabrightheavenlybody,thelastonetofadeinthemorninglight,andtheycalledit“Phosphorus”,thinkingtheywerenamingtwodifferentthings,wheninfactitwasjustonething,theplanetVenus(astheBabyloniansdiscovered).Now,thesentences“Phosphorusisbright”and“Hesperusisbright”arebothtrueandforexactlythesamereason:theballofrockthatisVenusisindeedbright.Butcompetentspeakerscouldacceptonesentenceastrueandrejectorbeindoubtabouttheother;or,afterhavingacceptedonesentenceastrue,theycouldfindtheothersentenceinformative.Thatisadifferenceincognitivesignificancebetweenthetwosentencesthatcannotbeexplainedjustbyappealtothereferencesofthetermsinvolved,forthosearethesame.Butitcanbenaturallyaccountedforifnamesareassociatedwithdefinitedescriptions:theterms“Hesperus”and“Phosphorus”,tiedtodifferentdescriptions,presentthereferentindifferentways.Accepting“Phosphorusisbright”istantamounttoacceptingthatthebeautifulheavenlybodyseeninthemorningskyisbright.LearningthenthatHesperusisbright,i.e.,thattheheavenlybodyseenintheeveningskyisbright,isobviouslyanewpieceofinformation.1089781472578235_txt_print.indd10808/08/201415:45\nReferenceAlltheseconsiderationsprovidestrongreasonstofavourtheviewthatnamesaresemanticallyguidedbydefinitedescriptions,andtoendorsetheclaimthatthebehaviourofdescriptionsprovidesauniformexplanationofthereferentiallinkthatconnectswordstothings.Infact,descriptivismwasforyearsthedominantviewinthephilosophyoflanguage.Andthencame1970.RigidityTheyear1970wasabadyearfordescriptivism.BothSaulKripke,intheseminallecturesNamingandNecessitydeliveredinJanuaryofthatyear,andKeithDonnellan,inhispaper“ProperNamesandIdentifyingDescriptions”,cameupindependentlywithpowerfulargumentsagainstthedescriptiviststance.6Semantictheoryissupposedtoprovide,amongotherthings,anaccountofthecontributionthatexpressionsmaketothetruthconditionsofthesentencesinwhichtheyoccur.InNamingandNecessity,Kripkearguesthatnamesanddefinitedescriptionsdonotoperatesemanticallythesameway,fortheircontributiontotruthconditionsisdifferent,somethingthatcanbeobservedbyreflectingonthemodalprofile(i.e.,thevariationoftruthvalueacrossdifferentcounterfactualcircumstances)ofsentencescontainingnamesversussentencescontainingdefinitedescriptions.SupposethatBenFranklinhadneverdevelopedaninterestinopticalinstruments,devotingallhisenergiestopoliticsafterdraftingtheDeclarationofIndependence,andthatsomesmall-townoptician,entirelyuninterestedinpolitics,hadcomeupwiththeideaofbifocalsinthelateeighteenthcentury.Iftheworldhadbeenthatway,“BenFranklinsignedtheDeclarationofIndependence”wouldstillbetrue,butthesentence“theinventorofbifocalssignedtheDeclarationofIndependence”wouldbefalse.So,itissimplynotthecasethat“BenFranklin”and“theinventorofbifocals”–oranyotherofthedescriptionsthatencapsulatetheinformationwehaveaboutBenFranklin–operatesemanticallythesameway,fortheircontributionstothetruthconditionsofsentencesaredifferent.Thetruthorfalsityof“BenFranklinsignedtheDeclarationofIndependence”dependsonwhetherthatman,ourbelovedandrespectedBenFranklin,puthissignatureonthedocument,anditdoesnotdependinanywayonwhetherhewasanaccomplishedinventor.Thetruthorfalsityof“theinventorofbifocalssignedtheDeclarationofIndependence”dependsonwhetherornotsomepersondidtwothings:inventbifocalsandsigntheDeclaration.That’swhythetwosentencescandifferintruthvaluewhenweconsiderthemagainstthebackgroundofacounterfactualsituation(orpossibleworld)thatisdifferentfromthewaytheworldhappenstobe.Propernames,simply,donotbehavelikedefinitedescriptionswhenitcomestodeterminingwhattheworldhasto1099781472578235_txt_print.indd10908/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguagebelikeinorderforagivensentencetobetrue.Propernames,Kripkeargued,arerigiddesignators:thereferentofanameistheindividualrelevantintheevaluationofsentencesinallpossibleworlds–or,asKripkeputit,propernamesdesignatethesameindividualwithrespecttoallpossibleworlds.Definitedescriptionssuchas“theinventorofbifocals”,“theRomanoratorthatdenouncedCatiline’sconspiracy”,“thetutorofAlexandertheGreat”arenot:theydesignatedifferentindividualswithrespecttodifferentwaystheworldmightbe,dependingonwhichindividualsatisfiesthedescriptiveprofileexpressedbythedefinitedescription.DescriptivistResponses:WidescopismandRigidificationAccordingtosomedescriptivists,Kripke’sargumentsdonotthreatendescrip-tivism.Mostfamously,MichaelDummettarguedsoonafter1970thatallKripkesucceededinshowingisthatthedescriptionsassociatedwithpropernamesalwaystakewidescopewhencombinedwithmodaloperatorsinsentencessuchas“itispossiblythecasethatBenFranklindidnotinventbifocals”.7Considerthedescription“theinventorofbifocals”inthesentence“itispossiblethattheinventorofbifocalsdidnotinventbifocals”Now,therearetwowaysofreadingthatsentence.Ononeofthem,thedescriptiontakesnarrowerscopeandthemodaloperatortakeswiderscope;onthatreadingthesentencesaysthatitispossible,orinotherwords,thattheworldcouldbesuchthatthepersonwhoinventedbifocalsdidnotinventbifocals.Nowobviously,thisisfalse;theworldcannotbesuchthatsomeonewhodoesFdoesnotdoF.Thereisanotherreadinginwhichthedescriptiontakeswiderscope;onthatreadingthesentencesaysthatthepersonwho(infact)inventedbifocalscouldhavenotdoneso,i.e.,thattheworldcouldbesuchthatthatman(theonewhohasinfactinventedbifocals)didnotdoso.Thereisnothingpuzzlingaboutsuchanassertion;infact,itisverylikelytrue.Now,accordingtoDummettthesentence“itispossiblethatBenFranklindidnotinventbifocals”mustbereadalwaysgivingthedescriptionsynonymousto“BenFranklin”widerscopeoverthemodaloperator.Thusnames,afterall,aresynonymouswithdefinitedescriptions;it’sjustthattheyalwaystakewiderscopeinthepresenceofmodaloperators.SaulKripkerespondedtoDummett’sargumentinthePrefacetothewrittenversionofNamingandNecessitythatappearedin1982.AsKripkepointedout,hisoriginalargumentswerenotbasedontheevaluationofsentencescontaininganymodaloperators.ThepointinKripke’sargument,andinthestoryaboutBenFranklindiscussedabove,hasnothingtodowithsentencesoftheform“itispossiblythecasethatS”or“itisnecessarilythecasethatS”.Theargumentreliesontheevaluationofsimplesentencessuchas“BenFranklininvented1109781472578235_txt_print.indd11008/08/201415:45\nReferencebifocals”or“BenFranklinsignedtheDeclarationofIndependence”,andithingesonwhatmakesthosesentencestrueorfalseunderdifferentcircum-stances.IfKripkeisright,thetruthconditionsof“BenFranklinsignedtheDeclarationofIndependence”dependontheman,BenFranklin,havingornothavingsignedtheDeclarationofIndependence,whereasthetruthconditionsof“theinventorofbifocalssignedtheDeclarationofIndependence”dependontherebeingapersonwhoinventedbifocalsandonhisorhersigningornotsigningtheDeclarationofIndependence.Clearlythetruthconditionsofthosesimplesentencesaredifferent,andtheycangodifferentways.Thequestionofscopedoesnotevencomeup.8AdifferentdescriptivistreactiontoKripke’sargumentsreliesontheobser-vationthat,althoughdefinitedescriptionsaretypicallynon-rigid,somedefinitedescriptionsarerigiddesignators.Forinstance“thesuccessorofeight”desig-natesthenumberninegiventhewaytheworldisandineveryotherpossiblecircumstance;althoughwecouldcallthenumberninebyanothername,itwouldstillbeamathematicaltruththatthenumbernine(bywhatevername)isthesuccessorofeight.So,althoughKripke’snotionofrigiditydrawsalinebetweennamesandmostdefinitedescriptions,itdoesnotjustbyitselfdrawalineseparatingthesemanticsofnamesfromthesemanticsofdefinitedescriptions.Somedescriptivists,sensitivetotheconsiderationsputforwardbyKripke,arguedthatnamesaresemanticallyconnectedtorigiddefinitedescriptions,forinstance,thatthedescriptiontiedto“BenFranklin”israthersomethinglike“theactualinventorofbifocals”,adescriptionthat,relativetoanycounterfactualsituation,willrefertowhoeveristheinventorofbifocalsintheactualworld.Inthatwayweobtainthesamedistributionoftruthvaluesrelativetoallcircumstancesofevaluationfor“BenFranklinsignedtheDeclarationofIndependence”and“theactualinventorofbifocalssignedtheDeclarationofIndependence”,defusingKripke’scriticism.9Thesuccessandplausibilityofsuchamovehasbeencontested.10But,arguably,theappealtorigidifieddescriptionsmissesthedeeperlessonofKripke’sargument,alessonthatmaynotbeobviousifwejustfocusontheconclusionthatnamesarerigiddesignators.For,evenifwegetthesamedistributionoftruthvaluesacrosspossibleworldswithanameandwitharigidifieddefinitedescription,ifKripkeisright,thecontributiontothedeterminationofthosetruthvalues(i.e.,thecontributiontotruthconditions)madeby“BenFranklin”andby“theactualinventorofbifocals”isdifferent.Thenamecontributesjustthereferent,thedescriptioncontributesacondition:thattherebeauniqueobjectthathappenedtoactuallyinventbifocals.1119781472578235_txt_print.indd11108/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageDirectReference:TheNewTheoryofReferenceDeclaringnamestoberigidfailstoclearlyseparatenamesfromalldescrip-tions.Butfocusingonhoweachcontributestothetruthconditionsofsentencesdoes.Goingbeyondtheclaimofrigidity,semanticistsmovedbyKripke’sargumentshavestressedthatnamesaredirectlyreferential,meaningthattheircontributiontothedeterminationoftruthconditionsofsentences–or,asitisoftenput,theircontributiontothepropositionsexpressedbysentences–isjustthereferent.KripkeandDonnellanofferedother,andperhapsstronger,argumentsagainstdescriptivism.Oftencompetentspeakersarenotinpossessionofsufficientinformationtouniquelyindividuatethereferentofaname.ManypeopleonlyknowaboutCicerothathewasanorator.ButsowereDemostenesandCato.Theusethatthosespeakersmakeof“Cicero”isnotguidedbyadescriptionthatuniquelyfitsCiceroandcanbethethreadthatconnectsthenametoitsreferent;yet,whenthosespeakersuse“Cicero”,theydorefertoCicero.11Moreover,speakersoftenattachthewrongdescriptiveinformationtoaname.KripkereportsthatmanypeoplethinkthatEinsteinwastheinventoroftheatomicbomb.Yet,whentheyuse“Einstein”,theyrefertoEinstein,nottosomephysicistmemberoftheManhattanProject.12Kripke’sandDonnellan’sconclusionisthatcompetentspeakersreferusingnameseveniftheyarenotinpossessionofuniquelyidentifyinginformation,adefinitedescription.Andtheymanagetorefertowhatintuitivelyistherightreferent,eventhoughtheyattachanerroneousdescriptiveprofiletoaname.Whatanamereferstoisnotdeterminedbyanassociateddefinitedescription.So,namesanddescriptionsnotonlymakedifferentcontributionstotruthconditions,somethingthatwouldentailthatanameisnotsynonymoustoadefinitedescription.Theconclusionoftheignoranceanderrorargumentsisthat,whenaspeakerusesaname,thereferenceofheruseofthenameisnotfixedbyadescription(neitheranon-rigidnorarigidifiedone).NeitherKripkenorDonnellandenythatspeakersattachdescriptiveinfor-mationtonames.Aswehavepointedout,weveryoftenlearnnamesbybeinggiveninformationaboutthereferent.Anditwouldbehardtodenythatthedescriptiveinformationwestore–evenifinsufficientorincorrect,orboth–playsanimportantroleinourcognitivelivesandintheexplanationofthecognitivesignificanceofsentencescontainingnames.ThepointthatKripkeandDonnellanpressisthatsuchdescriptiveinformationdoesnotdeterminethereferentofthename.Thetheoryofreferencethatemergesfromtheseconsiderations,directreferencetheory,isfundamentallyanti-descriptivist.Unlikedescriptions,propernamesarenotandarenotconnectedwithprofilesthatdotheworkofsearchingandidentifyingthereferent.Descriptions1129781472578235_txt_print.indd11208/08/201415:45\nReferencebehaveassearchmechanisms;theyare“prescriptionsforfindinganobject”,asRuthBarcanMarcusputit.Names,ontheotherhand,aremeretags:“Tocountasapropernameanexpressionmustreferwithoutbeingtiedtoanyparticularcharacterizationoftheobject.”13ThisapproachfindsinspirationinsomeremarksonnamesbyJohnStuartMill,andinparticularinhisexampleinvolving“Dartmouth”.14DartmouthisatthemouthoftheriverDart,andthisfactplayedaveryimportantroleinthedecisiontocallthetown“Dartmouth”.Buthadtheriverchangedcourse,itwouldnotbeincorrecttocontinuetoapply“Dartmouth”toDartmouth.Infact,werewetodiscoverthatDartmouthneverwasneartheriver,thenamewouldnotfailtonamethetown.Bycontrast,ifBenFranklindidn’tinventbifocallenses,“theinventorofbifocals”doesnotapplytohimandneverdid.Whatthiswayofthinkingaboutreferencestressesisthatthepropertiesthereferenthasorissupposedtohave,theprofilesthatitsatisfiesorappearstosatisfy,mayplayanimportantroleinthedecisiontouseacertainexpressiontonameit;thosepropertiesandprofilesareexploitedasmuchasotherbackgroundandcontextualfactorsintheprocessofintroducinganameandinusingitsubsequently,i.e.,inestablishingandinmaintainingreference.Buttheyarenotwhattheexpressionmeans,theyarenotanattachedsemanticsearchmechanismthatconnectstermandobject.15Butifthereferentofanameisnotdeterminedbyassociateddescriptiveinformation,thenhowdonamesconnecttotheirbearers?Descriptivismhadananswerforthisfundamentalquestion;itgaveusanaccountofthemechanismofreference.Howdoesanamerefer,accordingtodirectreferencetheory?TheCausal-HistoricalChainofCommunicationAswehavementioned,weusepropernamestorefertoverydistantobjects,distantinspaceandintime.IcannotpointatAristotleforthesimplereasonthathedoesnotexistanymore,sothemodelofauseofanameasanactofpointingdoesnotapplyinthiscase.Accordingtotheproponentsofdirectreferencetheory,thefactthatweareembeddedinachainofcommunicationthatconnectsouruseofagivennametoprevioususes,stretchingallthewaybacktothemomentofintroductionofthename,explainshowwemanagetorefertosuchdistantobjects.ImayhaveinsufficientorevenwronginformationaboutCicero,yetwhenIusethename“Cicero”myconnectiontootherspeakersinmycommunityfromwhichIhaveborrowedthename,andtheirconnectiontopreviousspeakersfromwhichtheyhaveborrowedthename,andsoforth,leadsallthewaybacktothereferentof“Cicero”,thepersonIamtalkingaboutwhenIusehisname,independentlyofmycapacitytosinglehimout,descriptivelyorotherwise.HereishowKripkepresentsthepicture:1139781472578235_txt_print.indd11308/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageSomeone,let’ssay,ababy,isborn;hisparentscallhimbyacertainname.Theytalkabouthimtotheirfriends.Otherpeoplemeethim.Throughvarioussortsoftalkthenameisspreadfromlinktolinkasifbyachain(NamingandNecessity,p.91).So,referenceisinitiallyfixed,ofteninsomekindofdubbingceremonyinwhichthereferentispresentandperceivedsothatthenamecanbebestoweduponit.Afterwardsthosepresentintheoriginalintroductionusethenametorefertotheobjectinquestionwhentheyaretalkingtoothersnotpresentinthedubbingceremony.Inthatwaythenamegetspassedfromspeakertospeakerandfromgenerationtogeneration,perhapssufferingchangesinitsspellinganditspronunciation,especiallywhenitispassedfromspeakersofonelanguagetospeakersofanotherlanguage.Eachnewlinkinthechainusesthenamewiththeintentionofreferringtowhateverobjectthepreviouslinkreferredtowhensheusedthename.16Eachnewlinkinthechainmayattachsome,thesameordifferent,descriptiveinformationtothename.Infactspeakersattachnotonlydescriptiveinformationtonames;theyalsoassociateimages,memories,smellsandallsortsofconnotations.Alltheseassociationsundeniablyplayaveryimportantroleincognitionandtheymayexplainthepsychologyofcommunicationandunderstanding.Whattheproponentofthenewtheoryofreferencedeniesisthattheydeterminewhatauserofanameisreferringtowhensheusesaname.Inusinganame,withwhateverassociations,aspeakerreferstotheobjectthatisattheendofthechainofcommunicationleadingtoherpresentuse.Whatexplainsreferenceisnotsomethingthatthespeakerhasinherhead(descriptiveorotherwise);itisthenon-mental,non-internal,veryworldlyfactthatthespeakerispartofanetwork.Itisimportanttostresswhatchainsofcommunicationareandwhattheyarenot.Achainofcommunicationmakesanexpressionpreservewhateverlinguisticfunctionithasfromonegenerationtothenext.Chainsaremecha-nismsofpropagationoflanguageandpreservationofmeaning,andassuchthechainpictureappliesnotspecificallytonamesbuttoanyexpressioninalanguage.Thereisachainofcommunicationthatconnectsmyuseof“Aristotle”toprevioususes.Thereisequallyachainofcommunicationthatconnectsmyuseof“table”toprevioususes,andmyuseof“not”.Changesdooccur:itmayhappenthatagroupofspeakerssystematicallymakeamistake;themistakemaycatchon,andbepassedtootherspeakers,untilintheendwhatwasamistakebecomesregularpracticeasifanewdubbing(oranewtermintroduction)hadoccurred.ThereisastoryaccordingtowhichtheinhabitantsoftheEastCoastofAfricacalledaportionofthemainland“Madagascar”.Portugueseexplorersgotthenamefromthenativesandmistakenlythoughtthatitwasthenameoftheisland.Theexplorerssurelyintendedtoconformtothenatives’practice,butbytheirrepeateduse1149781472578235_txt_print.indd11408/08/201415:45\nReferenceof“Madagascar”intheirdealingswiththeobviousintentiontosaysomethingabouttheisland(ausethatmusthaveleftthenativesinthemainlandflummoxediftheyeverinteractedwiththeexplorersatall)itwastheislandthatbecameknowntotheworldas“Madagascar”asifanewprotracteddubbinghadtakenplace.Nowadays,whenweuse“Madagascar”wearecertainlyreferringtotheisland.17Anotherpoint,whichmaybelessapparent,isthatthecausal-historicalexplanationisnotperseanalternativetodescriptivism,inthesenseofbeingatheorythatcontradictsdescriptivism.Thecausal-historicalapproachholdsthatachainofcommunicationpassesfromlinktolinkwhateveraccountsforthesemanticbehaviourofanexpression.Thedescriptivistcouldtakeonboardchainsofcommunication:whatispassedfromalinkalongsideaname,hethinks,ispreciselyadescriptionthatdeterminesthereferenceeachtimeaspeakerusesthename.Thedisagreementbetweenthedescriptivistandthenewtheoristisnotabouttheexistenceofchains.Whatcontradictsdescriptivismistheargumentthatreferenceispossiblewithoutthemediationofadescription,i.e.,thataspeakercanmanagetoreferwithouthavingtoacquirefromthepreviouslink,orinanywaymaster,adescriptionthatuniquelyidentifiesthereferent.Differentspeakersmayattachdifferent,rightorwrong,sufficientorinsuf-ficientdescriptiveinformationtoapropername(aswellasimagesandotherconnotations).Noneofthat,theanti-descriptivistthinks,determineswhatsheisreferringtowithauseofaname.Therelianceonthenetworkbyitselfprovideswhatisrequiredtoacquireanameanduseittorefertoitsbearer.18Somedescriptivistshavearguedthatthechainofcommunicationitselfprovidesadescription,implicitlyorexplicitlyknownbythespeaker,thatdeterminesthereferentofaname.Onthiskindofview,whenaspeakeruses“BenFranklin”,thereferenceofthenameisdeterminedbyadescriptionsuchas“theindividualattheendofthechainthatleadstothisuseof‘BenFranklin’”,or“thepersonthatthemembersofmycommunityfromwhichIinheritmyuseof‘BenFranklin’referto”,orsomethingsimilar.19Sincechainsofcommunicationarenotaname-onlyphenomenon,themechanismappearstobequitegeneral:whenaspeakeruses“table”thereisadescriptionofthesort“thekindofthingthatisattheendofthechainthatleadstomyuseof‘table’”,orsomesuchdescription.Itisnotentirelyclearthatbeinginpossessionofsuchadescriptiondeterminesreference(orapplication)ratherthanjustbeingtheexpressionofaspeaker’sawarenessthatshereliesonsharedmeaningsandconformstohercommunity’spractice,fornamesandforanyotherword.1159781472578235_txt_print.indd11508/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageOtherExpressionsDescriptivismisnotjustanapproachtothesemanticsofpropernames.Othersingularterms,suchasindexicals(“I”,“here”,“now”,“today”…)anddemon-stratives(“this”,“that”),werealsotakentofitthesemanticmouldofdefinitedescriptions.20Onadescriptivistapproach,“I”meanssomethinglike“thepersonwhoisspeaking”,“here”meanssomethinglike“theplacewherethespeakerislocated”,and“this”means“theobjectIampointingatorsomehowdemonstrating”.Indexicalsanddemonstrativesarecontext-sensitiveexpressions.Mypresentutteranceof“today”referstoOctober10thbutmyutteranceof“today”lastweekreferredtoOctober3rd;yourefertoyourselfwhenyouutter“I”andIrefertomyselfwhenIdothesame.Bysaying“Iamhungry”youandIaretalkingaboutdifferentpeople;moreover,youmaybesayingsomethingtruewhileI,havingjusthadbreakfast,amsayingsomethingfalse.Ifaspeakersayssomethingtruewhereasanotherspeakersayssomethingfalse,theyclearlyarenotsayingthesamething,eveniftheyareusingthesamewords.IfIsay“Iamhungry”andyouwanttoreportwhatIsaid,youwillsay“sheishungry”(probablypointingatme,ormakingsurethatIamsomehowsalientsothatitisclearyouaretalkingaboutme).Inthiscasetwospeakerssaythesamething,eventhoughtheyareusingdifferentwords–infact,theyhavetousedifferentwords:ifyouused“Iamhungry”youwouldnotbereportingwhatIsaid.Followinganargumentreminiscentofthetruth-conditionalconsiderationsaboutnamesanddescriptions,DavidKaplanarguedin“Demonstratives”21againstthedescriptivistapproachtonamesandindexicals.Supposethat“I”means“thepersonwhoisspeaking”,andthatIsay“Iamaphilosopher”.WeneedtoaskourselvesabouttheconditionsunderwhichwhatIsaidistrue.Iamaphilosopher,sowhatIsaidistrue.Thewaythingsare,itistruethatIamaphilosopheranditisalsotruethatthepersonspeakingisaphilosopher.ButnowsupposethatinsteadofbeingaphilosopherIhadbecomeachef.Iftheworldwerelikethat,itseemsquiteclearthatwhatIsaidwouldbefalse.Now,imaginethatinthatsituationIamnotspeakingandsomeoneelse,whoisaphilosopher,isutteringoutloud“Iamaphilosopher”.Inthiscaseitistruethatthepersonspeakingisaphilosopher,evenifitisfalsethatIamaphilosopher.Inotherwords:whenIutter“Iamaphilosopher”thetruthandfalsityofwhatIsaidisdeterminedbywhetherornotIamaphilosopher,notbywhetherornotIamspeaking;anutteranceof“Iamaphilosopher”andanutteranceof“thepersonspeakingisaphilosopher”havedifferenttruthconditions.But,asinthecaseofnames,theimportantpointisnotthatthetwosentences’truthvalueswithrespecttodifferentcircumstancesdonotcoincide.Theimportantpointisthatwhatdeterminestruthvalueacrossdifferentcircumstances(i.e.,1169781472578235_txt_print.indd11608/08/201415:45\nReferencewhatdeterminestruthconditions)differs.Whentheindexicalisused,thereferentdeterminestruthconditions.Indexicalsanddemonstratives,likenames,aredirectlyreferential.22However,unlikenames,indexicalsanddemonstrativesdohavesomedescriptivecontent.Forinstance,“I”referstothepersonwhoisspeakinginthecontextofutterance.Thatdescriptivecontentclearlyplaysaroleindeterminingthereferentofeachutteranceof“I”:thereisnowayIcanreferto“Hume”whenIuse“I”,simplybecauseIamtheobjectthatfitsthedescriptiveprofileexpressedby“thepersonwhoisspeakinginthecontextoftheutterance”.Moreover,masteryoftheuseofindexicalsanddemonstrativesrequiresmasteryofrulesofusethatarespecific.Therulethatgovernstheuseof“I”(somethinglike“inallcontexts,anutteranceof‘I’referstothespeakerinthatcontext”)isdifferentfromtheruleofusethatgovernstheuseof“today”(somethingsuchas“ineverycontextofuse,anutteranceof‘today’referstothedayinthatcontext”)and,eventhoughspeakersmaynotquiteputtherulesinthoseterms,itseemspatentthattheyhavetomasterthoserules.Inthisregard,indexicalsanddemonstrativesdifferfromnamestoo,foritdoesnotseemprimafacieplausibletosaythat“BenFranklin”and“BillClinton”areassociatedwithdifferentrulesofuse.Theyarenames:learntouseone,you’velearnttousethemall.AsRussellsaid(inmomentswhereheputasidehiscontentionthatpropernamesareabbreviationsofdefinitedescriptions),aname“ismerelyanoiseorshapeconventionallyusedtodesignateacertainperson”.23Contextuallysensitiveexpressionssuchasindexicalsanddemonstrativesaregovernedbyrulesofusethatcanbeexpressedasdescriptionsthat,inagivencontextofuse,determinethereferenceoftheuseofoneofthoseexpres-sions.Yet,itisnotthosedescriptionsthatconstitutethecontributiontothetruthconditionsofusesofsentencescontainingindexicalsordemonstratives.So,indexicalsanddemonstrativesarepartlylikenames(theycontributereferentstothedeterminationoftruthconditions)andpartlylikedefinitedescriptions(descriptiveinformationdeterminesthereferentineachoccasionofuse).Fromthepointofviewofthenewtheoryofreference,ifthecriterionofclassi-ficationoftermsiscontributiontotruthconditions,descriptionswouldbeinonecategory,names,indexicalsanddemonstrativesinanother.Butifthecriterionisthemodeofconnectionbetweentheuseofanexpressionanditsreferent,thenindexicalsanddemonstrativesfallinlinewithdefinitedescriptions,whereasnamesstandaloneaspurelyconventionalstand-insfortheirreferents.WhenReferenceFailsThefunctionofaname,accordingtodirectreferencetheory,istomakeanobject,thereferent,thesubjectofdiscourse,sothatsomething,trueorfalse,can1179781472578235_txt_print.indd11708/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguagebesaidaboutit,withoutpickingouttheobjectbyappealtoanyofitsproperties.TorepeatRuthBarcanMarcus’scharacterization,apropernameperformsthesemanticfunctionofreferring“withoutbeingtiedtoanyparticularcharac-terizationoftheobject”.Whenweutter,forinstance,“JohnStuartMillwasavegetarian”,“JohnStuartMill”justconjuresuptheindividualattheendofachainofcommunication,sothatwecanattributebeingvegetariantohim.Thetruthorfalsityofwhatwesaywhenweutterthatsentence,i.e.,theutterance’struthconditions,dependonthatindividual,andonwhetherhewas,orwasnot,avegetarian,sothecontributionofthenametowhatissaid,tothepropo-sitionexpressedbytheutterance,isjustthatindividual.But,observethatwhenweusepropernames,wedonotalwaysmanagetorefer.HavingdiscoveredNeptune,LeVerrierwascompletelysureithadtobeyetanotherplanetthatwascausinganomaliesintheorbitofMercury.Heandotherastronomersevenstartedusingthename“Vulcan”,thinkingthattheywereindeedreferringtoaplanet.24Buttherewasnosuchplanet.Yet,forquitesometimeLeVerrierandmanyotherswereusing“Vulcan”asconfidentlyastheywereusing“Neptune”–asconfidentlyasweuse“Aristotle”or“JuliusCaesar”.Thepointisthat,far-fetchedasinthelattertwocasesitmayseem,propernamesthatweoftenusemayfailtorefer.Now,thisgeneratesafewproblemsfordirectreferencetheories.Foronething,ifthefunctionofanameispreciselytobeastand-inforitsreferent,makingitasubjectofdiscourseandsoon,itwouldappearthatemptynameshavenofunctioninlanguage;competentspeakersshoulddismissthemasmeremeaninglessnoises,likethesoundofsnappingfingers.Buttheydonot;surely“Vulcan”didnotsoundtoLeVerrierlikeameaninglessnoisewhenhewasusingthename,notevenwhen,afterpointinghistelescopeandfindingemptyspace,heproclaimed“Imusthavemadeamistake,Vulcandoesnotexist”.And,foranotherthing,ifanamecontributesitsreferenttowhatissaidbyanutteranceofasentence,anemptynamecontributesnothing,soitisasif,inuttering“VulcaniscausinganomaliesintheorbitofMercury”,LeVerrierwasnotsayinganythingatall.25Emptynamesdonotappeartocausesuchproblemsforadescriptivistapproach.First,thefactthatanamefailstoreferdoesnotmakeitameaninglessnoise,forthesamereasonthat“thepresentKingofFrance”or“thelargestprimenumber”arenotmerenoises,eventhoughtheydonotdenote,i.e.,theyalsofailtopickoutanindividual.Thosedescriptionscontinuetobemeaningfulbecause,aswehaveexplained,theyhavepartsthatarethemselvesmeaningful,evenifnoobjectispresentlyKingofFranceandthereareaninfinitenumberofprimes.Asforthesecondproblem,sincedefinitedescrip-tionsarecomplexsymbolsthathaveothersymbolsasparts,theycontributethemeaningsofthosepartstopropositionstoo,sotheproblemoftherebeingnocontributiontowhatissaidbyanutteranceofasentencecontaininga1189781472578235_txt_print.indd11808/08/201415:45\nReferencenon-denotingdescriptiondoesnotarise.So,thereareissuesherespecificallyfortheoriesofdirectreferencetoaddress.26Now,aboutthefirstproblem,Donnellanhasnotedthatwhatmakesanamecommoncurrencyinacommunityand,wecouldadd,whatmakesitcognitivelysignificantforaspeakerinthatcommunity,isthefactthatthereisachainofcommunication,anetworkofusesofthename.Thechainitselfmayfailtobegroundedinaninitialactofreferring.Anamemaybeintroduced,andthenpassedaround,asaprankorbyhonestmistake,causedbyaperceptualerror,orby,asinthecaseofLeVerrier,becausetheintroducerisconvincedthatthereisayetundiscoveredobjectthatsatisfiescertainspecifications.Thedifferencebetweenareferringnameandanemptyoneisthat,inthelattercase,asDonnellanputsit,thechain“endsinablock”,27andspeakersmaycometodiscoverthis,aswhenLeVerrierfinallyfacesthedarknessofspacewhereheexpectedtofindaplanet.Butthechainofcommunicationexistsnevertheless,anditsexistenceiswhatexplainsthatspeakersinacommunitydonotdismissemptynamesasmerenoises;takeyourownperspectiveasaspeaker:thename“Aristotle”hasarrivedtoyoufromanotheruser,anditsarrivalisindependentofwhetheritisareferringnameornot.But,whatdoesatheoryofdirectreferencesayaboutthecontributionofanemptynametoaproposition?Letusthinkfirstofhowthingsaresupposedtoworkinproblemfreecases:whenaspeakerutters“PlatowasaGreekphilosopher”,accordingtodirectreferencetheory,shesaysaboutacertainindividual,Plato,themanattheendofthechainofcommunicationthatleadstothespeaker’suseofthename,thatheiswasaGreekphilosopher.Whatthespeakersays,thepropositionexpressedbyherutterance,consistsofanattributionofapropertytoanobject.Thepropositionourspeakerexpressesistrue,foritisafactintheworldthatPlatowasaGreekphilosopher.Butaspeakercanalsosucceedinattributingapropertytoanobjectandexpressafalseproposition.Forinstance,ourspeakerattributesthepropertyofbeingaplaywrighttoJaneAustenwhensheutters“JaneAustenwasaplaywright”,butthepropositionsheexpressesisfalse,foritisnotafactthatJaneAustenwasaplaywright.Withthisconceptionofwhatpropositionsareandofhowfactsintheworldmakethemtrueorfalse,theviewthatasentencecontaininganemptynamedoesnotexpressapropositionis,arguably,lessfar-fetchedthanitprimafacieappearstobe.For,afterall,aspeakerwhoutters“Vulcanisaplanet”isattributingthepropertyofbeingaplanettonothingatall.Thisdoesnotmeanthatthespeakerjustproducesmeaninglessnoises,noisesthataregoingtobeequatedtosoundslike,say,snoring,bytherestofthemembersofhercommunity:whatmakesherutterancesignificantisthefactthatsheisusinganame,forwhichachainofcommunicationexists,plusalltheconnotationsthat,asmentionedbefore,speakersassociatewiththewordsthatarecommon1199781472578235_txt_print.indd11908/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguagecurrencyintheircommunity.Herutteranceissignificantforitismadeupofwordsthatarepartofalanguage;itjustdoesnothaveapropositionalcontent,somethingthatcanbeevaluatedfortruthorfalsity.28ExperimentalDataandtheTheoryofReferenceTheoriginalcaseagainstdescriptivismreliedcruciallyonargumentssuchasKripke’sandDonnellan’serrorandignorancearguments.Thoseargumentsexploitedourintuitionsascompetentspeakersinordertoestablishthataspeakerreferstothebearerofanameevenifhedoesnotassociatecorrectnoruniquelyidentifyinginformationintheformofadescription.Thus,aspeakerreferstoEinsteinwhenheuses“Einstein”,evenifheattachestothenamethedescription“theinventoroftheatomicbomb”,andhewouldrefertoGödelwhenheuses“Gödel”evenifthedescriptiveinformationhehasaboutGödel–thatheprovedtheincompletenessofarithmetic–werefalse,evenif,unbeknownsttopracticallyallofus,Gödelhadactuallystolentheprooffromamancalled“Schmidt”,anobscuremathematicianwhoappeareddeadundermysteriouscircumstances.Supposingthelatter,completelyfictional,storytobetrue,Kripkeconvincinglycontended,aspeakerwhouses“Gödel”stillreferstoGödel,nottoSchmidt,eventhough“themathematicianwhoprovedtheincompletenessofarithmetic”isadescriptionthatappliestoSchmidtandnottoGödel.RecentlysomephilosophershavetakenissuewithKripke’scontention,andhavearguedthatthereisculturalvariationinthewaypeoplereacttoKripke’sstory:whereastheconclusionthatinthefictionalcircumstanceenvisagedspeakerswhouse“Gödel”refertoGödelisacceptedbyamajorityofWesterners,peopleinothercultures,EastAsiansforinstance,displayconsid-erablystrongerdescriptivistleanings.EdouardMachery,RonMallon,ShaunNicholsandStephenStichcomparedtheresponsesofagroupofWesternerstudentsintheUSandagroupofEastAsianstudentsinHongKongtotheGödelstory.29Intheirstudytheygavetheparticipantsthefollowingtwochoices:WhenJohn[aspeakerintheenvisagedsituation]usesthename“Gödel”,ishetalkingabout:(A)thepersonwhoreallydiscoveredtheincompletenessofarithmetic?or(B)thepersonwhogotholdofthemanuscriptandclaimedcreditforthework?MacheryandcolleaguesreportthatasignificantmajorityofWesternerschose(B),thusrevealingleaningstowardsKripke’scausal-historicalaccount,1209781472578235_txt_print.indd12008/08/201415:45\nReferencewhereasmostoftheHongKongparticipantschose(A),thusrevealingdescrip-tivistleanings.Thesignificanceofthoseexperimentalfindingshasbeendebated.Thereareotherstudiesthatdonotseemtosupporttheconclusion.30Moreover,somephilosophershavearguedthatthedatathatMacheryetal.relyondoesnottakeintoaccountsomeofthestrongestargumentsbyKripkeinfavourofthecausal-historicalpicture;soalthoughtheirresultsaresomewhatinterestingandpuzzling,theydonotconstituteaseriousthreatforthetheory.31AndthereisaquestionastotheadequacyofthedatathatMacheryetal.collect.Intheirstudytheyaskparticipantswhat,ontheirview,agivenspeakerreferstowhenheusesaname,anditwouldseemthatthedatathatshouldconstitutetheinputforsemantictheorizingaboutreferencearedataabouthowspeakersusenamesandotherreferentialexpressions,notdataabouthowspeakersthinkotherspeakersusethoseexpressions.32Inanycase,andindependentlyoftheultimaterelevanceoftheseresults,thediscussionaroundtheuseofexperimentsinthetheoryofreferencehasopenedupadebateaboutanissuethatusedtobeofconcerntophilosophersoflanguage,andthatwasforsometimesomewhatabandoned:thedebateabouttheempiricalevidenceonwhichthetheoryofreferenceand,ingeneral,thetheoryofmeaningissupposedtorelyon.33Notes1.BertrandRussell,ThePhilosophyofLogicalAtomism.LaSalle,OpenCourt,1985:111–12.Seealso“KnowledgebyAcquaintanceandKnowledgebyDescription”.MysticismandLogic.Totowa:BarnesandNoble,1917:152–67(p.163).2.“Commonwords,evenpropernames,areusuallyreallydescriptions.Thatistosay,thethoughtinthemindofapersonusingapropernamecorrectlycangenerallyonlybeexpressedexplicitlyifwereplacethepropernamebyadescription.”“KnowledgebyAcquaintanceandKnowledgebyDescription”.MysticismandLogic.Totowa:BarnesandNoble,1917,p.156.3.GottlobFrege,“OnSenseandReference”,1892,n.2.4.SeeJohnSearle,“ProperNames”,Mind67,1958:166–73,foradefenceoftheviewthatthereferentofanameistheobjectthatsatisfiesasufficientnumberofthedefinitedescriptionsinaclusterassociatedwiththename.SeealsoPaulZiff,“About‘God’”(PhilosophicTurnings.NewYork:CornellUniversityPress.1966:93–102).WhetherFregehimselfwasadescriptivistisamatterofsomecontroversy.GarethEvans(TheVarietiesofReference,Oxford:OxfordUniversityPress,1982,chapter1),JohnMcDowell(“OntheSenseandReferenceofaProperName”,Mind86,1977:159–85)andotherneo-FregeanphilosophershavearguedthatFregeansensesofpropernamesareobjectdependentinthatitisthereferentthatdeterminesthesense.Thatwouldmakesensesofpropernamesradicallydifferentfromothersenses,andinparticularfromsensesofdefinite1219781472578235_txt_print.indd12108/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguagedescriptions,sinceingeneralthesenseofatermdeterminesthereferent,accordingtoFrege.5.SeePerry’sreviewofHowardWettstein’sTheMagicPrisminPhilosophicalBooks,47,2006:34–9.6.Kripke’slectureswerepublished,withaPrefaceandAddendum,byHarvardUniversityPress(Cambridge,MA.:1982);Donnellan’spaperappearedinSynthese,21,1970:335–8.7.MichaelDummett,Frege:PhilosophyofLanguage.London:Duckworth,1973,especiallychapter5.Seealsothesecondedition(1981)inwhichDummettaddressesKripke’sresponse.8.SeeScottSoames’s“TheModalArgument:WideScopeandRigidifiedDescriptions”(Noûs,32,1998:1–22,reprintedinvolume2ofSoames’sPhilosophicalEssays.Princeton,NJ:PrincetonUniversityPress,2009)forargumentsagainstwidescopismandotherformsofdescriptivism.Proponentsofwidescopismhave,nevertheless,defendedtheviewagainstsomedetailsofSoames’sargument(although,inmyview,notagainstthesubstanceofKripke’sinitial1982response).See,forexample,DavidSosa’sdiscussionofSoames’spaperin“RigidityintheScopeofRussell’sTheory”(Noûs,35,2001:1–38),andmorerecentlyDavidHunter’s,“SoamesandWidescopism”inPhilosophicalStudies,123,2005:231–41.Seealsochapter2ofScottSoames,BeyondRigidity.NewYork:OxfordUniversityPress,2002,forresponses;andBenCaplan’s“AgainstWidescopism”,PhilosophicalStudies,125,2005:167–90.9.SeeAlvinPlantinga,“TheBoethianCompromise”,reprintedinAlvinPlantingaandMatthewDavidson(eds),EssaysintheMetaphysicsofModality.Oxford:OxfordUniversityPress,2003:122–39;andFelicia(Diana)Ackerman,“ProperNames,PropositionalAttitudesandNondescriptiveConnotations”,PhilosophicalStudies,1979:55–69.10.SeeagainScottSoames,“TheModalArgument:WideScopeandRigidifiedDescriptions”,Noûs,32,1998:1–22,reprintedinvolume2ofSoames’sPhilosophicalEssays,Princeton,NJ:PrincetonUniversityPress,2009;andchapter2ofBeyondRigidity,NewYork:OxfordUniversityPress,2002.SeealsoRobinJeshion,“DescriptiveDescriptiveNames”,inM.ReimerandA.Bezuidenhout(eds),DescriptionsandBeyond,Oxford:OxfordUniversityPress,2004:591–613.Foramorerecentvariantoftheviewthatequatesnamestorigidifieddescrip-tions–incorporatingalsosomeanti-descriptivistelements–seePhilippPettit,“Descriptivism,RigidifiedandAnchored”,PhilosophicalStudies,118,2004:323–38.11.Thishascometobeknownas“theargumentfromignorance”,aspresentedbyKripke.KeithDonnellanmakesthesamepoint,arguingthatbeinginpossessionofadescriptionisnotnecessaryforreferencetooccur.12.Thishascometobeknownas“theargumentfromerror”,aspresentedbyKripke.Donnellanputsforwardsimilarconsiderationswhenhearguesthatbeinginpossessionofadescriptionisnotsufficientforthenametorefertotheentitythatsatisfiesthedescription.13.Thequotesarefrom“PossibiliaandPossibleWorlds”,GrazerPhilosophischeStudien,25–26,1985–86:107–33,and“DoesthePrincipleofSubstitutivityRestonaMistake?”,A.AndersonandR.M.Martin,TheLogicalEnterprise,NewHaven:YaleUniversityPress,1973:31–8.1229781472578235_txt_print.indd12208/08/201415:45\nReference14ASystemofLogic,1843.BookII,chapter11,§5.15.Ihavearguedforthisviewin“TheQuestionofRigidityinNewTheoriesofReference”,Noûs,37,2003:161–79.16.Anamecanalsobeintroducedusingadescription.UrbainLeVerrierintro-ducedthename“Neptune”torefertotheplanetthatcausedcertainanomaliesintheorbitofUranus.Acknowledgingthatanamecanbeintroducedviaadescriptionisnotaconcessiontodescriptivism:thedescriptionisnotwhatispassedfromlinktolinkofthecommunicationchain.UsersofthenamecanrefertoNeptuneincompleteignoranceofthedescriptionthatwasoriginallyusedtofixitsreference.17.Somephilosophershavearguedthatcasessuchas“Madagascar”poseaseriousproblemforthecausal-historicalpicture,butthismaybeduetoanover-restrictiveinterpretationoftheclaimthatthereferentistheobjectattheendofthechainofcommunication,ortoaverystrictinterpretationofthepictureofadubbingceremony.Obviouslychangesinmeaningandreferencedooccur,andthosechangesaretraceabletoeitherintentionalintroductionofnewusesforwords(newdubbings,sotospeak),oragradualprocessalsoexplicableintermsofhowwordsaretransmittedfromlinktolink.SeeGarethEvans,“TheCausalTheoryofNames”,ProceedingsoftheAristotelianSociety,SupplementaryVolume47,1973:187–208;andMichaelDevittandKimSterelny,LanguageandReality,Oxford:Blackwell,1987,esp.chapter4.18.SeeMichaelDevitt,“SingularTerms”,JournalofPhilosophy,71,1974:183–205,andDesignation,NewYork:ColumbiaUniversityPress,1981.SeealsoJosephAlmog’s“SemanticalAnthropology”,MidwestStudiesinPhilosophy,vol.ix,1984:479–89,foradiscussionoftheroleofchainsofcommunication.19.Foranin-depthdiscussionanddefenceoftheview,seeFrederickKroon,“CausalDescriptivism”,AustralasianJournalofPhilosophy,65,1987:1–17;asimilarlinehasbeenendorsedmorerecentlybyManualGarcía-Carpinteroin“APresuppositionalAccountofReference-Fixing”,JournalofPhilosophy,97,2000:109–47.20.Evengeneralterms,termssuchas“gold”or“tiger”,havebeentreatedbysomesemanticistsusingthemodelofdescriptions.Theideaisthatageneraltermappliestowhateverinstancessatisfyassociateddescriptiveinformation.Thus,forinstance,“tiger”appliestothingsthatfittheprofile“largestripedfeline”orsomethingsimilar.Sincewearefocusingonreferencewewillnotdiscussdescriptivismaboutgeneraltermshereandthereactionofproponentsofdirectreferencetheories.Influentialanti-descriptivistargumentsasregardsgeneraltermscanbefoundinHilaryPutnam’s“MeaningandReference”,JournalofPhilosophy,70,1973:699–711.21.Kaplan’smanuscriptwaswidelycirculatedinthe1970s.Itappearedeventually,with“Afterthoughts”,inJ.Almog,J.PerryandH.Wettstein(eds),ThemesfromKaplan,NewYork:OxfordUniversityPress,1984:481–563.22.In“FregeonDemonstratives”,ThePhilosophicalReview,86,1977:474–97.JohnPerryappliestoindexicalsanargumentreminiscentofDonnellan’sargumentaboutthenon-necessityandnon-sufficiencyofdescriptiveinformationinthedeterminationofthereferenceofaname.AsPerryputsit,aspeakercanbeamnesiac,lackanyinformationaboutherself,andstillrefertoherselfwhenshe1239781472578235_txt_print.indd12308/08/201415:45\nTheBloomsburyCompaniontothePhilosophyofLanguageutters“Iamlost”.Similarly,adelusionalspeakerwhothinksheisHumemayhaveaveryrichdescriptiveprofilethatfitsHumeperfectly.Yet,whenhesays“IwrotetheTreatise”heissayingsomethingfalseabouthimself.HedoesnotmanagetorefertoHume.23.“KnowledgebyAcquaintanceandKnowledgebyDescription”,MysticismandLogic,Totowa:BarnesandNoble,1917:163.24.Weshouldbecarefulincharacterizingthephenomenonthatoccupiesushere.Wearenottalkingaboutusesofnamesinfictionortotalkaboutfiction.Thestatusoffictionaldiscourseisanextremelyinterestingandcomplextopic,butherewearefocusingratheroncaseswhereaspeakerintendstorefer,assumessheistalkingaboutarealobject,andfailstodoso.LeVerriermistakenlythoughtthathewastalkingaboutarealplanetwhenheused“Vulcan”.25.Foracatalogueoftheproblemsthatemptynamesposefortheoriesofdirectreference,andwaysofaddressingthoseproblems,seeDavidBraun,“EmptyNames”,Noûs,27,1993:449–69.SeealsoMargaReimer,“TheProblemofEmptyNames”,AustralasianJournalofPhilosophy,79,2001:491–506.26.Observethatneo-Fregeanswhocontendthatnamesexpressobject-dependentsensesalsofacesimilarissues.27.SeeKeithDonnellan,“SpeakingofNothing”,ThePhilosophicalReview,83,1974:3–31.28.In“EmptyNames”,DavidBraunarguesthatthespeakerexpressesagappyproposition,ratherthannopropositionatall.29.EdouardMachery,RonMallon,ShaunNicholsandStephenStich,“Semantics,cross-culturalstyle”,Cognition,92,2004:B1–B12.30.BarryLam(in“AreCantoneseSpeakersreallyDescriptivist?Revisingcross-culturalsemantics”,Cognition,2010)reportsresultsthatsuggestthatCantonesespeakershavethesamecausal-historicalintuitionsasWesternEnglishspeakers.TheresultsofamorerecentstudybyMacheryhimselfsuggestahighlevelofintra-cultural,notcross-culturalvariation(EdouardMachery,ChristopherY.OlivolaandMollyDeBlanc,“Linguisticandmetalinguisticintuitionsinthephilosophyoflanguage”,Analysis,69,4,2009:689–94.31.SeeMichaelDevitt,“ExperimentalSemantics”(PhilosophyandPhenomenologicalResearch82(2):418–435).32.SeeGenovevaMartí,“Againstsemanticmulti-culturalism”,Analysis,69,2009:42–8,andtheensuingdiscussion:E.Machery,C.OlivolaandM.deBlanc,“LinguisticandMetalinguisticIntuitionsinthePhilosophyofLanguage”,Analysis,69,2009:689–94,andG.Martí,“EmpiricalDataandtheTheoryofReference”,inM.O’Rourkeed.TopicsinContemporaryPhilosophy,Vol.10,Cambridge,MA:TheMITPress,forthcoming.33.ClassicalsourcesinthisdebateareRudolfCarnap,“MeaningandSynonymyinNaturalLanguages”,SupplementDofMeaningandNecessity,2ndedition.Chicago:TheUniversityPress,1956;andW.V.O.Quine,WordandObject,Cambridge,MA:TheMITPress,1960(especiallychapter2).1249781472578235_txt_print.indd12408/08/201415:45\n6IntensionalContextsMichaelNelsonAgeneralsubstitutionprincipleisvalidinclassicalquantificationallogic(QL).Ifwestipulatethatthereferenceofanindividualconstantisanindividual,thereferenceofapredicateisthesetofobjectsthatitistrueof,andthereferenceofasentenceisitstruthvalue,thenwecansaythatinQLco-referringexpres-sionsofanyofthesetypesaresubstitutablesalvaveritate.ThelanguageofQLisextensional.QLhastheseattractivefeaturesbecauseofitssemantics.AninterpretationIofasetofformulaeSinthelanguageofQLconsistsofanassignmentofanon-emptysetofindividualsasthedomainD,whichcanbeconceivedastheobjectsthelanguageisabout,anassignment,foreachindividualconstantintheformulaeofS,ofexactlyoneindividualfromD,andanassignmentofann-tupleofindividualsfromDforeachn-placepredicateFnintheformulaeofS,calledtheextensionofFninI.(ThereareinterpretationswithemptydomainsandemptyindividualconstantsinfreebutnotclassicalQLs.)Wedefinetruthintermsofanotionofsatisfaction.Wedefineanobjectualnotionofsatisfactionasfollows.Asequenceofobjects<1,…,n>fromDofIsatisfiesinIaconditionFn(x…x)justincase<1,…,n>isintheextensionofFninI.1nWecanthenemploythisnotionofsatisfactiontodefinetruthforanatomicsentenceasfollows:Fn(α…α),whereFnisann-placepredicateletterand1nα…αareindividualconstants,istrueinIjustincasethesequenceofvalues1nofα…αinI,whichisann-tupleofobjectsfromD,satisfiesthecondition1nFn(x…x)inI(andhenceisintheextensionofFninI).Forsimplicity,suppose1nouronlyprimitivequantifieristheexistentialquantifier$x.Wethendefinetruthforaquantifiedsentenceintermsofthenotionofsatisfactionasfollows.Aquantifiedformula$xYxistrueinIjustincasethereissomeobject1inthedomainDofIthatsatisfiesYxinI.Thetruth-functionaloperatorsaredefinedinthestandardwayintermsoftheirtruthtables.Oneparticularversionofthesubstitutionprincipleisthesubstitutionofco-referringindividualconstants.Supposethatαandαaretwoindividual12constantsthathavethesamevalue1inI.Then,foranypairofformulaeg,jinthelanguageofQLsuchthatgdiffersfromjexactlyinthereplacementofanoccurrenceofαingforanoccurrenceofα,gistrueinIifandonlyifjistrue12inI.Thisfollowsimmediatelyfromourdefinitions.Forsimplicity,suppose1259781472578235_txt_print.indd12508/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagethatgandjareatomicsentencescontaininga1-placepredicateF1.ThengistrueinIjustincasethevalueofαinI–i.e.,1–satisfiesF1x.jistrueinIjust1incasethevalueofαinI–i.e.,1–satisfiesF1x.So,gistrueinIifandonly1ifjistrueinI.So,inQLindividualconstantswiththesamevalueareinter-substitutablesalvaveritate.Indeed,substitutionsofpredicateswiththesameextension(withinaninterpretation)preservetruthvalue(withinthatinter-pretation)andsubstitutionsofsentences,evenwhentheyaresub-formulaeoflargerformulae,withthesametruthvalue(withinaninterpretation)preservetruthvalue(withinthatinterpretation).Ofcourse,thesesubstitutionsdonotobtainacrossinterpretations.Justbecauseαandαhavethesamevalue12inIdoesnotentailthattheyhavethesamevalueineveryinterpretationI’.Indeed,ifαandαarenonlogical,thenthereisaninterpretationinwhich12theyhavedifferentvalues.Thesubstitutionprincipleonlyobtainswithinaninterpretation.Extensionalismisasignofexpressiveshortcomings.Ahostofimportantnotions–thenotionsofnecessity(logical,metaphysical,andcausal),thenotionsofbeliefanddesire,temporalnotions,amongothers–cannotbeexpressedwithinapurelyextensionalframeworkwithindividualsasthevaluesofsingulartermsandvariables.Thisisclearestwhensentenceswiththesametruthvalueareinter-substitutable.WhileitistruethatIamwearingshorts,itisnotnecessary.Butiftruthsaresubstitutablefortruthsregardlessofthelinguisticcontextinwhichtheyareembedded,thenwecouldmovefromthetrue‘Itisnecessarythat2=2’tothefalse‘ItisnecessarythatIamwearingshorts’,substitutingthetrue‘Iamwearingshorts’forthetrue‘2=2’.Necessarytruthwould,inthatcase,collapsetomaterialtruth.Anymodallogicwithsuchaprinciplewouldbeequivalenttoitsnon-modalbaseandhencewouldn’tbeamodallogicworthitssalt,containingawhollyredundantoperatorwithnonewexpressivepower.Givenourassumptionthatsentencesrefertotheirtruthvalues,modellingnecessity(andtheothernotionsmentionedabove)requiresmovingtoanintensionallanguage.Intensionallogicspromisetomodelexpressionsthataresensitivetomorethanjusttheirreferences.Quotationisaclear,non-controversialexample.Eventhough“MayaAngelou”and“MargueriteJohnson”co-refer,(1)belowistrueand(2)isfalse.(1)‘MayaAngelou’has11letters.(2)‘MargueriteJohnson’has11letters.Wecansaythattheoccurrencesofthenamesin(1)and(2)arenonreferential,followingQuine’sterminology(Quine1980).Theoccurrencesofthosesametermsin(3)and(4)below,ontheotherhand,arereferential.1269781472578235_txt_print.indd12608/08/201415:46\nIntensionalContexts(3)MayaAngelouwroteautobiographies.(4)MargueriteJohnsonwroteautobiographies.(3)istrueifandonlyif(4)istrue,asbothnamesrefertothesameobjectandbothsentencesascribethesamepropertytothatoneobject.Whileapurelyextensionallogiccanhopetomodelsentenceslike(3)and(4),itislesshelpfulwithintensionalcontexts.Quotationisonlyonenon-extensionalcontext.Wehavealreadyseenthatmodalcontextsarealsointensional.Sotooaretemporalcontexts–whileitistrueboththatIamsittingandthatIhaveskin,onlyoneoftheseisalways,oratleastaslongasIamalive,let’shope,true–causalcontexts,andpropositionalattitudecontexts.Whereasaquotationalcontextissensitivetotheidentityofexpressionswithinitsscope,itisplausibletothinkthattheothercontextsonourlistaresensitivetothemeaningofthoseexpressions.Makingthispreciserequirestakingmeaningsseriously.Ignoringcontext-sensitivity,wherethesamesentencewithaconstantlinguisticmeaningexpressesdifferentthingsindifferentcontexts,suchasthesentence“Iamhappy”,themeaningofasentenceisaproposition.Twosentencescanbothhavethesametruthvalueandyetdifferentmeanings;theycanexpressdifferentpropositions.Itisplausibletothinkthattheabovecontextsaresensitivenotonlytothetruthvaluesofthesentenceswithintheirscope,asforexamplethetruth-functionsare,buttothepropositionsexpressedbythesentenceswithintheirscope.Standardmodalandtemporallogicsarebuiltonthisveryidea.(5)belowistruewhile(6)isfalse,eventhoughbothoftheirembeddedsentencesaretrue.(5)Itisnecessarythat2=2.(6)ItisnecessarythatIamwearingshorts.Theembeddedsentenceof(5)expressesadifferentpropositionfromtheembeddedsentenceof(6)and‘itisnecessarythat’issensitivetothepropo-sitionexpressedbythesentenceitgoverns.Thefirstexpressesapropositionthatisnecessarilytruewhilethesecondexpressesapropositionthatisonlycontingentlytrue.Whilemodalcontextsareintensional,inthesensethatsentenceswiththesametruthvaluearenotalwaysintersubstitutable,itmaybethatpropernamesareintersubstitutablewithinmodalcontexts.Whetherthisissodependsonanumberofimportantissues,andfullyexploringthisissueisbeyondthescopeofthepresentessay.Wewillhavetosettlewithaquicksketchofasetofthesesthatleadtosuchaview.Thefirstthesisisthethesisofdirectreference.Itiseasiesttostatethis1279781472578235_txt_print.indd12708/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagethesisifweassumethatpropositionsarestructured.Thenthethesisofdirectreferenceistheviewthatsimplesentenceswithpropernamesexpresssingularpropositions,whereapropositionissingularwithrespecttosomeindividualijustincaseitisaboutiinvirtueofhavingiasaself-representingconstituent.Propernamesaredirectlyreferential,expressingastheirsolesemanticcontenttheirreferences.Aconsequenceofthisviewisthattwosimplesentenceswithco-referringpropernamesexpressthesamesingularproposition.Thesecondthesisconcernsthesemanticsofthemodaloperator.Inshort,themodaloperatorisonlysensitivetothepropositionexpressedbytheembeddedsentenceitoperateson,andnotitsmodeofexpression.Thesetwothesesentailthatasubstitutionprincipleforco-referringpropernamesisvalidformodalcontexts.Thisisbecausesubstitutingco-referringpropernameswithinamodalcontextwillnotaffectthepropositiontheembeddedsentenceexpresses,bythefirstthesis.Andbythesecondthesis,themodalitselfisonlysensitivetothepropositionexpressedbytheembeddedsentenceitoperateson.Sothesubstitutionwillnotaffectthetruthvalueofthewhole.So,insofarasweaccepttheabovetheses,wearecommittedtotheclaimthatco-referringpropernames,butnotsentenceswiththesametruthvalue,areintersubstitutablewithinmodalenvironments.Inthefollowingsectionsweshalllookataparticularinstanceofaninten-sionalcontext–namely,propositionalattitudeconstructions–inmoredetail,spellingoutavarietyofpositions.PropositionalAttitudesOurpartiallistofintensionalcontextsincludedpropositionalattitudeconstruc-tions,whichincludeverbssuchas“believe”,“desire”,“hope”and“intend”,onatleastsomeoftheiruses.Theprimaryfocusofthischapterispropositionalattitudeverbs.Thestandard,thoughnotuniversal,viewofsuchverbsisthattheyexpressrelationsbetweenagentsandpropositions,wherepropositionsarelanguage-independententitiesthataretheprimarybearersoftruthvalue,whatareexpressedbysentences(orperhapsutterancesofsentences),andtheobjectsofbeliefs,desires,wishes,etc.Ishallassumeanumberofcontroversialthesesconcerningtheseexpressions.Thefirstispropositionalism,accordingtowhichsentences(orsentencesincontextorutterancesofsentences–Iwillremainneutralonissuesconcerningtheprimarybearersofsemanticcontent)expresspropositionsinvirtueofwhichtheyaretrueorfalseandpropositionsarethecontentsofcognitive(andsomeconative)attitudes.Thismayseemlikeatrivialassumption;theyare,afterall,calledpropositionalattitudes.Butacceptingtheassumptionistosetasideawiderangeofalternativeaccountsthathistoricallyhavebeenveryinfluential.Amongthemostimportantaresententialistaccounts1289781472578235_txt_print.indd12808/08/201415:46\nIntensionalContextsoftheattitudes,accordingtowhichsentences(orutterances)aretheprimarybearersoftruthvaluesandtheobjectsoftheattitudes.Sententialistsabouttheattitudesmaintainthatbelievingthatgrassisgreenorwantingworldpeace,forexample,isamatterofbelieving-truethesentence“Grassisgreen”ordesiring-truethesentence“Thereisworldpeace”.(See,forexample,Carnap1958,Quine1956,Davidson1967,1968,andHigginbotham1991.)Hereisaproblemforsententialism.(SeeChurch1950.)Intuitivelywecanreportanagent’sattitudeusingsentencesthattheagentherselfwouldnotusetoexpresstheattitudewearereporting.Oneclearcaseofthisisreportingtheattitudesofnonlinguisticbeings.“Look,thecatthinksthemouseisstillbehindthebooks.”Thisseemstrueonthebasisofthecat’sbehaviour,butitisprettyimplausiblethatthecatbelieves-truethesentence“Themouseisstillbehindthebooks”.ThesamepointcanbemadebyconsideringEnglishreportsoftheattitudesofnon-Englishspeakers.WemayknowthatJonaswantsabikeandtrulyreportthatfactbysaying“Jonaswantsabike”.ButifJonasonlyspeaksSwedish,thisisnotbecauseJonaswants-truetheEnglishsentence“Jonashasabike”.Finally,sometimeswereportEnglishspeakers’beliefsusingEnglishsentencestheywouldnotusetoexpresstheirbeliefs.Forexample,Icantrulysay,“BillbelievesthatIamsteppingonhistoes”,asIstandonhistoes.Butitisquitedubiousthatthisisbecausehebelieves-truethesentence“Iamsteppingonhistoes”.Iwouldreportthatlatterstatebysaying,“Billbelievesthatheissteppingonmytoes”.Inallofthesecases,itwouldseemthatthereportistruebecausetheembeddedsentence(thesentenceinthethat-clauseofthereport;thesentence“Themouseisstillbehindthebooks”,forexample)expressessomethinginthemouthofthereporterthattheagentofthereportbearstheappro-priateattitudeto.Billmaybelieve-truethesentence“Heissteppingonmyfoot”that,inhismouth,expressesthesamethingasthesentence“Iamsteppingonhisfoot”expressesinmymouth.Butitseemsthatwhatrelatesthesesentences(orsentence-tokens)isthefactthattheyexpressthesameproposition,whichleadsusbacktopropositionalism.Somesententialistshaveintroducedanotionofsame-saying,claimingthatanattitudereportisaccurateinsofarastheembeddedsentence,asusedbythereporter,same-sayssomesentencetheagentofthereportbelieves-(ordesires-,oretc.)true.(SeeDavidson1968andLeporeandLoewer1989.)Thesame-sayingrelationshouldnotbetakenasaprimitive.Buttheninvirtueofwhatdotwoutterancessame-say?Whentheyexpressthesameproposition?Ifso,wearebacktopropositionalism.Anditishardtoseehowelsetoexplicatethesame-sayingrelation.Finally,introducingasame-sayingrelationwillnothelpwiththefirstproblemofreportstonon-linguisticbeings,suchasnon-humananimalsandyoungchildren.Theissuesarecomplexandthereismoretosay.Idon’tpretendtohave1299781472578235_txt_print.indd12908/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageestablishedmyfirstassumption.ButIdothinkthatIhaveofferedsomemotivationforacceptingit.Second,Ishallassumeastructuralversionofpropositionalism,accordingtowhichpropositionshaveastructure,muchlikethesyntacticstructureofthesentencesthatexpressthem.