【语言哲学】语言学家的哲学 205页

\nPhilosophyforLinguists…amuchneededresource,boilingdownthedauntingmassofphilosophicalwritingsonlanguagetoaclearandreadableintroductiontowhosaidwhatandwhy.CatrionaMcPherson,UniversityofLeedsPhilosophyforLinguistsprovidesstudentswithaclear,conciseintroductiontothemaintopicsinthephilosophyoflanguage.Focusingonwhatlinguistsneedtoknowandhowphilosophyrelatestomodernlinguistics,thebookisstructuredaroundkeybranchesoflinguistics,includingsemantics,pragmaticsandlanguageacquisition.Assumingnopriorknowledgeofphilosophy,SiobhanChapmantracesthehistoryanddevelopmentofideasinthephilosophyoflanguage,andoutlinesthecontributionsofspecificphilosophers.Thebookishighlyaccessibleandstudent-oriented,andincludes:•ageneralintroductionandintroductionstoeachchapter•numerousexamplesandquotations•comprehensivesuggestionsforfurtherreading•anextensiveglossaryofphilosophicaltermsSiobhanChapmanisalecturerintheDepartmentofEnglishLanguageandLiteratureattheUniversityofLiverpool.\n\nPhilosophyforLinguistsAnintroductionSiobhanChapmanLondonandNewYork\nFirstpublished2000byRoutledge11NewFetterLane,LondonEC4P4EESimultaneouslypublishedintheUSAandCanadabyRoutledge29West35thStreet,NewYork,NY10001RoutledgeisanimprintoftheTaylor&FrancisGroupThiseditionpublishedintheTaylor&Francise-Library,2003.©2000SiobhanChapmanAllrightsreserved.Nopartofthisbookmaybereprintedorreproducedorutilisedinanyformorbyanyelectronic,mechanical,orothermeans,nowknownorhereafterinvented,includingphotocopyingandrecording,orinanyinformationstorageorretrievalsystem,withoutpermissioninwritingfromthepublishers.BritishLibraryCataloguinginPublicationDataAcataloguerecordforthisbookisavailablefromtheBritishLibraryLibraryofCongressCataloginginPublicationDataChapman,Siobhan,1968–Philosophyforlinguists:anintroduction/SiobhanChapmanIncludesbibliographicalreferencesandindex.1.Languageandlanguages—Philosophy.I.Title.P106.C5132000401–dc2199–462292ISBN0-203-45949-0Mastere-bookISBNISBN0-203-76773-X(AdobeeReaderFormat)ISBN0-415-20658-8(hbk)ISBN0-415-20659-6(pbk)\nForChris\n\nContentsAcknowledgementsxiIntroduction1Thephilosophy2Thelinguistics3Overview5Furtherreading71Wordsandthings8Introduction8Classicalphilosophy14TheBritishEmpiricists17Names22Useandmention23Millondirectreference25Fregeonsenseandreference27Russellondescriptions31Kripkeonnaming34Wordsandthingsinlinguistics36Furtherreading392Propositionsandlogic41Introduction41Propositions43Formallogic46Propositionallogic47Predicatelogic53Entailmentandpresupposition55Fregeonlogicalpresupposition57Russellondenoting58vii\nviiiContentsStrawsononreferring60Pragmaticpresupposition63Logicandlinguistics67Furtherreading703Truthandreality72Introduction72Truthconditions75Analyticandsyntheticsentences76Verification80Truththeories85Possibleworlds91Modality93Counterfactuals97MontagueGrammar99Linguisticsemantics101Furtherreading1034Speakersandhearers106Introduction106Wittgenstein’sphilosophyoflanguage110Ordinarylanguagephilosophy114Speechacts116Performatives117Illocutionaryforce120Searleonspeechacts121Meaningandintention123Conversationalimplicature126Naturallanguageandlogic126Theco-operativeprinciple130Floutingthemaxims135Pragmatics137Furtherreading1415Languageandmind144Introduction144Signsandstructures146Linguisticdeterminism148Empiricism151Bloomfield152Quine154\nContentsixTheInnatenessHypothesis158Anti-nativism161Thelanguagefaculty163Knowledgeanduseoflanguage165Empiricistandmentalistlinguistics167Furtherreading170Glossary170Notes173References184Index190\n\nAcknowledgementsTheideaforthisbookarosewhenIwasteachinglinguisticsandEnglishlanguageattheUniversityofKentatCanterbury.SomeofmystudentswereunderstandablydismayedbytheoccasionalreferencestophilosophyIwasmaking,andaskedifIcouldsuggestsomethingwhichwouldexplaintheseinmoredetail.WhenIfoundthatIcouldn’treallyrecommendabooktosuittheirexactneeds,IdecidedIwouldjusthavetowriteonemyself.IamthereforegratefultomystudentsatKentfor,ineffect,comingupwiththeidea,andalsotoTonyBex,mycolleagueatthetime,forconvincingmethatitwasagoodone,andencouragingmetogetitpublished.IwouldalsoliketothankthelinguisticsteamatRoutledge,whohaveseenthisprojectthroughfrominitialideatofinishedproductwithefficiency,cheerfulnessand,wherenecessary,patience.Finallybutmostimportantly,mythankstoChristopherRoutledgeandCatrionaMcPherson,whoreadthefirstandseconddrafts,respectively,andcommentedonthemindetail.Theirsuggestions,bothspecificandmoregeneral,wereinvariablyinsightful.Allthefaultsandanyerrorswhichremainare,ofcourse,myown,butwithouttheirhelptherewouldundoubtedlyhavebeenmanymore.xi\n\nIntroductionAnyonewhostudieslinguisticswill,soonerorlater,comeacrossreferencestophilosophy.Thesemaybetogeneralphilosophicalthemesandconcepts,ortotheideasofindividualphilosophersandschoolsofthought.Eitherway,studentsoflinguisticswillprobablyfindthemselveslookingforsomeguidanceinunderstandinghowthephilosophyreferredtorelatestotheirinterestinlanguage.Buteveniftheyknowwheretostart,theyarelikelytoendupgrapplingwithcomplexphilosophicalworks,orwithintroductionstothemwhichdon’tmakeanymentionoflinguistics.Thisbookisintendedforjustthesepeople.Itsaimistointroducesometopicsfromthehistoryofphilosophy,andtoexplaintheirrelevanceforpresent-daylinguists.It’snotsurprisingthatthereissomuchphilosophyinlinguistics.Linguisticsitselfisafairlyrecentacademicdiscipline,butmuchofourcurrentthinkingaboutlanguagehasdevelopedfromideaswhichdatefromthedecadesandcenturiesbeforeitcameintobeing.Manyofthesewereoriginallyputforwardbyphilosophers.Sosomeareasofthestudyoflanguagecanbeseen,atleastinpart,asoriginatingwithinphilosophyandonlylaterbecomingafocusofthenewdisciplineoflinguistics.Andallareashavebenefitedfromdiscussionsofthenatureoflanguage,andparticularlythenatureofmeaning,whicharetobefoundinphilosophy.Aswewillsee,thesediscussionsarerelevanteventothosewhoseethemselvesasprimarilyconcernedwithlanguageasameansofcommunication,oratypeofhumanbehaviour,andmaythereforebetemptedtoquestiontheimportanceofdetailedphilosophicalanalysistotheirinterests.Throughoutthisbook,nopriorknowledgeofphilosophyisassumed;philosophersandtheirideasareintroducedfromscratchastheyareneeded.However,becausethebookisprimarilyintendedforthosefollowingspecialistcoursesinlanguage,orthosewhoalreadyhaveageneralinterestinthesubject,thereisn’tmuchdetailedexplanationoftheideasandtheoriesofpresent-daylinguistics.Someofthesearementionedlaterinthisintroduction,andcanbefollowedupinthebooksrecommendedinthe‘furtherreading’sectionattheendofit.Butwewillbeginbylookingatthebranchofphilosophywewillbeconcernedwith,andthereasonswhywewillchieflybelimitingourselvestothisarea.1\n2IntroductionThephilosophyPhilosophyhasbeenpractised,invariousforms,forthousandsofyearsandinallpartsoftheworld.Inthisbook,however,wewillbeconcentratingchieflyonworkfromtwentieth-centuryBritainandAmerica.Thisiscertainlynotintendedtosuggestthattheworkswewillbelookingataretheonlyonesofinteresttolinguists,orthathighlyimportantworkonlanguagehasn’tbeenproducedatothertimesandinotherplaces.Buttwentieth-centuryphilosophyinEnglish,particularlyofthetraditionwhichhasbecomeknownasanalyticphilosophy,hashadthemostobviousanddirectinfluenceonthedevelopmentoflinguisticsasanacademicdiscipline.Thisisperhapssomeexcuseforsuchanarrowfocus;thereferencestophilosophywhichstudentsoflinguisticscomeacrossaremostlikelytobetoanalyticphilosophy.Twentieth-centuryphilosophydidn’t,ofcourse,emergefromavacuum,andwewillbelookingatsomeoftheworkwhichinfluenceditsdevelopment,particularlyworkfromAncientGreece,andfromseventeenth-andeighteenth-centuryEurope.Butwewon’tbelookingatimportantworkonlanguagefromotherperiods,suchasmedievalphilosophy,orfromnon-Westerntraditions,suchasIndianandChinesephilosophies.Analyticphilosophy,asitsnamesuggests,isanapproachtothesubjectsandproblemsofphilosophywhichreliesondetailedanalysis.Inparticular,complexnotionsandproblemsareanalysed,or‘brokendown’intosmallercomponentssothattheycanbebetterunderstood,orperhapsevenshownnottobeproblemsatall.Earlyanalyticphilosophywasconcernedwiththelogicalrelationshipsbetweenindividualconceptsandpropositions.ItcanarguablybeseenasoriginatingintheworkoftheGermanphilosopherGotlobFregeattheendofthenineteenthcentury,andbeingexemplifiedinEnglandbytheworkofphilosopherssuchasBertrandRussell.Duringtheearlytwentiethcentury,thisbranchofphilosophytookwhatissometimesdescribedasa‘linguisticturn’,theemphasisswitchingtothelanguageinwhichphilosophywasexpressed,andhencetolanguageitselfasalegitimatetopicofseriousphilosophicalinvestigation.Thistypeofanalyticphilosophyalsobecameknownas‘linguisticphilosophy’,atermwhichisappliedinparticulartotheworkofthephilosophersoftheBritishschoolof‘ordinarylanguagephilosophy’suchasPeterStrawsonandJohnAustin.Wewillbelookingatsomeoftheworkofthephilosophersmentionedinthisparagraph,andmanyothers,inthecourseofthisbook.Theothermajordevelopmentintwentiethcenturyphilosophyinwhichlanguagehasplayedacentralpart,andonewhichisoftencontrastedwithanalyticphilosophy,iscontinentalphilosophy.ThistermisappliedtoatypeofphilosophyproducedincontinentalEurope,andiscloselyassociatedwithnamessuchasFoucault,LyotardandDerrida.Thephilosophicalmethodofthisschool,whichisoftenreflectedinthestyleof\nIntroduction3writingused,isbasedonrhetoricandargument.Itrejectstheideaoffixed,certainmeaning,andthereforethepossibilityofreachingunderstandingbyrigorousanalysisofthelanguageinwhichphilosophicalproblemsareexpressed;thelanguageitselfisopentointerpretationandnegotiation.Thisphilosophicaltraditionhasprovedveryimportantinthedevelopmentofcriticaltheory.Continentalphilosophy,then,ishighlyrelevanttoanydiscussionofpresent-dayliterarycriticism,butlessimmediatelysotooneoflinguistics.Wewon’tbelookingatithere,butsomesuggestionsaremadeinthe‘furtherreading’sectionforthoseinterestedinthisarea.OfparticularrelevancetothedifferencebetweenanalyticalandcontinentalphilosophyisthewrittendebatebetweenJohnSearleandJacquesDerrida.Derrida’scriticismsof‘speechacttheory’,Searle’sresponsetothis,andDerrida’sreply,showupthedifferencesbetweentheirapproachestolanguageandtheirstylesofphilosophy.Derrida’srejectionofthepossibilityofreasonedanalysisandrationaldiscussionisreflectedinhisintricate,playfulstyleofargument.Itisn’thardtofindpointsofsimilaritybetweenanalyticphilosophyandpresent-daylinguistics.Bothdisciplinescouldbesaidtoplaceahighvalueonlogicalargument,onattentiontodetail,andonthecarefulanalysisofcomplexwholesintotheirconstituentparts.Thegrowingacceptanceoflanguageasalegitimatefocusofphilosophicalstudy,forwhichtheanalyticschoolwaslargelyresponsible,canbeseenasoneofthefactorswhichmadelinguisticsasanacademicsubjectpossible.ThelinguisticsUnlikemanyphilosophersoverthecenturies,linguiststakeitforgrantedthatnaturallanguageisaninterestingandimportantsubjecttostudyinitsownright.Thistermisgenerallyusedtodistinguishnaturallyoccurringhumanlanguagessuchas,forinstance,English,PolishandUrdu,fromartificiallanguages,suchasthoseusedinlogic,incomputingandinvariousformsofcoding.Languagesfromthesetwocategoriesmaywellhavealotincommon,butnaturallanguagesare,nevertheless,seenasbelongingtoaseparateanddefinableclass.Itisthetaskoflinguiststodescribeandanalysethefeaturesofthisclass.Almostanygeneralisationaboutnaturallanguagewillbecontroversial,becausesomanydifferentideasandopposingtheoriesarecurrentinlinguistics.Insofarasweneedtomakeassumptionsaboutnaturallanguage,andaboutthenatureoflinguistics,wewillbeusingthoseoutlinedverybrieflyinthenextthreeparagraphs.However,thisbookisn’tintendedtorelatephilosophicalideastoanyoneparticulartypeoflinguistics,andwe’llbeconsideringsomeothercurrentwaysofdescribinglanguageinthefinalsectionofeachchapter.\n4IntroductionOnoneview,then,thedescriptionofanaturallanguageisdividedintothreedifferentcategories:thephonological,thesemanticandthesyntactic.Thephonologicaldescriptionofalanguageconcernsthewayinwhichitssoundsarepatternedandrelated.Semanticsconcernsthemeaningsoftheindividualwordscontainedinalanguage,andalsothemeaningswhichresultwhenthesewordsarecombinedinvariousways.Syntaxisthestudyofthesewaysofcombiningwords:theregularitieswhichcanbeobservedintheconstructionofsentences.Thesethreebranchesoflinguisticdescriptionareseenastotallyseparateandindependentfromeachother,buttogethertheymakeupthegrammarofthelanguage.Agrammarcanbedescribedaswhatitisthatpeopleknowwhentheycanbesaidto‘know’alanguage,andiswhatlinguistsareattemptingtomodelindescribingalanguage.Grammar,inthisversionoflinguistics,issaidtobegenerative.Thatis,itconsistsofaseriesofruleswhicharesufficienttogenerate,orproduce,allthepossiblesentencesofalanguage.Inthisway,alanguagecanbedescribedasasetofsentences.Itisthesetofallthesentencesgeneratedbytherulesofthegrammar,ofallthegrammaticalsentences.Thesetisinfinite;youcouldneverproduceadefinitivelistofallthepossiblesentencesof,say,English.However,itisproduced,orgenerated,byafinitenumberofruleswhichitis,inprincipleatleast,possibletolist.Therearetwotypesofrulesinthegrammarofanynaturallanguage:generativerulesandtransformationalrules.Thegenerativerulesproduceaseriesofdeepstructures.Itisatthelevelofdeepstructurethatsemantics,orlogicalform,isdetermined.Thetransformationalrulesactonthedeepstructurestoproducesurfacestructures;atthislevelthedetailsofstructureandwordorderaredetermined.ThedescriptionoflinguisticsoutlinedaboveowesalottotheAmericanlinguistNoamChomsky,whoseworkwewillconsiderinChapter5.Asweshallsee,Chomskydevelopedhisinfluentialideasonlanguagefromthelate1950sonwards.However,someofthebasicsofpresent-daylinguistics,andsomewhichwouldbeacceptedevenbythosewhowouldargueagainsta‘Chomskyan’approachtothestudyoflanguage,datebackearlierinthecentury.Forinstance,in1938thephilosopherCharlesMorrissuggestedthatthestudyoflanguagecouldbedividedintoanumberofdistinctbranches.Particularlysignificantly,hearguedthatthestudyofsyntaxmustbecompletelyseparatefromsemantics,andthatitmustconsistof‘formationrules’and‘transformationrules’.Indeed,theideathatweneedtodistinguishbetweenthe‘underlying’,logicalformofasentence,whichischieflyresponsiblefordeterminingmeaning,andits‘surface’,grammaticalform,canbetracedbackevenfurther.Aswewillsee,relatedideascanbefoundintheworkofthephilosophersBertrandRussellandLudwigWittgenstein.Inthisbook,weshallbelookingatsomeofthewaysinwhichphilosophicalthinkinghascontributedtocurrentideasaboutthegrammarofnaturallanguage.Wewillbeconcentratingonsyntaxandsemantics;we\nIntroduction5won’tbelookingdirectlyattheinterfacebetweenphilosophyandphonology.Ofcourse,notalllinguistsareengagedinthedetailedstudyofthesyntax,thesemanticsorthephonologyofnaturallanguage.Thesearethethree‘core’areasoflinguisticdescription;theytellusalotaboutlanguage,butnothingatallabouthowpeopleactuallyuselanguage,howthecontextofuseaffectswhattheysayandmean,orwhatconsequencestheiruseofitcanhave.Theseissuesarestudiedinbranchesoflinguisticssuchaspragmatics,discourseanalysisandsociolinguistics.Wewillbejustasinterestedinthecontributionmadebyphilosophytothinkingintheseareasaswewillinitscontributionto‘core’areasoflinguisticstudy.OverviewEvenwithintherestrictionsoutlinedabove,therangeofphilosophicalwritingsrelevanttolinguisticsisvast.Whatwewillbestudyinginthisbookisbynecessityonlyaselectionofthese,intermsbothofsubjectmatterandofindividualphilosophers.Eachofthefivechaptersisconcernedwithone,necessarilybroad,topicfromphilosophy.Thephilosopherswhoseworkisusedarechosenbecausetheyhavebeenparticularlysignificantinthedevelopmentofthinkinginthisarea,orcanbeseenasrepresentativeofideaswhichareofcurrentrelevancetolinguistics.Inevitably,muchisleftout,butthedetailedstudyofphilosophicaltopics,andevenoftheworkofindividualphilosophers,isbeyondthescopeofthisbook.Eachchapterfollowsthesamegeneralformat.Intheintroductorysectionsomeproblemsorgeneralissuesrelatingtolanguagearepresented.Theseareproblemswhicharerelevanttolinguists,inthattheyareconcernedwithnaturallanguageasafocusofinterestinitsownright.Thebulkofeachchapterwillbeconcernedwiththeideasofphilosopherswhohavediscussedtheseissues,andwherenecessarywithacomparisonbetweendifferent,sometimescompeting,ideasonthesametopic.Eachchapterconcludeswithanassessmentofwhatthesehavecontributedtoourunderstandingoftheissuesidentifiedattheoutset,andtheirinfluence,directorindirect,onvariousbranchesofpresent-daylinguistics.Thereisalsoa‘furtherreading’sectionineachchapter,whichsuggestsdirectionsinwhichthemaintopicscanbeexploredfurther.Thisincludesbothsomeofthemorereadableoftheworkscovered,andalsointroductoryormoreadvanceddiscussionsofthem.Ifthereisageneralthemewhichbringstogetherallthetopicswewillbeconsidering,itisthenatureofmeaning.Meaningisofcentralimportancetoallbranchesofphilosophyconcernedwiththestudyoflanguage.Inthiswecanincludenotjusttheworkofthosephilosopherswhocanbelabelledasinterestedin‘thephilosophyoflanguage’,butofallthosewhohaveworkedinthephilosophyofsubjectssuchasmathematics,logic,knowledgeandthought,whohavehadcausetothinkaboutthelanguagein\n6Introductionwhichtheirideasareexpressed.Thequestionswhichhaveconfronted,andcontinuetoconfrontsuchphilosophers,arenotgenerally,ornotmoststrikingly,ofthe‘whatdoesthiswordmean?’type.Rather,theyhaveconsideredwhatitistosaythatonething‘means’anotherand,particularly,tosaythatsomepartofalanguagesystem‘means’somethingwhichisoutsideofthatlanguagesystem.Thisrelationshipbetweenthelinguisticandthenon-linguisticiscentraltoanyattempttodescribelanguage.Webegin,inChapter1,byconsideringwhathasbeensaidaboutthemostbasicaspectofthisrelationship:therelationshipbetweenwordsandthings.Wewilllookatwhatphilosophersfromclassicaltimesonwardshaveclaimedaboutwhatitisforawordto‘mean’somethingorconcept,andconsidertheinfluencetheirideashavehadonthepresent-daystudyoflexicalsemantics.InChapter2webroadenourstudytoincludethemeaningnotjustofindividualwordsbutofsentences,consideringthepropositionswhichtheyexpress.Wewillalsoconsiderthesignificanceofbasiclogic,includinglogicalrelationshipssuchasentailmentandpresupposition,inlinguistics.Chapter3isconcernednotsomuchwiththerelationshipswhichexistbetweenexpressionsofalanguage,buttherelationshipsbetweenthoseexpressionsandreality,andinparticulartherelationshipbetweenmeaningandtruth.Wewillbeconsideringanumberofwayswhichhavebeenproposedbyphilosophers,andsubsequentlybylinguists,toaccountforwhatitmeanstoattachthelabels‘true’and‘false’tosentences,orthepropositionstheyexpress.Chapter4broadensthediscussionstillfurthertoconsidernotjustlanguage,butthecontextsinwhichitisused.Wewillseethatmakingstatementsoffactwhichcanbelabelledeither‘true’or‘false’isjustoneofthemanytasksinwhichspeakersuselanguage,andthatafulllinguisticaccountthereforeneedstodorathermorethansimplyassigntruth-valuestopropositions.Wewillalsoseethatlogicalrelationshipssuchasentailmentsimplyaren’tadequateforthepurposeofexplainingthemanycomplicatedwaysinwhichpeopleuseandunderstandlanguage.Thereareotherrelationships,suchasvarioustypesof‘implicature’,whichoftenseemtobeatoddswithlogic,butwhichdescribemoreaccuratelywhatgoesoninconversation.Itisinthischapterthatwewilllookatthebranchofanalyticphilosophyknownasordinarylanguagephilosophy,andconsiderhowithascontributedtothedevelopmentofpragmatics.Finally,Chapter5isconcernedwiththerelationshipbetweenlanguageandmind.Agreatdealhasbeenwrittenonthisbybothphilosophersandlinguists.Wewillconcentratemainlyonthedebatebetween‘behaviourist’and‘nativist’accountsoflanguage:verygenerally,thequestionofwhetherlanguageisbestdescribedasatypeofbehaviourobservablewithinacommunity,oratypeofknowledgecompleteinanindividual.Wewillconsiderhowthesetwophilosophieshavecontributedtothedevelopmentof‘empiricist’and‘mentalist’traditionsinlinguistics.In\nIntroduction7eachofthesetraditions,workhasbeenproducedwhichattemptstoexplaintheprocessesoflanguageacquisition,andmoregenerallythenatureoflanguageitself.Thebookendswithaglossaryofsomeofthekeytermsintroduced.Thisconcentratesonphilosophicalterms,whicharelikelytobeunfamiliartothosestudyinglinguistics.FurtherreadingTheintroductiontoGlock(1997)TheRiseofAnalyticPhilosophyoffersaninterestingdiscussionofthedevelopmentandpossibledefinitionsofthetradition.Glockarguesthataninterestinlanguageisn’tadefiningcharacteristic;muchcontinentalphilosophyalsousestheterminologyoflinguistics.Thebook,whichconsistsofacollectionofessaysonthesubject,isitselfagoodillustrationofthedisagreementwhichexistsabouttheappropriatedefinition,andthesignificance,ofanalyticphilosophy.Dummett(1993)TheSeasofLanguageisamuchmoreextensive,anddifficult,collectionofessaysonanalyticphilosophy,concentratingparticularlyontheworkofWittgenstein.Intheseessays,Dummettisparticularlyinterestedintherelationshipbetweenlanguageandthought.Manyintroductionstocontinentalphilosophyhavebeenwritten,oftenconcentratingonitsapplicationtoliterarycriticism.SeeforinstanceEagleton(1983)LiteraryTheory,especiallyChapter4,andSeldon(1989)PractisingTheoryandReadingLiterature,alsoChapter4.ThedebatebetweenDerridaandSearlecanbefoundinthejournalGlyphfrom1977.AllofDerrida’scontributionstothedebatearecollectedinthemoreeasilyavailable(1988)LimitedInc.Chomskysetoutwhatcanperhapsbeseenasthebasicmanifestoforthetypeoflinguisticsdiscussedhere,particularlyfortransformationalgrammar,inhisfirst,shortbookSyntacticStructures(1957);seeparticularlyChapter3.Therearemanyintroductionstomodernlinguisticsavailable,forinstanceO’Gradyetal.(1997)ContemporaryLinguistics.Rathermoredetailed,butrestrictedinscope,isSmithandWilson(1990)ModernLinguistics.Foradetailedintroductiontotransformationaltheory,seeHaegman(1994)IntroductiontoGovernmentandBindingTheory,andRadford(1988)TransformationalGrammar.TheideasofCharlesMorrisreferredtointhischapterwerepublishedinhis(1938)article‘Foundationsofthetheoryofsigns’.WewilllookattheworkofRussellandWittgensteininChapters2and3.\n1WordsandthingsIntroductionMuchofwhatgoesoninlinguisticscanbedescribedunderthegeneralheadingof‘analysis’.Linguisticanalysisinvolves,forinstance,identifyinganddescribingthestructuresandsoundsofalanguage,andattemptingtoexplaintherelationshipsbetweenthem.Butthatisn’tthewholestory.Linguisticsisadisciplinewhichconcernsitselfwiththestudyoflanguageinallitsaspectsand,aslinguists,weneedtoconsiderhowtheindividualelementsofalanguagerelatenotjusttoeachother,buttotheworldoutside.Weneedtostudynotonlytherelationshipswithinalinguisticsystem,butalsothosebetweenthissystemandthethingsitdescribes.Afterall,itisonlythroughsuchrelationshipsthatlanguagefunctionsasitdoesincommunication.Tostudyonlytheinternalstructuresandrelationshipsofalanguagemighttellusalotaboutitscomplexitiesandregularities,butitwouldsomehow‘missthepoint’;wewouldbenonearertobeingabletoexplainhowlanguageactually‘works’thanwewerewhenwestarted.Partofthetaskofexplaininghowlanguageworksistoaccountfortheobviousdifferencebetweenexamplessuchas1)and2);ingeneral,1)isacceptedtobe‘true’,while2)islabelled‘false’.1)Akangarooisananimal.2)Zebrashavewings.Anotherpartofthetaskistodescribehowspeakersuselanguageforparticularpurposes:how3)mightcountsometimesasaquestionandsometimesasarequest,andhow4),althoughlabelled‘false’,mightsometimesbeusedtocommunicatesomethingwhichistrue.3)Canyoureachthetopshelf.4)Myflatmateisamachine.TheseareissueswhichwewilladdressinChapters3and4,butfirstweneedtoestablishhowthecomponentpartsofthesesentencescontributetoour8\nWordsandthings9interpretations.Ourexamplesaremadeupofindividualwordssuchaskangaroo,zebrasandflatmate.Beforewecandiscusswhatsentences‘mean’,andwhatspeakersmight‘mean’bywhattheysay,weneedtoestablishwhatwearetalkingaboutwhenwediscussthemeaningsofthesewords.Theremightnotseemtobemuchofaproblemhere.