(Itisafurtherthesis,andoneweshallconsiderbelow,whetherornotthestructureofthepropositionexpressedbyasentenceorutteranceofasentencemustmatchthesyntacticstructureofthesentenceitself.Myassumptionhereisneutralonthisissue.)Thisrulesoutviewsaccordingtowhichpropositionsaresetsofpossibleworldsorfunctionsfrompossibleworldstotruthvalues(seeLewis1972,1986andStalnaker1981,1984,1987,1988)orsituations(seeBarwiseandPerry1983).Whiletheseviewsarewidelyassumedinformalsemantics,therearepowerfulreasonstothinkthattheyareincapableofdoingjusticetothoughtandreasoning.(See,forexample,Soames1985.)That2+2=4andthatarithmeticisincompletearetrueinthesamesetofworlds–namely,allmathematicallypossibleones.Butmostfive-yearoldsknowthefirst,whilefewknow,orevenarecapableofgrasping,thesecond.Furthermore,someonemaywellbereasoningrationally,givenherconceptions,injudgingtrueanecessarilyfalseproposition.Forexample,Hilbertjudgedtruetheimpossiblewhenhethoughtthatmathematicscouldbefinitelyaxiomatizedwithaprovablyconsistentsystem,buthewasnotirrational.MathematicalknowledgebuildsonitselfanditwasprimarilythroughworkingontheHilbertprogrammethatGödelwasabletoprovehisincompletenesstheorems.NowsupposethatHilberthadbelievedthat2≠2.Thenhewouldhavebeenirrational.Inthatcase,hebelievedsomethingovertlyimpossible.So,itisonethingtobelievethatmathematicsisfinitelyaxiomatizedwithaprovablyconsistentsystemandanothertobelievethat2≠2.Ourjudgmentsofrationality,ofwhensomeoneisfailingtojudgeastheyought,requiresmakingthisdistinction.So,itseemsthatthesearedifferentpropositions,inwhichcasepropositionsarenotsetsofworldsorsituations.Thecontentsofourattitudesaremuchmorefine-grainedthanwhateitherthepossibleworldssemanticistorsituationtheoristgiveus.Again,Idonottakethistoestablishthesecondassumption,butIdothinkthatitmotivatesit.(We’llreturntothisformofargumentbelowinFrege’sPuzzle.)Withalloftheseassumptionsmade,onemightthinkthatIhavedefinedmyselfintoauniquespotinthelandscapeofpositions.Butplentyofinter-estingissuesremain.Andourassumptionscanhelpusbetterfocusonandarticulatethoseissues.Thefirstissueconcernsthecontentsofsentencesandthecontentsofattitudes.Isthecontentofasentenceandthecontentofanattitudeeverasingularproposition,orarethecontentsofsentencesandthecontentsofattitudesalwaysfullyconceptualized?AsingularpropositionaboutmetotheeffectthatIamsitting,forexample,containsmeandthepropertysitting.1309781472578235_txt_print.indd13008/08/201415:46\nIntensionalContextsThepropositionthattheauthorofthischapterissitting,ontheotherhand,plausiblydoesnotcontainmeasaconstituentbutratheraconceptualizationofme,astheauthorofthischapter.Thislatterpropositionconcernsmeinvirtueofmysatisfyingtheconditionofuniquelyauthoringthischapteranditisthisconditionandnotmethatformsthepropositionitself.Theformerpropositionconcernsmeinvirtueofhavingmeasadirectlyrepresentingconstituent.Ifsentencesofnaturallanguageexpresssingularpropositions,thenthereareexpressionsinnaturallanguagethataredirectlyreferential.Theirsemanticcontentsareexhaustedbytheirreferences.Referentialistsclaimthattherearesingularpropositionsandtheyarethecontentsofsentencesandofourcognitivestates.Fregeansdenythis.Thisdebateformsourfirstissue.Thesecondissueconcernswhetherornotasubstitutionprincipleisvalidforpropernames(andindexicalsanddemonstratives)thatoccurwithinthecontextofpropositionalattitudeverbs.FromthefactthatHesperusisPhosphorus,doesitfollowthatanyonewhobelievesthatHesperusisHesperusbelievesthatHesperusisPhosphorus?Thisquestionformsoursecondissue.Thetwoissuesarerelated.Themostpowerfulargumentsagainstadirectreferencetheoryrestonaveryrobustsetofintuitionsseemingtosupporttheinvalidityofthesubstitutionprinciple.InwhatfollowsIshallpresentsomeofthemainpositionsontheseissues.Frege’sPuzzleAllenStewartKonigsbergwasborninNewYorkCityin1935.Bytheageof15,hewassellinghisjokestoanewspaper.Bytheageof18,heenrolledinNYU,butsoondroppedout.Hewentontowrite,produceandactinsome30films.Janerememberstheboyofherearlyyears,theboyshecalled“AllenKonigsberg”.Butshelosttrackofhimasanearlyteenager.Still,shethinksofhimoften.Shealsothinksofthequirkyfilmdirectorwhogoesunderthename“WoodyAllen”.Shedoesn’trealize,itseems,thattheyareoneandthesame.“IfonlyIknewwhatbecameofAllenKonigsberg,”sheoftensaystohusband.“Hewasalwaysbuildingthings.He’sprobablyanengineer.Nothingcreative.Notafilmmaker,that’sforsure…Hey,let’sgoseeWoodyAllen’snewmovie.”OnthebasisofJane’slinguisticbehaviour,itisplausiblethatthefollowingaretrue.(7)JanebelievesthatWoodyAllenisafilmmaker.(8)JanebelievesthatAllenKonigsbergisnotafilmmaker.(9)JanedoesnotbelievethatAllenKonigsbergisafilmmaker.1319781472578235_txt_print.indd13108/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageGrant,forthetime,that(7)–(9)aretrue.Thentheprinciplethatco-referringsingulartermswithinthescopeofapropositionalattitude-ascribingverbaresubstitutablesalvaveritateisinvalid.Forsupposeotherwise.Then,as“WoodyAllen”and“AllenKonigsberg”areco-referring,thisprinciplewouldallowusinfer(10)fromthetruthof(9).(10)JanedoesnotbelievethatWoodyAllenisafilmmaker.But(10)isthecontraryof(7),whichwehavealreadyassumedistrue.So(10)isfalse.Contradiction.Thesubstitutionprinciplealsoallowsustoinferthetruthof(11)fromthetruthof(7).(11)JanebelievesthatAllenKonigsbergisafilmmaker.Butifboth(8)and(11)aretrue,thenJanebelievesapropositionanditsnegation.Thisisbecausetheembeddedsentencein(11)–i.e.,(12)–isthecontraryoftheembeddedsentencein(8).(12)AllenKonigsbergisafilmmaker.Rationalpeopledonot,atthesametime,believeapropositionanditsnegation.Janewould,inthatcase,notfulfiltherequirementsofgoodbelieving.Butthatisnotterriblyplausible,asnoamountofpurereflectionislikelytorevealtoherthemistakesheismaking.Hermistakeisnotlikethemistakeofthepersonwhobelievesthattherecanbeabarberwhoshavesallandonlythosewhodonotshavethemselves,forexample.Whilethiscontradictiondoesnotwearitscontradictorynatureonitssleeve,carefulreflectionofwhatwouldbeinvolvedinsuchasituation,andwonderingwhoshavesthebarberhimself,revealstothethoughtfulpersonthemistake.Thechargeofirrationality,ofnotbelievingasoneshould,sticksinthiscasebutnotinJane’s.ButthenJanedoesnotbelieveapropositionanditsnegation.So,thetruthof(11)doesnotfollowfromthetruthof(8).As(8)istrue,(11)isfalse.So,co-referringnamesarenotsubstitutablewithinthescopeof“believe”.(Indistinguishingthesetwoaspectsofthepuzzle,theoneinvolvingbelievingandnotbelievingandtheotherinvolvinganintuitivelyrationalagentbelievingapropositionanditsnegation,IamfollowingKripke1979.)Frege’spuzzleisbasedonthemorebasicinsightthatsentencesthatdifferonlyinthesubstitutionofoneco-referringnameforanother,suchas(12)and(13),differincognitivevalue.(13)WoodyAllenisafilmmaker.1329781472578235_txt_print.indd13208/08/201415:46\nIntensionalContextsTosaythatsuchpairsofsentencesdifferincognitivevalueistosaythatcompetent,reflectiveagentscanrationallyacceptonewhilerejectingtheother.ThisdifferenceisattheheartofFrege’spuzzle.Itisverytemptingtothinkthatthedifferenceentailsadifferenceincontent,orpropositionexpressed.(12)and(13)expressdifferentpropositions,whichaccountsforthedifferentattitudescompetent,reflectiveagentscantaketowardsthem.Butifthatisso,thentheydonotexpresssingularpropositions,astheywouldinthatcaseexpressthesameproposition.Moreover,propernamessuchas“WoodyAllen”arenotdirectlyreferential.Theircontentmustbeotherormorethantheirbarereferences.FregeanSolutionAFregeansolutiontoourpuzzletakesappearancesforgranted.(12)and(13)differincognitivevalueandexpressdifferentpropositionsand(7)–(9)aretrueand(10)–(11)arefalse.(7)and(11)reportdifferentbeliefs.For(7)tobetrue,Janemuststandinthebeliefrelationtotheproposition(WoodyAllenisafilmmaker).For(12)tobetrue,Janemuststandinthebeliefrelationtotheproposition(AllenKonigsbergisafilmmaker).TheFregeanclaimsthatthesearedistinctpropositions.(12)and(13),whilereferentiallyequivalent,expressdistinctpropositions,astheyexpresspropositionsthatcontaindifferentmodesofpresentationofWoodyAllen.Asingleobjectcanbepresentedtoanagentinmanydifferentways.Anagentcanbelieveofasingleobjectthatitissuchandsuchunderonemodeofpresentationbutnotunderanother.Thesebeliefsthenhavedistinctcontents.ThisistheheartofFrege’sdistinctionbetweensenseandreference.(SeeFrege1948,1956.)TheFregeansolutioniscomprisedoftwodistinctivetheses.Thefirstisthatdifferencesincognitivevaluecorrespondtodifferencesinthoughtorpropo-sitionexpressed.Thesecondisthatdifferencesinpairsofsentencessuchas(7)and(12)–thatis,pairsofpropositionalattitude-ascribingsentencesthatdifferonlyintheco-referringsubjectpositionexpressionintheembeddedsentencethatintuitivelydifferintruthvalue–aretobeexplainedintermsoftheirascribingdifferentbeliefcontentstotheagent,theoneofwhichtheagentbelievesandtheotherofwhichshedoesnot.So,itisbecause(12)and(13)expressdifferentpropositions–propositionsabout,tobesure,thesameindividualtotheeffectthatheisafilmmakerandhencepropositionswiththesamereferentialcontent–thattheyhavedifferentcognitivevalues.AnditisbecausebelievingthatWoodyAllenisafilmmakerisdistinct,invirtueofitscontent,frombelievingthatAllenKonigsbergisafilmmakerthat(7)isconsistentwith(9).OnewayofmakingmorepreciseabroadlyFregeansolution,onebasedon1339781472578235_txt_print.indd13308/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagethetwodistinctiveclaimsidentifiedabove,istoaccepttheRussellianclaimsthatordinarypropernamesare“disguiseddefinitedescriptions”,definitedescrip-tionsexpressdescriptiveconditions,andco-designatingdefinitedescriptionscanexpressdifferentdescriptiveconditions.Then(12)and(13)canbeseenasexpressingdifferentbutreferentiallyequivalentpropositions,eachconcerningWoodyAllenhimself.Forsimplicity,let’ssaythatanameisshortforitsownmetalinguisticdescription.Then“AllenKonigsberg”isshortfor“thepersonnamed‘AllenKonigsberg’”and“WoodyAllen”isshortfor“thepersonnamed‘WoodyAllen’”.(ThisisnotwhatRussellthought.)(12)and(13)canthenberegimentedassomethingasfollows.(Let“[ixFx]Yx”represent“TheFisY”andlet“Ax”abbreviate“xisnamed‘AllenKonigsberg’”,“Wx”abbreviate“xisnamed‘WoodyAllen’”,and“Fx”abbreviate“xisafilmmaker”.)(12)[ixAx]FxR(13)[ixWx]FxRThesearedistinctpropositions.(SeeManuelGarcia-Carpintero’schapterinthisvolumefordetailsonRussell’stheoryofdefinitedescriptions.)Moreover,itispossibleforarational,linguisticallycompetentpersontonotrealizethatthepersonnamed“AllenKonigsberg”isthepersonnamed“WoodyAllen”.So,wehaveanaccountoftheconsistencyof(7)–(9)andwhysubstitutionofco-referringpropernameswithinthescopeofpropositionalattitudeverbsfails.Suchsubstitutionscanaffecttheidentityofthebeliefbeingreportedbychangingthecontentoftheembeddedsingularterm.ProblemsfortheFregeanSolutionTheFregeansolutionfacesseveralproblems.HereIshalldiscussthree.ThefirstproblemarisesfromadetailinFrege’sownversionoftheview.Fregethoughtthatthesubstitutionprincipleisvalid,evenforexpressionswithinthescopeofattitudeverbs.Thename“WoodyAllen”isco-referentialwiththename“AllenKonigsberg”astheyoccurinsimplesentencessuchas(12)and(13).Butastheyoccurincomplexsentencessuchas(7)and(8),heclaimedthattheyarenotco-referential.Insimplesentences,thenamesrefertothemanhimselfandexpressamodeofpresentationorsense.Insidethescopeofanattitudeverb,however,theyrefertotheircustomarysenses,notthemanhimself.Asthecustomarysenseof“WoodyAllen”isdistinctfromthecustomarysenseof“AllenKonisberg”,theirreferencesin(7)and(8)aredistinct.Propositionalattitudeverbsinduce,then,areferenceshift.Termsthatoccurwithintheirscoperefertotheircustomarysenses.1349781472578235_txt_print.indd13408/08/201415:46\nIntensionalContextsSomehavefoundplausibleaprincipleofsemanticinnocence,accordingtowhichthelinguisticcontextofanexpressiondoesnotchangeitsreference.(Davidson1968istheclassicsource.SeealsoBarwiseandPerry1983.)Semanticinnocenceseemstoberequired,furthermore,toexplaincross-attitudinalanaphoricreference.Consider,forexample,thefollowingsentence.(14)WoodyAllendirectedVickiBarcelonaandJanebelievesthathestarredinit.Thepronoun“he”in(14)occurswithinthescopeoftheattitudeverb“believe”butisanaphoricontheoccurrenceof“WoodyAllen”,whichoccursoutsidethescopeofthatverb.So,theoccurrenceof“WoodyAllen”in(14)referstothemanhimself.Astandardtheoryofanaphoricrelationswoulddictatethatthereferenceoftheoccurrenceof“he”in(14)isdeterminedbythereferenceofthetermthatitisanaphoricon.So,theoccurrenceof“he”in(14)referstothemantoo.ButthatconflictswithFrege’sclaimthatpropositionalattitudeverbsinduceareferenceshift,inwhichcaseonewouldexpect“he”,asitoccursin(14),torefertoitscustomarysense,whateverthatmaybe.Attitudinalanaphora,then,challengestheFregeanclaimthatattitudeverbsinduceareferenceshiftandsupportssemanticinnocence.WhiletheFregeanmightattempttodefendthereferenceshiftingviewbyofferingamorecomplextheoryofcross-attitudinalanaphora,inwhichthereferenceof“he”issomesenseofthereferenceoftheexpressionuponwhichitisanaphoric,asopposedtothereferenceofthatexpression,itisimportanttoseethatFregeanismitselfisindependentofthereferenceshiftview.ThisisbecausethecoreinsightsofFrege’ssolution,namelythatdifferentco-referringnameshavedifferentsensesandthatthisdifferenceinsenseisrelevanttothedifferenceinthethoughtsagentshaveincasesinwhichtheysufferfromanidentityconfusion,canbeembeddedinatheoryconsistentwithsemanticinnocence.Thekeyistoclaimthatpropositionalattitudeverbsaresensitivenotonlytothereferencesoftheexpressionsintheirscopebutalsototheirsenses.Wethenhavethesameeffectsofreferenceshiftswithoutclaimingthattheexpressionsthemselvesrefertothecustomarysenses.TheFregeanclaimsthat(12)and(13)expressdifferentpropositions.Shecanthensaythatpropositionalattitudeverbsthemselvesaresensitivetothesensesofthetermswithintheirscope.(12)and(13),althoughhavingthesamereferentialcontent,willthennotbeintersubstitutablewithinthescopeofpropositionalattitudeverbs.Thereisnoneedtoclaimthattheembeddedtermsthemselvesshifttheirreferences.Theoccurrenceof“WoodyAllen”in(7)referstoWoodyAllen,justasitdoesin(12).So,anobjectiontoFregeanismthatrestsonsemanticinnocenceisnotsuccessful,assemanticinnocenceiscompatiblewithFregeanism.1359781472578235_txt_print.indd13508/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageTherearetwofurtherproblemswithFregeanismthatIshallconsider.ThefirstIwillcalltheproblemofsamenessofsenseandthesecondtheproblemofdescriptivism.ThesethreatenFregeanpositionsmoregenerally.Theproblemofsamenessofsensecanbeseenbyreturningto(14).Itseemstomethat(14)istrue.Butwhatarethetruthconditionsfor(14)?Inparticular,istheresomespecifiedwayinwhichJanemustthinkofWoodyAllenunderwhichshejudgesthathestarredinVickiBarcelona?It’shardtoseehowthatcouldbe.Unlike(7),inwhichitisatleastnotimplausibletothinkthattheterm“WoodyAllen”specifiessometypeofwayunderwhichJanemustthinkofWoodyAllenunderwhichshejudgesthatheisafilmmaker,itisnotplausibletothinkthattheoccurrenceof“WoodyAllen”in(14)specifiesoneparticularway.Ifthatisnotclear,thennotethat(14)wouldseemtoentail(15)below,as“WoodyAllen”occursoutsidethescopeofanintensionalverb.(15)AllenKonigsbergdirectedVickiBarcelonaandJanebelievesthathestarredinit.If(14)weretorequireforitstruththatJanethinkofWoodyAlleninaparticularsortofway,then(15)mustrequireforitstruththatJanethinkofWoodyAlleninthatsameparticularsortofway.Otherwiseitishardtoseehowtheonecouldentailtheotherandviceversa.Butthatisimplausible.Howcouldboth(14)and(15)requirethatJanethinkofWoodyAlleninthesamewaywhenthedifferentsingularterms,bytheFregean’sownlights,expressdifferentsenses?Thefundamentalproblemthatthisbringsoutisthatsomebeliefattributionsdonotseemtorequirefortheirintuitivetruththatthebelieverthinkabouttherelevantobjectsinanywaythatissystematicallyrelatedtothecomplementclauseofthebeliefattribution.Thissamepointcanbemadewithindexicalsanddemonstratives.Considerthefollowingsentence,asspokenbyWoodyAllen.(16)JanebelievesthatIamafilmmaker.Itisclearthatthetruthof(16)doesnotrequirethatJanebelievesomethingthatshewouldexpressbyutteringthesentence“Iamafilmmaker”.Toreportthatbelief,WoodyAllenwoulduse(17).(17)Janebelievesthatsheherselfisafilmmaker.So,thetruthof(16)doesnotrequirethatJanebedisposedtoexpressherbeliefusingtheembeddedsentenceof(16).Butitisalsoclearthatthetruthof(16)doesnotrequirethatJanebedisposedtoexpressherbeliefusingasentencewithanexpressionthatissystematicallyrelatedtothefirst-personpronoun.1369781472578235_txt_print.indd13608/08/201415:46\nIntensionalContextsForexample,itisimplausiblethatshemustbedisposedtoutterthesentence“Youareafilmmaker”inWoodyAllen’spresence.Shemightnotbeinapositiontousethesecond-personpronoun,notbeingintherightrelationstoWoodyAllen.Or,shemightbeinWoodyAllen’spresencebutnotrecognizehimasthepersonwhocreatedallthefilmssheassociateswiththename“WoodyAllen”.What(16)seemstoreport,instead,isthatJanehavesomebeliefwiththeequivalentreferentialcontenttoWoodyAllen’sutteranceof“Iamafilmmaker”.Thereseemstobenomorefine-grainedspecificationofarequiredwayofthinkingofWoodyAllenfor(16)tobetrue.Attheveryleast,thereisnotsystematic,context-independent,fine-grainedwayofthinkingofWoodyAllenthatisinvolvedin(16)’struth-conditions.Whiletheseproblemsmaybesurmountable,theyshowthathowthecomplementclauseofanattitude-ascribingsentenceintuitivelyrepresentshowtheagentconceivesoftheentitiestheattitudeisaboutisnotastraightforwardmanner.Itisimplau-siblethatthereisastablewayofthinkingofthereferencethatallcompetentspeakersassociatewithanamesuchthatanyattitude-ascribingsentencewiththatnameinthesubjectpositionofitscomplementclauseistrueonlyiftheagenthasabeliefunderthatwayofthinkingofthereference.ThefinalproblemIshallconsideristheproblemofdescriptivism.IhavedescribedFregeanismastheviewthatdifferencesincognitivevalue(betweensentencessuchas(12)and(13),forexample)aretobeexplainedintermsofadifferenceinpropositionexpressedandintuitivedifferencesinthetruthvalueofreferentiallyequivalentpropositionalattitude-ascribingsentences(suchas(9)and(10),forexample)aretobeexplainedintermsofadifferenceinthethoughtcontentascribed.Descriptivismisathesisaboutthesemanticsofpropernames,accordingtowhichthesemanticcontentofapropernameisgivenbysomedefinitedescription.Whilethereisnotadirectconnectionbetweenthetwotheses,descriptivismpromisesaconcretewayofestablishingthetwodistinctivethesesofFregeanism.Anddescriptivismissubjecttoafamousandinflu-entialsetofobjections–namely,SaulKripke’santi-descriptivismarguments(Kripke1981)andStrawson’sreduplicationargument(Strawson1959).Kripke’sprimarythreeargumentsagainstdescriptivism,themodalobjection,thesecondthesemanticobjection,andthefinaltheepistemologicalobjection,arediscussedinGenovevaMarti’schapterinthisvolumeon“Reference”,soIshallnotdiscussthemhere.ButIshalldiscussStrawson’sreduplicationargument.Strawson’sArgumentFirstsometerminology.Toidentifyanobjectistohavethecapacitytothinkdeterminatelyaboutthatveryobject.Descriptiveidentificationrequires1379781472578235_txt_print.indd13708/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagepossessingknowledgeaboutsomedescriptiveconditionthatobjectsatisfies.Demonstrativeidentificationismoredirect,beinggroundedinreal-worldconnectionsonebearstotheobjectasopposedtoone’sconceptionoftheobjectanditsrelations.Strawsonthenarguesthatthoughtabouttheexternalworldrequiresaprimitiveformofdemonstrativeidentification.Thisconclusionconflictswithathoroughgoingformofdescriptivism,accordingtowhichallourthoughtinvolvesonlydescriptiveidentificationofindividuals.Supposethatwelivedinareduplicationuniverse,inwhichthereisadistantregionoftheuniversethatisqualitativelyindiscerniblefromourregionoftheuniverse.Iamsittingatmydesk,thinkingaboutmycomputer.Ontheothersideoftheuniverse,thereisaqualitativelyidenticalsetofhappenings.Intuitively,Iamthinkingaboutthecomputerinfrontofmeandmyquali-tativetwinisthinkingaboutthecomputerinfrontofhim,and,intuitively,thesearedistinctthings.(Theclaimthattherearetwopeople,twodesks,andtwocomputersisdisputedbysome;see,forexample,O’Leary-Hawthorne1995.Theyclaimthatobjectsarebundlesofimmanentuniversalsandsowhatseemsliketwodistinct,spatiotemporallyseparate,qualitativelyindiscernibleobjectsisreallyoneobjectatsomespatiotemporaldistancefromitself.Iassumewithoutargumentthatthisiswrong.)Ifallidentificationweredescriptive,thiswouldnotbepossible,asthereisnopurelyqualitativedescriptiveconditionthatonlymycomputersatisfies.So,ifallidentificationisdescriptive,thentheoccupantsofthemassivereduplicationuniversedonothaveanythoughtsdeterminatelyabouttheparticularsintheirsurroundings,includingthemselves.Thisisimplausible.Strawsonmaintainsthatobjectsinourimmediateperceptualenvironmentcanbeidentifieddirectly–demonstratively–asthat.AsIamsittinginfrontofmycomputer,Icandirectlyanddemonstrativelyidentifythatobjectinvirtueofmyperceptualrelationtoit.Idonotneedtoconceiveofanyrelationsthatthatobjectuniquelybears.Rather,itissufficientforittosimplybethecasethatthatobjectisuniquelyperceptuallyrelatedtomeformetobeabletoidentifyit.BecauseIamperceptuallyrelatedtomycomputerandmytwintohis,wecaneachsucceedindemonstrativelyidentifyingourrespectivecomputersandthinkingdeterminatethoughtsaboutthem.Ithinkthatthereduplicationargumentshowsthatglobaldescriptivismisfalse.Itshowsthatwethinkaboutatleastsomeparticularsdirectlyordemon-stratively.Butitleavesopenwhatentitieswethinkaboutdemonstratively.ItdoesnotsupportStrawson’sviewthatwethinkdemonstrativelyabouttheobjectsinourperceptualenvironment.Toseewhy,considertheviewthatwethinkdemonstrativelyonlyaboutouroccurrentsense-data(Russell1910,1912).Asmysense-dataarenumericallydistinctfrommytwin’s,thedescriptionthecomputerthatisacausalsourceofthatsense-datumissatisfiedbymycomputerinsteadofbymytwin’s.Thisviewrespectstheintuitionthattheinhabitantsof1389781472578235_txt_print.indd13808/08/201415:46\nIntensionalContextsareduplicationworldthinkdeterminatelyaboutparticularsintheirenviron-ments.So,thereduplicationargument,whileestablishingtheneedforsomeirreducibledemonstrativethoughts,leavesopenthescopeofdemonstrativethought.(Formoreonthenatureofsingularthought,seetheessayscollectedinJeshion2010.)Thereduplicationargumentchallengesglobaldescriptivism.ButhowdoesthereduplicationargumentchallengeFregeanism?Letmesketchalineofargument.Grantthatthereduplicationargumentshowsthatthereisanirreducibleformofdemonstrativethoughtaboutsomeparticulars.Insofarasitispossibletomisidentifythoseparticulars,thentherewillbeFregecasesthatarenotsusceptibletoaFregeansolution.Thisisbecausetherewon’tbedistinctcontentsavailable,asourthoughtisreferentialwithrespecttothoseparticulars.ThisproblemisclearestifweconsidertheStrawsonianview,accordingtowhichtheobjectswithwhichweareperceptuallyrelatedaredemonstrativelyidentified.Clearlyitispossibletomisidentifymid-sizedobjectsofperception.IcanperceivethecomputerandnotrealizethatitisthesameastheobjectIearlierperceived.Ifmythoughtaboutthecomputerisdemonstrative,thenwecannotexplaintheconfusionintermsofadifferenceinmodeofpresentationthatleadstodifferentthoughtcontents.So,thereareasetofFregecasesforwhichtheStrawsoniancannotofferaFregeansolution.Wemaythensuspectthatthenon-FregeanaccountofthesecasescanbeappliedtoallFregecases.ReferentialismReconsideredTheaboveproblemsmaysuggestareturntoreferentialism.ButFrege’spuzzlechallengesthereferentialist.Aswehaveseen,theintuitionssupportingthepuzzlestronglysuggestthatpairsofsentencessuchas(12)and(13)expressdifferentpropositions.Butifthatisso,thenwhattheyexpressgoesbeyondtheirreferentialcontents.Thereferentialistcan’thavethat.Whatcanbesaid,then,aboutthosesupportingintuitions?Therearetwomainreferentialiststrategies.Thefirstemploysthedistinctionbetweenwhatissaidbyanutteranceofasentenceandwhatispragmaticallyimplicated.Thekeyisthat,whileutterancesofreferentiallyidenticalproposi-tionalattitude-ascribingsentencessaythesamething,theyoftenpragmaticallyimplicatedifferentinformationwithdifferingtruthvalues.ThehopeisthattheintuitionsgeneratingFrege’spuzzlecanbefullyaccountedforintermsofthispragmaticallyimplicatedinformation.Onthisstrategy,(7)and(9)aremutuallyinconsistent.Theintuitionthattheyareconsistentistobeexplainedintermsoftheinformationthatordinaryutterancesofthosesentencespragmaticallyimplicate.Thesecondstrategyrestsontheclaimthatpropositionalattitude-ascribingsentencesarecontext-sensitive.Onthisstrategy,(7)and(9)are1399781472578235_txt_print.indd13908/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageconsistent,despitethefactthattheybothinvolvethesamesingularproposition(oisafilmmaker),whereoisWoodyAllenhimself.Thesentencesinvolveanindexicalelementwithdifferentvalues.Mysaying,“Iamhungry”,andyoursaying,“Iamnothungry”,isperfectlyconsistent,becausethoseutterancesdonotexpressapropositionanditsnegation,asthedifferentusesof“I”expressdifferentcontents.Thecontextualistclaimsthatasimilar,althoughlessovert,phenomenonoccurswithutterancesof(7)and(9).Let’sbeginwiththefirststrategy.Muchofwhatwecommunicatebyourutterancesisindirect.Wearerarelyfullydirect,letalonefullyexplicit,abouteverythingwemean.Wesayjustenoughtogetourmeaningacross,relyingonouraudiencetoinferandfillintherest.Youaskme,aswearewatchinganNBAchampionshipgame,whichteamIamcheeringfor.Irespond,“IliveinLA”.IhavecommunicatedthatIamcheeringfortheLakers.ButIhaven’tsaidasmuch.WhatIsaidwasthatIliveinLA.IrelyuponyoutoinferthatIamcheeringfortheLakers,usingyourbackgroundknowledgethatpeoplelivinginLAtendtobeLakersfans.ThisinferencewasdrivenbythefactthatIwouldbeunresponsivetoyourquestionifImerelymeantthatIliveinLA,asyoudidn’taskmeaboutthat,andyetIamclearlyintendingyoutotakemeasbeingresponsive.ThisisaclearcaseofthephenomenonH.P.Gricestudiedinhisworkonconversationalimplicatures.(SeeGrice1975,1978,1981.)Accordingtoproponentsofthefirstreferentialiststrategy,asimilarstoryholdsforourusesofpropositionalattitude-ascribingsentences.(ProponentsofthisviewincludeBealer1982,§39;Berg1983;Braun1998,2000,2001a,2001b;McKay1981,1991;Nelson2002,2005;Reddam1982;Richard1983,1987;Salmon1986,1989,1995a;Soames1987,1989,1995;andTye1978.Bealerhaschangedhisview;seehis1993.Richardhasalsochangedhisviewin1989.Soameshasalsomodifiedhisview,althoughlessdrasticallythantheotherthree,inhis2002.)Althoughallutterancesof(7)and(11)saythesamething,typicalutterancesofthosesentenceswillcommunicatedistinctinformation.Onfirstpasswecansaythatthisisbecauseanutteranceofapropositionalattitude-ascribingsentencesuggestssomethingaboutthewayinwhichtheagentholdstheattitudethatshedoes,suggesting,inthecaseofanutteranceof(11),thattheagentwouldexpressherbeliefbeingreportedbyuttering(12),theembeddedsentence.So,onthisview,while(7)and(11)bothreportthatJanestandsinthebeliefrelationtothesingularproposition(oisafilmmaker),atypicalutteranceof(7)suggeststhatshewouldexpressthisbeliefbyutteringthesentence“WoodyAllenisafilmmaker”,whichistrue,whileatypicalutteranceof(11)suggeststhatshewouldexpressthisbeliefbyutteringthesentence“AllenKonigsbergisafilmmaker”,whichisfalse.Wesawabovethattheintuitivetruthofapropositionalattitude-ascribingsentencedoesnottracktheattitudetheagenttakestowardstheverysentenceembeddedinthatpropositionalattitude-ascribingsentence.Toaccommodate,1409781472578235_txt_print.indd14008/08/201415:46\nIntensionalContextstheproponentofthefirstreferentialiststrategyneedsamoreplausiblemetaphysicsofbelief.JohnPerrydistinguishedwhatanagentbelieves(thebeliefcontent)andhowshebelievesit(thebeliefstate).Perry(1977)arguedagainsttheFregean’sidentificationofbeliefcontents,thewhatofbelief,withbeliefstates,thehowofbelief.Perryarguedthatanadequateaccountofindexicalsanddemonstra-tivesinattitudesrequiresdistinguishingthesetwoaspectsofactsofthinkingandrecognizingthattheycanvaryindependently.OnPerry’sview,webelievecontentsbybeinginbeliefstateswhileembeddedinagivenenvironment.Wecanbelievedifferentcontentsindifferentenvironmentsbybeinginthesamebeliefstate–forexample,whenIissueathoughtIwouldexpressbyuttering“TodayisTuesday”onTuesday,Ientertainonethoughtcontent,andanotheronWednesday,eventhoughIaminthesametypeofbeliefstate–andwecanbelievethesamecontentindifferentenvironmentsbybeinginthedifferentbeliefstates–forexample,IentertainthesamecontentIentertainedonTuesdaybyissuingonWednesdayathoughtIwouldexpressbyuttering“YesterdaywasTuesday”.Thesedifferencesinenvironmentdonotneedtobeconceptualizedbytheagentforthemtomakeadifferencetohercognitivestates.Perry(1977)identifiedbeliefstates(whathetherecalled“senses”)withwhathecalledrolesandwhatKaplancalledcharacter.Inlaterwork(seeinparticularhis1980,1990,1997,and1998),Perrygivesthisidentificationupandindividuatesbeliefstatesintermsoftheirnarrowfunctionalrole,intermsoftherelationshipstoothercognitivestates,perceptionsandactionexecutions.Whilethedetailsneedtobefleshedout,wecanemploythisframeworktosaythatutterancesofpropositionalattitude-ascribingsentencespragmaticallyconveyinformationaboutthetypeofbeliefstatetheagentisinbyvirtueofwhatshebelieveswhatshebelieves.Atypicalutteranceof(7),then,pragmati-callyconveysinformationconcerningthebeliefstateJaneisininvirtueofwhichshestandsinthebeliefrelationtothesingularpropositionthatWoodyAllenisafilmmaker.Atypicalutteranceof(11)pragmaticallyconveysdifferentinformationconcerningJane’sbeliefstateinvirtueofwhichshestandsinthebeliefrelationtothatsamesingularproposition.Thispragmaticallyimpli-catedinformationcandifferintruthvalueandsopromisestoaccountfortheintuitionthat(7)istrueand(11)false,eventhoughthosesentencesalwaysexpressthesamepropositioninanycontext.Manyfinditunacceptabletorejecttheintuitionthat(7)istrueand(11)isfalse.Thereferentialist,however,canacceptthatintuition.Whattherefer-entialistmustrejectisthat(7)and(11)involveascribingtoJanebeliefswithdifferentcontents.Buttheremaybeotherdifferencesbetween(7)and(11)thatthereferentialistcanappealto.Namely,thereferentialistcanclaimthat(7)and(11)specifydistinctbeliefstates,usingtheterminologyintroducedabove,and1419781472578235_txt_print.indd14108/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagethatthatdifferenceaffectsthesemanticcontentof(7)and(11).Thisconstitutesthesecondreferentialiststrategy;whatIshallcallthecontextualiststrategy.Beforedevelopingthisstrategyfurther,let’spausetocontrastitwithwhathascomebefore.Asaneutralterminology,let’ssaythat(7)indicatesadistinctbeliefstatefrom(11)invirtueofwhichweintuitthat(7)istrueand(11)false.TheFregeanmaintainsthatthisdifferencemapsontoadifferenceincontent.Thereferentialist,ontheotherhand,maintainsthatthereisnodifferenceincontentbutonlyadifferenceinthebeliefstatebywayofwhichtheagentbelievesthecontentshedoes.Thecontextualreferentialistmaintainsthatthisdifferenceinindicatedbeliefstateaffectsthetruthof(7)and(11).Thatis,forthecontextualreferentialist,thisdifferenceissemanticallyrelevant,aspropo-sitionalattitude-ascribingsentencesaresemanticallysensitivetotheindicatedstate.Thepragmaticreferentialist,ontheotherhand,thinksthatthisdifferenceisnotsemanticallyrelevant.Propositionalattitude-ascribingsentencesaresemanticallyinsensitivetotheindicatedstate,althoughinformationabouttheindicatedstateisoftenpragmaticallycommunicatedassomekindofimplicature.Let’sreturntodevelopingthecontextualiststrategy.IbeginwithCrimminsandPerry’sversionofthestrategy.(SeeCrimminsandPerry1989,Crimmins1992a,1992b,1995a,1995b.)CrimminsandPerryarguethatpropositionalattitudereportsinvolve“unarticulatedconstituents”thatconcernhowthesubjectofthereportbelieveswhatshe(allegedly)believes.Atypicalutteranceof(7)expressesatruthwhileatypicalutteranceof(11)expressesafalsehood.ThefirstinvolvesimplicitreferencetoonewayJanehasofthinkingofWoodyAllenandthesecondtoadistinctwayJanehasofthinkingofWoodyAllen.BecauseJanebelievesthesingularproposition(WoodyAllenisafilmmaker)inthefirstwayandnotthesecond,thetwosentenceswilltypicallyexpressdifferentpropositionsthatdivergeintruthvalue.Extra-linguisticcontextdetermineswhatwayofgraspingisimplicitlyreferredtobyanutteranceofapropositionalattitude-ascribingsentence.Thecomplementclause(andnotjustthereferentsorsemanticcontentsoftheexpressionsofthecomplementclause)aretypicallyrelevanttowhatwayofgraspingisimplicitlyreferredto.So,althoughsubstitutionofco-referringnamesdoesnotaffectthepropositionthepropositionalattitude-ascribingsentenceclaimsthebelievertobelieve,asitdoesontheFregeanview,insomecasesitaffectswhatwayofgraspingisimplicitlyreferredtoandhenceiscapableofaffectingtruthvalueofapropositionalattitude-ascribingsentence.CrimminsandPerry’sviewhasbeensubjecttocriticism.(Forasampling,seeBach1993,Clapp1995,Reimer1995,Richard1993,Rieber1995,andSaul1992.Crimminsrespondstosomeoftheseobjectionsinhis1992and1995.SchifferobjectstoCrimminsandPerry’sviewinhis1992and1994onthebasisthat“believes”andotherpropositionalattitudeverbsdonotexpress1429781472578235_txt_print.indd14208/08/201415:46\nIntensionalContextsthree-placerelations,astheirviewrequires.LudlowrespondstoSchiffer’s“logicalform”objectioninhis1995and1996.SchifferrespondstoLudlowinhis1996.)Ishallherepresentonlyoneobjection,inordertomotivateanotherversionofthesecondreferentialiststrategytobediscussedbelow.Althoughtheissuesarecomplexandquicklydrawusintogeneralanddelicateissuesinthephilosophyoflanguage,manyaredrawntowhatwecancallaprincipleoflinguisticconstraint,accordingtowhichcontextualsupple-mentationofwhat’ssaidisalwaystraceabletosomesyntacticelement.(See,forexample,Stanley2000,2002foradefenceofthisconstraint.Recanati2002andCarston2002defendunarticulatedconstituents.)Thisprincipleisincom-patiblewiththeexistenceofunarticulatedconstituents.Onereasontoaccepttheprincipleisthatithelpstodelimitinformationthatisconveyedbyanutterancebutnotpartofwhatissaidbythesentenceutteredinawaythatsolves,inaprincipledway,anovergenerationproblemthatthreatenstheoriesviolatinglinguisticconstraint.Whileonecan,giventhecorrectconversationalsetting,usethesentence“2+2=4”tocommunicatethatoneisgoingtothestore,thesentenceitselfdoesnotsaythat.Aplausibleexplanationofwhyisthatnothinginthesentencecontributestherelevantcomponentsofthatcommuni-catedpieceofinformation.CrimminsandPerry’sviewisinconsistentwiththisprinciple,astheyarguethat“believes”issyntacticallytwo-placeeventhoughitexpressesathree-placerelation.Thismotivatesasearchforanaccountaccordingtowhichtheinformationencodedbyutterancesofbeliefsentencesissensitivetohowthesubjectofthereportgraspstheproposition,whileinsistingthatthepropositiongraspedisjustasingularproposition,withoutmakingwaysofgraspingunarticulatedconstituents.WecanbuilduptosuchaviewbylookingataviewdevelopedbyMarkRichard1990,1993,1995.LikeCrimminsandPerry,Richardoffersasemanticsofpropositionalattitudereportingsentencesthattreatsthemascontext-sensitive,is(quasi-)referentialist(Ireturntothisbelow),andyetblocksintersubstitutionofco-referringnameswithinthescopeofpropositionalattitudeverbs.UnlikeCrimminsandPerry,however,Richarddoesnotclaimthat“believe”(anditsfellowpropositionalattitudeverbs)expressesathree-placerelationanddoesnotappealtounarticulatedconstituents.Richard’sviewisthusconsistentwiththeprincipleoflinguisticconstraint.ThekeytoRichard’sviewishisclaimthat“believe”iscontext-sensitive,expressingdifferentrelationsindifferentcontexts.AccordingtoRichard,sentencesexpresswhathecallsRussellianannotatedmatrices(RAMs).Thesearerepresentedastuplesofpairsoflinguisticexpres-sionsandtheir“Russellianinterpretations”.(Richardnotesthat,strictlyspeaking,wedon’twantalinguisticexpressionbutrathersomekindofinternalrepresentation,akintoCrimminsandPerry’sideasandnotions.)Thepairsarecalledannotations.RAMsserveasboththesemanticcontentsofsentencesand1439781472578235_txt_print.indd14308/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagethecontentsofbeliefs.AccordingtoRichard’saccount,(7),forexample,istruerelativetosomecontextcandaworldofevaluationwjustincaseJanehasathoughtinwthatisproperlyrepresented(relativetothestandardsineffectinc)bythesentence“WoodyAllenisafilmmaker”.Theconversationalsettinggeneratesrestrictionsonwhatexpressionsarepropertranslationsoftherepre-sentationsoftheallegedbeliever.Richardcallstheserestrictionscorrelationfunctions.Correlationfunctionsmapannotations(usedbythebeliefreporter)onto(setsof)annotations(employedintheallegedbeliever’sbeliefs).Astherearecontextsinwhich“WoodyAllenisafilmmaker”and“AllenKonigsbergisafilmmaker”areassociatedwithdifferentrestrictions,wecannotsubstitute“AllenKonigsberg”for“WoodyAllen”in(7)andpreservetruth.Janemaywellhaveathoughtthatcanbeappropriatelyrepresented,giventherestrictionsinplaceinthecontext,bythefirstsentencebutnotthesecond.So,thereisnoprobleminsayingthat(7)istruewhile(11)isfalse.Richard’sviewisnotpurelyreferentialist.Thisisbecause,onhisview,thecontentsofbeliefsarenotsingularpropositions.Instead,thecontentofthebeliefthatWoodyAllenisafilmmakercontainsthepairofWoodyAllenhimselfandsomethingthatplaystheroleofamodeofpresentation–theannotation.ButRichard’sviewcanbealteredtoretainitscorefeatureswhilebeingreferen-tialist,treatingthecontentsofbeliefs(andthecontentsofsentencescontainingpropernames)assingularpropositions.TheresultingviewisinmanywayssuperiortoRichard’sown.BeginbyassumingthePerry-inspiredmetaphysicsofbeliefsketchedearlier,inwhichwedistinguishbeliefcontentsfrombeliefstates.WecanthenconceiveofRichardiancorrelationfunctionsasmapsfromcomplementclausestobeliefstates,preservingcontent.Thecontentofboth(12)and(13)isthesingularproposition(WoodyAllenisafilmmaker).Janebelievesthatpropositioninvirtueofbeinginabeliefstaterelatedto“WoodyAllenisafilmmaker”butnot“AllenKonigsbergisafilmmaker”.So,inabelief-ascribingcontextinwhichthecomplementclausesareintendedtorepresentthewaybywhichtheagentbelieveswhatsheisclaimedtobelieve,(7)istrueand(11)isfalse.Theviewpromisestoofferamoreplausibleaccountofwhentwoagents“believethesamething”,whichRichard’sownfine-grainedaccountofthecontentsofattitudescannot,asourintuitionsaboutwhentwoagentscountasbelievingthesamethingareofteninsensitivetodifferencesinhowagentsbelievewhattheybelieve.AnyviewlikeRichard’sfacesaseriousproblem:Theproblemofconflictingrestrictions.(SeeSider1995,Soames1995.)Tobuilduptotheobjection,considercasesinwhichaspeakerisascribingbeliefstodifferentagents.Itmaybecommongroundthatthoseagentsconceiveofmattersdifferently.Insuchacase,thespeakermayintendthesamecomplementclausetorepresentdifferenttypesofbeliefstatesfordifferentagents.SupposethatBobandSusanarelookingatapictureofBillClinton.Bobmightthenusethecomplement1449781472578235_txt_print.indd14408/08/201415:46\nIntensionalContextsclause“thathe(pointingatthepicture)waspresident”torepresentafirst-personbeliefstatewhenascribingbeliefstoBillClintonandadifferenttypeofbeliefstatewhenascribingbeliefstoGeorgeBush.Afterall,wecanimagineFrege-stylecasesinwhichthetypeofbeliefstateBobintendstospecifymightbeessentialtothepointofhisspeech.Nowtheproblemofconflictingrestrictions.Supposethattheascriberisconfusedabouttheidentitiesofthesubjectofthereport.So,forexample,supposethatJaneisascribingbeliefstoWoodyAllenandAllenKonigsberg.Becauseofheridentityconfusions,shetakesherselftobeascribingbeliefstotwodifferentpeople,dependingonhowsheconceivesofWoodyAllen.Butthenwecanimaginecircumstances,similartotheoneswithBobandSusandescribedabove,inwhichherintentionsgenerateconflictingrestrictionsonasinglecomplementclause.So,forexample,Janemightintend“he”,pointingataframeshotfromAnnieHall,in“thathelivesinNewYork”torepresentafirst-personstatewhenascribingbeliefstoWoodyAllenandathird-personbeliefstatewhenascribingbeliefstoAllenKonigsberg.So,whenJanesays,“WoodyAllenbelievesthathelivesinNewYork”,sheintendstoconveythatWoodyAllenhasafirst-personbelieftotheeffectthatWoodyAllenlivesinNewYork,andwhenshesays,“AllenKonigsbergbelievesthathelivesinNewYork”,sheintendstoconveythatAllenKonigsberghasathird-personbelieftotheeffectthatWoodyAllenlivesinNewYork.ThisisproblematicforRichardbecausetheserestrictionsconflictandsothereisnocorrelationfunctionthatrespectsalltheoperativerestrictions.Asaresult,anybeliefattributionJanemightmakeinsuchaconversationalsettingtoWoodyAllen/AllenKonigsbergwithacomplementclausecontaining“he”assubjectisfalse.Even“WoodyAllenbelievesthatheisself-identical”isfalse.Thisisintuitivelyunacceptable.Richardrespondedtothisproblem(1995)byclaimingthat,whenthereareconflictingrestrictionsoperativeinacontext,arelevantsentenceistrueinthecontextjustincaseitistrueoneveryresolutionoftheconflict.Asthesentence“WoodyAllenbelievesthatheisself-identical”istrueonanyresolutionoftheconflict,thesentenceiscountedastrue,onthismodifiedview.Nelson2002,2005criticisesRichard’sresponse,arguingthatitstilldoesnotaccommodateallintuitivetruth-valuejudgments,asitstillcountsfalsesomeutterancesthatareintuitivelytrue.Thisisbecauseoneoftheconflictingrestrictionsmayclearlybetheoperativerestriction,asfaraswhatthespeakerintendstobeconveyingisconcerned.OneaimofRichard’sviewistopresentasemanticsthatcorre-spondstotheinformationintuitivelyconveyedbyanutteranceofasentence,whichinthesecasesistheinformationthespeakerintendstoconvey.Consider,forexample,ourcasefromabove.JaneintendstobeascribingbeliefstoWoodyAllen,whomsheconceivesofastheverypersoninthephotosheispointingat.Shesays,“WoodyAllenbelievesthathelivesinNewYork”,intendingtoascribeafirst-personbelieftoWoodyAllen.Letussupposethathe1459781472578235_txt_print.indd14508/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagehassuchabelief.Butletussupposethathedoesnothaveathird-personbelieftotheeffectthathelivesinNewYork.Then,asweareimaginingJane’scontexttobeoneinwhichconflictingrestrictionsareineffect,Richard’sresponsecounselsustoseewhetherornottheattributionistrueoneveryresolutionoftheconflict.Asitisnot,thesentenceisnottrue.Butthisdoesnotrespectintui-tionsregardingwhatJaneintendstoconvey,whichisthetrueinformationthatWoodyAllenhasafirst-personbelieftotheeffectthathelivesinNewYork.Thisproblemwillinfectanysemantictheorythataimstocorresponddirectlytotheinformationthatisintuitivelyconveyedbyutterancesofpropo-sitionalattitude-ascribingsentences.Thatisbecauseanysuchaccountmustdealwithcasesinwhichtheattributorisconfusedabouttheidentitiesofthosetowhomsheisascribingattitudes.Inthatcase,whatsheintendstoconveywillbesensitivetothemannerinwhichtheagentoftheascription(asopposedtotheobjectoftheattitude,inastandardFrege-stylecase)isconceived.Thenthesubject-positionoftheattitude-ascribingsentenceitself,asopposedtothesubjectpositionoftheembeddedsentenceinthecomplementclause,isnotopentosubstitutionsalvaveritate.Thatisanundesirableresult,asitisaviolationoftheprincipleofnon-contradiction.Theseconsiderationssuggestthatweshouldnotdemandoursemanticsofpropositionalattitude-ascribingsentencestocorrespondtotheinformationintuitivelyconveyedbyutterancesofthosesentences.Thisinturnoffersreasontopreferourfirstreferentialismstrategyabove,eventhoughitinvolvesabandoningdeeplyheldintuitionsconcerningthetruthofattitude-ascribingsentences.Propositionalattitudeascriptionsarebutoneinstanceofintensionalcontexts.Ourfocusherehasbeenonthem,shortchangingotherintensionalcontextssuchasmodalandtemporaloperators,becausetheyraisesomeofthemostperplexingandfar-reachingofproblems.Wehavesurveyedthemaincompetingstructuredpropositionalistviews.WehaveseenFregeanviews,accordingtowhichallthoughtaboutanobjectisunderamodeofpresentation,andsotherearedistinctthoughtsaboutsomeobjecttotheeffectthatitissuchandsuch,dependingonhowtheobjectisconceived.Wehaveseenthat,whilethisviewiswell-suitedtohandlestandardFregecases,theviewfacesseveraldifficulties.Whilesomedifferencesinhowanagentconceivesoftheobjectofherbeliefdoseemtomattertoourintuitivejudgmentsofthetruthvalueofapropositionalattitudeascription,asstandardFregecasesdemonstrate,otherdifferencesdonot,asiswitnessedbyindexicalbeliefsandsingleattributionstomultipleagents.Thereferentialistcutsbeliefcontentsmorecoarsely.Onthereferentialist’sview,foracoresetofbeliefs,differencesinwaysofconceivingoftheobjectofthebeliefareirrelevanttowhatisbelieved.ThedifficultiesfortheFregeanarevirtuesforthereferentialist.ButthereferentialistmustworktoofferanaccountofFregecases.Wesurveyedtworeferentialiststrategies1469781472578235_txt_print.indd14608/08/201415:46\nIntensionalContextsfordealingwiththisproblem.Accordingtothefirst,theintuitiveresponsetoFregecasesrestsonconflatingwhatismerelyconveyedwithwhatthesentenceutteredsays.Whileutterancesof(7)and(11)typicallyconveydifferentinfor-mation,thefirstbeingtrueandthesecondfalse,thesentencesalwayssaythesamething.Accordingtothesecondreferentialiststrategy,theintuitiveresponsetoFregecasesiscorrect,butdoesnotentailthatthecomplementclausesof(7)and(11)pickoutdifferentpropositions.Onthisview,thereismoretoanaccuratebelief-ascribingsentencethanmerelyspecifyingthepropo-sitionbelieved;itmustalsoaccuratelyrepresenthowtheagentbelieveswhatshebelieves.Wesurveyedseveraltechniquesforhowthereferentialistcandothis,fromappealingtohiddenindexicalsandunarticulatedconstituentstopostulatingcontextsensitivityintheattitudeverbitself,claimingthattheverbpicksoutdifferentrelationsindifferentcontexts.BibliographyBach,K.(1993),“Sometimesagreatnotion:AcriticalnoticeofMarkCrimmins’sTalkaboutBeliefs”,MindandLanguage8:431–41.—(1994),ThoughtandReference.NewYork:OxfordUniversityPress.—(1997),“Dobeliefreportsreportbeliefs?”,PacificPhilosophicalQuarterly78:214–41.Barwise,J.andPerry,J.(1983),SituationsandAttitudes.Cambridge,MA:TheMITPress.Bealer,G.(1982),QualityandConcept.Oxford:OxfordUniversityPress.—(1993),“AsolutiontoFrege’spuzzle”,PhilosophicalPerspectives7:17–60.Berg,J.(1983),PragmaticsandtheSemanticsofBelief.Ph.Ddissertation,UCLA.Braun,D.(1991),“Propernames,cognitivecontents,andbeliefs”,PhilosophicalStudies62:289–305.—(1998),“Understandingbeliefreports”,PhilosophicalReview107:555–95.—(2000),“Russellianismandpsychologicalgeneralizations”,Noûs34:203–36.—(2001a),“Russellianismandprediction”,PhilosophicalStudies105:59–105.—(2001b),“Russellianismandexplanation”,PhilosophicalPerspectives14:253–89.—(2006a),“Illogical,butrational”,Noûs40:376–9.—(2006b),“NowyouknowwhoHonOakYunis”,PhilosophicalIssues16:24–42.Burge,T.(1977a),“Kaplan,Quine,andsuspendedbelief”,PhilosophicalStudies31:197–203.—(1977b),“Derebelief”,JournalofPhilosophy74:338–62.—(1986),“OnDavdison’s‘Sayingthat’”,inE.LePore(ed.)TruthandInterpretation:PerspectivesonthePhilosophyofDonaldDavidson.Oxford:Blackwell,190–210.Carnap,R.(1958),MeaningandNecessity:AStudyinSemanticsandModalLogic,2nd,enlarged,paperbackedition.Chicago:TheUniversityofChicagoPress.(Originallypublishedin1947.)Carston,R.(2002),ThoughtsandUtterances:ThePragmaticsofExplicitCommunication.Oxford:Blackwell.Castañeda,H.-N.(1966),“‘He’:Astudyinthelogicofself-consciousness”,Ratio,8:130–47.1479781472578235_txt_print.indd14708/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguage—(1967),“Indicatorsandquasi-indicators”,AmericanPhilosophicalQuarterly4:85–100.Chisholm,R.(1981),TheFirstPerson,Minneapolis:UniversityofMinnesotaPress.Church,A.(1950),“OnCarnap’sanalysisofstatementsofassertionandbelief”,Analysis10:97–9.Clapp,L.(1995),“Howtobedirectandinnocent:AcriticismofCrimminsandPerry’stheoryofattitudeascriptions”,LinguisticsandPhilosophy18:529–65.—(2002),“Davidson’sprogramandinterpretedlogicalforms”,LinguisticsandPhilosophy25:261–97.Crimmins,M.(1992a),TalkaboutBeliefs.Cambridge,MA:TheMITPress.—(1992b),“Contextintheattitudes”,LinguisticsandPhilosophy14:185–98.—(1993),“So-labeledneo-Fregeanism”,PhilosophicalStudies69:265–79.—(1995a),“Notionalspecificity”,MindandLanguage10:464–77.—(1995b),“Contextuality,reflexivity,iteration,logic”,PhilosophicalPerspectives9:381–439.Crimmins,M.andPerry,J.(1989),“Theprinceandthephonebooth:Reportingpuzzlingbeliefs”,JournalofPhilosophy86:685–711.Davidson,D.(1967),“Truthandmeaning”,Synthese17:304–23.—(1968),“Onsayingthat”,Synthese19:130–46.Dummett,M.(1973),Frege:PhilosophyofLanguage.London:Duckworth.—(1981),TheInterpretationofFrege’sPhilosophy.London:Duckworth.—(1991),TheLogicalBasisofMetaphysics.Cambridge,MA:HarvardUniversityPress.Fitch,G.(1984),“Twoaspectsofbelief”,PhilosophyandPhenomenologicalResearch45:87–101.—(1987),NamingandBelieving.Dordrecht:Reidel.Forbes,G.(1989),LanguagesofPossibility.Cambridge:BasilBlackwell.—(1990),“TheindispensabilityofSinn”,PhilosophicalReview99:535–63.—(1993),“ReplytoMarks”,PhilosophicalStudies69:281–95.—(1996),“Substitutivityandthecoherenceofquantifyingin”,PhilosophicalReview105:337–72.—(2000),“Objectualattitudes”,LinguisticsandPhilosophy23:141–83.—(2002),“Intensionality”,SupplementtotheProceedingsoftheAristotelianSociety76:75–99.Frege,G.(1904),“CorrespondencewithRussell”,inN.SalmonandS.Soames(eds),PropositionsandAttitudes.Oxford:OxfordUniversityPress,p.56.—(1948),“Senseandreference”,PhilosophicalReview57:209–30.(Originallypublishedin1892.)—(1956),“Thethought:Alogicalinquiry”,Mind65:289–311.(Originallypublishedin1918.)Grice,H.P.(1975),“Logicandconversation”,inP.ColeandJ.Morgan(eds),SyntaxandSemantics3:41–58.(OriginallydeliveredasWilliamJameslecturesatHarvardUniversityin1967.)—(1978),“Furthernotesonlogicandconversation”,SyntaxandSemantics9:113–28.—(1981),“Presuppositionandconversationalimplicature”,inP.Cole(ed.)RadicalPragmatics,NewYork:AcademicPress,183–98.Higginbotham,J.(1991),“Beliefandlogicalform”,MindandLanguage6:344–69.1489781472578235_txt_print.indd14808/08/201415:46\nIntensionalContextsJeshion,R.(ed.)(2010),NewEssaysonSingularThought.Oxford:OxfordUniversityPress.Kaplan,D.(1969),“Quantifyingin”,Synthese19:178–214.—(1970),“WhatisRussell’stheoryofdescriptions?”,inW.Yourgrau,A.BreckandH.Bondi(eds),Physics,Logic,andHistory,NewYork:Plenum,277–88.—(1977/1989),“Demonstratives”,inJ.Almog,J.PerryandH.Wettstein(eds),ThemesfromKaplan,Oxford:OxfordUniversityPress,481–563.—(1986),“Opacity”,inL.HahnandP.Schilpp(eds),ThePhilosophyofW.V.Quine,LaSalle:OpenCourt,229–89.Kripke,S.(1979),“Apuzzleaboutbelief”,inA.Margalited.MeaningandUse,Dordrecht:Reidel,239–83.—(1980),NamingandNecessity,Oxford:Blackwell.(OriginallydeliveredasaseriesoflectureatPrincetonUniversityin1970.)Larson,R.andP.Ludlow.(1993),“Interpretedlogicalforms”,Synthese95:305–55.Larson,R.andG.Segal.(1995),KnowledgeofMeaning:SemanticValueandLogicalFormCambridge,MA:TheMITPress.Lepore,E.andLoewer,B.(1989),“Youcansaythatagain”,MidwestStudiesinPhilosophy14:338–56.Lewis,D.(1972),“Generalsemantics”,inD.DavidsonandG.Harman(eds),SemanticsofNaturalLanguage,Dordrecht:D.Reidel,169–218.—(1979),“Attitudesdedictoanddese”,PhilosophicalReview88:513–-43.—(1981),“WhatpuzzlingPierrebelieves”,AustralasianJournalofPhilosophy59:283–9.—(1986),OnthePluralityofWorlds.Oxford:BasilBlackwell.Ludlow,P.(1995),“Logicalformandthehiddenindexicaltheory:AreplytoSchiffer”,JournalofPhilosophy92:102–7.—(1996),“Theadicityof‘believes’andthehiddenindexicaltheory”,Analysis56:97–101.—(2000),“Interpretedlogicalforms,beliefattribution,andthedynamiclexicon”,inM.Jaszczolted.ThePragmaticsofPropositionalAttitudeReports,Oxford:ElsevierScience,31–42.McKay,T.(1981),“Onpropernamesinbeliefascriptions”,PhilosophicalStudies39:287–303.—(1986),“Hisburningpants”,NotreDameJournalofFormalLogic27:393–400.—(1991),“Representingderebeliefs”,LinguisticsandPhilosophy14:711–39.Nelson,M.(2002),“Descriptivismdefended”,Noûs36:408–36.—(2005),“Theproblemofpuzzlingpairs”,LinguisticsandPhilosophy28:319–50.O’Leary-Hawthorne,J.(1995),“Thebundletheoryofsubstanceandtheidentityofindiscernibles”,Analysis55:191–6Perry,J.(1977),“Fregeondemonstratives”,PhilosophicalReview86:474–97.(ContainedinPerry1993and2000.)—(1979),“Theproblemoftheessentialindexical”,Noûs13:3–21.(ContainedinPerry1993and2000.)—(1980),“Beliefandacceptance”,MidwestStudiesinPhilosophy5:533–42.(ContainedinPerry1993and2000.)—(1990),“Self-notions”,Logos11:17–31.—(1993),TheProblemoftheEssentialIndexical,Oxford:OxfordUniversityPress.1499781472578235_txt_print.indd14908/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguage—(1997),“RipVanWinkleandothercharacters”,TheEuropeanReviewofAnalyticalPhilosophy2:13–39.(ContainedinPerry2000.)—(1998),“MyselfandI”,inM.Stammed.PhilosophieinSynthetischerAbsicht,Stuttgart:Klett-Cotta,83–103.(ContainedinPerry2000.)—(2000),TheProblemoftheEssentialIndexical.Expandededition,Stanford:CSLI.—(2001),ReferenceandReflexivity,Stanford:CSLI.Plantinga,A.(1974),TheNatureofNecessity,Oxford:Clarendon.—(1978),“TheBoethiancompromise”,AmericanPhilosophicalQuarterly14:129–38.Quine,W.(1956),“Quantifiersandpropositionalattitudes”,JournalofPhilosophy53,177–87.—(1980),“Referenceandmodality”(2ndrevisedversion),inFromaLogicalPointofView.2ndedition,revisedprinting,NewYork:HarperandRow,139–49.(Originallypublishedin1953.)Recanati,F.(1993),DirectReference:FromLanguagetoThought.Oxford:Blackwell.Reddam,P.(1982),PragmaticsandtheLanguageofBelief.Ph.Ddissertation,USC.Reimer,M.(1995),“Adefenseofderebeliefreports”,MindandLanguage10:446–63.Richard,M.(1983),“Directreferenceandascriptionsofbelief”,JournalofPhilosophicalLogic12:425–52.—(1987),“Attitudeascriptions,semantictheory,andpragmaticevidence”,ProceedingsoftheAristotelianSociety87:243–62.—(1988),“TakingtheFregeanseriously”,inD.Austined.PhilosophicalAnalysis.Dordrecht:Kluwer,219–39.—(1989),“HowIsaywhatyouthink”,MidwestStudiesinPhilosophy14:317–37.—(1990),PropositionalAttitudes:AnEssayonThoughtsandHowWeAscribeThem.CambridgeUniversityPress.—(1993a),“Attitudesincontext”,LinguisticsandPhilosophy16:123–48.—(1993b),“Sense,necessityandbelief”,PhilosophicalStudies69:243–63.—(1993c),“BoërandLycan’sKnowingWho”,Noûs27:235–42.—(1995),“Defectivecontexts,accommodation,andnormalization”,CanadianJournalofPhilosophy25:551–70.—(1997),“Propositionalattitudes”,inB.HaleandC.Wright(eds.),ACompaniontoPhilosophyofLanguage.Oxford:Blackwell,197–226.—(2001),“SeekingaCentaur,adoringAdonis:Intensionaltransitivesandemptyterms”,MidwestStudiesinPhilosophy25:103–127.Russell,B.(1903),ThePrinciplesofMathematics.Cambridge:CambridgeUniversityPress).—(1904/1988),“CorrespondencewithFrege”,inN.SalmonandS.Soames(eds),PropositionsandAttitudes.Oxford:OxfordUniversityPress,57.—(1905),“Ondenoting”,Mind14:479–93.—(1910),“Knowledgebyacquaintanceandknowledgebydescription”,ProceedingsoftheAristotelianSociety11:108–28.—(1912),TheProblemsofPhilosophy,Oxford:OxfordUniversityPress.Sainsbury,M.(1993),“Russellonnamesandcommunication”,inA.D.IrvineandG.A.Wedeking(eds),RussellandAnalyticPhilosophy,Toronto:TheUniversityofTorontoPress,3–21.Salmon,N.(1981),ReferenceandEssence.Princeton,NJ:PrincetonUniversityPress.—(1985),“Reflexivity”,NotreDameJournalofFormalLogic27:401–29.1509781472578235_txt_print.indd15008/08/201415:46\nIntensionalContexts—(1986),Frege’sPuzzle.Cambridge,MA:TheMITPress.—(1989),“Illogicalbelief”,PhilosophicalPerspectives3:243–85.—(1992),“Reflectionsonreflexivity”,LinguisticsandPhilosophy14:53–63.—(1995a),“Beingoftwominds:Beliefwithdoubt”,Noûs29:1–20.—(1995b),“Relationalbelief”,inP.LeonardiandM.Santambrogio(eds),OnQuine:NewEssays,Cambridge:CambridgeUniversityPress,206–28.—(1997),“Isderebeliefreducibletodedicto?”,inA.Kazmied.MeaningandReference(TheCanadianJournalofPhilosophySupplementaryVolume23),85–110.—(2006),“Theresilienceofillogicalbelief”,Noûs40:369–75.Saul,J.(1993),“Stillanattitudeproblem”,LinguisticsandPhilosophy16:423–35.—(1996),TheProblemwithAttitudes.Ph.Ddissertation,PrincetonUniversity.—(1997),“Substitutionandsimplesentences”,Analysis57:102–8.—(1998),“Thepragmaticsofattitudeascription”,PhilosophicalStudies92:363–89.—(1999a),“Theroadtohell:Intentionsandpropositionalattitudeascription”,MindandLanguage14:356–75.—(1999b),“Thebestofintentions:Ignorance,idiosyncrasy,andbeliefreporting”,CanadianJournalofPhilosophy29:29–48.Schiffer,S.(1977),“Namingandknowing”,MidwestStudiesinPhilosophy2:28–41.—(1987a),“The‘Fido’-Fidotheoryofbelief”,PhilosophicalPerspectives1:455–80.—(1987b),RemnantsofMeaning.Cambridge,MA:TheMITPress.—(1990),“Themode-of-presentationproblem”,inC.AndersonandJ.Owens(eds),PropositionalAttitudes,Stanford:CSLI.—(1992),“Beliefascription”,JournalofPhilosophy89:499–521.—(1994),“Aparadoxofmeaning”,Noûs28:279–24.—(1996),“Thehidden-indexicaltheory’slogical-formproblem:Arejoinder”,Analysis56:92–7.—(2006),“Aproblemforadirect-referencetheoryofbeliefreports”,Noûs40:361–8.Sider,T.(1995),“ThreeproblemsforRichard’stheoryofbeliefascriptions”,CanadianJournalofPhilosophy25:487–513.Soames,S.(1985),“Lostinnocence”,LinguisticsandPhilosophy8:59–72.—(1987),“Substitutivity”,inJ.Thomsoned.OnBeingandSaying,Cambridge,MA:TheMITPress,99–132.—(1989a),“Directreference,propositionalattitudesandsemanticcontent”,PhilosophicalTopics14:44–87.—(1989b),“Semanticsandsemanticcompetence”,PhilosophicalPerspectives3:575–96.—(1992),“Truth,meaning,andunderstanding”,PhilosophicalStudies65:17–35.—(1995),“Beyondsingularpropositions”,CanadianJournalofPhilosophy25:514–49.—(2002),BeyondRigidity:TheUnfinishedSemanticAgendaofNamingandNecessity,NewYork:OxfordUniversityPress.—(2003),PhilosophicalAnalysisintheTwentiethCentury,VolumeII:TheAgeofMeaning.Princeton,NJ:PrincetonUniversityPress.Sosa,D.(1996),“Theimportofthepuzzleofbelief”,PhilosophicalReview105:373–434.—(2001),“RigidityinthescopeofRussell’stheory”,Noûs35:1–38.Sosa,E.(1970),“Propositionalattitudesdedictoanddere”,JournalofPhilosophy67:883–96.1519781472578235_txt_print.indd15108/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageStalnaker,R.(1981),“Indexicalbelief”,Synthese49:129–42.—(1984),Inquiry.Cambridge,MA:TheMITPress.—(1987),“Semanticsforbelief”,PhilosophicalTopics14:177–90.—(1988),“Beliefattributionandcontext”,inR.GrimmandD.Merril(eds),ContentsofThought,Tucson:UniversityofArizona,146–81.Stanley,J.(1997a),“Rigidityandcontent”,inR.Hecked.Language,Thought,andLogic:EssaysinhonourofMichaelDummett,Oxford:OxfordUniversityPress,131–56.—(1997b),“Namesandrigiddesignation”,inB.HaleandC.Wright(eds),CompaniontothePhilosophyofLanguage,Oxford:Blackwell,555–86.—(2002),“Modalityandwhatissaid”,PhilosophicalPerspectives16:321–44.Strawson,P.F.(1959),Individuals:Anessayindescriptivemetaphysics.London:Methuen.Taschek,W.(1988),“WouldaFregeanbepuzzledbyPierre?”,Mind97:99–104.—(1992),“Frege’spuzzle,sense,andinformationcontent”,Mind101:767–92.—(1995a),“Belief,substitution,andlogicalstructure”,Noûs29:71–95.—(1995b),“Onbeliefcontentandthat-clauses”,MindandLanguage10:274–98.—(1997),“PuttingPierreandPeterincontext:Onascribingbeliefs”,inD.Jurtronic-Tihomiroviced.TheMariborPapersinNaturalizedSemantics,Maribor:Maribor.—(1998),“Onascribingbeliefs:Contentincontext”,JournalofPhilosophy95:323–53.1529781472578235_txt_print.indd15208/08/201415:46\n7ContextDependenceKentBachAllsortsofthingsarecontext-dependentinonewayoranother.Whatitisappropriatetowear,togiveortorevealdependsonthecontext.Whetherornotitisallrighttolie,harm,orevenkilldependsonthecontext.Ifyougooglethephrase“dependsonthecontext”,you’llgetseveralhundredmillionresults.Thischapteraimstonarrowthatdown.Inthiscontextthetopiciscontextdependenceinlanguageanditsuse.Itiscommonlyobservedthatthesamesentencecanbeusedtoconveydifferentthingsindifferentcontexts.Thatiswhypeoplecomplainwhensomethingtheysayis“takenoutofcontext”andinsistthatitbe“putintocontext”,because“contextmakesitclear”whattheymeant.Indeed,itispracti-callyaplatitudethatwhataspeakermeansinutteringacertainsentence,aswellashowheraudienceunderstandsher,“dependsonthecontext”.Butjustwhatdoesthatamountto,andtowhatextentisittrue?Philosophersandlinguistsoftensaythatcertainwords(andsentencescontainingthem)arecontext-sensitive,thatwhattheyexpressiscontext-dependent,asifitisperfectlyobviouswhatcontextdependenceis.Itisnot.Sowewillneedtoaskwhatcontextis,whatdependsonit,andwhatthisdependenceinvolves.Answerstothesequestionsarenotstraightforward.Itturnsoutthatthereismorethanonekindofcontextandthatdifferentsortsofthingsdependoneach.Atleasttheyseemto,for,aswewillsee,muchofwhatpassesforcontextdependenceisreallysomethingelse.Lookingatwhatgoesoninspecificcasessuggeststhatmuchofwhatisdoneincontextisnotdonebycontext.Whyshouldwelookintothesequestions?Therearetwomaintheoreticalreasons,thoughwewillnotdwellonthem.First,contextsensitivityposesachallengetothecommonviewthatthemeaningofasentencedeterminesitstruth-condition.Thisistheassumptionunderlyingthewidespreadviewthatthegoalofsemanticsistogiveasystematictheoryofthetruth-conditionsofsentences.However,atruth-conditionalsemanticshastoreckonwiththefactthatthesemanticcontentsofcontext-sensitiveexpressionsvaryfromonecontextofutterancetoanother,andthatispossibleonlyifthemeaningsofcontext-sensitiveexpressionsdeterminehowtheirsemanticcontentsvarywiththecontext.Onecomplication,aswewillsee,isthatmanysentencesdonotseemtohavetruth-conditions,evenrelativetocontexts.1539781472578235_txt_print.indd15308/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageThesecondworry,relatedtothefirst,isthatcontextsensitivitymightunderminetheprincipleofcompositionality.Thisisthecommonmethodo-logicalassumptionthatthesemanticpropertiesofcomplexexpressionsaredeterminedbythoseoftheirconstituentexpressionsandhowthesearerelatedsyntactically.Ifthisprincipledidnothold,soitisthought,wecouldnotunder-stand,muchlessknowinglyproduce,anyofthevirtuallyunlimitednumberofsentenceswehavenotpreviouslyencountered.Herethechallengeistoshowhowthecontentsofcomplexexpressionsaredeterminedbythecontentsoftheirpossiblycontext-sensitiveconstituents.Insomecases,thesemanticsofthecomplexexpressioncreatesproblemsforcompositionality.Toseethis,justcomparethemeaningsof“waterlily”and“tigerlily”orof“childabuse”and“drugabuse”.Therearemoredown-to-earthreasonsforinvestigatingcontextsensitivity.Ifthewordsinalanguageallhaduniqueanddeterminatemeanings(noambiguityorvagueness)andfixedreferences(noindexicality),andifusinglanguageweresimplyamatterofputtingone’sthoughtsintowords,under-standinganutterancewouldmerelybeamatterofdecipheringwhateverwordsthespeakeruttered.Butlanguageandouruseofittocommunicatearenotasstraightforwardasthat.Someexpressions,mostobviouslypronouns,suchas“I”,“they”and“this”,andtemporaltermssuchas“today”and“nextweek”,donothavefixedreferences.Forexample,whenIuse“I”itreferstome,butwhenyouuseititreferstoyou.Moreover,weoftenspeakinexplicitly,non-literally,orindirectly,andineachcasewhatwemeanisdistinctfromwhatcanbepredictedfromthemeaningsoftheexpressionsweutter.Wecanleavesomethingoutbutstillmeanit,useawordorphrasefiguratively,ormeansomethinginadditiontowhatwesay.Wecanevendoallthreeatonce.Here’sanexample.Supposeyouhaveafriendwhoseneighbouriswellknowntobeanincompetentdoctor.Yourfriendcomplainsaboutachroniccough,andyousay,“Youshouldseesomeonetoday,butnotthatgeniusnextdoor”.Youmeant,let’sassume,thatyourfriendshouldseeadoctorthatverydayforadiagnosis,butnottheincompetentneighbour.Youprobablymeantalsothathecouldwellhaveaseriousmedicalcondition.Whatistheroleofcontextinthiscase?Itdoesnotdeterminewhatyoumeant.Yourcommunicativeintentiondeterminesthat.Whatcontextdoesdoisprovideinformationthatyourfriendcoulduse,presumingyouintendhimto,tofigureoutwhatyoumeant.Inthatconnectioncontextplaysamerelyevidentialrole.However,itseemsthatcontextcanplayamoredirect,semanticrole,atleastinconnectionwithsuchwordsas“I”and“today”.Theyarecontext-sensitive,inthattheircontents,whattheycontributetothecontentsofsentencesinwhichtheyoccur,dependonthecontextinwhichtheyareused.Afterdiscussinghowthisissointheseparadigmcases,wewilllookatawiderangeofothersortsofexpressionsthathavealsobeenclaimedtobe1549781472578235_txt_print.indd15408/08/201415:46\nContextDependencecontext-sensitive.Toevaluatesuchclaims,whichfallunderthegeneralrubricofcontextualism,wewillneedtoaskwhetheritisthecontentoftheexpressionitselfthatvariesfromonecontextofutterancetoanotherand,crucially,whetheritdoessobecauseoffeaturesofthecontext.Thisquestionrarelygetsasked.Rather,contextualiststendtousephrasessuchas“context-dependent”and“context-sensitive”freelyanduncritically,asifitisobviousthatwhatisdoneincontextisdonebycontextandequallyobvioushow.Theprimaryaimofthischapteristoencouragethereadernottotakethe“obvious”forgranted.1.Introduction:Twokindsofcontext,tworolesforcontextOnethingshouldbeunderstoodfromthestart.Toholdthatcertaintermsarecontext-sensitiveisnottodenythattheyhavedictionarymeanings.Theclaimisnotthattheirmeaningsvarywiththecontext.Itisthattheirstandingmeaningsdeterminetheircontentsasafunctionofcontextsoftheiruse.Afterall,wewouldn’tlookwordsupinthedictionaryiftheydidn’thave(fairly)stablemeanings.Tobesure,werecognizethatagreatmanyareambiguousandexpecttheirdictionaryentriestogivetheirdistinctmeaningsseparately.Butambiguityisnotcontextdependence.Taketheambiguouswords“press”and“suit”astheyoccurinthesentence“Atailorpressedasuitincourt”.Itmightseeminacertaincontextthatthespeakerisusingittomeanthatthetailorsuedsomeone,butthisdoesnotshowthat“press”and“suit”arecontext-sensitive.Forshecouldinstead,howeverimprobably,haveusedthesentencetomeanthatthetailorironedasuitofclothesinacourtroom(forconvenience,Iwillgenerallyuse“she”forthespeakerand“he”forthehearer).Contextcannotpreventthat,althoughshemightnotbeunderstoodunlesssheclarifiedhowshewasusingherwords.Insomecases,thecontextmayleavethatopen,aswiththissentenceinaletterofrecommendation:“Ienthusiasticallyrecommendthiscandidatewithnoqualificationswhatsoever”.Thissentence,asitmightoccurinaletterofrecommendation,illustratesadifferentpoint:(1)MrTullyisaclearandforcefulspeaker.Iftheletterisforthejobofradioannouncer,thewritersurelymeansthatMrTullyhasanimportantqualificationforthejob.Ontheotherhand,iftheletterisforaradiologistjob,thewritercouldwelluse(1)toimplythatMrTullyisapoorcandidate.Inneithercasedoesthecontextdeterminewhatthewritermeansinuttering(1)–thatisamatterofherintention–butitdoesdeterminewhatthereadercouldreasonablytakehertomean.Therearedifferentthingsaspeakermightbedoinginutteringthissentence:1559781472578235_txt_print.indd15508/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguage(2)Thecopswillbreakuptheparty.Supposethere’saboisterouspartygoingonverylateatnightinanotherwisequietneighbourhood.Onecanimaginedifferentcircumstancesinwhich(2)couldbeusedto,andbetakenbythehearerto,makeaprediction,issueathreat,makeapromise,issueawarningorgiveanorder.Forexample,anearbyresident,confidentintheimpatienceofanotherneighbourandintherespon-sivenessofthepolice,mightutter(2)toherhusbandtomakeaprediction.Thatimpatientneighbour,tiredofthenoise,mightcallthepolice,andtherespondingofficermightuse(2)topromisethatthepolicewillbreakuptheparty.Andiftheimpatientneighbourhappenstobethemayor,shemightcallthepolicechiefanduse(2)togiveanorder.Theselastexamplesfallunderthedomainofpragmatics(seeChapter8),andtheissuestheyraisewillnotbepursuedhere.However,itisimportanttopointoutthattheroleofcontextinthesecasesisdifferentfromtheroleitplayswithexpressionsthataresemanticallysensitivetocontext.Indeed,twodifferentkindsofcontextareinvolved.Narrowcontextconsistsofmattersofobjectivefacttowhichthedeterminationofthesemanticcontentsofcertainexpressionsaresensitive.Broadcontextistheconversationalsetting,themutualcognitivecontextorsalientcommonground.Itincludesthecurrentstateoftheconversation(whathasjustbeensaid,whathasjustbeenreferredto,etc.),thephysicalsetting(ifthepartiesarefacetoface),salientpersonalknowledge,andrelevantbroadercommonknowledge.Playingapragmaticroleincommu-nication(onwhetheritcouldalsoplayasemanticrole,seeSection3.1),broadcontextincludestheinformationthatthespeakerexploitstomakewhatshemeansevidenttothehearerand,ifcommunicationistosucceed,thatthehearertakesintoaccount,ontheassumptionthatheisintendedto,tofigureoutwhatthespeakermeans.Thisdifferenceisoftenoverlooked.Forexample,Stalnaker,althoughhecontrastspropositionssemanticallyexpressedby(context-sensitive)sentencesandpragmaticeffectsofspeakers’utterances,treatssemanticandpragmaticcontextsensitivityonapar:Howshouldacontextbedefined?Thisdependsonwhatelementsofthesituationsinwhichdiscoursetakesplacearerelevanttodeterminingwhatpropositionsareexpressedbycontext-dependentsentencesandtoexplainingtheeffectsofvariouskindsofspeechacts.Themostimportantelementofacontext,Isuggest,isthecommonknowledge,orpresumedcommonknowledgeandcommonassumptionsoftheparticipantsinthediscourse.Aspeakerinevi-tablytakescertaininformationforgrantedwhenhespeaksasthecommongroundoftheparticipantsintheconversation.Itisthisinformationwhichhecanuseasaresourceforthecommunicationoffurtherinformation,and1569781472578235_txt_print.indd15608/08/201415:46\nContextDependenceagainstwhichhewillexpecthisspeechactstobeunderstood.(Stalnaker1999:67;myemphasis)However,beingusedasaresourceforcommunicationisverydifferentfromdeterminingwhatpropositionsareexpressed.Thepragmaticroleofcontextistoprovideabasis,asintendedbythespeaker,forthehearertofigureoutwhatthespeakermeans.Aspeaker’scommunicativeintentionisreasonabletotheextentthehearercanbeexpectedtorecognizeitonthebasisofwhatshesaysandthefactthatshesaysitinthecontext.However,itdoesnotliterallydetermine,inthesenseoffixing,whatthespeakermeans.Rather,itenablesthehearertodetermine,inthesenseofascertaining,whatshemeans.Thischaracterizationsuggeststhatbroadcontextimposesarationalconstraintonthespeaker’scommunicativeintention.However,itwouldbemisleadingtosaythatthisconstraintdetermineswhataspeakershouldintend,giventhatshesaysacertainthing.Afterall,aspeakercantrytocommunicateanythingshepleases.Theconstraint,rather,isthis:givenwhatsheintendstocommunicate,sheshouldsaysomethingthat,evenifsheisnotbeingfullyexplicit,makesevidenttothehearerwhatsheaimstoconvey.Inordertounderstandthespeaker,thehearerreliesontheassumptionthatsheintendshimtobeabletofigureoutwhatthisis,andtodothathemusttakeintoaccountcontextualfactsthathecanreasonablytakeherasintendinghimtotakeintoaccount(seeGrice1989,chs2and14,andBachandHarnish1979,chs1and5).Intheremainderofthischapterwetakeupavarietyofexamplesofexpres-sionsthathavebeenthoughttobecontext-sensitive.Manyturnoutnottobe.Insomecases,thisisbecausethesourceofthecontextualvariationisnottheexpressionitselfbuttheopenrangeofpossiblewaysinwhichspeakerscanusesimplesentencescontainingthatexpression.InSection2wewilldiscussso-calledindexicals,someofwhich,suchas“I”and“today”,doseemtobecontext-sensitiveinastraightforwardway,andcomparethemwithdemon-stratives,whosecontextuallyvariableusesseemtodependonthespeaker’sintentioninusingtheexpression,notonthecontextinwhichtheexpressionisused.Wewillalsobrieflyconsidertheclaimthatcertainsentencesthatwouldotherwisenothavetruth-conditionscontain“hiddenindexicals”,whosevaluesaresomehowprovidedbycontextandtherebycompletethepropositionbeingexpressed.Then,inSection3,wewilldiscussvariousissuesaboutspeakerintentions,contextsofutterance,andtherelationshipbetweenthetwo.Dointentionsreallytrumpcontexts,orisintentionpartofcontext?Finally,inSection4,wewilltakeupmoreexamples,mostlyinvolvingvariouskindsofadjectivesthathavebeenthoughttobecontext-sensitive.Oneaimofthischapteristoshowthatmanysupposedcasesofcontextsensi-tivityarereallyinstancesofsomethingelse,whichIcallsemanticincompleteness.1579781472578235_txt_print.indd15708/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageThatis,manysentencescontainingsuchexpressionsfallshortofexpressingapropositionandarethereforenotcapableofbeingtrueorfalse,evenrelativetoacontext.Theideaofsemanticincompletenessisstraightforwardifyouthinkintermsofstructuredpropositionsratherthantruth-conditions,asbuiltupoutofobjects,propertiesandrelations(seeKaplan1989a).Sincethesearemadeupofbuildingblocksassembledinaparticularway,itmakessensetosupposethatinsomecasessuchanassemblage,puttogethercompositionallyfromasentence’sconstituentsaccordingtoitssyntacticstructure,mightfailtocompriseaproposition(seeBach1994bandSoames2009).Thatisbecause,althoughthiscomprisestheentiresemanticcontentofthesentence,itlacksatleastoneconstituentneededforittobetrueorfalseandtobethecontentofathoughtorastatement.Evenso,likeamerephrase(seeStainton2006),asemanticallyincompletesentencecanbeusedtoassertaproposition.2.Indexicals:Automatic,discretionaryandhiddenThecommonphilosophicaltermforcontextuallyshiftytermsis“indexical”.Theobviousexamplesofindexicalsarepronouns(“she”,“we”,“you”,“they”,etc.)anddemonstratives(“this”,“those”,“there”,etc.),buttherearealsotemporalterms,suchas“now”and“today”,“lastweek”,and“threedaysago”,aswellasdiscourseindexicals,notably“theformer”and“thelatter”.Pronounsalsofunctionasdiscourseindexicals,whenusedanaphorically,aslinguistssay,toreferbacktopreviouslymentioneditems(forathoroughdiscussion,seeNeale2006).Infact,anaphoricpronounsillustrateoneofourmainthemes,aswiththeseexamples:(3)a.Acoparrestedarobber.Hewaswearingabadge.b.Acoparrestedarobber.Hewaswearingamask.Itisnaturaltosupposethatin(3a)“he”referstothecopandin(3b)totherobber.Itisnaturalallright,butnotinevitable.Thespeakerof(3a)couldbeusing“he”torefertotherobber,andthespeakerof(3b)couldbeusingittorefertothecop.Suchspeakerswouldprobablynotbeunderstoodcorrectly,atleastnotwithoutenoughstagesettingtooverridecommonsenseknowledgeaboutcopsandrobbers,butthatwouldbeapragmaticmistake.Nevertheless,thefactthat“he”couldbesousedsuggeststhatitisthespeaker’sintention,notthecontext,whichdeterminesthatin(3a)itreferstothecopandin(3b)totherobber.Thesamepointappliestotheseexampleswithtwoanaphoricpronouns:(4)a.Acoparrestedarobber.Hetookawayhisgun.1589781472578235_txt_print.indd15808/08/201415:46\nContextDependenceb.Acoparrestedarobber.Heusedhisgun.c.Acoparrestedarobber.Hedroppedhisgun.d.Acoparrestedarobber.Hetookawayhisgunandescaped.In(4a),presumably“he”wouldbeusedtorefertothecopand“his”totherobber,whereasin(4b)bothwouldbeusedtorefertothecop,in(4c)bothwouldbeusedtorefertotherobber,andin(4d)“he”wouldbeusedtotherobberand“his”tothecop.However,giventhedifferentusesofthepronounsinwhatisessentiallythesamelinguisticenvironment,whatexplainsthesedifferencesinreferencecanonlybethespeaker’sintention,notthenarrowcontext.Itisadifferent,pragmaticmatterhowtheaudiencefiguresoutthereferences,andthatdependsonthebroad,communicativecontext,whichcomprisestheextralinguisticinformationtobetakenintoaccount.Ifthespeakerdoesnotcorrectlyanticipatethehearer’sinference,herreferenceswillbemisunderstood.2.1AutomaticindexicalsMattersaredifferentwiththoseindexicalsthatreferindependentlyofthespeaker’sintention.Supposeyousaid(5)tome,(5)Iamrelaxedtoday.Theword“I”,thepresenttenseformoftheverb,andtheword“today”,asyouareusingthem,refertoyouandthetimeanddaywhenyouspoke.Indeed,theyseemtodosoautomatically,notbecauseofyourintention.ThatiswhyPerrycallsthem“automatic”ratherthan“discretionary”indexicals.Whattheyrefertodependsnotonwhatyouintendtorefertobut“onmeaningandpubliccontextualfacts”(Perry2001:58).So,ifinsteadIhaduttered(5)andhaddonesoonthefollowingday,thereferenceswouldhavebeentomeandtoadifferentday,andagainthereferenceswouldhavebeenautomatic.Also,ifIwerethentoremindyouwhatyousaidthepreviousday,Iwouldusedifferentindexicalsandutter(6),not(7).(6)Yousaidyouwererelaxedyesterday.(7)YousaidIamrelaxedtoday.Itisafactaboutthestandingmeaningof“I”that,asusedonagivenoccasion,itreferstoitsuserandafactaboutthemeaningof“today”that,asusedonagivenoccasion,itreferstothedayitisused.Thesefactsabouttheirmeanings1599781472578235_txt_print.indd15908/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageareessentiallyrulesfortheiruse.Thestablemeaningoftheexpression,orwhatKaplan(1989a)callsitscharacter,determineswhatcontextualparameterfixesitsreference,orwhatKaplancallsitssemanticcontent.Thiswasjustillus-tratedbythedifferencebetween“I”and“today”.Thecontent(reference)ofeachvaries,asamatterofitsrespectivelinguisticmeaning,withacertainsortoffactaboutthecontext.Moreover,asKaplanargues,theterms’characters(meanings)donotenterintothesemanticcontentsofsentencesinwhichthetermsoccur.Rather,theirreferencesenterin(Braun(2008)arguessimilarlyregardingdemonstrativephrases,suchas“thatguy”).Soifthedayafteryouuttered(5)Ispoke(8)toyou,(8)Youwererelaxedyesterday.Iwouldbesayingthesamethingyousaidthedaybefore.Thatis,despitetheirlexicaldifferences,sentence(5)asutteredbyyouononedayandsentence(8)asspokenbymetoyouthenextdaywouldhavethesamesemanticcontents,inthatwhatittakesforthemtobetruedependsonthesamething,yourbeingrelaxedonthedayinquestion.Therulesfortheuseof“you”and“yesterday”aredifferentfromthosefor“I”and“today”butinthesituationsdescribedtheirrespectivecontentsarethesame.2.2DiscretionaryindexicalsTermssuchas“now”,“then”,“here”,“we”,“you”,“she”and“that”seemtodifferinhowtheyworkfrom“I”,“today”and“lastweek”.Justcompare“now”with“today”.Whereasthereferenceof“today”asusedonagivenoccasionisstraightforwardlythedayofthatuse,thereferenceof“now”isnotstraightforwardlythetimeofitsuse.Thisisbecausetheextentofthattimeisnotfixed.Isitstrictlythemomentthat“now”isused,orcanthetimeinquestionbealargerdurationthatincludesthatmoment?Onemightinsistthatitisstrictlythemomentofuseandthatanylargerdurationmerelyincludingthatmomentisnot,strictlyspeaking,thereferenceof“now”.Thatmightseemplausibleuntiloneconsiders“here”.Theanalogousclaimwouldbethat“here”asusedonagivenoccasionreferstotheplaceofthatuse.Butwhichplaceisthat?Whatisthespatialanalogueofthemomentofutterance,evenifweexcluderegionsthatextendbeyondthespeaker’sbody?Thelocationofthespeaker’smouth,hercompletevocalapparatus,herentirebody?Thesituationissimilarwith“we”,butalsointerestinglydifferent.Fornotonlycantheextentofthereferencebelargerorsmaller,itcanbedisparate.“We”canbeusedonagivenoccasiontorefertoaplurality(orgroup)that1609781472578235_txt_print.indd16008/08/201415:46\nContextDependenceincludesthespeaker,butwhichpluralityisthat?Itcouldincludetheintendedaudience,butitmightnot,aswhenonepersonspeaksonbehalfofagroupofpeoplenotpresent.Whatdeterminesthereference?Afirstthoughtisthatthecontextdoes,butdoesit?Onsecondthought,itseemsthatthespeaker’sreferentialintentiondoesthetrick.Whatcontextdoesistoimposerationalconstraintsonthatintentionandonthehearer’sinferenceastowhatthatintentionis.Thiscontrastswithwhatdeterminesthereferencesof“I”,“today”and“lastweek”,asusedonagivenoccasion.Aspecific,objectivefactaboutthecontext(theidentityofthespeakerinthecaseof“I”andthetimeoftheutterancewith“today”and“lastweek”)determinesthereference.InSection3welookmoredeeplyintointentionsandcontextsinconnectionwithusesofdemonstrativessuchas“this”and“that”.2.3HiddenIndexicals?Severaltypesofsentencesraiseaninterestingissue.Asusedinagivencontext,theyseemintuitivelycapableofbeingtrueorfalse–theyseemtoexpressdefinitepropositions–andyettheyseemtobemissingsomethingnecessaryforthat.Toaccountforthisithasbeensuggestedthattheycontaina“hiddenindexical”.ThisphrasewasintroducedbySchiffer(1977:31–36)toaddressthelongstandingphilosophicalpuzzleaboutbeliefreports.Wewilltakeupsomelesspuzzlingcases.WeatherandotherenvironmentalreportsSo-calledweatherreportshaveattractedconsiderableattentioneversincetheywerefirstdiscussedbyPerry(1986)inconnectionwithhisnotionof“unarticu-latedconstituent”,hisphraseforanelementinthepropositionexpressedbythespeakerthatdoesnotcorrespondtoanyexpressionintheutteredsentence.Thedebatehasfocusedon“Itisraining”,buttherearesimilarsentences,someofwhichpertaintoaspectsoftheenvironmentotherthantheweather:(9)Itisraining.(10)Itiswindy.(11)Itisnoisy.(12)Itiseerie.1619781472578235_txt_print.indd16108/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageThesesentencesdonotspecifyalocationwhereitissaidtoberainingorwhatever.Moreover,theydonotseemtosaymerelythatitisraining(windy,etc.)somewhereorother,althoughthisisamatterofsomedebate(seeRecanati2002,Stanley2002andTaylor2001),atleastinthecaseof(9).Thislineseemsfarlessplausiblewiththeothersentences,especially(12).Itseemsthatthesesentences,astheystand,donotfullyexpresspropositions(arenottrueorfalse),evengiventhetimeofutterance.Thatis,itseemsthatalocationwhereitisbeingsaidtobewindy,noisyoreerieneedstobeunderstood.Thisisnotageneralrequirement,asTaylorpointsoutwithexamplessuchas“Johnisdancing”.Thatsentenceexpressesapropositioneventhoughdancingmusttakeplaceatalocation.However,somethingismissinginwhatsentences(10)–(12)express,withthesemanticallyempty“it”assubject.Comparethemwith“Chicagoiswindy”,“MidtownManhattanisnoisy”and“CarlsbadCavernsiseerie”.Speakersutteringsentencessuchas(9)–(12)mustintendsomelocationaspartofwhattheymean.Itisasiftheyutteredarichersentence,onethatincludedalocativephrase,ashere:(9+)ItisraininginSt.Andrews.(10+)ItiswindyinChicago.(11+)ItisnoisyinmidtownManhattan.(12+)ItiseerieinCarlsbadCaverns.Thesesentencesdoexpresspropositions,andspeakerscoulduse(9)–(12)toconveythesepropositions,butonlyintherightcontexts,wheretheycanreasonablyexpecttobetakentobetalkingaboutthelocationinquestion.Do(9)–(12)containahiddenindexical,somethinglike“(in)x”,fortheunspecifiedlocationthataspeakermustmean?Thatisaninterestinglinguisticquestion.Recanati(2004)andStanley(2005a)havecontinuedtheirdebateaboutwhethercontextualism,tobedefensible,musttaketheformofindexi-calism(thisisRecanati’stermfortheviewthatStanleydefends).However,neitheraddressestheunderlyingquestionofhowcontextdoeswhateachassumesitdoes,whetheritistoprovidevaluesforhiddenindexicalsormerelytosupplyunarticulatedconstituents.Regardingthisinterestinglinguisticquestion,themainreasonforpositingahiddenindexicalistoaccountforthemeaningofsentencessuchas(13a),asrepresentedin(13b):(13)a.WhereverJohnisonhisvacation,itrains.b.Wherever(x)Johnisonhisvacation,itrainsinx.1629781472578235_txt_print.indd16208/08/201415:46\nContextDependenceThatis,weseemtoneedthevariabletoaccountforthevariationintheplace,indicatedby“wherever”,whereitissaidtoberaining.Moreover,sotheso-calledbindingargumentgoes,ifthevariableispresentbecauseofaquantifierneedingavariabletobind,asin(13),thevariableispresent,functioningasanindexical,intheabsenceofaquantifier,asin(9).However,thisargumentdoesnotseemtoworkfor(14)and(15):(14)WheneverJohnisonhisvacation,itrains.(15)WheneverJohndoesn’thavehisumbrella,itrains.Inthesecasesthevariationinlocationisunderstood,butthereisnothinginthesentence(“whenever”hasreplaced“wherever”)torequirethatlocationbemarkedsyntactically.TermswithmissingcomplementsAsimilarsituationariseswithcertainexpressionsthatcanbeusedwithoutwhatlinguistscallcomplements,asinthesesentences:(16)Ronnieisready.(17)Lynnislate.Becauseonecan’tbejustplainreadyorjustplainlate(beingreadyandbeinglatearerelations,notproperties)andthesesentencesdonotindicatewhatRonnieandLynnarebeingsaidtobereadyorlatefor,thesesentencesseemnottoexpresspropositionsandnotcapableofbeingtrueorfalse.Yetthesesentencesareperfectlyusable.Howcanthatbe?Borg(2004)andCappelenandLepore(2005)maintainthatthesesentencesdoexpresspropositionsbutnottheonestheyareusedtoassert.Forexample,Borgthinksthat(16)expressesthepropositionthatRonnieisreadyforsomethingorother,eventhoughitcannotbeusedtoassertthatproposition.CappelenandLeporethinkthatitexpressesthe“minimal”propositionthatRonnieisready,fullstop.Theircritics(inPreyerandPeter2005)doubtthatthereissuchaproposition(onecan’tbejustplainready),andsomesuggestthat(16)mustbecontext-sensitive,perhapswithahiddenvariablewhosevalueiswhatRonnieisbeingsaidtobereadyfor.Evenifthereissuchavariable,ratherthanmerelyanunarticulatedconstituent,forreasonswehaveseenthecontextdoesnotprovidethatvalue.Themosteconomicalapproachdeniesthat(16)and(17)expresspropositions,1639781472578235_txt_print.indd16308/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagewhereasthefollowingaugmentedversionsofthesesentences,withtheextramaterialinitalics,doexpresspropositions(givenatimeofutterance):(16+)Ronnieisreadytogotoschool.(17+)Lynnislateforwork.Aspeakercanreasonablyuse(16)or(17)toassertthepropositionsexpressedby(16+)and(17+)evenwithoutspecifyingwhatshemeansRonnieisreadytodoorwhatLynnislatefor,providedshedoessoinacontextinwhichthehearercanfigureoutwhatsheistryingtoconvey.Contextdoesnotdeterminewhatcompletestheproposition,butenablesthehearertofigureouthowthespeakerintendsittobecompleted(seeBach1994b).RelationaltermsRelationalterms,suchas“neighbour”,“fan”,“enemy”,“local”and“foreign”,areso-called,unsurprisingly,becausetheyseemtoinvolvearelationtosomething.Youcan’tjustbeaneighbour,fullstop,butonlyaneighbourofothers.Youcan’tjustbeafan;youhavetobeafanofsomething,suchasateamoraperformer.Youcan’tbeanenemy,fullstop,butonlyanenemyofapersonoragroup.Thatiswhysentencessuchasthefollowingseemsemanticallyincompleteandhavebeenthoughttocontainhiddenindexicals:(18)Oliverisaneighbour.(19)Oscarisafan.(20)Osamawasanenemy.Theyseemtobemissingsomethingpresentintheseaugmentedversions:(18+)OliverisaneighbouroftheJoneses.(19+)OscarisafanofFCBarcelona.(20+)OsamawasanenemyofObama.Doesthisshowthatwhothepersonisbeingsaidtobeaneighbour,fanorenemyofisdeterminedbythecontext?No,itshowsonlythatthisisdeter-minedinthecontext.Whensomeoneuttersasentencesuchasoneofthese,1649781472578235_txt_print.indd16408/08/201415:46\nContextDependencewhatshemeansmustincludesuchrelationalinformation.Whetherornotsentences(18)–(20)containhiddenvariables,contextdoesnotdeterminewhatfillsthegap.PerspectivaltermsAsimilarpointappliestoperspectivalterms,suchas“left”,“distant”,“horizon”,“faint”and“occluded”.Sentencessuchasthefollowingseemtobesemanticallyincompleteforessentiallythesamereasonasthepreviousthree,andhavealsobeenthoughttocontainhiddenindexicals:(21)Thepostofficeisontheleft.(22)Onecanseeashiponthehorizon.(23)Theoldfirehouseisoccludedbyanapartmentbuilding.Inthesecasesthereisnoexplicitindicationofaperspectiveand,sincesomethingcan’tjustbeontheleft,fullstop,beonthehorizon,fullstop,etc.,theexplicitlyexpressedsemanticcontentsofthesesentencesseemtobesub-propositional.Incontrast,thesemanticcontentsoftheirexplicitlycompletedcounterpartsarefullypropositional:(21+)Thepostofficeisontheleftafteryoucrosstheintersectionof1standMain.(22+)Fromthetopofthehillonecanseeanislandonthehorizon.(23+)Fromthepostofficetheoldfirehouseisoccludedbyanapartmentbuilding.Onceagaincontextdoesnotdeterminewhatfillsthegap.Rather,itprovidesinformationtohelpthehearertofigureouthowthespeakerintendsthegaptobefilled.Wehavecertainlynotsettledthequestionsofwhethersentencessuchas(9)–(23),theonesconsideredinthissection,containhiddenindexicalsoraremerelyusedtoconveypropositionswithunarticulatedconstituents,buteitherwaytheyaresemanticallyincomplete.Thatdoesnotmeanthattheyarecontext-sensitive.Unfortunately,theteststhatCappelenandLepore(2005:ch.7)andCappelenandHawthorne(2009:ch.2)haveproposedforcontextsensi-tivityfailtodiscriminatebetweenthatandsemanticincompleteness.1659781472578235_txt_print.indd16508/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguage3.ContextandintentionSomephilosophersseenoimportantdifferencebetweencontentsthatare“determinedbythespeaker’sintentionsandthosethataredeterminedsolelybythenon-intentionalfeaturesofthecontextofutterance(suchasthespeaker,time,andplace).…[Yet]nosubstantiveorevenremotelyinterestingissuedependsuponthisdistinction”(Cappelen2007:8).However,theexamplesconsideredsofarsuggestthattheroleofthespeaker’sintentionistodeterminewhatthespeakermeans,notthecontentsofexpressionsthespeakeruses.Andwhereexpressioncontentisdetermined,asinthecaseofautomaticindex-icals,itisdeterminedby(narrow)context,withoutthehelpofthespeaker’sintention.However,somehavearguedthatcertainaspectsofwhatofwehavebeentreatingasbroadcontext,suchasrelevanceandsalience,arecapableofplayingaroleindeterminingthecontentsofcertainexpressions.Andothershavearguedthatthespeaker’sintentioncanitselfplaysucharoleandshouldbeconstruedasanaspectofnarrowcontext,hencethattheroleofcontextofthesemanticallyrelevantsortisnotlimitedtoautomaticindexicals.Inthissectionwewilltakeupseveraloftheseargumentsandlookattheroleofintentionintheuseofdemonstratives,theparadigmofdiscretionaryindexicals.3.1Contextvs.intentionGauker(2007;2010)hasarguedthatwhatiscontextuallysalientorrelevant,alongwithseveralother“accessibility”factors,cananddohelpdeterminesemanticcontent.ThisisinthespiritofLewis’s(1979,1980)liberalconceptionofthesemanticroleofcontext,onwhichevencontextuallyaptstandardsofprecisioncanplaysucharole.Gaukerineffecttreatssalienceandrelevanceaselementsofnarrowcontext,notmerelyfactorsforhearerstotakeintoaccounttofigureoutwhatthespeakermeans.Buthowcanthesefactorsbearonsemanticcontentunlessthemeaningsoftheexpressionsinquestionsomehowrequirethemtoplaysucharole?Gauker’sproposalmustineffecttreatmeaningascharacterinKaplan’ssense,butnotlimitedtomattersofobjectivefactasthespeakerandtimeofutterance.Thissortofproposalinvitesthefollowingobjection,statedoriginallybyBach(1994a:176–9)andforcefullyputbySchiffer:Meaning-as-charactermayinitiallyseemplausiblewhenthefocusisonawordsuchas“I”,butitlosesplausibilitywhenthefocusisonotherpronounsanddemonstratives.What“contextualfactors”determinethereferentofthepronoun“she”inacontextofutterance?…Evidently,themeaningof“she”(veryroughlyspeaking)merelyconstrainsthespeakerto1669781472578235_txt_print.indd16608/08/201415:46\nContextDependencerefertoafemale.Wedonotevenhavetosaythatitconstrainsthespeakertorefertoacontextuallysalientfemale,sincethespeakercannotintendtorefertoaparticularfemaleunlessheexpectsthehearertorecognizetowhichfemaleheisreferring,andtheexpectationofsuchrecognitionitselfentailsthatthespeakertakesthereferenttohaveanappropriatesalience.Whatfixesthereferentofatokenof“she”arethespeaker’sreferentialintentionsinproducingthattoken,andthereforeinorderforKaplantoaccommodate“she”,hewouldhavetosaythataspeaker’sreferentialintentionsconstituteonemorecomponentofthosen-tuplesthatheconstruesas“contexts”.ThetroublewiththisisthatthereisnoworkforKaplaniancontextstodoonceonerecognizesspeakers’referentialintentions.Thereferentofapronounordemonstrativeisalwaysdeterminedbythespeaker’sreferentialintention.(Schiffer2005:1141)Schiffer’sargumentappliesnotonlytopersonalpronounssuchas“she”(andobviouslytodemonstratives)butalsotoexpressionsofanyothersortforwhichsalienceorrelevanceisthoughttoplayacontent-determiningrole(inthesenseoffixingcontent,notmerelyascertainingit).Theonlyexceptionisautomaticindexicals.ItisimportanttoappreciatethatSchiffer’sargumentdoesnotbearonthepragmatic,thatisepistemicandnormative,roleofcontext,construedbroadlytoincludesalience,relevanceandsoon.Forthesefactorsdoconstrainwhataspeakercanreasonablyintend.However,thatroleshouldnotbeconfusedwiththesemanticroleofactuallydeterminingcontent.AsFodorandLeporeexplain,“Sincethespeaker’saccesstotheinterpretationofhisutteranceisepistemicallyprivileged,nothingaboutthebackgroundofanutteranceismetaphysicallyconstitutiveofitsinterpretation.Thefunctionofbackgroundknowledgeininterpretationis(only)toprovidepremisesforthehearer’sinfer-encesaboutthespeaker’sintentions”(2005:10).Insteadofappealingtosalienceandrelevance,whicharepropertiesthatrationalspeakerintentionsexploitbutwhichcannotplausiblyberegardedasactuallyfixingsemanticcontent,onemightarguethatthespeaker’sintentionfillsthesemanticgap.Stokke(2010)defends“intention-basedsemantics”,accordingtowhichthespeaker’sintentionisitselfanelementofnarrowcontext,hencecapableofdeterminingsemanticcontent.Focusingspecificallyondemonstrativereference,hearguesthatthespeaker’sintentioninusingademonstrativedeterminesitsreference.However,thereisaproblemwiththisproposal.Perhapsunawareoftheambiguityofthephrase“demonstrativereference”,whichcanmeaneitherreferencebyademonstrativeorreferencebyaspeakerinusingademonstrative,Stokkeevidentlyconflatesthetwo.Atbest,hisproposalrendersreferencebythedemonstrativeitselfasderivative,withnotheoreticalimportanceofitsown.As1679781472578235_txt_print.indd16708/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageNealeputsthepoint,inthespiritofStrawson(1950),“referringisanintentionalact,somethingspeakersdo,[and]talkofanexpressionitselfreferring,andeventalkofanexpressionreferringrelativetoanutterance,aspeechact,a‘context’,ora‘tokening’,is,atbest,derivative”(Neale2007:254).Thatis,wecanalwaysstipulatethattosaythatwhenthespeaker’sintentiondetermineswhatthespeakerisreferringto,theexpressiontherebyreferstoit.Butwhat’sthepointofsuchastipulation?Itjusttrivializesintention-basedsemanticsfordemonstratives.Thereisadeeperproblem,pertainingtothespeaker’sintention.Ifthespeakerintendstouseademonstrativetorefertoacertainobject,shedoessowiththeaudienceinmind.Sheuses“that”toenableheraudiencetofocusontheintendedobject,anddoessowiththeintentionthathetakethatastheobjectsheistalkingabout.Butthisispartofheroverallcommunicativeintentioninutteringthesentenceinwhich“that”occurs.Assuch,itisessen-tiallyanaudience-directed,hencepragmaticintention.Thespeakerdoesnothaveaseparatesemanticintentionthattheword“that”itselfrefertotheobjectsheisusingittoreferto.Asfarasreferencetothatobjectisconcerned,thereisjustheronepragmaticintention.Andrecognizingthatintentionisallthehearerhastodoforthereferencetobeconveyed.Itisnottheword“that”butthespeaker’suseofitthatmanifestsherreferentialintentionandtriggersthehearer’sinferencetowhatsheintendstoreferto.3.2Intentionsordemonstrations?Butisitreallythespeaker’sintentionthatdeterminesdemonstrativereference?Kaplan(1989a)initiallythoughtitwasthespeaker’sdemonstration(atleastanimplicitone,whenthereisnogesturebutaparticularobjectisuniquelysalient),butlater(Kaplan1989b)hedecidedthatitwasthespeaker’s“directingintention”,so-calledbecauseitguidestheactofdemonstration(ifthereisone)ofacertainobject.Thedemonstration,Kaplannowthought,playsmerelythepragmaticroleoffacilitatingcommunicationbymakingcleartothehearerwhatthespeakerintendstobereferringto(Braun1996modifiedKaplan’saccountofthecharacterofdemonstrativesthatfinessedthedifferencebetweenintentionsanddemonstrations).Reimer(1991)subsequentlyarguedthatKaplanshouldn’thavechangedhismind.Herargumentreliesmainlyoncaseswheretheintentionandthedemonstrationdiverge.Inoneexamplethespeakeruses“thatdog”withtheintentionofreferringtoFidobut,asFidoisfrolickingaroundwithotherdogs,sheinadvertentlypointstoSpot.ItseemsthatsheisthusreferringtoSpot,contrarytoherintention.Thisisclearfromthefactthatshewouldbespeakingfalselyifshesaid,“ThatdogisFido”.Theexplanationforthis,accordingtoReimer,isthatwhatiscontrollingisnottheintentionbuttheactofdemonstration.1689781472578235_txt_print.indd16808/08/201415:46\nContextDependenceWecanagreewithReimerthatwhatthespeakersaysinthisexampleisfalseandthatthisisbecausethereferenceisindeedtoSpot.However,asBach(1992)argued,thatisbecausetherelevantintentiononthepartofthespeakeristorefertothedogsheispointingat.ShebelievesthatthedogsheispointingatisFido–that’swhyshesays“ThatdogisFido”–andtothatextentintendstorefertoFido.However,theintentionthattheheareristorecognize,theonewhoserecognitionisessentialtothesuccessofthespeaker’sactofcommuni-cation,istheintentiontorefertothedogsheispointingat.Andthisintentionisdecisiveindeterminingthereference.Thatisbecausewhendemonstratingsomething,thespeakerintendsthehearertothinkofacertainitemastheonesheispointingatandtherebyintendingtobetalkingabout.IfsheintendedthehearertothinkofthedogasFido,shecouldnothavethoughtshewasprovidinghimwithnewinformation.Shewouldhavebeentellinghimthatthisdog,Fido,wasFido.So,althoughshefailedtorefertothedogsheintendedtoreferto,insofarassheintendedtorefertoFido,shesucceededinreferringtothedogsheintendedtodemonstrateanddiddemonstrate,theoneshewaspointingat.Theformerintentioncanbetrumpedbythedemonstrativegesture,butthelattercannotbe.3.3Contextofutterance?Atthebeginningweassumedthatsomeindexicals,suchas“I”and“today”,areautomatic(ratherthandiscretionary)andthatthecontextofutterancedeterminestheirreferences.However,thereareexceptionalcases,suchasthestandardansweringmachinemessage“Iamnotherenow”andthesimilarofficedoormessage“Iwillnotbeinmyofficetoday”.Themostobviouspartoftheproblemisthattheintendedtimeassociatedwith“now”,“today”orthepresenttenseisnotthetimeofspeakingorwriting.Notonlythat,thespeaker(orwriter)neednotbetherightperson(therecordedmessagemighthavecomewiththetelephone,andacolleaguemighthavepostedthenote).Similarsituationsarisewithadvertisements,signs,wills,fictionalandhistoricalnarra-tives,stenographersandtranslators.Suchexamplessuggestthatthecontextofutterance(theagentandthetimeoftheutterance)doesnotalwaysdeterminethereference,orelsethatthenotionofutterancecontextmustbeconstruedmoreabstractly.Eitherway,thequestionarisesofhowtherelevantnotionofcontexttiesinwiththemeaningsofwordssuchas“I”,“now”and“today”.Thatis,whatcoulditbeaboutthemeaningof“I”thatmakesitrefertothepersonbeingcalledortheoccupantoftherelevantoffice?Andwhatcoulditbeaboutthemeaningsof“now”and“today”thatcouldmakethemrefertothetimeofthecallorthedayofthevisittotheoffice.Indefenceofthestandardappealtocontextofutterance,one1699781472578235_txt_print.indd16908/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagemightappealtothedistinctionbetweenthesemanticcontentofthesentencesinwhichtheywereoriginallyutteredandthecontentstheyaretobetakentohavewhenheardbyacallerorreadbyavisitor.Todefendsuchaview,onewouldhavetoarguethatthesearenotliteralusesofthesentencesinquestion,oratleastthattheyaretobetreatedasiftheywerebeingutteredatthetimetheyareencountered.However,suchmoveswouldrequireinvokingintentionsandtherebyviewingtheseindexicalsaslessautomaticandmorediscretionarythanstandardlysupposed.Toaddressthisproblem,oratleastthepartinvolving“I”,DoddandSweeney(2010)refinethenotionofagentofthecontext.TheyconsiderdifferentversionsoftheansweringmachineandPost-itNoteexamplesandarguethatinsomecasestheagentofthecontextcanbesomeoneotherthanthepersonwhoactuallyproducedthetokenof“I”.Intheseasinstandardcases,“I”automati-callyreferstotheagentofthecontext.Inothercases“I”doesn’treferatall.Forexample,ifItakean“Iwillbebacksoon”noteleftfromyesterdayonmyneighbour’sdoorandputitonmydoor,Iamtheagentofthecontext.Butifmyneighbour,knowingthatIamawaybutwillreturnsoon,putshernoteonmydoor,Iamnottheagentofthecontext.Inthiscase,“I”doesnotmanagetorefertome,althoughitisnaturallytakenasifitdoes.4.AdjectivesandotheradditionalexamplesSofarwehaveattendedmainlytotheprimecandidatesforcontextsensi-tivity,indexicalsanddemonstratives,andtoseveralspecificsortsofsentencesthoughttocontainhiddenindexicals.Nowwewilltakeupseveralkindsofadjectivesthathavebeenclaimedtobecontext-sensitive.Afterthatwewillmentionafewothercandidates.4.1GradableadjectivesTherearethousandsofgradableadjectives,andtheyfallintotwobroadcategories.Relativeadjectives,exemplifiedby“tall”,“old”,“fast”and“smart”,arethemoreabundantbyfar.Lesscommonbutstillplentifulareabsoluteadjec-tives,suchas“flat”,“empty”,“pure”and“dry”.Relativeadjectivesapplyindegreesbut,unlikeabsoluteadjectives,nottoanymaximaldegree.Therearenoultimatedegreesoftallness,oldness,etc.Evenfastness,thoughpresumablylimitedinfactbythespeedoflight,isnotlimitedbythesemanticsof“fast”.Nothingcanbecompletelyorperfectlytall,old,fastorsmart.Incontrast,absoluteadjectives,astheirnamewouldsuggest,applytomaximaldegrees.Evenif,asamatterofphysicalfact,asurfacecannotbeperfectlyflatanda1709781472578235_txt_print.indd17008/08/201415:46\nContextDependencecontainercannotbecompletelyempty,thesephrases,“perfectlyflat”and“completelyempty”,makeperfectlygoodsemanticsense.RelativeadjectivesLetusbeginwithrelativeadjectives,astheyoccurinsentencessuchasthese:(24)Nojockeyistall.(25)RichardHughesistall.(26)Fidoisold.(27)MichaelPhelpsisfast.Allofthesesentencesmayseemtrue.Inparticular,(24)seemstrue.YetIrishjockeyRichardHughesis5’10”,andthatmakeshimtallforajockey.Doesthatmake(25)trueand(24)false?Notif“tall”isconstruedasacontext-sensitiveterm.Butisthattherightwaytolookatit?Thekeythingtonoticeisthattallnessisnotaproperty.Onecannotbejustplaintall.RichardHughesistallforajockeybutnottallforanadultCaucasianman–beingtallforajockeyisapropertyhehas,andbeingtallforanadultCaucasianmanisapropertyhelacks.Soitwouldseemthatboth(24)and(25),aswellas(26)and(27),aresemanticallyincomplete,arenotcapableofbeingtrueorfalse.Could(24)and(25)stillbetrueorfalse,albeitonlyrelativetocomparisonclasses(jockeys,adultmen,etc.)?Kennedy(2007)pointsoutthat,strictlyspeaking,therelativizationofagradableadjectiveistoastandard,notneces-sarilytoacomparisonclass(evenwithaspecifiedcomparisonclass,asin“tallforajockey”,astandardisstillneeded).Evenso,whenwerelativizeexplicitly,usuallyitistocomparisonclassesratherthantostandards,asintheseaugmentedversionsof(24)–(27):(24+)Nojockeyistallforanadulthuman.(25+)a.RichardHughesistallforajockey.b.RichardHughesistallforanadultCaucasianman.(26+)Fidoisoldforadog.(27+)MichaelPhelpsisfastforaswimmer.1719781472578235_txt_print.indd17108/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguage(25+a)and(25+b)arerelativizedtodifferentcomparisonclasses,andexpressdifferentpropositions,onetrueandonefalse(noticethatwhatmakesthedifferenceistheadditionalphrase,nottheword“tall”,whoseroleisthesameinboth).Speakerscanuse(25)toconveyoneofthose(orsimilar)propositions.Whichoneshesucceedsinconveyingdependsonthecontext,inthebroadsenseinwhichthepropositionshecanreasonablyexpecttobetakentoconveyandthehearercanreasonablytakehertoconveydependsonwhatinformationismutuallyavailable.Insomecasesitwillbenecessarytoexplicitlymentiontheintendedcomparisonclass.(Itshouldbenotedthatrelativeadjectivesaregenerallyvague,andvaguenessmightseemtoinvolveadifferentsortofapparentcontextdependence.Howthatistreateddependsonone’stheoryofvagueness,fartoobigatopictotakeupinthischapter.ItisbrieflydiscussedinChapter12.)AbsoluteadjectivesAbsoluteadjectives,unlikerelativeones,applytomaximaldegrees.Thereisamaximaldegreetowhich,forexample,asurfacecanbeflat,acontainercanbeempty,asampleofgoldcanbepure,oraclothcanbedry.Soitmightseemthatsentencessuchasthefollowingdeterminatelyexpresspropositionsandareperfectlycapableofbeingtrueorfalse.(28)Pooltablesareflat.(29)Somefreightcarsareempty.(30)ThegoldinFortKnoxispure.(31)Myraincoatisdry.However,itisdebatablewhethersuchpredicateshavetoapplytothemaximaldegreeforsentencessuchasthesetobetrue(nevermindwhethertheycanapplymaximallyintherealworld).Forexample,eventhoughpooltablesarenotperfectlyflat,itseemsthat(28)istrue.Itmayseemthatsuchsentencescanbetrueiftherelevantpredicateappliescloselyenough,butUnger(1975)arguedthatsomethingcanbeflat(empty,pure,dry)onlyifnothingcanbeflatter(emptier,purer,drier)thanit.However,helaterconjecturedthatthesetwocontrastingviews,whichhedubbedcontextualismandinvariantism,areequallytenable(Unger1984).Onlyonacontextualistviewaresentencessuchas(28)–(31)semanticallycontext-sensitive;onaninvariantistviewtheyarenot.Ifinvariantismiscorrect,suchsentencesarecategoricallytrueonlyiftheirpredicatesapplycompletelyorperfectly,andcategoricallyfalseotherwise.This1729781472578235_txt_print.indd17208/08/201415:46\nContextDependencemeansthatusesofthemarepragmaticallycontext-sensitive.Lasersohn(1999)describestheirusesasblessedwith“pragmatichalos”,whichmaybelargerorsmallerdependinghowtolerantorexactingtheiruse.Utterancesofthemcanbetreatedastrueiftheyarecloseenoughtoapplyingmaximally(asiftheyincludedanimplicit“approximately”,“forallpracticalpurposes”,orthelike),butwhatcountsascloseenoughvaries.Apooltablecanbeflatforallpracticalpurposesifitislevelenoughandsmoothenoughforplayinganunimpededgameofpool,assumingthat’stherelevantpurpose,butanairportrunwaycanbelessflatandstillcountasflatforitspurpose.WhydidUnger(1984)suggestthatthereisnoobjectivebasisfordecidingbetweenthetwoviewsontermssuchas“flat”and“empty”?Hepointedoutthatappealingtointuitionsisinconclusivebecausetheypointinbothdirec-tions.Forexample,itisintuitivelyplausiblethatsomethingcanbeflatonlyifnothingcanbeflatterthanit,andthatphrasessuchas“nearlyflat”andeven“almostperfectlyflat”implynotflat.Ontheotherhand,itseemsthatphrasessuchas“perfectlyflat”and“completelyflat”arenotredundant,hencethattheyarestrongerthanunmodified“flat”.SoUngerdoubtedthattheconflictbetweencontextualismandinvariantismaboutabsoluteadjectivescanberesolved.However,weshouldrememberthat“contextualism”isnotreallytherighttermforthefirstview,sinceitdoesnotreallyclaimthatcontextdeter-minesthestandardforapplyingtheterm.4.2Termsforresponse-dependentpropertiesAspecialcaseofgradableadjectivesarethose,suchas“edible”,“scary”and“comfortable”,usedtoascriberesponse-dependentproperties.Thesepropertiesareso-calledbecausetheyapplytothingsorsubstancesbecauseoftheeffectsthatthesethingsorsubstanceshaveonotherthings(sotheyarebetterdescribedasrelationsthanproperties).Thetermsthatexpressthemaregenerallyrelativeadjectives,butsome,suchas“lethal”,areabsolute.Theyareworthsinglingoutherebecausetheymightseemtobecontext-dependentquiteapartfrombeingstandard-relative(andvague).Considersentencessuchasthese:(32)Somerootsareedible.(33)Horrormoviesarescary.(34)Wassilychairsarecomfortable.Thesesentencesareusedtoindicateacertainresponsiveness(sensitivityorsensibility),eitherbiologicalorpsychological.Withedibility,obviouslythis1739781472578235_txt_print.indd17308/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageresponsivenessisbiological,buttypicallywhenaspeakermakesastatementusingasentencesuchas(32)itreflectstheresponsivenessofherbiologicaltype,thatis,Homosapiens.Withscarinessandcomfortableness,althoughbiologicallybased,itisprimarilypsychological,amatterofhowthingsoftherelevantsorts,horrormoviesorWassilychairsintheaboveexamples,areexperienced.Thingsarenotedible,scaryorcomfortable,fullstop,butonlyrelativetoanimalsorpeople.Thissuggeststhatsentencessuchas(32)–(34)arenotcontext-sensitivebut,rather,semanticallyincomplete.Thatdoesnotmean,ofcourse,thattheycannotbeusedtoconveypropositions.Typically,butnotnecessarily,statementsmadeinusingsentencessuchas(32)–(34)arefromthespeaker’sperspective,perhapsgeneralizedtoincludeagroupofwhichthespeakertakesherselftobetypical.Ifadvisedthatmostmenarenotscaredbyhorrormovies,afemalespeakermightretreatandclaimthathorrormoviesarescarytomostwomen.Perhapsmorerealistically,theintendedgroupcomprisespeoplesuchasthespeakerorperhapspeoplewithsensibilitieslikethespeaker’s.Theseobservationssuggestthatsentencessuchastheabove,iftheyaretoexpresspropositions,needtobeaugmentedinwayssuchasthese:(32+)a.Somerootsareedibleforhumans.b.Somerootsareedibleforgoats.(33+)a.Horrormoviesarescarytopeoplelikeme.b.Horrormoviesarescarytosmallchildren.(34+)a.Wassilychairsarecomfortable,atleasttopeopleofmysizeandshape.b.Wassilychairsarecomfortabletopeoplewithstrongbacks.Thesesentencesexpresspropositionsthatspeakersofthesemanticallyincom-pletesentences(32)–(34)mightmean.Typically,butnotnecessarily,statementsmadeinusing(32)–(34)arefromthespeaker’sperspective,perhapsgener-alizedtoincludeagroupofwhichthespeakertakesherselftobetypicalastoresponse.Theaugmented(a)sentencesexpresssuchpropositions.However,asthe(b)sentencesillustrate,theintendedgroupneednotincludethespeaker.Itisnotdifficulttoimaginecontexts,sayinresponsetoaquestionaboutwhetherlittleBillyshouldwatchHalloween5,inwhicharationalspeakeruttering(33)islikelytomean,andarationallistenerislikelytotakethespeakertomean,thathorrormoviesarescarytosmallchildren.However,thisdoesnotshowthatthecontextdeterminesthatthisiswhatismeant.Note,finally,thatthespeakermaynothaveaprecisecommunicativeintentionwhenutteringasentencesuchas(32)–(34),inwhichcasethereisnopropositionthatthespeakerdetermi-natelymeans(seeBuchanan2010).1749781472578235_txt_print.indd17408/08/201415:46\nContextDependence4.3PredicatesofpersonaltasteTheaboveproblemsareexacerbatedforpredicatesofpersonaltaste,aspecialcaseofadjectivesforresponse-dependentproperties.Theyincludesuchadjec-tivesas“fun”,“tasty”,“thrilling”and“funny”.Theydonotexpressabsolutepropertiesbut,soitissometimesthought,propertiesthatarerelativetoajudgeortoanassessor(seeLasersohn2005andMacFarlane2005).However,arguablytheyareexperiencer-relative,sincewhetherornotsomethingis,say,funortastydependsonhowitaffectstheperson,onwhatsortofexperienceitcauses,notonthejudgmentorassessmentthatthepersonmakesonthebasisofthatexperience(seeGlanzberg2007).Statementsmadeusingsimple,unrelativizedversionsofsentencescontainingpredicatesofpersonaltastearetypicallymadefromthespeaker’sperspective:(35)Pineapplesaretasty.(36)RaftingthroughtheGrandCanyonisthrilling.(37)JonStewartisfunny.However,aspeakercanalsoadopttheperspectiveofsomeoneelseorsomegroup,asperhapsinutteringthesesentenceswiththesamepredicates:(38)Alpoistasty.(39)Skydivingisthrilling.(40)ThreeStoogesmoviesarefunny.Someoneuttering(38)islikelytomeanthatAlpoistastytodogs,orperhapstoherowndog,nottoherself.Someoneuttering(39),sayifshewereexplainingwhysomepeopleengageinskydiving,couldwellmeanthatskydivingisthrillingtopeoplenotafraidtodoit,andsomeonemightutter(40)withtheperspectiveofFrenchpeopleinmind,sayifshewereexplainingwhyLesTroisImbecilesarestillpopularinFrance.Itisalwayspossible,andtoensuresuccessfulcommunicationoftennecessary,tomaketherelevantperspectiveexplicit,byutteringasentencerelativizedtothatperspective,aswiththesemoreelaborateversionsof(38)–(40):(38+)Alpoistastytodogs.