‘Surely’,acynicalnon-linguistmightsay,‘ifyouwanttoknowwhatkangaroomeansyouonlyhavetolookitupinadictionary—thewholepointofdictionariesistogivethemeaningsofwords’.Ifwedoasthecynicsuggestsandreachforourdictionary,wewillreadthatakangaroois‘alargeAustralianherbivorousmarsupial’.Butthishasn’tgotusanyfurtherforward.Wehaven’tgotoutsidethelinguisticsystematall;wehavesimplycomplicatedthingsbyintroducingotherwordssuchasmarsupial.Whatwehaveobtainedfromthedictionaryisnotinfactthemeaningoftheword,butadefinitionofonewordbymeansofaseriesofotherwords.Ourcynicalfriendisnotgoingtogiveupthateasily.Hetakesusonatriptoazooandpositionsusinfrontofacertainenclosure.‘There’,hesays,pointingataparticularanimal,‘Thatisakangaroo.That’swhatthewordmeans’.Againwehavetoexplainpatientlythatit’snotquiteassimpleasthat.Itmightseemreasonable(althoughit’sfarfromuncontroversial,aswewillsee)thatthebrownishcreatureoverthere,orthegroupofsuchcreatures,isthemeaningofthewordkangarooin1)above.Butourfriendmightfindithardertopointatthemeaningof,say,bird.Hewouldneedtodecideononeparticularcreaturetopointat,makingachoicebetween,say,thepenguins,theostrichesandtheparrots.Drawingapictureofabird-likefigurewouldhardlyhelp;ourfriendwouldneedtomakeaseriesofdecisionsaboutwhatcountedasanessentialfeatureof‘bird-hood’.Whateverhisfinaldrawinglookedlikeitwouldbeboundtobearlittleresemblancetosomeofthecreaturesinthezoolabelled‘bird’;ifhisdrawingboreareasonableresemblancetoacanaryitcouldhardlypassforasketchofaflamingo.Finally,nomatterhowcarefullyhesearchesthezoo,ourfriendwon’tbeabletopointatanythingandsay‘There,that’swhatunicornmeans’.Butthatishardlygroundsforsayingthatthewordunicornismeaningless;it’saperfectlyacceptablewordoftheEnglishlanguage.Atthispointourfriendisforcedtoadmitthatthereismoretolinguisticsthanhethought.Theproblemswhichwehavejustbeenconsideringhavebeendiscussedinphilosophyforwellovertwothousandyears.Manyofthephilosopherswhohavecontributedtothediscussionhavedonesobecausetheywereinterestedinitsimplicationsfortheirstudyofknowledge,oroflogic,orofthenatureofreality.Theircontributionshavethereforehadvaryingdegreesofrelevancetolinguistsand,aswewillsee,somehavebeentakenonasbasicpremisesoflinguistics,whileothershavebeenmoreorlessabandoned.Nevertheless,it’sworthwhileforlinguiststostudyadiscussionwhichhasconcerneditselfwithsomanyofthequestionsraisedbytherelationshipbetweenwordsandthings.\n10WordsandthingsOneoftheearliestpointsofdiscussion,andonewhichhasremainedcentral,isthequestionofwhetherwordscaninfactbesaidtoreferdirectlytothingsatall.Thismayatfirstappearanodd,evenanunnecessaryquestiontoask,butwesawearlierthatitwasonewhichcausedproblemsforourcynicalfriend,especiallywhenhetriedtorelatethewordbirddirectlytoanobject.Heknewverywellwhatitmeantforsomethingtobecalledabird,andcouldidentifyanynumberofindividualbirds;hecouldalsotalkaboutbirdswhentherewerenoneactuallypresent.Buttherewasnoonespecimenwhichmetthedefinitioncompletely.Onewayofansweringthequestionmightbetosaythatit’snotobjectsthemselvestowhichwordsrefer,butourideaofobjects;thewordbirdreferstoanidea,ormentalimage,ofwhatabirdislike.Thisanswerhasbeensuggestedbythosephilosopherswhosubscribetoanideationalaccountofmeaning.Accordingtosuchaccounts,weusewordstorefertoourinternalimpressions,whicharederivedfromourexperiencesoftheworld.Oneofthebiggestproblemsfacingideationalaccountsofmeaning,however,isrelatedtotheproblemourfriendencounteredwhenhetriedtodrawanexplanatorypictureofabird.It’stheproblemposedby‘generalideas’;anideaof‘bird’wouldhavetobegeneralenoughtobecompatiblewithanyindividualexampleofabird,butwouldthereforerunintodangerofbeingtoogeneraltocountasacoherentmeaning.Analternativeapproachtomeaning,onewhichavoidstheproblemofgeneralideas,isknownasthedirectreferenceaccount.Thisismoreorlesstheaccountadoptedbyourfriendwhenhetriedpointingatanimalstoexplainmeaning.Themeaningofawordsuchasbirdissimplythesetofindividualobjectstowhichitapplies.ThemeaningofanindividualnamesuchasCharlesDarwin,ordescriptionsuchastheheadkeeper,issimplytheparticularindividualreferredto.Thisistodescribemeaningintermsofdenotation.Awordorphrasedenotesacertainobjectorobjectsintheworld.However,inmanycasesdirectreferenceonitsownisnotenoughtoexplainmeaning.RemembertheproblemourfriendhadexplainingwhyunicornisalegitimatewordinEnglish.Ifmeaningweresimplyreferencethenunicornshouldbeanonsenseword,becausethereisnothingwhichitdenotes.Butwecouldtauntourfriendwith5)andnotbeaccusedoftalkingnonsense.5)Searchashardasyoulike,youwon’tfindanyunicorns.Andfurthermore,wecanassurehimofthetruthof6)withouthavinganyideawhatthedenotationofthebiggestanimalinthezoois.Tofindoutwhatthisphrasereferstowouldrequirealotofhardandpotentiallydangerousworkwithatapemeasure,butwecanusethephrase,anduseitperfectlycoherently,withoutknowing,orcaring,whatanimalactuallyfitsthedescription.\nWordsandthings116)Thatgatehastobewideenoughforthebiggestanimalinthezootogetthrough.Examplessuchasthesearegenerallyexplainedbysayingthatwordsdon’tjusthavedenotations,theyalsohaveconnotations.Thewordconnotationisusedinformallytodescribepropertieswhichareimpliedorsuggested;youmight,forinstance,beadvisedtoavoidusingawordbecauseithas‘badconnotations’.Inlinguisticsandphilosophythewordhasamuchmorespecificmeaning.Itisusedtodescribetheparticularpropertieswhichmakeawordorphraseapplicable.Thephrasetheheadkeeperwilldenoteoneparticularindividualatanyonetime,anditsdenotationwillvaryfromtimetotimedependingonretirements,promotionsandsoon.Butitconnotesaproperty,thepropertyofbeingheadkeeper,andthepropertyitselfremainsconstant.Asimilardistinctionismadebetweentheextensionandintensionofanexpression.Theextensionofasingulartermsuchastheheadkeeperis,again,justthepersonwhohappenstoholdthatpostatanyonetime,whiletheextensionofageneraltermsuchaselephantisthesetofallindividualelephants.Theintensionofawordorphrase,liketheconnotation,canbeseenasaproperty,orsetofproperties,whichremainconstant.Thesepropertiesdescribetherelevantindividualorindividuals,andcanthereforebeseenasasetofcriteriafordeterminingtheextensionatanygiventime.Wearenowinapositiontoexplainexamples5)and6).Thetermunicorndoesn’tdenoteanything;ithasnoextension.But5)isneverthelessameaningfulsentencebecauseunicornhasaconnotation,orintension;weknowwhattypeof(non-existent)animalwearetalkingabout,justasweknowthatnosuchanimalexists.Similarly,thebiggestanimalinthezooconnotesaparticularproperty,andwecandiscusstheanimalwhichhasthatpropertywithoutknowingwhatspecificanimalitis,withoutknowingthedenotationofthephrase.Inboththesecases,then,itisintensionratherthanextensionwhichiscentraltoexplainingthe‘meaning’contributedtosentencesbyindividualwordsandphrases.Thereare,ofcourse,situationsinwhichtheoppositeistrue:inwhichextensioniscentral.Imaginethequestionin7)appearinginahistoryexam.7)QWhowastheMonarchofEnglandin1600?AiElizabethI.AiiThepersonwithsupremepoweroverthelawsandgovernmentofthekingdom.Thefirstanswerwouldgetyoufullmarks.ItgivestheextensionofthetermtheMonarchofEnglandin1600,byindicatingtheactualindividualitrefersto.Thesecondanswer,however,wouldprobablybeconsideredtobefacetious,orelsetohaveseriouslymissedthepointofthequestion.Itgives\n12Wordsandthingstheintensionofthetermbydescribingthepropertieswhichdetermineitsapplication.Bothanswersmightbearguedtogive‘themeaning’oftheMonarchofEnglandin1600,butitisquiteclearthatthe‘right’answertothequestionistheextensional,ratherthantheintensional,meaning.Thereisaparticulartypeofsentencewhichhighlightsthedistinctionbetweenintensionandextension,andwhichwouldbeproblematicforanaccountofmeaningwhichdidn’tdistinguishbetweenthetwo.Considerthefollowingscenario.Astockbroker,let’scallhimClarence,frequentshislocalConservativeclubmostFridayevenings,andknowsseveraloftheothermembers,atleasttonodto.Oneinparticular,whomheknowsonlyasArchibald,heoccasionallyhasachatwithwhiletheywaitatthebar,andClarenceisinclinedonthewholetothinkthatArchibaldisagoodsort.NowitsohappensthatClarencefallsonhardtimes.Duetoanunprecedentedslumpinthestockmarkethisregularincomeallbutdriesup,andbeforeheknowsit,heisoverdrawnatthebank.Muchtohischagrinhereceivesaparticularlysternletterfromhisbankmanager,whomhedoesn’tknow,butwhosesignaturehemakesouttobe‘A.Braeburn-Twinsett’.HeconcludesthatBraeburn-Twinsettmustbeaveryrudeandunsympatheticfellow.However,ontheoccasionswhenhevisitstheConservativeclub,nowratherlessfrequently,hestilloccasionallychatstoArchibald,andstillgetsonwellwithhim.Nowitjustsohappensthat‘A.Braeburn-Twinsett’isArchibald,whohasnevertoldClarencewhathissurnameis,orwhathedoesforaliving.Inthiscontext,wewouldbejustifiedinsayingthat8)isanaccuratestatementofClarence’sviews,butwewouldhavetoadmitthat9)ismostdefinitelynottrue.8)ClarencebelievesthatArchibaldisajollynicechap.9)ClarencebelievesthatBraeburn-Twinsettisajollynicechap.TheproblemisthatthenamesArchibaldandBraeburn-Twinsettdenotethesameindividual,soif8)istrueitseemslogicalthat9)mustbetrueaswell,orelseClarencemustholdtwoopposingviewsaboutthesameperson.Thesolution,ofcourse,isthatalthoughthetwonameshavethesameextension,theydon’thavethesameintension.AsfarasClarenceisconcerned,Archibaldis‘thechapintheConservativeclub’andBraeburn-Twinsettis‘mybankmanager’.Itistheintensionofthenameswhichisrelevanttothe‘meaning’of8)and9).1Theseexamplescouldn’tbeexplainedifmeaningconsistedsimplyoftheobjectdenoted,asourcynicalfriendoriginallysuggestedwhenhetriedtoexplainmeaningbypointingtoanimalsinthezoo.Asweshallsee,examplessuchasthesehavebeendescribedasinvolving‘intensionalcontexts’,preciselybecausetheypresentsituationsinwhichtheintensionofanexpression,ratherthanitsextension,arerelevanttoquestionsofmeaningandrelateddecisionsabouttruthandfalsity.Examples8)and9)arealsoknownasreferentiallyopaquecontexts,orsimplyopaquecontexts.Tobeopaqueistheoppositeofbeing\nWordsandthings13transparent;inthesecontextsyoucan’t‘seethrough’tothereferenceoftherelevantnames.Beforewelookatsomeofthewaysinwhichphilosophershaveapproachedsuchquestions,weneedtoconsideroneotherfunctionwhichwordssometimesserve.Again,it’seasiesttodothisbyconsideringapairofexamples.10)Writersareoftentroubledpeople.11)Writersrhymeswithlighters.Example10)isnotaproblemfortheaccountofmeaningwehaveoutlinedsofar.Thewordwritersisherebeingusedtorefertoagroupofpeoplewhoaretheextensionoftheterm,andofwhomitissaidthattheyareoftentroubled.Butthesamecannotbesaidabout11).Herewriterscan’tbereferringtoagroupofpeopleatall;itmakesnosensetosaythatpeople‘rhyme’withanything.Itisonlywordswhichcanbesaidtorhyme,andso11)canonlybeunderstoodasastatementaboutthewordwritersitself.It’stheworditself,ratherthanitsextension,whichisbeingreferredto.Asfurtherevidenceofthis,notethat11)onlymakessensewiththesingularverb,rhymes.Thesentencewouldmakenosenseifwechangedittorhymetoagreewiththepluralnounwriters,asin‘Writersrhymewithlighters’.Thedistinctionbetweenexamplessuchas10)and11)isgenerallyreferredtoasthedifferencebetweentheuseofawordanditsmention.Example10)includesauseofthewordwriters;itreferstoitsusualextension.In11),however,thewordismentionedwithoutbeingused;itrefersonlytotheworditself.Notethatthewordlightersisalsomentionedhereratherthanused;11)isconcernedwiththewordlighters,notwithitsextension.Thisdistinctionisconventionallysignalledbyputtingawordwhichisbeingmentionedintoquotationmarks.Wethereforegetpairssuchas:12)Gladyshasanicename.13)‘Gladys’isanicename.Example12)isaboutaparticularperson,denotedbyGladys,ofwhomitissaidthatshehasanicename.Example13),however,isaboutaparticularname,denotedby‘Gladys’,ofwhichitissaidthatitisanicename.Later,wewilllookatsomeoftheconsequencesofthisaccount,andatsomeoftheobjectionswhichhavebeenraisedtoit.Inthischapterwewillbeconcernedmainlywiththerelationshipbetweenobjectsandthewordswhichnamethem.Inotherwords,wewillbeconsideringthemeaningsofconcretenounsandnounphrases.Thisinpartreflectstheemphasisofmuchofthephilosophywewillbestudying.Fromtheearliesttimes,philosophershavebeeninterestedintherelationshipbetweenphysicalobjects,togetherwiththeimpressionsthey\n14Wordsandthingsimparttothefivesenses,andthewordswhichpeopleuseforthem.Aswewillsee,however,thereisalsoalongtraditionofdiscussingthetermswhichareusedtodescribeproperties,suchasjustandgood.Aslinguists,weareusedtolabellingtheseadjectives,butinphilosophicaldiscussiontheyaremoreoftendescribedaspredicates;theycanbepredicatedofnounsinstatementssuchas‘Caesarisjust’and‘peaceisgood’.Wewilltouchonthemeaningofpredicateswhererelevantinwhatfollows,butwewillreservemoredetailedstudyofthemforChapter2.ClassicalphilosophyInterestintherelationshipbetweenwordsandthingscanbetracedbacktoclassicalphilosophy,oftendescribedasthefoundationsoftheWesternphilosophicaltradition.Theterms‘classicalphilosophy’and,lesscommonnowadays,‘thephilosophyoftheAncients’,areusedtorefertothephilosophicalwritingswhichhavesurvivedfromancientGreeceandRome.Perhapsthemostinfluentialofclassicalphilosophers,aswellasthebestknown,arePlatoandhispupilAristotle,wholivedandworkedinancientGreeceinthefifthandfourthcenturiesBC.Theoriginsofthephilosophicaltreatmentofsuchtopicsaspolitics,ethics,science,aestheticsandindeedlanguagehavebeentracedtotheirwork.MostofPlato’sphilosophicalwritingsareintheformofdialogues.Inthese,twoormorecharactersdiscussanddebatephilosophicalissues,ofteninaninformalandconversationalstyle.IngeneralinPlato’sdialoguesthereisonecharacterwhoactsasteacher,andguidesoneormorepupilstowardsabetterunderstandingofthechosentopicbyaskingaseriesofleadingquestions.InPlato’searlywritingsatleast,theteacheriscalled‘Socrates’,andisgenerallytakentorepresentthecharacterandideasofPlato’sownteacherSocrates.It’sonlythroughthesedialoguesthatweknowoftheteachingsofSocrates,whodidn’thimselfleaveanywrittenrecordofhisideas.TounderstandPlato’saccountoftherelationshipbetweenwordsandobjects,weneedtolookatoneofhismostfamousideas,the‘theoryofForms’.PlatopositstheexistenceofIdealForms,sometimesknownasPlatonicIdeals.Thesearetheideal,orperfectversionsofqualitiesandobjectswhichenableustomakesenseoftheworldaroundus.Ourknowledgeofthemisinnate,meaningthatwearebornwithitaspartofourhumannatureratherthangainingitthroughexperience.PlatonicIdealsarenotthemselvestangibleobjects;it’sperhapseasiesttothinkofthemasbeingthetemplates,orpatterns,ofallindividualobjects.Assuch,theyareabsoluteandunchanging.Thereexistsanidealversionof,forinstance,Justice,althoughwedon’texperiencethisIdealwhenweexperiencesomeindividualmanifestationofjustice.Similarly,thereareIdealversionsofTable,DogandTriangle,butallweeverexperienceareindividual,\nWordsandthings15imperfecttables,dogsandtriangles.TheseFormsaresometimesalsodescribedas‘Ideas’.Thistermshouldn’tbetakentosuggestthatPlatonicFormsexistonlyinthehumanmind.PlatoseemstounderstandIdealFormsasbeinguniversalpropertieswhichhaveareal,butintangibleexistence.Plato’saccountofIdealFormsisdevelopedoveranumberofhisdialogues.OnesuchdialogueisPhaedowhich,likemanyofthem,isnamedafteroneofthepupilswhomSocratesinstructs.Socratesconductsadialoguewithagroupofhispupils,inwhichheleadsthemtounderstandthatanyjudgementthattwoobjectsdisplay‘equality’mustbebasedonknowledgeofaperfectversionofEquality,whichservesasan‘absolutestandard’.Hegoesontoarguethatthesamemustholdforothercharacteristics,suchasbeauty,goodness,uprightnessandholiness.OurjudgementsofallthesequalitiesdependonknowledgeofIdealForms.Socratesdistinguishesbetweenindividual,‘concrete’objects,observablebymeansofthesenses,andtheForms,whichare‘constant’:‘Andthese[concreteobjects]youcantouchandseeandperceivewiththesenses,buttheunchangingthingsyoucanonlygraspwiththemind—theyareinvisibleandarenotseen’(79a).2Plato’saccountofhowwordsrelateto,orname,objectsandqualitiesdependsontherelationshipbetweenimperfectconcreteindividualsandconstantIdealForms.Ifweseesomethingwiththeappropriateproperties,wecallitbeautiful,notbecauseitisthesameastheIdealofBeauty,butbecauseitissimilarenoughtoittobegiventhatname.InthediscussionreportedinPhaedo,Socratesestablishesthat‘theFormsexistindividually,andthatotherthingsparticipateinthemandderivetheirnamesfromthem’(102b).Inourexample,wemightsaythattheindividualderivesitsnamefrom,oriscalledaftertheIdealofBeauty,althoughitisnot,andcouldnotbe,ofthesameformastheIdeal.SothemeaningofthewordbeautyispreciselythatIdealFormwhichservesasatemplateagainstwhichanyindividualcanbemeasured.Thisearlyaccountofmeaningis,therefore,verydifferentfromthedirectreference,orextensionalaccountswehaveconsidered.Theextensionofthewordbeautyisthesetofindividualentitiestowhichitcanappropriatelybeapplied.Butthemeaningoftheword,accordingtoPlato,isnotthisactualset,buttheIdealFormofBeauty,whichtheindividualsall,imperfectly,resemble.Platomakesclearthatwedon’texperienceIdealFormsexceptbythinkingaboutthem,butit’snecessarytohisaccountthatwehavesome‘inborn’or‘innate’knowledgeofthem,towhichourindividualexperiencescanbecompared.Inotherwords,wemusthaveaninnateknowledgeoftheproper,orrealmeaningsofwordsbeforeweeverexperienceindividualswhichthosewordscanname.Plato’saccountisconcernedwithabstracts.Inordertoexplainthewayinwhichwordsareappliedtoobjectsandconcepts,hetalksnotjustoftheactualobjectsbutofothertypesofentity,entitieswhichhavenoconcreteexistenceandwhichwecanthereforeneverexperiencebymeansofour\n16Wordsandthingssenses.Itisthereforeametaphysicalaccount,andindeedisregardedasoneofthefirstmetaphysicaltheoriesintheWesterntradition.Metaphysicscanbeverygenerallydefinedasthediscussionofthatwhichexists,butexistsoutofreachofoursensesandthereforecan’tbedirectlyobservedandstudied.So,forinstance,religioustheoriesaremetaphysicalinthat,ingeneral,theyareconcernedwiththeexistenceofthingswecan’tsee:thingswhichare‘otherthan’physicalreality.LikeSocratesbeforehim,Platowasdedicatedtotheteachingofphilosophy.HeestablishedtheAcademyofAthens,sometimesdescribedastheearliestuniversityintheWesternworld.ByfarhismostfamouspupilattheAcademywasAristotle.MostoftheworkofAristotlewhichhassurvivedisintheformofpublishedlectures,originallydeliveredattheAcademyandelsewhere.Aristotle’sworkwasheavilyinfluencedbyhisteacherPlato,buthebynomeanssubscribedtoallofPlato’sideas.Perhapsmostsignificantly,herejectedthenotionofIdealFormsuponwhichPlato’sphilosophyofobjectsandnamesdepended.Thisrejectionwasconsistentwithhisattemptstoexplaintheworldfromwhatissometimesdescribedasa‘commonsense’pointofview;therewasnoplaceinhisdescriptionforentitiesofwhichwehavenodirectevidence.ForAristotle,theultimaterealityiswhatweexperiencethroughoursenses.Inthis,hisapproachcanbedescribedasessentiallyempirical.Heseesourknowledgeoftheworldasbeingderivedfromourexperienceofit,ratherthanfromunobservablemetaphysicalconcepts.Wehaveaconceptofapropertysuchas‘justice’becausewehavelearnedaboutitfromourexperienceofindividual,actualactsofjustice.ThisisverydifferentfromPlato’saccount,inwhichindividualactsofjusticearerecognisedassuchbecauseoftheextenttowhichtheypartakeintheuniversalandunchangingconceptofJustice.ThedifferentaccountsofrealityofferedbyPlatoandAristotlenaturallyhavedifferentconsequencesforthenatureofwords,andtheirrelationshiptoobjects.WehaveseenthatPlato’smetaphysicalaccountsuggeststhatwordsarethenamesofIdealForms.Concreteobjectsmaybegivennamesbybeing‘namedafter’theIdealFormtowhichtheyapproximate,butcanneverbeidentical.ForAristotlethepictureisreversed.Allwehaveaccesstoisinformationfromoursenses,inotherwordsourexperiencesoftheappearance,smell,taste,soundandtextureofactualobjects.SowordsmuststandnotforabstractIdeals,norevenforconcreteobjects,butfortheimpressionwhichwehavereceivedfromoursenses.Whenwedescribesomethingas,say,adog,wearegivinganametotheimpressionwereceivefromthesightofafour-leggedanimalofaparticulartype,aswellperhapsasacharacteristic‘doggy’smelland‘shaggy’textureofwhichwearealsoaware.Andwhenintheabsenceofanysuchimpressionswetalkgenerallyabout‘dogs’,wearereferringtothecollectionofexperienceswehavegainedfromindividualdogs.Similarly,inusinganabstractnounsuchas‘justice’,werefertotheimpressionswehavereceivedfromindividualactsofjustice.\nWordsandthings17ThecollectionofimpressionswhichAristotledescribesusasreceivingfromourencounterswiththeworldaresometimeslabelled‘Ideas’.ThetermisusedratherdifferentlyherethanwhenitisappliedtoPlatonicForms,becauseinthiscasewereallyaredealingwithsomethingwhichexistsinthehumanmind:withwhataresometimescalled‘mentallikenesses’.AristotlediscussestheroleoftheseIdeasin,forinstance,theworkentitleddeInterpretatione.Hedescribes‘signs’asvaryingbetweenindividuals,but‘affectationsofthesoul’,ormentalexperiences,asbeingthesameforeveryone:Andjustaswrittenmarksarenotthesameforallmen,neitherarespokensounds.Butwhattheseareinthefirstplacesignsof—affectationsofthesoul—arethesameforall;andwhattheseaffectationsarelikenessesof—actualthings—arealsothesame.(Chapter1)3SoforAristotle,wordsareeffectivebecausetheyaresigns,orsymbols,ofthecollectionofimpressionswhichformamentalexperience.Theseexperiencesinturnarethe‘likenesses’ofactualentities,whichhavearealityandanexistenceoftheirown,althoughwecanhaveaccessonlytotheexperiences.InAristotle’swork,then,wecanseethebeginningsoftheideationalaccountofmeaning.Themeaningofawordisaparticularidea,ormentalimage,whichwehaveformedasaresultofourexperienceoftheworld.Hisviewsonexperience,andalsoontherelationshipbetweenwordsandobjects,have,aswewillseenext,influencedphilosophicaldebateinmorerecentcenturies.ButPlato’saccounthasalsoremainedimportant,andwewillbereturningtoit,andtosomeoftheideasitinfluenced,inlaterchapters.TheBritishEmpiricistsForAristotle,itwas‘ideas’,inthesenseofmentalimages,orimpressionsofexternalobjects,whichwerethemeaningsofwords.Thistypeofdefinitionofmeaningwasfundamentaltotheworkofanumberofindividualphilosophersoftheseventeenthandeighteenthcenturieswhoareoftendescribedassettingtheagendaformodernphilosophy,muchasPlatoandAristotledidforancientphilosophy.ThesephilosophersaresometimesreferredtocollectivelyastheBritishEmpiricists,buttheydidn’tworkasagroupandtheirworkinfactdiffersfromeachother’sinanumberofways.Whattheydohaveincommon,however,asthetitlesuggests,isanempiricalapproachtophilosophy.Theywerecommittedtoexplainingourknowledgeandunderstandingoftheworldaroundusintermsofourexperienceofit.TheBritishEmpiricistswerephilosophersoftheEnlightenment.Thisis\n18Wordsandthingsthenamewhichisoftengiventotheintellectualandpoliticaldevelopmentsoftheperiodbetweenthe1680sand1780s.ThiswasatimeofgreatchangeinBritainandinEuropegenerally,whenmanyofthetraditionalbeliefsandvaluesofthemedievalandearlymodernperiodswerebeingchallengedbythedevelopingnaturalsciences.Inparticular,peopleweregraduallybecominglesshappytoaccepttheauthorityofthechurchoverallaspectsoflife,andwerethereforebecomingreluctanttoacceptthe‘divineright’ofmonarchstorulewithabsolutepower.Instead,ideasofthenaturalequalityofallpeoplewereadvanced,whichinturnledtoaninterestinwhatwewouldnowcall‘humanrights’.Duringthisperiod,variousprogrammesofsocialandeducationalreformwereattempted.IfthepoliticalmovementsoftheEnlightenmentsawarejectionofauthorityinfavourofegalitarianism,theninphilosophytootherewasatendencytorejecttraditionalwisdominfavouroftheindividual’suseofreason.Therewasagrowingbeliefinthepowerofhumanprogress,andtheresultantobligationonpeopletomakesenseofthephysicaluniverseforthemselves,ratherthanrelyingonmysticismandsuperstition.Oneoftheearlyphilosopherstovoicesuchviews,andaleadingBritishEmpiricist,wasJohnLocke.Bornin1632,LockestudiedtraditionalphilosophyatOxford,buthisinfluencesalsoincludedthenewideaswhichwerebecomingestablishedinEurope,andthedevelopingexperimentalsciences.Hewroteextensivelyonscience,onpoliticsandeconomics,andalsoonthephilosophyofknowledge.ProbablyhisbestknownworkisthelengthytreatiseonscienceandknowledgeentitledAnEssayConcerningHumanUnderstanding.Thiswasfirstpublishedin1690,butunderwentseveralextensiverevisions,thefourtheditionappearingin1700,justfouryearsbeforeLocke’sdeath.Inthiswork,Lockeconsidersthenatureofhumanunderstanding,orknowledge,andattemptstoexplainhowitispossible.Heemphasisesthecentralimportanceofobservationandofreason.Insodoing,herejectstheviewcurrentatthetimethathumanbeings,byvirtueoftheirstatusastheculminationofCreation,havecertaininnateknowledgewhichenablesthemtomakesenseoftheworld.WecanrecogniseinthisaviewsimilartoPlato’sideathatwemustbebornwithknowledgeofuniversalForms,beforewecanmakejudgementsaboutactual,concreteobjects.Lockeallowsthatourbasiccognitivecapacities,ourabilitiestothink,reason,andremember,mustbeinnate,inthesenseofbeinganecessarypartofhumannature.Buthearguesthatallactualknowledgemustbeacquired.Hedeclaresthathisintentionistoshow‘howmen,barelybytheuseoftheirnaturalfaculties,mayattaintoalltheknowledgetheyhave,withoutthehelpofanyinnateimpressions,andmayarriveatcertainlywithoutanysuchoriginalnotionsorprinciples’(BookI,ChapterII,1).4Lockeholdsthatallthatwehavedirectaccesstoareourideas,whichwecancontemplatebyuseofourunderstanding,anditistotheseideasthatwereferwhenweusewords.SoforLockethemeaningsofwordsarethe\nWordsandthings19varioussimpleandcomplexideaswhichhumansarecapableofforming.Indeed,inasectionofhisAnEssayConcerningHumanUnderstandingdedicatedtothediscussion‘OfWords’Lockeintroducestheideathataspeakerisabletousewords‘assignsofinternalconceptions,andtomakethemstandasmarksfortheideaswithinhisownmind,wherebytheymightbemadeknowntoothers,andthethoughtsofmen’smindsbeconveyedfromonetoanother’(BookIII,ChapterI,2,originalitalics).Sowhenweuseaword,saytree,wearedescribingtheidea,ortheimpression,wereceivefromoursenseexperiencesofatree.NotethatLockeseesthepurposeoflanguageasbeingtoconveyideasfromthemindofthespeakertothatofthehearer.Thispictureofthefunctionoflanguage,whichhasbeenvoicedindifferentwaysbymanyphilosophersoverthecenturies,isonewewillreturntolater.Wordscanalsobeusedtoconvey‘generalideas’.When,inthecomfortofourstudy,wetalkaboutthepropertiesof‘trees’,wearereferringtotheideawehaveformedfromallourexperiencesoftrees.Indeed,it’sonlybymeansofthewordtreethatweareabletoclassifyalltheseindividualexperiencesasbeinginsomeway‘thesame’.AsLockeexplainsit,theintroductionofgeneraltermsmeantthat‘onewordwasmadetomarkamultitudeofparticularexistences’(BookIII,ChapterI,3).Theexistenceofawordtreegroupstogetheracollectionofsimpleideasderivedfromseparatesensoryexperiences.Thisaccountofgeneralideasrunsintosimilarproblemstothosewhichfacedourfriendatthestartofthischapterwhenhetriedtopresentuswiththemeaningofbird.Rememberthatwhenhetriedtodrawapicturetoexplainthemeaningheencounteredallsortsofdifficulties.Nowimaginethatthepicturehedrawsistorepresentthegeneralideawhichhehasinhishead.Hewillneedtoproduceanimagewhichisidentifiablybird-like,buthasnospecificcharacteristicswhichwouldmakeitincompatiblewithparticularbirds.Itcouldn’tbeofanyparticularcolour,sincebirdscomeinallsortsofcolours.Thiswouldseemtodemandtheimpossible:thatwhenweusethewordbirdgenerally,wehaveaccesstoanimageofabirdwhichisofnoparticularcolour,orforthatmattersizeorshape.Yetevenacolourlessimagewillnotdo,ifsuchathingispossible,becauseallbirdsareofsomecolour,andtherefore‘colour’mustbeoneofthepropertiesofthegeneralidea.ThisveryproblemwaspointedoutafewyearsafterLocke’sdeathbyGeorgeBerkeley,anotheroftheBritishEmpiricists,butonewhodifferedfromLockeonseveralimportantpoints.InhisPrinciplesofHumanKnowledge,publishedin1710,Berkeleytakesasanexamplehisowninabilitytocontemplateasuitablygeneralideaof‘man’:TheideaofmanthatIframetomyself,mustbeeitherofawhite,orablack,oratawny,astraight,oracrooked,atall,oralow,ora\n20Wordsandthingsmiddle-sizedman.Icannotbyanyeffortofthoughtconceivetheabstractideaabovedescribed.