(39+)Skydivingisthrillingtopeoplenotafraidtodoit.1759781472578235_txt_print.indd17508/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguage(40+)TheThreeStoogesarefunnytoFrenchpeople.Insomecasesitisunnecessarytospecifytheexperiencersfromwhoseperspectivebeingtasty,thrillingorfunnyisbeingconsidered.Forexample,ifyouareaskingsomeonewhethersomethingistasty,thrillingorfunny,oryouarereportingthatsomeonefindssomethingtasty,thrillingorfunny,normallyyouareaskingfororreportinghowitisfromtheirperspective.Lasersohn(2005)andStephenson(2007)haveproposedrelativistaccountsofsentencessuchas(35)–(40),accordingtowhichsuchsentencesexpressproposi-tionsthatarenottrueorfalseabsolutelybutonlyrelativetojudgesorassessors.Leavingasidegeneralchallengestotheideaofrelativetruth(seeGlanzberg(2009)andSoames(2011),andseveralcontributionstoGarcía-CarpinteroandKölbel(2008)),Boghossian(2006)andWright(2008)haveraiseddoubtsthattruthrelativisminagivenarea,suchaswherepredicatesofpersonaltasteareinvolved,canamounttoanythingmorethanpropertyrelativisminthatarea.Theirbasicideaisthatcertainostensiblymonadicproperties,asexpressedbyone-placepredicates,areactuallyrelational.Thisissupportedbyourobservationthatwhentherelatumisactuallyspecified,asinthesentencesin(38+)–(40+),thereisnoquestionofrelativetruth.Oncewetakethosesentencesintoaccount,wecansee,bysubtractionasitwere,that(38)–(40)aresemanti-callyincomplete.Insteadofsupposingthattheyexpresspropositionswhosetruthisjudge-orexperiencer-relative,wecanrejectthepresuppositionthattheyexpresspropositionsatall.Thisleavesopen,ofcourse,whatspeakersdowhentheyusesuchsentences.Theymayormaynothaveaparticularperspectiveinmind,andtheymayormaynotbeunderstoodasassertingsomethingthatinvolvesaparticularperspective,buttheseareseparatequestions,questionsaboutusingunrelativizedsentencescontainingpredicatesofpersonaltaste.Assentences,(35)–(40)arenodifferentfromtheonescontainingtermsforresponse-dependentpropertiesor,forthatmatter,otherrelativeadjectives.4.4OddsandendsWehavediscussedavarietyoftypesofexpressionsthatareoratleasthavebeenthoughttobecontext-sensitive.Thefollowinglistincludesthemandafewmore,which,duetospacelimitations,cannotbetakenuphere:•automaticindexicals:I,today,nextweek,lastyear[tense]•discretionaryindexicalsanddemonstratives:we,she,they,you,now,here,then,there,that,those•weatherandotherenvironmentalreports:(Itis)raining,hot,humid,windy,noisy,eerie,crowded1769781472578235_txt_print.indd17608/08/201415:46\nContextDependence•expressionsthatcanbeusedwithoutcomplements:ready,late,finish,strongenough,legal,eligible,incompetent,experienced,applicable,relevant,difficult•relationalterms:neighbour,fan,enemy,local,foreign,employee,mentor•perspectivalterms:left,distant,up,behind,foreground,horizon,faint,occluded•gradableadjectives,bothrelativeandabsolute:tall,old,fast,smart;flat,empty,pure,dry•termsforresponse-dependentproperties:edible,poisonous,scary,nause-ating,comfortable•predicatesofpersonaltaste:fun,funny,thrilling,boring,tasty,delicious,tasteful,cute,sexy,cool•possessivephrases,adjectivalphrases,noun-nouncompounds:John’scar,John’shometown,John’sboss,John’scompany;fastcar,fastdriver,fasttyres,fasttime;waterlily,tigerlily;childabuse,drugabuse;vitaminpill,dietpill,sleepingpill,painpill•subsententialutterances:“Ashark!”,“Inthecupboard”,“Scalpel!”,“Water!”,“Coffeeortea?”•prepositions:in,on,to,at,for,with•lightverbs:do,have,put,get,go,make,take,give•implicittemporal,spatial,andquantifierdomainrestriction•certainphilosophicallyinterestingterms:know,might,probable,necessary,explain,and,or,if,obligatory,permissible,ought,free,responsibleFinally,somephilosophersandlinguistshavesuggestedthatall,oratleastavastpreponderance,ofsentencesarecontext-sensitive,irrespectiveofwhichparticularwordstheycontain(Searle1978,SperberandWilson1986,Travis2000,Carston2002,Recanati2004).Radicalcontextualismisbasedonintuitionsaboutthetruth-valuesofutterancesofagivensentenceinvariouscontexts.Bydescribingvariouscontextsinwhichthesamesentencecanbeusedtomeandifferentthings,asevidencedbyintuitionsthatinsomecontextsanutteranceofthesentenceistrueandinothersfalse,thesewritersconcludethatthemeaningofthesentence,evenifitcontainsnoobviouslyindexicalexpressions,iscontext-sensitive.Therearethreeproblemswiththislineofreasoning(formoredetaileddiscussionseeBach2005:29–34).Firstofall,themostthatitshowsisthatagreatmanysentencesaresemanticallyincomplete.Itdoesnotshowthattheyarecontext-sensitive.Secondly,thesentencesusuallyconsideredarenotrandomsentencesbut,rather,sentencesofthetypicallyshortsortthatpeopleuseineverydayspeech.Evenifthosesentencesaresemanticallyincomplete,andevenifthatshowedthattheyarecontext-sensitive,itdoesnotfollowthattherearen’tother,moreelaboratesentences,oneforeachcontext,whoseutterancewouldhavemadewhatthespeakermeantfullyexplicit.Andthere1779781472578235_txt_print.indd17708/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagearealotmoreofthoselonger,moreelaboratesentencesthanthereareoftheshortsentencesthatmotivatethelineofreasoninginthefirstplace.Thirdly,thislineofargumenthasaratherdrasticimplication.Ifitworkednotonlyfortheshortsentencesunderconsiderationbutalsoforthemoreelaborateones,wewouldhavetoconcludethatthoughtsareessentiallyineffable.Thatisbecausetheargumentwouldshowthatnoamountofelaboration,ofspellingoutthethoughtthespeakeristryingtoexpress,couldevermakethatthoughtfullyexplicit.Perhapssomewouldwelcomethisconsequence,butitseemstobeanunfortunateconsequenceoftheradicalcontextualistposition.Ittakesalottoshowthatexpressionsofagiventypearecontext-sensitive.Itisnotenoughtopointoutthatwhataspeakermeanswhenusingtheexpressioncanbedifferentindifferentcontexts.Ithastobethecontentoftheexpressionitselfthatvaries,andithastobethecontext,inawaydeterminedbythemeaningoftheexpression,thatmakesthedifference.Thefactthatasentencedoesnotexpressapropositionindependentlyofthecontextdoesnotshowthatitiscontext-sensitiveandthatwhichpropositionitexpressesdependsonthecontext.Foritmightnotexpressapropositionatall.Whatpassesforasentence’s“intuitive”contentisoftenthepropositionaspeakerusesittoconvey.Appendix:GuidetoFurtherReadingTolookmoredeeplyintothetopicstouchedoninthischapterandintosomerelevantonesomittedforreasonsofspace,thereadershouldconsulttherefer-enceslistedbelow.Adjectivesandcontextsensitivity:Unger(1984),Lahav(1989),Ludlow(1989),Heal(1997),Kennedy(2007),Glanzberg(2007),Reimer(2002),RothschildandSegal(2009)Bindingargument:Partee(1989),StanleyandSzabó(2000),Bach(2000),CappelenandLepore(2005:ch.6),Stanley(2005b),CappelenandHawthorne(2007),CohenandRickless(2007),Neale(2007:354–8),Elbourne(2008),Sennet(2008)Contextandcompositionality:KingandStanley(2005),FodorandLepore(2005),Pagin(2005),Recanati(2010:ch.1)Contextsandintentions(anddemonstrations),Lewis(1979;1980),Kaplan(1989a;1989b),Reimer(1992),Bach(1992),Braun(1996),García-Carpintero(1998),Stalnaker(1999),King(2001),FodorandLepore(2005),Bach(2005;2006a),Gauker(2007),Åkerman(2010),Gauker(2010),Montminy(2010),Mount(2010),Stokke(2010)Contextualism(epistemic),PreyerandPeter,eds.(2005,PartI)Contextualism(semantic),Strawson(1950),Searle(1978;1980),Lewis(1979),SperberandWilson(1986),Travis(2000),Bezuidenhout(2002),Carston(2002),Recanati(2004;2010)1789781472578235_txt_print.indd17808/08/201415:46\nContextDependenceHiddenindexicals(variables,arguments)andunarticulatedconstituents:Perry(1986),Stanley(2000),Taylor(2001),Recanati(2002),Stanley(2002),Martí(2006),Neale(2007),Hall(2008),Recanati(2010,ch.4),Landau(2010)Hiddenindexicalsandbeliefreports:Schiffer(1977:31–36;1992),Richard(1990),Crimmins(1992),Bach(1997)Indexicalsandtheirdiverseuses:Smith(1989),Sidelle(1991),Predelli(1998),Corazzaetal.(2002),Romdenh-Romluc(2002),Weatherson(2002),Predelli(2002),Corazza(2006),Mount(2008),Åkerman(2009),Egan(2009),Stevens(2009),DoddandSweeney(2010),Recanati(2010:ch.6)Pragmaticintrusion(intosemanticcontent),Levinson(2000),Recanati(2004),KingandStanley(2005),KortaandPerry(2008)Relativetruth:Kölbel(2002),Lasersohn(2005),MacFarlane(2005),Stephenson(2007),Wright(2007),García-CarpinteroandKölbel,eds.(2008),CappelenandHawthorne(2009),Glanzberg(2009),Soames(2011),EganandWeatherson,eds.(2011:Intro.andchs.4–8)Semanticincompleteness/underspecification:Atlas(1989:Ch.2),Bach(1994b),Recanati(2004),Atlas(2005),Stainton(2006),Soames(2009),andworkslistedunder“Contextualism(semantic)”aboveSemanticminimalismvs.contextualism:Unger(1984),Borg(2004),CappelenandLepore(2005),Bach(2005;2006b),Montminy(2006),PreyerandPeter,eds.(2007)Vaguenessandcontextsensitivity:Raffman(1996),Lasersohn(1999),Fara(2000),Stanley(2003),Shapiro(2006),Kennedy(2007)ReferencesÅkerman,Jonas(2009),“Apleaforpragmatics”,Synthese170:155–67.—(2010),“Communicationandindexicalreference”,PhilosophicalStudies149:355–66.Atlas,JayDavid(1989),PhilosophyWithoutAmbiguity,Oxford:OxfordUniversityPress.—(2005),“Semanticalunderdeterminacy”,inLogic,Meaning,andConversation,Oxford:OxfordUniversityPress,pp.3–44.Bach,Kent(1992),“Intentionsanddemonstrations”,Analysis52:140–46.—(1994a),ThoughtandReference,pbk.edn,Oxford:OxfordUniversityPress.—(1994b),“Conversationalimpliciture”,MindandLanguage9:124–62.—(1997),“Dobeliefreportsreportbeliefs?”PacificPhilosophicalQuarterly78:215–41.—(2000),“Quantification,qualification,andcontext:AreplytoStanleyandSzabo”,MindandLanguage15:262–83.—(2005),“Contextexmachina”,inZ.Szabóed.Semanticsvs.Pragmatics,Oxford:OxfordUniversityPress,pp.15–44.—(2006a),“Whatdoesittaketorefer?”inE.LeporeandB.Smith(eds),TheOxfordHandbookofPhilosophyofLanguage,Oxford:OxfordUniversityPress:pp.516–54.—(2006b),“Theexcludedmiddle:Semanticminimalismwithoutminimalproposi-tions”,PhilosophyandPhenomenologicalResearch73:435–42.Bach,Kent,andRobertHarnish(1979),LinguisticCommunicationandSpeechActs,Cambridge,MA:MITPress.1799781472578235_txt_print.indd17908/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageBezuidenhout,Anne(2002),“Truth-conditionalpragmatics”,PhilosophicalPerspectives16:105–34.Boghossian,Paul(2006),“Whatisrelativism?”inP.GreenoughandM.Lynch(eds),TruthandRealism,Oxford:OxfordUniversityPress,pp.13–37.Borg,Emma(2004),MinimalSemantics,Oxford:OxfordUniversityPress.—(2007),“Minimalismversuscontextualisminsemantics”,inG.PreyerandG.Peter(eds),ContextSensitivityandSemanticMinimalism:NewEssaysonSemanticsandPragmatics,Oxford:OxfordUniversityPress,pp.546–71.Braun,David(1996),“Demonstrativesandtheirlinguisticmeanings”,Noûs30:145–73.—(2008),“Complexdemonstrativesandtheirsingularcontents”,LinguisticsandPhilosophy31:57–99.Buchanan,Ray(2010),“Apuzzleaboutmeaningandcommunication”,Noûs44:340–71.Cappelen,Herman,andJohnHawthorne(2007),“Locationsandbinding”,Analysis67:95–105.—(2009),RelativismandMonadicTruth,Oxford:OxfordUniversityPress.Cappelen,Herman,andErnieLepore(2005),InsensitiveSemantics:ADefenseofSemanticMinimalismandSpeechActPluralism,Oxford:Blackwell.—(2007),“Themythofunarticulatedconstituents”,inM.O’RourkeandC.Washington(eds),SituatingSemantics:EssaysInHonorofJohnPerry,Cambridge,MA:MITPress,pp.199–214.Carston,Robyn(2002),ThoughtsandUtterances:ThePragmaticsofExplicitCommunication,Oxford:Blackwell.Cohen,JonathanandSamuelRickless(2007),“Bindingargumentsandhiddenvariables”,Analysis67:65–71.Cohen,L.J.(1971),“SomeremarksonGrice’sviewsaboutthelogicalparticlesofnaturallanguage”,Y.Bar-Hilleled.PragmaticsofNaturalLanguage,Dordrecht:Reidel,pp.50–68.Corazza,Eros(2006),“Ontheallegedambiguityof‘now’and‘here’,”Synthese138:289–313.Corazza,Eros,WilliamFishandJonathanGorvett,J.(2002),“WhoisI?”PhilosophicalStudies107:1–21.Crimmins,Mark(1992),TalkaboutBeliefs,Cambridge,MA:MITPress.Dodd,DylanandPaulaSweeney(2010),“Indexicalsandutteranceproduction”,PhilosophicalStudies150:331–48.Egan,Andy(2009),“Billboards,bombs,andshotgunweddings”,Synthese251–79.Egan,Andy,andBrianWeatherson(eds)(2011),EpistemicModality,Oxford:OxfordUniversityPress.Elbourne,Paul(2008),“Theargumentfrombinding”,PhilosophicalPerspectives22:89–110.Fara,DeliaGraff(2000),“Shiftingsands:Aninterest-relativetheoryofvagueness”,PhilosophicalTopics20:45–81(originallypublishedunderthename“DeliaGraff”).Fodor,JerryandErnieLepore(2005),“Outofcontext”,ProceedingsandAddressesoftheAmericanPhilosophicalAssociation78:3–20.García-Carpintero,Manuel(1998),“Indexicalsastoken-reflexives”,Mind107:529–64.1809781472578235_txt_print.indd18008/08/201415:46\nContextDependenceGarcía-Carpintero,Manuel,andMaxKölbel(eds)(2008),RelativeTruth,Oxford:OxfordUniversityPress.Gauker,Christopher(2007),“Notoleranceforpragmatics”,Synthese175:359–71.—(2010),“Contextsinformalsemantics”,PhilosophyCompass5:568–78.Glanzberg,Michael(2007),“Context,content,andrelativism”,PhilosophicalStudies136:1–29.—(2009),“Semanticsandtruthrelativetoaworld”,Synthese166:281–307.Grice,Paul(1989),StudiesintheWayofWords,Cambridge,MA:HarvardUniversityPress.Hall,Alison(2008),“Freeenrichmentorhiddenindexicals?”Mind&Language23:426–56.Heal,Jane(1997),“Indexicalpredicatesandtheiruses”,Mind106:619–40.Kaplan,D.(1989a),“Demonstratives”,inJ.Almog,J.PerryandH.Wettstein(eds),ThemesFromKaplan,Oxford:OxfordUniversityPress,pp.481–563.—(1989b),“Afterthoughts”,inJ.Almog,J.PerryandH.Wettstein(eds),ThemesFromKaplan,Oxford:OxfordUniversityPress,pp.565–614.Kennedy,Christopher(2007),“Vaguenessandgrammar:Thesemanticsofrelativeandabsolutegradableadjectives”,LinguisticsandPhilosophy30:1–45.King,Jeffrey(2001),ComplexDemonstratives:AQuantificationalAccount,Cambridge,MA:MITPress.King,Jeffrey,andJasonStanley(2005),Semantics,Pragmatics,andtheRoleofSemanticContent,inZ.Szabóed.,Semanticsvs.Pragmatics,Oxford:OxfordUniversityPress,pp.111–64.Kölbel,Max(2002),TruthWithoutObjectivity,London:Routledge.Korta,Kepa,andJohnPerry(2008),“Thepragmaticcircle”,Synthese165:347–57.Lahav,Ran(1989),“Againstcompositionality:thecaseofadjectives”,PhilosophicalStudies,55:111–29.Landau,Idan(2010),“Theexplicitsyntaxofimplicitarguments”,LinguisticInquiry41:357–88.Lasersohn,Peter(1999),“Pragmatichalos”Language75:522–51.—(2005),“Contextdependence,disagreement,andpredicatesofpersonaltaste”,LinguisticsandPhilosophy28:643–86.Levinson,Stephen(2000),PresumptiveMeanings:TheTheoryofGeneralizedConversationalImplicature,Cambridge,MA:MITPress.Lewis,David(1979),“Scorekeepinginalanguagegame”,JournalofPhilosophicalLogic8:339–59.—(1980),“Index,context,andcontent”,inS.KangerandS.Ohman(eds),PhilosophyandGrammar,Dordrecht:Reidel,pp.79–100Ludlow,Peter(1989),“Implicitcomparisonclasses”,LinguisticsandPhilosophy12:519–33.MacFarlane,John(2005),“Makingsenseofrelativetruth”,ProceedingsoftheAristotelianSociety105:321–39.Martí,Luisa(2006),“Unarticulatedconstituentsrevisited”,LinguisticsandPhilosophy29:135–66.Montminy,Martin(2006),“Semanticcontent,truth-conditionsandcontext”,LinguisticsandPhilosophy29:1–26.1819781472578235_txt_print.indd18108/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguage—(2010),“Contextandcommunication:Adefenseofintentionalism”,JournalofPragmatics42:2910–18.Mount,Allyson(2008a),“Theimpurityof‘pure’indexicals”,PhilosophicalStudies138:193–209.—(2008b),“Intentions,gestures,andsalienceinordinaryanddeferreddemon-strativereference”,MindandLanguage23:145–64.Neale,Stephen(2006),“Pronounsandanaphora”,inM.DevittandR.Hanley(eds),TheBlackwellGuidetothePhilosophyofLanguage,Oxford:Blackwell,pp.335–73.—(2007),“Onlocation”,inM.O’RourkeandC.Washington(eds),SituatingSemantics:EssaysinHonorofJohnPerry,Cambridge,MA:MITPress,pp.251–395.Pagin,Peter(2005),“Compositionalityandcontext”,inG.PreyerandG.Peter(eds),ContextualisminPhilosophy:Knowledge,Meaning,andTruth,Oxford:OxfordUniversityPress,pp.303–48.Partee,Barbara(1989),“Bindingimplicitvariablesinquantifiedcontexts”,inC.Wiltshire,B.MusicandR.Graczyk(eds),CLS25:PapersfromtheTwentyFifthMeetingoftheChicagoLinguisticSociety,Chicago:ChicagoLinguisticSociety,pp.342–65.Perry,John(1986),“ThoughtwithoutRepresentation”,ProceedingsoftheAristotelianSociety,Supp.Vol.60:263–83.—(2001),ReferenceandReflexivity,Stanford:CSLIPublications.Predelli,Stefano(1998),“Iamnotherenow”,Analysis58:107–15.—(2002),“Intentions,indexicals,andcommunication”,Analysis62:310–16.Preyer,Gerhard,andGeorgPeter(eds)(2005),ContextualisminPhilosophy:Knowledge,Meaning,andTruth,Oxford:OxfordUniversityPress.—(2007),ContextSensitivityandSemanticMinimalism:NewEssaysonSemanticsandPragmatics,Oxford:OxfordUniversityPress.Raffman,Diana(1996),“Vaguenessandcontext-sensitivity”,PhilosophicalStudies81:175–92.Ravin,Yael,andClaudiaLeacock(eds)(2000),Polysemy:Theoreticalandcomputa-tionalapproaches,Oxford:OxfordUniversityPress.Recanati,François(2002),“Unarticulatedconstituents”,LinguisticsandPhilosophy25:299–345.—(2004),LiteralMeaning,Cambridge:CambridgeUniversityPress.—(2010),Truth-conditionalPragmatics,Oxford:OxfordUniversityPress.Reimer,Marga(1992),“Dodemonstrationshavesemanticsignificance?”Analysis51:177–83.—(2002),“Doadjectivesconformtocompositionality?”PhilosophicalPerspectives16:183–98.Richard,Mark(1990),PropositionalAttitudes:AnEssayonThoughtsandHowWeAscribeThem,Cambridge:CambridgeUniversityPress.Romdenh-Romluc,Komarine(2002),“NowtheFrenchareinvadingEngland!”Analysis62:34–41.Rothschild,Daniel,andGabrielSegal(2009),“Indexicalpredicates”,Mind&Language24:467–93.Schiffer,Stephen(1977),“Namingandknowing”,MidwestStudiesinPhilosophy2:28–41.—(1992),“Beliefascription”,JournalofPhilosophy89:499–521.—(2005),“Russell’stheoryofdescriptions”,Mind114:1135–83.Searle,John(1978),“Literalmeaning”,Erkenntnis13:207–24.1829781472578235_txt_print.indd18208/08/201415:46\nContextDependence—(1980),“Thebackgroundofmeaning”,inJ.R.Searle,F.KieferandM.Bierwisch(eds),SpeechActTheoryandPragmatics,Dordrecht:Reidel,pp.221–32.Sennet,Adam(2008),“Thebindingargumentandpragmaticenrichment,or,whyphilosopherscareevenmorethanweathermenabout‘raining’,”PhilosophyCompass3:135–57.Shapiro,Stewart(2006),VaguenessinContext,Oxford:OxfordUniversityPress.Sidelle,Alan(1991),“Theansweringmachineparadox”,CanadianJournalofPhilosophy21:525–39.Smith,Quentin(1989),“Themultipleuseofindexicals”,Synthese78:167–91,Soames,Scott(2009),“Thegapbetweenmeaningandassertion”,inPhilosophicalEssays,Volume1,NaturalLanguage:WhatItMeansandHowWeUseIt,Princeton,NJ:PrincetonUniversityPress,pp.278–97.—(2011),“Trueat”,Analysis71:124–33.Sperber,Dan,andDeirdreWilson(1986),Relevance:CommunicationandCognition,Cambridge,MA:HarvardUniversityPress.Stainton,Robert(2006),WordsandThoughts:Subsentences,Ellipsis,andthePhilosophyofLanguage,Oxford:OxfordUniversityPress.Stalnaker,Robert(1999),ContextandContent:EssaysonIntentionalityinSpeechandThought,Oxford:OxfordUniversityPress.Stanley,Jason(2000),“Contextandlogicalform”,LinguisticsandPhilosophy23:391–434.—(2002),“Makingitarticulated”,MindandLanguage17:149–68.—(2003),“Context,interest-relativity,andthesorites”,Analysis63:269–80.—(2005a),“Review:FrançoisRecanati,LiteralMeaning”,NotreDamePhilosophyReviews,availableonlineathttp://ndpr.nd.edu/news/24857-literal-meaning/—(2005b),“Semanticsincontext”,inG.PreyerandG.Peter(eds),ContextualisminPhilosophy:Knowledge,Meaning,andTruth,Oxford:OxfordUniversityPress,pp.221–53.Stanley,Jason,andZoltánGendlerSzabó(2000),“Onquantifierdomainrestriction”,MindandLanguage15:219–61.Stephenson,Tamina(2007),“Judgedependence,epistemicmodals,andpredicatesofpersonaltaste”,LinguisticsandPhilosophy30:427–525.Stevens,Graham(2009),“Utteranceatadistance”,PhilosophicalStudies143:213–221.Stokke,Andreas(2010),“Intention-sensitivesemantics”,Synthese175:383–404.Strawson,P.F.(1950),“Onreferring”,Mind59:320–44.Szabó,ZoltánGendler(2001),“Adjectivesincontext”,inI.KeneseiandR.M.Harnish(eds),PerspectivesonSemantics,Pragmatics,andDiscourse,Amsterdam:JohnBenjamins,pp.119–46.Taylor,Kenneth(2001),“Sex,breakfastanddescriptusinterruptus”,Synthese128:45–61.Travis,Charles(2000),UnshadowedThought,CambridgeMA,HarvardUniversityPress.Unger,Peter(1975),Ignorance:ACaseforScepticism,Oxford:OxfordUniversityPress.—(1984),PhilosophicalRelativity,Minneapolis:UniversityofMinnesotaPress.Weatherson,Brian(2002),“Misleadingindexicals”,Analysis62:308–10.1839781472578235_txt_print.indd18308/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageWright,Crispin(2008),“Relativismabouttruthitself:Haphazardthoughtsabouttheveryidea”,inManuelGarcía-CarpinteroandMaxKölbel(eds),RelativeTruth,Oxford:OxfordUniversityPress,pp.157–85.1849781472578235_txt_print.indd18408/08/201415:46\n8PragmaticsFrançoisRecanatiPragmaticsandOrdinaryLanguagePhilosophySemantics,thestudyofmeaning,andpragmatics,thestudyoflanguageinuse,aretwoimportantareasoflinguisticresearchthatowetheirshapetothegroundworkofphilosophers.Althoughthetwodisciplinesarecomplementary,thephilosophicalmovementsoutofwhichtheygrewwereverymuchincompetition.Inthemiddleofthetwentiethcentury,thereweretwoopposing“camps”withintheanalyticphilosophyoflanguage.Thefirstcamp–“ideallanguagephilosophy”,asitwasthencalled–wasthatofthepioneers,Frege,Russellandthelogicalpositivists.Theywere,firstandforemost,logiciansstudyingformallanguagesand,throughthem,“language”ingeneral.Workinthistradition(especiallythatofFrege,Russell,Wittgenstein,Carnap,TarskiandlaterMontague)gaverisetocontemporaryformalsemantics.Theothercampwasthatofso-called“ordinarylanguagephilosophers”,whoadvocatedamoredescriptiveapproachandemphasizedthe“pragmatic”natureofnaturallanguageasopposedto,say,the“language”ofPrincipiaMathematica.Theirownwork(especiallythatofAustin,Strawson,GriceandthesecondWittgenstein)gaverisetocontemporarypragmatics.Thepragmaticinvestigationsundertakenbyordinarylanguagephiloso-phershadbeennotablyanticipatedbyvariousresearchersbelongingtoothertraditions(phenomenologistssuchasMartyorReinach,linguistssuchasBallyorGardiner,psychologistssuchasBühler,andanthropologistssuchasMalinowski).However,whatinfluencedordinarylanguagephilosophersmostwastheconceptionoflanguageadvocatedby“ideallanguagephilosophers”,whichtheyreactedstronglyto.Centralintheideallanguagetraditionhadbeentheequationof,oratleastthecloseconnectionbetween,themeaningofasentenceanditstruth-condi-tions.Thistruth-conditionalapproachtomeaningisoneofthethingswhichordinarylanguagephilosophersfoundquiteunpalatable.Theirownemphasiswasonthedistinctionbetween“language”and“speech”(Gardiner1932)and,correlatively,between“sentence”and“statement”(Austin1950,Strawson1950).Itisthesentence(aunitof“language”)thathasmeaning,accordingtoordinarylanguagephilosophers;whereasitisthestatementmadebyuttering1859781472578235_txt_print.indd18508/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagethesentenceinaparticularcontextthathastruth-conditions.Thesentenceitselfdoesnothavetruth-conditions.Truthcanonlybepredicatedofsentencesindirectly,viatheconnectionsbetweenthesentenceandthespeechactitcanbeusedtoperform.Ratherthanequatingthemeaningofasentencewithitsallegedtruth-conditions,somephilosophersinthepragmatictraditionhavesuggestedequatingitwithits“speechactpotential”(whichmayinclude,asaproperpart,acertaintruth-conditionalpotential;seeAlston1964,2000).Supposethatwepositabstractobjects,namely“propositions”,whichareessentiallytrueorfalse.Thenthepointmadebyordinarylanguagephiloso-pherscanbeputbysayingthatsentencesdonotexpresspropositionsinvacuo,butonlyinthecontextofaspeechact.Giventhatthesamesentencecanbeusedtomakedifferentspeechactswithdifferentcontents,the“proposition”whichisthecontentofthespeechactmustbedistinguishedfromthelinguisticmeaningofthesentencequaunitofthelanguage(“sentencemeaning”).Itmustalsobedistinguishedfromthecontextuallydeterminedmeaningofaparticularutteranceofthesentence(“utterancemeaning”),forthisincludesmuchmorethanmerelythepropositionalcontentofthespeechactperformedinutteringthesentence.Utterancemeaningincludesarichnon-truth-conditionalcomponent:Besidesthepropositionitexpresses,anutteranceconveysindica-tionsconcerningthetypeofthespeechactbeingperformed,theattitudesofthespeaker,theplaceoftheutterancewithinthediscourse,itspresuppositions,andsoforth.Moreoverthereisasecondarylayerofmeaningwhichincludesthecontextualimplicationsofthespeechact,andinparticularwhatPaulGrice,himselfanordinary-languagephilosopher,calledthe“conversationalimplica-tures”oftheutterance.SpeechActsSpeechacttheory(Austin1975,Searle1969)isconcernedwithcommunication–notcommunicationinthenarrowsenseofthetransmissionofinformation,butcommunicationinabroadersensewhichincludestheissuingoforders,theaskingofquestions,themakingofapologiesandpromises,etc.Accordingtothetheory,aspeechactismorethanmerelytheutteringofagrammaticalsentenceendowedwithsenseandreference.Tospeakisalsotodosomethinginafairlystrongsense–itistoperformwhatAustincalledan“illocutionaryact”.Inperforminganillocutionaryact,aspeakertakesonacertainroleandassignsacorrespondingroletothehearer.Bygivinganorder,thespeakerexpresseshisdesirethatthehearerfollowacertaincourseofconductandpresentshimselfashavingtherequisiteauthoritytoobligethehearertofollowthecourseofconductinquestionsimplybecauseitisthespeaker’swill.Thesocialroletakenonbythespeakerwhogivesanorderisembodiedintheorganizational1869781472578235_txt_print.indd18608/08/201415:46\nPragmaticsnotionof“superiorrank”.Austinstressedsuchinstitutionalembodimentsofillocutionaryrolesinordertoshowthatlanguageitselfisavastinstitutionincorporatinganarrayofconventionalrolescorrespondingtotherangeofsociallyrecognizedillocutionaryacts.Fromthispointofview,assertion,theactofmakingastatement,isonlyoneillocutionaryactamongmanyothers.Illocutionaryacts(includingassertions)have“felicityconditions”–conditionsthatmustbecontextuallysatisfiedfortheillocutionaryacttobesuccessfullyperformed.Thestudyoffelicityconditionsisacentralconcernofspeechacttheory,alongwiththetaxonomyofillocutionaryacts.Butthemostcentralconcern,perhaps,relatestothecharacterizationoftheverynotionofanillocutionaryact.Illocutionaryactsaregenerallyintroducedostensively,byexamples–asIdidabove;andtheyaredistinguishedbothfromtheactofmerelysayingsomething(“locutionary”act)andfromtheactofcausingsomethingtohappenbysayingsomething(“perlocutionary”act,e.g.frightening,convincing,etc.).Thenatureoftheintermediatecategoryof“illocutionaryacts”remainsamatterofdebate,however.Thepioneersofspeechacttheory,AustinandSearle,advocatedaninstitutionalorconven-tionalapproach.Inthisframeworktheillocutionaryactsperformedinspeech,liketheactsthatareperformedingames(e.g.“winningaset”intennis),aregovernedbyrules,andexistonlyagainstabackgroundofconventions.Butanalternative,“intentionalist”view,originatingfromGrice(1957)andStrawson(1964),developedandhasbeenforsometimethedominanttrendinspeechacttheory(seecommunicativeintentionsonpages199–200).ContextualImplicationsThenotionofacontextualimplicationitselfisaspeech-acttheoreticnotion.If,besidethemeaningorcontentofanutterance,thereisanotherrealm,viz.thatoftheillocutionaryacttheutteranceservestoperform,thenalongwiththeimplicationsofwhatissaidtherewillbeafurthersetofimplicationsderivablefromtheutterance,namelytheimplicationsoftheillocutionaryactitself.Someofthese“pragmatic”implicationsarefairlytrivial.Thus,accordingtoordinarylanguagephilosophers,itisaruleofthelanguagegameofassertionthatwhoeverassertssomethingbelieveswhatshesaysandhassomeevidenceforit;eventheliar,whodoesnotobeythisrule,hastopretendthatshedoesifshewantstoparticipateinthegame.Thisrulegeneratespragmaticimplications:Byassertingsomethingandthereforeengaginginthelanguagegame,thespeaker“implies”thatsheobeystherulesofthegameand,therefore,thatshebelieveswhateversheisasserting.Thespeakercannotdisavowtheseimplicationsofherspeechactwithout“pragmaticcontradiction”.Apragmaticcontradictionisaconflictbetweenwhatanutterancesaysandwhatitpragmaticallyimplies.1879781472578235_txt_print.indd18708/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageThusMoore’sfamousparadoxicalutterance,“ItisrainingbutIdonotbelieveit”,isnotself-contradictoryinthelogicalsense:thestateofaffairsitdescribesislogicallypossible.Butthespeaker’sassertingthatitrainsimpliesthatshebelievesit,andthiscontradictsthesecondpartoftheutterance.(Thetwinnotionsofpragmaticimplicationandpragmaticcontradictionor“pragmaticparadox”havebeenusedtoilluminateavarietyofphilosophicalissues,includingthenatureofDescartes’scogito.)LesstrivialarethecontextualimplicationsdiscussedbyGriceunderthelabel“conversationalimplicatures”(Grice1989).AccordingtoGrice,thespeakermakinganutterancedoesnotmerelyimplythatsherespectstherulesofthelanguagegame;amongthepragmaticimplicationsoftheutterance,wefindanumberofadditionalassumptionscontextuallyrequiredinordertomaintainthesuppositionthattherulesofthegamearebeingobserved.SupposethatIamaskedwhetherIwillgoout;Ireply,“Itisraining”.AsIsaidabove,itisaruleofassertionthattheassertorbelieveswhathesaysandhassufficientevidenceforit.Byvirtueofthisrule,myutteranceimpliesthatIbelievethatitisraining,andthatIhavesomeevidenceformyassertion.Consideredasananswertoaquestion,myutterancealsoimpliesthatitprovidestheinformationrequestedbytheaddressee,foritisaruleoftheQuestion-and-Answerlanguagegamethattheanswerermustprovidetherequestedpieceofinformation.Nowinordertomaintainthesuppositionthatthespeaker’sutteranceactuallyprovidestherequestedinformation,additionalpremissesareneeded,forexampletheassumptionthatthespeakerwillnotgooutifitrains.Inconjunctionwiththiscontextualassumption,theutteranceimpliesthatthespeakerwillnotgoout,therebyprovidinganegativeanswertothequestion.Insofarastheyservetorestoretheutter-ance’sconformitytotherulesofthegame,theconclusionthatthespeakerwillnotgooutandthecontextualassumptionthroughwhichitisderivedarefurtherpragmaticimplicationsoftheutterance.Gricecalledthem“conversa-tionalimplicatures”.Contrarytothemoretrivialpragmaticimplications,theycanbedisavowedbythespeakerwithoutpragmaticcontradiction(atleastifthereisanotherwayofmakingtheutterancecompatiblewiththesuppositionthattherulesofthegamearebeingrespected).Thisdistinguishingfeatureofconversationalimplicaturesisreferredtoastheir“cancelability”.Implicaturesthatarenotdisavowedarelegitimatelytakenaspartofwhattheutterancecommunicates.Theyconstituteasecondlayerofmeaning,additionaltowhatisliterallysaid.Grice’stheoryoflayersofmeaninghasprovedveryfruitfulbothinlinguisticsandinphilosophy.Itsstrategicimportancewillbestressedbelow(seepages198–99).1889781472578235_txt_print.indd18808/08/201415:46\nPragmaticsNon-Truth-ConditionalAspectsofMeaningLikepragmaticimplications,non-truth-conditionalaspectsofmeaningareeasytoaccountforifspeechisconsideredasarule-governedactivity(Stenius1967).Whatisthemeaningof,say,theimperativemood?Arguably,thesentences“Youwillgotothestoretomorrowat8”,“Willyougotothestoretomorrowat8?”and“Gotothestoretomorrowat8”representthesamestateofaffairs.Thedifferencebetweenthemispragmaticratherthandescriptive;itrelatestothetypeofillocutionaryactperformedbytheutterance.Thustheimperativemoodindicatesthatthespeaker,inutteringthesentence,performsanillocu-tionaryactofa“directive”type.(SuchanactisgovernedbytherulethatifthespeakerperformsadirectiveactwithcontentP,theaddresseeistomakeitthecasethatP.)Toaccountforthis“indication”,whichdoesnotbelongtotheutterance’sdescriptiveorpropositionalcontent,wecanpositaruleorconventiontotheeffectthattheimperativemoodistobeusedonlyifoneisperformingadirectivetypeofillocutionaryact.Thisrulegivesconditionsofusefortheimperativemood.Byvirtueofthisrule,aparticulartokenoftheimperativemoodinanutteranceuindicatesthatadirectivetypeofspeechactisbeingperformedbyu.This(token-reflexive)indicationconveyedbythetokenfollowsfromtheconditionsofusethatgovernthetype,whichconditionsofuseconstitutethelinguisticmeaningofthetype(Recanati1987:15–17).Pragmaticindicationsareaspeciesofpragmaticimplication:theyarewhattheuseofaparticularexpressionpragmaticallyimplies,byvirtueofacertainconditionofuseconventionallyassociatedwiththeexpression.Incontrasttomorestandardpragmaticimplications,however,pragmaticindicationsarelinguisticallyencoded,viatheconditionofuseconventionallyassociatedwiththeexpression.Gricecalledsuchconventionalpragmaticimplications“conventionalimplicatures”,asopposedtoconversationalimplicatures(Grice1989).Whethertheyconcernthetypeoftheillocutionaryact,asintheexampleIhavegiven,orsomeotheraspectofthecontextofutterance,pragmaticindicationscanalwaysbeaccountedforintermsofconditionsofuse.Theyare“use-conditional”aspectsofmeaning.Theirexplorationisoneoftheempiricaltasksofsemanticsconstruedasthestudyoflinguisticmeaningunderallitsaspects.(SeePotts2005forrecentdevelopmentsinthisarea.)IndexicalsUse-conditionalmeaningisnotincompatiblewithdescriptivecontent,inthesensethatoneandthesameexpressioncanbeendowedwithboth.Thereareexpressionsthathaveapurelyuse-conditionalmeaninganddonotcontributetotruth-conditionalcontent.Illocutionarymarkerssuchastheimperative1899781472578235_txt_print.indd18908/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagemood,ordiscourseparticlessuchas“well”,“still”,“afterall”,“anyway”,“therefore”,“alas”,“oh”andsoforth,fallintothiscategory.Thusthefollowingutteranceshavethesametruth-conditionalcontent,andaredistinguishedonlybythepragmaticindicationstheyrespectivelyconvey:Well,PeterdidnotshowupStill,PeterdidnotshowupAfterall,PeterdidnotshowupTherefore,PeterdidnotshowupAlas,PeterdidnotshowupOh,PeterdidnotshowupButtherearealsoexpressionsthathaveatwo-layeredmeaning.Indexicalsareacaseinpoint.Aruleofuseisclearlyassociatedwithindexicals:thus“I”isgovernedbyaconventionofuse–itistobeusedtorefertooneself.Byvirtueofthisconventionalrule,auseuof“I”token-reflexivelyindicatesthatitreferstothespeakerofu.Butualsocontributestotheutterance’struth-conditionalcontent.“I”beingadirectlyreferentialexpression,itstruth-conditionalcontri-bution(its“content”)isitsactualreferent,nottheruleofusethatcontextuallydeterminesthereferent(Kaplan1989a).Besidethehorizontaldistinctionbetweentruth-conditionalandnon-truth-conditionalaspectsofmeaning,weseethatthereisaverticaldistinctionbetweentwolevelsofmeaningforindexicalexpressions(Strawson1950,Kaplan1989a).Atthefirstlevel–correspondingtothelinguisticmeaningoftheexpression-type–wefindtheruleofuseconventionallyassociatedwiththeexpression.Atthesecondlevel–correspondingtothecontext-dependentsemanticvalueofthetoken–theruleofusedeterminestheexpression’s“content”.LevelsofMeaningThetwodistinctionswehavemade,betweentruth-conditionalandnon-truth-conditionalaspectsofmeaningontheonehand,andbetweenlevelsofmeaningontheotherhand,cross-cuteachother.Inthesamewayasthetruth-condi-tionalcontentofanindexicalsentenceiscontext-dependentand,therefore,belongstothesecondlevelofmeaning,thepragmaticindicationsconveyedbyanexpressiongovernedbyaruleofusealsoarecontext-dependentandbelongtothesecondlevelofmeaning.Inotherwords,adistinctionmustbemadebetweentheruleofuse(firstlevelofmeaning)andthepragmaticindicationsitcontextuallygenerates,justaswedistinguishbetweentheruleofuseandthetruth-conditionalcontentitcontextuallydetermines.1909781472578235_txt_print.indd19008/08/201415:46\nPragmaticsThatpragmaticindications,thoughconventional,arecontext-dependentisshownbyexamplessuchas(1):(1)Theweatherisnice,butIhavealotofworkTheconjunction“but”isgovernedbyacertainconditionofusewhichdistin-guishesitfrom“and”.AccordingtoDucrot,“but”istobeusedonlyifthefollowingconditionsarecontextuallysatisfied(Ducrot1972:128ff):(i)Thefirstconjunct(P)arguesinfavourofacertainconclusionr;(ii)Thesecondconjunct(Q)arguesinfavourofnot-r;(iii)QisconsideredstrongerthanP,i.e.thewholeutterancearguesinfavourofnot-r.Uttering(1)pragmaticallyimpliesthattheconditionsofuseassociatedwith“but”aresatisfied,i.e.thatthereisaconclusionrsuchthatthefirstconjunctarguesinfavourofrandthesecondconjunctprovidesastrongerreasontoconcludethatnot-r.Butthepragmaticimplicationconveyedbyaparticularutteranceof(1)ismuchmorespecific.Incontext,thevariable“r”isassignedaparticularinterpretation,e.g.weshouldgoforawalk.(1)thereforepragmaticallyimpliessomethinglike(2):(2)Weshouldnotgoforawalk(becauseofalltheworkIhavetodo),despitetheniceweather.Insofarasitiscontext-dependentandconveyedbythetoken,thispragmaticimplicationistobelocatedatthesecondlevelofmeaning,alongsidethecontentofindexicals.Eveninacaseinwhichthepragmaticindicationisfullyconventionalandnotinneedofcontextualspecification,itisconveyedbythetoken,notbythetype.Thusaparticularuseuofthepronoun“I”indicatesthatit(u)referstothespeakerofu.Thistoken-reflexiveindicationisdistinctfromtheruleofuse,totheeffectthatforallx,ifxisatokenof“I”itmustbeusedtorefertothespeakerofx.ThepictureisfurthercomplicatedbytheGriceandistinctionbetweenwhatisliterallysaidandwhatisnon-literallyorindirectlycommunicated.Weenduphavingathree-folddistinctionbetweenthefollowinglayersofmeaning:Firstlevel:rulesofuse(“character”,inKaplan’sframework)Secondlevel:truth-conditionalcontent+pragmaticindicationsThirdlevel:conversationalimplicaturesTheneedforathirdlevelofmeaningcomesfromthefactthatthecontextual1919781472578235_txt_print.indd19108/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageprocessresponsibleforconversationalimplicatures(andnon-conventionalpragmaticimplicationsingeneral)takesthesecond-levelmeaningoftheutteranceasinput.Whenanexpressionisgovernedbyaconditionofuse,usingthatexpressionpragmaticallyimpliesthattheconditionissatisfied.Butconversationalimplicatures,inconstrasttoconventionalimplicatures,arenotgeneratedbyvirtueofaconditionofusedirectlyassociatedwithaparticularlinguisticexpression;theyarenormallygeneratedbyvirtueofconversationalnormswhichconcernthecontentofutterancesratherthantheexpressionswhichareusedtoconveythatcontent.Forexample,aspeakershouldnotsaywhathebelievestobefalse(“maximofquality”,inGrice’sterminology);asaresult,sayingthatPpragmaticallyimpliesthatthespeakerbelievesthatP.Thegenerationofthispragmaticimplicationpresupposesthatthepropositionexpressedhasbeenidentified:fromthefactthatthespeakerhassaidthatP,togetherwiththedefaultassumptionthatthemaximofqualityisrespected,wecaninferthatthespeakerbelievesthatP.Theimplicature-generatingprocessthereforedeservestobecalleda“secondarypragmaticprocess”(Recanati1993,2004).Therearethreebasiclevelsofmeaning,withthecontextcontrollingthetransitionfromthefirstonetothesecondandfromthesecondtothethird.Thepropositionexpressedbytheutterancemustfirstbecontextuallyidentified(primarypragmaticprocess)inorderforthenon-conventionalpragmaticimplicationstobederived(secondarypragmaticprocess).OpenTextureForordinarylanguagephilosophers,thetruth-conditionalor“descriptive”contentofanutteranceisapropertyofthespeechact,notapropertyofthesentence.Asentenceonlyhastruth-conditionsinthecontextofaspeechact.Thisissonotmerelybecauseofindexicality–thefactthatthereferenceofsomewordsdependsonthecontextofutteranceinasystematicway.Indexicalityisonlyoneformofcontext-dependence.Thereisanotherone,nolessimportant,whichaffectsthesense(theconditionsofapplication)ofwordsratherthantheirreference.AccordingtoAustinandWittgenstein,wordshaveclearcondi-tionsofapplicationonlyagainstabackgroundof“normalcircumstances”correspondingtothetypeofcontextinwhichthewordshavebeenusedinthepast.Thereisno“convention”toguideusastowhetherornotaparticularexpressionappliesinsomeextraordinarysituation.Thereasonforthisisthattheapplicationofwordsultimatelydependsontherebeingasufficientsimilaritybetweenthenewsituationofuseandpastsituations.Therelevantdimensionsofsimilarityarenotfixedonceforall–thisiswhatgenerates“opentexture”(Waismann1951).Ultimately,itisthecontextofutterancewhichdeter-mineswhichdimensionofsimilarityisrelevant,hencewhichconditionshave1929781472578235_txt_print.indd19208/08/201415:46\nPragmaticstobesatisfiedforagivenexpressiontoapply(Recanati2004:ch.9).Itfollowsthatthesenseofordinarydescriptivewordsiscontext-dependent,likethereferenceofindexicals,thoughnotquiteinthesameway.Onthis“contextu-alist”approachadvocatedbyTravis(1975,1981,2008)andSearle(1978,1983),truth-conditionscannotbeascribedtosentence-typesbutonlytoutterances.Contextualismwasacentraltenetinthepragmaticconceptionoflanguagedevelopedbyordinarylanguagephilosophers(thoughsomeatypicalordinarylanguagephilosophers,suchasGrice,rejectedit).TheSemantics/PragmaticsDistinctionIfmuchofcontemporarypragmaticsderivesfromtheworkofordinarylanguagephilosophers,thename“pragmatics”–contrastedwith“semantics”–wascoinedbyaphilosopherintheideallanguagetradition,CharlesMorris(Morris1938).ThegeneralconceptionassociatedwithMorris’sdistinction(hereafter“thetraditionalconception”)hasbeenveryinfluential,thoughitisnowoutdated.Onthetraditionalconception,thesemantics/pragmaticsdistinctionrestsontwomorebasicdistinctions:(a)semanticsstudiestherelationsbetweenwordsandtheworld(i.e.,whatwordsdenoteorrepresent),whilepragmaticsstudiestherelationsbetweenwordsandtheirusers(e.g.whatwordsareusedtodo);(b)semanticsdealswiththestandingmeaningwhichexpression-typespossessinvirtueoftheconventionsofthelanguage,whilepragmaticsisconcernedwithwhatthespeakermeanswhensheutterssomethinginaparticularcontext.Onthisconception,thelinguisticmeaningofthesentenceistakentobeitsrepresentationalcontent,andthecontextistakentocontributeonlyinfor-mationpertainingto“therelationsbetweenthewordsandtheirusers”.Aswehaveseen,however,theequationofthemeaningofthesentencewithitsrepresentationalcontentcannotbeaccepted,fortworeasons:1Becauseofindexicality,open-texture,etc.,sentencesrepresentaspecificstateofaffairsonlyincontext;theirrepresentationalcontentdependsonsomefeaturesoftheircontextofuse,henceitcannotbeequatedwiththecontext-independentmeaningofthesentence-type.2Thereareuse-conditionalaswellastruth-conditionalaspectsofthemeaningofsentence-types.Toaccountforthem,itseemswemustgiveupthepurelytruth-conditionalconceptionofsemanticsandmakeroomforanon-truth-conditionalcomponentwithinthelatter.Facedwitheachoftheseobjections,therehavebeenattemptstodefendthetraditionalconception,oratleastitsspirit.Asforthefirstobjection,ithasbeen1939781472578235_txt_print.indd19308/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagearguedthattheconventionalmeaningofasentencedeterminesitsrepresenta-tionalcontent,evenifitdoessoonly“withrespecttocontext”.Therolecontextplaysinthedeterminationofcontentissaidtobeminimal,incontrasttotheroleitplaysinthedeterminationofspeaker’smeaning.Asforthesecond,therehavebeenattemptstoforceuse-conditionalaspectsofmeaningintothemouldoftruth-conditionalsemantics.Take,forexample,theimperativemood.Onecanusethepragmaticequivalencebetweentheimperative“Closethedoor!”andthe“explicitperformative”“Iaskyoutoclosethedoor”tosupporttheclaimthatnon-declarativesentenceshaveadeclarativeparaphrasethroughwhichtheycanbegivenatruth-conditionalanalysis(seeLewis1970:54–61).Noneoftheseattemptshavebeenquitesuccessful.Asforthefirst,thesharpdistinctionbetweenspeaker’smeaningand“whatissaid”hasbeencalledintoquestion.Ithasbeenshownthatweneedtoappealtowhatthespeakermeans(whathe“hasinmind”)tofixthevalueofindexicals,hencethecontentofwhatissaid(seepage196).Eventhe“minimalistprinciple”,accordingtowhichcontextcomesintoplayinthedeterminationofsemanticcontentonlywhensomeexpressionconventionallyrequiresit,hasbeenquestioned.(Onthesedebates,seeCarston2002;Borg2004;Recanati2004,2010;CappelenandLepore2005;Predelli2005;Stanley2007.)Asforthesecond,itiscommonlyacknowl-edgednowadaysthatnotallaspectsoflinguisticmeaningaredescriptiveortruth-conditional.Withthedownfallofthetraditionalconception,thesemantics/pragmaticsdistinctionhaslostmuchofitsbite,anddifferentauthorsdrawitdifferently.(Someauthorsproposetodispensewithitaltogether–seee.g.Cappelen2007.)Still,thereissomecontinuitywiththetraditionalconception.Asthe“PragmaticsvsSemantics”entryoftheHandbookofPragmatics(HornandWard2004)putsit,themorecontext-dependentanaspectofmeaningis,andthelessdescriptiveortruth-conditionalitis,thelesswearetemptedtocallit“semantic”andthemorewearetemptedtocallit“pragmatic”.Butnothingmorespecificthanthatcanbesaidwithoutrelyingonstipulation,andforthatreasonthe“borderwars”overthesemantics/pragmaticsinterfaceareoftensuspectedofbeingverbaldisputes.ContextandPropositionalAttitudesThedescriptivecomponentofsentencemeaningcanbeequatedwithafunctionfromcontextstopropositions(Stalnaker1970,1999;Kaplan1989a).The“context”isoftenconstruedasapackageofvarioussituationalfactorsrelevanttodeterminingthesemanticvaluesofthecontext-sensitiveconstit-uentsofthesentence.(Seethenexttwosectionsbelow,pages196–98,foranalternativenotionofcontext.)Thustheplaceofutterance,theidentityof1949781472578235_txt_print.indd19408/08/201415:46\nPragmaticstheparticipantsinthespeechepisodeandthetimeofutteranceareamongthefactorsonwhichthepropositionexpressedbyanindexicalsentencedepends.Itwouldbeamistake,however,toholdthatonlysuch“external”(i.e.non-intentional)featuresofthesituationofutterancehavearoletoplayinthedeterminationofwhatissaid.Inmanycases,whatthespeaker“hasinmind”istherelevantfactor.Thus“John’sbook”canmeanthebookthatJohnwrote,thebookthathebought,thebookheisreading,orwhatever.Thesentenceinwhichtheexpression“John’sbook”occursexpressesadefinitepropositiononlywhenaparticularrelationbetweenJohnandacertainbookhasbeencontextuallydetermined,butthereisno“rule”whichenablestheinterpretertodeterminethelatterexceptthatitmustbetherelationwhichthespeaker“hasinmind”.Likewise,inthecaseofdemonstratives,oneofthecrucialcontextualfactorsseemstobethereferentialintentionthatthespeakermanifestsbytheutteranceofthedemonstrativeandtheassociatedgestures(Kaplan1989b;Bach1992).Thissuggeststhatthecontextagainstwhichanutteranceisinterpretedincludesfactorssuchastheintentions,expectations,beliefsandotherproposi-tionalattitudesofthespeakerandheraudience.Especiallyimportant,onthatsortofview,arethebeliefsthataresharedand“mutuallyknown”tobeshared;theyconstitutea“commonground”whichcanbeexploitedindiscourse(Stalnaker1999,2002;Clark1992).Thisintroducesustothetopicof“presup-position”whichisgenerallymentioned,alongwithspeechacts,indexicalsandimplicatures,asoneofthecentralissuesinpragmatics.PresuppositionThereisabasicsenseinwhichpresupposingisapragmaticattitudetowardsaproposition–theattitudeconsistingintakingitforgranted.The“context”issometimesdefinedasasetofpresuppositionsinthissense,i.e.asetofproposi-tionsthataretakenforgrantedatagivenpointindiscourse(seee.g.Karttunen1974;Stalnaker1974).Inordertobepartofthecontextthusdefined,apresup-positionmustnotonlybesharedbytheparticipantsinthespeechepisode,butalsorecognizedbythemtobesoshared.Besidethepragmaticnotionofpresupposition(wherepresupposingissomethingaspeechparticipantdoes),istherealsoapurelysemanticnotion,wherepresupposingissomethingwhichasentencedoes?Forexample,doesthesentence“Johnstoppedteachingundergraduates”carrythepresuppositionthatJohnusedtoteachundergraduatesaspartofitssemanticcontent?Thesemanticnotionofpresuppositionhasbeenquestionedontwogrounds.First,ithasbeenpointedoutthatpresuppositions,likeconversationalimpli-catures,seemtobedefeasibleorcancelable.Thismightsuggestthatthebasic,1959781472578235_txt_print.indd19508/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagepragmaticsenseistheonlysensewecangivetothenotionofpresupposition.Sentencesdonothavepresuppositions;onlytheparticipantsinaspeechepisodecanpresupposesomething.Thisconclusion,however,istoostrong,fortheconventionalnatureof(some)presuppositionsismanifestandwell-documented.Arguably,whatthedefeasibilityofpresuppositionsshowsisnotthatpresuppositionsarenon-conventional–noonedoubtsthatsomepresup-positionsareencodedandbelongtothelinguisticmeaningofexpressions–butratherthattheycanbeoverriddenifcertainconditionsaresatisfied(Gazdar1979).Anotherissueiswhetherpresuppositionsaffectthetruth-conditionalcontentofutterancesandshouldbecharacterizedprimarilyinsemanticterms.AccordingtoStalnaker,thatneednotbethecase.Likepragmaticindicationsingeneral,thelinguisticpresuppositionsassociatedwithcertainexpressions(e.g.theverb“stop”inthepreviousexample)canbeconstruedasconditionsofuseorconstraintsonthecontext.Thelinguisticpresuppositionencodedbytheverb“stop”isacertainconstraintonthecontext,namelytherequirementthatthespeakerofthecontextpresupposeacertainproposition,or(ifthecontextisdirectlyconstruedasasetofpropositionsor“commonground”)therequirementthatitcontainacertainproposition,namelythepropositionthatJohnusedtoteachundergraduates.Anutteranceof“Johnstoppedteachingundergraduates”isappropriateorfelicitousonlyinacontextinwhichthisconstraintissatisfied.InterpretationandContext-ChangeThenotionof“context”ascommongroundthatfeaturesindiscussionsofpresuppositionmakesitpossibletoseetherelationbetweencontextandcontentastwo-wayratherthanone-way.Thepropositionexpressed,whichdependsonthecontext,itselfchangesthecontext.Accordingtodynamictheoriesofdiscourse,thecontentofanassertionisnormallyfedintothecontextagainstwhichthenextutterancewillbeinterpreted(Karttunen1974;Stalnaker1978;Kamp1985;Heim1988).Thecontextofinterpretationconstantlychanges–Stalnakerspeaksofan“everchangingcontext”–becauseitevolvesasdiscourseproceeds.Thusitispossibleforthecontexttoshiftinthemiddleofanutterance.Thispossibilityistakentoaccountforanumberofpuzzlingfacts,includingthedefeasibilityofpresuppositionsincomplexsentences(seeSchlenker2008foracriticaldiscussion,andforarevisedversionofthepragmaticapproachtothe“projectionproblem”pioneeredbyStalnaker).Ifthepropositionexpressedbyanutteranceisnormallyfedintothecontext,theassumptionthatthispropositionhasbeenexpressedalwaysbecomespartofthecontextasaresultoftheinterpretationoftheutterance.Itisthis1969781472578235_txt_print.indd19608/08/201415:46\nPragmaticsassumption,togetherwiththedefaultassumptionthatthespeakerrespectsthenormsofconversation(plusvariousotherassumptionsincludedinthecontext),whichmakeitpossibletoinfertheconversationalimplicaturesthatenrichtheoverallmeaningoftheutterance.Itfollowsthatthecontextualchangesinducedbyanutterancebyvirtueofitsexpressingacertainpropo-sitionaffectnotonlytheinterpretationoftheutterancesthatfollow,butequallytheoverallmeaningoftheveryutteranceresponsibleforthecontextualchange.Anothersortofcontext-changeinducedbyanutterancehasbeendescribedbyDavidLewis(1979).Sometimesthedefaultassumptionthatthespeakerrespectsthenormsofconversationpreventstheutterancefrombeinginter-pretedwithrespecttothecontextathandbecause,ifitweresointerpreted,itwouldviolatethenormsinquestion.Thisleadstoamodificationofthecontextinordertoreachamoresatisfactoryinterpretation.ThusiftheutterancepresupposesthatP,andPisnotpartofthecontextathand,itisintro-ducedintothecontextinordertorendertheutteranceconformtothenorms(“accommodation”).TheStrategicImportanceofConversationalImplicaturesIfsemanticsandpragmaticsbothstudythecontextualdeterminationofthepropositionexpressed(insofarasitdependsbothonthelinguisticmeaningofthesentenceandonthecontext),conversationalimplicaturesfallwithinthesoledomainofpragmatics,fortheyarenotconstrainedbythelinguisticmeaningofthesentenceinthewaythepropositionexpressedis.Yetthetheoryofimpli-catureshasimportantconsequencesforsemantics.ThankstoGrice’stheory,manyintuitiveaspectsofmeaningcanbeputintothe“pragmaticwastebasket”asimplicatures,ratherthantreatedasgenuinedataforsemantics.Take,forexample,thesentence[PorQ].Itcanreceiveaninclusiveoranexclusiveinter-pretation.Insteadofsayingthat“or”isambiguousinEnglish,wemayconsideritasunambiguouslyinclusive,andaccountfortheexclusivereadingbysayingthatinsomecontextstheutteranceconversationallyimplicatesthat[P]and[Q]arenotbothtrue.Whenthereissuchaconversationalimplicature,theoverallmeaningoftheutteranceisclearlyexclusive,eventhoughwhatisstrictlyandliterallysaidcorrespondstothelogicalformula[PvQ].Itisherethatthecomplementarycharacterofsemanticsandpragmaticsisparticularlymanifest.Semanticsissimplifiedbecausealotofdatacanbeexplainedawayas“impli-catures”ratherthangenuineaspectsofthe(literal)meaningoftheutterance.Grice’stheoryofimplicatureshasbeenextremelypopularamongseman-ticistspreciselybecauseitenablesthetheorist,whencertainconditionsaresatisfied,toshifttheburdenofexplanationfromsemanticstopragmatics.Fromthispointofview,themostinterestingnotionisthatofageneralized1979781472578235_txt_print.indd19708/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageconversationalimplicature(Grice1989;Gazdar1979;Levinson2000).Whenaconversationalimplicatureisgeneralized,i.e.generatedbydefault,ittendstobecomeintuitivelyindistinguishablefromsemanticcontent.Grice’stheoryhastaughtthesemanticistnottotakesuch“semantic”intuitionsatfacevalue.Evenifsomethingseemstobepartofthesemanticcontentofanutterance,thepossibilityofaccountingforitpragmatically,intermsofconversationalimpli-cature,mustalwaysbeconsidered.Grice’stheoryisimportantalsobecauseithasprovidedaninfluentialargumentagainstthecontextualismprofessedbyordinarylanguagephiloso-phers.Forexample,Strawsonhadclaimedthatthetruth-conditionsof[PandQ]inEnglisharecontextuallyvariable:thenotionoftemporalsuccession,orthatofcausalconnection,oranumberofothersuggestionsconcerningtheconnectionbetweenthefirstandthesecondconjunctcanenterintotheinter-pretationof[PandQ],dependingonthecontext(Strawson1952:81–2).“Theygotmarriedandhadmanychildren”meansthattheyhadchildrenaftergettingmarried;“Socratesdrankthehemlockanddied”meansthathediedasresultofdrinkingthehemlock.Thoseaspectsoftheinterpretationareverymuchcontext-sensitive;yettheyaffecttheutterance’struth-conditions.Thetruth-conditionsof[PandQ],therefore,arenotfixedbyarigidrule,butdependonthecontext.Asagainstthisview,Gricehasarguedthatthetruth-conditionsof[PandQ]arefixedandcontext-independent.[PandQ]istrueifandonlyif[P]and[Q]arebothtrue.Thus“Theygotmarriedandhadmanychildren”wouldbetrue,eveniftheyhadthechildrenbeforegettingmarried.Tobesure,theutteranceconveysthesuggestionthatthechildrencameafterthemarriage.Butthissuggestionisnothingotherthanaconversationalimplicature,accordingtoGrice.Itdoesnotaffecttheutterance’ssemanticcontent–itsliteraltruth-condi-tions.Gricecriticisedhisfellowordinarylanguagephilosophersforconfusingthetruth-conditionsofanutterancewithitstotalsignificance.Thoughcontro-versial(Travis1985;Recanati1994,2004),thisargumenthasbeenverypopular,andithasplayedamajorroleinthesubsequentdownfallofordinarylanguagephilosophy.CommunicativeIntentionsThepioneersofpragmatics(e.g.MalinowskiandAustin)usedtoinsistonthesocialdimensionoflanguageasopposedtoitscognitiveorrepresentationalfunction.Aspragmaticsdeveloped,however,itisthepsychologicaldimensionoflanguageusethatcametotheforefrontofdiscussions,inpartasaresultofGrice’sworkonmeaningandcommunication.Inhisfamous1957article,Gricedefinedapragmaticnotionofmeaning:thenotionofsomeone’smeaningsomethingbyapieceofbehaviour(agesture,an1989781472578235_txt_print.indd19808/08/201415:46\nPragmaticsutterance,orwhatnot).Grice’sideawasthatthispragmaticnotionofmeaningwasbasicandcouldbeusedtoanalysethesemanticnotion,i.e.whatitisforalinguisticexpressiontohavemeaning.StrawsonsoonpointedoutthatGrice’spragmaticnotionofmeaningcouldalsobeusedtocharacterizetheelusivenotionofanillocutionaryact(Strawson1964).InthesectionSpeechacts(page188),wementionedtheviewthatillocutionaryactsareessentiallyconven-tionalacts(liketheactswhichowetheirexistencetotherulesofaparticulargame).ThisconventionalistapproachwasdominantinspeechacttheoryuntilStrawsonestablishedabridgebetweenGrice’stheoryofmeaningandAustin’stheoryofillocutionaryacts.Illocutionaryacts,inthenewframework,canbeanalysedintermsoftheutterlynon-mysteriousnotionofaperlocutionaryact.Aperlocutionaryactconsistsinbringingaboutcertaineffectsbyanutterance.Forexample,bysayingtoyou“Itisraining”,Ibringitaboutthatyoubelievethatitisraining.Now,accordingtothesuggestedanalysis,toperformtheillocutionaryactofassertingthatitisrainingis(inpart)tomakemanifesttotheaddresseeone’sintentiontobringitabout,bythisutterance,thattheaddresseebelievesthatitisraining.