(Introduction,10)5BerkeleypayshomagetoLocke’swork,describinghimas‘alatedeservedlyesteemedphilosopher’(Introduction,11)butarguesthathehasbeenledintotheerrorofproposinggeneralideasbythenatureoflanguageitself;thefactthattherearegeneraltermsseemstosuggestthattheremustbegeneralideastowhichtheyrefer.Berkeleyarguesthat,onthecontrary,generalwordsmustbecapableofbeinggivenbroaddefinitions,forinstancethatatriangleisa‘plainsurfacecomprehendedbythreerightlines’,butthatsuchdefinitionsarecompatiblewithavastnumberofindividual,separate,ideas.Hesumsup:‘Itisonethingfortokeepanameconstantlytothesamedefinition,andanothertomakeitstandeverywhereforthesameidea:theoneisnecessary,theotheruselessandimpracticable’(Introduction,18).GeorgeBerkeleyisoftenreferredtoasBishopBerkeley,thepositionherosetointheAnglicanChurch.HewasborninIrelandin1685andspentmuchofhislifethere,althoughhealsospenttimeontheContinent,inAmericaandinEngland,dyinginOxfordin1753.Hislivelypersonalityandnovelphilosophicalideasmadehimpopularinsociety,andheforgedmanyfriendshipsamongtheintellectualandliteraryfiguresofhisday.Berkeley’sreligiousbeliefswerethefoundationofmuchofhisphilosophy.LikeLocke,hearguedthatthemindhasdirectaccessonlytoitselfanditsownideas.HealsoclaimedthatknowledgeofGodispresentintheminditself.Hisaccountofthemindgaverisetohisphilosophyof‘immaterialism’,forwhichheisprobablybestknown,andwhichhasledtovariousparodiesofhiswork.Theseparodiestendtoplayonhisinsistencethatwehavenoconclusiveevidencethatthematerialworldexistsintheabsenceofourperceptionofit.Berkeleywasanxioustocounter‘materialist’philosophieswhichclaimthattheonlyrealexistenceistheexistenceofphysicalmatter.Suchaccountsruleoutthevalidityofdiscussingminds,spiritsand,indeed,God.InthisBerkeleycanbedistinguishedfromLocke.Locke,aswehaveseen,describedwordsasstandingonlyforourideasofthematerialworld.Butheallowedthatwegenerallyassumethatthoseideasareformedfrominformationfromoursensesaboutrealobjects:‘Itisthereforetheactualreceivingofideasfromwithoutthatgivesusnoticeoftheexistenceofotherthingsandmakesusknowthatsomethingdothexistatthattimewithoutuswhichcausesthatideainus’(BookIV,ChapterXI,2).Locke’sviewsherecanperhapsbelikenedtothoseofAristotle,whoaswehaveseen,maintainedthatwordsstandforindividualideas,butallowedthatthematerialobjectswhichformedthoseideaswereactualandconstant.Berkeleydrewthecontraryconclusion,namelythat,sincetheonlythingswehaveaccesstoareourideas,theonlythingswhichwecanbesurehaverealexistencearethoseideas.Wehavenosufficientevidencethatthe\nWordsandthings21materialworldexistsindependentlyofourideaofit.Justaswehavenoreasontobelievethattheobjectsofourimaginationexistotherthaninourmind,sowehavenoreasontobelievethatourideas,howeverformed,haveanyexistenceotherthaninourmind.Soforinstance,wecan’tsaywithcertaintythatatreewhichweareobservingwillcontinuetoexistonceweceaseobservingit;tosaysoisonlytoimaginethatwecouldobserveitifwewerepresent:Forastowhatissaidoftheabsoluteexistenceofunthinkingthingswithoutanyrelationtotheirbeingperceived,thatseemsperfectlyunintelligible.Theiresseispercipi,norisitpossibletheyshouldhaveanyexistence,outofthemindsofthinkingthingswhichperceivethem.(PrinciplesofHumanKnowledge,3)TheLatininthispassageisgenerallytranslatedas‘tobeistobeperceived’.Berkeleyoffersanumberofargumentsinfavourofthisconclusion,encouraginghisreadertoperformaseriesofmentalexercisesasproofthatitisimpossiblefor‘unthinking’,orinanimatebeingstohaveanyrealityindependentofthemindsofthosewhoobservethem.Hesuggests,forinstance,thatweconsiderwhatdifferenceswewouldbeabletoobserve,firstiftherewereanexternalrealityandtheniftherewerenone;hissuggestionisthattherewouldbenodifference,hencethatwehavenoreasontobelieveinanexternalreality.Hegoesontoarguethatitisnotinfactpossibletoimaginetheexistenceofanunobservedobjectwithouttherebyimaginingitobserved.Hepresentsalltheseideasasifheissimplyremindinghisreadersofwhatmustbecommonsense:Butsayyou,surelythereisnothingeasierthantoimaginetrees,forinstance,inapark,orbooksexistinginacloset,andnobodybytoperceivethem.Ianswer,youmayso,thereisnodifficultyinit.Butwhatisallthis,Ibeseechyou,morethanframinginyourmindcertainideaswhichyoucallbooksandtrees,andatthesametimeomittingtoframetheideaofanyonethatmayperceivethem?Butdoyounotyourselfperceiveorthinkofthemallthewhile?Thisthereforeisnothingtothepurpose.(23)DespiteBerkeley’sinsistencethatwehavenovalidevidencefortheexistenceofthematerialworld,hewasbynomeansasceptic.Thatistosay,hedidn’tholdwiththeviewthatnothingcanbeknownforcertainandthatallthatispossiblearejudgementsofprobabilitybasedonobservation.Thescepticalapproachtoknowledge,whichcanbedatedtoancienttimes,wascurrentwhenBerkeleywaswriting,andindeedoneofhischiefmotiveswastocounteractsuchtrends.Itscurrencywasrelatedtothechallengeposedtooldorthodoxies,andparticularlytothevalidityofreligious\n22Wordsandthingsknowledge,duringtheEnlightenment.Berkeleywaskeentoprovideanalternativetoscepticism,withitsclaimsthatnoknowledgecanbecertainandthereforethatwecanhavenocertaintyoftheexistenceofGod.Alongwithothertheologiansofhistime,heequatedthispositionwithatheism.Aswehaveseen,heclaimedthattheminddoesindeedhavecertainknowledge,butonlyofitselfandofGod.Materialobjectsexistonlyinbeingperceived,butthis,heargued,doesn’tmeanthatweneeddoubttheirexistence.Allobjectsareconstantlybeingperceived,sotheyallcontinuetoexist.Evenwhennohumansarearound,thematerialworldisperceivedbyGod,andexistsinthemindofGod.BeforeleavingBerkeley’sresponsetoLocke,itisworthtakingnoteofoneotherpointheraises.RememberthatLockedescribedlanguageasexistingmainlyforthepurposeoftransferringideasfromthemindofthespeakertothatofthehearer.Berkeleyobserves,however,that:Thecommunicatingofideasmarkedbywordsisnotthechiefandonlyendoflanguage,asiscommonlysupposed.Thereareotherends,astheraisingofsomepassion,theexcitingto,ordeferringfromanaction,theputtingthemindinsomeparticulardisposition;towhichtheformerisinmanycasessubservient,andsometimesentirelyomitted,whenthesecanbeobtainedwithoutit,asIthinkdoesnotinfrequentlyhappeninthefamiliaruseoflanguage.(Introduction,20)Berkeleyisherechieflyconcernedwithoccasionswhenwordsmayhavesomeeffectonthehearer,particularlyanemotionalone,withoutthateffectnecessarilybeingassociatedwithanyidea.Buthisobservationcanperhapsbeseenasaprecursortothediscussionofmeaninginuse,atopicwhichwewillconsiderfurtherinlaterchapters.WewillalsoreferbacktotheworkoftheBritishEmpiricistsinthefinalchapter,whenwewillconsideritsimplicationsforthequestionofhowchildrenfirstlearnlanguage.First,however,weneedtoconsidersomeoftheresponsestotheirviews,andthewaysinwhichtheyinfluencedlaterthought.Inthesewecanseethebeginningsofarecognisably‘modern’accountofmeaning.NamesGeorgeBerkeleywasoneofalargenumberofphilosophersandtheologianstorespondtoLockesoonafterthepublicationofAnEssayConcerningHumanUnderstanding.AnothersuchresponsewasthatoftheGermanphilosopherLeibniz,whowroteNouveauxEssaissurl’EntendementHumain,translatedintoEnglishasNewEssaysonHumanUnderstanding,between1703and1705,althoughitwasnotpublisheduntil1765.6\nWordsandthings23GottfriedWilhelmLeibnizlivedandworkedinGermanyinthelateseventeenthandearlyeighteenthcenturies.LikeLocke,hewasdeeplyinterestedinthescientificdevelopmentsofhistime.Indeed,hewasatruepolymath,familiarwithmanydisciplinesbesidesphilosophy,andwritingontopicsasdiverseasmedicineandtheology.Hisprofessionallifewasspentincourtratherthaninuniversityappointments;despitebeingofferedanumberofprofessorships,hepreferredtoremainintheserviceoftheDukeofBrunswick,wholaterbecameGeorgeIofEngland.Hisofficialdutiesseemtohavelefthimplentyoftimeforhisstudies.TheonlymainphilosophicalworkwhichwaspublishedduringhislifetimewasTheodicy,whichfirstappearedin1710,buthewroteextensively,andmanyotherbookswerepublishedafterhisdeath,includingNewEssaysonHumanUnderstanding.Leibniz’sresponsetoLockeispresentedintheformofadialoguebetweentwocharacters:PhilalethespresentingLocke’sargumentsandTheophilusreplyingtothemwithLeibniz’sresponses.Leibnizoffersarationalistaccountofunderstanding;heseesknowledgeasbeingderivedfromtheapplicationofreason,ratherthanadoptingthe‘commonsense’viewoftheempiriciststhatknowledgeisderivedfromexperience.HealsochallengesLocke’sclaimsaboutwhatwordsmean.PhilalethesrepeatsLocke’sclaimthatwordsreferonlytoourideasofobjects;Theophilusrespondsthat:‘Ineachcasewordsindicatethethingsaswellastheideas’(287).ThroughTheophilus,Leibnizisclaimingthatwordsstandnotjustforourideas,orimpressions,ofthings,butforthethingsthemselves.Hedoesn’tgoasfarastoclaimthatwordsreferjusttoactualobjectsorsetsofobjects.This,aswehaveseen,isknownasthe‘directreference’accountofmeaning,inwhichwordsaresaidtoreferdirectlytomind-externalobjectsoreventswithoutthemediationofourideas.OnenamecloselyassociatedwiththedirectreferenceaccountofmeaningisthatofJohnStuartMill.Millcanperhapsbeseenasbeginningadiscussionof‘names’,whichwasdevelopedovertheperiodofacenturybyanumberofdifferentphilosophers.BeforeweconsiderMill’saccountandthedebateitgenerated,however,it’sworthlookingatthepassagewhichimmediatelyfollowstheclaimfromLeibnizwehavejustquoted.Itcanperhapsbeseenasanearlyreferencetothedistinctionwehavealreadyencounteredbetween‘use’and‘mention’.UseandmentionInNewEssaysonHumanUnderstanding,aftermakinghisclaimaboutwords,thingsandideas,Theophilusgoesontomakethefollowingobservation,whichisnotsubjectedtoanyfurtherdiscussion:Sometimeswordsthemselvesarespokenofmaterially,andinsuchacontextonecannotpreciselyreplacethewordbyitssignification,i.e.by\n24Wordsandthingsitsrelationtoideasortothings.Thishappensnotonlywhenonespeaksasagrammarianbutalsowhenonespeaksasalexicographer,givingtheexplanationofaname.(287)Leibnizisnoticingherethatwordsarenotalwaysusedtorefertoobjectsorideas;sometimesweneedtorefertowordsthemselves,mostoftenwhen,ashesuggests,weareconcernedwiththeactuallinguisticpropertiesofwords.Theuseofwordstotalkaboutwords,ormoregenerallyoflanguagetotalkaboutlanguage,issometimesdescribedas‘metalinguistic’;languageisboththemeansofdescriptionandtheobjectdescribed.Wehavealreadylookedatsomeexamplesofthistypewhenweconsideredthedifferencebetweentheuseandthementionofaword,atthestartofthischapter.Wecouldsaythatwhenawordisbeinguseditreferstosomethingoutsideofthelinguisticsystem,butwhenitismentioned,whichLeibnizdescribesasthewordbeingused‘materially’,itreferstotheworditself,partofthelinguisticsystem.Aswehaveseen,ithasbecomecustomarytodistinguishbetweentheuseandthementionofawordbypresentingthelatterinquotationmarks.TheimplicationsofthisarediscussedbythephilosopherLeonardLinskyintheintroductiontoabookwhichheeditedin1952entitledSemanticsandthePhilosophyofLanguage.Hecommentsthat‘inordertosaysomethingabout(ormention)anythingitisnecessarytouseanameorothermeansofdesignationforthatthing’(4).Metalinguisticstatementsthereforeposeapossibleproblem,becausetherearenoindependentnamesbywhichwordscanbedesignated.Linskyarguesthatthereisnopossibleconfusionbetweenathinganditsnameinanexamplesuchas14):14)Johnistall.Thissentencemakesreferenceto,ormentionsthepersonJohn,butit’snotJohnhimselfbuthisnamewhichappearsinthesentence.Johnismentionedinthesentence;John’snameisusedinit.Thisdistinctionisnotsoeasilymade,however,inthecaseof15):15)‘John’consistsoffourletters.ThissentenceisclearlynotaboutthepersonJohn,butaboutthewordJohn;itisthewordJohnwhichisbeingmentioned.YetthewordJohn,thenameforthepersonJohn,doesnotitselfhaveaseparatename.Aswehaveseen,theuseofquotationmarksisaconventionalwayofindicatingthatthewordisbeingmentioned,notused.Anotherwayofexplainingthisistosaythat‘John’isinfactthenameofthewordJohn;wecancreateanameforanywordbyplacingtheworditselfinsidequotationmarks.Thisgivesusawayofmentioningthewordwhenwewanttosay,forinstance,thatit\nWordsandthings25hasfourletters,justaswehaveawayofmentioningthepersonwhenwewanttosay,forinstance,thatheistall.Thedistinctionbetweenuseandmention,andthenotationwhichaccompaniesit,havenotalwaysbeenacceptedwithoutcriticism.Oneofthemostcommonobjectionstoberaisedistheratherunsatisfactorylogicalconsequenceofthisaccount.In15)wesawanexampleoftheuseofthenameforthewordJohn;nowimaginethatwewanttomentionthisname.Followingtheconventionwehaveestablished,wewouldneedtogivethenameanameofitsownbyplacingitinsidequotationmarks.Thiswouldgiveus“John”,andthenameforthis,shouldwewanttomentionit,wouldhavetobe‘“John”’,andsoon.Severalwritershaveexpressedreservationsabouttheinfiniteprocessofnaming-of-nameswhichthedistinctionbetweenuseandmentionseemstoimply.Someofthesearelistedinthe‘furtherreading’sectionattheendofthischapter.InhiscritiqueofLocke,then,Leibnizwasperhapsthefirsttoraisethepossibilityofthis‘special’functionofwords.Healso,aswehaveseen,differsfromLockeinhisaccountofwordsused‘normally’,andthewaytheyrelatetoobjects.Intherestofthissection,wewillconsiderthediscussionbyJohnStuartMillandothersofthisrelationshipbetweennamesandobjects.InthecontextofMill’sworkthetermnameisused,asindeedithadbeenintheworkofpreviousphilosophers,toindicateanyword,orindeedanynounphrase,whichcanbeusedtoidentifyaparticularpersonorobject.Thisusageisnotconsistentthroughoutthedebatethatfollowed;thetermisalsosometimesusedwiththerestrictedmeaningof‘propername’,suchasJohn,Mary,orMrJones,andwewillneedtobearthesedifferentusesinmindinwhatfollows.MillondirectreferenceJohnStuartMillwasborn,justoveracenturyafterJohnLocke’sdeath,in1806.LikeLocke,andindeedlikeLeibniz,hedidn’tbasehiscareeratauniversity;heworkedasanadministratorintheEastIndiaCompanyandlaterasaMemberofParliament,beforehisdeathin1873.Nevertheless,hepublishedmanybooksandpamphletsduringhislifetime.Heisprobablybestknownforhismoralandpoliticalphilosophyandhisviewsonsocialandeconomicreform.Hearguedforascientificapproachtotheunderstandingofthesesubjects,andthisisoneconcernofhisASystemofLogic,firstpublishedin1843.Inthisbook,Millalsoengagesinthedevelopmentofaprecisemethodoflogicalanalysisandreasoning,seeingaclearaccountoflanguageasanecessaryprerequisiteforthis.Indeed,hebeginswithachaptercalled‘Ofthenecessityofcommencingwithananalysisoflanguage’,arguingthatlanguageisanecessarytoolofscientificandlogicalanalysis,andthereforeneedstobeaspreciseandaccurateasanyotherscientificinstrument.\n26WordsandthingsMillposeshimselfthequestionofwhetherwordsarebestdescribedasnamesofthingsorofourideasofthings.Hedescribesthefirstasthedescription‘incommonuse’andthesecondasaninventionby‘metaphysicians’.Millsuggeststhat‘thereseemsgoodreasonforadheringtothecommonusage,andcalling…thewordsunthenameofthesun,andnotthenameofourideaofthesun’(23–4).Hisargumentinfavourofthisdecisionisthe‘commonsense’viewthat,whenweusewordstosaysomething,forinstanceinsaying‘thesunisthecauseoftheday’,wearetalkingaboutthephysicalentitiesinvolved,andnotaboutourideaofthem.Milldistinguishesbetween‘general’and‘individual’names.Theformerarewordswhichcanbeappliedtoanindefinitelylargeclassofobjects,whilethelattercaneachbeappliedtoonlyoneindividual,asisthecasewithpropernames:ThusmaniscapabletobeingtrulyaffirmedofJohn,George,Mary,andotherpersonswithoutassignablelimit;anditisaffirmedofalloftheminthesamesense;forthewordmanexpressescertainqualities,andwhenwepredicateitofthosepersons,weassertthattheyallposessthesequalities.ButJohnisonlycapableofbeingtrulyaffirmedofonesingleperson,asleastinthesamesense.(27)7Millfurtherdistinguishesbetween‘connotative’and‘non-connotative’names,describingthisasoneofthedistinctions‘whichgodeepestintothenatureoflanguage’(31).Tounderstandthisdistinctionweneedtorecalltheexamplesusedabovetoillustratethedifferencebetween‘denotation’and‘connotation’,thedenotationofawordbeingsimplythethingreferredto,andtheconnotationbeingthepropertyorgroupofpropertieswhichdeterminethedenotation.Forinstance,wedistinguishedbetween‘ElizabethI’asthedenotation,and‘thepersonwithsupremepoweroverlawsandgovernmentofthekingdom’astheconnotationofthephrasethemonarchofEnglandin1600.Indistinguishingbetweenconnotativeandnon-connotativenames,Millissuggestingthatnotallnameshaveconnotations.Thosewhichheidentifiesasbeingnon-connotativearewhatwewouldcall‘propernames’.Milldescribespropernamesaswordswhichserveonlytoidentifyindividuals.Apartfromthispurpose,itisinappropriatetoaskwhattheword‘means’,ortoconsiderthe‘reason’foranindividualtohaveacertainname.Ageneraltermlikeman,however,denotesallindividualhumanbeings,andalsoconnotestheattributesofbeinghuman;‘itisaconnotativename’(32).Millstressesthataconnotativenamesuchasthisisstillanameofwhatitdenotes;theconnotationcanofferanexplanationofwhyitcanserveasanameforthis.HeconsidersthepropernameSophroniscusandthedenotationallyequivalentexpressionThefatherofSocrates:\nWordsandthings27Boththesearenamesofthesameindividual,buttheirmeaningisaltogetherdifferent;theyareappliedtothatindividualfortwodifferentpurposes:theone,merelytodistinguishhimfromotherpersonswhoarespokenof;theothertoindicateafactrelatingtohim,thefactthatSocrateswashisson.(38)SoMillclassifiesthepropernameSophroniscus,andthenounphrasethefatherofSocratesastwodifferenttypesofindividualname.Usedsuccessfully,bothservetopickoutanindividual.Butthepropernamedoesnomorethanthis,whilethenounphrasealsogivessomedefinitionofthatindividual.It’spossibletoaskwhysomedescriptivephraseisapplicabletoanindividual,butnottoaskwhysomepropernameis.WecanexplainwhythedescriptionthefatherofSocratesisappropriatebysaying‘becausehewasthefatherofSocrates’,butifaskedthesamequestionaboutSophroniscus,wecanonlysay‘becausethat’shisname’.Itis,ofcourse,possibletostateexplicitlytherelationshipbetweenthenameandthedescriptionwhichMilldiscusses.Wecansay‘SophroniscusisthefatherofSocrates’.Butif,asMillclaims,thepropernameSophroniscusisnon-connotative,thisstatementcanbeconcernedonlywiththedenotationofthetwoterms.Ineffectitsaysthattwoexpressionswhichdenotethesameindividualareequivalent.Thestatementshould,therefore,betotallytrivialanduninformative;itshouldn’tevenbeworthsaying.However,expressionssuchasthisareinfactoftenused,andusedtoconveynon-trivialinformation;theyareoftenworthsaying.ThisparticularproblemwasaddressedbytheGermanphilosopherGottlobFrege,whoseworkwewillconsidernext.FregeonsenseandreferenceBornin1848,GottlobFregewasProfessorofMathematicsattheUniversityofJena.Theworkforwhichheisnowbestknown,onlogicandonlanguage,developedlateinhiscareeroutofhisinterestinmathematics.In1892hepublishedanarticleentitled‘UberSinnundBedeutung’.ThishasbeentranslatedintoEnglishvariouslyas‘Onsenseandmeaning’and‘Onsenseandreference’.8ThetermmeaninghasbeenusedinsuchanumberofdifferentwaysthatitcanbecomeconfusingtotrytoreserveitfortheverypreciseideaofFrege’sBedeutung.Inparticular,Frege’stermisappliedtosomethingquitedifferentfrommeaningasusedbyMillwho,aswehaveseen,commentsthattwoexpressionscanrefertothesameindividualbuthavedifferentmeanings.ForthisreasonwewillusethetermssenseandreferencewhendiscussingFrege’swork.Frege’sstartingpointiswithwhathedescribesas‘expressionsofequality’.UsingMill’sexampletoillustrateFrege’spoint,wecouldsaythat\n28Wordsandthingswhile16)canoffersomevaluableinformation,17)isasimple,necessarytruth,whichtellsusnothingnew.16)SophroniscusisthefatherofSocrates.17)SophroniscusisSophroniscus.Fregepointsoutthat,ifthis‘equality’isarelationbetweentheindividualsdenoted,16)shouldbenomoreinformativethan17);bothwouldmerelystatethatacertainindividualisidenticalwithhimself.Rather,hesuggests,whatwearesayingin16)isthatSophroniscusandthefatherofSocratesbothdenotethesameindividual;wearemakingastatementabouttherelationshipbetweentwodifferentwaysofnaminganindividual,notbetweentheindividualandhimself.Hence,whatwesayissignificantandfarfromatrivialtautology.Thedifferencebetweenthetwoexpressionswhichmakes16)significantisnotadifferenceintheirdenotation,but‘adifferenceinthemodeofpresentationofthethingdesignated’(57).9ThisobservationleadsFregetoproposethateverynamehasanelementconcernedwiththismodeofpresentation,whichhecallsitssense,aswellasadenotation,whichhecallsitsreference.Toillustratethis,Fregeusesthepairofnamestheeveningstarandthemorningstar.Thesenamesweregiventothebrightlightsseenatdifferentpointsintheskyintheeveningandthemorning,respectively.Noweventuallyitwasdiscoveredthatthesetwolightswereactuallythesamebody,seenindifferentplacesbecauseoftherotationoftheearth(theyareboth,infact,theplanetVenus).Whenthisdiscoverywasmade,itwouldhavebeenpossibletoreportitusing18):18)Theeveningstaristhemorningstar.Likeourexample16)above,thisisaperfectlyinformativestatement.Thisisbecause,asFregenotes,althoughtheeveningstarandthemorningstararenowknowntohavethesamereference,theyhavedifferentsenses;thesameobjectisidentifiedinbothcases,buttwodifferentwaysofidentifyingitarebeingused.Wecouldequallywellidentifythissameobjectyetanotherwayandsay:19)TheeveningstarisVenus.SofaritmaynotappearthatFregeissayinganythingsoverydifferentfromMill.Heisusingdifferentterminology,buthisreferenceseemsveryclosetoMill’sdenotation,andhissensetoMill’sconnotation;ifconfrontedwithanexamplelike18),Millwouldpresumablyexplainthattheeveningstarandthemorningstardenotethesameindividualbuthavedifferentconnotations.Andindeed,likeMill,Fregeadoptsa‘directreference’account;heexplicitlystatesthatthemeaningofanameisaphysicalobject,notanidea:\nWordsandthings29Themeaningandsenseofasignaretobedistinguishedfromtheassociatedidea.Ifwhatasignmeansisanobjectperceivablebythesenses,myideaofitisaninternalimage….Theideaissubjective:oneman’sideaisnotthatofanother.