(Thisisnotthefullstory,ofcourse.)Anillocutionaryactthereforeinvolvesthemanifestationofacorrespondingperlocutionaryintention.ButthereisaspecialtwistthatthesuggestedanalysisinheritsfromGrice’soriginalconceptionofmeaning:theintentionmustbemademanifestinaspecially“overt”manner.Notonlymustthespeaker’sintentiontobringaboutacertainbeliefintheaddresseeberevealedbyhisutterance,buthisintentiontorevealitmustalsoberevealed,anditmustberevealedinthesameovertmanner.Thischaracteristic(ifpuzzling)featureofovertnessisoftencapturedbyconsideringtherevealedintentionitselfasreflexive:Acommuni-cativeintention,i.e.thetypeofintentionwhosemanifestationconstitutestheperformanceofanillocutionaryact,istheintentiontoachieveacertainperlo-cutionaryeffect(e.g.bringingaboutacertainbeliefintheaddressee)viatheaddressee’srecognitionofthisintention.Thenotionof“mutualknowledge”(Lewis1969;Schiffer1972),atworkinthecharacterizationofcommonground(seeContextandpropositionalattitudes,page196),isalsorelevanttothecharacterizationofovertness.TheIntentional-InferentialModelEventhoughtheconventionalistapproachtocommunicationisstillalive(aswitnessedby,e.g.,Millikan,1998),theGrice–Strawson“intentionalist”approachhasgainedwidecurrencyinpragmatics.Typicalinthisrespectaretheneo-GriceantheoriesofferedbyBachandHarnish(1979)andSperberandWilson(1995).Theyhaveputforwardaninferentialmodelofcommunicationintendedtosupersedethe“codemodel”inspiredbyShannonandWeaver.1999781472578235_txt_print.indd19908/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageAccordingtothecodemodel,communicationproceedsasfollows.Thespeakerwishestocommunicateacertaincontent;sheencodesthecontentinquestionintoasentence,usingthegrammarofthelanguageasa“code”,pairingcontentsandsentences(possiblywithrespecttoacontextofutterance).Theinterpreter,byvirtueofhisknowledgeofthesamegrammar(and,perhaps,ofthecontext),iscapableofdecodingthesentenceandrecoveringtheintendedcontent.Thealternative,inferentialmodelofcommunicationisverydifferent.Anutteranceisseenasameaningfulaction,i.e.anactionthatprovidesinterpreterswithevidenceconcerningtheagent’sintentions.Whatdistin-guishescommunicativeactsfromothermeaningfulactionsiswhatcanbeinferredfromtheevidence:Acommunicativeactisanactthatprovidesevidenceofacertaincommunicativeintentiononthepartofthespeaker.Inotherwords,thespeaker’sintentiontocommunicatesomethingiswhatexplainshisutterance,consideredasapieceofbehaviour.Fromthispointofview,thecontentofthecommunicativeact–whatiscommunicated–isthetotalcontentofthecommunicativeintentionsthatcanbeinferredfromit.Letuscallthistheutterance’scommunicativemeaning,distinctfromtheliteralorconventionalmeaningofthesentence(determinedbythecode,i.e.thegrammar).Understandingisessentiallyaninferentialprocessinthisframework,andtheconventionalmeaningofthesentenceprovidesonlypartoftheevidenceusedindeterminingthecommunicativemeaningoftheutterance.1Notes1.Thischapterisalightlyupdatedandshortenedversionoftheentry“Pragmatics”thatappearedinthe1998RoutledgeEncyclopediaofPhilosophy.ReferencesAlston,W.(1964),PhilosophyofLanguage,EnglewoodCliffs,NJ:PrenticeHall.—(2000),IllocutionaryActsandSentenceMeaning,Ithaca,NY:CornellUniversityPress.Austin,J.(1950/1971),“Truth”,inJ.Austin,PhilosophicalPapers,2ndedn,Oxford:ClarendonPress,pp.117–33.—(1971),PhilosophicalPapers,2ndedn,Oxford:ClarendonPress.—(1975),HowtoDoThingswithWords,2ndedn,Oxford:ClarendonPress.Bach,K.(1992),“PavingtheRoadtoReference”,PhilosophicalStudies67:295–300.Bach,K.andHarnish,M.(1979),LinguisticCommunicationandSpeechActs,Cambridge,MA:MITPress.2009781472578235_txt_print.indd20008/08/201415:46\nPragmaticsBorg,E.(2004),MinimalSemantics,Oxford:ClarendonPress.Bühler,K.(1990),TheoryofLanguage,trans.byD.Goodwin,Amsterdam:J.Benjamins.Cappelen,H.(2007),“SemanticsandPragmatics:SomeCentralIssues”,inG.PreyerandG.Peter(eds),Context-SensitivityandSemanticMinimalism,Oxford:ClarendonPress,pp.3–22.Cappelen,H.andLepore,E.(2005),InsensitiveSemantics,Oxford:Blackwell.Carston,R.(2002),ThoughtsandUtterances:ThePragmaticsofExplicitCommunication,Oxford:Blackwell.Clark,H.(1992),ArenasofLanguageUse,Chicago:TheUniversityofChicagoPressandCSLI.Davis,S.(ed.)(1991),Pragmatics:AReader,NewYork:OxfordUniversityPress.Ducrot,O.(1972),Direetnepasdire:principesdesémantiquelinguistique,Paris:Hermann.Gardiner,A.(1932),TheTheoryofSpeechandLanguage,Oxford:ClarendonPress.Gazdar,G.(1979),Pragmatics:Implicature,PresuppositionandLogicalForm,NewYork:AcademicPress.Grice,P.(1957),“Meaning”,PhilosophicalReview66:377–88.—(1989),StudiesintheWayofWords,Cambridge:CambridgeUniversityPress.Heim,I.(1988),“OntheProjectionProblemforPresuppositions”,reprintedinS.Davised.Pragmatics:AReader,NewYork:OxfordUniversityPress,pp.397–405.Horn,L.(2005),“TheBorderWars:ANeo-GriceanPerspective”,inK.vonHeusingerandK.Turner(eds),WhereSemanticsMeetsPragmatics:theMichiganPapers,Oxford:Elsevier,pp.21–48.Horn,L.andWard,G.(eds)(2004),HandbookofPragmatics,Oxford:Blackwell.Kamp,H.(1985),“Context,ThoughtandCommunication”,ProceedingsoftheAristotelianSociety85:239–61.Kaplan,D.(1989a),“Demonstratives”,inJ.Almog,H.WettsteinandJ.Perry(eds),ThemesfromKaplan,NewYork:OxfordUniversityPress,pp.481–563.—(1989b),“Afterthoughts”,inJ.Almog,H.WettsteinandJ.Perry(eds),ThemesfromKaplan,NewYork:OxfordUniversityPress,pp.565–614.Karttunen,L.(1974),“PresuppositionandLinguisticContext”,TheoreticalLinguistics1:181–94.Levinson,S.(2000),PresumptiveMeanings:TheTheoryofGeneralizedConversationalImplicature.Cambridge,MA:MITPress.Lewis,D.(1969),Convention,Cambridge,MA:HarvardUniversityPress.—(1970),“GeneralSemantics”,Synthese22:18–67.—(1979),“ScorekeepinginaLanguage-Game”,JournalofPhilosophicalLogic8:339–59.Malinowski,B.(1923/1949),“TheProblemofMeaninginPrimitiveLanguages”,supplementtoC.OgdenandI.A.Richards,TheMeaningofMeaning,10thedn,London:Routledge,pp.296–336.Millikan,R.(1998),“ProperFunctionandConventioninSpeechActs”,inL.Hahned.ThePhilosophyofP.F.Strawson,LaSalle:OpenCourt,pp.25–43.Morris,C.(1938),FoundationsoftheTheoryofSigns,Chicago:UniversityofChicagoPress.2019781472578235_txt_print.indd20108/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagePotts,C.(2005),TheLogicofConventionalImplicatures,Oxford:OxfordUniversityPress.Predelli,S.(2005),Contexts:Meaning,Truth,andtheUseofLanguage,Oxford:OxfordUniversityPress.Recanati,F.(1987),MeaningandForce:ThePragmaticsofPerformativeUtterances,Cambridge:CambridgeUniversityPress.—(1993),DirectReference:FromLanguagetoThought,Oxford:BasilBlackwell.—(1994),“ContextualismandAnti-ContextualisminthePhilosophyofLanguage”,inS.Tsohatzidis,FoundationsofSpeechActTheory:PhilosophicalandLinguisticPerspectives,London:Routledge,pp.156–66.—(2004),LiteralMeaning,Cambridge:CambridgeUniversityPress.—(2010),Truth-ConditionalPragmatics,Oxford:OxfordUniversityPress.Schiffer,S.(1972),Meaning,Oxford:ClarendonPress.Schlenker,P.(2008),“BeArticulate!APragmaticTheoryofPresuppositionProjection”,TheoreticalLinguistics34:157–212.Searle,J.(1969),SpeechActs,Cambridge:CambridgeUniversityPress.—(1978/1979),“LiteralMeaning”,inJ.Searle,ExpressionandMeaning,Cambridge:CambridgeUniversityPress,pp.117–36.—(1979),ExpressionandMeaning,Cambridge:CambridgeUniversityPress.—(1983),Intentionality,Cambridge:CambridgeUniversityPress.Sperber,D.andWilson,D.(1995),Relevance:CommunicationandCognition,2ndedn,Oxford:BasilBlackwell.Stalnaker,R.(1970),“Pragmatics”,Synthese22:272–89.—(1974/1991),“PragmaticPresuppositions”,reprintedinS.Davised.Pragmatics:AReader,NewYork:OxfordUniversityPress,pp.471–82.—(1978/1999),“Assertion”,reprintedinR.Stalnaker,ContextandContent,Oxford:OxfordUniversityPress,pp.78–95.—(1999),ContextandContent,Oxford:OxfordUniversityPress.—(2002),“CommonGround”,LinguisticsandPhilosophy25:701–21.Stanley,J.(2007),LanguageinContext,Oxford:ClarendonPress.Stenius,E.(1967),“MoodandLanguage-Game”,Synthese17:254–74.Strawson,P.(1950/1971),“OnReferring”,inP.Strawson,Logico-LinguisticPapers,London:Methuen,pp.1–27.—(1952),IntroductiontoLogicalTheory,London:Methuen.—(1964/1971),“IntentionandConventioninSpeechActs”,inP.Strawson,Logico-LinguisticPapers,London:Methuen,pp.149–69.—(1971),Logico-LinguisticPapers,London:Methuen.Travis,C.(1975),SayingandUnderstanding,Oxford:BasilBlackwell.—(1981),TheTrueandtheFalse:theDomainofthePragmatic,Amsterdam:J.Benjamins.—(1985),“OnWhatisStrictlySpeakingTrue”,CanadianJournalofPhilosophy15:187–229.—(2008),Occasion-Sensitivity:SelectedEssays,Oxford:OxfordUniversityPress.Tsohatzidis,S.(1994),FoundationsofSpeechActTheory:PhilosophicalandLinguisticPerspectives,London:Routledge.Waismann,F.(1951),“Verifiability”,inA.Flewed.LogicandLanguage,1stseries,Oxford:BasilBlackwell,pp.17–44.Wittgenstein,L.(1953),PhilosophicalInvestigations,Oxford:BasilBlackwell.2029781472578235_txt_print.indd20208/08/201415:46\nSemanticNormativityand9Naturalism1JoséL.ZalabardoKripke’sNormativityArgumentInWittgensteinonRulesandPrivateLanguage,SaulKripkepresentedanargument,inspiredbyhisreadingofWittgenstein,againstthepossibilityoffactsastowhatsomeonemeansbyalinguisticexpression.Theargumentproceededbyconsideringaseriesofproposalsastowhatthesefactsmightconsistin,arguingineachcasethatthecandidateaccountofmeaningfactswasunsatisfactory.AproposalthatfiguresprominentlyinKripke’sargumentisthedispositionalaccountofmeaning.Accordingtothedispositionalaccount,theconditionsofcorrectuseassociatedwiththemeaningofanexpressionaredeterminedbyspeakers’dispositions:howitwouldbecorrecttouseanexpression,asmeantbyaspeaker,isdetermined,onthisaccount,byhowthespeakerisdisposedtousetheexpression.Kripkeraisedseveralobjectionstodispositionalism.Onethathasreceivedmuchattentionintheliteraturearisesfromthenormativestatusoffactsaboutcorrectuse.Herearesomerepresentativepassages,focusingonthecaseofthemeaningofthesign“+”:Thedispositionalistgivesadescriptiveaccountofthisrelation[betweenthemeaningIascribeto“+”andhowIamdisposedtoanswerthequestion“68+57=?”]:if“+”meantaddition,thenIwillanswer“125”.Butthisisnottheproperaccountoftherelation,whichisnormative,notdescriptive.Thepointisnotthat,ifImeantadditionby“+”,Iwillanswer“125”,butratherthat,ifIintendtoaccordwithmypastmeaningof“+”,Ishouldanswer“125”.[…]therelationofmeaningandintentiontofutureactionisnormative,notdescriptive(Kripke1982:37).Acandidateforwhatconstitutesthestateofmymeaningonefunction,ratherthananother,byagivenfunctionsign,oughttobesuchthat,whateverinfactI(amdisposedto)do,thereisauniquethingthatIshould2039781472578235_txt_print.indd20308/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagedo.Isnotthedispositionalviewsimplyanequationofperformanceandcorrectness?(Kripke1982:24).Theinterpretationoftheseandrelatedpassagesisbynomeansstraight-forward,butmanyhavereadthemasadvancingargumentsthatapplytosemanticslinesofreasoningoriginallyaimedatnaturalisticaccountsofethicalconcepts(Boghossian1989;Wright1984;Blackburn1984;McDowell1984).InthischapterIamgoingtoconsidertwooftheselinesofreasoning.IamnotgoingtobeconcernedwiththequestionwhethertheycanbeplausiblyattributedtoKripke.2Mygoalwillbetodeterminewhethertheselinesofreasoningcanbeconstruedasprovidingcogentargumentsagainstsemanticnaturalism.TheNaturalisticFallacyThefirstlineofreasoningthatIwanttoconsiderisbasedonthethoughtthatthereisanunbridgeablelogicalgapbetweenstatementsabouthowthingsareandstatementsaboutwhatsomeoneshoulddo.DavidHumeappearstoendorsethisthoughtinthefollowingfamouspassageoftheTreatise:Ineverysystemofmorality,whichIhavehithertometwith,Ihavealwaysremark’d,thattheauthorproceedsforsometimeintheordinarywaysofreasoning,andestablishesthebeingofaGod,ormakesobservationsconcerninghumanaffairs;whenallofasuddenIamsurpriz’dtofind,thatinsteadoftheusualcopulationsofpropositions,is,andisnot,Imeetwithnopropositionthatisnotconnectedwithanought,oranoughtnot.Thischangeisimperceptible;butishowever,ofthelastconsequence.Forasthisought,oroughtnot,expressessomenewrelationoraffirmation,’tisnecessarythatitshou’dbeobserv’dandexplain’d;andatthesametimethatareasonshouldbegiven;forwhatseemsaltogetherinconceivable,howthisnewrelationcanbeadeductionfromothers,whichareentirelydifferentfromit(Hume1978:469).Humeisherepresentingathoughtaboutargumentsfrompremisesabouthowthingsaretoconclusionsaboutwhatsomeoneshoulddo.Thethoughtappearstobethattheseargumentscannotbedeductive,or,asIproposetoconstruethethoughtinitially,thattheycannotbeunderwrittenexclusivelybylogicandconceptualanalysis.Wecanformulatethethoughtasconcerningthecharacteroftheinfer-entialprinciplesthatcanvalidatethesearguments.Iamgoingtorefertotheprinciplesthatconcernusasprescriptiveprinciples.Prescriptiveprincipleswillbestatementsoftheform:2049781472578235_txt_print.indd20408/08/201415:46\nSemanticNormativityandNaturalism1.(∀x)(∀S)(φ(x,S)⊃Sshould/shouldn’tdox)whereSrangesoversubjectsandxoveractions,andφisapropositionalfunctionthatdescribesanactionandpossiblyasubjectandtherelationshipbetweensubjectandaction.Iwanttoconcentrateonaparticulartypeofprescriptiveprinciples–thoseinwhichthedescriptionfromwhichaprescriptionisderiveddoesn’tincludetheascriptiontothesubjectofadesire,goalorpro-attitude.ForthispurposeIamgoingtointroduceadistinctionbetweenhypotheticalandcategoricalprescriptiveprinciples.WeshallsaythataprescriptiveprincipleishypotheticaljustincaseinorderforxandStosatisfyφ,Sneedstohavesomedesire,goalorpro-attitude.Thatis,ifxandSsatisfyφ,thereisadesire,goalorpro-attitudethatShas,suchthatinitsabsencexandSwouldn’tsatisfyφ.Aprescriptiveprinciplewillbecategoricalifitisnothypothetical.Thus,theobligationsexpressedbycategoricalprescriptiveprinciplesareindependentofthesubject’svolitions.However,theyaresubjecttobeoverriddenbythedemandsplacedonthesubject’sbehaviourbyotherprescriptiveprinciples.Let’sfocusonaninstanceof1specifyingconditionsunderwhichyoushoulddox.Evenifitweretrue,andevenifSandxsatisfiedφ,itmightbethat,allthingsconsidered,Sshouldnotdox,ifforStodoxwouldbeincompatiblewithwhatotherprescriptiveprinciplesdemandofS.Iwanttoformulatethethoughtthatwecannotdeduceaconclusionaboutwhatsomeoneshoulddofrompremisesabouthowthingsareasaclaimaboutthestatusofcategoricalprescriptiveprinciples.IshallrefertotheclaimasHume’sthought:Hume’sthought:Categoricalprescriptiveprinciplescannotbeestablishedsolelybymeansoflogicandconceptualanalysis.NoticethatHume’sthoughtisinprincipleneutralonwhetherornottherearetruecategoricalprescriptiveprinciples.Ontheonehand,anadvocateofHume’sthoughtcanmaintainthattherearetruecategoricalprescriptiveprinciples,solongasshedoesn’ttakethemtobederivablebylogicandconceptualanalysis.Ontheotherhand,anadvocateofHume’sthoughtcandenytheexistenceoftruecategoricalprescriptiveprinciples.Therejectionofcategoricalprescriptiveprinciplescantaketwoforms.Onepossibilityistodenytheexistenceoffactsastowhatoneshoulddooverandabovewhatfollowsfromhypotheticalprescriptiveprinciples.Theotheristoasserttheexistenceofcategoricalprescriptivefacts,whiledenyingthattheysuperveneondescriptivefacts.Onthisview,whenSshoulddoxthereisn’tadescriptionofSandxsuchthatwheneveranactionandasubjectsatisfythedescriptionthesubjectshouldperformtheaction.2059781472578235_txt_print.indd20508/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageHume’sthoughtcanbeseenasposinganinsuperableobstacletotheprovisionofnaturalisticanalysesofethicalnotions.Supposethefollowingisaconceptualtruthaboutmoralrightness:2.Ifitwouldbemorallyrightforyoutoperformanaction,theninprinciple(i.e.unlesstheobligationisoverriddenbyotherimperatives)youshouldperformtheaction.Nowconsideranaturalisticanalysisoftheconceptofmoralrightness,accordingtowhichthereisa(categorical)descriptionNofarelationshipbetweenanactionandasubjectsuchthatforanactionandasubjecttosatisfythisdescriptioniswhatitmeansforittobemorallyrightforthesubjecttoperformtheaction.ThisanalysiswouldresultinastraightforwardviolationofHume’sthought.(∀x)(∀S)(N(x,S)⊃Sshoulddox)wouldbeatruecategoricalprescriptiveprinciplederivablefromtheconceptualanalysisofmoralrightness.IfHume’sthoughtiscorrect,andifitisanalyticthatoneshoulddowhat’smorallyright,thennonaturalistic(categorical)analysisofmoralrightnesscanbecorrect.Thisisnotasseriousablowtoethicalnaturalismasitmightsound,sincecontemporaryethicalnaturalistsdon’ttypicallyseethemselvesasprovidingconceptualanalysesofethicalconcepts.Theyseethemselvesmorecommonlyastryingtodisclosethehiddenessenceofethicalconcepts,notunlikethewayinwhichsciencedisclosesthehiddenessenceofwaterorgold.Onthisconceptionofthetask,anaturalisticaccountofanethicalconceptwouldexpressametaphysicallynecessarytruththatcouldonlybeestablishedaposteriori.Ifanaturalisticaccountofmoralrightnessisconceivedalongtheselines,itwillnolongerlicenseviolationsofHume’sthought,sinceitwon’tallowustoestablish(∀x)(∀S)(N(x,S)⊃Sshoulddox)byconceptualanalysis.AsfarasIcansee,theonlywaytocounterthismovewouldbetostrengthenHume’sthoughtinordertoruleoutthepossibilityofderivingcategoricalprescriptiveprinciplesfrommetaphysicallynecessarytruths:Hume’sthought(secondversion):Categoricalprescriptiveprinciplescannotbeestablishedsolelybymeansoflogic,conceptualanalysisormetaphysicalnecessities.IamnotgoingtodiscusstheplausibilityofeitherversionofHume’sthought.InsteadIamgoingtoassumeforthesakeoftheargumentthatthesecondversionoftheprincipleiscorrect,inordertoaddressthequestion,whetherwecanobtainfromthisacogentargumentagainstsemanticnaturalism.2069781472578235_txt_print.indd20608/08/201415:46\nSemanticNormativityandNaturalismTheNormativityofLinguisticMeaningDiscussionoftheconsequencesofHume’sthoughtforsemanticshasgenerallyfocusedonnaturalisticaccountsoftherelationpairingeachpredicatewiththepropertywhoseinstancessatisfythepredicate,asmeantbyaspeakeratatime.Whenapredicate,asmeantbyaspeaker,isrelatedinthiswaytoaproperty,Ishallsaythatthepredicatereferstotheproperty,orthatthespeakerreferstothepropertybythepredicate.Usingthisnotionofpredicatereference,wecannowdefinepredicatesatisfactionasfollows:3.P,asmeantbyS,issatisfiedbyajustincaseainstantiatesthepropertytowhichSrefersbyP.Supposenowthatthefollowingisaconceptualtruthaboutpredicatemeaning:4.IfP,asmeantbyS,issatisfiedbya,thenSshouldapplyPtoa.ItfollowsthatforeverypredicateP,subjectSandobjecta,thefollowingisaconceptualtruth:5.(∀x)(∀S)(xisanascriptionofPtoabyS&P,asmeantbyS,issatisfiedbya⊃Sshoulddox)3Nowconsideranaturalisticanalysisofpredicatereference,accordingtowhichthereisanaturalrelationrbetweenpredicates,propertiesandspeakerssuchthatforP,PandStober-relatediswhatitmeansforStomeanPbyP.AndletNbethenaturalizationofpredicatesatisfactionthatresultsfromthefollowingdefinition:N(S,P,a)iffr(P,P,S)&Pa.4Nowtheanalysisofpredicatereferencewillenableustoderivethefollowingfrom5:6.(∀x)(∀S)(xisanascriptionofPtoabyS&N(S,P,a)⊃Sshoulddox)Clearly,theantecedentof6isadescriptionofxandS,ontheplausibleassumptionthatbeinganascriptionofPtoaisadescriptionofx.Thenitfollowsthat6isacategoricalprescriptiveprinciple.Furthermore,ifpredicatereferencewerecorrectlyanalysableintermsofr,6wouldbeatruecategoricalprescriptiveprinciplederivablebyconceptualanalysis,andwewouldhaveaviolationofHume’sthought.Hume’sthoughtappearstoruleoutnaturalisticanalysesofpredicatereference.Thesamesituationwillensueifwethinkoftheclaimthatpredicatereferenceisrasanecessaryaposterioritruth,solongasweinvokethesecondversionofHume’sthought.Onecouldtrytoresistthislineofreasoning,aspresented,byobjectingto2079781472578235_txt_print.indd20708/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguage4.TheproblemisthatitrequiresthesubjecttoascribePtoeveryobjectthatsatisfiesit.Thisis,ingeneral,notapossibility.Hence,ontheassumptionthatoughtimpliescan,4cannotexpressagenuineprescription(seeHattiangadi2006:226).Thereareseveralwaysinwhichtheargumentmightbemodifiedtoovercomethisobjection.Anappealingpossibilityistoreplace4withaninjunctionagainstascribingPtoobjectsthatdon’tsatisfyit:7.(∀x)(∀S)(xisanascriptionofPtoabyS&P,asmeantbyS,isnotsatisfiedbya⊃Sshouldn’tdox)7doesn’tfallfouloftheought-implies-canprinciple,sincerefrainingfromascribingPtoobjectsthatdon’tsatisfyitisinprinciplewithinS’spowers.And7,togetherwithanaturalisticanalysisofpredicatereferencewillyieldaviolationofHume’sthoughtsimilarto6.However,onecouldobjectto7,aswellas4,ondifferentgrounds.Onecouldarguethattheresimplyisnocategoricalobligationforspeakerstospeakthetruthortorefrainfromspeakingfalsehoods.Forspeakerswhohavethisastheirgoal,7isalegitimatemaxim,buttothosewhohavenodesiretoavoidsayingfalsehoods,7doesn’tapply.Ifthisisright,thentogetauniversallyvalidprincipleweneedtoreplace7withaclaimalongthefollowinglines:8.(∀x)(∀S)(xisanascriptionofPtoabyS&P,asmeantbyS,isnotsatisfiedbya&Swantstoavoidsayingwhat’sfalse⊃Sshouldn’tdox)Ithinkthatthestatusof8asaconceptualtruthisunobjectionable.However,8wouldnotbringnaturalisticaccountsofpredicatereferenceintoconflictwithHume’sthought.Itwouldindeedforceustotreatthefollowingasaconceptual/metaphysicaltruth:9.(∀x)(∀S)(xisanascriptionofPtoabyS&~N(S,P,a)&Swantstoavoidsayingwhat’sfalse⊃Sshouldn’tdox)Theproblemisthat,while9isaprescriptiveprinciple,itisobviouslyofthehypotheticalvariety,andHume’sthoughtisnotviolatedbytreatinghypotheticalprescriptiveprinciplesasconceptualtruths.5Itisimportanttobearinmindthatthisobjectionto7cannotsimplyrestonthepossibilitythatthedemandsthatitmakesonthesubject’sbehaviourmightbeoutweighedbyotherprescriptiveprinciples,asthiswouldbecompatiblewith7,onourunderstandingofprescriptiveprinciples.Theclaimisthatthesedemandsapplyonlytothosethathaveacertaindesire,andthisisthesuresignthatthecategoricalprescriptiveprincipleisillegitimate.62089781472578235_txt_print.indd20808/08/201415:46\nSemanticNormativityandNaturalismAnotherapproachtodevelopingtheargumentagainstnaturalisticaccountsofpredicatereferencewouldbetofocusonprescriptiveprinciplesthatexpress,notanobligationtospeakthetruth,ortorefrainfromspeakingfalsehoods,butanobligationtospeakinwaysthatexpressthemeaningsthatwewanttoconvey.7Thisapproachcomestoprominencewhenwereflectonthecontrastbetweenfactualmistakes,e.g.,applyingapredicatetoanobjectthatdoesn’tinstantiatethepropertythatthepredicaterefersto,andsemanticmistakes,e.g.,applyingtoanobjectapredicatethatdoesn’trefertothepropertythatthespeakerwantstoattributetotheobject.Theclaimthatmistakesofthesecondkindaretobeavoidedcanbeexpressedalongthefollowinglines:10.(∀x)(∀S)(xisanascriptionofPtoabyS&SreferstoPbyP&SwantstoavoidattributingPtoainx⊃Sshouldn’tdox)Theclaimthat10isaconceptualtruthhasmoreplausibilitythanthecorre-spondingclaimfor7.However,10doesn’tbringnaturalistaccountsofpredicatereferenceintoconflictwithHume’sthought.Ifpredicatereferencecanbeanalysedinnaturalisticterms,andifaparallelnaturalistictreatmentisavailableofthenotionofwhichpropertyasubjectwantstoattributewithagivenpredicateascription,then10isundeniablyaprescriptiveprinciple.Butjustasclearly,giventhatitsantecedentmentionsS’sdesiretoavoidattributingacertainpropertytotheobjectofpredication,10isaprescriptiveprincipleofthehypotheticalkind,and,asweknow,Hume’sthoughtisnotinconflictwiththeconceptualstatusofhypotheticalprescriptiveprinciples.TheNormativityofMentalContentThenormativecharacterofsemanticnotionscanalsobeinvokedtounderminenaturalisticaccounts,notoflinguisticmeaning,butofmentalcontent,andshiftingourattentiontothemindmightremovesomeoftheobstaclesencoun-teredinthelinguisticversionoftheenterprise.Thisisnotafundamentalchangeofsubjectmatter.Itwouldbesurprisingifsemanticnotionsweretoreceiveindependentexplicationsinthelinguisticandthementalrealm.Rather,weshouldexpectthatonesetofnotionsisexplicatedintermsoftheother.Theonlyremainingquestioniswhetheroursemantictheorytreatslinguisticmeaningormentalcontentasfundamental,andwhilebothoptionshaveadvocates,startingwiththementalisthefavouredrouteamongsemanticnaturalists.8Accordingtotherepresentationaltheoryofmind,theprimarybearersofmentalcontentaresyntacticallyspecifiablementalitemsanalogoustolinguisticexpressions.Onthisaccount,wewillhaveacategoryofmentalrepresentations2099781472578235_txt_print.indd20908/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagethatplaythesameroleinthecompositionofthoughtsthatthepredicatesofalanguageplayinthecompositionofsentences.Thementaleventofattributingapropertytoanobjectwillconsist,onthisaccount,intokeninginacertainwayamentalpredicatethatreferstothisproperty.9Ifweassumethispictureofmentalrepresentation,wecaneasilydevelopamentalversionoftheargumentwepresentedaboveagainstnaturalisticaccountsofpredicatereference.However,thisassumptionisnotindispen-sable.10Wecanalsoformulateaversionoftheargumentformentalcontentthatdoesn’tpresupposetherepresentationaltheoryofmind.Theargumentwouldbebasedonthethoughtthatthefollowingclaimisaconceptualtruth:11.(∀x)(∀S)(xisamentalattributionbySofPtoa&~Pa⊃Sshouldn’tdox)Ifthisthoughtiscorrect,itwillposeanobstacletoanynaturalisticaccountofthenotionofmentalattributionofproperties,whetherornottheaccountconstruesthenotionasinvolvingsyntacticallyidentifiabletypes,alongthelinesoftherepresentationaltheoryofmind.Anysuchaccountwouldturn11intoaconceptuallytruecategoricalprescriptiveprinciple,inviolationofHume’sthought.Noticethat11istherestrictiontoaclassofbeliefswithaparticularlysimplestructureoftheprescriptionnottobelievefalsehoods–tobelieveonlytruths–towhichI’mgoingtoreferasthetruthprescription.Henceourquestionis,ineffect,whetherwecanderivethetruthprescriptionfromlogic,conceptualanalysisormetaphysics.Someauthorshavesuggestedthatevenif11wereaconceptualtruth,itwouldn’tfollowdirectlythatmentalcontentisnormativeinasensethatwouldposeanobstacletoanaturalistaccountofthenotion.Thethoughtisthattheconceptualtruthof11mightinprinciplemanifestthenormativecharacter,notofmentalcontent,butofbelief.11Wecanseetheissuemoreclearlyifwefocusontheversionof11thatwecanformulateifweassumetherepresentationaltheoryofcontent:12.(∀x)(∀S)(xisanascriptionbySofmentalpredicateMtoa&MreferstoP&~Pa⊃Sshouldn’tdox)Thethoughtisthattheconceptualcharacterof12couldbemadecompatiblewithHume’sthoughteitherbyrejectingthepossibilityofanaturalisticaccountofthereferenceofmentalpredicates(thesecondconjunct)orbyrejectingsuchanaccountofthenotionofamentalstatebeinganascriptionofamentalpredicate(thefirstconjunct).Intheterminologyusedbyadvocatesofthe2109781472578235_txt_print.indd21008/08/201415:46\nSemanticNormativityandNaturalismrepresentationaltheoryofmind,thisisthecontrastbetweenaccountsofmentalcontentandaccountsofpropositionalattitudes.Theidea,then,isthatthesemanticnaturalistcanacceptHume’sthoughtandtheconceptualcharacterof12,solongassheispreparedtoforgoanatural-isticaccountofthenotionofamentalstatebeinganascriptionofamentalpredicate–ofamentalpredicatebeing“tokenedinthebeliefmode”.Thispointis,strictlyspeaking,correct,butIsuspectthenaturalistwon’tfindthisrouteparticularlyappealing.Theproblemconcernsthestatusthattheresultingpositionwouldaccordtofactsaboutwhichmentalstatestokenagivenpredicateinthebeliefmode.Thesefactswouldhavetobeconstruedasnon-descriptive,butthisconstrualwouldhavetobemadecompatiblewithanaturalisticaccountoffactsaboutwhichpropertyeachmentalpredicaterefersto.Iamnotgoingtoexploretheissuehere,butitishardtoseehowtheseviewscouldbecombinedintoanappealingpackage.Ishallassume,accordingly,thatifwecouldestablishtheconceptualstatusof11,Hume’sthoughtwouldberenderedincompatiblewithanaturalisticaccountofmentalcontent.Clearly,fortheanti-naturalistit’snotenoughtoestablishtheconceptualtruthofahypotheticalimperativeconcerningmentalattributionsofproperties,makingtheobligationtoavoidattributingPtoobjectsthatdon’tinstantiateitcontingentonShavingthedesiretoavoidbelievingfalsehoods.Anotherunpromisingavenuewouldbetogroundtheobligationtoavoidfalsementalpropertyattributionsontheuntowardpracticalconsequencesofbelievingfalsehoods.Forwecannotexpecttoestablishthepracticalundesirabilityofbelievingfalsehoodsexclusivelyonlogical,conceptualormetaphysicalconsiderations,unlessapragmatistaccountoftruthisadopted.Onanynon-pragmatistaccountoftruth,theconnectionbetweentruebeliefandthepromotionofourgoalswillbeacontingentmatter.Theanti-naturalistneedstoestablishonconceptual/metaphysicalgroundsacategoricalobligationtorefrainfrombelievingwhat’sfalse,independentofthesubject’svolitionsorofthepracticalconsequencesofsuchbeliefs.ShahandVellemanonWhyweShouldBelievetheTruthThemostprominentrecentattemptstoestablishthetruthprescriptiononconceptualgroundsareduetoNishiShahandDavidVelleman.Theyconsistintwoseparatearguments,eachpresentedinthefirstinstancebyoneoftheseauthorsindividually,butlaterdevelopedjointlyinco-authoredwork.MygoalinthissectionistoprovideanassessmentofShahandVelleman’sarguments,summarizingamoredetaileddiscussionthatIhavepresentedelsewhere(Zalabardo2010).IwanttostartbyconsideringShah’sargumentforthetruthprescription.It2119781472578235_txt_print.indd21108/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageisbasedonthecontentionthatacceptingthetruthprescriptionisaconditionforpossessingtheconceptofbelief.Hewrites:“itisoneoftheconditionsforpossessingtheconceptofbeliefthatoneaccepttheprescriptiontobelievethatponlyifpistrue”(Shah2003:470).Itfollowsfromthisthat“acompetentuseroftheconceptofbeliefmustaccepttheprescriptiontobelievethatponlyifpistrueforanyactivitythatheconceivesofasbelief-formation”(Shah2003:470).ItisimportanttounderstandthecontrastbetweenShah’scontentionandtheclaimthatacceptingthetruthprescriptionisaconditionforhavingbeliefs.Shahisverycarefultopointoutthathisargumentdoesn’testablishthelatterclaim.Hisargumentconcernstheconditionsforhavingtheconceptofbelief,nottheconditionsforformingbeliefs,and,asShahconcedes,havingtheconceptofbeliefisnotrequiredforformingbeliefs.ThispointrenderstheconnectionbetweenShah’scontentionandthetruthprescriptionslightlyproblematic.Ifacceptingtheprescriptionwereaconditionforhavingbeliefs,thelinkwouldbestraightforward:allbelieverswouldbesubjecttotheprescriptionbecausethosewhodidn’tacceptitwouldn’tcountasbelievers.ButwithShah’sclaimthingsareslightlymorecomplicated,since,accordingtohim,someonewhodoesn’tacceptthetruthprescriptionmightstillcountasabeliever.Itwouldseem,then,thatShah’sclaimwouldestablishthetruthprescriptiononlyforsubjectswhohavetheconceptofbelief.Butthisrestrictionwon’tprotectthenaturalistfromtheconsequencesofHume’sthought.AllitmeansisthatShah’sclaimwillgroundaprescriptionofthefollowingform:13.(∀x)(∀S)(xisamentalattributionbySofPtoa&Shastheconceptofbelief&~Pa⊃Sshouldn’tdox)Sincethenaturalistcanalsobeexpectedtoendorseadescriptiveaccountofwhocountsashavingtheconceptofbelief,hernaturalisticaccountofmentalpropertyattributionwillturn13intoacategoricalprescriptiveprinciple,thusbringingitintoconflictwithHume’sthought.Thus,forShah,youarenotconceivingofanattitudeasbeliefunlessyouaccepttheprescriptiontoavoidfalsehoodinit.Shahdefendsthisclaimwithaninference-to-the-best-explanationargument.Hearguesthatthereisaphenomenonthatwouldbeadequatelyexplainedbytheviewthattheconceptofbeliefrequiresacceptingthetruthprescription,butwouldotherwisegounexplained,orreceiveonlyinferiorexplanations.Theexplanandumthatplaysthisroleinhisargumentisaphenomenontowhichherefersasthetransparencyofdoxasticdeliberation.Doxasticdeliberationis“deliberationaboutwhattobelieve”(Shah2003:447),andthetransparencyofdoxasticdeliberationisthefollowingfeatureofthisactivity:2129781472578235_txt_print.indd21208/08/201415:46\nSemanticNormativityandNaturalismThedeliberativequestionwhethertobelievethatpinevitablygiveswaytothefactualquestionwhetherp,becausetheanswertothelatterquestionwilldeterminetheanswertotheformer.Thatis,theonlywaytoanswerthequestionwhethertobelievethatpistoanswerthequestionwhetherp(ShahandVelleman2005:499).AccordingtoShah,treatingacceptanceofthetruthprescriptionasnecessaryforhavingtheconceptofbeliefenablesustoprovideasatisfactoryexplanationoftransparency,butifthisviewisrejectedtransparencywillgounexplained.HereishowShahproposestoexplaintransparency:[…]acompetentuseroftheconceptofbeliefmustaccepttheprescriptiontobelievethatponlyifpistrueforanyactivitythatheconceivesofasbelief-formation.Becauseoneacceptsthisprescriptioninsofarasoneisdeliberatingaboutwhethertobelievethatp,determiningwhetherpistruewillbeimmediatelyimperative,totheexclusionofanyotherquestion,foranyonewhoentertainsthedeliberativequestionwhethertobelievethatp(Shah2003:470).Shah’sthoughtisthat,sincedoxasticdeliberationinvolvesconceivingofthecognitiveactivitythatitgeneratesasbeliefformation,itrequiresacceptingthetruthprescriptionforthisactivity,andthismakesthequestionwhethertobelievethatpgivewaytothequestionwhetherp.Shah’sopponentissomeonewhothinksthatacceptingthetruthprescriptionisnotnecessaryforhavingtheconceptofbelief–thatyoucouldinprinciplecountashavingtheconceptofbeliefevenifyouadheredtodoxasticpoliciesthatareincompatiblewiththetruthprescription,e.g.havingonlybeliefsthatmaximizeyouroverallutility.Iamgoingtorefertothispositionasrelativism.12InorderforShah’sargumenttosucceed,heneedstoachievetwogoals.First,heneedstoconvincetherelativistthattransparencyisagenuinephenomenonforwhichanexplanationisneeded.Second,heneedstoconvinceusthattreatingacceptanceofthetruthprescriptionasnecessaryfortheconceptofbeliefaffordsabetterexplanationoftransparencythananyavailabletotherelativist.Shah’sfirsttaskthenistoconvincetherelativistthattransparencyisarealphenomenoncallingforanexplanation.Noticethattransparencypostulatestheexistenceoftwodifferententerprises:ontheonehand,thecognitiveenterpriseoftryingtoanswerthequestionwhetherpand,ontheother,thedeliberativeenterpriseoftryingtoanswerthequestionwhethertobelievethatp.Transparencyisthephenomenonthatobtainsifthelatterenterpriseimmedi-atelygiveswaytotheformer.Butontheconceptionofthedeliberativeenterprisethattherelativistcan2139781472578235_txt_print.indd21308/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagebeexpectedtoendorseattheoutset,transparencydoesn’tseemtohold.Fortherelativist,wetrytoanswerthequestionwhethertobelievethatpbytryingtodeterminewhetherbelievingthatpwouldsatisfyourdoxasticcriteria.Forsomeonewhosubscribestothetruthprescription,thiswillconsistintryingtoanswerthequestionwhetherp,butsomeonewhosubscribestoanalternativedoxasticprescription,e.g.tobelieveonlywhatmaximizesherutility,wouldtrytoanswerthedeliberativequestionbytryingtoansweradifferentfactualquestion–whetherbelievingpwouldmaximizeherutility(callthisenter-priseutilitarianreflection).Onthisconceptionofdoxasticdeliberation,itisnotgenerallyatransparententerprise.Thus,establishingthattransparencyisagenuinephenomenonrequiresinvokingadifferentaccountofdoxasticdeliberation.ShahandVellemanmakeaproposaltothiseffectintheirjointpaper.Todeliberateaboutwhethertobelievethatp,ShahandVellemantellus,istoengage“inreasoningthatisaimedatissuingornotissuinginone’sbelievingthatpinaccordancewiththenormforbelievingthatp”(ShahandVelleman2005:502).13Aswehaveseen,Shahneedstoshowthatutilitarianreflectiondoesn’tcountasdoxasticdelib-erationonp.Inlightofthisconstrualofthedeliberativeenterprise,Shahwouldachievethisgoalifhecouldshowthatutilitarianreflectioncannotbeaimedatissuing(ornotissuing)inyourbelievingthatpinaccordancewiththenormforbelievingthatp.Inadifferentpaper,Shahseemstoputforwardanargumentthatwouldyieldthisconclusion.Hewrites:InthesenseIhaveinmind,deliberatingwhethertobelievethatpentailsintendingtoarriveatbeliefastowhetherp.Ifmyansweringaquestionisgoingtocountasdeliberatingwhethertobelievethatp,thenImustintendtoarriveatbeliefastowhetherpjustbyansweringthatquestion.Icanarriveatthebeliefjustbyansweringthequestionwhetherp;however,Icannotarriveatthebeliefjustbyansweringthequestionwhetheritisinmyinteresttoholdit(Shah2006:482).HereShahseemstoarguethatthereasonwhyIcannotaimatarrivingatabeliefastowhetherpthroughutilitarianreflectionisthatIcannotachievethisgoaljustbymeansofutilitarianreflection.Thisargumentreliesonaquestionableprincipletotheeffectthatanactivitycannotaimatanoutcomeifitcan’tbringitabout.ButevenifweconcedethisprincipletoShah,hisreasoningrunsintotrouble.Let’sassumeforthesakeoftheargumentthattheonlyactivitythatcanenablemetoarriveatabeliefastowhetherpintherequisitewayisansweringthequestionwhetherp,andthatitfollowsfromthisthatansweringthequestionwhetherpistheonlyactivitythatcanbeaimedatissuingornotissuinginbeliefastowhetherp.OnShah2149781472578235_txt_print.indd21408/08/201415:46\nSemanticNormativityandNaturalismandVelleman’sconstrualofdoxasticdeliberation,thisoutcomewouldentailthattheenterpriseistransparent.TheproblemforShahisthattheargumentthathehasusedtoestablishtransparencycanalsobeusedasaseeminglyadequateexplanationofthephenomenonthatdoesn’trestontheassumptionthatacceptingthetruthprescriptionisnecessaryforhavingtheconceptofbelief.Thealternativeexplanationwouldgoasfollows:(a)tryingtoanswerthequestionwhetherpistheonlyactivitythatcanenablemetoarrive,intherequisiteway,atabeliefastowhetherp;(b)itfollowsfromthisthatitistheonlyactivitythatcanbeaimedatissuingornotissuinginabeliefastowhetherp,and(c)thisentails,inturn,thatdoxasticdeliberationcanonlytaketheformoftryingtoanswerthequestionwhetherp,astransparencyrequires.Intryingtoconvincehisopponentthattransparencyisarealphenomenon,Shahhasalsosuppliedherwithaseeminglyadequateexplanationofit.HenceShahhasfailedtoconvincetherelativistthattransparencyisagenuinephenomenonthatcanonlybeexplainedbytheassumptionthatacceptingthetruthprescriptionisnecessaryforhavingtheconceptofbelief.Thereforehehasfailedtoprovideacogentargumentforthelegitimacyof11(or13).Letmeturnnow,morebriefly,totheargumentoriginallydevelopedbyDavidVelleman.Hehasarguedthat“[…]anattitudedoesn’tqualifyasabeliefunlessit[…]hasatendencytobeconstrainedbyinputinwaysdesignedtoensurethatitistrue”(Velleman2000:255).Itwouldfollowfromthisthatabidingbythetruthprescriptionisnecessaryforhavingbeliefs.Vellemandefendsthisviewbycontendingthat,unlessweintroducethisconditioninourdefinitionofbelief,wewon’tbeabletodistinguishbeliefsfromothercognitiveattitudes,suchasimaginingsandassumings(Velleman2000:247).Thealternativethatheconsidersisthemotivationalaccountofbelief,whichseekstoaccountforbeliefintermsofthedispositionstobehaviourthatbeliefsmotivate.Vellemanisundoubtedlyrighttoclaimthatothercognitiveattitudesbesidesbeliefcanmotivatebehaviour.Butthisisnotenoughtoestablishhispoint.Heneedstoarguethat,say,assumingorimaginingthatphasthesamemotiva-tionalroleasbelievingthatp,andIcan’tseethatheprovidesadequatesupportforthispoint.TotakeoneoftheexamplesVellemanconsiders,imaginingthatyouareanelephantaspartofagameofmake-believemaymotivateyoutobehaveincertainways,butIdon’tthinkonecanplausiblyclaimthatitmotivatesyoutobehaveinexactlythesamewaysasbelievingthatyouareanelephantwould.14Andsolongasthemotivationalroleofbeliefisdifferentfromthemotivationalrolesofotherpropositionalattitudes,itwillbepossibleinprincipletousethesedifferencestosingleoutbelief,obviatingtheneedtointroduceregulationfortruthinouraccount.IconcludethatVelleman’s2159781472578235_txt_print.indd21508/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageargumentdoesn’tprovideadequatesupportfortheclaimthathavingbeliefsrequiresabidingbythetruthprescription.Stability15AsIhavearguedintheprecedingsection,IthinkthatShahandVellemanhavefailedtosupporttheclaimthatthereisauniversalprescriptiontobelieveonlytruths.Iamnotawareofanyotherextantproposalsthatsucceedinthisenterprise.However,thereisanareathatIthinkholdssomepromisefortheview.InthissectionIwanttosketchalineofreasoningthatmightgosomewaytowardsestablishing11onpurelyconceptualgrounds.Iamgoingtoarguethatthedoxasticpolicybasedontheprescriptionexpressedby11doesn’thavemanycoherentalternatives.LetmerefertothedoxasticcriterionaccordingtowhichSshouldbelievethatpjustincasepsatisfiesconditionC(writtenCp)astheC-criterion.16Andlet’ssaythatShasadoptedtheC-criterionasherdoxasticpolicyjustincase,foreverypropositionponwhichshewantstoformanopinion,SaimstobringitaboutthatshebelievesthatpifandonlyifCp.Iamgoingtoarguethattheconditionsthatacriterionhastosatisfyinordertogenerateanadequatedoxasticpolicymightconferonthetruthcriterionaspecialstatus.Considerfirsttheeffectthatyourdoxasticpoliciescanbeexpectedtohaveonyourpracticalpolicies.SupposethatyourpolicyastowhentodoφistodoφonoccasionxjustincasexsatisfiesconditionC.Whatyouendupdoinginpursuitofthispolicyonaparticularoccasion(whetheryouφoryoudon’tφ)willdependonthebeliefthatyouformastowhetherCissatisfied.Hence,ifyournominalpolicyastowhethertoφistodosojustincaseCissatisfied,youreffectivepolicywillbetodosojustincaseyoubelievethatCissatisfied.Ifyousubscribetothetruthcriterion,thispointdoesn’thaveanyremarkableconsequences:youaimtoφonoccasionxjustincaseyoubelieveCx,butyouaimtobelieveCxjustincaseCx.HenceyouroverallpolicyisstilltoφonoccasionxjustincaseCx.Thingsaredifferent,however,ifyousubscribetoadoxasticcriterionotherthanthetruthcriterion.ThenthepolicyofφingonoccasionxjustincaseyoubelieveCxdoesn’timmediatelycollapseintothepolicyofφingonoccasionxjustincaseCx.Supposethat,forpropositionsconcerningwhetherCissatisfied,yourdoxasticpolicyistheD-criterion(tobelievepjustincaseDp)ThenyoureffectivecriterionwithrespecttoφwillbetoφonoccasionxjustincasethepropositionCxsatisfiesconditionD(writtenDCx).Inthisway,whichdoxasticpolicyyouadoptwillaffectyoureffectivepracticalpolicies.Suppose,e.g.,thatyournominaldietarypolicyistoeatsomethingjustincaseitisgoodforyou,butyourdoxasticpolicywithrespecttowhetherafoodisgoodforyouistobelievethatitisjustincasethismakes2169781472578235_txt_print.indd21608/08/201415:46\nSemanticNormativityandNaturalismyoufeelgoodaboutyourself.Thenyoureffectivedietarypolicywillbetoeatsomethingjustincasebelievingthatitisgoodforyoumakesyoufeelgoodaboutyourself.Doxasticpoliciescanhavethiskindofeffect,notonlyonpracticalpolicies,butalsoonotherdoxasticpolicies.SupposethatyournominaldoxasticpolicywithrespecttoapropositionpistobelievepjustincaseCp,butyourdoxasticpolicywithrespecttothepropositionCpistobelieveitjustincaseitsatisfiesconditionD.ThenyoureffectivedoxasticpolicywithrespecttopwillbetobelieveitjustincaseDCp–orrathertobelievepjustincaseDCpsatisfiesyourdoxasticcriterionforthisproposition…Theseconsiderationshaveimportantconsequenceswhenwefocusonglobaldoxasticpolicies–policiesdeterminingthecircumstancesunderwhichasubjectwouldbelieveanyproposition.SupposethatyouhaveadoptedadoxasticpolicybasedontheC-criterion.ThenyouwillaimtobelievepjustincaseCp,butyouwillaimtobelieveCpjustincaseCCp,andsoonadinfinitum.Inthiscase,whatwillbeyoureffectivedoxasticpolicywithrespecttop?WhetherthisquestioncanreceiveasatisfactoryanswerdependsonthebehaviourofC.Let’ssaythattheC-criterioniscompletelystablejustincase,necessarily,foreveryp,Cpifandonlyifp.IftheC-criterioniscompletelystable,wecanplausiblyarguethattheinfiniteregressgeneratedbyaglobaldoxasticpolicydoesn’tposeaproblem.WhentheC-criterionisappliedtop,itdictatesthatyoushouldbelievethatpjustincaseCp,i.e.,bythestabilityoftheC-criterion,justincasep.WhenthescopeofthecriterionisextendedtoCp,theverdictthatityieldsforpdoesn’tchange.ItdictatesthatyoushouldbelievethatpjustincaseyoushouldbelievethatCp,andyoushouldbelievethatCpjustincaseCCp.Hence,bytwoapplicationsofthestabilityoftheC-criterion,wehavethatyoushouldbelievethatpjustincasep.Bythesamereasoning,ifweextendthescopeoftheC-criterionalongthishierarchyanynumberoftimes,thesameverdictwillensue–youshouldbelievethatpjustincasep.Thissuggeststhat,whenthecriterionisappliedalongthewholeinfinitehierarchy,wewillobtainthesameoutcome:youshouldbelievethatpifandonlyifp.Hencetheinfiniteregresswon’tpreventthepolicyfromissuingaverdictforeveryproposition.Inothercases,though,theregresswillbeproblematic.LetmeuseC0(p)todenotep,andforeveryn,letCn+1(p)denoteCCn(p).LetmesaythatCisultimatelyunstablejustincase,foreverynthereisanmgreaterthannsuchthatitisnotthecasethatnecessarilyCm+1(p)justincaseCm(p).IfCisultimatelyunstable,thenadoptingaglobaldoxasticpolicybasedontheC-criterionwillhavedisastrousconsequences.Nomatterhowfaralongthehierarchyweextendtheapplicationofthecriterion,theverdictthatwehavereachedatthatpointconcerningwhetheroneshouldbelievepmightbereversedatsomelaterpointaswecontinuetoextendthescopeofthecriterion.Atnopointwillwereachadefinitiveverdictthatcan’tbereversedfurtheralongthehierarchy.2179781472578235_txt_print.indd21708/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageInthesecircumstances,weseemforcedtoconcludethatthecriteriondoesn’tyieldaverdictonwhetheroneshouldbelievethatpwhenitisappliedtothewholeinfinitehierarchy.Henceanultimatelyunstablecriterionwon’tsustainacoherentglobaldoxasticpolicy.Thisisunquestionablyanauspiciousoutcomeforsupportersofthetruthprescription.Thetruthcriterioniscompletelystable.Hencetheinfiniteregressgeneratedbyglobaldoxasticpolicieswon’tposeaproblemforthepolicybasedonthetruthcriterion.Furthermore,onecouldplausiblyarguethatthetruthcriterionistheonlycompletelystabledoxasticcriterion.Let’sassumethisisright.Then,ifcompletelystablecriteriaweretheonlydoxasticcriteriaforwhichtheregressdoesn’tposeaproblem,wewouldhavefoundawayofvindicatingtheideathatthetruthprescriptionappliesuniversally:onlydoxasticpoliciescompatiblewithitwouldseemtoovercometheproblem.However,thisisnotquiteright.Completelystablecriteriaarenottheonlydoxasticcriteriatosurvivetheregress,astheyarenottheonlydoxasticcriteriathatarenotultimatelyunstable.TheC-criterionisultimatelystable(i.e.notultimatelyunstable)justincasethereisannsuchthat,foreverymgreaterthanorequalton,necessarilyCm+1(p)justincaseCm(p).Now,completelystablecriteriaareultimatelystableforn=0.Butcriteriathatareultimatelystableonlyforhighervaluesfornarenotcompletelystable.Nevertheless,thereasoningthatwehavepresentedtoshowthatcompletelystablecriteriadonotfaceaproblemwiththeregressgeneratedbyglobaldoxasticpoliciescouldbeusedtoestablishthisresultforeveryultimatelystablecriterion.IftheC-criterionisultimatelystableforn,thenoncethescopeofthecriterionhasbeenextendedtothenthtermofthehierarchy,theverdictthatityieldsconcerningwhetheroneshouldbelievethatpwillnotbechangedbyfurtherextensions.ThepolicywillcontinuetodictatethatoneshouldbelievethatpjustincaseCn(p).Inlightofthis,wecanconcludethatthisistheverdictthatthepolicywillsanctionwhenthecriterionisappliedtothewholeinfinitehierarchy.Everyultimatelystabledoxasticcriterioncanavoidthedifficultyposedbytheinfiniteregress.Arethereanyultimatelystablecriteriathatarenotcompletelystable?Iwanttosuggestthattheremightbeatleastone.Itiscertainlynotthecaseforeverypandeverypossiblestateofinformationthattheevidencesupportspjustincasep.Misleadingevidenceisadistressinglycommonoccurrence.Itfollowsthattheevidencecriterionisnotcompletelystable.Neverthelessonecouldplausiblyarguethat,necessarily,foreverypandeverypossiblestateofinformation,theevidencesupportspjustincasetheevidencesupportsthepropositionthattheevidencesupportsp.Theright-to-leftdirectionofthisbiconditionalisfairlyuncontroversial,asevidenceforthehypothesisthatthereisevidenceforpsurelycountsasevidenceforp.Andtheleft-to-rightdirectionalsohassomeplausibility.Foronecouldarguethatitisinthenatureofevidencetobeaccessible.Andthiswouldseemtoentailthattheevidence2189781472578235_txt_print.indd21808/08/201415:46\nSemanticNormativityandNaturalismwon’tsupportpunlessthereisevidencethatitdoes.Ifthisreasoningiscorrect,theevidencecriterionwillbeultimatelystable(forn=1).Inanycase,eveniftheevidencecriterionisultimatelystable,andevenifotherultimatelystablecriteriaexist,thelineofreasoningthatIhavesketchedhastocountasapartialvictoryfortheuniversalityofthetruthprescription.Mostdoxasticpoliciesincompatiblewithitareultimatelyunstable,andhenceunsuitableasglobaldoxasticpolicies.Take,forexample,doxasticcriteriabasedonthepracticalconsequencesofbelief.ItseemsreasonabletosupposethatitisnotthecasenecessarilyforanypropositionpandsubjectSthatbelievingthatpmaximizesS’sutilityjustincaseS’sutilityismaximizedbythebeliefthatbelievingthatpmaximizesherutility.However,thediscreditthattheargumentbringsonultimatelyunstablecriteriahaslimits.Thereisnothingwrong,asfarastheargumentgoes,withassessingbeliefswithanultimatelyunstablecriterion,solongaswedon’ttrytobaseourowndoxasticpolicyonit.Eventheiruseasthebasisforadoxasticpolicyisnotruledoutbytheargument,solongasitsscopeissuitablyrestricted.Icanuseanultimatelyunstablecriterionforp,forthepropositionthatthiscriterionissatisfiedbybelievingthatp,andsoon.Allthat’sneededisthatsoonerorlaterwereachapointinthissequencebeyondwhichmydoxasticpolicyisbasedonanultimatelystablecriterion.TheOpen-QuestionArgumentLetmeturnnowtothesecondlineofreasoningagainstsemanticnaturalismthatIwanttoconsiderinthispaper.Kripke’sappealtothenormativecharacterofmeaninghassometimesbeenseenasputtingforwardaversionoftheopen-questionargument.TheargumentwasoriginallypresentedbyG.E.Mooreagainstthepossibilityofdefiningoranalysinggood.Moorewrites:Thehypothesisthatdisagreementaboutthemeaningofgoodisdisagreementwithregardtothecorrectanalysisofagivenwhole,maybemostplainlyseentobeincorrectbyconsiderationofthefactthat,whateverdefinitionbeoffered,itmaybealwaysasked,withsignificance,ofthecomplexsodefined,whetheritisitselfgood(Moore1903:15).Mooreishappytoacceptthattheremaybenecessaryandsufficientconditionsforsomethingbeinggood.Whathedoesn’tacceptisthattheseconditionscouldbetakenasdefiningoranalysinggood.Thereasonthatthequotedpassagegivesforthisisthatwecanmeaningfullyaskwhethertheseconditionsaresufficientor,presumably,necessaryforbeinggood.Theargumentfeaturesprominentlythenotionofwhetheritispossible2199781472578235_txt_print.indd21908/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagetoaskmeaningfullywhether,say,maximizingutilityisgood.Iproposetoconstruethisnotionintermsofwhetherornothavingacertainbeliefinvolvingaconceptisanecessaryconditionforthistobetheconceptofgood.Inlinguisticterms,thequestioniswhetherhavingabeliefexpressiblewithasentenceinvolvingapredicate,say“good”,isanecessaryconditionforthispredicatetomeangood.Onthisconstrual,thequestion“IsCgood?”couldnotbemeaning-fullyaskedifsomeonewhodoesn’tbelievethatCisasufficientconditionforthesatisfactionof“good”cannotmeangoodby“good”.Havingthisbeliefshouldn’tbetakentorequireentertainingthepropositioninconsciousness.Thebeliefcouldbemanifestedinsteadinyourlinguisticdispositionsbyyourunwillingnesstocontemplatethepossibilityofwithholding“good”fromsomethingthatyoubelievetosatisfyC.Wecannowprovideaformulationfortheopen-questionargument,concen-tratingagainontheconceptmorallyright:1714.ForeverydescriptionD,itispossibleforsomeonewhomeansmorallyrightby“morallyright”tobelieve,forsomeactionx,bothD(x)and“~Morallyright(x)”.1815.IfsatisfyingDisanalyticallysufficientforbeingmorallyright,then,foreveryx,believingD(x)and“~Morallyright(x)”isincompatiblewithmeaningmorallyrightby“morallyright”.Therefore:16.ForeverydescriptionD,satisfyingDisnotanalyticallysufficientforbeingmorallyright.Andaparallelargumentcanbeformulatedagainstthepossibilityofanalysingpredicatesatisfaction:17.ForanydescriptionD,itispossibleforsomeonewhomeanssatisfiesby“satisfies”tobelieve,forsomeS,P,a,bothD(S,P,a)and“~satisfies(S,P,a)”.18.IfsatisfactionofDbyS,PandaisanalyticallysufficientforasatisfyingP,asmeantbyS,thenforallS,P,a,believingD(S,P,a)and“~satisfies(S,P,a)”isincompatiblewithmeaningsatisfiesby“satisfies”.Therefore:19.ForeverydescriptionD,D(S,P,a)isnotanalyticallysufficientforasatis-fyingP,asmeantbyS.2209781472578235_txt_print.indd22008/08/201415:46\nSemanticNormativityandNaturalismClearly,thecogencyofeachoftheseargumentsdependsonwhetherwearecapableofprovidingadequatesupportforitsfirstpremise.Weshalladdressthispointlateron.Butamoreimmediateproblemisthateveniftheseargumentssucceededinestablishingtheirconclusions,thenaturalistcouldeasilydismissthemasirrelevanttoherproject.Aswementionedabove,contemporarynaturalistsdon’ttypicallyseethemselvesasprovidinganalyticallynecessaryandsufficientconditionsforthesatisfactionofconcepts.Instead,theytendtoseethesatisfactionoftheconceptstheyareaccountingforasconnectedbyanecessaryaposteriorilinktotheconditionsthatfigureintheaccount.Themodelforthisconceptionoftheirtaskisthesemanticsofnaturalkindtermsasexpli-catedbySaulKripkeandHilaryPutnam(Kripke1980;Putnam1975b).Onthisconceptionofnaturalistaccounts,anaccountof“morallyright”or“satisfies”willseektoachievethekindofunderstandingof,say,“water”providedbythediscoverythatwaterisHO.Clearly,onthisconceptionoftheenterprise,2theconclusionofanopen-questionargumenthasnopowertoundermineanaccountofwhatdeterminesthesatisfactionconditionsof“water”.ThepointismadeveryclearlybyGilbertHarmanforthecaseofethics:[…]asitstandstheopenquestionargumentisinvalid.Ananalogousargumentcouldbeusedonsomeonewhowasignorantofthechemicalcompositionofwaterto“prove”tohimthatwaterisnotHO.Thisperson2willagreethatitisnotanopenquestionwhetherwateriswaterbutitisanopenquestion,atleastforhim,whetherwaterisHO.Sincethisargument2wouldnotshowthatwaterisnotHO,theopenquestionargumentinethics2cannotbeusedasitstandstoshowthatforanacttobeanactthatoughttobedoneisnotforittohavesomenaturalcharacteristicC(Harman1977:19).Onourconstrualofthenotionofwhatcountsasanopenquestion,Harman’spointisthatthesatisfactionconditionsof“water”aredeterminedbyHO,even2thoughmeaningwaterby“water”doesnotrequirebelievingthatbeingHOis2necessaryforsatisfying“water”.AsimilarpointismadebyScottSoamesconcerningtheeffectoftheargumentonnaturalisticaccountsinsemantics:[…]IamwillingtograntthattheskepticmightberightinmaintainingthatclaimsaboutwhatImeantarenotaprioriconsequencesofnonin-tentionaltruths.IfitwereclearthatanynecessaryconsequenceofasetofclaimsPwasalsoanaprioriconsequenceofP,thenthisadmissionwouldprovidethescepticwithjustwhatheneeds;forthenhecouldforcemetoadmitthatclaimsaboutmeaningmightnotbenecessaryconsequencesofnonintentionaltruths.