(59)Sowhileideasofanobjectmayvaryfromindividualtoindividual,theobjectitself,themeaningofaname,isaconstant,mind-externalentity.ButFregedoesdifferfromMill;briefly,heallowsthatpropernameshavesenseaswellasreference.RememberthatMilldistinguishedbetweenconnotativeandnon-connotativenames;Sophroniscusdenotesanindividualbuthasnoconnotation,whereasthefatherofSocrates,whiledenotingthesameindividual,alsoconnotesacertainpropertywhichdistinguishesthatindividual.Fregedoesn’tmakethisdistinction.Heinfactusesthetermpropernametorefertoanythingwhichservestodenoteaparticularobject.Thedesignationofasingleobjectcanalsoconsistofseveralwordsorotherthings.Forbrevity,leteverysuchdesignationbecalledapropername’(57).SoforFrege,thereisnoneedtodistinguishbetweenSophroniscusandthefatherofSocrates;theybothhavesenseaswellasreference.It’sfairlyeasytoworkoutthesenseofadescriptivenounphrase.In19),forinstance,theobjectdenotedbytheeveningstarisidentifiedbymeansofoneparticularproperty;wemightparaphraseandsayitis‘thelightwhichcanbeseenintheskyinaparticularplaceintheevening’.Butthesenseofapropernameislessapparent.Wemayknowwhat,orwho,theydenote,butwecan’tnecessarilysayhowanindividualispickedoutbyVenusorSophroniscusor,touseFrege’sownexampleAristotle.Frege’sansweristhatsuchnamesworkinexactlythesamewayasthedescriptions;theydenoteanindividualbymeansofsomeidentifyingproperty.Accordingtothisaccount,ifyouknowthemeaningofanamethenyouassociatesomeparticularpropertywiththatmeaning;this,foryou,isthesenseofthename.Thisaccountworkswellenoughformanynames,particularlythoseofhistoricalfigures.ItisfairlylikelythatformostpeoplenowadaysthenameLloydGeorge,forinstance,willrefertoaparticularpersonandwillhaveasensealongthelinesof‘theBritishPrimeMinisterduringtheFirstWorldWar’.But,asFregeacknowledges,theremaynotalwaysbeagreementaboutthesenseofnames.Tousehisexample,wemayfindthatforonepersonthenameAristotlehasthesense‘thepupilofPlatoandteacherofAlexandertheGreat’,whileforanotherithasthesense‘theteacherofAlexandertheGreatwhowasborninStagira’.Frege’sresponsetothisisthat‘solongasthethingmeantremainsthesame,suchvariationsofsensemaybetolerated’,althoughhestipulatesthatthey‘oughtnottooccurinaperfectlanguage’(58n).Fregeinfactseessenseastheessentialpropertyofallnames;itispossiblefornametohavesensebut\n30Wordsandthingsnoreference,butnotvice-versa.Thisisanissuetowhichwewillreturninthenextchapter.FregethenconsiderstheimplicationsforhisaccountofwhatisknownasLeibniz’sLaw,ortheLawofSubstitution.Thiscanbeparaphrasedasfollows:20)Iftwoexpressionsaredenotationallyequivalent,onecanbesubstitutedforanotherinaformulasalveveritate.SalveveritateisLatinfor‘withtruthunchanged’,sotheprinciplestatesthatsubstitutinganexpressionforanotherwiththesamereferenceshouldnotalterthetruth,orfalsity,ofthesentenceinwhichitoccurs.Andthisworksinpracticeformanyexamples.IfwecanagreethatCharlesDickensandtheauthorofGreatExpectationshavethesamereference,weshouldagreethat21)and22),forinstance,mustbothbethesameintermsoftruth(andthattheyare,infact,bothtrue),asmust23)and24)(whicharebothfalse):21)CharlesDickenslivedinthenineteenthcentury.22)TheauthorofGreatExpectationslivedinthenineteenthcentury.23)CharlesDickenswasFrench.24)TheauthorofGreatExpectationswasFrench.However,weseemtorunintodifficultieswhenwetrytoapplytheLawofSubstitutiontoexampleslike25)and26):25)Joesaid‘CharlesDickenswasFrench’.26)Joesaid‘theauthorofGreatExpectationswasFrench’.Hereagainwehaveformed26)byreplacinganexpressionin25)withonewiththesamereference,butthemeaningofthewholeformuladoesnotremainthesame;itstruthhaschanged.IfJoeinfactsaid‘CharlesDickenswasFrench’,then25)isanaccuratewayofreportingthisbut26)isnot.Fregeexplainsthisbymeansofatypeofuse-mentiondistinction.Whatwearetalkingaboutinanexamplelike25)isnottheusualmeaningofJoe’swords,butthewordsthemselves.Wethereforehave‘signsofsigns’,indicatedassuchbythepresenceofthequotationmarks:‘Accordingly,awordstandingbetweenquotationmarksmustnotbetakenashavingitsordinarymeaning’(58–9).ButtheproblemsfortheLawofSubstitutiondonotendwithdirectquotations,withtheirhelpfulquotationmarks.Wecanrecallfromtheintroductiontothischapterthattherearecertaintypesofsentence,knownasintensionalcontexts,inwhichitisn’tpossibletosubstituteextensionallyequivalentexpressions.Asareminder,let’sconsiderthefollowingpair:\nWordsandthings3127)PipbelievesthattheauthorofGreatExpectationsisagenius.28)PipbelievesthatCharlesDickensisagenius.It’spossiblethatif27)istruethen28)isalsotruebutthisisn’tnecessarilythecase.Pipmayhavereadandadmiredthenovelwithoutnoticingwhowroteit;hemaynotrememberwhowroteit;hemayevenbelievethatWalterScottwroteit.Inallthesecaseswewouldwanttosaythat27)wastruebut28)false.Similarly,PipmayhavereadonlyOliverTwist,andonthisbasis28)mightbetrue.Butifhedoesn’tknowthatCharlesDickensalsowroteGreatExpectations,thenwecertainlycouldn’tclaimthat27)wastrue.Theseexamplesinvolvewhatareknownas‘prepositionalattitudes’.Theyareconcernedwithanattitudeofthesubjecttoaparticularproposition.Inthiscasetheattitudeisoneofbelief,butthesameeffectisachieved,withanumberofotherattitudes,suchashoping,fearing,wishingandsoon.Frege’sexplanationofexamplesofthistypereliesonhisdistinctionbetweensenseandreference,andonhisclaimthateverythingwhichhedescribesasa‘propername’hasasense.Heclaimsthatincasessuchas27)and28),aswellasinthedirectquotationexamplessuchas25)and26),therelevantnames‘donothavetheircustomarymeaningbutdesignatewhatisusuallytheirsense’(59).Soin28),forinstance,thename‘CharlesDickens’isbeingusedtodenotenotitsreference,theindividual,butitssense.Aswesawearlier,whatcountsasthesenseofanamecanvaryfromonepersontoanother.CharlesDickensisherebeingusedtodenotethesensewhichPipattachestothisname.Aswehavesuggested,thissensemaybe‘theauthorofOliverTwist’,andmaynotbe‘theauthorofGreatExpectations’,dependingontheextentofPip’sindividualknowledge.Fregelabelstheseuses,whereaworddenotesitssense,asindirectuses,whichhedistinguishesfromcustomaryuses,inwhichawordisusedtodenoteitsreference.‘Theindirectmeaningofawordaccordinglyisitscustomarysense’(59).Wearenowfamiliarwiththreesetsofexpressionstodiscussthedifferencebetweentheactualobjectidentifiedbyaname,andthepropertieswhichdeterminehowthenameisapplied:extensionandintension,Mill’sdistinctionbetweendenotationandconnotation,andFrege’sreferenceandsense.10Thesesamedistinctions,particularlythelastone,havebeenemployedtotalkaboutthemeaningofsentences,atopicwhichwewillinvestigateinthenextchapter.RussellondescriptionsThenextmajorcontributiontothediscussionofthenatureofnamesanddescriptionscamefromBertrandRussell.Russellwasoneofthemostinfluentialfiguresintwentieth-centuryBritishphilosophy,publishingmanybooksandarticlesduringhislongcareer.Hewasalsoanotedpolitical\n32Wordsandthingsactivistandsocialreformer,ahigh-profilepacifistand,inlaterlife,aleadingmemberoftheCampaignforNuclearDisarmament.Bornin1872intoanaristocraticfamily,RussellspentmuchofhisprofessionallifeatCambridge.However,hispacifismduringtheFirstWorldWar,whichledtohisbeingbrieflyimprisoned,causedhimtobedismissedfromhispost,andhedidn’treturntotheUniversityuntil1944,whenhebecameaFellowofTrinityCollege.LikeFrege,Russell’sfirstandprimaryphilosophicalinterestwasinmathematicallogic.Hisapproachwasessentiallyananalyticone,concernedwiththeanalysisofthelanguageinwhichtheideasoflogicareexpressedand,ultimately,witharefinementofthatlanguage.Thisinturnledtoaninterestinmeaning,andinparticularinthelogicoflinguisticmeaningasitrelatestoissuesofhumanknowledgeandunderstanding.RussellclaimsinhisAutobiography,publishedin1967,thatitwaswhilehewasinprisonduringtheFirstWorldWarthathisinterestinmeaningdeveloped.Certainlyitwastherethathewrotethemanuscriptwhichwaspublishedafterthewarin1919asIntroductiontoMathematicalPhilosophy.BertrandRussellwasthefirsttousethenow-familiartermdefinitedescriptionforphrasessuchasthefatherofSocrates,theeveningstar,andsoon.Suchphrasesofferadefiningdescriptionofanindividualand,crucially,conveytheideathatthereisonlyonesuchindividual.11Theyareintroducedbywhatisknowninlinguisticsasthedefinitearticle,anditisthiswhichintroducestheideaof‘uniqueness’.Withtheindefinitearticletheeffectisdifferent;aneveningstardoesn’tpickoutanyparticularstar,inRussell’stermitintroducesonlythe‘concept’ofeveningstars,whileafatherofSocratesisbizarrepreciselybecauseitseemstosuggestthatSocratesdidn’thavejustoneparticularfather.RememberthatFregedifferedfromMillinarguingthatbothdefinitedescriptionsandnames(forFregebothclassedas‘propernames’)haveasenseaswellasareference.Inthechapteron‘Descriptions’inhis1919book,however,Russellarguesthatdefinitedescriptionsandpropernamesshouldbeclassedasverydifferententities.Thatisn’ttosaythatRusselladvocatessimplyreturningtoMill’sversion.UnlikeMill,Russellproposestorestrictthepropertyofdirectreferencetoactualpropernames(and,aswewillsee,tooneparticulartypeofuseofpropernames)andtoexcludedefinitedescriptionsfromtheclassofreferringexpressionsaltogether.Insupportofthisclaim,Russellarguesthatitisn’tpossiblesimplytosubstitutenamesfordefinitedescriptionswhichapparentlyhavethesamereference.LikeFrege,heconsiderstheevidencefrom‘identitystatements’.Both29)and30)mayinfactbetruebut,asRussellputsit,‘thefirstisafactofliteraryhistory,thesecondatrivialtruism’(174).29)ScottistheauthorofWaverley.30)ScottisScott.\nWordsandthings33AccordingtoRussell’sexplanation,thedifferencebetween29)and30)isduetoadifferencebetweenScottononehandandtheauthorofWaverleyontheother,whichismoresignificantthanjustadifferenceinFregean‘sense’.ThisdifferenceisbasedontheapparentlystraightforwardfactthattheauthorofWaverleyisamorecomplexexpressionthanScott.Anordinarypropername,suchasScott,iswhatRusselldescribesasa‘simplesymbol…directlydesignatinganindividualwhichisitsmeaning’(174);itcan’tbeanalysedintoindividualparts.AdefinitedescriptionsuchastheauthorofWaverley,however,ismadeupofanumberofparts,eachofwhichitselfhasmeaning;themeaningofthewholeexpressionisdependentonthefixedmeaningsoftheseindividualparts.SoRussellsubscribestothe‘directreference’accountofmeaningforpropernames;theydenotesomeparticularindividual,suchastheparticularauthor.Definitedescriptions,however,don’treferto,butratherdescribe,anindividual.Whentheyappearinasentencetheireffectistospecifythatanindividualmeetingacertaindescriptionexists,sothatsomethingcanthenbesaidaboutthisindividual.Sincepropernamesservesimplytodenotesomeindividual,it’sessentialthattheremustexistsomeindividualforthemtodenote.Anamewhichdoesn’tdenoteissimplymeaningless.Thesameisnotthecasefordefinitedescriptionswhichdon’tdescribe.Russell’sexamplesarethegoldenmountainandtheroundsquare.Therearen’tinfactindividualswhichcorrespondtothesedescriptions,butthedescriptionsthemselvesarestillmeaningful,‘thereasonbeingthatitisacomplexsymbol,ofwhichthemeaningisderivedfromthatofitsconstituentsymbols’(179).Russellalsodiscusseswhathappenswhendescriptionssuchastheseoccurinsentences;wewilllookatthispartofhistheoryinChapter2.Suchanaccountofpropernameswouldseemtoraiseproblemsfortheuseoffictionalorhistoricalnames.Itmightbequiteeasytoagreethatacompletelymade-upname,whichreferstonothing,ismeaningless,butwecanhardlyclaimtobepuzzledastothemeaningofexamplessuchasthefollowing:31)Hamletsawhisfather’sghost.32)Homerwasblind.Russelldiscussesboththesepropernames,althoughnottheseactualexamples.Weknowwhat31)means,eventhoughweknowthatHamletisacharacterinfictionwhoneveractuallyexisted.Weknowwhat32)means,eventhoughweacceptthatthereisnoconclusiveevidencethatthepersonwecallHomereverlived.ButRussell’sexplanationisnottosaythatHamletandHomerdoexist,ifnotinreallifethenatleastinimaginationorinfiction;hevehementlyrejectsanysuchexplanation,callingit‘amostpitifulandpaltryevasion’(169).Rather,hesuggeststhatinsuchcaseswhatappeartobepropernamesarenotinfactbeingusedaspropernames,to\n34Wordsandthingsreferdirectlytoindividuals,butratheras‘abbreviateddescriptions’;thenamesactuallystandinforthedescriptionsweassociatewiththem,andservetodescribeanindividualsothatsomethingcanbesaidofit.Whatthesedescriptionsarewillofcoursevaryfromindividualtoindividual,asFregenotedinhisdiscussionofAristotle,butforalotofpeopleHamletmightbeanabbreviationfor‘theheroofShakespeare’smostfamoustragedy’,andHomerfor‘theauthoroftheIliadandtheOdyssey’.Usingthesenamesastheyareusedinexamples31)and32)thereforeinvolvesthementalassociationofthenameswiththesedescriptions.Russellinfactsuggeststhatpropernamesfunctionasabbreviateddescriptionsinmostuses.Theoccasionswhentheyfunctionaspropernamesmustbelimitedtothoseoccasionswhenyouactuallyhavedirectsensoryexperienceofsomeone,forinstanceyoucanseethem,andyouassociateanamewiththatperson.Inthesecasespropernamescanbesaidtoreferdirectlytoindividuals.Butinmanyothercases,evenwhenusinganameforsomeoneyouknowpersonally,youwillprobablyactuallyassociateadescriptionwiththatname,suchas‘mybrother’,‘thelocalpostman’,‘thewomanIsitnexttoateveningclass’,andsoon.Inthesecases,usingthename(John,MrJones,Evelyn)willinvolveapplyingthisdescription,ratherthanreferringdirectlytotheindividual.Russellsuggeststhat‘thesameconsiderationsapplytoalmostallusesofwhatlooklikepropernames’(179).RememberthatRussell’sinterestinlanguagewasderivedfromhisworkonlogicontheonehandandthetheoryofknowledgeontheother.Hewasnotparticularlyconcernedwithlanguageasameansofcommunication;theseclaimsabouttheusesofpropernameshavemoretodowithwhatgoesoninthemindsofindividualswhentheyuse—oreventhinkabout—aname,thanwiththewaysinwhichnamescanbeusedtocommunicateinformation.It’srelativelyeasytoseehowanamemaybeassociateddirectlywithsomeonewhoispresent,orindirectlywiththedescriptionofsomeonewhoisabsent,whentheseassociationstakeplaceinyourownmind.It’sperhapshardertoseeanyoftheseprocessesgoingonwhenyouusesentencessuchasthosewehavebeenconsideringtocommunicatewithsomeoneelse.Aswewillseeinthenextchapter,Russellwasexplicitinhispurely‘philosophical’,orlogicalinterestinlanguage.Nevertheless,it’stheissueofhownamesareinfactused,andtheirroleincommunication,whichhasledtosomeofthestrongestcriticismofhisaccount.KripkeonnamingPerhapsthebestknownaccountofnamesofferedasanalternativetothoseputforwardbyFregeandRussellisthatproposedbySaulKripkeanAmericanphilosopherwhobecameaprofessoratPrincetonUniversityin1977.Inanarticlepublishedin1972called‘Namingandnecessity’,Kripke\nWordsandthings35identifiesproblemsforanyaccountofnameswhichrelatesthemtodescriptions.12TheseareproblemsforFrege’sclaimthatnames,aswellasdescriptions,havebothreferenceandsense,andforRussell’ssuggestionthatmostofwhatappeartobepropernamesareactuallyactingas‘abbreviateddescriptions’.Wehavealreadyseenthatsuchdescriptionsmustbeallowed,innaturallanguage,tovaryfromindividualtoindividual.FregeallowedthatforonepersonthenameAristotlemayhavethesense‘thepupilofPlatoandteacherofAlexandertheGreat’andforanother‘theteacherofAlexandertheGreatwhowasbornatStagira’.It’salsothecasethatsuchdescriptionsaren’talways‘reliable’.Wecanimaginecertainancientdocumentscomingtolightwhichindicate,forinstance,thatAlexandertheGreatwasnevertaughtbyAristotle.Thismightwellmakeusrevisecertainofourviewsonclassicalhistory,butitwouldhardlyforceustorethinktheactualmeaningofthewordAristotle.Wemightthinktoourselves‘soitturnsoutthatAristotlewasn’ttheteacherofAlexandertheGreat’,butnot‘soitturnsoutthatAristotlewasn’tAristotle’.Kripkeoffersthefollowingexampleofthis:ColumbusmightbeidentifiedbymanypeopleasbeingthefirstEuropeantolandinAmerica.It’squitelikelythatthiswasn’tinfactthecase;it’squitelikelythatAmericawasreachedbytheVikingslongbeforeColumbus.However,peoplewhoassociatethatdescriptionwithColumbusdon’tusethetermColumbustoreferto‘someNorseman’(264).13TheyuseittorefertotheparticularhistoricalfigurecalledColumbus;theyjusthappentohaveafalsebeliefabouthim.Suchbeliefsneedn’tinfactbetrueofanyindividualforanametobemeaningful:Biblicalscholars…thinkthatJonahreallyexisted.Itisn’tbecausetheythinkthatsomeoneeverwasswallowedbyabigfishorevenwenttoNinevehtopreach.Theseconditionsmaybetrueofnoonewhatsoeverandyetthename‘Jonah’reallyhasareferent.(264)Kripke’sexplanationisthatnames,unlikedescriptions,are‘rigiddesignators’.LikeMillbeforehim,heseesnamesasdirectlydenotingindividuals,withoutthemediationofanysenseordescription.Theconnectionbetweennameandindividualisunaffectedbyanyincidentalchangesinwhatdescriptionscanappropriatelybeappliedtotheindividual,andindeedisthesameforallspeakersofalanguage,regardlessofindividualdifferencesinknowledgeorpointofview.Kripke’semphasisisthereforeonthewayinwhichanameisusedbyacommunityofspeakers,oftenthesetofspeakersofaparticularlanguage,torefertoaparticularindividual.Centraltohisaccountistheuseofanameincommunication,andinparticularthewayinwhichanindividualspeakercomestouseanameonacertainoccasionsothatitisunderstood.Thisaccountissometimesdescribedasthe‘causation’theoryofnames,andcanbesummarisedasfollows.\n36WordsandthingsEveryindividualwhichhasanameisgiventhisnameonaparticularoccasion.Kripkereferstothisasan‘initialbaptism’,butitneedn’ttaketheformofanofficialceremony.Itmaybe,forinstance,whenparentsfirsttelltheirfamilywhattheyhavedecidedtocalltheirchild,orwhenatowncouncilpassesamotiontogiveaparticularnametoanewroad.Aftersuchanoccasion,thenameispassedonwithinacommunityofspeakerssimplybybeingusedtorefertothatindividual.Sowhenanyspeakerusesanametorefertoanindividual,itisintheorypossibletotraceacontinuous‘chainofcommunication’,inwhichthenamehasbeenusedwiththesamereferencerightfromthetimeatwhichitwasoriginallygiven.InKripke’swords,‘it’sinvirtueofourconnectionwithotherspeakersinthecommunity,goingbacktothereferenthimself,thatwerefertoacertainman’(265).Themeaningofanameisthereforeassignednotbyindividuals,asitisinRussell’saccountofnamesasabbreviateddescriptions,butbyanentirecommunityofspeakers.Thisexplainswhyspeakerscanusenames,andusethemsuccessfully,evenwhentheywouldn’tbeabletoofferadefinitedescriptionwiththesamereference.Forinstance,Kripkesuggeststhat‘themaninthestreet’mightusethenameFeynman,andwhenasked,maysay‘wellhe’saphysicistorsomething’.Kripke’sclaimisthateventhoughhecan’tproduceauniquereferringdescription,heisusingthenameappropriately,thatis‘heusesthename“Feynman”asanameforFeynman’(262).Thisispreciselybecauseofthewayinwhichhehasheardthenameusedbyotherspeakersinthecommunity.Kripke’sworkonnamingwasproducedmorethanhalfacenturyafterRussellstartedtakinganinterestinmeaning,andlongerstillafterFregeappliedhisinterestinmathematicstotheproblemofidentitystatements.Wecertainlyshouldn’tdismissFregeandRussell’saccounts;indeedwewillreturntotheminthenextchapter,andassesstheirimpactonanotherareaoflinguisticenquiry.ButitisKripke’sworkwhichhasbeenthefocusofinterestforlinguists.Althoughhetooisprimarilyalogician,Kripkedemonstratesanimplicitunderstandingoftherelevanceof‘speakers’and‘communities’whenitcomestodiscussingmeaning.Heshowsanawareness,whichistakenforgrantedinmanybranchesofmodernlinguistics,thatmeaningisn’tjustsomethingwhichisattachedtoacertainword,orcombinationofwords,butsomethingwhichinvolvestheindividualspeakers,andthecommunityofspeakers,ofalanguage.Aswehaveseen,Frege’sinterestinmeaningrelatescloselytohisinterestinmathematics,andRussell’stohisinterestinlogicandknowledge.Inneitheroftheseinterestsiscommunicationofprimaryrelevance.WordsandthingsinlinguisticsThephilosophicalworkswehavestudiedinthischapter,andparticularlytheaccountsofnamesinthelastsection,havebeeninfluentialinanumber\nWordsandthings37ofareasoflinguistics.Aswewillseeinlaterchapters,theyhaveinformedworkonthewaysinwhichwordscontributetomakingsentences‘true’or‘false’,andontherelationshipbetweenlanguageandmind.But,aswesawattheoutsetofthischapter,linguistsaswellasphilosophershavecausetobeinterestedintherelationshipbetweenwordsandthingsinitsownright.Theareaoflinguisticsconcernedwiththemeaningsofindividualwordsofalanguage,andtherelationshipsbetweenthem,isthebranchofsemanticsknownaslexicalsemantics.Theserelationshipsinclude,forinstance,thoseofsynonymy(twowordswiththesamemeaning),antonymy(twowordswithoppositemeanings),polysemy(onewordwithmanymeanings)andhomophony(twoseparatewordswhichsoundthesame).Theclassificationofsuchrelationshipsmightappeartoberelativelystraightforward,butinmakingdecisionsaboutthemtheissueswhichwehaveexploredinthischapterarehighlyrelevant.Iftwowordsaresynonymous,forinstance,theyshouldintheorybeinterchangeableinallcontexts.Wemightwanttoclaimthatthewordsladandboyaresynonymsbecausetheyhavethesamemeaning;afterall,theycanbothbeparaphrasedas‘younghumanmale’.Butoncewebecomeawareoftheissueswehavebeenstudyinghere,thecasebecomesalotlessclear-cut.Tobeginwith,thetwowordscertainlydon’tseemtobeexactlyintensionallyequivalent.Toseethisweonlyneedtotrythemoutinasetofintensionalcontexts,suchasthestatementsofprepositionalattitudein33)and34):1433)Guslikestothinkthatheisabitofalad.34)Guslikestothinkthatheisabitofaboy.Nowitcertainlyseemstobethecasethat33)couldbetruewhile34),inwhichanapparentlysynonymousexpressionhasbeensubstituted,couldbefalse.Ladandboyarenotintensionallyequivalentbecauseonehasassociationswhichtheotherlacks.Andindeed,onceweconsidertheextensionsofthewords,theexactsetofindividualstheycouldbeusedtoreferto,wewouldprobablybereluctanttoagreethattheyhaveexactlythesameextensionalmeaningeither.Thewordlad,forinstance,mightgenerallybeagreedtoallowforahigherupperagelimitthanboy.Thissuggeststhatitsextensionisalargersetofindividuals,whichinturnsuggeststhatthetwoarenotextensionallyequivalent.Thesameseemstohold,toagreaterorlesserdegree,formanypairsofwordswhichareconventionallyclassedassynonyms.Thinkofpairssuchassnake/serpent,carlautomobile,hall/vestibule,oranyofthewordsgroupedtogetherinathesaurus,andyouwillprobablybeabletodetectintensionaland/orextensionaldifferencesbetweenthem.Ithasbeensuggestedthatnonaturallanguageinfactincludesanyexactsynonyms,pairsofwordswhicharebothextensionallyandintensionallyequivalent.Indeed,itwouldseemtointroduceanunnecessaryredundancyintoalanguage;iftwowordshadexactlythesamemeaning,oneofthemwould\n38Wordsandthingsbeunnecessary.Forsimilarreasons,ithasbeensuggestedthatexacttranslationbetweentwolanguagesisneverinfactpossible.Nowordinthefirstlanguagecouldeverhaveaprecisecounterpart,intermsofbothextensionandintension,inthetargetlanguage.Definitedescriptions,ofthetypeidentifiedbyBertrandRussell,havecontinuedtobeasourceofinteresttophilosophersandlinguistsalike.Wewilllookatsomeoftheirwork,particularlyasitrelatestothewaysinwhichspeakersusedefinitedescriptionsindifferentcontexts,inlaterchapters.Assuggested,muchofthecriticismofRussell’saccounthasbeenfocusedonitsinabilitytodealwith‘reallife’usesofdefinitedescriptions.Asjustoneexampleofthetypeofproblemwhichhasbeenraised,considerthefollowingquestion:35)WhoisthemandrinkingaMartini?Wecanimaginethisquestionbeingaskedattwopartiesbytwodifferentpeople,givingtwoverydifferentinterpretations.Inonecase,atafashionablereception,awomannoticesamanwithacommandingairdrinkingaclearliquidfromaMartiniglass.Sheutters35)toafellowguestinordertofindoutwhetherheisanyonefamous.SheusesthedefinitedescriptionthemandrinkingaMartinisuccessfullytorefertothatparticulardistinguished-lookingpersonifherfellowguestrecogniseswhoshemeans;itdoesn’tmatterthatitinfactturnsoutthatallhehasintheglassiswater.Thesecondsceneissetattheannualgatheringofalocaltemperancesociety.SomeonehasjustinformedthepresidentthatamanhasbeenspottedslylydrinkingaMartini.Theoutragedpresidentutters35);hisintentionistodiscovertheidentityofthemiscreant,andheusesthedefinitedescriptiontorefertothischaracter,whoeverhemaybe.ThisexampleisdiscussedbythephilosopherKeithDonnellaninhis1966essay‘Referenceanddefinitedescription’.Helabelsthefirstcaseanexampleofthereferentialuseofadefinitedescription.Thesecond,whichhearguesisacommontypeofuseofdefinitedescriptions,butonenotconsideredbyRussell,helabelsanexampleoftheattributiveuse.Beforewemoveontostudytherelationshipbetweenlanguageandtruth,it’sworthsparingathoughtfortheideationalaccountsofmeaning.Recallthattheseareaccountswhichclaimthat,sinceallwehavecertainknowledgeofisourownideasandimpressions,itmustbethesewhicharethemeaningsofwords.Thelastweheardofideationalaccountsofmeaningwasintheeighteenthcentury,whentheyseemedrathertobefounderingontheproblemsposedbygeneralideas,andontheextremesofimmaterialism.Buttheyweren’tentirelysunk.Indeed,thenotionof‘mentalimage’asmeaningbecamecentraltotheworkoftheStructuralistlinguistsofthetwentiethcentury,ofwhomFerdinanddeSaussureisperhapsthemostfamous.Wewillbestudyingsomeofthisworkinthefinalchapter.\nWordsandthings39FurtherreadingTheBritishEmpiricistsLocke’sAnEssayConcerningHumanUnderstandingisavailableinanumberofmoderneditions.TheEverymanversionusedhereisslightlyabridged,butcontainsallthepartsrelevanttoLocke’saccountoflanguage.ThisissetoutchieflyinBookIIIofthework,‘Ofwords’,partofwhichisreprintedinMartinich(1996):500–5.GeorgeBerkeleyactuallywrotetwopartstoThePrinciplesofHumanKnowledgebutlostthemanuscriptofPartIIbeforepublicationandneverrewroteit.Thepublishedversion,whichisavailableinanumberofmoderneditionsincludingPenguin,thereforeconsistsonlyoftheIntroductionandPartI.Itisshortandlively.IdeationalaccountsofmeaningareconsideredinHacking(1975)WhyDoesLanguageMattertoPhilosophy?inhissectionon‘Theheydayofideas’,includingreferencetosomephilosophersnotdiscussedhere.UseandmentionSearle(1969)SpeechActsincludesadiscussionof‘Referenceasaspeechact’whichbeginswithastaunchcriticism(73–6)ofthenotionthattomentionawordbyproducingitininvertedcommasistogiveanameforthatword.Searlearguesthattheworditselfisproduced,butnotinthenormalway;itispresentedfordiscussion.Anotherattackonthedistinctionbetweenuseandmention,thistimeonewhichdealsspecificallywithLinsky’saccount,canbefoundinRoyHarris’(1996)Signs,LanguageandCommunication(100–4).NamesMill’saccountofnamesissetoutinBookIofASystemofLogic,mainlyinChapter2.ReferencesheretoFrege’s‘Onsenseandreference’aretoGeachandBlack(1980)ThePhilosophicalWritingsofGottlobFrege.Russell’sessay‘Description’formsachapterofhis(1919)IntroductiontoMathematicalPhilosophy.Kripke’s‘Namingandnecessity’wasoriginallypublishedinabookofthesamenamein1972.Russellconsidersreferenceinanumberofhisotherphilosophicalworks.Forinstance,namesanddescriptionsarediscussedinPartIIofHumanKnowledge(1948)andpropernamesarethesubjectofChapter6ofAnInquiryintoMeaningandTruth(1940).Alotofworkhasbeenproducedinthisarea.Carnap(1956)MeaningandNecessitygivesadetailedanalysisofthetermsintensionandextension,andcomparesthemwiththeworkofRusselland,inparticular,ofFrege.\n40WordsandthingsEvans(1982)TheVarietiesofReferenceoffersadetailedbutdifficultaccount,withseparatechaptersofPartIdevotedtoFrege,toRussellandtorecentworksuchasthatofKripke.HelinksRussell’saccounttohisinterestinthephilosophyofthemindindetail.Evanshimselfarguesthatthereare‘non-Russellianreferringexpressions’,inotherwordsthatnamesandexpressionswhichRussellwouldclassasdescriptionscaninfactbeexplainedintermsofreference.OnephilosopherwhohaswrittenabouttheparticularproblemsposedbyopaquecontextsistheAmericanW.V.O.Quine,whoseworkwewilllookatinthefinalchapter.Hedealswiththesetopicsinhis(1961)bookFromaLogicalPointofView.Muchofhisargumentthere,includingthesuggestionthatitisnecessarytoconsultintensionwhenconsideringstatementsofbelief,canbefoundinhis(1956)article‘Quantifiersandpropositionalattitudes’.Manyexpositoryaccountsofthediscussionofnamesareavailable.R.Martin(1987)TheMeaningofLanguagediscussesRussell’sandKripke’saccountsofpropernamesinChapters16and17.Stainton(1996)PhilosophicalPerspectivesonLanguage,Chapter3,isconcernedwithdirectreferenceaccountsingeneralandRussell’saccountofdescriptions.ThedebateistreatedinsomedetailinPartIIofDevittandSterelny(1987)LanguageandReality.\n2PropositionsandlogicIntroductionSofarwehavebeenconcernedalmostentirelywiththewayinwhichlanguagenames,orpicksout,individualobjects.Wehavelookedatsomepossibleexplanationsoftherelationshipbetweenthematerialworldandthewordsusedtodescribeit.Butofcourselanguagedoesn’tjustprovidenamesforobjects.Beingabletousealanguageinvolvesmuchmorethansimplyknowingalonglistofwordsandmeanings.Imaginethatagroupofaspiringlinguistssucceedinlistingallthewordsofalanguage,andgivingtheirintensionalandextensionalmeanings.Theywouldstillhavealotofworklefttodobeforetheycouldclaimtohavedescribedwhatitisthatpeople‘know’whentheyareablesuccessfullytouseandunderstandthelanguage.Ifweweretryingtoexplaintothestrugglinglinguistswhatwasmissingfromtheirdescriptionofalanguage,wewouldprobablyfindourselvessayingsomethinglike,‘Allyou’vedoneisproducealistofthewordsinthelanguage.Youhaven’tsaidanythingaboutthewaysinwhichthewordscancombinetogether,ortheeffectsonmeaningwhentheyarecombined’.Ineffect,wewouldbecomplainingthatthelinguistshadsaideverythingtheycouldaboutthewordsofthelanguage,butnothingatallaboutitsphrasesandsentences.Theseare,ofcourse,thesubjectsofagreatdealofanalysisanddescriptioninlinguistics.Thewaysinwhichwordscanlegitimatelycombinetogethertoformphrases,clausesandsentencesarestudiedinsyntax.Themeaningswhichresultfromthesecombinationsarethesubjectofsemanticanalyses.Andhere,asinthestudyoftherelationshipbetweenwordsandthings,linguistshavetakenuptopicsfirstdiscussedinphilosophy.Atthebeginningofthelastchapterweconsideredlinguisticsasadisciplinewhichnecessarilyconcernsitselfwithboththerelationshipsbetweenlanguageandtheoutsideworldandtherelationshipswithinalinguisticsystem.Inthestudyofsentencemeaning,boththesetypesofrelationshiparerelevant,andbothhavebeenconsideredbyphilosophersoverthecenturies.Weareconcernedwiththerelationshipbetween41\n42Propositionsandlogiclanguageandtheworldwhenweaskwhatitisthatmakesasentenceanappropriatedescriptionofaparticularstateofaffairs:essentially,whenweconsiderwhatitisforasentencetobe‘true’.Thiswillbeourmainconcerninthenextchapter.Whenweconsidertherelationshipswhichexistbetweenthesentencesofalanguage,wearealsoprimarilyconcernedwithrelationshipswhicharebasedonthepropertiesoftruthandfalsity.However,inthiscaseweareconcernednotsomuchwithwhatmakesindividualsentencestrueorfalse,butwiththewaysinwhichtruth-valuescanbepredictedforgroupsofrelatedsentences.Forthetimebeing,let’sassumethateverysentenceinalanguagecanbelabelledeither‘true’or‘false’.Thetruth-valueofanysentence,then,iswhicheveroftheselabelsapplies.Thereareonlytwopossibletruth-valuesforanysentence:trueandfalse.Speakersofalanguageinfactknowalotaboutrelationsbetweentruth-values.AsaspeakerofEnglish,youknowthatif1)belowistrue,then2)mustalsobetrue,while3)mustbefalse.1)Braeburn-TwinsettownsaVolvoEstate.2)Braeburn-Twinsettownsacar.3)Braeburn-Twinsettdoesn’townacar.IfyouaskedanyspeakerofEnglishtoexplaintheconnectionbetween1)and2),theywouldprobablysaythat2)‘followslogically’from1),orthat‘2)hastobetrueif1)istrue’.Thisparticularrelationshipbetweensentencesisknownasentailment.Itiswhatisknownasatruth-functionalrelationship;thefunction,oreffect,ofentailmentistopredictthetruth-valueofonesentencefromwhatisknownofthetruth-valueofanother.Example3)containsnegation,whichisalsotruth-functional,becausewecanpredictthetruth-valueof3)ifweknowthetruth-valueof1);if1)istruethen3)mustbefalse.Ontheotherhand,knowingthat1)istruewillleaveusnonethewiseraboutthetruth-valueofanexamplesuchas4).4)Braeburn-Twinsettlivesinadetachedhouse.Wecouldexplainthisbysayingthatthereisnotruth-functionalrelationshipbetween1)and4).Wemightbeabletomakeinformedguessesaboutwhether4)islikelytobetrue,butwewouldhavenofirmevidence.Truth-functionalrelationshipsbetweensentencesarestudiedinthebranchofphilosophyknownaslogic.Logicisprimarilyconcernednotwithlanguagebutwithformalrelationsinvolvingconstantsandvariables.Butbecausetheserelationscantellussomethingaboutwhatisgoingonwhenweinterpretsentencessuchas1)–3)above,logicisanimportantareaofphilosophyforlinguists.Thereareothertypesofrelationshipbetweensentenceswhichspeakersofalanguagecan,ingeneral,agreeon,andwhichlinguiststhereforeneedtobeabletoexplain.Youmayhavenoticed,forinstance,thatifyouare\nPropositionsandlogic43assuredthat1)isatruesentenceyoulearnmorefromitthansimplythecontentsofBraeburn-Twinsett’sgarage.YouarealsoledtounderstandthatBraeburn-Twinsettmustactuallyexist.Afterall,ifhedidn’texist,wecouldn’ttruthfullysayanythingabouthim.Inotherwords,fromourexample1)wealsounderstandthefollowingpieceofinformation:5)Braeburn-Twinsettexists.Nowthewayinwhich1)leadsustounderstand5)isratherdifferentfromtherelationshipofentailmentwhichholdsbetween1)and2).Braeburn-Twinsett’sexistence,andhisownershipofacar,arebothnecessaryconditionsfor1)tobetrue.Butwhileusingsentence1)whenyouknewBraeburn-Twinsettdidn’townacarwouldjustinvolvesayingsomethingfalse,usingitwhenyouknewthatnosuchpersonactuallyexistedwouldinvolvedoingsomethingveryoddindeed.Thisrelationshipbetween1)and5)isknownaspresupposition.Asfurtherevidencethatpresuppositionis,atleast,differentfromlogicalentailment,considerthefollowing:6)Braeburn-Twinsettdoesn’townaVolvoEstate.Example6)isthenegationof1),andasaresultitdoesn’tshareitsentailments.Specifically,2)isnotalogicalentailmentof6).But,incontrast,6)doesstillseemtosuggestthetruthof5).Sowhilenegationhasastrongeffectontherelationshipbetweenasentenceanditsentailments,ithasnosucheffectonpresupposition.Presuppositionhasbeenthesubjectofmuchdiscussioninlinguistics,andistalkedofvariouslyasarelationbetweenwords,betweensentencesandbetweenspeakers.Buttherootsofthisdiscussionareplantedveryfirmlyinphilosophyandinfact,asweshallsee,intheworkofsomephilosopherswithwhomwearealreadyfamiliar,includingGottlobFregeandBertrandRussell.Thischapter,then,willbelargelyconcernedwiththedifferenttypesofrelationshipwhichcanexistbetweensentences,andwiththewaysinwhichwhatweknowaboutthetruth-valueofonecantellusaboutthetruth-valueofother,relatedsentences.Inordertounderstandwhathasbeensaidabouttheserelationships,wewillneedtoexploresomebasicideasfromphilosophicallogic.First,though,weneedtobeabitmorefamiliarwithwhatcanbesaidingeneralaboutsentencemeaning.Inparticular,weneedtothinkaboutpropositions,andtheirsignificanceindistinguishingbetweenindividualsentences.PropositionsWesawinthepreviouschapterthatthemeaningsofwordsareoftendescribedasbeingmadeupofbothanextension,anactualobjectorpropertynamed,andanintension,thecharacteristicsbywhichthatobject\n44Propositionsandlogicorpropertyisdistinguished.Infactaverysimilarclaimhasbeenmadeforsentences.Inthecaseofasentence,theextensionisnotanobjectorproperty,orevenastateofaffairs;itisatruth-value.Wehaveseenthattherearegenerallysaidtobejusttwotruth-values:trueandfalse.Soitfollowsthateverysentenceinalanguagehasoneoftwoextensions;itiseithertrueorfalse.Now,wecan’texplainsentencemeaningjustintermsofthesetwoextensions.Ineffect,thiswouldleaveusnowayofdistinguishingbetweenalltruesentencesontheonehand,andallfalseonesontheother.Ifweexplainedsentencemeaningonlyintermsofextension,andevenifwecouldacceptthat7)–9)arealltrue,wewouldn’tbeabletosayanythingaboutthedifferencesinmeaningbetweenthem.7)Pandasareblackandwhite.8)TheEarthislargerthanMars.9)ColumbusdiscoveredAmerica.Thiscertainlyseemstobeanunacceptablestateofaffairs.Despiteallhavingthesameextension,it’sclearthat7),8)and9)haveindividualmeaningswhicharedistinctfromeachother.Furthermore,it’sclearthat8)meansthesameas10),while9)meansthesameas11),butthatneither10)nor11)meanthesameas7).10)MarsissmallerthantheEarth.11)AmericawasdiscoveredbyColumbus.Whenweareconcernedwithmeaningwhichdiffersfromonetruesentencetothenext,weareconcernedwithadifferentaspectofsentencemeaning:withintensionalmeaning.TheproposalthattheoverallsignificanceofasentenceneedstobedividedintotwoseparateelementscanbefoundintheworkofGottlobFrege.Inhis1892article‘Onsenseandreference’whichwestudiedinthelastchapter,heproposesthatthesenseofasentenceisaparticular‘thought’.Healsodescribesthe‘truth-valueofasentenceasconstitutingwhatitmeans’(204).RememberthatforFregewhatsomethingmeansisits‘reference’,orextension.Hegoesontoexplain:Ifnowthetruth-valueofasentenceisitsmeaning,thenontheonehandalltruesentenceshavethesamemeaningandso,ontheotherhand,doallfalsesentences.Fromthiswecanseethatinthemeaningofasentenceallthatisspecificisobliterated.Wecanneverbeconcernedonlywiththemeaningofasentence;butagainthemerethoughtaloneyieldsnoknowledge,butonlythethoughttogetherwithitsmeaning,i.e.itstruth-value.(204)\nPropositionsandlogic45Fregespecifiesthatwhatformsthesense,orintension,ofasentenceistobeunderstoodasthe‘content’ofathought.Thisiswhatisoftendescribedasaproposition.Inphilosophyandlinguistics,asinsomeofits‘everyday’uses,thetermisoftenassociatedwiththoughts,statements,claimsandbeliefs.Propositionsaresometimesdescribedasabletobeintroducedby‘that…’clauses.If,forinstance,youbelievethatMrsSolomon,yournext-doorneighbour,iswise,thenthepropositionthatMrsSolomoniswiseisthecontentofyourbelief,andyoucanexpressthisbeliefinanyofthefollowingways,amongothers.12)MrsSolomoniswise.13)Mynext-doorneighbouriswise.14)MadameSolomonestsage.Thesentences12)–14)arecertainlynotidentical,buttheyallcontain,orexpress,thesameproposition.Toputitanotherway,13)and14)areintensionallyequivalentto12):13)bymeansofparaphrasewithinthesamelanguage,and14)bymeansoftranslationintoanotherlanguage.1Ifwewereaskedwhether12)–14)weretrue,wewouldneedtofindoutaboutthetruthorfalsityofthisoneproposition.IfitinfactturnsoutthatMrsSolomonisnotwise,thenthebelief,andthereforeallthesentencesusedtoexpressit,willbefalse.Sopropositionsareentitieswhichcanbetrueorfalse,andthereforehaveconsequencesforthetruthorfalsityofthesentenceswhichexpressthem.Rememberthattheintensionofawordcanbesaidtodeterminewhatcountsasitsextension.Injustthesamewaytheintensionofasentence,aproposition,candeterminewhetheritsextensionalmeaningis‘true’or‘false’.Itmayhelptothinkofatruepropositionasbeingequivalenttoa‘fact’.Ifyoubelieveorstateafactthenyourbelieforstatementwillbetrue.Ifwhatyoubelieveorstateisnotafact,yourbelieforstatementwillbefalse.Noticealsothatourexamples12)–14)involveasubject,bymeansofwhichanindividualisnamed,andapredicateinwhichsomepropertyisascribedtothatindividual.Theassociationofanindividualwithapropertyisoftentakentobethebasicformofaproposition.Infact,itispartofsomedefinitionsofpropositionality.Suchanassociationisgenerallyexpressedinsimpledeclarativesentences,asinexamples12)–14)above.It’simportanttobearinmindthatdeclarativesentencesarenotthemselvespropositions,butthattheygenerallyexpresspropositions.ThisimportantpointismadeparticularlyclearlybytheGermanphilosopherRudolfCarnap.Inhis(1956)bookMeaningandNecessity,originallypublishedin1947,Carnapdefinesthetermpropositionasbeingusednot‘forsentencesorforsentencestogetherwiththeirmeaningbutforthoseentitieswhichthemselvesareextra-linguisticbutwhich,iftheyfindexpressioninalanguage,areexpressedby(declarative)sentences’(26–7).\n46PropositionsandlogicCarnapisherestressingthattheintensionalmeaningsofsentences,likethoseofindividualwords,can’tthemselvesbeexplainedwithinthelinguisticsystem;forwordsandsentencestobemeaningfultheymustrelatetotheworldoutsidethissystem.Ofcoursethereistheaddedcomplicationthatwhenwewanttoproduceaproposition,forpurposesofdiscussion,wegenerallyneedtodosobyproducingadeclarativesentencewhichexpressestheproposition.Alltheexampleswehaveconsideredinthissectionhavetakentheformofdeclarativesentences.Thisisoneinstanceofthegeneralproblemwhichlinguistsandphilosophersfaceofhavingtouselanguageforthepurposeofdiscussinglanguage.Wewillbereturningtopropositionalitythroughoutthefollowingchapters,wherewewillconsideritsimplicationsinmoredetail.Forthetimebeing,however,weareequippedwithenoughinformationtoenableustoexaminethewaysinwhichpropositionshavebeenusedinformallogic.FormallogicThetermlogicisoftenusedinnormal,non-philosophicalconversationtomeansomethingsimilarto‘reasoning’or‘argument’.Peoplesay‘Idon’tfollowyourlogic’or‘Idon’tseethelogicinthat’whentheymeanthattheycan’tunderstandhowaparticularconclusionwasreached,oragreewithaparticularlineofargument.Inphilosophy,thetermisusedwitharelated,butmoreprecisemeaning,totalkabouthowwecangetfromaparticularstartingpoint,saythetruth-valueofonesentence,toaseparateconclusion,suchasthetruth-valueofanother,related,sentence.Weareinterestedinlogicbecauseofitssignificanceinlinguistics,butmanyofthephilosopherswhohaveworkedinthisarea,someofwhoseworkwewillconsiderinthischapter,havebeenconcernedprimarilywithmathematics,andonlysecondarilywithlanguage.Withthisinmind,it’sperhapsnotsurprisingthattheexpressionsusedinlogiccanlookratherlikemathematicalequations.Thiscanmakethemlookdistinctlyoff-puttingatfirstglancetostudentsoflanguage,butwithabitofdecodingweshouldfindthatthepartsoflogicweneedtounderstandforourpurposesarenottoodaunting.The‘mathematical’lookispartlyduetothefactthatlogicemploysvariablestostandinforspecificentities,soastobeabletomakegeneralstatements,orformulae,whichwillholdgoodwhateverspecificvaluesaregiventothevariables.Youareprobablyfamiliarwithmathematicaltermssuchasvxwhichmeans‘thesquarerootofanynumber’andx+1whichmeans‘anynumberplusone’.InjustthesamewayinlogicwefindformulaesuchasépÙqùwherethevariablespandqstandforpropositions,andthesymbolÙisalogicalversionof+.Noticethataswellastheuseofvariablestostandforpropositionsweareusingthesymbolséùinplaceofthequotationmarks‘’.Theseareknownascornerquotes,andareusedtoindicatethatwhatisbeingquotedisageneralformularatherthanaspecificsentence.Theuseofcorner\nPropositionsandlogic47quotesgenerallyimpliesthatwecouldbetalkingaboutanyexpressionwhichhasthisparticularform:anyexpressionwhichcanbeobtainedbyreplacingthevariableswithappropriatewordsorphrases.Withappropriatesubstitution,theformulaépÙqùcouldbeinstantiated,orexemplified,byanyofthefollowing,andofcoursebyanyothersimilarexpression:15)RoverhasfourlegsandTommyhastwolegs.(wherep=Roverhasfourlegsandq=Tommyhastwolegs)16)AliceistenandBenisfive.17)Myneighbour’scarrunsonpetrolandmycarrunsondiesel.Thebranchoflogicweareconsideringhere,theonemostcentrallyconcernedwithrelationshipsbetweenpropositions,isknownaspropositional,orsententiallogic.PrepositionallogicPropositionssuchas‘Roverhasfourlegs’and‘Benisfive’areknownassimplepropositions;theyeachrepresentwhatFregewoulddescribeasthe‘content’ofasinglethought.Propositionallogicislargelyconcernedwiththewaysinwhichsimplepropositionscancombinetogethertoformcompoundpropositions,propositionswhichcanbeanalysedasinvolvingthecontentsofmorethanonethought.Examples15)–17)aboveareallcompoundpropositions.Propositionallogicisalsoknownaspropositionalcalculus;itinvolvesworkingout,orcalculating,thetruth-valueofacomplexpropositionfromthetruth-valuesofitsparts.Sodoingpropositionalcalculussometimesseemsabitlike‘doingsums’withpropositions.Ifitlooksattimesasifwemightbegettingtoofarawayfromlanguage,thenrememberthatfortheoutputofpropositionalcalculustomakesenseitmustalwaysbepossibletoreplacethevariablessuchasp,q,rwithsimplepropositions.Infactthisisoftenagoodwayofcheckingwhatapropositionalformulameans.Soinstudyingpropositionallogicweareconsideringthepossibleoperations,orprocedures,whichcanbeperformedonpropositions.Propositionallogicincludesanumberofsymbolsforthese,whichareknownasoperators,orsometimesastruth-functionaloperators,becausetheytellusthetruth-valueofthecompoundpropositionfromwhatweknowtobethetruth-valuesofthesimplepropositions.Unlikethesymbolssuchasp,q,r,thesesymbolsalwaysstandforthesameoperations;theyeachhaveauniquevalue.Theseoperatorsarethereforeknownaslogicalconstants.Theoperatorswhichwewillconsiderhereareconjunction,negation,disjunctionandcondition.Theycanberelatedtothelinguisticexpressionsand,not,orandif…then,althoughtheexactnatureoftheserelationshipsissomethingwewillneedtolookintoinmoredetaillater.