[…]However,thisargumentativestrategyfails.ThankstotheworkofKripkeandothers,ithasbecomeclearthatmany2219781472578235_txt_print.indd22108/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagenecessaryconsequencesofpropositionsarenotaprioriconsequencesofthem.Consequently,myadmissionthatclaimsaboutmeaningsmaynotbeaprioriconsequencesofnonintentionaltruthsneednotunderminemybeliefthattheyarenecessaryconsequencesofthosetruths(Soames1998:231).Thuswhattheseauthorsareproposingisthatnaturalistscansidesteptheopen-questionargumentbymodellingtheconnectionbetweentheirexpli-candaandtheirexplicantiaontheconnectionbetweennatural-kindtermsandthepropertiesthatdeterminetheirsatisfactionconditions.Analysisoftheconceptwaterdoesn’tsufficefordeterminingthatitissatisfiedbysamplesofHO.Similarly,fortheenlightenednaturalist,analysiswon’tsufficefordeter-2miningthatmorallyrightissatisfiedby,say,actionsthatmaximizeutility,orthatrefersissatisfiedbyapredicateandapropertywhenever,say,tokeningsofthelattercausetokeningsoftheformerundernormalconditions.Itfollowsthatevenifthesenaturalisticaccountsarecorrect,thedescriptions“xmaximizesutility”or“tokeningsofxcausetokeningsofyundernormalconditions”won’tprovidecounterexamplestotheconclusionoftheopen-questionargument.Thispointstrikesmeasincontestable.Iwanttosuggest,however,thattheantinaturalistmightstillhavesomeroomformanoeuvre.Goingbacktowater,thepointthatthedescriptionHOisnotanalyticallysufficientfor2thesatisfactionofwaterisnotopentoquestion.Buttheantinaturalistmightacceptthispoint,assheshould,andstillcontendthatthebehaviourofwaterasanatural-kindtermrendersotherdescriptionsanalyticallysufficientforitssatisfaction.LetmefollowPutnaminusingthetermoperationaldefinitiontorefertothearrayofcriteriathataspeaker(oracommunity)usesforapplyinganatural-kindterm(Putnam1975b).Onthestandardaccountofthesemanticsofanatural-kindterm,thepropertytowhichitrefersissingledoutastheunder-lyingkindpresentinmostsamplesinthespeaker’senvironmentthatsatisfyheroperationaldefinitionfortheterm.Mysuggestionisthattheantinaturalistcouldarguethatthefactthatthereferenceofanatural-kindtermisdeterminedinthiswayisdiscoverablebyanalysis–independentlyofanyfactthatcanbeascertainedonlybyempiricalinvestigation.Weneedempiricalresearchtoascertainwhichpropertyplaysthisrole,butnottoascertainthatthepropertythatplaysthisrole,ifthereisone,willdeterminetheextensionoftheterm.Butifthisiscorrect,itturnsoutthatthereis,afterall,adescriptionthatisanalyticallysufficientforthesatis-factionofwater–namely,xisaninstanceofthekind,ifthereisone,presentinmostofthesamplesinthespeaker’senvironmentthatsatisfyheroperationaldefinitionofwater.19Providingadequatesupportfortheclaimthatthisdescriptionshouldbetreatedasanalyticallysufficientforthesatisfactionofwaterlieswellbeyondthescopeofthispaper.20WhatIamarguingisthatifthisclaimcouldbe2229781472578235_txt_print.indd22208/08/201415:46\nSemanticNormativityandNaturalismadequatelyestablished,thenthethreatoftheopen-questionargumenttonaturalistaccountswouldbereinstated.Forthenaturalistwouldhavetoexplainhow,say,maximizingutility,issingledoutasthepropertythatdeter-minestheextensionofmorallyright.Thisexplanationmayormaynotbethesameastheonegivenforwater,butineithercaseitcanbeexpectedtoinvokeasecond-orderpropertyMwhosesatisfactionbyutilitymaximizationconfersonthelatteritsextension-fixingrole.AndiftheconnectionbetweenmorallyrightandMcanbeclaimedtobeanalytic,wewillhave,asinthecaseofwater,ananalyticallysufficientdescription:xisaninstanceoftheproperty,ifthereisone,thatsatisfiesM.21Thesamesituationcouldbeclaimedtoobtaininsemantics.Onceagain,asecond-orderpropertyTwouldhavetobeinvokedtoexplainhow,say,tokeningsofxcausetokeningsofyundernormalconditionsissingledoutfortheroleoffixingtheextensionofrefers.AndiftheconnectionbetweenTandrefersisanalytic,wewillappeartohaveananalyticallysufficientdescription:beingapredicate-propertypairinstantiatingtherelation,ifthereisone,thatsatisfiesT.Hence,evenifmorallyrightandrefersaretreatedasnatural-kindterms,theywillhaveanalyticallysufficientdescriptions.Therefore,invokingthissemanticmodelwon’tenabletheenlightenednaturalisttosidesteptheconclusionoftheopen-questionargument.Heraccountwillstillbeinconflictwiththeconclusionoftheargument.Normativity?IntheprecedingsectionIhavearguedthattheconclusionofanopen-questionargumentforaconceptCmightstillbeinconflictwithanaturalisticaccountofC,eveniftheaccountisconstruedalongthelinesofthenatural-kindmodel.Obviouslythiswouldposeathreatforthenaturalistonlyiftheargumentweresound,andthesoundnessoftheargumentineachcasecannotbeplausiblyasserteduntilsupportisprovidedforitsfirstpremise–theprinciplethatthereisnodescriptionthathastobebelievedtobesufficientforthesatisfactionoftheconceptbyanyonewhohastheconcept.Iwanttoclosebyconsideringbrieflyhowthisprinciplecouldbesupportedinethicsandinsemantics.Thestandardapproachtothistaskinethicsistoappealtothenormativedimensionofethicalconcepts.Forthecaseofgood,Darwall,GibbardandRailtonexpressthethoughtinthefollowingterms:Attributionsofgoodnessappeartohaveaconceptuallinkwiththeguidanceofaction,alinkexploitedwheneverweglosstheopenquestion“IsPreallygood?”as“Isitclearthat,otherthingsbeingequal,wereallyoughtto,ormust,devoteourselvestobringingaboutP?”(Darwall,Gibbard,andRailton1992:117).2239781472578235_txt_print.indd22308/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageFormorallyright,thesamethoughtisexpressedby2,above(page207).Thislinkenablesustoinvoketheintuitionthatitisperfectlypossibletoimagine,foranyactiondescription,someonewhoacceptsthedescriptionbutdoesn’tthinkthattheactionoughttobeperformed.AsDarwall,GibbardandRailtonputit:Ourconfidencethattheopennessoftheopenquestiondoesnotdependuponanyerrororoversightmaystemfromourseemingabilitytoimagine,foranynaturalisticpropertyR,clear-headedbeingswhowouldfailtofindappropriatereasonormotiveforactioninthemerefactthatRobtains(orisseentobeintheoffing)(Darwall,Gibbard,andRailton1992:117).Theseconsiderationssuggestanargumentfor14withthefollowingpremises:20.ForeveryactiondescriptionD,itispossibletobelieve,forsomex,bothDxandthatxdoesnothavetobedone.21.Youdon’tmeanmorallyrightby“morallyright”unless,foreveryx,ifyoubelievethatxdoesn’thavetobedone,youbelieve“~Morallyright(x)”.Forpredicatereferencewemighttrythesameapproach,deriving17fromthefollowingpremises:22.ForanydescriptionD,itispossibletobelieve,forsomeS,Panda,bothD(S,P,a)andthatSdoesn’thavetoascribePtoa.23.Youdon’tmeansatisfiesby“satisfies”unless,forallS,P,a,ifyoubelievethatSdoesn’thavetoascribePtoathenyoubelieve“~satisfies(S,P,a)”.Notice,though,thatthisapproachwouldtakeusbacktothedifficultiesthatweencounteredtryingtouseHume’sthoughttounderminesemanticnaturalism.For23andrelatedprincipleswouldhavenoplausibilityunlessthetruthprescriptioncouldbeestablishedbyconceptualanalysis.However,Iwanttosuggestthatinvokingthenormativityofmeaningatthispointisnotmandatory.ThealternativethatIwanttoputforward,inclosing,isthattherolethatnormativityplaysintheethicalcasecouldbeplayedinsemanticsbydisquotation.WhatwewouldneedtoinvokeistheclaimthathavingtheconceptofsatisfactionrequiresbelievingthataisnotsatisfiedbyP,asmeantbyyou,wheneveryoubelieve~Pa.Thisclaimwouldsupplytheargumentwiththefollowingpremise:2249781472578235_txt_print.indd22408/08/201415:46\nSemanticNormativityandNaturalism24.Youdon’tmeansatisfiesby“satisfies”unless,forallP,a,ifyoubelieve~Payoubelieve“~satisfies(I,P,a)”.22Now,tocompletetheargumentfor17wewouldonlyneedtoarguethatforanydescriptionD,itispossibletobelieveD(I,P,a)whilenotbelievingPa.Thiswouldgiveusthefollowingprinciple:25.ForanydescriptionD,itispossibletobelieve,forsomePanda,bothD(I,P,a)and~Pa.And25and24wouldestablish17inthesamewayinwhich20and21establish14.Followingthisapproachwewouldbeabletomountaversionoftheopen-questionargumentinsemanticsthatdoesn’trelyontheassumptionthatmeaningisnormative.Notes1.IamgratefultoMarioGómezTorrente,ChristopherJay,MarkKalderon,GenovevaMartí,MikeMartinandtheeditorsofthisvolume.2.InZalabardo1997Iarguedagainstthisattribution.3.InanascriptionofPtoa,asIamusingthephrase,SassertsthatasatisfiesP.4.HereandelsewhereIusepredicativetermsassingulartermsinpropositionsinwhichtheyfigureasobjectsofpredication.5.Noticethatthesameobjectionwouldapplytoaversionoftheargumentthatreplacedtheobjectiveobligationexpressedby7withasubjectiveobligation:(∀x)(∀S)(xisanascriptionofPtoabyS&SbelievesthatP,asmeantbyher,isnotsatisfiedbya⊃Sshouldn’tdox)6.ThispointismadeinHattiangadi2006.SeeWhiting2007forareply.7.ThisapproachisdiscussedinWikforss2001:209–12.8.ThisistheapproachassociatedwithPaulGrice.SeeGrice1957.9.See,forexample,theAppendixtoFodor1987.10.SeeinthisconnectionthedebatebetweenColinMcGinnandPaulBoghossian.McGinn(1984:147)arguedthatitwasnotpossibletoraiseaproblemofthenormativityofmentalcontentanalogoustotheproblemofthenormativityoflinguisticmeaning.Boghossian(1989:514)offersareplythatrestsontheassumptionthatmentalcontenthassyntacticallyidentifiablebearers.Iamsuggestingthatthisassumptionisunnecessary.11.Boghossian(2003:41–5)offersaninterestingdiscussionofthisissue.12.Thepositionisindependentofrelativismabouttruth.Itisaviewaboutthegoalofcognition,whichisperfectlycompatiblewithanon-relativistnotionoftruth.13.SeealsoShah2006:489.14.Vellemanappearstoconcedethatthebehaviouraloutcomewillbedifferentineachcase(Velleman2000:272),butheclaimsthatthedifferencesareduenottoadifferenceintheunderlyingbehaviouraldisposition,buttothepresence2259781472578235_txt_print.indd22508/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageinthecaseofimaginingofcountervailingbeliefs.However,hedoesn’tofferanargumentforexplainingthedifferenceinthisway,ratherthanasarisingfromdifferentbehaviouraldispositions.15.Thissectioncanbeskippedwithoutlossofcontinuity.16.Iamusingpropositionsastheirownnames.17.IamnotclaimingthatMoorewouldhaveappliedtomorallyrighttheargumentthathedevelopsforgood.18.Believing“f(x)”,forsomex,isbelievingthepropositionexpressedbyasubsti-tutioninstanceof“f(x)”.19.Noticethattheanalyticityofthisconnectioniscompatiblewiththepossibilitythatdifferentspeakershavedifferentoperationaldefinitionsofwater,solongas,intheirrespectiveenvironments,theyalltrackthepresenceofHO.220.Putnam,forone,thinksthatit’snotpossibletoobtainanalyticallysufficientconditionsforthesatisfactionofanatural-kindterm(Putnam1975b),butthedifficultiesthatheraisesforthecandidatesthatheconsiders(Putnam1975a)donotinvalidatemyproposal.21.AnaccountofmoralconceptsalongtheselineshasbeendefendedbyFrankJackson.SeeJackson1997.22.“I”hereisthefirst-personpronoun.ReferencesBlackburn,Simon(1984),“TheIndividualStrikesBack”,Synthese58:281–302.Boghossian,Paul(1989),“TheRule-FollowingConsiderations”,Mind98:507–49.—(2003),“TheNormativityofContent”,PhilosophicalIssues13:31–45.Darwall,Stephen,AllanGibbard,andPeterRailton(1992),“TowardFindesiècleEthics:SomeTrends”,PhilosophicalReview101:115–89.Fodor,JerryA.(1987),Psychosemantics.TheProblemofMeaninginthePhilosophyofMind.Cambridge,MA:MITPress.Grice,H.P(1957),“Meaning”,PhilosophicalReview66:377–88.Harman,Gilbert(1977),TheNatureofMorality.AnIntroductiontoEthics.NewYork:OxfordUniversityPress.Hattiangadi,Anandi(2006),“IsMeaningNormative?”MindandLanguage21:220–40.Hume,David(1978),ATreatiseofHumanNature.2ndedn.Oxford:ClarendonPress.Jackson,Frank(1997),FromMetaphysicstoEthics:ADefenseofConceptualAnalysis.Oxford:OxfordUniversityPress.Kripke,Saul(1980),NamingandNecessity.Oxford:Blackwell.—(1982),WittgensteinonRulesandPrivateLanguage.Oxford:Blackwell.McDowell,John(1984),“WittgensteinonFollowingaRule”,Synthese58:325–63.McGinn,Colin(1984),WittgensteinonMeaning.Oxford:BasilBlackwell.Moore,G.E(1903),PrincipiaEthica.Cambridge:CambridgeUniversityPress.Putnam,Hilary(1975a),“IsSemanticsPossible?”InMind,LanguageandReality.PhilosophicalPapers,Volume2.Cambridge:CambridgeUniversityPress.—(1975b),“TheMeaningof‘Meaning’”,InMind,LanguageandReality.PhilosophicalPapers,Volume2.Cambridge:CambridgeUniversityPress.2269781472578235_txt_print.indd22608/08/201415:46\nSemanticNormativityandNaturalismShah,Nishi(2003),“HowTruthGovernsBelief”,PhilosophicalReview112:447–82.—(2006),“ANewArgumentforEvidentialism”,PhilosophicalQuarterly56:481–98.Shah,Nishi,andJ.DavidVelleman(2005),“DoxasticDeliberation”,PhilosophicalReview114:497–534.Soames,Scott(1998),“SkepticismaboutMeaning:Indeterminacy,Normativity,andtheRule-FollowingParadox”,CanadianJournalofPhilosophySupplementaryVolume23:211–49.Velleman,J.David(2000),“OntheAimofBelief”,InThePossibilityofPracticalReason.Oxford:ClarendonPress.Whiting,Daniel(2007),“TheNormativityofMeaningDefended”,Analysis67:133–40.Wikforss,ÅsaMaria(2001),“SemanticNormativity”,PhilosophicalStudies102:203–26.Wright,Crispin(1984),“Kripke’sAccountoftheArgumentagainstPrivateLanguage”,JournalofPhilosophy81:759–77.Zalabardo,JoséL(1997),“Kripke’sNormativityArgument”,CanadianJournalofPhilosophy27(4),467–88.—(2010),“WhyBelievetheTruth?ShahandVellemanontheAimofBelief”,PhilosophicalExplorations13:1–21.2279781472578235_txt_print.indd22708/08/201415:46\nAnalyticity,Apriority,10ModalityAlbertCasulloIntheintroductiontohisCritiqueofPureReason,ImmanuelKantintroducesaconceptualframeworkthatinvolvesthreedistinctions:theepistemicdistinctionbetweenaprioriandempirical(oraposteriori)knowledge;themetaphysicaldistinctionbetweennecessaryandcontingentpropositions;andthesemanticdistinctionbetweenanalyticandsyntheticpropositions.Kantutilizeshisframeworktoposefourquestions:(1)Whatisaprioriknowledge?(2)Isthereaprioriknowledge?(3)Whatistherelationshipbetweentheaprioriandthenecessary?(4)Istheresyntheticaprioriknowledge?Inresponse,he(1965:43)maintainsthataprioriknowledgeis“independentofexperience”,contrastingitwithaposterioriknowledge,whichhasits“sources”inexperience.Heofferstwocriteriaforaprioriknowledge:necessityandstrictuniversality;andarguesthat,sincemathematicalpropositionsarenecessary,mathematicalknowledgeisapriori.Kant’sclaimthatnecessityisacriterionoftheapriorientails(K1)Allknowledgeofnecessarypropositionsisapriori,butnot(K2)Allpropositionsknownaprioriarenecessary.Finally,Kantmaintainsthatallpropositionsoftheform“AllAareB”areeitheranalyticorsynthetic:analyticifthepredicateiscontainedinthesubject;syntheticifitisnot.Utilizingthisdistinction,hearguesthat(K3)Allknowledgeofanalyticpropositionsisapriori;and(K4)Somepropositionsknownaprioriaresynthetic.Insupportof(K4),Kantclaimsthatthepredicatetermsof“7+5=12”and“Thestraightlinebetweentwopointsistheshortest”arenotcontainedintheirrespectivesubjects.2289781472578235_txt_print.indd22808/08/201415:46\nAnalyticity,Apriority,ModalityKant’sconceptualframeworkandthefourquestionsthatheposeswithinitareatthecentreofmuchcurrentphilosophicaldiscussion.First,thereisdisagreementoverhowtoarticulateKant’scharacterizationofaprioriknowledgeandwhetherthatcharacterization,howeverarticulated,isadequate.Second,themostfundamentaldivisionincontemporaryphilosophyisbetweenthosewhoacceptandthosewhorejecttheexistenceofaprioriknowledge.Kant’ssupportingargumentplaysacentralroleinthedebate.Third,SaulKripkerejectsboth(K1)and(K2),buthisexamplesofnecessaryaposterioriknowledgeandcontingentaposterioriknowledgeremaincontro-versial.Finally,thedenialof(K4)byproponentsoflogicalempiricismandW.V.Quine’ssubsequentrejectionoftheanalytic/syntheticdistinctioncontinuetodominatediscussionsofaprioriknowledge.TheConceptofaprioriKnowledgeKant’scharacterizationoftheaprioriisnotfullyarticulated.Hedoesnotspelloutthesenseinwhichaprioriknowledgemustbe“independent”ofexperienceorthesenseinwhichaposterioriknowledgehasits“source”inexperience.Itisgenerallyacceptedthat,byasourceofknowledge,Kantmeansasourceofjustification.SotheKantianconceptionofaprioriknowledgecomesto:(APK)SknowsapriorithatpifandonlyifS’sbeliefthatpisjustifiedaprioriandtheotherconditionsonknowledgearesatisfied;and(APJ)S’sbeliefthatpisjustifiedaprioriifandonlyifS’sjustificationforthebeliefthatpdoesnotdependonexperience.(APJ)hasbeencriticisedfromtwodirections.First,somemaintainthatitisnotsufficientlyinformative;ittellsuswhatapriorijustificationisnot,butnotwhatitis.Hence,LaurenceBonJour(1985)rejectstheKantianconceptionofapriorijustificationinfavourofthetraditionalrationalistconception:(AP1)S’sbeliefthatpisjustifiedaprioriifandonlyifSintuitively“sees”orapprehendsthatpisnecessarilytrue.AlvinPlantinga(1993)andBonJour(1998)offervariantsof(AP1).Second,othersmaintainthatthesenseof“dependence”relevanttoapriorijustificationrequiresarticulation,andhaveofferedtwocompetingaccounts.AlbertCasullo(2003)endorses(AP2)S’sbeliefthatpisjustifiedaprioriifandonlyifS’sbeliefthatp2299781472578235_txt_print.indd22908/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageisnon-experientiallyjustified(i.e.,justifiedbysomenon-experientialsource).HilaryPutnam(1983)andPhilipKitcher(1983)favour(AP3)S’sbeliefthatpisjustifiedaprioriifandonlyifS’sbeliefthatpisnon-experientiallyjustifiedandcannotbedefeatedbyexperience.(AP1)and(AP3)faceseriousobjections.Theterm“see”isusedmetaphoricallyin(AP1).Letusassumethatitsharesonebasicfeaturewiththeliteraluseof“see”:“Sseesthatp”entails“Sbelievesthatp”.Hence,(AP1)hastheconsequencethatifS’sbeliefthatpisjustifiedapriorithenSbelievesthatpisnecessarilytrue.Thisconsequencefacestwoproblems.SupposethatSamisamathematicianwhobelievessomegenerallyacceptedtheoremTonthebasisofavalidproof.Presumably,Sam’sbeliefisjustified.ButsupposethatSamisalsoaseriousstudentofphilosophywhohascometodoubtthecogencyofthedistinctionbetweennecessaryandcontingentpropositionsand,asaconse-quence,refrainsfrommodalbeliefs.ItisimplausibletomaintainthatSam’sbeliefthatTisnotjustifiedapriorimerelybecauseofhisviewsaboutacontroversialmetaphysicalthesis.(AP1)isalsothreatenedwitharegress.ItentailsthatifS’sbeliefthatpisjustifiedapriorithenSbelievesthatnecessarilyp.MustS’sbeliefthatnecessarilypbejustified?Ifnot,itishardtoseewhyitisanecessaryconditionofhavinganapriorijustifiedbeliefthatp.Ifso,thenpresumablyitisjustifiedapriori.ButinorderforS’sbeliefthatnecessarilyptobejustifiedapriori,Smustbelievethatnecessarilynecessarilyp,andthesamequestionariseswithrespecttothelatterbelief.Mustitbejustifiedornot?Hence,(AP1)eitherfacesaninfiniteregressofjustifiedmodalbeliefsoriscommittedtotheviewthathavinganunjustifiedbeliefthatnecessarilypisanecessaryconditionofhavingajustifiedbeliefthatp.(AP3)isalsoopentoseriousobjection.Kripke(1980)andKitcher(1983)maintainthatanadequateconceptionofaprioriknowledgeshouldallowforthepossibilitythatapersonknowsempiricallysomepropositionthatheorshecanknowapriori.(AP3)precludesthispossibility.Assumethat(A)SknowsempiricallythatpandScanknowapriorithatp.Fromtheleftconjunctof(A),itfollowsthat(1)S’sbeliefthatpisjustifiedempirically,kwhere“justified”abbreviates“justifiedtothedegreeminimallysufficientkforknowledge”.Considernowtheempiricalsourcesthathavebeenalleged2309781472578235_txt_print.indd23008/08/201415:46\nAnalyticity,Apriority,Modalitytojustifymathematicalpropositions:countingobjects,readingatextbook,consultingamathematician,andcomputerresults.Eachofthesesourcesisfallibleinanimportantrespect.Thejustificationeachconfersonabeliefthatpisdefeasiblebyanempiricallyjustifiedoverridingdefeater;thatis,byanempiricallyjustifiedbeliefthatnot-p.IfS’sbeliefthatpisjustifiedbycountingacollectionofobjectsandarrivingataparticularresult,thenitispossiblethatSrecountsthecollectionandarrivesatadifferentresult.IfS’sbeliefthatpisjustifiedbyatextbook(mathematician,computerresult)thatstatesthatp,thenitispossiblethatSencountersadifferenttextbook(mathematician,computerresult)thatstatesthatnot-p.Ineachcase,thelatterresultisanempiricallyjustifiedoverridingdefeaterforS’soriginaljustification.Hence,giventhefalliblecharacterofempiricaljustification,itfollowsthat(2)S’sempiricaljustificationforthebeliefthatpisdefeasiblebyanempiri-callyjustifiedbeliefthatnot-p.(2),however,entailsthat(3)S’sbeliefthatnot-pisjustifiableempirically,dwhere“justifiable”abbreviates“justifiabletothedegreeminimallysufficientdtodefeatS’sjustifiedbeliefthatp”.Furthermore,theconjunctionof(AP3)andktherightconjunctof(A)entails(4)ItisnotthecasethatS’snon-experientialjustificationforthebeliefthatkpisdefeasiblebyS’sempiricallyjustifiedbeliefthatnot-p.(4),however,entailsthat(5)ItisnotthecasethatS’sbeliefthatnot-pisjustifiableempirically.dTheconjunctionof(3)and(5)isacontradiction.Hence,(AP3)isincompatiblewith(A).(AP2),however,iscompatiblewith(A)sincetheconjunctionof(AP2)andtherightconjunctof(A)doesnotentail(4).SinceboththetraditionalrationalistconceptionandthePutnam–KitcherarticulationoftheKantianconceptionofapriorijustificationareopentoseriousobjectionswhich(AP2)avoids,(AP2)providesthesuperiorarticulationoftheconceptofapriorijustification.2319781472578235_txt_print.indd23108/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageTheExistenceofaprioriKnowledgeTraditionalapriorism:KantKantoffersthemostinfluentialtraditionalargumentfortheexistenceofaprioriknowledge.Kant(1965:43)holdsthatnecessityisacriterionoftheapriori:“ifwehaveapropositionwhichinbeingthoughtisthoughtasnecessary,itisanapriorijudgment;…”He(1965:52)thengoesontoarguethat“mathematicalpropositions,strictlysocalled,arealwaysjudgmentsapriori,notempirical;becausetheycarrywiththemnecessity,whichcannotbederivedfromexperience.”Kant’sargument,theArgumentfromNecessity,canbepresentedasfollows:(N1)Mathematicalpropositionsarenecessary.(N2)Onecannotknowanecessarypropositiononthebasisofexperience.(N3)Therefore,onecannotknowmathematicalpropositionsonthebasisofexperience.Thephrase“knowanecessaryproposition”in(N2)maskssomeimportantdistinctions:(A)Sknowsthetruth-valueofpjustincaseSknowsthatpistrueorSknowsthatpisfalse.(B)SknowsthegeneralmodalstatusofpjustincaseSknowsthatpisanecessaryproposition(i.e.,necessarilytrueornecessarilyfalse)orSknowsthatpisacontingentproposition(i.e.,contingentlytrueorcontingentlyfalse).(C)SknowsthespecificmodalstatusofpjustincaseSknowsthatpisneces-sarilytrueorSknowsthatpisnecessarilyfalseorSknowsthatpiscontingentlytrueorSknowsthatpiscontingentlyfalse.(A)and(B)arelogicallyindependent:onecanknowonebutnottheother.OnecanknowthattheGoldbachConjectureiseithernecessarilytrueorneces-sarilyfalsebutnotknowwhetheritistrueorfalse.Similarly,onecanknowthatthePythagoreanTheoremistrue,butnotknowwhetheritisnecessarilytrueorcontingentlytrue.Thespecificmodalstatusofapropositionisjusttheconjunctionofitstruth-valueanditsgeneralmodalstatus.Therefore,onecannotknowthespecificmodalstatusofapropositionunlessoneknowsbothitstruth-valueanditsgeneralmodalstatus.2329781472578235_txt_print.indd23208/08/201415:46\nAnalyticity,Apriority,ModalityWecannowdistinguishtworeadingsof(N2):(N2A)Onecannotknowthetruth-valueofanecessarypropositiononthebasisofexperience;and(N2B)Onecannotknowthegeneralmodalstatusofanecessarypropositiononthebasisofexperience.Kant(1965:52)supports(N2)withtheobservationthat“Experienceteachesusthatathingissoandso,butnotthatitcannotbeotherwise.”Thisobservationsupports(N2B)butnot(N2A),sinceKantallowsthatexperiencecanprovideevidencethatsomethingisthecase,butdeniesthatitcanprovideevidencethatsomethingmustbethecase.Theconclusionoftheargument,however,isthatknowledgeofthetruth-valueofmathematicalpropositions,suchasthat7+5=12,isapriori.Kant’sargument,theKantianArgument,cannowbearticulatedasfollows:(N1)Mathematicalpropositionsarenecessary.(N2B)Onecannotknowthegeneralmodalstatusofanecessarypropositiononthebasisofexperience.(N3A)Therefore,onecannotknowthetruth-valueofmathematicalpropositionsonthebasisofexperience.TheKantianArgumentturnsonthisprinciple:(KP)Ifthegeneralmodalstatusofpisknowableonlyapriori,thenthetruth-valueofpisknowableonlyapriori.(KP),however,isfalse.Ifonecanknowonlyapriorithatapropositionisnecessary,thenonecanknowonlyapriorithatapropositioniscontingent.Theevidencerelevanttodeterminingthelatteristhesameasthatrelevanttodeterminingtheformer.Forexample,ifonedeterminesthat“2+2=4”isnecessarybytryingtoconceiveofitsfalsehoodandfailing,onedeterminesthat“Kantisaphilosopher”iscontingentbytryingtoconceiveofitsfalsehoodandsucceeding.Butfromthefactthatonecanknowonlyapriorithatthepropo-sition“Kantisaphilosopher”iscontingent,itdoesnotfollowthatonecanknowonlyapriorithattheproposition“Kantisaphilosopher”istrue.Clearly,itisknowableaposteriori.RoderickChisholm(1977)suggeststhefollowingreformulationoftheArgumentfromNecessity,theModalArgument:2339781472578235_txt_print.indd23308/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguage(N1)Mathematicalpropositionsarenecessary.(N2B)Onecannotknowthegeneralmodalstatusofanecessarypropositiononthebasisofexperience.(N3B)Therefore,onecannotknowthegeneralmodalstatusofmathematicalpropositionsonthebasisofexperience.TheModalArgumentfacesadifferentproblem.Whyaccept(N2B)?Kantmaintainsthatexperiencecanteachusonlywhatisthecase.Butagooddealofourordinarypracticalknowledgeandthebulkofourscientificknowledgeprovideclearcounterexamplestotheclaim.MyknowledgethatmypenwillfallifIdropitdoesnotprovideinformationaboutwhatisthecasefortheantecedentiscontrary-to-fact.Scientificlawsarenotmeredescriptionsoftheactualworld.Theysupportcounterfactualconditionalsand,hence,provideinformationbeyondwhatistrueoftheactualworld.Intheabsenceoffurthersupport,(N2B)shouldberejected.ModerateApriorism:LogicalEmpiricismAsecondstrategyfordefendingtheexistenceofaprioriknowledgeisofferedbyproponentsoflogicalempiricism,suchasA.J.Ayer(1952)andCarlHempel(1972),whorejectJohnStuartMill’scontentionthatknowledgeofbasicmathematicalpropositions,suchasthat2x5=10,isbasedoninductionfromobservedcases.Bothdrawattentiontothefactthat,ifoneisjustifiedinbelievingthatsomegeneralpropositionistrueonthebasisofexperience,thencontraryexperiencesshouldjustifyoneinbelievingthatthepropositionisfalse.Butnoexperienceswouldjustifyoneinbelievingthatamathematicalproposition,suchasthat2x5=10,isfalse.Suppose,forexample,thatoneweretocountwhatappeartobefivepairsofshoesandarriveattheresultthattherewereonlynineshoes.Ayer(1952:75–6)contendsthat[o]newouldsaythatIwaswronginsupposingthattherewerefivepairsofobjectstostartwith,orthatoneoftheobjectshadbeentakenawaywhileIwascounting,orthattwoofthemhadcoalesced,orthatIhadcountedwrongly.Onewouldadoptasanexplanationwhateverempiricalhypothesisfittedinbestwiththeaccreditedfacts.Theoneexplanationwhichwouldinnocircumstancesbeadoptedisthattenisnotalwaystheproductoftwoandfive.SinceAyermaintainsthatwewouldnotregardanyexperiencesasevidence2349781472578235_txt_print.indd23408/08/201415:46\nAnalyticity,Apriority,Modalitythatamathematicalpropositionisfalse,heconcludesthatnoexperiencesprovideevidencethattheyaretrue.Ayer’sargument,theIrrefutabilityArgument,canbestatedasfollows:(A1)Noexperiencesprovideevidencethatmathematicalpropositionsarefalse.(A2)Ifnoexperiencesprovideevidencethatmathematicalpropositionsarefalse,thennoexperiencesprovideevidencethattheyaretrue.(A3)Therefore,noexperiencesprovideevidencethatmathematicalproposi-tionsaretrue.Ayer’sexampleprovidesveryweaksupportfor(A1)because(a)itdoesnottakeintoaccountthenumberofexperiencesthatconfirmthepropo-sitioninquestion,(b)itinvolvesonlyasingleexperiencethatdisconfirmstheproposition,and(c)thehypotheseswhichareinvokedtoexplainawaythedisconfirmingexperienceasapparentarenotsubjectedtoindependentempiricaltest.Inasituationwherethereisastrongbackgroundofsupportingexperientialevidenceforaninductivegeneralizationandanisolateddiscon-firmingexperience,itisreasonabletodiscountthedisconfirmingexperienceasapparentandtoexplainitawayonwhateverempiricalgroundsaremostplausible.Butitdoesnotfollowthatthegeneralizationinquestioncannotbedisconfirmedbyexperience.Inordertoprovidestrongersupportfor(A1),Ayer’sexamplemustberevisedasfollows:increasethenumberofexperiencesthatdisconfirmthepropositionsothatitislargerelativetothenumberofexperiencesthatconfirmit;andsubjectthehypothesesinvokedtoexplainawaythedisconfirmingexperiencesasapparenttoindependentteststhatfailtosupportthem.Letusnowsupposethatonehasaverylargenumberofexperiencesthatdisconfirmthepropositionthat2x5=10and,furthermore,thatempiricalinvestigationsofthehypothesesinvokedtoexplainawaythesedisconfirmingexperiencesasapparentproduceverylittle,ifany,supportforthehypotheses.Giventheserevisions,Ayercancontinuetoendorsepremise(A1)onlyattheexpenseofholdingempiricalbeliefsthatareatoddswiththeavailableevidence.InductiveRadicalEmpiricism:MillRadicalempiricismistheviewthatdeniestheexistenceofaprioriknowledge.Onestrategyfordenyingtheexistenceofaprioriknowledgeistoofferradicalempiricistaccountsofthosedomainsofknowledgethatproponentsofthea2359781472578235_txt_print.indd23508/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageprioriallegetobeknowableonlyapriori.Sincemathematicalknowledgehasreceivedthemostattention,wewillfocusonit.Radicalempiricistaccountsofmathematicalknowledgefallintotwobroadcategories:inductiveandholistic.JohnStuartMill(1973)offersaninductiveempiricistaccountofmathematicalknowledge.Inductiveempiricismwithrespecttoadomainofknowledgeinvolvestwotheses:(1)somepropositionswithinthatdomainareepistemi-callymorebasicthantheothers,inthesensethatthenon-basicpropositionsderivetheirjustificationfromthebasicpropositionsviainference;and(2)thebasicpropositionsareknownbyinductiveinferencefromobservedcases.Mill’sprimarythesisisthatthebasicpropositions,theaxiomsanddefinitions,ofarithmeticandgeometryareknownbyinductionfromobservedcases.Mill’saccountfacesseriousobjections,suchasthoseofferedbyGottlobFrege(1974).Letusassume,however,thattheseobjectionscanbedeflectedandthatMillprovidesadefensibleinductiveempiricistaccountofmathe-maticalknowledge.Doesthisshowthatmathematicalknowledgeisnotapriori?IfMill’saccountisdefensible,thenitfollowsthatKant’sclaimthatonecannotknowmathematicalpropositionsonthebasisofexperienceisfalse.Itdoesnotfollow,however,thattheweakerclaimthatthereisaprioriknowledgeofmathematicalpropositionsisfalse.Fromthefactthatoneknows(orcanknow)mathematicalpropositionsonthebasisofexperience,itdoesnotimmediatelyfollowthatonedoesnot(orcannot)knowmathe-maticalpropositionsapriori.Mill(1973:231–2)addressesthegapinhisargumentwiththefollowingconsiderations:Theycannot,however,butallowthatthetruthoftheaxiom,Twostraightlinescannotincloseaspace,evenifevidentindependentlyofexperience,isalsoevidentfromexperience.…Wherethenisthenecessityforassumingthatourrecognitionofthesetruthshasadifferentoriginfromtherestofourknowledge,whenitsexistenceisperfectlyaccountedforbysupposingitsorigintobethesame?…Theburdenofproofliesontheadvocatesofthecontraryopinion:itisforthemtopointoutsomefact,inconsistentwiththesuppositionthatthispartofourknowledgeofnatureisderivedfromthesamesourcesaseveryotherpart.HeattemptstoclosethegapbyappealingtoaversionoftheExplanatorySimplicityPrinciple:Ifaputativesourceofknowledgeisnotnecessarytoexplainknowledgeofthepropositionswithinsomedomain,thenitisnotasourceofknowledgeofthepropositionswithinthatdomain.Mill’sargument,theExplanatorySimplicityArgument,canbearticulatedasfollows:2369781472578235_txt_print.indd23608/08/201415:46\nAnalyticity,Apriority,Modality(M1)Inductiveempiricismprovidesanaccountofmathematicalknowledgebasedoninductivegeneralizationfromobservedcases.(M2)αisasourceofknowledgeforsomedomainDonlyifαisnecessarytoexplainknowledgeofsomepropositionswithinD.(M3)Therefore,mathematicalknowledgeisnotapriori.Theburdenoftheargumentiscarriedby(M2),theExplanatorySimplicityPrinciple.Casullo(2005)arguesthattheExplanatorySimplicityPrincipleisfalsebecauseitrulesoutthepossibilityofafamiliarformofepistemicoverdetermi-nation.Thejustificationofsomeofourbeliefsisoverdeterminedbydifferentsources.Therearesomebeliefsforwhichwehavemorethanonejustification,eachofthosejustificationsderivesfromadifferentsource,andeach,intheabsenceoftheothers,issufficienttojustifythebeliefinquestion.Forexample,I’vemisplacedmywalletandwonderwhereImighthaveleftit.Isuddenlyrecallhavingleftitonthekitchentablelastnight.Myrecollectionjustifiesmybeliefthatmywalletisonthekitchentable.But,justtobesure,Iwalkouttothekitchentocheck.Tomyrelief,Iseemywalletonthetable.Myseeingmywalletonthetablealsojustifiesmybeliefthatmywalletisonthetable.Soheremyjustificationisoverdeterminedbydifferentsources.Ifthejustificationofmybeliefisoverdeterminedbytwodifferentsources,itfollowsthatmybeliefisjustifiedbytwodifferentsources.Hence,intheabsenceofanargumentagainstthepossibilityofepistemicoverdeterminationbydifferentsources,Mill’sappealtotheExplanatorySimplicityPrinciplesimplybegsthequestion.HolisticRadicalEmpiricism:QuineQuinerejectsinductiveempiricism.Herejectstheideathattherearebasicmathematicalpropositionswhich,takeninisolation,aredirectlyjustifiedbyobservationandinductivegeneralization.Quine’saccountofmathematicalknowledgeisaversionofholisticempiricism.Mathematicalpropositionsarecomponentsofscientifictheories.Theyarenottesteddirectlyagainstobser-vation,butonlyindirectlyviatheirobservationalconsequences.Moreover,theydon’thaveobservationalconsequencesinisolation,butonlyinconjunctionwiththeotherpropositionsofthetheory.Hence,accordingtoholisticempir-icism,entirescientifictheories,includingtheirmathematicalcomponents,areindirectlyconfirmedordisconfirmedbyexperienceviatheirobservationalconsequences.OurmainconcerniswhetherQuine’saccountofmathematicalknowledge2379781472578235_txt_print.indd23708/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageprovidesanargumentagainsttheexistenceofaprioriknowledge.TheargumentofQuine’s(1963)classicpaper,“TwoDogmasofEmpiricism”,remainscontro-versial.HisattackisdirectedatavariantofFrege’sconceptionofanalyticity:astatementisanalyticifitcanbeturnedintoalogicaltruthbyreplacingsynonymswithsynonyms.Hisprimarytargetisthenotionofsynonymyandhisleadingcontentionscanbesummarizedasfollows.First,synonymycannotbeexplainedintermsofdefinition,interchangeabilitysalveveritate,orsemanticrules.Second,theverificationtheoryofmeaningdoesprovideanaccountofstatementsynonymy;butthetheorypresupposesradicalreductionism,whichisafailedprogramme.Avestigeofthatprogrammesurvivesintheviewthatindividualstatementsadmitofconfirmationordisconfirmation.Quineobjectstothisvestigesinceitlendscredencetotheideathattherearestatementsconfirmednomatterwhat,whichhe(1963:43)rejectsonthegroundsthat“nostatementisimmunetorevision”.TherearetwostrandstoQuine’sargument.Thefirstchallengesthecogencyofsemanticconceptssuchassynonymy.Thesecondchallengestheremainingvestigeofreductionism.Neithercontention,however,isexplicitlydirectedataprioriknowledge.Hence,ifQuine’sargumentdoespresentachallengetotheexistenceofaprioriknowledge,thensomeadditionalpremiseisnecessarythatconnectsoneofitsexplicittargetstotheapriori.OnestandardreadingofQuine’sargumentisthathisgoalistounderminethecentraltenetoflogicalempiricism,(LE)Allaprioriknowledgeisofanalytictruths,byshowingthattheanalytic/syntheticdistinctionisnotcogent.Supposewegrantthat(LE)isindeedQuine’stargetandthathisargumentsestablishthattheanalytic/syntheticdistinctionisnotcogent.Itdoesnotfollowthateithertheclaimofproponentsof(LE)thatthereisaprioriknowledgeortheirsupportingargumentforthatclaimisnotcogent.Logicalempiricists,suchasAyer,donottake(LE)tobeconstitutiveoftheconceptofaprioriknowledge.Moreover,theydonotbasetheircasefortheexistenceofaprioriknowledgeonapremise,suchas(LE),thatinvolvestheconceptofanalytictruth.TheyendorsetheKantianconceptionofaprioriknowledgeandbasetheircaseforaprioriknowledgeontheIrrefutabilityArgument.Theythengoontoofferindependentargumentstoshowthatpropositionsknownaprioriareanalytic.Hence,Quine’sargumentestablishesonlythattheirthesisaboutthenatureofthepropositionsknownaprioriisnotcogent.Butfromthisitdoesnotfollowthateithertheirclaimthatthereisaprioriknowledgeortheirsupportingargumentforthatclaimisnotcogent.OnemightattempttobolsterQuine’sargumentbymaintainingthat(LE)isconstitutiveoftheconceptofaprioriknowledge.Iftheconceptofapriori2389781472578235_txt_print.indd23808/08/201415:46\nAnalyticity,Apriority,Modalityknowledgeinvolvestheconceptofanalytictruthandthelatterconceptisincoherent,thentheformerisalsoincoherent.Therearetwowaysinwhichtheconceptofaprioriknowledgemightinvolvethesemanticconceptofanalytictruth:explicitlyorimplicitly.Aswesawinsection1,neitherKant’sconceptionofaprioriknowledge,(APK),norhisconceptionofapriorijustification,(APJ),explicitlyinvolvestheconceptofanalytictruth.Theonlyplausiblecaseformaintainingthattheconceptofaprioriknowledgeimplicitlyinvolvesthatconceptisbasedontwopremises:(1)theconceptofaprioriknowledgeinvolvestheconceptofnecessarytruth;and(2)theconceptofnecessarytruthisanalysableintermsoftheconceptofanalytictruth.Bothpremisesareproblematicsince(APK)doesnotinvolvetheconceptofnecessarytruth,andthereisnoavailableanalysisoftheconceptofnecessarytruthintermsoftheconceptofanalytictruth.Putnam(1983)proposesanalternativeconnectionbetweenQuine’sconten-tionsandtherejectionoftheapriori.HemaintainsthatQuine’scontentionsaredirectedtowardstwodifferenttargets.Hisinitialcontentionsaredirectedtowardsthesemanticconceptofsynonymy.Hislatercontentions,however,aredirectedtowardstheconceptofastatementthatisconfirmednomatterwhat,whichisnotasemanticconcept.Itisanepistemicconcept;itisaconceptofapriority.Kitcher(1983:80)endorsesPutnam’sreadingofQuine’sargument:“Ifwecanknowapriorithatpthennoexperiencecoulddepriveusofourwarranttobelievethatp.”Hence,thePutnam–KitcherversionofQuine’sargument,theUnrevisabilityArgument,canbestatedasfollows:(Q1)Nostatementisimmunetorevisioninlightofrecalcitrantexperience.(Q2)IfS’sbeliefthatpisjustifiedapriori,thenS’sbeliefthatpisnotrationallyrevisableinlightofanyexperientialevidence.(Q3)Therefore,noknowledgeisapriori.Theargumentfails.Premise(Q2)isopentotheobjectionpresentedagainst(AP3)onpages231–2.TheRelationshipBetweenaprioriKnowledgeandNecessaryTruthCurrentinterestintherelationshipbetweenaprioriknowledgeandnecessarytruthisduetoKripke(1971,1980),whomakestwostrikingepistemologicalclaims:(E1)Therearenecessaryaposterioritruths;and2399781472578235_txt_print.indd23908/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguage(E2)Therearecontingentaprioritruths.Kripkemaintainsthat(E1)isaconsequenceofoneofhisprimarymetaphysicaltheses:(MT)Identitystatementsinvolvingpropernamesarenecessarilytrueiftrue,andthat(E2)isaconsequenceofoneofhisprimarysemantictheses:(ST)Adefinitedescriptionthatisemployedtointroduceanamefixesthereferenceofthatnameratherthanprovidingitssense.Healsoacknowledgesthatitisawidelyheldview,onethatheassociateswithKant,that(K)Allknowledgeofnecessarytruthsisaprioriandallaprioriknowledgeisofnecessarytruths.Therefore,hearguesagainst(K)inordertodefuseapotentialobjectionto(MT)and(ST).InordertoassesshowKripke’sclaimsbearonKant’saccountoftherelationshipbetweenaprioriknowledgeandnecessarytruth,fourpreliminaryobservationsareinorder.First,(K)istheconjunctionoftwoprinciples:(K1)Allknowledgeofnecessarytruthsisapriori;and(K2)Allaprioriknowledgeisofnecessarytruths.Second,Kant’sconceptionofaprioriknowledge(KAP),(KAP)SknowsapriorithatpifandonlyifS’sjustificationforthebeliefthatpisindependentofallexperienceandtheotherconditionsonknowledgearesatisfied,doesnotunderwriteeither(K1)or(K2)sincenecessityisnotconstitutiveof(KAP).Third,Kant’scontentionthatnecessityisacriterionofaprioriknowledge,whereacriterionisasufficientconditionthatisnotconstitutiveoftheconceptofaprioriknowledge,underwrites(K1).Fourth,neitherKant’sconceptionofaprioriknowledgenorhiscontentionthatnecessityisacriterionoftheaprioriunderwrites(K2).(K2)playsnoroleintheframeworkfordiscussingtheapriorithatKantarticulatesinhisintroductiontotheCritique.2409781472578235_txt_print.indd24008/08/201415:46\nAnalyticity,Apriority,ModalityCasullo(2010)maintainsthat(K2)drawsitssupportfromadifferentsource,thetraditionalrationalistconceptionofaprioriknowledge:(RAP)SknowsapriorithatpjustincaseSintuitively“sees”(orapprehends)thatpisnecessarilytrueandtheotherconditionsonknowledgearesatisfied.Since“‘seeing’thatpisnecessarilytrue”entails“pisnecessarilytrue”,itfollowsfrom(RAP)thataprioriknowledgeisrestrictedtonecessarytruths.Therefore,only(E1)bearsonKant’saccountoftherelationshipbetweenaprioriknowledgeandnecessarytruth.Kripkeinitiallyprovidestwodifferentexamplesinsupportof(E1):(a)state-mentsinwhichanessentialpropertyisattributedtoaphysicalobject;and(b)identitystatementsinvolvingdifferentco-referentialpropernames.Helaterextendshisdiscussionofidentitystatementstoincludetheoreticalidentitystatements.Wewillfocuson(a)and(b).Let“a”bethenameofaparticularlecternand“F”bethepropertyofbeingmadeofwood.SupposethatsomeoneknowsthatFa–i.e.,thatthislecternismadeofwood.Suchknowledgeisaposteriorisinceoneknowsthatsomethingismadefromwoodasopposedto,say,waterfrozenfromtheriverThamesonthebasisofhowitlooksandfeels.Yet,ifFaistrue,itisnecessarilytruesinceFisanessentialpropertyofa.Inanypossibleworldinwhichaexists,aisF.Hence,onewhoknowsthatFahasaposterioriknowledgeofanecessarytruth.ToassesstheimplicationsofKripke’sexample,wemustkeepinmindthattheexpression“aposterioriknowledgeofanecessarytruth”isambiguoussinceitdoesnotdistinguishbetween(A)aposterioriknowledgeofthetruth-valueofanecessaryproposition,(B)aposterioriknowledgeofthegeneralmodalstatusofanecessaryproposition,and(C)aposterioriknowledgeofthespecificmodalstatusofanecessaryproposition.Kripke’scaseisanexampleofaposterioriknowledgeofthetruth-valueofFasinceonediscoversviaexperiencethatthelecternismadeofwood.Whataboutknowledgeofitsgeneralmodalstatus?HereKripke(1971:153)isexplicitinmaintainingthatweknowby“aprioriphilosophicalanalysis”thatifFaistrue,thenitisnecessarilytrue.Hence,Kripke’scaseisnotanexampleofaposterioriknowledgeofthegeneralmodalstatusofanecessaryproposition.Kripkemaintainsthatsuchknowledgeisapriori.Finally,onewhoknows(aposteriori)thatFaand(apriori)thatifFa,thennecessarilyFacaninfer,andtherebyknow,thatnecessarilyFa.KnowledgethatnecessarilyFaisknowledgeofthespecificmodalstatusofFa.SinceknowledgeofthespecificmodalstatusofFaisbased(inpart)onaposterioriknowledgeofitstruth,itisalsoaposteriori.ThesameobservationsapplytoKripke’sexampleofidentitystatementsinvolvingpropernames.Since,accordingtoKripke,ordinarypropernames,2419781472578235_txt_print.indd24108/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagesuchas“Hesperus”and“Phosphorus”,arerigiddesignators,eachpicksoutthesameobjectinallpossibleworldsinwhichitpicksoutanyobject.Therefore,ifbothpickoutthesameobjectintheactualworld,bothpickoutthesameobjectinallpossibleworldsinwhichtheypickoutanyobject.Hence,if“HesperusisPhosphorus”istrue,itisnecessarilytrue.Ontheotherhand,itwasanastronomicaldiscoverythatHesperusisPhosphorus.So,onceagain,Kripkehasprovidedanexampleofaposterioriknowledgeofthetruth-valueofanecessaryproposition.Moreover,he(1980:109)maintainsthatweknow“byaprioriphilosophicalanalysis”thatsuchidentitystatementsarenecessarilytrueiftrue.Hence,Kripke’scaseisnotanexampleofaposterioriknowledgeofthegeneralmodalstatusofanecessaryproposition.Finally,onewhoknows(aposteriori)thatHesperusisPhosphorusand(apriori)thatifHesperusisPhosphorus,thennecessarilyHesperusisPhosphoruscaninfer,andtherebyknow(aposteriori),thatnecessarilyHesperusisPhosphorus.Howdoes(E1)bearonKant’saccountoftherelationshipbetweenaprioriknowledgeandnecessarytruth?Thequestioncannotbeansweredstraightfor-wardlybecauseneitherKantnorKripkemakestheappropriatedistinctions.(K1)isambiguous.Therearetwowaysofreadingit:(K1A)Allknowledgeofthetruth-valueofnecessarypropositionsisapriori;and(K1B)Allknowledgeofthegeneralmodalstatusofnecessarypropositionsisapriori.AlthoughKantendorsesboth(K1A)and(K1B),theargumentheoffersinsupportof(K1)supportsonly(K1B).Kripke’sexamplesofnecessaryaposterioritruthsareexamplesofaposterioriknowledgeofthetruth-valueofanecessaryproposition.He,however,deniesthattheyareexamplesofaposte-rioriknowledgeofthegeneralmodalstatusofanecessaryproposition.Hence,Kripke’sclaimschallenge(K1A)butnot(K1B).BothKantandKripkecontendthatknowledgeofthegeneralmodalstatusofpropositionsispossible.YetKripke’sclaimthattherearenecessaryaposte-rioritruthspresentsasignificantchallengetothatcontention.Priortohisargumentstothecontrary,mostheldthefalsebeliefthatnecessaryaposterioritruths,suchasthatHesperusisPhosphorus,arecontingenttruths.Moreover,thereremainsastrongintuitionthatappearstosupportthatfalsebelief.Thissuggeststhatmodalintuitionsaresystematicallyunreliableandthattheyresultinwidespreaderrorregardingthegeneralmodalstatusofpropositions.Suchwidespreaderror,inturn,threatensmodalknowledge.Kripkerecognizesthechallengetohispositionandrespondstoitinamannerthatishospitabletomodalknowledge.Kripkemaintainsthat,giventhatHesperusisPhosphorus,thereisnopossibleworldinwhichHesperus2429781472578235_txt_print.indd24208/08/201415:46\nAnalyticity,Apriority,ModalityisnotPhosphorus.SoitisfalsethatitmightturnoutthatHesperusisnotPhosphorus.Yethe(1980:103)acknowledgesthat“thisseemsverystrangebecauseinadvance,weareinclinedtosay,theanswertothequestionwhetherHesperusisPhosphorusmighthaveturnedouteitherway.”Kripke(1980:103–4)attemptstoresolvethistensionasfollows:TheevidenceIhavebeforeIknowthatHesperusisPhosphorusisthatIseeacertainstaroracertainheavenlybodyintheeveningandcallit“Hesperus”,andinthemorningandcallit“Phosphorus”.Iknowthesethings.Therecertainlyisapossibleworldinwhichamanshouldhaveseenacertainstaratacertainpositionintheeveningandcalledit“Hesperus”andacertainstarinthemorningandcalledit“Phosphorus”;andshouldhaveconcluded–shouldhavefoundoutbyempiricalinvestigation–thathenamestwodifferentstars,ortwodifferentheavenlybodies.…Andsoit’struethatgiventheevidencethatsomeonehasantecedenttohisempiricalinvestigation,hecanbeplacedinasenseinexactlythesamesituation,thatisaqualitativelyidenticalepistemicsituation,andcalltwoheavenlybodies“Hesperus”and“Phosphorus”,withouttheirbeingidentical.Sointhatsensewecansaythatitmighthaveturnedouteitherway.NotthatitmighthaveturnedouteitherwayastoHesperus’sbeingPhosphorus.Thoughforallweknewinadvance,Hesperuswasn’tPhosphorus,thatcouldn’thaveturnedoutanyotherway,inasense.Kripke(1980:142)generalizeshisanswertothepuzzleasfollows:Anynecessarytruth,whetherapriorioraposteriori,couldnothaveturnedoutotherwise.Inthecaseofsomenecessaryaposterioritruths,however,wecansaythatunderappropriatequalitativelyidenticalevidentialsituations,anappropriatecorrespondingqualitativestatementmighthavebeenfalse.Sothereisasenseinwhichanecessaryaposterioritruthmighthaveturnedouttobefalse,butthatsensedoesnotentailthatitisnotanecessarytruth.Therearetwodifferentsensesinwhichpmightturnouttobefalseor,alternatively,twosensesinwhichitispossiblethatpisfalse.Thefirstismetaphysicalsinceitpertainstowhetherthereisapossibleworldinwhichpisfalse.Thesecondisepistemicsinceitpertainstowhetherthefalsehoodofp(or,moreprecisely,p*,wherep*istheappropriatequalitativeanaloguetop)iscompatiblewithone’squalitativeevidence.AccordingtoKripke,wherepisanecessaryaposterioritruthandonehasanintuitionthatpmightturnouttobefalse,onedoesnothaveanintuitionthatthefalsehoodofpismetaphysicallypossible.Instead,suchanintuition,whenproperlyunderstoodandaccuratelyreported,isanintuitionthatthefalsehoodofpisepistemicallypossible.In2439781472578235_txt_print.indd24308/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageotherwords,whereEisone’soriginalqualitativeevidenceforp,onehastheintuitionthatthefalsehoodofp*iscompatiblewithE*,wherep*isaqualitativestatementthatappropriatelycorrespondstopandE*isanevidentialsituationthatisappropriatelyqualitativelyidenticaltoE.Therefore,theintuitiondoesnotcallintoquestionthenecessarytruthofp.Kripke’saccountishospitabletomodalknowledgefortworeasons.First,whenmodalintuitionsareproperlyunderstoodandaccuratelyreported,themodalbeliefsthattheysupportaretrue.Modalerrorariseswhenoneconfusesepistemicpossibilitywithmetaphysicalpossibility.Second,modalerroristractableinthat(a)itissystematicandwidespreadonlyinthecaseofaposteriorinecessities,but(b)Kripke’saccountidentifiesthesourceoftheerror,whichenablesustoavoiditor,atleast,tocorrectit.Thesefeaturesofmodalintuitionalignitfavourablywithotherfalliblesourcesofknowledge,suchasperception.Two-dimensionalsemanticsoccupiesaprominentplaceinthecontem-porarydiscussionofmodalknowledge.Oneofitsvirtues,accordingtoitsproponents,isthatitprovidesaperspicuousaccountofthetwotypesofpossi-bilitydistinguishedbyKripke.DavidChalmers(2006:59)lucidlysummarizesthetwo-dimensionalapproachasfollows:Thecoreideaoftwo-dimensionalsemanticsisthattherearetwodifferentwaysinwhichtheextensionofanexpressiondependsonpossiblestatesoftheworld.First,theactualextensionofanexpressiondependsonthecharacteroftheactualworldinwhichanexpressionisuttered.Second,thecounterfactualextensionofanexpressiondependsonthecharacterofthecounterfactualworldinwhichtheexpressionisevaluated.Correspondingtothesetwosortsofdependence,expressionscorrespondinglyhavetwosortsofintensions,associatingpossiblestatesoftheworldwithextensionsindifferentways.Thetwosortsofintensions,accordingtoChalmers(2006:59),yieldtwodifferentwaysofthinkingaboutpossibilities:Inthefirstcase,onethinksofapossibilityasrepresentingawaytheactualworldmightturnouttobe:orasitissometimesput,oneconsidersapossibilityasactual.Inthesecondcase,oneacknowledgesthattheactualworldisfixed,andthinksofapossibilityasawaytheworldmighthavebeenbutisnot:orasitissometimesput,oneconsidersapossibilityascounterfactual.Thetwodifferentwaysofthinkingaboutpossibilitiescanresultindifferentextensionsbeingassignedtoanexpressionrelativetoapossibleworld.ConsideragainthepossibleworlddescribedbyKripke,inwhichsomeonesees2449781472578235_txt_print.indd24408/08/201415:46\nAnalyticity,Apriority,Modalityacertainstarinacertainpositionintheeveningskyandcallsit“Hesperus”,andalsoseesacertainstarinacertainpositioninthemorningskyandcallsit“Phosphorus”,butthetwostarsarenotidentical.Ifwethinkofthispossi-bilityascounterfactual,then“Hesperus”and“Phosphorus”bothpickoutVenusgiventhatbothpickoutVenusintheactualworld.Hence,inthatworldconsideredascounterfactual,HesperusisPhosphorus.Moreover,inanyworldconsideredascounterfactual,“Hesperus”and“Phosphorus”bothpickoutVenusgiventhatbothpickoutVenusintheactualworld.Sothereisnoworld,consideredascounterfactual,inwhichHesperusisnotPhosphorus.Ontheotherhand,ifwethinkofKripke’spossibilityasactual,then“Hesperus”and“Phosphorus”pickoutdifferentobjects.Hence,inthatworldconsideredasactual,HesperusisnotPhosphorus.So,thinkingofpossibilitiesascounter-factual,capturesthesenseinwhichitcouldnothaveturnedoutthatHesperusisnotPhosphorus;butthinkingofthemasactualcapturesthesenseinwhich,forallweknewinadvance,itmighthaveturnedoutthatHesperusisnotPhosphorus.SyntheticaprioriKnowledgeKant’smostenduringcontributiontothecontroversysurroundingaprioriknowledgeishisdefenceof(K4)Somepropositionsknownaprioriaresynthetic.Theliteratureontheapriorioverthepast150yearsisdominatedbythisissue.Inaddressingthatliterature,onequestionimmediatelyarises:Whyistheexistenceofsyntheticaprioriknowledgeepistemologicallysignificant?Kantregardsitassignificantbecauseitsetsthestageforhisprimarytheoreticalundertaking,whichistoanswerthequestion:Howissyntheticaprioriknowledgepossible?Kant’squestion,however,ispuzzlinginonerespect.Havingestablishedthatthereisaprioriknowledge,heisinapositiontoposethequestion:Howisaprioriknowledgepossible?Thefactthathedeemsitnecessarytodrawtheanalytic/syntheticdistinctionandtodefend(K4)indicatesthatKantdoesnotthinkthataprioriknowledgeingeneralisproblematic.Inparticular,heviewsanalyticaprioriknowledgeasunproblematic.Ifsyntheticaprioriknowledgeisepistemologicallyproblematicbutanalyticaprioriknowledgeisnot,thentheymustdifferinsomeway.What,accordingtoKant,isthedifference?Kantmaintainsthatknowledgeofanalyticpropositionsrequiresonlypossessionoftherelevantconceptsandtheprincipleofcontra-diction.Syntheticaprioriknowledge,however,requiresmore.Forexample,inordertoknowthat7+5=12,Kant(1965:53)maintains:“Wehavetogooutside2459781472578235_txt_print.indd24508/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagetheseconcepts,andcallintheaidoftheintuitionwhichcorrespondstooneofthem.”Syntheticaprioriknowledgeraisesspecialepistemologicalproblemsbecauseofitsallegedsourceinintuition.Thesignificanceof(K4)isrootedintheassumptionthatthesourceofsyntheticaprioriknowledgeisdifferentfromthesourceofanalyticaprioriknowledge.Kant,however,doesnotdefendthisassumption.Althoughhemaintainsthatknowledgeofanalyticpropositionsrequiresonlyknowledgeoftheprincipleofcontradictionandthecontentofconcepts,hedoesnotexplicitlyaddressthesourceofsuchknowledge.Sincehedoesnotexplicitlyaddressthesourceofanalyticaprioriknowledge,Kanthasnobasisforclaimingthatthesourceofsuchknowledgeisdifferentfromthesourceofsyntheticaprioriknowledge,letalonethatthelatterisepistemologicallymoreproblematicthantheformer.Consequently,theepistemologicalsignificanceof(K4)ispresup-posedratherthanestablished.Reactionsto(K4)fallintothreebroadcategories.Thoseinthefirstendorse(K4)buttakeissuewithsomeofKant’sexamples.Frege,forexample,agreesthatthetruthsofgeometryaresyntheticaprioributmaintainsthatthetruthsofarithmeticareanalytic.Thoseinthesecondreject(K4).Logicalempiricists,suchasAyer,arguethatallegedexamplesofsyntheticaprioritruthsareeitheranalyticoraposteriori.Thereactionsinthethirdcategory,whichdrawtheirinspirationfromQuine,denythecogencyoftheanalytic/syntheticdistinctionand,afortiori,thecogencyof(K4).Theepistemologicalimportofthesereactionsisminimal.Fregeendorses(K4),butcontendsthatthetruthsofarithmeticareanalytic.HisdefenceofthiscontentionrequiresamodificationofKant’sconceptionofanalytictruth.Frege(1974:4e)explicatestheconcept,withrespecttomathe-maticalpropositions,intermsoffeaturesoftheirproof:“If,incarryingoutthisprocess[offollowingtheproofofaproposition],wecomeonlyongenerallogicallawsandondefinitions,thenthetruthisananalyticone.”TheresultingconceptionofanalytictruthisbroaderthanKant’s.Itdoesnotrestrictsuchtruthstothoseinwhichthepredicateiscontainedinthesubject.Anymathe-maticaltruthwhoseproofconsistssolelyofgenerallogicallawsanddefinitionsqualifiesasanalytic.Armedwiththisbroaderconceptionofanalyticity,Frege’sprojectistodemonstrate(F1)Allarithmetictruthsareanalytic.Thisprojectfacesanumberofformidabletechnicalobstacles.Butwewillassumethattheycanbeovercomeinordertoassessitsepistemologicalconsequences.Asuccessfuldemonstrationof(F1)hasnosignificantepistemo-logicalconsequences.Ademonstrationthatallarithmetictruthscanbeproved2469781472578235_txt_print.indd24608/08/201415:46\nAnalyticity,Apriority,Modalityfromgenerallogicallawsanddefinitions,takenbyitself,tellsuslittleaboutknowledgeofthosetruthssinceitissilentwithrespecttotheissueofhowoneknowstheprimitivegenerallaws,definitions,andlogicalprinciplesemployedinsuchproofs.Inparticular,(F1)iscompatiblewiththeclaimthatthetruthsofarithmeticareknowableonlyviaintuition.Onemightsuggestthatalthough(F1)failstoestablishthatarithmeticknowledgeisnotgroundedinintuition,itdoeshaveasignificantconsequenceregardingsuchknowledge.(F1)establishesthatifknowledgeoflogicanddefinitionsdoesnothaveitssourceinintuitionthenknowledgeofarithmeticdoesnothaveitssourceinintuition.Thisresultissignificantsinceitestablishesthatthereisauniformexplanationofknowledgeoflogic,definitions,andarithmetic.Theclaimthat(F1)establishesthatthereisauniformexplanationofknowledgeoflogic,definitions,andarithmeticrestsonanunsubstantiatedassumption:theonlyroutetoarithmeticknowledgeisthroughprooffromgenerallogicallawsanddefinitions.Thisassumptionhasanunwelcomeconse-quence.Itentailsawide-rangingscepticismwithrespecttotheelementarytruthsofarithmetic.Iftheonlyroutetoarithmeticknowledgeisthroughprooffromgenerallogicallawsanddefinitionsthenveryfew,ifany,havesuchknowledge.Kanttookforgrantedthatmostliterateadultsknowapriorithat7+5=12,andsetouttoprovideanaccountofsuchknowledge.Ifmostliterateadultshavesuchknowledge,thentheremustbearoutetoitotherthanthetypeofproofenvisionedbyFrege.Therefore,Fregefailstoshowthatthereisauniformexplanationofthetypicalliterateadult’sknowledgeoflogic,definitions,andarithmetic.Therearetwopossibleexplanationsofthetypicalliterateadult’sknowledgeofarithmetic:eitheritssourceisthesameasthesourceofknowledgeoflogicanddefinitions,oritisdifferent.Ifitisthesame,thenFrege’sprogrammeforestablishing(F1)isunnecessary.Ifitisdifferent,thenwearestillfacedwiththeproblemofexplaininghowsuchknowledgeispossible.IfKantisrightaboutthescopeofaprioriarithmeticknowledge,thenFregefailstoprovideanexplanationofsuchknowledge.Ayer(1952:73)rejects(K4)onthegroundsthateitherwemust“acceptitasamysteriousinexplicablefact”thatthereissyntheticaprioriknowledge,orwemust“accepttheKantianexplanationwhich…onlypushesthemysteryastagefurtherback.”Instead,heendorses(LE)Allaprioriknowledgeisofanalytictruths,whichheregardsasepistemologicallysignificantbecauseitprovidesanexpla-nationofaprioriknowledgethatisfreeofthemysterythatplaguesKant’saccount.2479781472578235_txt_print.indd24708/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageAyer(1952:78)rejectsKant’saccountoftheanalytic/syntheticdistinction,andoffersanalternative:“apropositionisanalyticwhenitsvaliditydependssolelyonthedefinitionsofthesymbolsitcontains,andsyntheticwhenitsvalidityisdeterminedbythefactsofexperience.”Ayer’sconceptionofanalyticityisbroaderthanKant’ssinceitdoesnotrestrictanalyticproposi-tionstothosewhosepredicateiscontainedinthesubject.Anypropositionthatistrueinvirtueofthedefinitionsofthesymbolsitcontainsqualifiesasanalytic.Ayer’s(1952:78–9)mostexplicitdefenceof(LE)ispresentedinthecontextofdiscussinglogicaltruths:[T]heproposition“Eithersomeantsareparasiticornoneare”isananalyticproposition.Foroneneednotresorttoobservationtodiscoverthatthereeitherareorarenotantswhichareparasitic.Ifoneknowswhatisthefunctionofthewords“either”,“or”,and“not”,thenonecanseethatanypropositionoftheform“Eitherpistrueorpisnottrue”isvalid,indepen-dentlyofexperience.Hisargumentcanbestatedasfollows:(AJ1)Oneneednotresorttoobservationtodiscoverthatthereeitherareorarenotantswhichareparasitic.