\n48PropositionsandlogicConjunctionLogicalconjunctionacts,asitsnamesuggests,tojointwoormorepropositionstogether.ThesymbolforlogicalconjunctionisÙ,sotheconjunctionoftwosimplepropositions,pandq,isrepresentedbytheformulaépÙqù.Othersymbolswhicharesometimesusedforthisareép·qù,éKpqù,ép&qùOrsimplyépqù.Theyallmeanthesame;wewillstickwithépÙqùforsimplicity.Wehaveseenthatoperatorssuchasconjunctionaresometimesdescribedas‘truth-functional’becausetheydeterminethetruthofacompoundpropositionfromthetruthofsimplepropositions.Nowconsiderthesimplepropositions18)and19),andthecompoundproposition20):18)Maryishappy.19)Fredissad.20)MaryishappyandFredissad.Thecompoundpropositioncombinestheinformationcontainedinboththesimplepropositions.Ifthesepiecesofinformationarebothtruethenthecompoundpropositionwillbetrue.Butifeitherpieceofinformationisfalse,thatisifeitherMaryisnothappyorFredisnotsad,thenthecompoundpropositionin20)willbefalse,becauseitwillcontainapieceoffalseinformation.IfMaryisnothappy,forinstance,thenitsimplycan’tbethecasethatMaryishappyandFredissad.Andofcourseifbothsimplepropositionsarefalsethenthecompoundpropositionwillbefalse.Inotherwords,theonlysituationinwhichépÙqùcanbetrueisifbothpandqaretrue.Inanyothersituation,épÙqùwillbefalse.Thereisamoreconcisewayofexpressingthesortofinformationcontainedinthelastparagraph.Describingthetruth-functionaleffectsofanoperatorinallpossiblesituationscanbealong-windedprocess.Thisinformationcaninsteadbecontainedinatruthtable,achartwhichgivesthetruth-valueofthecompoundpropositionineverypossiblesituation.Themeaningofanoperatorconsistsinitseffectsontruth-values,sosuchatableconstitutesadefinitionoftheoperator.Atruthtableemployscolumnstolistthepossibletruth-valuesandrowstoshowhowtheoperatorrelatesthemtoeachother.Hereisthetruthtableforconjunction:Eachcolumnisheadedwithasimpleorcomplexproposition.Eachrowrepresentsapossiblesituation,andstatesthetruth-valuesofeachpropositioninthatsituation.Sothefirstrowundertheheadingscanbe\nPropositionsandlogic49readas‘ifpistrueandqistruethenépÙqùistrue’,andthesecondas‘ifpistrueandqisfalsethenépÙqùisfalse’,andsoon.NegationNegationisdifferentfromconjunctioninthatitoperatesononlyonepropositionatatime.Informallywemightsaythatifweaddnottoapositivestatementwegetthedenial,ortheopposite,ofthatstatement.So22)and23)bothacttodeny,ortonegate,whatissaidin21):21)Themoonismadeofgreencheese.22)Themoonisnotmadeofgreencheese.23)Itisnotthecasethatthemoonismadeofgreencheese.Inpropositionalcalculus,thesymbolforlogicalnegationis~.Soifwereplacethesimplepropositionin21)withthevariablep,wecanrepresent22)or23)asé~pù.Nowwehavesaidinformallythatanegativepropositionwillbethe‘opposite’ofthecorrespondingpositive.Intermsoftruth-values,theoperationofnegationinvertsthetruth-valueofaproposition.Soifwetakeatruepropositionandnegateitwemustgetafalseone,butifwetakeafalseproposition(suchaswhatweinfacthavein21])andnegateitwegetatrueproposition(suchas22]or23]).Inotherwords,it’snotpossibleforapropositionanditsnegationtobebothtrueorbothfalse.Hereisthetruthtablefornegation:Astheheadingsindicate,eachrowgivesapossiblevalueforpandthecorrespondingvalueforé~pù.Sothetablestatesthatwheneverpistrue,é~pùisfalse,andwheneverpisfalse,é~pùistrue.Thatisthemeaningoflogicalnegation.Certainotherfactsfollowfromthis,ofcourse.Thetablerulesoutthepossibilitythatpandé~pùcouldeverbothbetrue.Rememberwesawthataconjunctionwillbefalsewheneveratleastoneofthesimplepropositionsconjoinedisfalse.ItfollowsthatacompoundpropositionoftheformépÙ~pùmustalwaysbefalse.Example24),andindeedanypropositionwiththisform,isacontradiction.224)AliceisachildandAliceisnotachild.Noticealsothatifwetakeapropositionoftheformé~pùandfeeditbackthroughthetable,negationwillagainreversethetruth-value;ifé~pùistruethené~~pùwillbefalse.Byhavingtheoppositevaluefromé~pù,é~~pùwill\n50Propositionsandlogicalwayshavethesametruth-valueasp.Soinlogic,ifnotalwaysinnaturallanguage,twonegativescanceleachotherout,or‘makeapositive’.Thiscanbeexpressedlogicallybytheformulain25).Althoughitsoundsdistinctlyodd,itisquiteeasytointerpret26)asfalse,like21):25)~~p=p26)Itisnotthecasethatthemoonisnotmadeofgreencheese.DisjunctionDisjunctionislikeconjunction,ratherthannegation,inthatitcombinestwoormorepropositions.Thesepropositionsaredescribedasdisjuncts.Thedisjunctionoftwosimplepropositions,pandq,issymbolisedasépÙqù.Disjunctioncanberelatedinmeaningtooralthough,asisoftenthecasewithlogicaloperatorsandnaturallanguageexpressions,theirmeaningsarenotalwaysequivalent.However,asthesimilaritytoorsuggests,disjunctiondiffersfromconjunctioninthatonlyone,notboth,ofthedisjunctsneedbetrueforthedisjunctionasawholetobetrue.Toillustratethis,wecanusethesimplepropositions27)and28),andthedisjunction29).27)Williamhasgonetoplaytennis.28)Marthahasgonetoherviolinlesson.29)WilliamhasgonetoplaytennisorMarthahasgonetoherviolinlesson.ImaginethatyoupassbyWilliamandMartha’shouse,spotthattheircarisnotonthedrive,andproduce29)bywayofexplanation.Ifitturnsoutthatoneofthesimplepropositions27)and28)istrue,youwouldbejustifiedinconcludingthatwhatyousaidin29)wastrue.If,however,itturnsoutthatneither27)nor28)istrue,thenyouweremistakenandwhatyousaidin29)wasfalse.Thisistheonlysituationinwhichalogicaldisjunctioncanbefalse.Thetruthtablefordisjunctionisasfollows:Thisshowsupthedifferencebetweenlogicaldisjunctionandnaturallanguageor.Thisdifferenceoccursinthefirstlineofthetruthtable.If,inourexample,itturnsoutthatinfactWilliamwasattennisandMarthawasatherviolinlesson,youmightwellbeaccusedofhavingbeenwronginstating29);youdidn’tallowforthepossibilitythatbothdisjunctsmightbe\nPropositionsandlogic51true.Butthetruthtableforlogicaldisjunctionallowsforthedisjuncttobetruewhenbothdisjunctsaretrue.ConditionFinally,weneedtoconsiderlogicalconditionalswhich,asnotedearlier,canberelatedtonaturallanguageif…then.Astatementoftheform‘ifp,thenq’isrepresentedinlogicasépÙqù.Aswithconjunction,therearesomealternativewaysofsymbolisingthisoperator,andyoumaysometimescomeacrosstheformsépÙqùoréCpqù,whichareexactlyequivalent.Thetwopropositionsinvolvedinaconditionalareknownastheantecedent(p)andtheconsequent(q).Imaginethatafriendsaystoyou:30)IfyouwashthecarIwilltakeyououtforlunch.Itwould,ofcourse,beperfectlyreasonabletoexpectthat,sometimeafteryouhavewashedthecar,youwillbetakenoutforlunch.Soreasonable,infact,thatifyouwashedthecarbutwerethendisappointedinthisexpectation,youwouldbejustifiedinclaimingthatyourfriendhadlied:that30)wasfalse.Fromthiswelearnthattheconditionalasawholeisfalsewheretheantecedentistruebuttheconsequentfalse.Nowimaginethatyoudecidethatyourfriendisnotofferingaverygoodbargain,thatyoudeclinetowashthecar,andthatyouarethentakenoutforlunchanyway.Youwillprobablybepleasantlysurprised,butthisturnofeventsdoesn’tmake30)false.Infactwecangofurtherandsaythatiftheantecedentisfalse(ifyoudon’twashthecar),thentheconditionalasawholewillbetruewhateverhappens.Thismayatfirstseemtobearatherstrangeclaimtomake.Wemightbeinclinedtowonderhow30)canbeatruestatementifyourefusetowashthecar.Well,itwillbetruejustbecauseitisconcernedonlywiththeconsequencesofyourwashingthecar;ithasnothingtosayaboutwhatwillhappenifyoudon’twashit,sonopossibleconsequencecanmakeitfalse.Sothetruthtableforlogicalconditionalisasfollows:Thereisoneparticularconsequenceofthiswhichseemsbizarrewhentranslatedintonaturallanguagesentences,butisperfectlyconsistentlogically.Aswehaveseen,aconditionalwillalwaysbetruewhenitsantecedentisfalse.Itfollowsfromthisthatanyconditionalwithafalse\n52Propositionsandlogicantecedentwillbetrue,regardlessofthenatureofitsconsequent.So31)and32)mustbothbedescribedastruestatements:31)IftheworldisflatthenLondonisthecapitalofEngland.32)Ificesinksinwaterthenwhalesarefish.Thisisbecauseapurelylogicalconditionalisconcernedonlywithtruth-functionalrelationships,andnotatallwiththenatureof,orpossibleconnectionsbetween,thepropositionspandq.Itissometimesreferredtoasamaterialconditional,anamesuggestedbyBertrandRussell.Withnaturallanguageif,ontheotherhand,wegenerallyexpecttheantecedentandtheconsequenttoberelatedinsomeway,forinstanceintermsofcauseandeffect.Beforewefinishourbriefsurveyoflogicaloperatorsweshouldconsiderthecurious-lookingexpressioniff,whichalsooccursincomplexpropositions,andwhichneedstobeclearlydistinguishedfromtheexampleswehaveconsideredsofar.Putbriefly,iffisanabbreviationforifandonlyif.Nowtheexpressionifandonlyifdoesoccurinnaturallanguage,althoughit’slesscommonthanthesimpleif,anddescribesadifferentrelationship.Imaginethat,insteadof30)above,yourfriendhadsaidtoyou:33)Iwilltakeyououtforlunchif,andonlyif,youwashthecar.Youwouldstillbejustifiedinexpectingtobetakenoutforlunchafterwashingthecar,butthistimeyouhavenoreasontohopethatyourfriendmighttakeyououtforlunchevenifyoudecline.Infact,33)wouldprovefalseifyoudecidednottowashthecarbutweretakenoutforlunchanyway.Anexpressionoftheform‘qif,andonlyif,p’isrepresentedinlogicaséqiffpù,oralternativelyépºqù,ép«qù,oréEpqù.Itstruthtableisasfollows:Propositionallogiccan,ofcourse,getalotmorecomplicatedthantheexampleswehaveconsideredhere,butwehavenowdefineditsgeneralnotationandbasicoperators.Aswellasthewaysinwhichsimplepropositionscanbecombinedtoformcomplexones,however,logicisalsoconcernedwiththeformofthesimplepropositionsthemselves.Thisistheconcernofthebranchoflogicknownaspredicatelogic,whichwewillconsidernext.\nPropositionsandlogic53PredicatelogicPredicatelogicoffersamoredetailedformofanalysisthanpropositionallogic,inthatitdescribesnotjustrelationshipsbetweenpropositions,butalsotheinternalstructureofpropositionsthemselves.Itenablesustodescribehowapropositioncan,forinstance,makestatementsabouttheexistenceofentities,associateanentitywithaparticularproperty,andgeneraliseoverclassesofentities.Italso,inasense,includespropositionallogic,inthatitusesthelogicalconstantswehaveconsideredsofartoexplainrelationshipsbetweenpropositions.Inpredicatelogic,variablesareusedtostandinforentitiesandthepredicateswhichcanbeappliedtothem.SoinaformulasuchaséFxù,Fstandsforapredicateandxforanentity.SoFcanbereplacedbyanypredicateandxbyanyentitygivingus,amongnumerousothers,examplessuchas:34)Thishousehasthreebedrooms.(wherex=thishouseandF=threebedrooms)35)Thefloorisrotten.(wherex=thefloorandF=rotten)3Aswellasassigningapredicatetooneentity,predicatelogicallowsustodescribearelationshipbetweentwoentities.Togivetheformulaforthis,weplaceavariablestandingfortherelationshipfirst,followedbyvariablesstandingforthetwoentities,forinstance:éFxyù.Justaswiththesimpleformula,wecansuggestahugenumberofinstantiationsofthisformula,suchas:36)GeorgewasthefatherofElizabeth.(whereF=fatherof,x=Georgeandy=Elizabeth)37)ClarenceadmiresArchibald.(whereF=admires,x=Clarenceandy=Archibald)38)Marthaplaystheviolin.(whereF=plays,x=Marthaandy=violin)Predicatelogicalsoincludesanumberofquantifiers.Inlinguistics,thistermisusedtorefertowordswhichexpressquantity,wordssuchasall,every,each,any,most,many,some,no,etc.Thereareoperatorsinpredicatelogicwhichcorrespondtowordsofthistype,themostcommonlydiscussedofwhicharetheexistentialandtheuniversalquantifiers.ExistentialstatementsTheexistentialquantifier,asitsnamesuggests,isusedtomakelogicalstatementsabouttheexistenceofentities.Itisrepresentedbythesymbol?;astatementtotheeffectthatanentity,x,exists,wouldtaketheform??x?.\n54PropositionsandlogicThiscanbeaddedtoanyformulainwhichaparticularpredicateisassignedtoavariable,givingsomethingwhichcanbegivenadefinitetruth-value.Soforinstance?redx?specifiesofanyindividualwhichcanbesubstitutedforxthatitisred.Asitstands,withoutknowingwhatindividualisunderdiscussion,wecan’tsayanythingaboutitstruth-value.Butnowlet’saddtheexistentialquantifier,givingus39):39)($x)(redx)Wenowhaveaformulawhichstatesthatthereexistssomeparticularentity,andfurtherthatthisentityisred.Wecouldparaphraseitinformallyas‘thereissomethingwhichisred’.Thisstatementcanbeassignedatruth-value.Infact,wecouldpointatanyoneredthingtoindicatethatitistruethatthereexistsanentitywhichisred.Notethatthisformuladoesn’tsayanythingaboutwhetherthereisoneormorethanonerelevantentitywhichfitsthedescription.Soalongside39)abovewecouldsuggest:40)($x)(Popex)Again,wecouldpointtoanindividualwhoexistsandwhoisthePope,todemonstratethatthisistrue.Infact,thereisonlyoneindividualwecouldpointtoinordertoprove40)true,butachoiceofmanyfor39);thereisonlyonePopebuttherearemanyredthings.Theexistentialquantifierdoesn’tallowforthisdistinction;itcanbestbereadasstipulatingthatthereisatleastoneentityofacertaintype.UniversalstatementsTheotherquantifiercommonlyusedinpredicatelogicistheuniversalquantifier.Itisusedtomakestatementsaboutthewholeofanysetofentities,ortogeneraliseoveraclass.Itsnaturallanguagecounterpartsincludeall,every,eachandany.Thesymbolfortheuniversalquantifieris".Thiscanbecombinedwiththeotheroperatorswehaveconsideredtomakeuniversalstatements.Supposewewanttogeneraliseoveraparticularclassofentities,sayallpoliticians,andtostatethattheyallhaveaparticularproperty,saythattheyareallcorrupt.UsingPtostandforthepredicateisapoliticianandCtostandforthepredicateiscorrupt,wecanexpressourgeneralisationasfollows:41)("x)(Px®Cx)Thiscanbereadas‘foreveryentityx,ifxisapoliticianthenxiscorrupt’or,moreinformally,‘allpoliticiansarecorrupt’.\nPropositionsandlogic55Theuniversalquantifiercanalsobeusedtogeneraliseoverrelationshipsbetweenentitiesor,moreprecisely,overrelationshipsbetweenclassesofentities.Rememberthatwesawabovethat‘GeorgewasthefatherofElizabeth’isonepossibleinstantiationoftheformula?Fxy?,whereF=fatherof.NowofcoursethepairElizabeth,Georgeisonlyoneofahugenumberofpairswhichwouldfitthatformula.Ifwewantedtosuggestauniversalstatementfromthiswecouldsaythateverybodyhasafather,orthatforeveryperson,thereissomeonewhoishisorherfather.Nowwealreadyhaveaquantifierfordescribingexistence,theexistentialquantifier.Soourgeneralisationcouldbestatedinlogicalformasfollows:42)("y)($x)(Fxy)Thiscanbereadas:‘foreveryentityythereexistsanentityxsuchthatxisthefatherofy’,orsimply‘everyonehasafather’.4Theseareofcourseonlyafewofthemanytypesofpropositionwhichcanbeexpressedusingpredicatelogic.Buttheoperatorswithwhichwearenowfamiliarareenoughtoenableustounderstandsomeofthemainideaswhichhavebeenputforwardconcerninglogicalrelations.Fortherestofthischapterwewillconcentrateonthetwoparticularrelationshipsweconsideredintheintroductiontothischapter:entailmentandpresupposition.EntailmentandpresuppositionInourstudyofpropositionallogic,welookedatsomeofthewaysinwhichthetruthorfalsityofonepropositioncanhaveconsequencesforthetruthorfalsityofanother.Inparticular,weconsideredanumberofsituationsinwhichthetruth-valueofasimplepropositionaffectsthetruth-valueofthecompoundpropositionofwhichitisapart.So,forinstance,fromwhatweknowaboutthelogicalpropertiesofconjunction,wecansaythatif44)isfalsethen43)mustalsobefalse;or,toputitanotherway,for43)tobetrueitisnecessaryfor44)tobetrue.43)It’sasunnydayandTommyhasfourticketsforthetestmatch.44)Tommyhasfourticketsforthetestmatch.Thetruth-functionalpropertiesofconjunctiontellusthatifapropositionqisfalsethenapropositionépÙqùisalsofalse,soifTommydoesn’thavefourticketsforthetestmatchthennotonly44)butalso43)mustbefalse.Noticethatif44)istrue,however,wecan’tsayanythingdefiniteaboutthetruth-valueof43);itmightbetrue,buttodecidewhetheritisornotweneedtoknowsomethingelse,thatiswhetherit’sasunnydayornot.Whatwehavejustestablishedabouttherelationshipbetween43)and44)is,infact,that43)entails44),orthat44)isanentailmentof43).Ifa\n56Propositionsandlogicpropositionpentailsapropositionq,wecanrepresentthisrelationshipasép®qù.Rememberthatthisisoneofthesymbolswhichissometimesusedfortherelationshipofmaterialcondition.Infactthetwologicalrelationshipsareverysimilar;ifpentailsqthenthetruthofqfollowslogicallyfromthatofp,justasthetruthofaconsequentfollowsfromthetruthofanantecedent.Wecansummarisetherelationshipofentailmentbysayingthat,ifpentailsq,thenthetruthofqisanecessarypreconditionforthetruthofp.Therearetwopredictionswecanmakeaboutthetruth-values:45)Ifpistruethenqistrue.46)Ifqisfalsethenpisfalse.Aswehaveseen,wecan’tpredictanythingaboutthetruth-valueofqifpisfalse,orthetruth-valueofpifqistrue.Apropositionmayhaveanynumberofentailments.Forinstance,43)aboveentailsboth44)andalsothepropositionthatit’sasunnyday.Eachoneoftheseentailmentsmustbetrueinorderforptobetrue.Thetruthofeachoneofitsentailmentsisanecessaryconditionforthetruthofp,butonlytakenaltogetherdoesthetruthoftheentailmentscountasasufficientconditionforthetruthofp.TherearepropositionsotherthanlogicalentailmentswhichfollowfromthetruthofsentencessuchasourtwostatementsaboutTommy.Wewouldprobably,forinstance,understandthattherewassomeonecalledTommy,evenifwewerenotpreviouslyawareofhisexistence;wewouldunderstand47):47)Tommyexists.Aswesawintheintroductiontothischapter,wecandescribethisasapresupposition.Unlikealogicalentailment,itissharedbyapositivestatementanditsdenial;43)and48)share47)asapresupposition:48)It’sasunnyday,butTommydoesn’thavefourticketsforthetestmatch.This,infact,sumsupthemostcommondefinitionofsemanticpresupposition.Wewilllookatthisdefinitionanditsimplicationsinmoredetailsoon.Butfirstweneedtoconsidersomeofthephilosophicalworkwhichwasinfluentialinshapingourunderstandingofpresuppositionanditsrelationshiptologic.Todoso,weneedtoreturntothedebateaboutthenatureofnameswhichwestudiedinthelastchapter,andspecificallytotheopposingviewsofGottlobFregeandBertrandRussell.\nPropositionsandlogic57FregeonlogicalpresuppositionFregewasperhapsthefirsttousethetermpresuppositionforthistypeoflogicalrelationship,althoughhedoessoonlyinpassing.In‘Onsenseandreference’,henoticesthesharedelementbetweenastatementanditsnegation.Hereishisexplanation:Ifanythingisassertedthereisanobviouspresuppositionthatthesimpleorpropernamesusedhavemeaning.Ifthereforeoneasserts‘Keplerdiedinmisery’,thereisapresuppositionthatthename‘Kepler’designatessomething;butitdoesnotfollowthatthesenseofthesentence‘Keplerdiedinmisery’containsthethoughtthatthename‘Kepler’designatessomething.IfthiswerethecasethenegationwouldhavetorunnotKeplerdidnotdieinmiserybutKeplerdidnotdieinmisery,orthename‘Kepler’hasnoreference.Thatthename‘Kepler’designatessomethingisjustasmuchapresuppositionfortheassertionKeplerdiedinmiseryasforthecontraryassertion.(69)RememberthatinFrege’sterms,tosaythataname‘hasmeaning’istosaythatitreferstosomeactualentity.SoFregecanbeparaphrasedassayingthatanassertionwhichcontainsanameintroducesapresuppositionthatthisnamedoesinfactrefer.Thisisnot,however,partofthepropositionexpressed(partofthe‘senseofasentence’).‘Keplerdidnotdieinmisery’doesn’tentailthatKeplerdiedinmisery,butitdoespresupposethatKeplerexists.5Fregefurtherarguesthatthedenotationofacomplexexpressionisdependentonthedenotationsofitsparts.SoifKeplerdoesn’tdenote,ifthereisnosuchpersonasKepler,thesentence‘Keplerdiedinmisery’can’tdenoteeither.Now,accordingtoFrege,thedenotationofasentenceisitstruth-value,soasentencewhichhasnodenotationhasnotruth-value.Thisisnot,ofcourse,tosaythat‘Keplerdiedinmisery’makesnosense,orexpressesnoproposition.ForFregeall‘propernames’havesense,orconnotation.Sowecanstillattributesensetothename,andhencetothesentencewhichcontainsit,evenifKeplerdoesn’texist.AnassertionofthesentencewhenKeplerdoesn’texistwouldbeanexampleofwhatissometimesknownas‘presuppositionfailure’.Suchusesareinfactquitecommoninnaturallanguage.ButFregewasinterestedindevelopingalogicallyperfectlanguage.Hesawtheexistenceofreferringexpressions\n58Propositionsandlogicwhichfailtodenote,andhenceofsentenceswithouttruth-value,asan‘imperfectionoflanguage’(70).Frege’saccountofthistypeofexamplewaschallengedtwenty-threeyearsafter‘Onsenseandmeaning’wasfirstpublished,inanarticlebyBertrandRussellwrittenin1905andentitled‘Ondenoting’.InthelastchapterwelookedatRussell’slaterarticle,‘Descriptions’,andsawthat,unlikeFrege,hedrewadistinctionbetweenpropernames,suchasKepler,anddefinitedescriptions,suchasthePrimeMinisterofBritain.Heclaimedthattheselatterdon’tinfactrefertoindividuals;theydescribebutdonotdenote.Asevidenceofthis,Russelldrewattentiontodefinitedescriptionssuchastheroundsquareandthegoldenmountain.Wepostponeddiscussionofthequestionofwhathappenswhensuchphrasesoccurinsentences,aquestionwhichRusselladdressesinhis1905article.RussellondenotingIn‘Ondenoting’,Russellobservesthat‘aphrasemaybedenoting,andyetnotdenoteanything;e.g.,“ThepresentkingofFrance”’(471).Byvirtueofitsform,justbecauseitisadefinitedescription,thisphrasewouldappeartodenoteauniqueindividual.Yetin1905,andindeedtoday,thereisnoindividualnamedbythisphrase;thereisnothingtowhichtheexpressionrefers.ThequestionthenishowtointerpretasentencewithanexpressionsuchasthekingofFranceasitssubject.HereisRussell’sown,nowfamous,example:49)ThekingofFranceisbald.It’sclearhowaFregeanaccountofthisexamplewouldgo.Fregedoesn’tinfactdiscussdefinitedescriptionsinthiscontext,buthisexplanationwouldnecessarilybethesameasforhis‘Keplerdiedinmisery’example,becauseheclassesdescriptionsandnamestogetheras‘propernames’,havingbothsenseandreference.ForFrege,anassertionof49)wouldintroduceapresuppositionthatthedefinitedescriptiondoesrefer,thattheindividualexists.Insituationswhenthisisnotinfactthecase,49)wouldfailtohaveatruth-value,anunfortunateexampleofanimperfectioninlanguage.Russelllooksattheprobleminaratherdifferentway.Hisclaimisthat49)doeshaveatruth-value:‘itisplainlyfalse’(484).Inexplaininghow49)canbedescribedasfalse,Russellconcentratesonthemeaningofthesentenceasawhole,ratherthanjustthatofthesubject.Indeed,hisclaimisthatthemeaningofadefinitedescriptioncanonlybedescribedintermsofthemeaningofexpressionsinwhichitoccurs;‘adenotingphraseisessentiallypartofasentence,anddoesnot,likemostsinglewords,haveanysignificanceofitsown’(488).Hedistinguishesbetweenthegrammaticalformofasentence,apurelylinguisticmatter,and\nPropositionsandlogic59itslogicalform,somethingwhichcanbeexpressedinpredicatelogicofthetypewehavebeenstudying.TheexpressionthekingofFrancemaybethegrammaticalsubjectof49),butitisn’tthelogicalsubject.Indeed,thissentencehasnologicalsubject.Unlikeasentencewithapropernameinthesubjectposition,49)isnot,despiteappearances,logicallyinsubjectpredicateform.Instead,themeaningofthesentenceinvolvesquantifiers,variablesandidentity.Thesentencestatesthatthereexistsone,andonlyone,kingofFrance,andalsothatallentitieswhicharekingofFrancearebald.Itisfalsebecauseoneofitsentailments,thepropositionthatthereexistsakingofFrance,isfalse.ThelogicalformwhichRussellassignstohisexamplecanbesummarisedinthefollowingformulaofpredicatelogic.50)($x)(kFxÙ("y)(kFy®y=x)ÙBx)Thismayatfirstappearratherdaunting,butthisisonlybecauseit’sabitlongerthanthelogicalformswehaveconsideredsofar.It’sentirelymadeupofvariablesandlogicaloperatorswithwhichwearefamiliar.Soit’shelpfultoconsideritasaseriesofsmall‘chunks’oflogic:Informally,thiscanbeparaphrasedas‘thereisone,andonlyone,kingofFranceandheisbald’.AsRussellexplains,thisoffersawayofreducinganypropositionwithadenotingphraseasaconstituenttoonewhichdoesn’tinvolveadenotingphrase.Thesentenceneedn’tbedismissedasmeaninglessbecauseitisn’tasentence‘about’aboutthekingofFrance.It’saseriesofstatementsaboutexistenceandidentity.Suchstatementscanindeedbemeaningful,evenintheabsenceofakingofFrance.RememberthatFregeusedtherelationshipbetweenastatementandthedenialofthatstatementtoarguethatcertainpropositionsmustbepresupposedratherthanentailed.Russellalsoconsiderstheeffectofnegation,butheusesittosupporthisanalysisofexamplessuchas49)asnotbeinglogicallyofsubject-predicateform.Hearguesthat51),thenegationof49),isinfactlogicallyambiguous.51)ThekingofFranceisnotbald.Hisclaimisthatthiscouldbeinterpretedintwoways,asadenialofeitheroftwoofthelogicalentailmentsof49):thatis,ofthebaldnessorofthe\n60Propositionsandlogicexistenceoftheking.Theacceptabilityofbothofthefollowingillustratesthisapparentambiguity:52)ThekingofFranceisnotbald;hehasafullheadofhair.53)ThekingofFranceisnotbald;thereisnokingofFrance.Thisisoftendescribedasanambiguityinthe‘scope’ofthenegation.Usingthelogicalformin50)above,wecanexpressthisambiguityinthescopeofnegationasanambiguitybetween52b)and53b):52b)~($x)(kFx&("y)(kFy®x=y)&Bx)53b)($x)(kFx&("y)(kFy®x=y)&~Bx)Again,theseformulaslookcomplicatedatfirstglance,butoncloserinspectionconsistonlyofaseriesofsymbolswithwhichwearefamiliar.In52b)thenegationisattachedtothestatementofexistence;thisformulastatesthatthereisnouniqueentitywhichisthekingofFranceandisbald.In53b)however,thenegationisattachedonlytothestatementofbaldness;thereisauniquekingofFrance,butheisnotbald.Nowofcoursein1905,andalsotoday,theinterpretationofferedin52b)makes51)true,whilewiththeinterpretationin53b)itisfalse.Russelldoesn’trecognisepresuppositionasaseparatelogicalrelationship.HeanalysesboththeexistenceofthekingofFranceandhisbaldnessaslogicalentailmentsof49).Ineffect,forRussell,51)canbereadas‘EitherthekingofFranceisnotbaldorthereisnouniquekingofFrance’.Russell’saccountofdefinitedescriptions,andof‘denotingphrases’whichfailtodenote,waswidelyacceptedandremainedpracticallyunchallengedforforty-fiveyears.Thenin1950PeterStrawsonpublishedanarticleentitled‘Onreferring’,inwhichheofferedanalternativeapproachtothequestionsRussellhadbeenconcernedwith.Strawsonwasoneoftheleadingmembersofwhatcametobeknownastheordinarylanguageschoolofphilosophy,theschoolwhicharguedthatnaturallanguageandlanguageuseareasvalidsubjectsforphilosophicalinvestigationsasaretheformalisedlanguagesoflogicandmathematics.StrawsononreferringStrawson’sinterestinnaturallanguageuseisapparentinhisapproachtotheproblemofdenotingexpressionswhichfailtorefer:the‘thekingofFrance’problem.Hebeginshisarticlewiththecommentthat‘Weverycommonlyuseexpressionsofcertainkindstomentionorrefertosomeindividualperson’(320,emphasisadded).Itisthespeaker,nottheexpressionsthemselves,whichStrawsonregardsasdoingthereferring.Strawsonarguesthat‘“Mentioning”,or“referring”,isnotsomethingan\nPropositionsandlogic61expressiondoes;itissomethingthatsomeonecanuseanexpressiontodo’(326).Itisthisverypointwhich,hesuggests,hasbeenmissedby‘logicians’whentheyhavediscussedmeaning.Strawson’sexplanationoftherelevantexamplesisremarkablysimilartoFrege’saccount.Bothdependontheideathatpropernamesanddescriptionsintroduceapresuppositionoftheexistenceoftheindividualdenoted.Infact,philosophersandlinguistsnowsometimestalkaboutthe‘Frege-Strawson’definitionoflogicalpresupposition,almostasiftheyhadworkedonittogether.Thissimilaritymayatfirstseemrathersurprising.Afterall,thetwophilosophers,workingmorethanfiftyyearsapartinseparatecountries,tookverydifferentapproachestothestudyoflanguage.Fregewasprincipallyinterestedindevelopingalogicallyperfectlanguagewhichwouldbeadequateforthepurposesofmathematicalanalysis.HewouldseemtobeanobviouscandidateforthecriticismlevelledbyStrawsonat‘logicians’interestedinmeaning.Strawson,ontheotherhand,wasdedicatedtotheideathatnaturallanguagewasalegitimatetopicofstudyinitsownright.Tocriticiseitforbeing‘imperfect’would,accordingtoStrawson,betomissthepoint;languagedoesn’tneedtobe‘purified’or‘perfected’inorderforittobeworthyofseriousstudy.Nevertheless,bothphilosophersreachedtheconclusionthattherelationshipbetweenadenotingexpressionandthepropositionthatitsreferentexistsisoneofpresuppositionratherthanentailment.Thisconclusion,however,followedforeachofthemfromverydifferentpremises.ForFrege,itwasalogicalconsequenceofhisdistinctionbetweensenseandreference,itselfanattempttoexplainlogicallycertaintypesofsentence,mostimportantlystatementsofidentity.ForStrawson,theconclusionwaspartofhisrejectionofRussell’saccountofsentencescontainingdefinitedescriptions:anaccountintermsofalogicalformverydifferentfromgrammaticalform.Inparticular,StrawsoncriticisesRussellforhis‘preoccupationwithmathematicsandformallogic’(337).Russell,hesuggests,insistsonaclassicalsystemoflogic,oneinwhicheverypropositionmustbeeither‘true’or‘false’;heinsistsonapplyingthistowhatare,infact,specifictypesofnaturallanguageuse.RussellthereforefailstotakeaccountofhowmostpeoplewouldreactonhearinghisthekingofFranceexample(inwhich,forsomereason,Strawsonchangesbaldtowise).Strawsonexplicitlyrejectstheclaimthatanexpressionsuchas‘thekingofFranceiswise’mustbeeithertrueorfalse.Heallowsforcasesinwhichaspeakerusesthissentencebut‘thequestionofwhetherhisstatementwastrueorfalsesimplydidn’tarise’(330,originalemphasis).Theseareinstancesinwhichthestatementfailstosayanythingtrueorfalsesimplybecauseitfailstomentionanyindividual.InStrawson’sanalysis,then,thepropositionthatthereexistsauniquekingofFranceisnotanentailmentofthesentence.Itis,rather,impliedinwhathedescribesasa‘special’way.InIntroductiontoLogicalTheory,publishedin1952,heusesthetermpresuppositionforthisspecialtypeof\n62Propositionsandlogicimplication.Itisherethatheoffersadefinitionofpresuppositionastruth-functionallydistinctfromentailment.Heremindsusthatifaproposition,sayq,isalogicalentailmentofanotherproposition,sayp,thenqisanecessaryconditiononlyforthetruthofp.Hecomparesthiswiththerelationshipbetweentwostatements,whichhelabelsSandS’,inwhich‘S’isanecessaryconditionofthetruthorfalsityofS…letussay,asabove,thatSpresupposesS’’(175,originalemphasis).AnotherwayofexpressingthisistosaythatS’ispresupposedbySifitislogicallynecessaryforbothS(thetruthofS)and~S(thefalsityofS).Thisiswhatisgenerallyacceptedtobethestandardlogical,ortheFrege-Strawsondefinitionofpresupposition.Ineffect,Strawsonabandonstheideathatwhatisknownasbivalentlogicmustbesufficienttoexplainordinarylanguage.Bivalentlogic,alsoknownasclassicallogicbecauseofitsoriginsinancientphilosophy,isthelogicwehavebeenworkingwithinthischapter.Itallowsachoicebetweenpreciselytwotruth-valuesforanyproposition:‘true’and‘false’.Strawsonarguesthatinsomecontextsastatementmayhaveneitherofthesevalues,becausethequestionofitstruthorfalsitymayjustnotarise.Somelogicianshaveproposedathree-valued,ortrivalentlogicwhichissometimesdescribedascontainingthevalues‘true’,‘false’and‘neither-true-nor-false’.AnexamplesuchasRussell’sthekingofFrancesentence,whenthepresuppositionnecessaryforittobeeithertrueorfalsefails,mustbeclassifiedashavingthisthirdtruth-value.Otherlogicianshavesuggestedthatitispossibletoretainabivalentsystem,aslongasa‘truth-valuegap’isincludedtoaccountforsuchexamples.ItisatestimonytoBertrandRussell’slongevityandenduringinterestinphilosophythatin1957,fifty-twoyearsafteroriginallypublishing‘Ondenoting’,heproducedaresponsetoStrawson’schallenge,inanarticleentitled‘MrStrawsononreferring’.Thisresponseisshortandpolemical,andservestohighlightthedifferencesbetweentheapproachestothestudyofmeaningadoptedbyRussell,withhisinterestinmathematicallogic,andbyStrawson.Russellsumsupthisdifference,fromhisownpointofview,bydescribinghowmanyphilosophersofthetime,includingStrawson,Arepersuadedthatcommonspeechisgoodenoughnotonlyfordailylife,butalsoforphilosophy.I,onthecontrary,ampersuadedthatcommonspeechisfullofvaguenessandinaccuracy,andthatanyattempttobepreciseandaccuraterequiresmodificationofcommonspeechbothasregardsvocabularyandasregardssyntax.(387)Russellsawthetaskofthephilosopherasbeing,inpart,tomodifyimprecisenaturallanguagetoreflect‘correct’logicalstructure.Inhisresponse,hereiterateshiscommitmenttotheideathateverymeaningfulsentencemust,inaccordancewithclassicallogic,beeithertrueorfalse.Hisaimhadalwaysbeentofindanadequatewayofexplainingthislogic;‘My\nPropositionsandlogic63theoryofdescriptionswasneverintendedasananalysisofthestateofmindofthosewhouttersentencescontainingdescriptions’(388).Russell’sresponsetoStrawsonindicatesthatthedebatebetweentheirdifferentaccountsofnon-denotingdefinitedescriptionswasnotresolved,andthatitinfactremainsunresolvable.Thetwophilosophers,andtheschoolsofthoughtwhichtheycanbeseenasrepresenting,approachtheproblemsfromtwoverydifferentperspectives.Strawson’saccountistheonewhichhasbeenofmostenduringinteresttolinguists.Thisisperhapsnotsurprising,givenStrawson’sinterestinspeakersandtheiruseoflanguage.Afterall,Russellaccusedhimofbeingconcernedwith‘commonspeech’andwith‘thestateofmind’ofthespeaker,twotopicswhichhavebecomeofcentralimportanceinmodernlinguistics.Thereis,however,anotherreasonforthepopularityofStrawson’saccountoflogicalpresupposition,clearlydistinguishedasitisfromlogicalentailment.Itseemstocapturetheintuitiveresponsethatmanypeoplehavetoexamplessuchas‘thekingofFranceisbald’.Thatis,thatthereareimportantdifferencesbetweenthepropositionthatthekingofFranceisbaldandthepropositionthatheexists.Thefirst,alogicalentailment,iswhatisexplicitlystated;itissetoutfordiscussion.Thesecond,however,seemstobenotsomuchstatedasassumed,ortakenforgranted.Thedistinctionissometimesdiscussedintermsofforegrounded(entailed)propositionsandbackgrounded(presupposed)ones.Russell’saccount,dependentasitisonaseriesoflogicalentailments,doesn’tallowforanyexplanationofthisapparentdifference.ForRussell,thisisn’treallyaproblemsince,aswehaveseen,hisinterestisn’tinspeakers’(orhearers’)statesofmind.However,it’simportantforanexplanationofthisdifferencetobeavailablewhenpresuppositionisdiscussed,asithasbeen,asarelationshipbetweenspeakers.Suchaccountsinvolvewhatisknownascontext-boundorpragmaticpresupposition.PragmaticpresuppositionThestoryofworkonpresuppositioninlinguisticscanverybroadlybesummarisedasadebatebetweentwoopposingpointsofview.Somelinguistshaveseenpresuppositionasasemanticfeatureofthelanguage,determinedbythestructureandvocabularyofsentences,whileothershavecharacteriseditasapragmaticfeatureofspeakers’behaviour,determinedbytheutterancestheyproduceinparticularcontexts.Whenpresuppositionwasfirstdiscussedinlinguistics,itwasthe‘logical’aspectsofpresuppositionwhichreceivedmostattention,generallydiscussedintermsoftheFrege-Strawsonaccountofthedifferencebetweenentailmentandpresupposition.Linguistsinthistraditionhaveconcentratedonanalysingthepresuppositionalpropertiesandbehaviourofdifferentwords,structuresandcomplexsentences.Wewillconsiderafewoftheseinthefinalsectionofthischapter.\n64PropositionsandlogicAroundthebeginningofthe1970s,linguistsbegantoworkontheideathatpresuppositionisbestdescribedasarelationshipbetweenutterances,oreventhespeakerswhoproducethem.Suchdescriptionsdrewontheclaimthatwhatispresupposedisoftendependentnotjustonlogicalproperties,butoncontextofuse.Theywerebasedontheideathatthereasonswhysomeinformationmaybepresupposedratherthanasserted,andthereforebepresentedas‘background’,aredependentoncontextualfactorssuchasthespeaker’sbeliefs,thepurposeofthediscourseandsoon.Despitethecleardifferenceofemphasisbetweensemanticandpragmaticaccounts,bothcanbeseenasdrawingonideasputforwardbyStrawson.Strawson’sinfluenceonthedevelopmentofapragmaticaccountofpresuppositionshouldn’tinfactbetoomuchofasurprise.Hewas,aswehaveseen,keentoemphasisetheimportanceofdistinguishingbetweensentencesandtheusestheyareputto,orthestatementstheycanbeusedtomake.InIntroductiontoLogicalTheory,hestatesexplicitlythathis‘presuppositional’logicappliestostatements,nottosentences:‘Itisaboutstatementsonlythatthequestionoftruthorfalsitycanarise;andabouttheseitcansometimesfailtoarise’(175).Inassigningtruth-valuestostatementsratherthantosentences,hemakesitclearthatheseesthem,andrelatedpropertiessuchaspresupposition,asdependentoncontext.Andin‘Ondenoting’hecommentsonthepresuppositionofuniqueexistenceattachedtotheuseofthedefinitearticle(inaphrasesuchasthekingofFrance)inawaywhichagainhighlightstheimportanceofcontext.Theoccurrenceofthedefinitearticleactsasasignaleitherthatpreviousreferencehasbeenmadetotheentityinquestion,orthat‘thecontext(includingthehearer’sassumedknowledge)isexpectedtoenablethehearertotellwhatreferenceisbeingmade’(342,originalemphasis).Itisfactswhichcanbetreatedasbackground,suchastheexistenceofcertainindividuals,whicharemostlikelytobepresupposed.Oneclearreasonwhyafact,inaparticularcontext,mightbetreatedinthiswayisifitis‘oldinformation’,inthatitisalreadyknown.InStrawson’sterms,thisisinformationwhichhasalreadybeenmentioned,orisobviousfromthecontext.Indeed,linguistshaveoftenequated‘background’information,whichisgenerallypresupposed,with‘old’,or‘given’information,whilewhatis‘foregrounded’isgenerallyinformationwhichis‘new’.Strawsonhimselfillustratesthispointbyencouraginghisreaderstoconsiderthedifferentcontextsinwhich54)and55),whichexpressthesameproposition,wouldbeappropriate.54)Thatisthemanwhoswamthechanneltwiceinoneday.55)Thatmanswamthechanneltwiceinoneday.Example54),unlike55),containsthedefinitedescriptionthemanwhoswamthechanneltwiceinoneday,togetherwithitsassociatedpresuppositionofexistence.Strawson’sexplanationofthedifference\nPropositionsandlogic65betweenthetwoisphrasedintermsofknowledgeandbeliefonthepartofthespeakerandhearer.Hesuggeststhat:‘Youwouldsay[54]insteadof[55]ifyoukneworbelievedthatyourhearerkneworbelievedthatsomeonehadswumthechanneltwiceinoneday’(343,originalemphasis).NotethatStrawsonis,ineffect,discussingthedifferencebetweenthesetwoexamplesintermsofthesituationsitwhicheachwouldbeappropriate,anotionwhichhasbecomecentraltopragmaticdescriptionsofuse.AccordingtoStrawson’saccount,then,whatispresupposedcanbeseenasareflectionofthespeaker’sbeliefsaboutthesituationinwhichsheisspeaking.Inparticular,whatispresupposediswhatthespeakerbelievestobealreadyknownbythehearer.Therewould,afterall,belittlepointin‘foregrounding’,anddrawingattentionto,whatisalreadyknown.However,asStrawsonhimselfobserves,itsimplydoesn’tseemtobethecasethataspeakerwillonlyeverpresupposewhatshehasreasontobelievetheheareralreadyknows.Informationwhichisinfact‘new’maysometimesbepresupposed.Strawson’sownaccountofthisisin‘Ondenoting’.Henotesthatifheweretouseasentencesuchas‘allmychildrenareasleep’,hewouldintroducethepresuppositionthathehaschildren.Hegoesontoobservethathemayusethissentence:withtheintentionoflettingsomeoneknowthatIhavechildren,orofdeceivinghimintothinkingthatIhave.Norisitanyweakeningofmythesistoconcedethatsingularphrasesoftheform‘theso-and-so’maysometimesbeusedwithasimilarpurpose.NeitherAristoteliannorRussellianrulesgivetheexactlogicofanyexpressionofordinarylanguage;forordinarylanguagehasnoexactlogic.(344)Strawsonhereobservesthatpresupposingexpressionsmaysometimesintroduceinformationwhichisnewtothespeaker,butdoesn’tofferanyexplanationofthecontextsinwhichtheycanbeusedinthisway.VarioussubsequentattemptshavebeenmadetodefinepresuppositioninawaywhichmakesitconsistentwithusesofthetypeStrawsonnoticed.Manyofthesedatefromthe1970s,wheninterestwasrenewed,inphilosophyandlinguistics,inpragmaticaspectsofpresupposition.OnephilosophertotakeuptheissueofpresupposingexpressionswhichareusedintheabsenceofthenecessarybackgroundknowledgewasPaulGrice.GricewasanotheroftheOxfordphilosophersoftheordinarylanguageschool.InitiallyPeterStrawsonwashisstudent,andthetwophilosopherslatercollaboratedonjointwork.Griceisnowbestknownforhistheoryof‘conversationalimplicature’,whichwewillconsiderinChapter4.Buthiscontributiontothediscussionofpresuppositioncanbefoundin‘Presuppositionandconversationalimplicature’,anarticlewhichheworkedonduringthe1970s,althoughitwasfirstpublishedin1981.\n66PropositionsandlogicInthisarticleGriceconsidersthehypothesis,whichhemakesquiteclearhedoesn’twanttocommithimselfto,thatRussell’stheoryofdescriptiongivesanaccurateaccountof‘thephrases’,andthatcertainpragmaticprinciplescanbefoundasalternativestothe‘theoryofpresupposition’.Weneedn’tconcernourselvesheretoomuchwiththedetailsofthisaccount,whichisnotasfullydeveloped,andhasnotbeennearlyasinfluential,asRussell’sorStrawson’s.However,Grice’sgeneralapproach,andhiscommentaryonsomeofhisexamples,areinterestingdevelopmentsofthetopic.Significantly,hisaccountisdependentonconversationaluseoflanguage.Hesuggeststhatinconversationspeakerstendtoexpresstheirmessageinsuchawayasto‘facilitate’anappropriateresponse.Onetypeofappropriateresponsetoastatementoffactwouldbeadenial.Oneofthepropositionstowhichtheuseofasentencecommitsaspeaker‘is,insomeway,singledoutastheonethatisspeciallylikelytobedenied’(273).6Otherinformationhas‘common-groundstatusand,therefore,isnotsomethingthatislikelytobechallenged’.Thiscommongroundcouldbesharedknowledge,butneednotbe.Gricearguesthatwemightwellhearsomeoneutteringanexamplesuchas56)inacontextinwhichaparticularconcertisbeingdiscussed,butthespeakerhasnoreasontobelievethatheraudienceknowsthatheraunthasacousin,oreventhatshehasanaunt.56)Myaunt’scousinwenttothatconcert.Theuseofthereferringexpressionmyaunt’scousinpresupposestheexistenceofthatrelative,justastheuseoftheexpressionthekingofFrancepresupposestheexistenceofthemonarch.Grice’sexplanationisthatapropositionneedonlybe‘non-controversial’,or‘somethingwewouldexpectthehearertotakefromus(ifhedoesnotalreadyknow)’(274).Wemightexpandonthisandsaythat,evenifwedon’tknowthatthespeakerhasanauntwhohasacousin,thisfactisperfectlyconsistentwithourbackgroundknowledgeandassumptions;peopletendtohavefamilies,whicharequitelikelytoincluderelativessuchasauntsandcousins.Grice’saccountsuggeststhat,intermsoflanguageuse,thedifferencebetweenpresuppositionandassertionisperhapsbestcharacterisedasadifferenceinlevelsofcommitment.Onereasonwhyaspeakermightbesodeeplycommittedtoapropositionthatshepresupposesratherthanassertsmay,ofcourse,bethatitisknownbybothspeakerandhearer.Itwouldbesimplypointless,orredundant,tosay,forinstance57)insteadof58)ifyouaresurethatyourhearerhassomebasicknowledgeaboutFrenchpolitics.57)ThereisapresidentofFrance,andhelivesinParis.58)ThepresidentofFrancelivesinParis.\nPropositionsandlogic67Similarly,toborrowGrice’sownexample,itwouldbeunnecessarilycomplicatedtouse59)insteadof60).59)Ihaveanauntwhohasacousinwhowenttothatconcert.60)Myaunt’scousinwenttothatconcert.Evenifyourhearerdoesn’tknowmuchaboutyourfamilyconnections,60)willtellhimallheneedstoknowwithoutbeingunnecessarilywordy.Itwillalsosteerhimintherightdirectionasfarasaresponseisconcerned.Whatyouaretalkingaboutistheconcertinquestion.Yourfamilytreeisnotreally‘thepoint’;yourhearerwouldberatherchangingthesubjectifherespondedwith‘Gosh,Ineverknewthatyouhadanaunt’.Thebalancebetweenpresupposedandassertedinformationseemstoofferonewayofexamininghowspeakersuselanguagetopresenttheirideassuccinctly,todirecttheirhearerstowardsrelevantinformation,andalsotoexercisecontroloverconversations.Someworkinthisareaismentionedunder‘furtherreading’,below.Thefinalsectionofthischapterisconcernedwithsomemoregeneralwaysinwhichlinguistshavepickeduponthis,andonotherareasoriginallydiscussedinlogic,andhaveconsideredtheirsignificanceforthestudyofnaturallanguage.LogicandlinguisticsTheworkofFrege,RussellandStrawsonwhichwehaveconsideredinthischapterhasbeenveryinfluentialinlinguistics.Assuggested,itisFregeandStrawson’sworkonlogicalpresuppositionwhichhasbeenthefocusofmostinterest,butRussell’sworkshouldn’tbeignored.Inparticular,Russell’sdistinctionbetween‘apparent’grammaticalformand‘hidden’logicalformhasbeenseenasabasisfordevelopmentsinlinguisticssuchasChomsky’sdistinctionbetween‘surface’structureand‘deep’,logicalstructure.Meanwhile,bytheendofthe1970s,logicalpresuppositionhadreceivedsomuchattentionthatGricewasabletocomment,in‘Presuppositionandconversationalimplicature’,thatInrecentyears,linguistshavemadeitincreasinglydifficultforphilosopherstokeeptheireyesgluedonahandfulofstockexamplesof(alleged)presuppositions,suchthekingofFrance’sbaldnessandtheinquiryaboutwhetheryouhaveleftoffbeatingyourwife.(279)Muchoftheworkwhichwasconductedinlinguisticswasaimedatidentifyingexpressionsandconstructionswhichmightbesaidtointroducepresuppositions.AsGricesuggests,philosophershadonthewholebeencontenttodiscusspresuppositionintermsofaveryfewexamples.Indeed,\n68Propositionsandlogicwehavebeenabletofollowthediscussionsofarbyconsideringonlywhatareknownasexistentialpresuppositions,presuppositionsofexistenceassociatedwithusingexpressionssuchaspropernamesanddefinitedescriptions.Ifpresuppositionistobeavaluabletoolforexplaininglanguage,wereallyneedafewmoreexamplesthanthat.Let’sstartwithGrice’sveiledreferencetowife-beating.Whathehasinmindhereisthepresuppositionwhichhadbeennoticedinconnectionwiththeuseoftheverbstop.Ifyousaythatyouhavestoppeddoingsomethingyougenerallypresuppose,butdon’tactuallyassert,thatitissomethingwhichyouusedtodo.Soifyou’representedwith61),it’simpossibletoanswereitherwaywithoutcommittingyourselftohavingbeenaviewerofdaytimeTVatsomepoint(we’lltakethelibertyofchangingwife-beatingtodaytimeTV-watching).61)HaveyoustoppedwatchingdaytimeTVyet?Thisexamplecontainswhatisknownasapresuppositiontrigger;thepresenceoftheverbstopintroducesthepresuppositionthatthehabithasexisted.Manyotherpresuppositiontriggershavebeenidentified.7Afewexamplesoftheseareillustratedinexamples62)–65);ineachcasethetrig-geringwordishighlighted.62)Johnregretssendingadefamatorye-mailtothewholelist.63)Theworkersknew/believedthatthemanagementwerelying.64)ThestudentsmanagedtofinishtheiressaysduringReadingWeek.65)Johnread‘Ondenoting’beforehewentdownthepub.Regretandknowarebothexamplesoffactiveverbs;theyintroducethepresuppositionthattheactiondescribedinthesub-clausedidactuallyoccur,inthiscasethatJohndidactuallysendthee-mailandthatthemanagementwereactuallylying.Notethatthispresuppositiondoesn’tarisewiththeverbbelieve;ifweread63)withbelieveinsteadofknowwedon’tactuallylearnwhetherthemanagementwerelyingornot.Theverbmanage,asin64),entailsthatthetaskwascompleted,butitalsopresupposesthatthetaskwasadifficultone.Finally,thepresenceofbeforein65)introducesthepresuppositionthattheactiondidactuallytakeplace;Johndidgodownthepub.Onewayinwhichthesehaveallbeenidentifiedaspresuppositional,onewhichiseasilytested,isthattheyareall‘preservedundernegation’;ifyoutrynegatinganyoftheseexamplesyou’llfindthattheentailmentsarelostbutthepresuppositionswehaveidentifiedremain.Thesametestcanbeappliedtopresuppositionswhichhavebeenidentifiedasbeingattachedtoparticulargrammaticalstructures,ratherthantoindividualwords.Examples66)and67)bothincludenegationbutsharethesamepresuppositionsastheirpositivecounterparts.\nPropositionsandlogic6966)Itwasn’tColumbuswhodiscoveredAmerica.67)NewEssaysConcerningHumanUnderstanding,whichisaverygoodread,wasn’tpublishedduringitsauthor’slifetime.In66)thepresenceofthecleftconstructionintroducesthepresuppositionthatsomeoneperformedtheactiondescribed,inthiscasethatsomeonediscoveredAmerica.In67)it’sthepresenceoftherestrictiverelativeclausewhichintroducesapresupposition:inthiscasethatthebookinquestionisagoodread.Ithasalsobeennoticedthatthepresenceofaparticularwordorconstructionisn’talwaysenoughtoensurethatapresuppositionisintroduced.Semanticpresuppositionsofthistypecanbedefeatedinappropriatecontexts.Asjustoneverysimpleexampleofthis,considerthedifferencebetweenthefollowing:68)Hamletwashedhimselfbeforehekilledtheking.69)Hamletstoppedhimselfbeforehekilledtheking.In68),thepresenceofbefore,asexpected,introducesthepresuppositionthattheaction(thekillingoftheking)didinfacttakeplace.Butin69)thisisn’tthecase;intheinterpretationof69)asawholewesimplydon’tgettheimpressionthatHamletkilledtheking.Tofindthereasonforthis,weneedtolookbeyondthepresenceofthebeforeclause,andconsultwhatweknowaboutthemeaningoftheverbstopped.Becausethisverbisincompatiblewiththeactionhavingbeenperformed,thepresuppositionwhichwewouldexpecttobeassociatedwithbeforeisn’tassociatedwiththesentenceasawhole.Examplesofthistypehavebeenusedtoarguethatpragmaticpresuppositionmustatleasttakeprecedenceover,andperhapscompletelysupercede,semanticpresupposition.Itisthewidercontextinwhichthebeforeclauseoccurswhichdetermineswhatispresupposed.Pragmaticaccountshavebeensuggestedinvolvingmuchlargercontexts,inthiswayusingpresuppositionasameansofanalysingdiscourse.Theuseoflanguageindiscourse,andtheeffectofcontextonthelogicalpropertiesofsentences,issomethingthatwewillreturntoinChapter4.First,however,wewilllookatsomeofthewaysinwhichphilosophersandlinguistshaveaccountedforthefactthatweareable,oftenwithoutthinking,toacceptsomestatementsas‘true’andtodismissothersas‘false’.Inotherwords,wewillconsidertherelationshipbetweensentences,orthepropositionstheyexpress,andtherealitywhichtheydescribe.\n70PropositionsandlogicFurtherreadingPropositionsSomediscussionofpropositionalityoccursinmanygeneralbooksonphilosophyandthephilosophyoflanguage.Ayer(1971)Language,TruthandLogicisagoodexample(seeespeciallyChapter5);andSearle(1969)SpeechActsisparticularlyreadable(seeparticularlyChapter2).Thebasicsofpropositionality,correspondence,truthconditions,etc.areintroducedinmanytextbooksonsemantics.AgoodexampleistheintroductorychaptertoCann(1993)FormalSemantics.LogicWehaveonlyscratchedthesurfaceinourstudyofformallogic.AgoodintroductiontothesubjectisGuttenplan(1986)TheLanguagesofLogic.TwobooksparticularlyaimedatexplaininglogicaltheoryforstudentsoflinguisticsareMcCawley(1981)EverythingthatLinguistsHaveAlwaysWantedtoKnowaboutLogic,andJ.Martin(1987)ElementsofFormalSemantics.Bothdealwithfarmorecomplexmaterialthanisdealtwithhere,includingtheroleofsettheoryinlogic,butbotharedesignedtobeaccessibletothosewithoutamathematicalbackground.McCawley’sbookiscomprehensive;thefirstthreechaptersinparticulardealwithpropositionallogicandthelogicaloperators,whileChapter9isconcernedwithpresupposition,bothlogicalandpragmatic.MartinintroducespropositionallogicinChapter1andrelatesittoproofconstruction,atopicnotconsideredhere.HetakesthediscussionfurtherinChapter5.Chapter7isconcernedwithpresupposition;Martinconcentratesinparticularonpresuppositionfailure,thenotionofthree-valuedlogicandtheprojectionproblem.PresuppositionSincethethreemainarticlesbyFrege,RussellandStrawsononthesubject,muchoftheworkonthetopicofpresuppositionhasbeenundertakenbylinguists,andtheliteratureinthisareaisvast.Forthisreason,mostintroductionstoandcommentariesonpresuppositionarefoundinbooksonlinguisticsratherthanonphilosophy.However,someintroductionstothephilosophyoflanguagedoincludediscussionoftheprimaryphilosophicaltexts,andinparticularofRussell’saccountofdefinitedescriptions.See,forinstance,R.Martin(1987)TheMeaningofLanguage,Chapter13;DevittandSterelny(1988)LanguageandReality,Chapter3;andStainton(1996)PhilosophicalPerspectivesonLanguage,Chapter3.\nPropositionsandlogic71Aswehaveseen,workinlinguisticsonpresuppositionhasdevelopedinanumberofdifferentdirections,andeachdirectionhasitsownbodyofliterature.Chomsky’s(1972)bookStudiesonSemanticsinGenerativeGrammar,andparticularlythechapter‘Deepstructure,surfacestructureandsemanticinterpretation’,introducestheideathatintonationalhighlighting,whichChomskycalls‘stress’,canaffectwhatispresupposed.Chomskysees‘stress’asbeinggrammaticallydetermined,andthereforearguesthatthesepresuppositionalphenomenamustbeexplainedsemantically.Stalnaker’s(1974)article‘Pragmaticpresuppositions’isanearlyattempttoexplaintheroleofspeakersandofcontextinwhatispresupposed.AlongerstudyofthisisofferedbyvanderSandt(1988)ContextandPresupposition.Burton-Roberts(1989)TheLimitstoDebate,ontheotherhand,proposesapurelysemanticaccountofpresupposition,butonewhichmustbesupplementedbyapragmaticaccount.Heoffersarevisedversionofthe‘standard’(Frege-Strawson)logicaldefinitionofpresupposition,andarguesthatathirdtruth-valueneednot,andshouldnot,beidentifiedwitha‘truth-valuegap’,orlackoftruth-value.SurveysofworkonpresuppositioncanbefoundinGazdar(1979)Pragmatics,whoisofcourseconcernedmainlywiththedevelopmentofworkinpragmaticpresupposition,andWilson(1975)PresuppositionsandNon-Truth-ConditionalSemantics,whooffersapurelysemanticapproach,inparticularconcentratingonanumberofindividuallexicalitems.Manyintroductionstosemanticsorpragmaticscontainachapteronpresupposition,althoughofcoursetheperspectiveandemphasisofthesevary.Levinson(1983)PragmaticsconsiderstherelevancetopragmaticsofworkonpresuppositionfromtheRussell/Strawsondebateonwards.McCawley(1981)EverythingthatLinguistsHaveAlwaysWantedtoKnowaboutLogic,Chapter9isashortbutdetailedanalysisofsemanticpresupposition,andparticularlyofthestatusandconsequencesofathree-valuedlogic.