(AJ2)Therefore,theproposition“Eithersomeantsareparasiticornoneare”isananalyticproposition.(AJ1)isanepistemologicalpremise:itassertsthatthepropositioninquestioncanbeknownapriori.(AJ2),however,isasemanticconclusion:itassertsthatthepropositioninquestionisanalytic.Thevalidityoftheargumentdependsonthefollowingprinciple,whichlinkstheepistemicpremiseandthesemanticconclusion:(AJ3)Allpropositionsknowableaprioriareanalytic.Hence,Ayer’sdefenceof(LE)iscircular.Supposewegrant(AJ2).Does(AJ2)provideanexplanationofaprioriknowledgeoflogicaltruths?Ifwereturntothepreviouslycitedpassage,Ayeroffersthefollowingpremiseinsupportof(AJ1):(AJ4)Ifoneknowswhatisthefunctionofthewords“either”,“or”,and“not”,thenonecanseethatanypropositionoftheform“Eitherpistrueorpisnottrue”isvalid,independentlyofexperience.2489781472578235_txt_print.indd24808/08/201415:46\nAnalyticity,Apriority,ModalityAyerexplainsaprioriknowledgeoflogicaltruthsintermsofanabilityto“see”thattheyaretrueindependentlyofexperience.Thesenseof“see”invokedbyAyertoexplainknowledgeoflogicaltruthsisnottheliteralsense.Hisexpla-nationappealstoametaphoricalsenseof“see”thatisnotfurtherexplained.Therefore,Ayer’sexplanation,likeKant’sappealtointuition,“onlypushesthemysteryastagefurtherback”.Quine’srejectionofthecogencyoftheanalytic/syntheticdistinctionhasbeenwidelyviewedaschallengingtheexistenceofaprioriknowledge.Wehaveexaminedtwolinesofargumentinsupportofthatview(seepages239–40)andconcludedthatbothfail.Thereis,however,anotherargument,theExplanatoryArgument,whichdrawsitsinspirationfromQuineandchallengestheexistenceofaprioriknowledge:(E1)Atheoryofknowledgehastwogoals:(a)toarticulatethesourcesandextentofhumanknowledge;and(b)toexplainhowthosesourcesgeneratetheknowledgeinquestion.(E2)Therefore,ifatheoryofknowledgeendorsesacategoryofknowledgebutcannotexplainhowthatknowledgeispossible,thenthetheoryisunacceptable.(E3)Theonlyavailablenon-mysteriousexplanationofhowaprioriknowledgeispossibleinvolvestheanalytic/syntheticdistinction.(E4)ButQuinehasshownthatthedistinctionisincoherent.(E4)Therefore,atheoryofknowledgethatendorsestheaprioriisunacceptable.TheExplanatoryArgumentfocusesattentionontheexplanatoryrequirementsofanadequatetheoryofknowledge.Itscentralpremiseis(E3).Althoughwidelyendorsed,thereislittlesupportfor(E3).Aswesawearlier,however,neitherKant’snorFrege’snorAyer’sconceptionofanalyticityoffersmuchintermsofanexplanationofaprioriknowledge.Nevertheless,thechallengeposedbytheExplanatoryArgumentremains.Atheoryofknowledgeendorsingtheapriorimustoffersomeexplanationofhowsuchknowledgeispossible.Oncewerecognizethattheconceptofanalyticityofferslittleintermsofexplanatorypower,weareinapositiontorecognizethattheexplanatoryproblemgoesbeyondthecoherenceoftheanalytic/syntheticdistinction.Themoregeneralproblemthatmustbeaddressedishowaprioriknowledgeispossible.Kant’sintroductionoftheanalytic/syntheticdistinction,alongwiththeassumptionthatthesourceofanalyticaprioriknowledgeisdifferentfromand2499781472578235_txt_print.indd24908/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagemoreproblematicthanthesourceofsyntheticaprioriknowledge,islargelyresponsibleforthecontinuedcontroversysurroundingtheexistenceofaprioriknowledge.Byfocusingattentiononthequestionofhowsyntheticaprioriknowledgecanbepossible,Kantinitiatedatraditionwhoseprimarypreoccu-pationwaswiththesemanticquestionsofwhetherthereisacogentanalytic/syntheticdistinction,howtodrawthatdistinction,andwhichpropositionsfallintoeachcategory.Providinganswerstothesesemanticquestions,however,willnotsettlethecontroversyovertheexistenceofaprioriknowledge.Thequestionthatmustbeaddressedtoresolvethatcontroversyisthemoregeneralepistemologicalquestion:Howisaprioriknowledgepossible?ReferencesAyer,A.J.(1952),Language,TruthandLogic,NewYork:DoverPublications.BonJour,Laurence(1985),TheStructureofEmpiricalKnowledge,CambridgeMA:HarvardUniversityPress.—(1998),InDefenseofPureReason,Cambridge:CambridgeUniversityPress.Casullo,Albert(2003),APrioriJustification,NewYork:OxfordUniversityPress.—(2005),“EpistemicOverdeterminationandAPrioriJustification”,PhilosophicalPerspectives19:41–58.—(2010),“KnowledgeandModality”,Synthese172:341–59.Chalmers,David(2006),“TheFoundationsofTwo-DimensionalSemantics”,inM.Garcia-CarpinteroandJ.Macia(eds),Two-DimensionalSemantics,Oxford:OxfordUniversityPress.Chisholm,R.M.(1977),TheoryofKnowledge,2ndedn,EnglewoodCliffs,NJ:Prentice-Hall.Frege,Gottlob(1974),TheFoundationsofArithmetic,2ndedn,revised,trans.J.L.Austin,Evanston:NorthwesternUniversityPress.Hempel,Carl(1972),“OntheNatureofMathematicalTruth”,inR.C.Sleighed.NecessaryTruth,EnglewoodCliffs:Prentice-Hall.Kant,Immanuel(1965),CritiqueofPureReason,trans.NormanKempSmith,NewYork:StMartin’sPress.Kitcher,Philip(1983),TheNatureofMathematicalKnowledge,NewYork:OxfordUniversityPress.Kripke,Saul(1971),“IdentityandNecessity”,inM.K.Munitzed.IdentityandIndividuation,NewYork:NewYorkUniversityPress.—(1980),NamingandNecessity,CambridgeMA:HarvardUniversityPress.Mill,JohnStuart(1973),ASystemofLogic,J.M.Robsoned.Toronto:UniversityofTorontoPress.Plantinga,Alvin(1993),WarrantandProperFunction,NewYork:OxfordUniversityPress.Putnam,Hilary(1983),“‘TwoDogmas’Revisited”,inRealismandReason:PhilosophicalPapers,Vol.3.Cambridge:CambridgeUniversityPress.Quine,W.V.(1963),“TwoDogmasofEmpiricism”,inFromaLogicalPointofView,2ndedn,revised,NewYork:HarperandRow.2509781472578235_txt_print.indd25008/08/201415:46\nNewDirectionsinthe11PhilosophyofLanguage1MaxKölbelMuchrecentworkinthephilosophyoflanguagehasbeenconcernedinonewayoranotherwithquestionsconcerningtheinteractionbetweenthestandingmeaningofexpressionsandthecontextinwhichtheyareused.Thereareatleasttwodifferenttypesofsourceforthisrenewedinterestincontextdependence.Ontheonehand,therearelong-standingcontroversiesconcerningtherighttreatmentofcontextandaboutthescopeandviabilityoftraditionalsemantics(seeChapter7,ContextDependence,inthisvolume).Ontheotherhand,thereareanumberofphilosophicalcontroversiesoutsidenaturallanguagesemanticsinwhichappealtosomekindofcontextdependencehasplayedarole,andwhichdriveaninterestincorrespondingsemanticissuesconcerningcontextdependence.Alltheseareasseemtoconvergeonacertaincomplexofproblems,andtheseinturnpointtowardsfoundationalandmethodologicaluncertaintyinsemanticsandphilosophyoflanguage.Independently,method-ologicalreflectionisreceivingimpetusfromageneralphilosophicaltrendtoreflectonmethod(“Metaphilosophy”),andthisisleadingtoarenewedinterestinfoundationaldebatesabouttheempiricalstatusofsemanticsandthecorrecttreatmentofcontextdependence.Ishallbrieflyadumbratethekindsofquestionsthatareinplay,andhowtheyinteract.Ishallthenmoveontodiscussaseriesoffurther,relatedtopics,whichseemrecentlytoattracttheinterestofresearchers.DoubleIndexSemanticsStandardtreatmentsofcontext-dependentexpressionssuchas“I”,“she”,“tomorrow”and“here”followthemodelofadoubleindexsemantics(Kamp1971;Kaplan1977;formoredetails,seeChapter7,pages159–66inthisvolume).Onthismodel,thesameexpressioncanexpressdifferentcontentsindifferentcontextsofuse–whichcontentthisisisdeterminedbytheexpression’scharacter,i.e.(partof)itsstablelinguisticmeaning.Thus,ifthesentence“Iamhungrynow”isusedinacertaincontext,itwillexpressabouttheuttererofthat2519781472578235_txt_print.indd25108/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagecontextandthetimeofthatcontext,thattheformerishungryatthelatter.Inotherwords,thesemanticpropertiesofsentencesaretreatedascharacters,i.e.functionsfromcontextsofusetosemanticcontents,wheresemanticcontentsinturnare(oratleastdetermine)functionsfromcircumstancesofevaluationtotruth-values.(InKaplan1977,contextsareorderedsequencesconsistingofanagent,aplace,atimeandaworld,suchthatatthetime,theagentisattheplaceattheworld.Circumstancesareorderedsequencesconsistingofaworld,atimeandpossiblyfurtherparameterssuchasalocation.)Correspondingly,subsententialexpressionshavecharactersthatensurethatsentenceswillhaveappropriatecharactersandexpresstheappropriatecontentsincontext.Adoubleindexsemanticsofthiskindcanbeseenaspredictingthetruth-valuesofactualandpotentialutterancesinthefollowingoversimplifiedway.Supposeanactofutteringasentencesoccursatacontextc.Thesemanticswillassigntosacharacter,i.e.afunctionfromcontexttocontent.Theutteranceisthentreatedasexpressingthecontentthatisthevalueofs’scharacterforcasargument.Thiscontentofsatccannowbeevaluatedwithrespecttothecircumstanceofevaluationdeterminedbyc.Thesemanticspredictstheutterancetohavethattruth-value,i.e.thevalueofthecontentofsatcforthecircumstanceofcasargument(seeKaplan1977:522).OnthissimplifiedunderstandingoftheKaplanianframework,wheneveroneutteranceofasentenceistrueandanotherutteranceofthesamesentenceisfalse,thenthetwoutterancesmusthaveoccurredatdifferentcontextsofuse.Thedivergenceoftruth-valuemusthaveariseninoneoftwoways:eitherthesentenceexpressestwodifferentcontentsatthedifferentcontexts,or,whileitexpressesthesamecontentinthedifferentcontexts,thiscontentreceivesdifferentvalueswithrespecttothedifferentcircumstancesofevaluationcorrespondingtothesecontexts.Thus,togivenamestothetwopossibilities,anysentencethatcanbeutterednowtrulyandnownot,mustexhibiteithercontext-sensitivityorcircumstance-sensitivity(seeChapter7,andseeRecanati,2007:34).Forexample,ifwehaveanearlierandalaterutteranceof“Obamaisasleep”,onetrue,onefalse,thenthiscouldbeduetotherespectivecontextsdeterminingdifferentsemanticcontents,oneaboutthetimeatwhichtheearlierutteranceoccurred,andoneaboutthetimeatwhichthelaterutteranceoccurred(inthecaseoftensedsentences,King2003,e.g.,followsthismodel).Thiswouldbeacaseofcontext-sensitivity.Alternatively,itmightbeacaseofcircumstance-sensitivity:whilethesemanticcontentexpressedby“Obamaisasleep”inthetwoutterancesisthesame,therelevantcircumstanceofevalu-ationisdifferentinthetwocases:thefirstutterance’struth-valueisthevalueofthissharedcontentwithrespecttothetimeofthefirstutterance,whilethesecondutterance’struth-valueisthevalueofthissharedcontentwithrespecttothetimeofthesecondutterance(inthecaseoftensedsentences,Kaplan1977andRecanati2007,e.g.,followthismodel).2529781472578235_txt_print.indd25208/08/201415:46\nNewDirectionsinthePhilosophyofLanguageGrice’sBuffer:WhatisSaidandImplicaturesThisviewofthepredictionsgeneratedbyasemanticsinaKaplanianframeworkis,however,questionable,forthereareseveraldifferentwaysinwhichagivenutterancecanbeevaluatedfortruth,anddifferentsuchwayscanresultindifferentevaluations:forexampleonecanevaluatetheutteranceinaliteralorinanon-literalway,onecanevaluateamessagedirectlyexpressedoronethatisindirectlyconveyed.TakeoneoftheexamplesfromChapter7:anutteranceof“Heisagenius”couldexpressaliteralmessagethatisfalse,yetanironicalmessagethatistrue.Whichoftheseistoserveasadatumtobepredictedbyasemantictheory?Again,ifthesentenceisutteredinresponsetothequestion“Canhesolvethisproblem?”,thenitmaywellconveythecontentthatthereferentof“he”cansolvetheproblem.Accordingly,theutterancemaybejudgedtobetrueifheisindeedabletosolvetheproblem–evenifheisnotagenius.Wecouldevenencounterbothphenomenaatonce:anironicaluseof“heisagenius”inresponsetothequestionwhetherhecansolvetheproblem.Itseemsclearthatourdescriptionofthesemanticpropertiesoftheexpres-sionsinvolvedinthesentence“Heisagenius”shouldnotneedtomakespecialprovisionforeachandeverymessagethatthissentencecouldbeusedtoconvey.Thiswouldmakethetaskofsemanticsimpossible.Itseemsthatweshouldarticulatemorecarefullywhichdatawetakesemantictheoriestopredict.Wemightinsistthatthereisagoodsenseinwhichalltheseutterances,whetherironicalorindirect,dependfortheirtruth-valueonlyonwhetherthereferentof“he”isagenius.Whattheystrictlyandliterallysayisthesame.Itseemsamorereasonableandfeasibletaskforacompositionalsemantictheorytopredictthetruth-valueofwhatutterancessaystrictlyandliterally.ItwasGrice(1989:especiallychs.2and15)whopioneeredandexploredthepotentialofthisthoughtandtriedtomakeitfruitfulforsemantics.Heintro-ducedthetechnicalnotionofwhatissaidbyanutteranceandcontrasteditwithwhatismerelyimplicated.Hetriedtoprovideageneraltheoryofconversationthatwouldexplainhowlanguageusersconvey(“implicate”)informationoverandabovewhattheirutterances“say”.AccordingtoGrice,languageexchangesaregovernedbyanumberofcooperativeprinciplesor“conver-sationalmaxims”thatparticipantswillbeexpectedtoadhereto.Thegeneralmaximrequiresparticipantsinaconversationtomaketheircontributions“suchasisrequired,atthestageatwhichitoccurs,bytheacceptedpurpose…ofthetalkexchangeinwhich[they]areengaged”(Grice1989:26).Whatanutteranceconveys,orinGrice’swords,whatit“conversationallyimplicates”,canbeworkedoutorinferredbyappealtothesemaxims.Anotherexample:ifIsay“MywatchlookslikeaRolex”,youmayreasonablyconcludethatmywatchisnotinfactaRolex.Grice’stheoryallowsustoofferasemanticsofthissentenceaccordingtowhichthecontentitexpressesiscompatiblewiththe2539781472578235_txt_print.indd25308/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguagewatchinfactbeingaRolex,andtoexplainthereasonablenessoftheinferenceintermsoftheaudience’sassumptionthatIobserveGrice’scooperativeprinciple(seeChapter8,Pragmatics,inthisvolumeformoredetails).Grice’stheoryofconversationalimplicatureswasoriginallymeanttoshieldtraditionalcompositionalsemanticsfromthedifficultythat,giventherightcircumstances,almostanysentencecanbeusedtogetacrossalmostanymessage.Postulatinganintermediatesemanticcontent(“whatissaid”)tobepredictedbysemantics,fromwhichtheultimatemessageconveyedcanbepragmaticallyinferred,allowsustomaintainthesystematicityandcompo-sitionalityofsemantics.However,thereistodayconsiderablecontroversyaboutthebestimplementationofGrice’sideasconcerningwhatutterancessay.Griceoriginallycharacterizedhisnotionbysaying(i)thatwhatissaidiscloselyrelatedtotheconventionalmeaningoftheexpressionsused,and(ii)thatwhatisimplicatedbyanutterancecanbeworkedoutfromwhathasbeensaid,ontheassumptionthatthecooperativeprincipleisbeingadheredto.Nowadaysmanytheoristsdoubtthatsemanticcontentscanmeetbothrequire-ments.Some,suchasRecanati(2001,2004),haveinsistedthatwhatissaidbyanutterancemustbeapropositionthatis“available”toconsciousthought,becauseitconstitutestheinputtoa“process”ofderivingconversationalimpli-catures.Asaconsequence,thesetheoristsargue,whatissaid(thesemanticcontent)cannotbepredictedbyastandardsemantictheorysuchasKaplan’s.Inotherwords,theconventionalsemanticpropertiesofthewordsusedinanutteranceareoftennotsufficienttodetermine,togetherwiththeKaplaniancontext,whathasbeensaidbythatutterance.Others,suchasBach(2001;Chapter7inthisvolume),havemaintainedthatsemanticcontents(whatissaid)arepredictedbysemantictheories,butconcludethatthesecontents(whatissaid)areoftensubpropositionalor“semanticallyincomplete”.YetotherssticktoGrice’soriginalpositionaccordingtowhichtheconventionalmeaningoftheexpressionsusedinanutterance,asdescribedbythesemantics,issuffi-cienttopredictthesemanticcontent(whatissaid)oftheutterancegivenitsKaplaniancontext.Inordertomaintainthis,some(Borg2004;CappelenandLepore2004)claimthattherearesuchthingsas“minimalcontents”,whileothers(Stanley2000andStanleyandSzabó2000)claimthatthereismoretothesentencesusedintheutterancesinquestionthantheirsuperficialphoneticappearancesuggests.Withoutthesimplifyingassumptionthatsemantictheoriesaredirectlypredictiveofourintuitionsastothetruth-valuesofutterancesofsentences,therearesomedeepmethodologicaldisagreementsamongstsemanticists.Forthereisnoconsensusastowhatexactly“semanticcontents”are,whethertheyarepredictablemerelyonthebasisofsemanticsandcontext,andhowwecantracksemanticcontentsonthebasisofempiricalorquasi-empiricaldata.2549781472578235_txt_print.indd25408/08/201415:46\nNewDirectionsinthePhilosophyofLanguageUnarticulatedConstituentsandCompositionalityDisagreementsabouttheappropriatenotionofsemanticcontentorwhatissaidaredirectlylinkedtoadebateaboutthequestionofwhetherthereareso-called“unarticulatedconstituents”(anotionoriginatingwithPerry1986).Thus,consideranutteranceofthesentence“Itisraining”.Twosimultaneousutterancesofthissentencecanconcerndifferentplaces,AandB,andthus(atleastapparently)differintruth-value.Towhatdoweattributethedifference?Itdoesnotseemtoopromisingtosaythatinfactthetwoutteranceshavethesametruth-valueandthattheimpressionofadifferenceiscreatedonlybyaconver-sationalimplicature.Amorepromisingoptionistoclaimthattheutterancesexpressdifferentsemanticcontents:oneapropositionaboutAandtheotherapropositionaboutB.Now,thedefendersofunarticulatedconstituents(e.g.Carston2002;Recanati2004)willclaimthatnoelementofthesentenceusedhasthislocationasitssemanticvalue.Thisconstituentofthesemanticcontentisnot“articulated”explicitlybyanysyntacticelementoftheexpressiontypeused.Thisviewpitsthemagainsttwodifferenttypesofopponents.Onetypeofopponentofunarticulatedconstituents(StanleyandSzabó2000;Stanley2000,2007)willagreethatthesemanticcontentsexpressedhavealocationasaconstituent.Buttheyclaimthatthesentenceuttereddoes,atadeepsyntacticlevel(“logicalform”),containasyntacticelementthathasthelocationasitssemanticcontent.Inthisway,thelocationis,afterall,articulated.Onthisview,thesentence“Itisraining”exhibitsstandardcontext-sensitivitywithrespecttotheplacerelevantforeachutterance.Theothertypeofopponentwoulddenythatthecontentexpressedhasalocationasaconstituent.Yetanotherposition,whichplayshardlyanyroleinthisparticulardebate,wouldmaintainthatthesemanticcontentofthetwoutterancesisthesame,butthatitisacontentthatiscircumstance-sensitive,sothatthetwodifferentutterancesareevaluateddifferentlybecausethecontenttheyshareisevaluateddifferentlyatthedifferentcircumstancesofevaluationofeachutterance(cf.Predelli2005).Thecontextualists’viewthatthereareunarticulatedconstituents,i.e.thatsomeconstituentsofsemanticcontents(whatissaid)donotcorrespondtoanyconstituentoftheexpressiontypeused,hasledthemtodeny(inlinewithotherdeniersofcompositionalitysuchasLahav1989;Travis1997)thattherecanbeacompositionalsemantics,i.e.atheorythatallowsderivationofthemeaningsofcomplexexpressions(incontext)fromthemeanings(incontext)oftheirconstituentparts(Carston2002;Recanati2004;forsomecriticismseePaginandPelletier2007).Morerecentlyoneofthecontextualists,Recanati,hasexplainedthathissemanticsstillrespectscertaincompositionalityconstraints,suchasforexampletheconstraintthatthesemanticcontentofanutteranceisdeterminedbythemodulatedmeaningsofthesyntacticpartsoftheutterance(Recanati2010a:44).Anemergingminimalconsensusaboutcompositionality2559781472578235_txt_print.indd25508/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageseemstobethat(i)“compositionality”canbedefinedinmanyinterestingyetsignificantlydifferentwaysandthat(ii)meetingsomecompositionalityconstraintsisinthebackgroundofanyexplanationoflinguisticcreativity(seeSzabó2007,andPaginandWesterståhl2010aand2010b,foroverviewsandreferencestotheliterature).Tosummarize,then,thereisconsiderabledebateaboutthenatureofsemanticcontentsandthereforeaboutthekindofpredictionsasemantictheoryforanaturallanguageshouldbeexpectedtoprovide.AccordingtotheoristssuchasStanleyandSzabó,thesemanticsforalanguageLprovidesinformationaboutthesemanticpropertiesofexpressionsofLwhichsufficestopredictforeachutteranceofasentenceofLwhatitssemanticcontentis,andthiscontentiswhathasbeensaidbytheutteranceandrepresentsthestartingpointforderivinganyconversationalimplicatures.Inthosecaseswhereitseemsthatthesamesentenceisusedtoexpressdifferentcontentsindifferentutteranceswithoutrelevantvariationinthecontext,thesetheoristspostulateextradeepsyntaxandclaimthatstrictlythesentencesutteredwereindexicalinsuchawaythatthevariationincontextbroughtaboutadifferenceinsemanticcontent.Thepreferredstrategyofanothergroup(minimalists,Borg,CappelenandLepore)istosaythat,justasthesentenceisthesame,thesemanticcontentexpressedisalsothesame.Theappearanceofadifferenceinsemanticcontentisduetoadifferenceinspeech-actcontent,andspeech-actcontentislargelyindependentofsemanticcontent.Bachagreesthatthesentenceusedhasnohiddenstructureanddeterminesthesamesemanticcontentinbothcases,butaccordingtohimthiscontentisnotfullypropositional(notsemanticallycomplete),for,onhisview,propositionsareevaluableintermsoftruth(whichthesestablecontentsaren’t),andsemanticallycompletecontentsarepropositions.Contextualists(Carston,Recanati),again,denyanyextrahiddensyntacticstructure,buttheyinsistthatsemanticcontentsarefullypropositionalandgeneratedby“modulation”or“freeenrichment”fromsentenceswhosesemanticpropertiesbythemselvesunderdeterminesemanticcontent.Afurtherposition(Predelli2005)triestoavoidalltheproblemsraisedbyvariationsofsemanticvaluefromoneutterancetoanotherbypointingoutthataKaplaniansemanticscanbesuccessfulevenifwehavenotbeenabletosayexactlywhichsentence-contextpairsarerepresentativeof,ormodel,anygivenutterance.ContextualismandRelativismInsomeareasoutsideofnaturallanguagesemanticsinastrictsense,questionsofcontextdependenceandcorrectassignmentofsemanticcontent,haverecentlybecomeprominent.Inepistemology,theterm“contextualism”hasadifferentsenseanddenotesamuchdebatedview.Epistemologicalcontextualistsclaim2569781472578235_txt_print.indd25608/08/201415:46\nNewDirectionsinthePhilosophyofLanguagethatwhetherapersoncanbetrulysaidtoknowsomepropositionwilldependoncertaincontextualfactors.Thatis,theverysamepersonwiththesamebeliefs,evidence,informationetc.canbetrulysaidtoknowinonebutnotinanothercontextofutterance.Forexample,accordingtothecontextualist,itmightbetrueinmostordinarycontextstosayofPeterthatheknowsthathehashands.Butinacontextinwhichasceptichasraisedcertainpossibilitiesoferrortosalience,it’snottruetosaythatheknows.Theideaisthatwhetherathinkerknowssomepropositionisrelativeto,say,standardsofknowledge.Interestingfromasemanticpointofviewisinthisconnectionwhethertheepistemologicalcontextualistwantstoclaimthattheverb“toknow”istobetreatedlikeanindexical,i.e.asexpressingadifferentrelationbetweenknowingsubjectandknownpropositionindifferentcontextsofuse,orwhetheritistobetreatedasexpressingthesamerelationeachtime,however,arelationthatholdsinrelationtosomestandards,butnotinrelationtoothers.(See,e.g.,Cohen1986;DeRose1992;Lewis1996;aswellasHawthorne2006;Stanley2007;MacFarlane2005a).Thisdebateaboutthepurportedcontextdependenceofknowledgeclaimsismotivatedbyspecificallyepistemologicalconcerns,namelytheproblemofknowledge-underminingscepticalarguments.However,structurallysimilardebateshavearisenindifferentareas,fordifferentphilosophicalreasons.Thus,metaphysicalandepistemologicalworriesofadifferentkindhaveledtotheviewthatevaluativeclaimsorjudgmentsofvarioussortscanbeevaluatedastrueorfalseonlyinrelationtocertainparameters,suchasastandardoraperspective(see,e.g.,Kölbel2003;Lasersohn2005,2008;Stephenson2007;Stojanovic2007;Francén2007;MacFarlaneforthcominga;compareChapter7,Section3inthisvolume).Similarly,inmetaphysicaldiscussionsaboutthenatureofcausation,onepositionhasbeentosaythatwhetheroneeventcausesanotherdependsonapurposeorcontrast(see,e.g.,Hitchcock1996;Woodward2003;Schaffer2005,2011).Again,whethersomethingispossibleintheepistemicsenseorwhetheritisprobable,isfrequentlyclaimedtodependonastateofinformation(e.g.Price1983;Eganetal.2005;MacFarlaneforthcomingb;Yalcin2007;Egan2007,2011).Ineachofthesecasesasemanticdebatecanbehadconcerningtheprecisenatureofthesedependencies(seeKölbel2008foramoredetailedoverviewandsomemorereferencestotheliterature).Thebasicdatathetheoristsaretryingtoaccountforareverysimilartotheonesgivingrisetodebates,discussedabove,abouttherightconstrualofsemanticcontent:wehavedifferentutterancesofthesamesentence,whichseemtoreceivedifferentevaluations.Butthesentencedoesnotseemtobeindexicalonthesurface.Here,aspreviously,theoptionsare:(i)denyadifferenceintruth-valueandexplaintheappearanceofsuchadifferenceinotherways;(ii)concedeadifferenceintruth-valuewhiledenyingadifference2579781472578235_txt_print.indd25708/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageinsemanticcontent,byconstruingcontentsasreceivingtruth-valuesinrelationtoanextrafactor;or(iii)postulateimplicitsyntaxwithindexicalelementsandthusconcede(andexplain)adifferenceinsemanticcontent.Oncertainassumptions,thisseeminglytechnicalissueinsemanticscanturnintoaprofoundmetaphysicaldivide.Supposewesaythatrealityismadeupofallthetruepropositions,i.e.allthefacts.Thenanti-contextualistsincategory(i)andcontextualistsincategory(iii)canmaintainthatrealityisobjective,thateveryonefacesthesamefacts.However,thesecondoftheseviewswillbecommittedtorealitydependingonstandardsforknowledge(standardsoftaste,explanatorycontrasts,statesofinformation,etc.);i.e.thatinsomeareasofthoughtandspeech,thecontentsofthoughtandspeechdonothaveabsolutetruth-valuesandarenotmadetrueorfalsebyobjectivefacts(seeEinheuser2008;Wright2008).Reflectiononso-called“relativist”approachesalongthelinesof(ii)haveledsometheoriststothinkaboutwhatitmightbetoassertasemanticcontentorpropositionthatvariesinitstruth-valuenotjustwithapossibleworldbutalsowithastandardforknowledge,astateofinformation,astandardoftaste,acontrast,andsoon.Inparticular,thequestionhasbeenraisedwhetherapracticeofassertingpropositionsmakessenseifthereisnosingleworld,time,standard,stateetc.withrespecttowhichassertionswillbeevaluatedastrueorfalse(e.g.Kölbel2002;MacFarlane2005b;Recanati2007;seealsoEvans1985).MacFarlane(2003,2008)hasarguedthat,inthecaseofutterancesconcerningthecontingentfuture,thesemanticcontentofanutterancecannotbeconstruedashavinganabsolutetruth-value,roughlybecausethiswouldconflictwiththemetaphysicalviewthatthefutureisgenuinelyopen.Whatnorms,then,governthepracticeofassertingsuchcontents?Thishasalsoledtorenewedinterestinwhatitistoassert,orcommunicatewith,propositionsconstruedassetsof“centredworlds”withinaStalnakerianmodelofconversation(e.g.Egan2007;Stalnaker2008:ch.3;Torre2010;Ninan2010;Einheuserforthcoming;Mossforthcoming;cf.Stalnaker1978,2002).Vaguenessandthesoritesparadox,afieldofresearchthathasreceivedenormousattentioninrecentyears,isalsorelevanttothesedebatesconcerningthecorrectconstrualofsemanticcontents.Contextualistsaboutvaguenessproposethatvague(i.e.sorites-prone)predicatesshouldbeconstruedascontext-dependentinaveryspecificway,onewhichexplainswhyitseemstousthatasinglegrainorhaircannotturnanon-heapintoaheap,orabaldmanintoonethatisn’tbald(Kamp1981;Raffman1996;Soames1999;Shapiro2006;Fara2000,2008).Somehaveconstruedthiscontext-dependenceasakindofindexicality(Soames1999),othershavestrenuouslydeniedthis(Raffman2005;cf.Stanley2003).Othershaveproposed“vaguepropositions”inanalogywithtensedorcontingentpropositions,i.e.approach(ii)above(e.g.Kölbel2011).Acloselyrelatedfieldofenquirywithinlinguisticsisthesemanticsofgradable2589781472578235_txt_print.indd25808/08/201415:46\nNewDirectionsinthePhilosophyofLanguageadjectives,wheretheexactnatureofthecontext-dependenceofsuchadjectiveisdiscussed(e.g.Szabó2001;Kennedy1999).MethodologicalandFoundationalUncertaintiesNeitherinphilosophyoflanguageandsemanticsofnaturallanguagesmorenarrowlyconstrued,norinotherfieldsofphilosophywithrelatedinterests,isthereaconsensusaboutabasicsemanticframeworkwithinwhichtoanchorclaimsaboutsemanticcontent.Itisthereforenotsurprisingthat,increasingly,theoristsbegintomakeexplicittheirmethodologicalassump-tionsaboutthetheoreticalaimofsemanticsandtheframeworkwithinwhichtoconductit.Many,probablymost,theoristsoperatewiththeassumptionthatsemantictheoriesfornaturallanguagesinsomesensemodelthecompe-tenceofspeakersofthatlanguage,andtheminimumconsensusaboutwhatcompetenceinvolvesseemstobeknowledgeoftheconditionsunderwhichasentence(incontext)wouldbetrue(seeChapter4,TheoriesofMeaningandTruthConditions).InBarbaraPartee’swords:“semanticcompetenceiswidelyconsideredtoconsistinknowledgeoftruth-conditionsandentailmentrelationsofsentencesofthelanguage”(Partee2011:section2).However,thereisconsiderablevariationastoanyadditionalaspectsofcompetencethatsemantictheoriesaretomodel.Some,moretraditionallyminded,semanticistsaffordafairlyminimalroletosemantics:allsemanticsdoesispredictorexplainliteraltruthorcorrectnessofutterancesofsentences,asjudgedbycompetentspeakersunderidealizedconditionsofomniscienceofthecircumstancesofutterance(e.g.Carnap1955;Lewis1980;Stanley2007:Introduction).Onthisview,therewouldbelittletochoosebetweensemantictheoriesthatpredictthesametruth-conditions,exceptperhapseleganceorexplanatoryconvenience.Othersgofurtherandclaimthatsemantictheoriesarequiteliterally“known”byamentalfaculty,thelanguagemodule(LarsonandSegal1995).Somegosofarastohypothesizethatsemantictheorieshave“psychologicalreality”andshouldbeviewedasdetailedmodelsofpsychologicalprocessesofinterpretinglanguage.Carston(2002)exploresthehypothesisthatsemantictheoriesmodelinterpretivebrainprocessesinsuchawaythattheypredictcomparativeprocessingspeeds.Recanati(2004)claimsthatcertainstagesofutteranceinterpretationare“available”toconsciousness.PaulPietroski(2011)takesthecognitiveconceptionofsemanticsmuchfurtherinthatheclaimsthatthetraditionaltruth-conditionalconceptionofsemanticshaslittlevalueinempiricalsemantics.Thereisafurthercontrastbetweenthosewhoemphasizeadependenceoncurrentlinguistictheory(Higginbotham1985;LarsonandSegal1995;HeimandKratzer1998;Pietroski2003;Stanley2007)andthosewhoenvisageamoreaprioristicapproachforsemantics(e.g.2599781472578235_txt_print.indd25908/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageKatz1985;Soames1985;Devitt2003).FormoredetailsseeChapter2,OntheNatureofLanguageinthisvolume.Asreportedabove(Chapter5,Reference),anewapproachtomethodhasalsoledsomephilosopherstoquestiontheso-calledcausaltheoryofreference,whichhadforalongtimeenjoyedthestatusoforthodoxy.Somephilosophershavesuggestedthatcertainnon-Westernpopulationsdonotsharetheintui-tionsonwhichKripke’sfamousargumentswerebased(seeMacheryetal.2004).Theyhavetriedtoestablishthisbyexperimental,i.e.questionnaire,methods.Moreover,theyhavefoundthatthereisahighdegreeofdivergenceevenwithinthesameculturalgroup(Macheryetal.2009).Thisnovelmethod-ologyisbynomeanswidelyaccepted,noristheinterpretationoftheresultsoftheseexperiments(seeChapter5inthisvolume,andalsoHansen&Chemlaforthcoming).However,thediscussionisanothermanifestationofanewreadinesstoprobemethodologicalassumptions.OtherDevelopmentsThisbringsmetoanotherrelativelynewdevelopment:independentlyofthechallengesposedbythenewexperimentalphilosophersjustmentioned,thecurrentofthosewhowanttoresistKripke’slandmarkargumentsagainst“descriptivism”hasgrown.Evenifthisisnotamassmovement,severalphiloso-phershaveconsistentlytakenupDummett’slineofresistancetoKripkeandhavedevelopedformsof“causaldescriptivism”,theviewthatallexpressions,includingpropernames,doexpresssomeminimaldescriptivecontent–evenifnotthetypeofdescriptivecontentthatKripkewasenvisaging.Thesuggestion,forexample,isthatthesemanticcontentofauseofanamesuchas“JohnSmith”canbeexpressedby“themanwhoisatthebeginningorsourceofthename-usingpracticetowhichthistokenof‘JohnSmith’belongs”(see,e.g.,Stanley1997;Noonan2001;García-Carpintero2006).InsomecasesthisdescriptivistviewispartofamoregeneraloutlookthattriestoreviveanddevelopFrege’sandCarnap’sviewsonanalyticityandapriority(e.g.Chalmers2002,2010).Similarly,opponentsofDirectReferencetheoriesofsingularthought(seeagainChapter5,above)havebeendeveloping“latitudinarian”viewsofsingularthought,asforexampleSosa1970,1995,andHawthorneandManley(forthcoming).Thisisnottosaythatthedominantviewdoesnotcontinuetobeonethatassignsaspecialnon-descriptiveroletoacertainclassofsingulartermsthatincludespropernamesanddemonstratives.Theseviewsarealsobeingdevelopedfurther,forexampleinSainsbury2005;Soames2005;Recanati2010b;Jeshion2010.Thesearesomeoftheareasinwhichphilosophersoflanguagehaverecentlybeenactive,andinwhichnodoubtwewillseefurtherdevelopmentsin2609781472578235_txt_print.indd26008/08/201415:46\nNewDirectionsinthePhilosophyofLanguagecomingyears.TherearealsomanyotherareasoflivelyresearchactivityinthephilosophyoflanguagethatIhavenotbeenabletosurveyhere,suchasdiscus-sionsofnon-truth-conditionalaspectsofmeaning(see,e.g.,DavidKaplan(ms)),thenatureofpresupposition(see,e.g.,Schlenker2009;vonFintel2008),conventionalimplicature(see,e.g.,Potts2007),Davidson’sprogramme(see,e.g.,LeporeandLudwig2005),anddynamicsemantics(see,e.g.,Beaver2001).Note1.Thetitleofthischaptersmacksofcrystalballgazing:pronouncingonwhichtopicsandviewsarecurrentlyfashionable,andwherethespotlightoffashionmightmovenext.ItistruethatIhavenotcarriedoutastatisticalanalysisofrecentresearchinphilosophyoflanguage,norhaveIusedanyothersystematicmethodsinarrivingatajudgmentofwhatmightbethedirectionsinwhichthephilosophyoflanguageiscurrentlymoving.Instead,whatIwillbedoingismakesomeobservationsonsomerecentdevelopmentsinthephilosophyoflanguage,developmentsthatseemtometoreflectcertaingeneraltrends.However,Iclaimneitherauthoritynorobjectivitynorexhaustivenessformyparticularselectionoffocus.IamgratefultoManuelGarcía-Carpinteroforhishelpwithpartsofthischapter.TheresearchleadingtotheseresultswassupportedbyMICINN,SpanishGovernment,I+D+iprogramme,grantFFI2009-13436andalsoCONSOLIDERINGENIOProgramme,grantCSD2009-0056,aswellastheEuropeanFP7programme,grantno.238128.ReferencesBach,Kent(2001),“YouDon’tSay?”,Synthese128,15–44.Beaver,David(2001),PresuppositionandAssertioninDynamicSemantics.Stanford:CSLI.Borg,Emma(2004),MinimalSemantics,Oxford:OxfordUniversityPress.Cappelen,HermanandErnieLepore(2004),InsensitiveSemantics,Oxford:Blackwell.Carnap,Rudolf(1955),“MeaningandSynonymyinNaturalLanguages”,PhilosophicalStudies6,33–47.Carston,Robyn(2002),ThoughtsandUtterances,Oxford:Blackwell.Chalmers,DavidJ.(2002),“OnSenseandIntension”,PhilosophicalPerspectives16,135–82.—(2010),“ConstructingtheWorld”,TheJohnLockeLectures,deliveredattheUniversityofOxfordinMay2010.Availableonlineathttp://www.philosophy.ox.ac.uk/lectures/john_locke_lectures/past_lectures/Cohen,Stewart(1986),“KnowledgeandContext”,TheJournalofPhilosophy83,574–83.DeRose,Keith(1992),“ContextualismandKnowledgeAttributions”,PhilosophyandPhenomenologicalResearch52,913–29.2619781472578235_txt_print.indd26108/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageDevitt,Michael(2003),“LinguisticsIsNotPsychology”,inA.Barbered.EpistemologyofLanguage,Oxford:OxfordUniversityPress,pp.107–39.Egan,Andy(2007),“EpistemicModals,RelativismandAssertion”,PhilosophicalStudies133,1–22.—(2011),“RelativismaboutEpistemicModals”,inStevenD.Halesed.ACompaniontoRelativism,Oxford:Blackwell.Egan,A.,J.HawthorneandB.Weatherson(2005),“EpistemicModalsinContext”,inG.PreyerandG.Peter(eds),ContextualisminPhilosophy,Oxford:OxfordUniversityPress,pp.131–70.Einheuser,Iris(2008),“ThreeFormsofTruth-Relativism”,inManuelGarcia-CarpinteroandMaxKölbel(eds),RelativeTruth,Oxford:OxfordUniversityPress.—(forthcoming),“RelativizedPropositionsandtheFregeanOrthodoxy”,PhilosophyandPhenomenologicalResearch.Evans,Gareth(1985),“DoesTenseLogicRestonaMistake?”,inhisCollectedPapers,Oxford:ClarendonPress,pp.341–63.Fara,DeliaGraff(2000),“ShiftingSands:AnInterest-RelativeTheoryofVagueness”,PhilosophicalTopics28,45–81.—(2008),“ProfilingInterestRelativity”,Analysis68,326–35.Fintel,Kaivon(2008),“WhatIsPresuppositionAccommodation,Again?”,PhilosophicalPerspectives22,137–70.Francén,Ragnar(2007),MetaethicalRelativism:AgainsttheSingleAnalysisAssumption.Dissertation,UniversityofGothenburg.García-Carpintero,Manuel(2006),"Two-DimensionalSemantics:ANeo-FregeanInterpretation”,inM.García-CarpinteroandJ.Macià(eds),Two-DimensionalSemantics,Oxford:OxfordUniversityPress,pp.181–204.García-Ramírez,EduardoandMarilynShatz(2011),“OnProblemswithDescriptivism:PsychologicalAssumptionsandEmpiricalEvidence”,Mind&Language,26,53–77.Grice,Paul(1989),StudiesintheWayofWords,Cambridge,MA:HarvardUniversityPress.Hansen,NathanielandEmmanuelChemla(forthcoming).“ExperimentingonContextualism”.Hawthorne,John(2006),KnowledgeandLotteries.Oxford:OxfordUniversityPress.Hawthorne,JohnandDavidManley(forthcoming),TheReferenceBook,Oxford:OxfordUniversityPress.Heim,IreneandAngelikaKratzer(1998),SemanticsinGenerativeGrammar,Oxford:Blackwell.Higginbotham,James(1985),“OnSemantics”,LinguisticInquiry16,547–93.Hitchcock,ChristopherRead(1996),“TheRoleofContrastinCausalandExplanatoryClaims”,Synthese107,395–419.Jeshion,Robin(2010),“SingularThought:AcquaintanceSemanticInstrumentalismandCognitivism”,inRobinJeshioned.NewEssaysonSingularThought,Oxford:OxfordUniversityPress,pp.105–40.Kamp,Hans(1971),“FormalPropertiesof‘Now’”,Theoria37,227–74.—(1981),“TheParadoxoftheHeap”,inU.Mönniched.AspectsofPhilosophicalLogic,Dordrecht:Reidel,pp.225–77.2629781472578235_txt_print.indd26208/08/201415:46\nNewDirectionsinthePhilosophyofLanguageKaplan,D.(1977),“OnDemonstratives”,inJ.Almog,J.PerryandH.Wettstein(eds),ThemesfromKaplan(1989),Oxford:OxfordUniversityPress,pp.481–563.Kaplan,David(nd),“TheMeaningof‘Oops’and‘Ouch’”.Unpublishedms.Katz,JerroldJ.(1985),“OutlineofaPlatonistGrammar”,inJ.Katzed.ThePhilosophyofLinguistics,Oxford:OxfordUniversityPress,pp.172–203.Kennedy,Christopher(1999),ProjectingtheAdjective:TheSyntaxandSemanticsofGradabilityandComparison,NewYork:GarlandPress.King,Jeffrey(2003),“Tense,Modality,andSemanticValues”,PhilosophicalPerspectives17,195–245.Kölbel,Max(2002),TruthWithoutObjectivity,London:Routledge.—(2003),“FaultlessDisagreement”,ProceedingsoftheAristotelianSociety104,53–73.—(2008),“MotivationsforRelativism”,inManuelGarcia-CarpinteroandMaxKölbel(eds),RelativeTruth,Oxford:OxfordUniversityPress,pp.1–38.—(2011),“VaguenessasSemantic”,inR.DietzandS.Moruzzi(eds),CutsandClouds:IssuesinthePhilosophyofVagueness,Oxford:OxfordUniversityPress,pp.304–26.Lahav,Ran(1989),“AgainstCompositionality:TheCaseofAdjectives”,PhilosophicalStudies57,261–79.Larson,RichardandGabrielSegal(1995),KnowledgeofMeaning,Cambridge,MA:MITPress.Lasersohn,Peter(2005),“ContextDependence,Disagreement,andPredicatesofPersonalTaste”,LinguisticsandPhilosophy28,643–86.—(2008),“QuantificationandPerspectiveinRelativistSemantics”,PhilosophicalPerspectives22,305–37.Lepore,ErnestandKirkLudwig(2005),DonaldDavidson.Meaning,Truth,LanguageandReality,Oxford:OxfordUniversityPress.Lewis,David(1980),Index,Context,andContent,inStigKangerandSvenÖhman(eds),PhilosophyandGrammar,Dordrecht:Reidel.ReprintedinDavidLewis,PapersinPhilosophicalLogic,Cambridge:CambridgeUniversityPress,1998.—(1996),“ElusiveKnowledge”,AustralasianJournalofPhilosophy74,549–67.—(1998),PapersinPhilosophicalLogic,Cambridge:CambridgeUniversityPress.MacFarlane,John(2003),“FutureContingentsandRelativeTruth”,PhilosophicalQuarterly53,321–36.—(2005a),“TheAssessmentSensitivityofKnowledgeAttributions”,inTamarSzaboGendlerandJohnHawthorne(eds),OxfordStudiesinEpistemology1,Oxford:OxfordUniversityPress,pp.197—233.—(2005b),“MakingSenseofRelativeTruth”,ProceedingsoftheAristotelianSociety105,321–39.—(2008),“TruthintheGardenofForkingPaths”,inMaxKölbelandManuelGarcía-Carpintero(eds),RelativeTruth,Oxford:OxfordUniversityPress.—(forthcominga),Assessmentsensitivity:relativetruthanditsapplications.—(forthcomingb),“EpistemicModalsareAssessment-Sensitive”,inBrianWeathersonandAndyEgan(eds),EpistemicModals,Oxford:OxfordUniversityPress.Machery,E.,ChristopherY.OlivolaandMollyDeBlanc(2009),“Linguisticandmetalinguisticintuitionsinthephilosophyoflanguage”,Analysis,69,689–94.2639781472578235_txt_print.indd26308/08/201415:46\nTheBloomsburyCompaniontothePhilosophyofLanguageMachery,E.,R.Mallon,S.NicholsandS.Stich(2004),“Semanticscross-culturalstyle”,Cognition92,B1–B12.Mallon,R.,E.Machery,S.NicholsandS.Stich(2009),“Againstargumentsfromreferente”,PhilosophyandPhenomenologicalResearch,79,332–56.Martí,Genoveva(2009),"AgainstSemanticMulti-Culturalism”,Analysis69,42–8.Moss,Sarah(forthcoming),“UpdatingasCommunication”,PhilosophyandPhenomenologicalResearch.Ninan,Dilip(2010),“DeSeAttitudes:AscriptionandCommunication”,PhilosophyCompass5,551–67.Noonan,Harold(2001),Frege:aCriticalIntroduction,Cambridge:Polity.Pagin,PeterandDågWesterståhl(2010a),“Compositionality:DefinitionsandVariants”,PhilosophyCompass5,250–64.—(2010b),“Compositionality:ArgumentsandProblems”,PhilosophyCompass5,265–82.Pagin,PeterandJeffPelletier(2007),“Content,ContextandComposition”,inG.PeterandG.Preyer(eds),ContentandContext.EssaysonSemanticsandPragmatics,Oxford:OxfordUniversityPress,pp.25–62.Partee,Barbara(2011),“FormalSemantics”,inPatrickColmHoganed.TheCambridgeEncyclopediaoftheLanguageSciences,Cambridge:CambridgeUniversityPress.Perry,J.(1986),“ThoughtwithoutRepresentation”,ProceedingsoftheAristotelianSociety,supplementaryvolume60,137–52.ReprintedinJ.Perry,TheEssentialIndexicalandOtherEssays,Oxford:OxfordUniversityPress,1993.Pietroski,Paul(2003),“TheCharacterofNaturalLanguageSemantics”,inAlexBarbered.TheEpistemologyofLanguage,Oxford:OxfordUniversityPress.—(2011),“MinimalSemanticInstructions”,inCedricBoeckxed.OxfordHandbookofLinguisticMinimalism,Oxford:OxfordUniversityPress.Potts,Christopher(2007),“ConventionalImplicature:ADistinguishedClassofMeanings”,inG.RamchandandCh.Reiss(eds),OxfordHandbookofLinguisticInterfaces,Oxford:OxfordUniversityPress,pp.475–501.Predelli,Stefano(2005),Context:Meaning,TruthandtheUseofLanguage,Oxford:OxfordUniversityPress.Price,Huw(1983),“Does‘probably’modifySense?”,AustralasianJournalofPhilosophy61,396–408.Quine,WillardVanOrman(1960),WordandObject.Cambridge,MA:MITPress.Raffman,Diana(1996),“VaguenessandContextRelativity”,PhilosophicalStudies81175–92.—(2005),“HowtoUnderstandContextualismaboutVagueness:ReplytoStanley”.Analysis65,244–8.Recanati,François(2001),“WhatisSaid”,Synthese128,75–91.—(2004),LiteralMeaning,Cambridge:CambridgeUniversityPress.—(2007),PerspectivalThought,Oxford:OxfordUniversityPress.—(2010a),Truth-conditionalPragmatics,Oxford:OxfordUniversityPress.—(2010b),“SingularThought:inDefenceofAcquaintance”,inRobinJeshioned.NewEssaysonSingularThought,Oxford:OxfordUniversityPress,pp.141–89.Sainsbury,Mark(2005),ReferencewithoutReferents,Oxford:OxfordUniversityPress.Schaffer,Jonathan(2005),“ContrastiveCausation”,PhilosophicalReview114,327–58.—(2011),“CausalContextualisms”,inMartijnBlaauwed.ContrastivisminPhilosophy,London:Routledge.2649781472578235_txt_print.indd26408/08/201415:46\nNewDirectionsinthePhilosophyofLanguageSchaffer,JonathanandJoshuaKnobe(forthcoming),“ContrastivismSurveyed”,Noûs.Schlenker,Philippe(2009),“LocalContexts”,Semantics&Pragmatics2,1–78.Shapiro,Stewart(2006),VaguenessinContext,Oxford:OxfordUniversityPress.Soames,Scott(1985),“SemanticsandPsychology”,inJ.Katzed.ThePhilosophyofLinguistics,Oxford:OxfordUniversityPress,pp.204–26.—(1999),UnderstandingTruth,Oxford:OxfordUniversityPress.—(2005),ReferenceandDescriptions:TheCaseagainstTwo-Dimensionalism.Princeton,NJ:PrincetonUniversityPress.Sosa,Ernest(1970),“PropositionalAttitudesDeDictoandDeRe”,JournalofPhilosophy67,883–96.—(1995),“FregeanReferenceDefended”,PhilosophicalIssues6,91–9.Stalnaker,Robert(1978),“Assertion”,inP.Coleed.SyntaxandSemantics,ix.Pragmatics,315–22.—(2002),“CommonGround”,LinguisticsandPhilosophy25,701–21.—(2008),OurKnowledgeoftheInternalWorld,Oxford:OxfordUniversityPress.Stanley,Jason(1997),“NamesandRigidDesignation”,inBobHaleandCrispinWright(eds),ACompaniontothePhilosophyofLanguage,Oxford:Blackwell,pp.555–85.—(2000),“ContextandLogicalForm”,Linguistics&Philosophy23:391–434.—(2003),“Context,InterestRelativityandtheSorites”,Analysis63,269–81.—(2006),KnowledgeandPracticalInterests,Oxford:OxfordUniversityPress.—(2007),LanguageinContext:SelectedEssays,Oxford:OxfordUniversityPress.Stanley,JasonandZoltanSzabó(2000),“OnQuantifierDomainRestriction”,MindandLanguage15,219–61.Stephenson,Tamina(2007),“Judgedependence,epistemicmodals,andpredicatesofpersonaltaste”,LinguisticsandPhilosophy30,487–525.Stojanovic,Isidora(2007),“TalkingaboutTaste:Disagreement,ImplicitArgumentsandRelativeTruth”,LinguisticsandPhilosophy30:691–706.Szabó,Zoltán(2001),“AdjectivesinContext”,inR.HarnishandI.Kenesei(eds),PerspectivesonSemantics,Pragmatics,andDiscourse,Amsterdam:JohnBenjamins,pp.119–46.—(2007),“Compositionality”,inE.N.Zaltaed.TheStanfordencyclopediaofphilosophy.Availableonlineathttp://plato.stanford.edu/entries/compositionality/—(2010),“TheDeterminationofContent”,PhilosophicalStudies148,253–72.Torre,Stephan(2010),“CenteredAssertion”,PhilosophicalStudies150,97–114.Travis,Charles(1997),“Pragmatics”,inBobHaleandCrispinWright(eds),ACompaniontothePhilosophyofLanguage,Oxford:Blackwell.Woodward,James(2003),MakingThingsHappen:ATheoryofCausalExplanation,Oxford:OxfordUniversityPress.Wright,Crispin(2008),“RelativismaboutTruthItself”,inManuelGarcía-CarpinteroandMaxKölbel(eds),RelativeTruth,Oxford:OxfordUniversityPress.Yalcin,Seth(2007),“EpistemicModals”,Mind116,983–1026.2659781472578235_txt_print.indd26508/08/201415:46\nA–ZofKeyTermsThepropercharacterizationofmanyofthefollowingnotionsiscontroversial.Hereweonlypurporttogivethereaderaninitialimpressionofwhattherelevantnotionisintendedtoencompass.apriorivs.aposteriori(orempirical)knowledge:Onehasaposterioriorempiricalknowledgeofone’sweightorthecolouroftheclothespeoplearoundonearewearing.Aprioriknowledgeissupposedtobeknowledgethatwehavewithoutthusrelyingonexperience,perceptualorevenintrospective.Usualcandidatesaretheknowledgethateveryobjectisidenticaltoitself,thattwoplusthreeequalsfive,orthatnosurfacecanbebothredalloverandgreenallover.analyticvs.synthetic:Analyticstatementsaresupposedtobestatementstrueinvirtueofthemeaningofsomeoftheconstituenttermsofthesentenceswithwhichtheyaremade,andknowntobesoonthebasisofunderstandingsuchterms.Purportedexamplesareallvixensarefemalefoxesorallbodiesareextended.Syntheticstatementsarenon-analyticones.bivalence:Theviewthatpropositionsareeithertrueorfalse,withnothirdalternative,evenpropositionsthatarecontentsofvaguestatements,statementswithemptynames,oraboutapossiblyindeterminatecontingentfuture.causal-historicalchainofcommunication:One’sabilitytousetermssuchas“GustaveFlaubert”or“bouillabaisse”or“palmtree”isacquired,throughaspecificcausalprocess,fromone’sfellowspeakers.Thesameapplies,inturn,totheabilityofthoseotherspeakers,andsoon.Inthiswaywecanideallytracedifferenthistoriesofusebacktoeventsinwhichthosetermswerefirstintroducedfortheiruseinlinguisticcommunication(“initialbaptisms”;eventssuchastheonesinwhich,say,iPhonesornewbornsorpetanimalsacquiretheirnames.Thisseriesofinterconnectedlanguageusersthatgoesbackfromagivenuseofatermtotheoriginalintroductionofthetermiscalleda“causal-historicalchain”.causaltheoryofreference:Theviewthatcausalrelations–suchascausal-historicalchainsofcommunication–determinewhatusesoftermsreferto;itcontrastswiththedescriptivistviewthatthereferenceisdeterminedbyidentifyingknowledge2669781472578235_txt_print.indd26608/08/201415:46\nA–ZofKeyTermsofthereferentsthatcompetentspeakershave,becausethereference-determiningcausalrelationsarenotsupposedtobeknownbycompetentspeakers.(subjunctive)conditional:Itisacommonemotiontoregretsomethingsthathappened,andtowishthattheyhadbeendifferent;wetypicallyexpressitbymeansofstatementssuchas“ifIhadrememberedtotakemypassport,Iwouldnothavemissedtheflight”.Thesearesubjunctiveconditionals,incontrastwithindicativeconditionalssuchas“ifhedidnotremembertotakehispassport,hehadtimeenoughtogobackhomeandpickitup”.Subjunctiveconditionalsaresometimescalled“counterfactualconditionals”,buttheantecedentneednotexpressanythingcontrarytofact:predictionsaresometimesexpressedbymeansoftheseconditionals,evenwhentheantecedentisactuallysatisfied:“ifhehadcaughtmeasles,hewouldshowexactlythosesymptoms”.compositionality:Whenweencounterawordforthefirsttime,wehavetolookitupinthedictionary,orotherwisefindoutitsmeaning.However,muchmorefrequentlythanweareawareof,weencountersentencesorphrasesthatwehaveneverheardbefore,whosemeaningswehavenodifficultywhatsoeverinunderstanding.Thisseemstosuggestthat,whileparticularwordsorlexicalitemsgettheirmeaning,asitwere,onebyone,compoundphrasesandsentencesgettheirmeaningfromthemeaningoftheconstituentexpressionsandthesyntacticmannerofcomposition.content(semantic,propositional):Utterances,giventheirlinguisticmeaning,allowustoperformdifferentspeech-acts,suchasorders,promisesorclaims;someofthem–aswhenonetellsPeter,“Peter,returnthebooktothelibrarytoday”,whenPeterreplies“IpromiseI’llreturnthebooktothelibrarytoday”,andanobservercontends“Peterwillreturnthebooktothelibrarytoday”–appeartorepresentthingsinthesameway.Thiscontentorpropositionpotentiallyrepresentedbydifferentspeech-acts(andalsobydifferentmentalattitudes)isofcoursenotsomethingthatmustobtainintheactualworld,fortheorderorthepromisemightwellnotbesatisfied,andtheclaimmightnotbetrue,andstilltheywouldrepresentthingsinthesameway;itiscommontothinkofitasatruth-condition,aconditionthattheworldmustmeet,fortheorderandthepromisetobesatisfied,ortheclaimtobetrue.contextualism(aboutknowledge,indebatesaboutsemanticcontents,etc.):Althoughthereisasenseinwhichutterancesof“Iamhungry”,madebydifferentsubjects,shareacommonmeaning,itisnaturaltothinkthattheirsemanticcontentisnotthesame:onemaybetrueincompleteindependenceoftheother,becausetheyrepresentdifferentthings;theconditionsthatshouldobtainintheworldforeachtobetrueareprimafacieindependentofeachother.2679781472578235_txt_print.indd26708/08/201415:46\nA–ZofKeyTermsIntheexample,itisnaturaltothinkthattheexpressionsresponsibleforthisare“I”,andthepresenttenseoftheverb,whicharecontext-dependentexpres-sions:theircontributiontosemanticcontentdependsonthecontextinwhichtheyareused,onwhothespeakerisinthecontextinthecaseof“I”orwhatthetimeofthecontextisinthecaseofthepresenttense.Acontextualistviewaboutagivenpieceofdiscoursepurportstodetectinstancesofcontextdependenceinunexpectedplaces.Thus,contextualistsabout“toknow”contendthat“PeterknowsthatthebankisopenonSaturdays”expressesdifferentsemanticcontentsindifferentcontexts,allowingfordifferencesintruth-value;thus,inonecontextitmaybetheclaimthatPeterknowsin-a-way-that-requires-very-little-evidencethatthebankisopenonSaturdays,andisthustrue;inanotherthatPeterknowsitin-a-way-that-requires-very-strong-evidence,andhenceisfalse.contingent:Propositions(thecontentsofthoughtorspeech)canbetrueorfalsedependingonthestateoftheworld.Someofthesepropositionsarenecessarilytrue,theycannotfailtoobtain:e.g.,(puttingasidemattersofvagueness)thateitherthereareatmosttwophonesonthetable,ortherearethreeormorephonesonthetable.Otherpropositionsobtain,butnotnecessarilyso,theymighthavefailedtoobtain:e.g.,thatthereareatmosttwophonesonthetable.Thelatterarecontingentpropositions.Inanextendedsense,itissometimessaidthatabeingisnecessary,whenthepropositionthatitexistsisnecessary,orcontingent,whenthatpropositionistrue,butnotnecessarilyso.convention:Someregularitiesinthebehaviourofagroupareessentiallysustainedbythe(perhapstacit)agreementofitsmembers,because,althoughtheyserveacoordinationfunctionusefulforthegroup,somealternativeregularitymighthavebeenequallyserviceable.Suchregularitiescanbecalled“conventions”.Theagreementneednothavebeenexplicit,butmighthavebeentacitlyadoptedafterthegroupsomehowstumbledontheregularity:theconventionofdrivingonaparticularsideoftheroadmighthavebeenadoptedinsuchaway.conversationalmaxim:AsGricepointedoutinelaboratinghistheoryofconver-sationalimplicatures,conversationsarerationalactivities,whicharetherebysubjecttonormsorprinciplesdeterminingwhentheyarepursuedrationally.Griceidentifieda“supermaxim”,ormoregeneralregulating“cooperativeprinciple”:Makeyourconversationalcontributionsuchasisrequired,atthestageatwhichitoccurs,bytheacceptedpurposeordirectionofthetalkexchangeinwhichyouareengaged;andseveralsubmaximsdevelopingfurtherthisgeneralprincipleinwaysmoreadequateforspecificconversations,inparticularthoseconsistingof2689781472578235_txt_print.indd26808/08/201415:46\nA–ZofKeyTermsexchangesofinformation.Oneofthesesubmaxims,however,soundsalmostasgeneralasthecooperativeprinciple:Berelevant!creativity:Whenitcomestolanguagemastery,Chomskyandothershaveused“creativity”torefertothefact,thatweareabletoproduceandunderstandsentencesandphraseswehavenevercomeacrossbefore.Thisfactisoftentakenasevidenceforcompositionality.Thetermcanalsobeproperlyusedtorefertoourcapacitytogivetooursentencesmeaningstheydonotconventionallyhave,andtounderstandthemwhenotherspeakersdoso,aswithmetonymyandmetaphor,conversationalimplicaturesandindirectspeech-acts.denotation:Russellintroducedthisterminhisdiscussionofdenotingphrases,suchas“everyman”,“aman”,“thekingofFrance”,etc.;andsomepeoplereserveitfortheentitysatisfyingadefinitedescription(NeilArmstronginthecaseof“thefirstpersontosetfootontheMoon”),butnowadaysitismostlyusedassynonymouswith“extension”.descriptivism:Theviewthat(usesof)namesandotherreferentialexpres-sions,likeindexicalssuchas“he”or“thatbook”,aremeaningfulthroughtheirassociationwithdefinitedescriptionssuchasthefirstheavenlybodyseenatduskinthewesternsky,or“thefirstpersontosetfootontheMoon”,whichidentifytheirreferentamongallpossiblecandidates.The“association”mightbeastightassynonymy,ormerelyreference-fixingwithoutsynonymy.determiner:Asyntacticcategoryofexpressionsincludingthedefinitearticle“the”andquantifierssuchas“every”,“some”,“most”,“the”,etc.Determinerscombinewithcommonnounstoformnounphrases,asin“mostdogs”or“someman”.directreference:AclaimsuchasthecapitalofanydemocraticcountryistheseatofitsparliamentisaboutWashingtonDC,butonlyindirectlyso;thesemanticcontentoftheclaimdirectlyinvolvesonlypropertiessuchasbeingademocraticcountry,beingacapital,plusthecontributionofdeterminerssuchasanyorthe.Directreferenceistheviewthatclaimsmadewithpropernamesandindexicalsaredirectlyabouttheirreferents,inthat,evenifthoseexpressionsaresomehowassociatedwithdefinitedescriptions,itisthereferentsthemselves,andnotthepropertiesexpressedbysuchdescriptions,thatidentifythecontents.doubleindexsemantics:Semanticcontentsaretypicallyconditionsthatobtaininsomepossibleworldsandnotinothers;theycanthusberepresentedasfunctionsfromindexes(possibleworlds)totruth-values.Anassertionofoneofthesecontentsistrueifthefunctionyieldstruthwhenappliedtotheactual2699781472578235_txt_print.indd26908/08/201415:46\nA–ZofKeyTermsworld,thevaluefortheindexsuppliedbythecontextinwhichthesentenceisuttered.Operatorssuchasitispossiblethatoritisnecessarythatchangetheindexofevaluation:“possibly,snowisblue”(or,moresimply,“snowmighthavebeenblue”)istrueoftheactualworldsuppliedbythecontextifsnowisblueistrueofsomeworld,inthiscaseadifferentone.Someexpressionsinthesamecategory(modalexpressionsinthiscase),however,donottaketheirvaluefromtheindexrelativetowhichtheyarebeingevaluatedastheyappearembeddedinasentence,butalwaysfromthevalueprovidedbythecontextofutterancefortheindex.Thus,consider“actually”in“itispossiblethatsnowisblue,eventhoughsnowisnotactuallyblue”.Thisistrueoftheactualworld,asbefore,ifsnowisblue,eventhoughsnowisnotactuallyblueistruewithrespecttosomeworldw,againinthiscasenottheactualone.However,toproperlyevaluatethepartsnowisnotactuallyblue,wecannotconsiderthevaluethatsnowisnotbluehaswithrespecttow;wemustestablishthevalueithaswithrespecttotheactualworld,thevalueprovidedbythecontextofutterance.Somethingsimilarhappenswithtemporalindexicalssuchas“now”or“today”,ifwetreatcontentsofatomicsentencesasfunctionsfromworldsandtimestotruth-values,andtensessuchasthepastorthefutureasoperatorsoverthem:weneedtokeeptrackofthetimeofthecontext,because“now”or“today”arealwaysinterpretedrelativetoit,asshownbyexamplessuchas“somedayitwillbethecasethatthecolourofsnowisblue,althoughtoday(now)itisnotblue”.Dealingwiththeseexamplesrequiresthatoursemanticshasnotjustoneindexfortheworldorthetimeofevaluation,buttwo:oneshiftablebyoperatorssuchas“possibly”or“itwillbethecasethat”,anothergivenbythecontextofutterancerequiredtointerpretindexicalssuchas“actually”and“now”nomatterwheretheyappearembeddedinacomplexsentence.empiricism(logical,radical):Empiricistsingeneraltakeperceptualexperienceandintrospectionasthefundamentalsourceofknowledgeandevidence,deflatingtheroleofreasonandaprioriknowledge;mostresearchersinthe“logicalempiricism”movementstartingaroundtheViennaCircleinthe1920sfollowedCarnapinacceptingaspecialcategoryoflogicalandmathematicaltruths,buttheytriedtoexplainthemasanalytictruthsresultingfromlinguisticstipulationsorconventions.OthersfollowQuine’smoreradicalview,rejectingthecategoryofapriorioranalytictruthsandknowledgealtogether.emptynames:Namessuchas“Vulcan”or“SherlockHolmes”thatlackthekindofreferent(extensionordenotation)inthephysicalworldthat“Venus”or“BenjaminFranklin”dohave(although,onsomeviews,theydohavereferents,extensionsordenotations,ofamoreabstractkind).2709781472578235_txt_print.indd27008/08/201415:46\nA–ZofKeyTermsexperimentalsemantics(pragmatics,philosophy):Linguiststakeforgrantedthatspeakers,theythemselvesinparticular,havelinguisticintuitionsthatarereliableindicatorsofthepropertiesofthelanguagestheyinvestigate,andthustheytypicallyrelyontheirownintuitionsinsettingup,confirmingordisconfirmingtheirtheories.Experimentalsemanticsistheresorttotheempiri-callyrefinedmethodologiesthatpsychologists,neurologistsandcognitivescientistshavedevisedinrecentyearstostudyinamoresystematicwaydataforsemantictheories–forinstance,dataaboutthetruth-valueofutterancesrelativetodifferentsituations.Experimentalpragmaticsdoesthesamefordataaboutconversationalimplicatures,presuppositions,speech-acts,etc.;experi-mentalphilosophy,fordatarelevantforphilosophicalaccountsofknowledge,reference,moralvalueandobligation,etc.extension:Inprovidingacompositionalcharacterizationofthemeaningcontributionofsententialconnectivessuchas“and”,“or”,“if...then...”,FregeandRussellfoundthatweshouldassumethatsentencessignifytheirtruth-values,theTrueandtheFalseonbivalentassumptions(theywereconsideringformallanguages,buttheyhadinviewatleastsomeusesofthoseexpressionsintheirnaturallanguages).Similarly,inprovidinganaccountofthecontributionofquantifierssuchas“every”,“some”,etc.,theyfoundthatweshouldassumethatpredicatessuchas“isaman”or“isGreek”signifytheirextensions,theclassofobjects(perhapsinagivendomainofobjects)ofwhichtheyaretrue,andsingulartermssuchas“BarackObama”or“he”orevenperhaps“thefirstmanlandingontheMoon”,theobjectstowhichtheyrefer.Truth-values,extensions(inadomain)andreferentsaretheextensionsofsentences,predicatesandsingularterms,respectively;theseexpressionsmayofcoursehaveadditionalsemanticvalues,buttheyareassignedextensionsofthekindindicatedinanypromisingsemantictheory.first-orderlogic(predicatelogic):Thelogicoflogicalconstantsincludingthesententialconnectives“not”,“and”,“ifthen”,thequantifiers“forall”and“forsome”,andidentity;i.e.,thedelineationoftheclassofvaliditiesdeterminedbythemeaningofsuchexpressions,suchas“allGreeksadmiresomepoet,SocratesisGreek,thereforeSocratesadmiressomepoet”.formallanguage:Alanguagewhosesyntaxandsemanticshasbeenexplicitlyspecified,usingtheresourcesofsomemathematicalmetatheorysuchassettheory.Thelanguageoffirst-orderlogicisagoodexample.formalsemantictheory:Asemantictheorythatusesformallanguagesasidealizedmodels(“frictionlessplanes”)fornaturallanguages,inthehopeof2719781472578235_txt_print.indd27108/08/201415:46\nA–ZofKeyTermsthusmakingmoreperspicuousitsessentialsemanticproperties,abstractingawayfromsemanticallyirrelevantfeatures.foundationalvs.descriptivesemantics:Adescriptivesemanticspurportstocharacterizethesemanticpropertiesofagivenlanguage,mostlytryingtoperspicuouslycharacterizeitscompositionalstructure:thesemanticvaluesofitslexicalitems,thoseoftheirrelevantstructuralrules,andhowtheycombineforcomplexexpressionstohavetheirresultingvalues.Afoundationalsemanticspurportstoexplainwhyagivenlanguageinfacthasthedescriptivesemanticsitdoeshave(whetherinvirtueofconventionsitsspeakersfollow,invirtueofsomeintentionsofitsspeakers,invirtueofdeeppropertiesoftheirmind/brain,etc.).Fregeanism:Fregedefendeddifferentviews,onthenatureofnumbersandofarithmeticingeneral,onlogic,etc.,andthustherearedifferentformsofFregeanism;butinphilosophyoflanguagecontextsitusuallyreferstoadescriptivistconceptionofreference.illocutionaryforce:Inspeech-acts,contentsareputforwardwithdifferentconstitutivepurposes,goals,pointsorforces:asthecontentofanassertion,anorder,apromiseorarequest.Notjustanypurposethatmaybethepointofaspeech-actisanillocutionaryforce,formanypurposes(“perlocutionary”ones)arenotessentialorconstitutiveofthespeech-actassuch,asforinstancefright-eningtheaudienceorimpressingthem.implicature(conversational-,conventional-,generalized-):Technicaltermintro-ducedbyH.P.Grice.Inmakingspeech-acts,wesometimesconveyorimplicatecontentsadditionaltothoseconstitutingthesemanticcontentoftheexpres-sionsweissue;touseGrice’sfamousexample,ifinwritingaletterofreferenceforaphilosophyjobonejustsaysthatthecandidateispoliteandhasgoodhandwriting,weconveythatwedonottakehimorhertobecommendableforthejob.Thisisaconversationalimplicature,“particularized”inthatitisessen-tiallyconveyedbyparticularutterancesofthecontext.Appealingtohistheoryofimplicatures(whichessentiallyappealstoconversationalmaxims),Gricesuggestedthatthenaturallanguage“and”hasthesamesemanticcontentasitscounterpartinfirst-orderlogic,andthussemanticallylackstemporalimplica-tions:“hegotsickandhetookapill”issynonymouswith“hetookapillandhegotsick”;thetemporalimplicationisalsoaconversationalimplicature,inthiscaseconveyedinalmostanycontextandtothatextentgeneralized.Finally,Gricesuggestedthat“and”and“but”havethesamesemanticcontent;theimplicationin“pbutq”thatthetruthofpisinsometensionwiththatofq,absentin“pandq”,isalsoanextra-semanticimplicature,althoughinthiscaseoneconventionally2729781472578235_txt_print.indd27208/08/201415:46\nA–ZofKeyTermsassociatedwiththeexpression.Recentlydifferentwritershavemadesimilarclaimsaboutpejoratives:“boche”wouldhavethesamesemanticcontentas“German”,theadditionalderogatoryaspectsbeingaconventionalimplicature.indexical:Anexpressionsuchas“I”,“he”or“now”,whichyieldsasemanticcontentrelativetopropertiesofthecontextinwhichitisused.Seedoubleindexsemanticsforsomemoredetails.inferentialrole:Apatternofinferencesassociatedwithagivenexpression,perhapsasamatterofitsmeaning.Forinstance,theintroductionrule(frompontheonehandandqontheother,inferp&q)andeliminationrule(fromp&q,inferp;fromp&q,inferq)associatedwithfirst-orderlogicconjunction.intension:Semantictheorymustascribeextensionstoexpressions.Expressionshavedifferentextensionswithrespecttodifferentpossibleworlds;theintensionofanexpressionisafunctionfromworldstoextensions,whicharealsorequiredforsemanticpurposessuchastheaccountofthemodaloperatorsofnecessity,possibilityorcontingency.is–oughtdistinction:Itisoftenthoughtthatnormativeorprescriptiveconclusionscannotbevalidlyderivedfrompurelydescriptivepremisses.Inotherwords,claimsorjudgmentsaboutwhatsomeoneoughttododonotfollowdeductivelyorconceptuallyfromclaimsorjudgmentsaboutwhatisthecaseasamatteroffact.Inafamouspassage(TreatiseIII,I,I,27),Humearticulatesthisprinciplebystatingthatisandoughtaredistinctrelations,andthatthelatterdoesnotfollowbydeductionfromtheformer.Hencetheterm“is–oughtdistinction”.→naturalisticfallacyliteralmeaning:Whatanexpressionliterallymeansonanoccasioncanbecontrastedwithwhatiscommunicated,forexample,metaphorically,indirectly,ironically,asaconversationalimplicature,etc.Thus,whensomeonesays“Ihavenothingtowear”,theydonotusuallywishtoconveythattheyhaveliterallynothingwhatsoeverthattheycanwear,whichiswhattheirwordsliterallymean.Rather,theyareconveyingsomethingnon-literally,e.g.thattheydon’thaveanythingsuitabletowearfortheupcomingoccasion,etc.Thisintuitivedifferencebetweenliteralandnon-literaluseoflanguageissometimesgivenamorespecifictheoreticalroleintheoriesaboutlanguage,cf.Recanati’sbookLiteralMeaning(2004).massterm:Masstermsormassnounsareusedonlyinthesingular.Examples:“milk”,“gold”,“pepper”;theytypicallyapplytoproperpartsofwhattheyapplyto,properpartsthereof,andsoon.Theycontrastwithcountnouns,suchas“chair”or“tiger”.2739781472578235_txt_print.indd27308/08/201415:46\nA–ZofKeyTermsmeaningtheory:Atheorythatpredictsorexplainsthemeaningoflinguisticexpressions,i.e.thoseoftheirpropertiesthatallowonetocommunicatewiththem.Onecandistinguishbetweenthetheoryofmeaningingeneral,i.e.thestudyoflinguisticmeaninginabstractionofanyparticularlanguage,andatheoryofmeaningforaparticularlanguage,suchasCatalanorSwedish.Atheoryofmeaningforaparticularlanguageisatheorythatcontainstheinfor-mationneededforunderstandingthelanguage.Generalinsightsaboutwhatindividualtheoriesofmeaningforparticularlanguageswouldneedtosaywillamounttoinsightsaboutlinguisticmeaningingeneral.modality(metaphysical,epistemic,deontic):Inadditiontoallowingustosayhowthingsare,manylanguagesallowonetosayhowthingscanormightbe,orhowthingscannotfailtobeormustbe.Inthefirstcase,onestatesapossibility,inthesecond,anecessity.Possibilityandnecessityarecalled“modalities”,andlanguagesthatcanexpressthesearecalled“modal”.Philosophersdistinguishdifferentspeciesofnecessityandpossibility:itisphysicallypossiblethatp,justifitiscompatiblewiththelawsofphysicsthatp.Itisphysicallynecessarythatpjustifitisincompatiblewiththelawsofphysicsthatnot-p.Metaphysicallypossibleiswhatiscompatiblewiththelawsofmetaphysics.Epistemicallypossibleiswhatiscompatiblewithwhatisknown.Theso-calleddeonticmodalitiesconcernwhatiscompatiblenotwithdescriptivelaws,butwithprescriptiverequirements,i.e.whatispermittedandrequiredrespectively.naturallanguage:Naturallanguagesarethoselanguagesthatareacquiredbyinfanthumansinthenaturalway,e.g.English,Japanese,etc.;orlocalversionsofEnglish,Japanese,etc.(Sometimes,naturallanguagesareacquirednotinthenaturalwayininfancy,butbyarduoustraininginadulthood,e.g.whenanativespeakerofEnglishtakesJapaneseclasses.)Naturallanguagescontrastwithartificiallanguages,i.e.languagesthathavebeendesigned,suchasprogramminglanguagesortheformallanguagesthatlogiciansandtheoreticallinguistsbuildasmodels.naturalisticfallacy:Theterm“naturalisticfallacy”wasintroducedbyG.E.MooreinhisbookPrincipiaEthicaasatermforthe(alleged)mistakemadebyanyonewhothinksthat“good”canbedefinedinnaturalterms.Mooreusedthe“openquestionargument”toarguethatnosuchdefinitioncanbecorrect.Buttheterm“naturalisticfallacy”isalsooftenusedmorelooselytodenoteanymistakeninferenceofprescriptiveornormativeconclusionsfrompurelydescriptivepremisses.→is–oughtdistinctionnecessity(metaphysical-,epistemic-,deontic):→modality2749781472578235_txt_print.indd27408/08/201415:46\nA–ZofKeyTermspossibility(metaphysical-,epistemic-,deontic):→modalitypossibleworld:Inadditiontothewaytheworldactuallyis,therearemanynon-actual(counterfactual)waystheworldmightbe.Apossibleworldisacompletewaytheworldmightbe.Thus,theactualworldisoneofthepossibleworlds.Inmodernsemantics,thenotionofapossibleworld(andsimilarnotions,suchasCarnap’sstate-descriptions)hasbeenusedinmodellingorexplainingvariousmodalorintensionalphenomena.Thus,necessityisoftenmodelledastruthinallpossibleworlds,andbeliefisoftenmodelledasthesetofpossibleworldsthatareregardedas“livepossibilities”.Standardintensionsarefunctionsfrompossibleworldstoextensions.Thereissomecontroversyaboutthemetaphysicalstatusofpossibleworlds.predicate:Traditionalgrammardistinguishesthesubjectandthepredicateasthemainpartsofasentence.Thus,inthesentence“Socratesiswise”,“iswise”wouldbethepredicate.Morerecentlinguisticsusesdifferentsyntacticcategories,suchasVP(verb-phrase)andNP(noun-phrase).presupposition:WhenIsaytoyou“Yourinferioritycomplexhasdisap-peared”,thenIampresupposingthatyouusedtohaveaninferioritycomplex.Presupposingthisisdifferentfromstatingit,e.g.whenIsay“Youusedtohaveaninferioritycomplex.Nowithasdisappeared.”Presuppositionsalsodifferfromotherkindsofimplicationintheircharacteristic“projection”properties:presuppositionsarepreservedunderacertainrangeofembeddingsormodificationsunderwhichotherimplicationsarenotpreserved.Thus,thepresuppositionthatyouhaveaninferioritycomplexdoesnotgoawaywhenIturnmyinitialclaimintoaquestion:“Hasyourinferioritycomplexdisap-peared?”;orwhenIembeditmodally:“Possiblyyourinferioritycomplexhasdisappeared”;orevenwhenIaddasimplenegation:“Yourinferioritycomplexhasnotdisappeared.”proposition:Propositionsaretheoreticalabstractentitiesthathavebeenpostulatedasthe“objects”or“contents”ofspeechandthought.Thusitissaidthatwhenyoudesirethattherebearevolutionnextyear,andIsupposethattherewillbearevolutionnextyear,andathirdpersonbelievesandassertsthattherewillbearevolutionnextyear,thenthedesire,thesuppo-sition,thebeliefandtheassertionallsharethesamecontent,namelythepropositionthattherewillbearevolutionnextyear.Thereisphilosophicalcontroversyabouttheexistenceofpropositionsaswellasabouttheirnature.Astotheirnature,theoristsaredividedintothosewhowanttoconstruepropositionsassetsofpossibleworlds,andthosewhopreferatheoryof“structuredpropositions”.2759781472578235_txt_print.indd27508/08/201415:46\nA–ZofKeyTermspropositionalattitude:Propositionalattitudesarethosementalstatesandmentalactsthatcanorhavebeenconstruedasinvolvingarelationwithaproposition.Similarly,manyspeech-actsareconstruedaspropositionalacts.quantifier(generalized-,restricted-):Quantifiersareexpressionsthatbindvariables,suchastheexistentialquantifier“$x”andtheuniversalquantifier“∀x”inclassicallogic.Roughly,ifwehaveapredicate“exists”,aformula“$x:xexists”istruejustifoneofthevaluesforthevariablexinthedomainofquantificationisintheextensionof“exists”.Similarly,“∀x:xexists”istrueifallthevaluesofxinthedomainmeetthiscondition.Thus,quantifierscanbeusedtosaysomethingabouttheproportionofmembersofthedomainthatsatisfythepredicate(oropensentence)towhichthequantifierisapplied.Generalizedquantifiertheorystudiesallpossiblequantifiers,whereasrestrictedquantifiersarequantifiersthatrelatetwopredicates(oropensentences)tooneanother:e.g.(Mostx:Fx)Gxwhichwouldroughlybeinterpretedas:mostthingsinthedomainthatsatisfyFx,alsosatisfyGx.Restrictedquantifiersallowamorefaithfulformalmodelofnaturallanguagequantifierphrases,suchas“everywoman”or“mostpoliticians”.rationalism:Viewsthataffordgreatimportancetopurereasonasasourceofknowledgeareoftencalled“rationalism”.Theexactextentoftheroleofreasoncanvaryfrom“someknowledgeisacquiredbyreasonalone”to“allknowledgeisacquiredbyreasonalone.”rigidity:Anexpressioniscalled“rigid”ifitsextensiondoesnotvaryfromonepossibleworldtoanother.Thus,KripkehasfamouslyarguedinNamingandNecessity(1982)thatpropernamesarerigid:theynamethesameindividualinallpossibleworlds.Semanticnormativity:Thethesisthatthereareirreduciblenormsthatgoverntheuseoflinguisticexpressionsiscalledathesisof“semanticnormativity”.semanticvalue:ItwasFrege’sideatoemploythemathematicaltheoryoffunctionsinthedescriptionoflanguages,treatingsomeexpressionsasreferringtofunctionsandotherstoargumentsthataremembersofthedomainofthesefunctions.Whenoneofthefunction-expressions,e1,isthensyntacti-callyappliedtooneoftheargument-expressions,e2,theresultingcompoundexpressione1(e2)referstothevalueofthereferentofe1fortheargumentthatisthereferentofe2.ThisallowedFregeasystematicdescriptionofhowthereferentsofcompoundsaredeterminedbythereferentsoftheirpartsandthewaytheyarecombined(→compositionality).Fregeusedthetheoreticalnotionofa“referent”insuchasystematicdescriptionofalanguage,i.e.in2769781472578235_txt_print.indd27608/08/201415:46\nA–ZofKeyTermsasemantics.Othertheoristslaterintroducedtheterm“semanticvalue”asagenerictermfortheentitiesthatarethussystematicallyassignedinasemantics.semantics:Asystematicdescriptionofalanguagethatmodelsatleastanimportantaspectofthemeaning-relatedpropertiesoftheexpressionsofthatlanguage.Formalsemanticsdescribes(andtherebystipulates)themeaningpropertiesofsomeformallanguage.Naturallanguagesemanticsprovidesadescriptionofafragmentofanaturallanguage,orperhapsdesignsformallanguagesthataresupposedtomodelcertainaspectsofnaturallanguages.singularterm:Thereisnogenerallyagreedupondefinitionof“singularterm”.“JohnStuartMill”and“theinventoroflogic”areusuallythoughtofasexamplesofsingularterms.Bycontrast,“philosopher”and“inventor”areusuallyclassifiedasgeneralterms.Mill(inASystemofLogic,1843,p.17a)providesadefinitionthatisagoodguide:hesaysthatasingularterm(or,inMill’sterminology,“singularname”)isa“namewhichisonlycapableofbeingtrulyaffirmed,inthesamesense,ofonething”.Bycontrast,hedefines“generalterm”asfollows:“namewhichiscapableofbeingtrulyaffirmed,inthesamesense,ofeachofanindefinitenumberofthings”.Thepropersemantictreatmentofsingularterms,inparticularthequestionofwhethertheyconstituteasemanticallyhomogeneouscategory,ishistoricallyamongstthemostdebatedissuesinthephilosophyoflanguage.speech-act:Justlikehittinganailwithahammerisakindofaction,performedwithacertainpurposebyanagent,theuseoflanguageincommunicationcanalsobethoughtofasatypeofaction,whichcanbedeliberate,serveapurpose,andbeassessedintermsofitsrationality.Forexample,theeventwhichconsistsinapersonutteringthewords“Onecheeseburger,please”inafastfoodoutletqualifiesasanaction,justlikehiswalkingtotheoutletorblowinghisnose.It’stheactionoforderingacheeseburger.Bycontrast,sneezingandbreathingdonotseemclassifiableasactions.J.L.Austiniswell-knownforhavingdistinguishedmanyspeech-acttypes,suchasasserting,promising,commandingorasking.substitutabilitysalvaveritate:“Salvaveritate”isLatinfor“underpreser-vationoftruth”.So,tosaythatoneexpressionissubstitutablesalvaveritateforanothermeansthatsubstitutingtheonefortheotherwillnotaffectthetruth-value.syntax:Syntaxofalanguagedefinesordescribeswhichexpressionsarewell-formed.Inthecaseofaformallanguage,thisisamatterofdefinition.Inthecaseofanaturallanguage,thisisamatterofprovidingasystematicorprincipledaccountofwhichexpressionsaregrammaticallycorrect.Thus,2779781472578235_txt_print.indd27708/08/201415:46\nA–ZofKeyTermsasyntaxofEnglishthatpredictsthat“Andcougheddrankbeerhe”iswell-formedwouldseemtobeincorrect.synthetic:Notanalytic.truth-condition:Thetruth-conditionofasentence(incontext)istheconditionunderwhichitistrue.Theexpression“truth-condition”isoftenusedbyadherentsoftruth-conditionalsemanticsinordertocharacterizethesemanticvaluesofsentences(incontext).However,thetermmasksacertainunclarity.ItisoftenthoughtthataTarski-styleextensionalsemantictheoryforalanguageLspecifiesthetruth-conditionsofsentencesinLbygeneratingaT-theoremoftheform“sisTiffp”foreachsentencesinL.However,truth-conditionscouldonlyfigureasthesemanticvaluesofsentencesiftheygobeyondwhatsuchtheoremsstate.truth-value:Insomesemantictheories,thesemanticvaluesofsomeex-pressionsare“truth-values”.Frege’soriginalsemanticstreatedsentencesashavingtruth-valuesastheirsemanticvalue(referent).InFrege,thereareonlytwotruth-values:theTrueandtheFalse.Butinprinciplewecouldhavemoretruth-values.vagueness.Imagineaseriesof400.001men,startingwithonewith0hairsonhishead,anotherwith1,andsoonuntilthelastonewhohas400.000hairs.Thefirstisclearlybald,thelastclearlynon-bald.Inbetween,wewouldfindborderlinecasesfortheapplicationof“bald”:menofwhomitisunclearwhethertheyarebald,andunclearwhethertheyarenot.Relativetothisseries,wecouldbuildasoritesargument:thefirstmanisbald;ifamanwithnhairsisbald,anotherwithn+1hairsisstillbald;therefore,thelastmanisbald.Predicatessuchas“bald”,whichhaveborderlinecasesandallowforsoritesparadoxes,arevague.variable:Avariableisanexpressionthathasvariablesemanticvalues.Thus,inastandardTarskiansemanticsforaformallanguage,variablesareassignedsemanticvaluesonlyrelativetoan“assignmentofvaluestovariables”.Variablesthereforecontrastwithconstants.Constantexpressionshavethesamesemanticvaluewithrespecttoeveryassignment.Innaturallanguage,variablesareoftenclaimedtooccuratthelevelofdeepstructureorLF,withoutbeingvisibleatthesurfacelevel.Theevidencefortheirexistencecomesfromphenomenasuchasbinding.Pronounsaresometimesclaimedtobesurfaceexpressionsthatfunctionlikevariables.verification:Toverifysomethingistofindevidencethatitistrue.2789781472578235_txt_print.indd27808/08/201415:46\nSelectBibliographyThefollowingcontainssuggestionsforfurtherreadingcorrespondingtoeachofthemainchaptersofthisvolume.1.EditorialIntroduction:HistoryofthePhilosophyofLanguage(ManuelGarcía-Carpintero)Dever,Josh(2006),“Compositionality”,inE.LeporeandB.Smith(eds),TheOxfordHandbookofPhilosophyofLanguage,Oxford:OxfordUniversityPress,pp.633–66.[Anexcellentintroductiontocontemporarydiscussionsofcomposition-ality(whichcanbecompletedwiththetwoarticlesbyPaginandWesterståhlmentionedbelowforChapter4)]García-Carpintero,ManuelandJosepMacià(eds)(2006),Two-DimensionalSemantics,Oxford:OxfordUniversityPress.[Containsmanyinfluentialpapersontwo-dimensionalsemantics,withanintroductoryexposition.]Sainsbury,Mark(1993/2002),“RussellonMeaningandCommunication”,inDepartingfromFrege,London:Routledge,pp.85–101.[AnuancedexaminationofRussell’sactualviewsonpropernames.]Soames,Scott(2003),PhilosophicalAnalysisintheXXthCentury,vol.1:TheDawnofAnalysis,Princeton,NJ:PrincetonUniversityPress.[AveryclearpresentationoftheTractarianviews,inthecontextofequallyclearintroductionstotheviewsofRussellandMoore(needlesstosay,someoftheinterpretationssuggestedinthetextdifferfromSoames’s).]Williamson,Timothy(2007),ThePhilosophyofPhilosophy,Oxford:Blackwell.[Afirst-classprovocativeviewonthemethodologyofphilosophy.]2.OntheNatureofLanguage(JamesHigginbotham)Baker,MarkC.(2002),TheAtomsofLanguage:TheMind’sHiddenRulesofGrammar,NewYork:BasicBooks.[Containsageneralviewofparametricvariationwhichdoesnotrequirespecializedknowledge.]Chomsky,Noam(2000),NewHorizonsintheStudyofLanguageandMind,Cambridge:CambridgeUniversityPress.[RecentmaterialthatlinksChomsky’stechnicalworkwithhisviewsongeneralissuesofscientificinquiryanditsmodernhistory.]Guasti,MariaTeresa(2004),LanguageAcquisition:TheGrowthofGrammar,Cambridge,MA:MITPress.[Excellentsurveyofcontemporarydevelopmentalpsycholinguistics.]2799781472578235_txt_print.indd27908/08/201415:46\nSelectBibliographyGuéron,JacquelineandLilianeHaegeman(1998),EnglishGrammar:AGenerativePerspective,Oxford:Blackwell.[Excellentsurveyofcontemporarycomparativegenerativegrammar.]Heim,IreneandAngelikaKratzer(1998),SemanticsinGenerativeGrammar,Oxford:Blackwell.[Standardvolumeonsemanticsandtheinteractionbetweensemanticsandsyntax.]Larson,RichardandGabrielSegal(1995),KnowledgeofLanguage:AnIntroductiontoSemanticTheory,Cambridge,MA:MITPress.[Standardvolumeonsemanticsandtheinteractionbetweensemanticsandsyntax.]3.FormalSemantics(JoshDever)Barker,C.andP.Jacobson(2007),DirectCompositionality,Oxford:OxfordUniversityPress.[Ananthologyofpapersdevelopinganalternativeframeworktothatgivenhere,inwhichquantifierambiguitiesareexplainedwithoutsyntacticmovementrules.]Cresswell,M.(2006),“FormalSemantics”,inM.Devitt,M.andR.Hanley(eds),TheBlackwellGuidetothePhilosophyofLanguage,Oxford:Blackwell.[Arecentoverviewofthenatureofformalsemantics.]Heim,I.andA.Kratzer(1998),SemanticsinGenerativeGrammar,Oxford:Blackwell.[Amajortextbookinformalsemantics,developinginmoredetailsomeoftheideaspresentedhere.]Kamp,H.andU.Reyle(1993),FromDiscoursetoLogic:IntroductiontoModel-TheoreticSemanticsofNaturalLanguage,FormalLogic,andDiscourseRepresentationTheory,London:Springer.[Adetailedoverviewofdiscourserepresentationtheory,analternativetothesemanticapproachdevelopedhere.]Montague,R.(1974),FormalPhilosophy:SelectedPapers,NewHaven:YaleUniversityPress.[Containsseveralseminalworksbyoneofthefoundingfiguresofformalsemantics.]Portner,P.andB.Partee(2002),FormalSemantics:TheEssentialReadings,Wiley-Blackwell.[Acollectionofanumberofthemostinfluentialpapersinformalsemantics.]Stanley,J.(2008),“PhilosophyofLanguageintheTwentiethCentury”,inTheRoutledgeCompaniontoTwentiethCenturyPhilosophy,London:Routledge.[Asurveyofthehistoricaldevelopmentofthephilosophyoflanguage,whichhelpsgiveabroadercontextfortheprojectsofformalsemantics.]4.TheoriesofMeaningandTruthConditions(KathrinGlüer)Davidson,D.(1967/1984),“Truthandmeaning”,inInquiriesintoTruthandInterpretation,Oxford:ClarendonPress,1984,pp.17–36.[Davidson’sseminalarticlearguingthataTarskiantheoryoftruthcanbeusedasaformalsemantictheory.]Davidson,D.(1973/1984),“Radicalinterpretation”,inInquiriesintoTruthand2809781472578235_txt_print.indd28008/08/201415:46\nSelectBibliographyInterpretation,Oxford:ClarendonPress,1984,pp.125–39.[Davidson’sclassicalmeaning-theoreticalarticle.]Dummett,M.(1974/1993),“Whatisatheoryofmeaning(I)”,inTheSeasofLanguage,Oxford:OxfordUniversityPress1993,pp.1–33.[ThefirstofDummett’sclassicalmeaning-theoreticalarticles.]Dummett,M.(1976/1993),“Whatisatheoryofmeaning(II)”,inTheSeasofLanguage,Oxford:OxfordUniversityPress,1993,pp.34–93.[ThesecondofDummett’sclassicalmeaning-theoreticalarticles.]Grice,P.(1957/1987),“Meaning”,inStudiesintheWaysofWords,Cambridge,MA:HarvardUniversityPress,1987,pp.213–23.[Grice’sseminalpaperprovidingtheclassicalanalysisofnon-naturalmeaningintermsofintentions.]Pagin,P.(2006),“Meaningholism”,inE.LeporeandB.Smith(eds),HandbookofPhilosophyofLanguage,Oxford:OxfordUniversityPress,pp.213–32.[Usefuldiscussion,andthemostcompleteoverview,ofdifferentversionsofsemanticholism.]Pagin,P.andWesterståhl,D.(2010a),“CompositionalityI:DefinitionsandVariants”,PhilosophyCompass5,250–64.[Togetherwith2010b,ausefulintroductionto,overviewover,anddiscussionofdifferentversionsoftheideathatnaturallanguagehasacompositionalsemantics.]Pagin,P.andWesterståhl,D.(2010b),“CompositionalityII:DefinitionsandVariants”,PhilosophyCompass5,265–82.[Togetherwith2010a,ausefulintro-ductionto,overviewover,anddiscussionofdifferentversionsoftheideathatnaturallanguagehasacompositionalsemantics.]Quine,W.V.O.(1960),WordandObject,CambridgeMA:MITPress.[ChapterTwointroducesradicaltranslationandarguesfortheindeterminacyoftranslation.]Stalnaker,R.(1997),“Referenceandnecessity”,inB.HaleandC.Wright(eds),ACompaniontothePhilosophyofLanguage,Oxford:Blackwell,pp.534–54.[Usefuloverviewoftheconnectionsbetweenreferenceandnecessity.Coinstheterm“foundationalsemantics”forthetheoryofmeaningdetermination.]5.Reference(GenovevaMartí)Braun,David(1993),“EmptyNames”,Noûs27,449–69.[Explores,fromoneperspective,theproblemthatemptynamesposefornewtheoriesofreference.]Devitt,Michael(2011),Theoria.[Alsodiscussestheroleofempiricalevidenceinthetheoryofreference.]Devitt,MichaelandKimSterelny(1987),LanguageandReality,Oxford:Blackwell.[Anexplorationoffundamentalissuesinthetheoryofreference,fromtheperspectiveofacausal-historicalandanti-descriptiviststandpoint.]Donnellan,Keith(1970),“ProperNamesandIdentifyingDescriptions”,Synthese21,335–8.[Anotherclassic.Donnellanarguesthatthereferenceofauseofanameisnotdeterminedbyanassociateddefinitedescription.]Evans,Gareth(1973),“TheCausalTheoryofNames”,ProceedingsoftheAristotelianSociety,SupplementaryVolume47,187–208.[Anin-depthdiscussionofthecausal-historicalpictureofreference.]Kripke,Saul(1982),NamingandNecessity,Cambridge,MA:HarvardUniversity2819781472578235_txt_print.indd28108/08/201415:46\nSelectBibliographyPress.[Kripke’sclassic1970lectures,inwhichheattacksdescriptivismandarguesthatnamesarerigiddesignators.]Machery,Edouard(2011),“ExpertiseandIntuitionsaboutReference”,Theoria.[Discussestheroleofempiricalevidenceinthetheoryofreference.]Marti,Genoveva(forthcoming2012),“EmpiricalDataandtheTheoryofReference”,inM.O’Rourkeed.TopicsinContemporaryPhilosophy.Vol10.Cambridge,MA:MITPress.[Alsodiscussestheroleofempiricalevidenceinthetheoryofreference.]Reimer,Marga(2001),“TheProblemofEmptyNames”,AustralasianJournalofPhilosophy,79,491–506.[Explores,fromanotherperspective,theproblemthatemptynamesposefornewtheoriesofreference.]Soames,Scott(2002),BeyondRigidity,NewYork:OxfordUniversityPress.[AnanalysisofthelessonsofNamingandNecessity.ThebookcontainsdiscussionsofsomeofthedescriptivistanswerstoKripke’sarguments,soitisalsoagoodbibliographicalsourcetofollowthehistoryofthedevelopmentofthediscussionafter1970.]6.IntensionalContexts(MichaelNelson)Braun,D.(1998),“Understandingbeliefreports”,PhilosophicalReview107,555–95.[Adefenceofthesubstitutionofco-referringnameswithinpropositionalattitudecontexts.]Davidson,D.(1968),“Onsayingthat”,Synthese19,130–46.[Aclassicdefenceofsententialism,thethesisthatpropositionalattitudeverbsreportrelationstosentences.]Frege,G.(1918/1956),“Thethought:Alogicalinquiry”,Mind65,289–311.[Aclassicdiscussionofindexicalthoughts.]Kaplan,D.(1986),“Opacity”,inL.HahnandP.Schilpp(eds),ThePhilosophyofW.V.Quine,LaSalle:OpenCourt,pp.229–89.[Adiscussionofintensionalcontexts,ofbothbeliefandalethicmodality,andQuine’sargumentsagainstquantifyingin.]Kripke,S.(1979),“Apuzzleaboutbelief”,inA.Margalited.MeaningandUse,Dordrecht:Reidel,239–83.[AclassicargumentthatFrege’spuzzleisnotdrivenbyanythesesconcerningthesemanticsofpropernames.]Perry,J.(2000),TheProblemoftheEssentialIndexical,expandededition,Stanford:CSLI.[AcollectionofPerry’simportantcontributionstopropositionalattitudes.]Quine,W.(1956),“Quantifiersandpropositionalattitudes”,JournalofPhilosophy53,177–87.[Aclassicdiscussionoftheproblemsofquantifyingintointensionalcontexts.]Quine,W.(1952/1980),“Referenceandmodality”,2ndrevisedversion,inhisFromaLogicalPointofView,2ndedn,revisedprinting,NewYork:HarperandRow,pp.139–59.[Aclassicargumentagainstquantifyingintomodalcontexts.]Richard,M.(1990),PropositionalAttitudes:AnEssayonThoughtsandHowWeAscribeThem,Cambridge:CambridgeUniversityPress.[Anaccessiblediscussionofpropositionalattitudereports.]Russell,B.(1910),“Knowledgebyacquaintanceandknowledgebydescription”,2829781472578235_txt_print.indd28208/08/201415:46\nSelectBibliographyProceedingsoftheAristotelianSociety11,108–28.[Aclassicdiscussionofthedistinctionbetweenknowledgebyacquaintanceandknowledgebydescription.]Salmon,N.(1989),“Illogicalbelief”,PhilosophicalPerspectives3,243–85.[Adiscussionofrationalbeliefinapropositionanditsnegation.]7.ContextDependence(KentBach)Bach,Kent(2005),“Contextexmachina”,inZ.Szabóed.Semanticsvs.Pragmatics,Oxford:OxfordUniversityPress,pp.15–44.[Debunksextravagantclaimsabouttheroleofcontext.]Braun,David(2007),“Indexicals”,StanfordEncyclopediaofPhilosophy,availableonlineathttp://plato.stanford.edu/entries/indexicals/[PresentsKaplan’stheoryofindexicalsandsomealternativetheories,andthendiscussesvariousissuesraisedbyindexicalsanddemonstratives.]Glanzberg,Michael(2007),“Context,content,andrelativism”,PhilosophicalStudies136:1–29.[Arguesforacontextualistandagainstarelativistsemanticsforpredi-catesofpersonaltaste,andsuggeststhatthesearenotimportantlydifferentfromothergradableadjectives]Kennedy,Christopher(2007),“Vaguenessandgrammar:Thesemanticsofrelativeandabsolutegradableadjectives”,LinguisticsandPhilosophy30:1–45.[Onthesemanticsofgradableadjectivesandtheirapparentcontextsensitivity,contrastingthestructureofscalesassociatedwithrelativeandabsoluteadjectives]Neale,Stephen(2006),“Pronounsandanaphora”,inM.DevittandR.Hanley(eds.),TheBlackwellGuidetothePhilosophyofLanguage,Oxford:Blackwell,pp.335–73.[Acomprehensivediscussionofvarioussortsofoccurrencesofpronounsthat,becauseofhavingantecedents,seemsensitivetotheirlinguistcontext]Perry,John(2006),“Usingindexicals”,inM.DevittandR.Hanley(eds.),TheBlackwellGuidetothePhilosophyofLanguage,Oxford:Blackwell,pp.314–34.[Presentsthedistinctionbetweenautomaticanddiscretionaryindexicalsanddiscussestheiruses,includingproblemcasesliketheansweringmachinepuzzle]Soames,Scott(2010),“Thelimitsofmeaning”,ch.7ofPhilosophyofLanguage,Princeton:PrincetonUniversityPress,pp.145–73.[DiscussesKaplan’saccountofindexicalsanddemonstrativesandappliesthenotionofsemanticincompletenesstogiveanalternativeaccount]8.Pragmatics(FrançoisRecanati)Austin,J.(1975),HowtoDoThingswithWords,2ndedn,Oxford:ClarendonPress.[Austin’spioneeringlecturesonspeechacttheory.]Clark,H.(1996),UsingLanguage,Cambridge:CambridgeUniversityPress.[Aninvestigationintothepragmaticdimensionoflanguage,byaleadingpsychologist.]Davis,S.(ed.)(1991),Pragmatics:AReader,NewYork:OxfordUniversityPress.[Aninclusivecollection,wheremanypapersmentionedinthisentrycanbefound.]2839781472578235_txt_print.indd28308/08/201415:46\nSelectBibliographyFillmore,C.(1997),LecturesonDeixis,Stanford :CSLI.[Indexicalityinhumanlanguages.]Grice,P.(1989),StudiesintheWayofWords,Cambridge:CambridgeUniversityPress.[ContainsGrice’sfamouslectures“LogicandConversation”andacollectionofhispapersonthephilosophyoflanguage,includingtheseminal“Meaning”.]Kadmon,N.(2001),FormalPragmatics,Oxford:Blackwell.[Asurveyoftheinterfacebetweenpragmaticsandformalsemantics,withspecialattentiontopresuppo-sitionandfocus.]Recanati,F.(2004),LiteralMeaning,Cambridge:CambridgeUniversityPress.[Asurveyofthecontextualism/literalismdebate.]Searle,J.(1969),SpeechActs,Cambridge:CambridgeUniversityPress.[Aclassicofpragmatics.]Sperber,D.andD.Wilson(1995),Relevance:CommunicationandCognition,2ndedn,Oxford:BasilBlackwell.[AcognitivelyorientedtheoryintheGriceantradition.]Stalnaker,R.(1999),ContextandContent,NewYork:OxfordUniversityPress.[ContainsseveralofStalnaker’sinfluentialpapersonthefoundationsofpragmatics.]9.SemanticNormativityandNaturalism(JoséL.Zalabardo)Boghossian,P.(1989),“TheRule-FollowingConsiderations”,Mind98,507–49.[Defendstheviewthatmeaningisnormative.]Boghossian,P.(2003),“TheNormativityofContent”,PhilosophicalIssues13,31–45,[Rejectstheauthor’searlierviewthatlinguisticmeaningisnormativeandarguesthatthenormativityofmentalcontentisdefensibleonlyifitcanbederivedfromthenormativityofbelief.]Brandom,R.(1994),MakingitExplicit,Cambridge,MA:HarvardUniversityPress.[Animportantdefenceoftheviewthatmeaningisnormativefromapointofviewthathasn’tbeencoveredinthischapter.]Glüer,K.andÅ.Wikforss(2009a),“AgainstContentNormativity”,Mind118,31–70.[Rejectssemanticnormativity.]Glüer,K.andÅ.Wikforss(2009b),“TheNormativityofMeaningandContent”,TheStanfordEncyclopediaofPhilosophy(Summer2009Edition),EdwardN.Zalta ed.availableonlineathttp://plato.stanford.edu/archives/sum2009/entries/meaning-normativity/[Usefulsurveyoftheliteratureonsemanticnormativity.]Kripke,S.(1982),WittgensteinonRulesandPrivateLanguage,Oxford,Blackwell.[Widelyseenasthecontemporarysourceoftheideathatthenormativityofmeaningposesaproblemforsemanticnaturalism.]10.Analyticity,Apriority,Modality(AlbertCasullo)Ayer,A.J.(1952),Language,TruthandLogic,NewYork:DoverPublications.[AyerdefendstheviewthatsomeknowledgeisaprioriagainstMill’scontentionthat2849781472578235_txt_print.indd28408/08/201415:46\nSelectBibliographyallknowledgeisaposteriori,andoffersaclassicaldefenceofthelogicalempiricistviewthatthereisnosyntheticaprioriknowledge.]BonJour,L.(1998),InDefenseofPureReason,Cambridge:CambridgeUniversityPress.[BonJourdefendsthetraditionalrationalistviewthatrationalinsightintothenecessarytruthofapropositionisthesourceofaprioriknowledge.Hearguesagainstbothlogicalempiricism,whichlimitsaprioriknowledgetoanalytictruths,andradicalempiricism,whichmaintainsthatallknowledgeisaposteriori.]Casullo,A.(2003),APrioriJustification,NewYork:OxfordUniversityPress.[Casullooffersanarticulationoftheconceptofaprioriknowledgeandarguesthatthetraditionalargumentsforandagainsttheexistenceofaprioriknowledgeareinconclusive.Hemaintainsthataresolutionofthecontroversyovertheexistenceofsuchknowledgerequiresempiricalinvestigation.]Kant,I.(1965),CritiqueofPureReason,translatedbyNormanKempSmith,NewYork:St.Martin’sPress.[Kantintroducesthefourprimaryquestionsregardingaprioriknowledgethatframethecontemporarydiscussionofaprioriknowledge.Hisanswerstothosequestionsremaincontroversialandcontinuetodominatethecontemporarydiscussion.]Kitcher,P.(1983),TheNatureofMathematicalKnowledge,NewYork:OxfordUniversityPress.[Kitcheroffersawide-rangingsurveyoftraditionalaprioriaccountsofmathematicalknowledge.Herejectssuchaccountsandarticulatesaversionofmathematicalempiricism.]Kripke,S.(1980),NamingandNecessity,Cambridge:HarvardUniversityPress.[Kripkesetsthestageforthecontemporarydiscussionofaprioriknowledgebycarefullydistinguishingtheconceptsofaprioriknowledge,necessarytruthandanalytictruth.Healsochallengestraditionalassumptionsabouttherelationshipbetweenaprioriknowledgeandnecessarytruthbyofferingexamplesofnecessaryaposterioriknowledgeandcontingentaprioriknowledge.]Peacocke,C.(2004),TheRealmofReason,Oxford:OxfordUniversityPress.[Peacockedefendsaversionofmoderaterationalism,whichrejectsthetraditionalration-alistaccount,accordingtowhichaprioriknowledgeistobeexplainedbyadistinctpsychologicalfaculty.Heoffersinitsplaceametasemanticaccount,whichexplainssuchknowledgeintermsoffeaturesofconceptpossession.]Quine,W.V.(1963),“TwoDogmasofEmpiricism”,inFromaLogicalPointofView,2ndedn,revised,NewYork:HarperandRow,pp.20–46.[Quinerejectsthelogicalempiricisttraditionbyarguingagainstthecogencyofthedistinctionbetweenanalyticandsynthetictruths.HealsorejectsMill’sinductiveempiricistaccountofmathematicalandlogicalknowledge,andarticulatesanholisticempiricistaccountofsuchknowledge.]2859781472578235_txt_print.indd28508/08/201415:46\nIndexaprioriknowledge3,5–7,13–16,19–22,beliefreports49–50,55–7,66,74–80,161;221–2,228–50,259–60,266,284–5seealsoFrege’spuzzle;intensionalofcontingenttruths15–16,229,240,285contextsAckerman,F.122n.9Bennett,J.101n.15adjectives26–7,44n.7,61,89Berg,J.140gradable170–7,258–9,283bivalence103n.32,266,271adverbs58,61,81n.14Blackburn,S.204Ajdukiewicz,K.81n.12Boghossian,P.176,204,225nn.10–11,284Almog,J.123n.18BonJour,L.229,285Alston,W.186Borg,E.100n.4,163,194,254,256ambiguity12,32,43,44n.7,49–50,Bowerman,M.40–1,44n.866–71,74–80,80n.1,81n.26,154–5,Brandom,R.284197,280Braun,D.124n.25,124n.28,140,160,168,analytictruths13,206,223,226n.19,228,281–3238–9,245–50,260,270,285Buchanan,R.174vs.synthetictruths40–1,228–9,238,Bühler,K.185245–6,248–50,266,285;Burge,T.7seealsoconceptualanalysisanaphora135,158,283Caplan,B.122n.8anti-realism;seetruth,theoriesof,Cappelen,H.100n.4,163,165–6,194,254,epistemic256appositivevs.descriptiveinterpretationsCarnap,R.13,124n.33,129,185,259–60,44n.7270,275Aristotle1,16,107Carston,R.143,177,194,255–6,259artificiallanguages,seeformallanguagesCasullo,A.229,237,241,285assertions3–4,17–18,32,38–9,55,88,chainsofcommunication113–15,118–19,90,96–7,101n.10,102n.22,110,158,123nn.16–18,266163–4,176,187–8,196,199,225n.3,Chalmers,D.16,244,260258,269–70,272,275,277characters,Kaplanian15,88,141,160,166,Austin,J.17–18,101n.12,185–7,192,168,191,251–2198–9,277,283charity,theprincipleof93–5,99,102n.24Ayer,A.J.13,234–5,238,246–9,284Chemla,E.260Chisholm,R.233Bach,K.142,157–8,164,166,169,177,Chomsky,N.16,18,26,28–32,44n.2,195,199,254,256,28381nn.20–1,101n.12,269,279Baker,M.6,279Church,A.81n.13,129Bally,C.185circumstancesofevaluation252–3,255,Barwise,J.81nn.16–17,130,135257–8,270Bealer,G.19,140Clapp,L.142Beaver,D.261Clark,H.195,283beliefcontentsvs.beliefstates141–2,codemodelofcommunication,the144–5199–2002869781472578235_txt_print.indd28608/08/201415:46\nIndexCohen,L.J.45n.11reference-fixing13–15,123n.16,240,Cohen,S.257269commonground144,156,195–6,199referentialuseof9,12,18communicativeintentions89–91,114–15,rigidified12–13,18,111–12,122123n.17,145,154–9,161,166–70,174,seealsodescriptivism195,198–200,272,281demonstratives13,27,39,106,116–17,compositionality,theprincipleof16,20–1,131,136,138–9,141,157–8,160–1,81n.22,86–7,91,100n.7,101n.12,166–70,176,195,260,283;seealso154,254–6,267,269,276,279indexicalslocal42–3,45n.13DeRose,K.257seealsosemantics,compositionalDescartes,R.188conceptualanalysis16,19–21,203–26;seedescriptivism7,11–13,18–19,22,23n.10,alsoanalytictruths106–24,134,136–9,266,269,272,282conditionals89,98,267causal115,123n.19,260contextdependence9,15,17–18,33,determiners8,27–8,34,44n.9,46,63–6,38–40,43,47–8,53–5,81n.9,88,68,77,269116,127,139,143,147,153–79,Dever,J.81n.22,279185–200,251–9,267–8,283;seealsoDevitt,M.123nn.17–18,124n.31,260,contextualism;indexicals281contextualism12,18,100n.4,140,142–3,discourserepresentationtheory85,280155,162,193,198,255–8,267–8,283–4disjunctivism13epistemological256–7,268displacement33–6,67–8radical86,177–8dispositionalaccountofmeaning,thevs.invariantism172–3203–4seealsocontextdependencedisquotation101n.20,224;seealsotruth;convention1,13,18,101n.15,117,187,theoriesof;deflationary189–94,196,199–200,254,268–70,272;Divers,J.80n.2seealsoimplicatures,conventionalDodd,D.170conventionalism13,199;seealsoDonnellan,K.7,12,109,112,119–20,empiricism,logical122n.6,122nn.11–12,123n.22,124n.conversationalmaxims55,192,253–4,27,281268–9,272doxasticdeliberation212–19Cooper,R.81nn.16–17Ducrot,O.191correlationfunctions144–5Dummett,M.44n.10,92–3,96–9,103nn.creativity256,26931–2,110,122n.7,260,281Cresswell,M.280Crimmins,M.142–3Egan,A.81n.24,257–8Einheuser,I.258Daniels,N.21empiricism,Darwall,S.223–4logical13,185,229,234–5,238,246,270,Davidson,D.16,80n.3,80n.14,84–9,284–593–6,99,100nn.2–3,101n.12,101n.radical235–9,270,28518,102nn.23–5,103n.28,129,135,Enderton,H.81n.23261,280–2epithets41,45n.11,273Davies,M.16euphemisms41,45n.11deBlanc,M.124n.30,124n.32Evans,G.11,13,121n.4,123n.17,258,dedictovs.derebeliefs50,76–80,81n.25281definitedescriptions7–13,23n.8,66,77,experimentalphilosophy21,120–1,124n.80,81n.16,106,26930,260,2712879781472578235_txt_print.indd28708/08/201415:46\nIndexfalsificationism98Harman,G.103n.33,221Fara,D.G.258Harnish,R.157,199felicityconditions:seesatisfaction,Harris,Z.27–8contextualHattiangadi,A.208,225n.6Fillmore,C.284Hawthorne,J.138,165,257,260Fodor,J.167,225n.9Heim,I.16,196,259,280formallanguages3,28–9,32,43,84–5,87,Hempel,C.234185,271,274,277–8Higginbotham,J.44n.3,44n.9,45n.12,formalsemantics,seesemantics129,259Foster,J.102n.27Hintikka,J.81n.10Francén,R.257Hitchcock,C.257Frege,G.1–4,7,12,16–17,22n.3,28,Hobbes,T.186–7,96,101n.12,107,121–2nn.Horwich,P.933–4,133–5,185,236,238,246–7,Humberstone,L.16249,260,271–2,276,278,282;seeHume,D.204,273alsoFregeanism;Frege’spuzzle;Hunter,D.122n.8neo-FregeanismFregeanism12,87,131,133–7,139,141–2,Iatridou,S.81n.7146,272;seealsodescriptivism;Ichikawa,J.21neo-Fregeanismillocutionaryacts17,22n.3,186–7,189,Frege’spuzzle131–47,282199,272;seealsoassertionsfulfilmentconditions17imperativesentences17,189,194implicatures,García-Carpintero,M.123n.19,260conventional189,192,261,273García-Ramírez,E.22conversational12,140,186,188–92,195,Gardiner,A.185197–8,253–5,268–9,271,272–3Gauker,C.166generalized197–8,272Gazdar,G.196,198seealsopragmaticimplicationsgeneralterms9,23n.10,123n.20,277;seeimplicitknowledge48,51,98,103n.31,alsonaturalkindterms115generalizedquantifiertheory8,81n.16,indexicals7,9–13,15,17,23n.12,55,27688–9,106,116–17,123n.22,131,136,Gettier,E.19,21140–1,146,157–8,189–95,256–8,273,Gibbard,A.223–4282–4Glanzberg,M.175–6,283automaticvs.discretionary159–61,Gleitman,L.45n.14166–7,169–70,176,283Glüer,K.102n.24,284hidden140,147,157,161–5,170,256,Goodman,N.4258grammars28–32,42–3,200seealsodemonstratives;semantics,Greenberg,M.103n.33doubleindexGrice,H.P.12,18,90–1,101n.9,101nn.inferentialmodelofcommunication,the12–13,101nn.15–16,140,157,185–9,199–200191–3,197–9,225n.8,253–4,268,272,inferentialroles98–9,273281,284intensionalcontexts55–7,76–7,79,81n.Guasti,M.T.44n.1,27910,81n.26,89,91,125–47,282;seeGuéron,J.26,31,33,280alsobeliefreportsintuitions6,10,12,18–22,120–1,124n.30,Haegeman,L.26,31,33,280131,138–41,144–7,173,177–8,197–8,Hansen,N.260224,229,241–7,249,254,260,2712889781472578235_txt_print.indd28808/08/201415:46\nIndexirony88,154,253,273literalvs.non-literalmeaning41,85–6,is-oughtdistinction,the204,273–4;see88,101n.9,154,170,188,191,197–8,alsoopen-questionarguments200,253,259,273,284;seealsoirony;metaphorJackson,F.16,226n.21Locke,J.1Jacobson,P.81n.26locutionaryacts17,187Jarvis,B.21logicalform(LF)67–71,74,89,143,255–6,Jeshion,R.22,122n.10,139,260278logicalpositivism;seeempiricism;logicalKadmon,N.284Ludlow,P.143Kamp,H.81n.26,196,251,258,280Ludwig,K.103n.30,261Kant,I.23n.4,228–9,231–4,236,238–42,245–50,285MacFarlane,J.81n.24,175,257–8Kaplan,D.7,15–16,81n.9,88,116,123n.Machery,E.120–1,124nn.29–30,124n.21,141,158,160,166–8,190–1,194–5,32,260,282251–4,256,261,282–3Malinowski,B.185,198Karttunen,L.195–6Mallon,R.120,124n.29Katz,J.260Manley,D.260Keenan,E.81n.16Marcus,R.B.113,118Kennedy,C.171,259,283Margalit,A.45n.11King,J.252Martí,G.124n.32,282Kitcher,P.230–1,239,285Marty,A.185Kölbel,M.81n.24,102n.23,102–3n.28,massterms89,273176,257–8May,R.81n.19Kratzer,A.16,81nn.6–8,259,280McDowell,J.13,97,99,121n.4,204Kripke,S.5–7,9–16,18–19,21,23n.8,80n.McGinn,C.225n.105,81n.25,109–13,120–1,122nn.6–8,McKay,T.140122nn.11–12,132,137,203–4,219,221,Merchant,J.44n.4229–30,239–45,260,276,281–2,284–5metaphor45n.11,88,154,269,273Kroon,F.123n.19methodology21–2,69,79–80,154,251,254,259–60,271,279Ladusaw,W.81n.18Mill,J.S.113,234,236–7,277,284–5Lahav,R.255missingcomplements163–4,177Lam,B.124n.30modalbases54–5languageacquisition31,41,43–4,51,modalillusions14–15,242–497–9,274,279modalknowledge2–7,13–16,228,232–7,Larson,R.16,259,280239–50,259,285Lasersohn,P.173,175–6,257modifiers27–8,34–5,38–9,44n.5,44n.7Lepore,E.100n.4,101n.10,103n.30,129,modulation88,101n.9,255–6163,165,167,194,254,256,261Montague,R.16,81nn.11–12,185,280levelsofmeaning186,188,190–2Moore,G.E.188,219,226n.17,274,279Levinson,S.198Morris,C.193Lewis,D.23n.12,80n.2,81n.12,81n.15,Moss,S.258101n.15,130,166,194,197,199,257,259names7,9–13,18–19,23n.8,57–60,62,linguisticcompetence7,16–17,20,31,37,71–2,106–24,128,131–47,240–3,260,41,43,45n.11,51,86–7,91–2,96–9,269,276,279,281–2103n.31,114,259;seealsocreativityco-referring3,10–11,14–15,108,125–8,linguisticconstraint,theprincipleof143131,241–5,282;seealsoFrege’spuzzle2899781472578235_txt_print.indd28908/08/201415:46\nIndexempty10–13,118–20,124n.25,124n.28,perlocutionaryacts187,199,272266,270,281–2Perry,J.7,23n.12,108,122n.5,123–4n.seealsodescriptivism;reference,the22,130,135,141–4,159,161,255,causaltheoryof;rigiddesignators282–3narrowvs.broadcontext156–7,159,perspectivalterms165,177166–7Pettit,P.122n.10naturalkindterms1,13,221–3,226n.20picturetheoryofintentionality,the3–7,naturalisticfallacy,the204,273–4;seealso13,22–3n.3,23n.5open-questionargumentsphonetics47–8,254Neale,S.7–8,158,168,283phonology32,47–8,67necessarytruths3,5–6,13–16,19–22,phrasestructures29–3340–1,53,126–7,130,206,228–49,268,Pietroski,P.259274–5,285Pinker,S.26aposteriori14–16,206–7,221–2,229,Plantinga,A.122n.9,229241–4,285Plato1negativepolarityitems65,81n.18Port-Royallogicians,the1Nelson,M.140,145Potts,C.189,261neo-Fregeanism13,121n.4,124n.26pragmaticimplications139–42,186–92,Nichols,S.120,124n.29196;seealsoimplicaturesNinan,D.258pragmatics12,17–18,48,55,72,86,100n.nominals27–9,36,44n.74,100–1n.9,139–42,156–9,167–8,non-truth-conditionalaspectsofmeaning173,185–200,254,271,283–4186–7,189–94,196,261Predelli,S.194,255–6Noonan,H.260predicatesofpersonaltaste175–7,283prescriptiveprinciples204–12;seealsoOlivola,C.124n.30,124n.32truthprescription,theopen-questionarguments219–24presuppositions9,12,186,192,195–7,opentexture192–3261,271,275,284;seealsocommonoperationaldefinitions222–3,226n.19groundoperators,Price,H.257lambda-extract60,81n.13pronouns27,36,39,135–7,154,158–9,modal30,53–5,81n.8,89,110,126,128,166–7,191,278,283146,270,273propernames;seenamesnon-extensional72propositionalattitudeverbs81n.25,propositionalattitude89128,131–6,142–3,147,282;seealsotense89,146,270operators,propositionalattitudetruth-functional125Putnam,H.7,44n.10,123n.20,221–2,ordinarylanguagephilosophy185–7,226n.20,230–1,239192–3,198quantifiednounphrases62–80,81nn.Pagin,P.99,100n.4,100n.7,101n.14,15–16,81nn.25–6102n.24,103n.32,255–6,279,281monotoneincreasing/decreasing64–5,Parikh,R.80n.481n.17Partee,B.259quantifierraising(QR)67–9,71,74,79,passive,the35–681n.19Patterson,D.93,102n.21Quine,W.V.O.13,20,80n.1,95,103n.Peacocke,C.16,28529,124n.33,126,129,229,237–9,246,Pelletier,J.100n.4,255249,270,281–2,285Peregrin,J.103n.33quotation126–7;seealsodisquotation2909781472578235_txt_print.indd29008/08/201415:46\nIndexradicalinterpretation95,102n.23satisfaction72,87,100n.8,101n.15,Raffman,D.258125–6,205–6Railton,P.223–4contextual187,191–3,196rationalism229,231,241,276,285ofpredicates123n.20,207–8,220–1,Recanati,F.100n.4,101nn.9–10,143,224–5,276162,177,189,192–4,198,252,254–6,ofsingularterms9,76,106,110,113,258–60,273,284119,121n.4,122n.12,131,138,269Reddam,P.140Saul,J.142reference,Schaffer,J.257thecausal(‘new’)theoryof7,11–12,Schiffer,S.101n.13,101nn.15–16,142–3,106–24,260,266,281–2;seealso161,166–7,199reference,direct;referentialism;rigidSchlenker,P.196,261designatorsSearle,J.17,100n.4,121n.4,177,186–7,thedescriptivisttheoryof,see193,284descriptivismSegal,G.16,259,280direct10–11,19,112–13,117–19,123n.semanticholism95,97,99,28120,124n.25,127–8,131,133,138,190,semanticincompleteness157–8,162–5,260,269;seealsoreference,thecausal171,174,176–8,254,283theoryof;referentialismsemanticinnocence,theprincipleof135failure;seenames,emptysemanticminimalism100n.4,256ofpredicates23n.10,39,125,207–11,semanticnaturalism84,204–26,284220–4semanticnormativity18,23n.8,203–26,seealsodefinitedescriptions,reference-276,284fixing;definitedescriptions,semantics,referentialuseof;demonstratives;compositional18,37–40,73,87,91,indexicals;names;pronouns;100n.6,101n.18,253–5,271–2,276,singularterms281;seealsocompositionality,thereferentialism131,139–47;seealsoprincipleofreference,directcontextualist;seecontextualismReimer,M.124n.25,142,168–9,282doubleindex251–2,269–70Reinach,A.185dynamic196,261;seealsodiscourserelationalterms164–5,176–7representationtheoryrepresentationaltheoryofmind,theexperimental;seeexperimental209–11philosophyReyle,U.280foundationalvs.descriptive84,Richard,M.140,142–6,28299–100n.1,272,281Rieber,S.142game-theoretic85rigiddesignators5,10,13–14,18–19,intention-based167–823n.8,109–12,242,276,282;seealsopossibleworlds23n.12,50–63,71–6,descriptions,rigidified80n.5,80n.10,85,89,101n.10,102n.Rumfitt,I.101n.1926,130,275Russell,B.1–3,7–12,16,22n.3,23n.9,proof-theoretic85,96,98,102n.2323n.11,81n.16,107,117,121nn.1–2,relativist176,258,283;seealsotruth,134,138,143,185,269,271,279,282–3theoriesof,relativistRussellianannotatedmatrices(RAMs)situation85,130143–4Tarskian71–2,278truth-conditional4–5,9–10,15–18,39,Sainsbury,M.13,23n.9,23n.11,260,27985–106,109,111–12,116–18,136–7,Salmon,N.140,283153,157,185–6,190–4,196,198,259,2919781472578235_txt_print.indd29108/08/201415:46\nIndex267,278;seealsosemantics,possiblestructuredpropositions128,130,146,158,worlds275two-dimensional13–16,244–5,279Sweeney,P.170type-theoretic57–63,66,71–5,81n.12,Szabó,Z.254–6,25981n.15vs.pragmatics48,100–1n.9,185,189,tabooterms41–2193–4,197;seealsopragmaticsTarski,A.81n.23,87–8,185;seealsosensedata11,138semantics,Tarskian;truth,theoriessenses,Fregean12,107,121–2n.4,124n.of,Tarskian26,133–6,186,192–3,240Taylor,B.101n.16,sententialismvs.propositionalism128–9,Taylor,K.162282tense30–1,36,38,159,169,176,252,258,Shah,N.211–16,225n.13268;seealsooperators,tenseShannon,C.199Torre,S.258Shapiro,S.258traces35–6,67–8,71,73,75,81nn.20–1Shatz,M.22transformations31,33;seealsoSider,T.144displacementsingularclaims/propositions/thoughtsTravis,C.18,100n.4,177,193,198,2559–10,128,130–1,133,139–44,260truth,theoriesof,singularterms9,13,23n.10,91,106,116,correspondence93126,132,134,136,225n.4,260,271,deflationary92–3,102n.23277;seealsodefinitedescriptions;epistemic92,103n.32demonstratives;indexicals;names;pragmatist211rigiddesignatorsrelativist72,81n.24,176,225n.12,258Soames,S.20,102n.27,122n.8,122n.10,Tarskian(T-theories)85,87–9,93–5,97,130,140,144,158,176,221–2,258,99,102n.23,102–3n.28,278,280260,279,282–3truthprescription,the210–19,224Sosa,D.122n.8Tye,M.140Sosa,E.19,260speechacts4,17–18,40,48,88,90–1,unarticulatedconstituents142–3,147,100–1n.9,156–7,168,186–7,189,192,161–3,165,255195,199,256,267,269,271–2,276–7,Unger,P.172–3283;seealsoassertions;illocutionaryacts;locutionaryacts;perlocutionaryvagueness32,154,172–3,258,266,268,acts278Sperber,D.177,199,284Velleman,J.D.211,213–16,225–6n.14Stainton,R.158Vergnaud,J.R.36Stalnaker,R.16,80nn.4–5,99n.1,130,verificationism92,98–9,100n.2,238;see156–7,194–6,258,281,284alsosemantics,proof-theoreticStanley,J.143,162,194,254–60,280vonFintel,K.81n.7,261Stenius,E.189Stephenson,T.176,257Waismann,F.192Sterelny,K.123n.17,281weatherreports161–3,176Stern,J.45n.11Weaver,W.199Stich,S.120,124n.29Westerståhl,D.81n.16,100n.7,256,279,281Stokke,A.167Wettstein,H.122n.5Strawson,P.11,18,89–92,94,101n.12,Whiting,D.225n.6137–9,168,185,187,190,198–9Wikforss,Å.225n.7,284Williams,M.92,102n.232929781472578235_txt_print.indd29208/08/201415:46\nIndexWilliamson,T.20–2,100n.2,279Woodward,J.257Wilson,D.177,199,284Wright,C.92,176,204,258Wittgenstein,L.1–7,10–11,13,16–18,22–3nn.3–5,23n.12,86,92,96,100n.Yalcin,S.81n.7,2575,101n.12,185,192,203wh-phrases33–5,67–8Zalabardo,J.211,225n.2Whitehead,A.2Ziff,P.121n.42939781472578235_txt_print.indd29308/08/201415:46