\n3TruthandrealityIntroductionAspartofourinvestigationinthelastchapterintohowwordscancombinetogethertomakemeaningfulstatements,westudiedthebasicsofpropositionalandpredicatelogic.Theseprovideduswithasystemfordescribingthetruth-functionalpropertiesofsentences,orofthepropositionstheyexpress.Theyalsogaveusaninsightintosomeoftherelationshipswhichcanexistbetweenpropositions.Theydidn’t,however,offerusanyhelpinexplaininghowlanguagerelatestoreality.Aswehaveseen,languageisn’tjustanautonomoussysteminwhichwordsarestrungtogethertomakesentences.Inordertogiveafullaccountoflanguageingeneral,orindeedofanyparticularlanguage,thelinguistneedstoexplainnotjusttherelationshipswithinalanguage,butthosebetweenalanguageandtheworldoutside.InChapter1weaddressedonepartofthisissuebyconsideringhowitisthatwords,byvirtueofwhatwemightcalltheir‘lexicalmeaning’,canidentifyactualobjectsintheworld.Toputthisanotherway,weconsideredhowitisthattheintensionalpropertiesofawordareabletodeterminewhatobjectorobjectsformitsextension.Wearenowinapositiontoaskthesamequestionaboutsentences.Wehavedefinedtheintensionofasentenceastheproposition,orthought,thatitexpresses.Thequestionwenowneedtoaskishowthispropositiondeterminestheextensionofthesentence,inotherwordswhetheritis‘true’or‘false’.There’snoquestionthatwefrequentlyneedtolookattheworldoutsideofthelanguagesystemwhenmakingdecisionsabouttruth-values.Takinganywellformedsentence,weareunlikelytobeabletosaywhetherthepropositionwhichitexpressesistrueorfalsewithoutlookingatwhatwemightlooselydescribeas‘reality’,orperhaps‘theworld’.1That’snottosaythattherearen’tsomesentencestowhichwefeelconfidentassigningthevalue‘true’or‘false’straightaway,withoutfirsthavingtofindouthowthingsareintheworld.Wemay,forinstance,feelthatwhilewe’renotpreparedtomakeajudgementaboutthetruthof1)withoutfirstgoingto72\nTruthandreality73havealookinthegarden,wearequiteconfidentinsayingthat2)isevidencebeforemakingadecision.1)Thecherrytreeisinblossom.2)Anunhappybankmanagerisnotabankmanager.3)Allgreenfrogsaregreen.Insayingthatwedon’tknowwhether1)istrueorfalseweare,ineffect,admittingthatwedon’tknowwhatitsextensionor,inFrege’sterms,itsreferenceis.Butthisisn’t,ofcourse,tosaythatthesentencemakesnosensetous.Onthecontrary,weknowexactlywhatpropositionitexpressesandwecanprobablypicturetoourselvesthesituationitdescribes.Wemightsaythatweknowwhatwouldbethecaseifitweretrue.Insayingthiswearesayingthatweknowwhatconditionswouldhavetoholdfor1)tobeclassifiedas‘true’.Inotherwords,althoughwedon’tknowitstruth-value,wedoknowitstruthconditions.Thereisacertainstateofaffairs,whichwouldhavetobeinplaceinorderfor1)tobeatruesentence,andthedescriptionofthisstateofaffairsformsthetruthconditionsof1).Wemightsayinformallythattherelevantstateofaffairsisthatthecherrytreeunderdiscussionmustbeinblossomatthetimeofspeaking.It’seasiertomakedecisionsaboutthetruth-valuesofthesecondtwoexamples.Wemightsaythat2)isacontradiction.Ithastobefalse,whateverthestateofaffairs;itisnecessarilyfalsebecauseofthewordsitcontains.2But3),ontheotherhand,isatautology;ithastobetruebecause,inasense,itsaysthesamethingtwice.Inphilosophicalterminology,3)isanexampleofananalyticsentence.Theadjectiveanalyticis,ofcourse,relatedtothenounanalysis.Ananalysisofsyntacticstructure,forinstance,involvesbreakingdownacomplexwhole,suchasasentence,intoitscomponentpartsinordertounderstandhowitisconstructed.Thesamecouldbesaidofsemanticanalysis;itinvolvesidentifyingtheindividualelementswhichmakeupthewhole.Example3)isnecessarilytrue;weknowitistruejustbecauseweknowthemeaningoftheindividualwordsitcontains.Specifically,ifweanalysethesubjectwefindthatgreenisacomponentpart,andfurtheranalysisrevealsthatgreenalsoformsthepredicate.Wewillbelookingatwhathasbeensaidaboutanalyticsentenceslaterinthischapter.Thecentralideawewillbeconcernedwith,however,isthatthesense,or‘prepositionalmeaning’ofasentencecanbedefinedintermsofitstruthconditions,orthespecificcircumstanceswhichhavetoholdforittobetrue.Aswewillseethroughoutthechapter,variousproblemspresentthemselvesfora‘truth-conditional’accountofmeaning.Tostartwith,noticethedifferencebetweenexampleslike1)aboveand4)below.Intuitively,weknowwhattheybothmean,butwhenwesetaboutdescribinghowwewouldknowiftheyweretrueorfalse,weareforcedtoadmitthattheyareinfactverydifferent.\n74Truthandreality4)Therearefairiesatthebottomofmygarden,buttheyareinvisibleandmakenosound.Weknowwhatthetruthconditionsof4)are;weknowwhatfacts,however,improbable,wouldmakeittrue.Whatmakes4)differentfromanexamplelike1)isnotsomuchtheimprobabilityofwhatisdescribed,asthefactthatwehavesimplynoideahowtofindoutwhatitstruth-valueis.Weknowwellenoughhowtouseourknowledgeofthesenseof1)todetermineitstruth-value;wecangointothegardenandhavealookatthetree.Butwhilewearetherewewon’tbeabletocheckonthetruth-valueof4),becausewecan’timaginewhatmightcountasevidenceforacceptingitastrueorrejectingitasfalse.Examplessuchasthisshowupanimportantconsequenceofaccountsofsentence-meaningentirelyintermsoftruth.Ifyoucanneverdecidewhether4)istrueorfalse,andmeaningisultimatelydependentontruth,itmightbelegitimatetoquestionwhether4)canbesaidtobemeaningfulatall.Thisistheargumentwhichhasbeenputforwardinverificationisttheoriesofmeaning,whichwewillconsiderinthischapter.Theseareinteresting,atleastinpart,becauseitwasthereactionagainstthemwhichledtoagrowinginterestinthewaylanguageisactuallyused,andinthesignificancesentencesmayhavebeyondthestarkfactofbeingtrueorfalse.Evenifweconcentrateonsentencesforwhichwecanmakedefinitedecisionsabouttruth-value,wecomeacrosssomeimportantdistinctionswhichastraightforwardtruth-basedtheoryofmeaningwouldmiss.Let’sconsiderthedifferencebetweenthefollowingpairofsentences:5)Oilfloatsonwater.6)Thirty-fourisalargernumberthantwenty.Weshouldn’thavemuchdifficultyinagreeingthatthesearebothtrue,butwemaywellspotthatthereisadifferent‘type’oftruthinvolvedineachcase.Whileweknowthatoilfloatsonwater,andcouldproveitifnecessary,wecanimagineasituationinwhichitisn’ttrue.Wecanimagineacontext,sayinaworldcreatedinsciencefiction,inwhichoilhasverydifferentpropertiesfromthoseweknowittohaveinfact,andinwhichitsinksinwater.Incontrast,it’simpossibletoimagineaworldinwhich6)isnottrue.3Wecansumupthisdifferencebysayingthat6)isanecessarytruth,while5)isacontingentone;itsohappensthat5)istruebutitneedn’tbe.Thisdifference,andtheidea,whichwehavesuggestedinformally,thatwecanimaginedifferent‘worlds’inwhichthingsareotherthanweknowthemtobe,hasbeentakenupbyanothertruth-basedaccountofmeaning,knownaspossibleworldsemantics.Aswewillsee,suchaccountshavesuggestedthatmeaningcanbedefinedintermsoftruth,butonlywithreferencetothetotality,orset,ofpossibleworldsinwhichasentenceistrue.\nTruthandreality75First,however,weneedtogobacktothebeginning,andtostartbylookingatwhathasbeensaidaboutthewaysinwhichlanguagerelatestotheworld.Indoingso,weneedtobeawareofthedifferencewhichhasbeendescribedbetweenapropositionhavingtheproperty‘true’or‘false’,anditshavingconditionswhichcandeterminetruthorfalsity.TruthconditionsInChapter2,whenweexaminedthenatureof‘propositionality’,wesawthatapropositioncanbethecontentofathought,theobjectofabelief,ormostsignificantlyfromourcurrentpointofview,theintensionofasentence.Wehavejustestablishedthatonewayofdescribingthecontentofapropositionisasasetoftruthconditions.Now,iftheintensionalmeaningofasentencecanbecharacterisedasaseriesofconditionswhichmustbefulfilledforittobetrue,wehaveaclearwayofexplaininghowtheintensionofasentencedeterminesitsextension,ortruth-value.Discussionofmeaningintermsoftruthconditionsis,ofcourse,onefocusofsemantics.Studiesofmeaningconcernedwithphrasesandsentences,butnotwithcontexts,issometimesdescribedas‘truth-conditionalsemantics’;itexplainssemanticmeaningasthesetofconditions,orrestrictions,whichthelinguisticformoftheexpressionplacesonwhatitcantruthfullybeusedtosay.Butbeforelinguisticsemanticsbecameaseparatetopicofstudy,suchaccountshadalreadybeenproposedinphilosophy.Insearchofatruth-conditionalaccountofmeaning,wecangorightbacktoAristotle.IndeInterpretionehedistinguishesbetweensentenceswhichmakestatements(orexpresspropositions)andthosewhichdon’t;sentenceswhichmakestatementsare‘thoseinwhichthereistruthorfalsity’(4,17a,3).HemakesitclearinCategoriaethat,whileapropositionissomethingwhichhasthepotentialtobetrueorfalse,anystatementorbeliefofthatpropositionmaywelldependforitsactualtruth-valueoncontext.‘Statementsandbeliefsthemselvesremaincompletelyunchangeableineveryway;itisbecausetheactualthingchangesthatthecontrary[beingfalseratherthantrue]comestobelongtothem’(5,4a,35).Propositions,then,areindependentofcontext,andspecifyconditionsfortruth.Theindividualqualityofbeingtrueisdependentontheseconditionsbeingfulfilled.Aristotle’sdefinitionisanearlyversionofwhatbecameknownasthecorrespondenceaccountoftruth;aproposition,orsentence,istrueifitcorrespondstoreality.4Suchanaccountbecamecurrentagaininthetwentiethcentury,whenitwasgenerallysummedupasfollows:7)‘p’istrueiffp.Notethatthereissomeroomforconfusionhereastowhether7)isaruleaboutsentencesoraboutpropositions.Theoccurrenceofpinquotation\n76Truthandrealitymarkscouldbeseenasreferringeithertoanyparticularproposition,ortotheforminwhichapropositionisexpressed,suchasabelieforasentence.Correspondenceaccountsoftruthhavebeencriticisedforbeingunclearonthispoint,andit’scertainlypossibletofinddifferentphilosophersworkingwitheachoftheseinterpretations.Forinstance,thephilosopherLudwigWittgensteinexplainsinhisTractatusLogico-Philosophicus,firstpublishedinEnglandin1922,that:Tounderstandapropositionmeanstoknowwhatisthecaseifitistrue’(4.024,emphasisadded).BertrandRussell,inHumanKnowledge(1992),adoptstheotherinterpretation.Hedescribestruthandfalsehoodas‘externalrelations’,inthatyoucan’ttellwhetherasentenceorbelief,forinstance,istruesimplybylookingatitscontent.ButRussellthenpointsoutthat:‘althoughunderstandingthesentencedoesnotenableyoutoknowwhetheritistrueorfalse,itdoesenableyoutoknowwhatsortoffactwouldmakeittrueandwhatsortwouldmakeitfalse’(128,emphasisadded).Forourpresentpurposes,however,weneedn’tworryaboutthesedifferentinterpretations,butcanconcentrateonsomeoftheconsequencesofcorrespondenceaccountsofmeaning.Aswehaveseen,therearesomesentences,describedasanalytic,forwhichitdoesn’tseemtobenecessarytostateanyparticularconditionsfortruth.Inotherwords,theydon’tneedtobeshowntocorrespondtoanyparticularsituationinordertobedescribedas‘true’.Inaccountsofmeaningdependentontruthandcorrespondence,aspecialcaseisoftenmadefortheclassofanalyticsentences,onthegroundsthatourjudgementsaboutthemdon’tdependonexperience.AnalyticandsyntheticsentencesWesawintheintroductiontothischapterthatcertainsentencesareknownasanalyticbecauseaprocessofanalysisisallweneedtodecidethattheyaretrue;theindividualwordstheycontainmakethemnecessarilytrue.Incontrast,formanysentences,suchasour‘cherrytree’example,weneedtolookbeyondtheindividualmeaningsofthewordstodeterminetruth-value.Sentencessuchastheseareknownassynthetic.Ifanalysisisconcernedwithbreakingacomplexwholedownintoitsconstituentparts,synthesisistheoppositeprocessofbuildingsomethingfromseparateelements.Asyntheticstatement,then,ismadeupofseparate,independentparts.Thereisnonecessaryconnectionbetweenthemeaningsoftheseparts,soitisn’tpossibletoestablishthetruthofasyntheticsentencejustbyconsideringtherelationshipbetweensubjectandpredicate.Knowledgeofthetruthofanalyticsentencesissaidtobeapriori.Tosaythisistosaythatnootherknowledgeneedstoprecedeit;itislogically‘prior’tootherknowledge.Inotherwords,aswehaveseen,wedon’tneedtorefertoanyexperienceoftheworldbeforewecanjudgesucha\nTruthandreality77sentencetobetrue.Itusedalsotobecommontoclaimthatsyntheticsentencescouldn’tbeapriori,preciselybecauseweneedtoknowsomethingabout‘howthingsare’beforewecandecideontheirtruth-value.Thisapparentlyclear-cutdistinctionwaschallengedinthelateeighteenthcenturybytheGermanphilosopherKant.ImmanuelKant’slifewasapparentlyuneventful,buthiseffectonmodernphilosophyprofound.HelivedinKnigsberginEastPrussiaforallofhiseightyyears,gainingareputationasateacherattheuniversitythere.KantwasaphilosopheroftheEnlightenment.Hewasbornin1724,justtwentyyearsafterthedeathofJohnLocke,who,aswesawinChapter1,arguedthattheonlyphenomenaofwhichwecanhavedirectknowledgewerethosewecanexperience:ineffect,ourideasofobjectsratherthanactualobjectsthemselves.Kantcommentsseveraltimesinhiswritingsonthecontributionmadeby‘thecelebratedLocke’.However,hedrawsattentiontothelimitationsofLocke’saccountinexplainingtheoriginsofknowledge.ForLocke,allknowledgeandunderstandingmustbederivedfromexperience;knowledgeofgeneralprinciplesmustbederivedfromanaccumulationofexperiencesofparticularinstances.Kant,however,arguesthatmanyofthemeansbywhichwederiveknowledgecan’tbeempiricalinorigin.In1781,Kantpublishedacollectionofhiswritings,includinghisresponsetowhathesawasthescepticismoftheempiricists,inhisCritiqueofPureReason,aworkwhichherevisedandrepublishedin1787.Thetitlereflectshisinterestindeterminingthoseaspectsofhumanknowledgewhichmustbeaccountedforasdependenton‘pure’,ornaturalunderstanding,asopposedtoexperience.Followingtheconventionofthetime,Kantdistinguishesbetweenclassesofknowledgewhichareaprioriandthosewhichareaposteriori,ordependentonandfollowingfromexperience.However,hedepartsfromconventionbyarguingthatit’snotonlyanalyticstatementsofwhichwehaveaprioriknowledge.Rather,certainsyntheticstatementscanbeknowntobetrue,andnecessarilytrue,withoutreferencetoexperience,aswouldbedemandedbyempiricism,andwithoutdemonstrationbylawsofreason,aswouldberequiredbyrationalism.Kanthimselfusestheexamples8)and9),towhichwecouldaddtheratherlesstechnicalpair10)and11),todistinguishbetweenanalyticandsyntheticjudgements.Kantexplainedthatin8)thepredicateiscontainedintheconceptofthesubject,whilein9)thepredicateisoutsideoftheideascontainedinthesubject.Anyanalysisofthemeaningof‘body’(oranythingwhichhassubstance)wouldnecessarilyincludetheideaof‘extended’(ortakingupsomeamountofspace),butnottheideaof‘heavy’.Similarlyallpuppiesarenecessarily,bydefinition,youngdogs,buttheyarenotbydefinitionadorable.8)Allbodiesareextended.\n78Truthandreality9)Allbodiesareheavy.10)Allpuppiesareyoungdogs.11)Allpuppiesareadorable.Anotherwayoflookingatthisistosaythatananalyticstatementsuchas10)islogicallynecessary;thenegationofit,asin12),isself-contradictory.Asyntheticstatementsuchas11),however,islogicallycontingent;thenegationofitin13)isnotalogicalcontradictionbecausetheconcept‘adorable’isnotpresentintheconcept‘puppy’.12)Notallpuppiesareyoungdogs.13)Notallpuppiesareadorable.Kantexplainsthat‘judgementsofexperience,assuch,areoneandallsynthetic’(B11).5The‘purereason’withwhichheisconcerned,however,isthatclassofknowledgewhichissyntheticbutisnotbasedonexperience:knowledgewhichissyntheticandapriori.Wemightdescribeouraprioriknowledgeasour‘intuitions’abouthowtheworldis,butitisinfact,andcanonlyeverbe,knowledgeoftheworldasweperceiveit.Thisisbecauseweexperienceonlyourownperceptionsoftheworld,themselvesbasedoncertainpreconceivedideas,suchasthoseaboutquantity,spaceandrelation.TheworldofwhichKantwrites,therefore,consistsinhumanexperienceofrealityratherthanrealityitself.Indeed,forKanttheworldissimplynotavailabletouswithoutthemediationofourperceptionofit.Heclaimsthat‘whatobjectsmaybeinthemselves,andapartfromallthisreceptivityofoursensibility,remainscompletelyunknowntous’(A42).These‘thingsinthemselves’,ofwhichweashumanbeingscanneverobtainanycertainknowledge,aredescribedbyKant,andbymanyphilosopherswhohavefollowedhim,astranscendent.Thisadjectiveisrelatedtotheverbtranscend,whichmeanstogo,ortobe,beyondcertainlimits.Inthisparticularcase,itisusedtorefertothatwhichgoes,oris,beyondthelimitsofexperienceandresultantknowledge.Inthis,Kantisinterestedinwhatweunderstand,orinfer,tobethecasebyexerciseofourreasonbutwithoutanyfirmbaseinexperience.Inarelatedusage,hereferstoourintuitionsabouttheworld,oraprioriconditionsforknowledge,astranscendental;suchaspectsofhumanunderstandingarenecessarypreconditionsforourinterpretationofanyexperience,butareindependentofit,sothattheytooexistbeyondthelimitsofexperience.Themoregeneraltermtranscendentalismisnowadaysusedtorefertoanyphilosophyorbeliefwhichemphasisestheimportanceofunderstanding,orofemotion,abovethatofactualexperienceoftheworld.Wehaveyettospecifywhatsortsofknowledgecanbeatonceapriori,ortranscendental,andsynthetic.Kantsuggeststhatknowledgeofthistypecanbefoundinmathematics.Hedescribesourknowledgeofmathematicalpropositionsasbeingapriori,butthepropositionsthemselvesasnotbeing\nTruthandreality79analytic.Weknowthat14)istruebecauseitisconsistentwithourviewofhowmathematicsworks.Ourknowledgeofthisisdependenton,andderivedfrom,ourwayofunderstandingtheworld.Nodegreeofanalysisoftheindividualelementsof14)can‘prove’itstruth.So,unlikewithanalyticsentencessuchas8)and10),themeaningofthepredicatefourisnotactuallycontainedinthemeaningofthesubjecttwoplustwo.14)Twoplustwoequalsfour.Ifwerefertothe‘realworld’,forinstancebycountingtwobeanstogetherwithtwootherbeansandfindingthatwehavefourbeans,wewon’tactuallyhaveprovedthat14)istrue.Wefindthatwehavefourbeansjustbecauseoursystemofmathematics,oforganisingentitiesandquantities,tellsusthatwehavefour.Whethertheworldasidefromourperceptionofitactuallymakes‘2+2=4’trueorfalseisnotsomethingtowhichwehaveaccess.Kant’sownpresentationofthisargumentissummedupasfollows:Theassertionthat7+5isequalto12isnotananalyticproposition.Forneitherintherepresentationof7,norinthatof5,doIthinkthenumber12’(A164).ForKant,suchmathematical,aprioriknowledgepresentsafurtherproblemforempiricism.Wemightbesaidtohavegeneralknowledgeofthetruthofsuchpropositions,butwecanonlyeverhaveparticularexperiencesoftheworld.Ofcourse,it’snotjustmathematicalpropositionsofwhichwehavegeneralknowledge,buttheyneedtobedistinguishedfromgenuineempiricalgeneralisations,suchas15).15)Allbirdshavefeathers.Thegeneralisationin15)mightbedescribedas‘probable’—webelieveittobetruebecauseforeveryinstanceofbirdthatwehaveexperienced,thepredicatefeathershasbeenapplicable.Wecanimaginethetypeofexperiencewhichwouldcauseustodoubtthevalidityof15),howeverunlikelywethinkthismightbe.Butwesimplycannotimagineanyexperiencewhichwouldcauseustodoubtthetruthof14).Wecanimagineobservingafeatherlessbird,butwecan’timagineobservingtwoobjectsplustwoobjectsmakingfiveobjects.Thedifferencebetweenthesetwotypesofknowledgeis,Kantargues,somethingwhichapurelyempiricalaccountofknowledgejustcan’texplain.‘Experienceteachesusthatathingissoandso,butnotthatitcannotbeotherwise’(B3);experiencealonecan’texplainthecertaintyofourknowledgeof14).Kant’schiefinterestwasnotinlanguagebutinthenatureofknowledgeandunderstanding,anditisinthisareaofphilosophythathisinfluencehasbeenmoststronglyfelt.However,hisworkalsoprovidesagoodstartingpointforaconsiderationofdifferenttypesoftruth.Inwhatfollowswewillbeconcernedmainlywiththemeaningofsentenceswhich\n80Truthandrealityaresynthetic,butofwhichwecan’tclaimaprioriknowledge.Inotherwords,wewillbethinkingaboutsentenceswhichareneithernecessarilytruenornecessarilyfalse:sentenceswhichcanbeusedtomakeinformativestatementsabouttheworld.ItfollowsfromKant’saccountofthesethattodescribethemas‘true’cannevermeanforcertainthattheycorrespondwithreality,butatbestthattheycorrespondwithourparticular,humanperceptionofit.Next,weneedtoconsiderthewaysinwhichtheseproblems,andthegeneralthemeoftherelationshipbetweenmeaningandtruth,wereapproachedbysomeofthemainschoolsofphilosophyinthetwentiethcentury.Wewillstartwiththeconceptofverification,whichreceivedmuchattentioninthemiddlepartofthatcentury.VerificationInordertostudyexplanationsofmeaningintermsofverification,weneedtomoveforwardsome150yearsfromthetimeinwhichKantwaswriting,andconsidertheworkoftwentieth-centuryphilosophersknownaslogicalpositivists.Wewill,however,stillbeconcernedwiththetopicsofunderstanding,knowledgeandexperience.Logicalpositivistsaresometimesreferredtoaslogicalempiricists;theiraccountofmeaningis,crucially,dependentonexperienceoftheworld.Itisn’thardtotracethephilosophicalpredecessorsofthelogicalpositivists.Thebranchofphilosophicalthoughtknowngenerallyaspositivismhadbeencurrentsincethenineteenthcentury.Thistimingwasnocoincidence;thegrowthofpositivismwascloselylinkedtotheexpansioninscientificdiscoveryandknowledge.Positivistswereinterestedinknowledgewhichcouldbegainedthroughscientificinvestigation,notinknowledgewhichdidn’trelyonobservationandcouldn’tbetestedbyexperiment.Theythereforerejectedallclaimstometaphysicalknowledge,suchasthoserelatingtoreligionorsuperstition.WecanspotasimilarityherewiththeideasweconsideredinChapter1inconnectionwiththeBritishEmpiricists,andparticularlywithJohnLocke.Lockeclaimedthatwehaveaccessonlytoourideasofobjects,dependentoninformationfromthesenses,nottoobjectsthemselves.Thepositivists,somedecadeslater,claimedthattheonlytypeofknowledgewhichispossibleisknowledgebasedontheexperiencesofthesenses,andthereforethattheonlystatementswhichwecanmakewithanyconfidencearethoseaboutsensedata.Positivism,then,wasaradicallyempiricalschoolofthought.Itsnameisrelatedtotheeverydaymeaningofthewordpositiveas‘absolute’or‘definite’.Theknowledgeinwhichthepositivistswereinterestedwastheabsoluteknowledgederivedfromobservationoftheworld.Thetermslogicalpositivismandlogicalempiricismareappliedtothedevelopmentofpositivisminthetwentiethcenturyconcernedparticularlywithmeaningandtruth.Thisdevelopmentismostcloselyassociatedwitha\nTruthandreality81groupofphilosophersknownastheViennaCircle,whoheldweeklymeetingsinViennainthe1920sand1930storeadpaperstogetheranddiscussideas.Thesewerenotexclusivelyphilosophicalideas;thegroupwasalsoconcernedwithpolitics,andpublishedworkonsocialreformanddemocracy.ThisbroughtthemintoconflictwiththerisingpowerofNationalSocialism,andmanyofthemwereforcedtoleaveEuropeforAmericaintheyearsleadinguptothestartoftheSecondWorldWar.Asaresult,theViennaCircleceasedtomeetinthe1940s.ButbythistimeithadbecomeextremelyinfluentialonContinentaland,byinfluence,onBritishphilosophy.TheindividualmembersoftheViennaCirclewereconcernedwithavarietyofdifferentareasofscienceandmathematics.Theiraimasagroupwastodevelopaformoflanguageforexpressingscientificobservation.Theywereconcernedwithidentifyingthe‘meaningful’statementsofscience,asopposedtothe‘meaningless’statementssuchasthoseofmetaphysics,religion,ethicsandaesthetics.Thestatementswhichtheyclassifiedasmeaningfulwereofthreetypes.First,therewereanalyticalstatements,ofthetypewehavejustconsidered,inwhichtruthisapparentfromananalysisofthestatementitself,withoutanyneedtorefertoexternalreality.Second,therewerethestatementsofmathematicsandoflogicwhich,aswehaveseenfromtheworkofKant,can’tbeprovedbyobservationoftheworld.6Thethirdcategoryofstatementswhichtheyclassifiedasmeaningfulweresyntheticstatementswhichcouldbesubjecttoanexplicitprocessofverification.Thisthirdcategoryisthelargestandmostdiverse,andcontainsallotherstatementswhicharepermissibleinscientificdiscussion.Astatementwhichcanbesubjecttoverification,orisverifiable,isonewhichiscapableofbeingtested,andhenceeitheracceptedastrueorrejectedasfalse,byobservation.Soforalogicalpositivist,16)and17)wouldcountasmeaningfulstatements,while18),19)and20)aremeaningless,andthereforehavenoplaceinscientificdiscussion.16)WaterfreezesatzerodegreesCelsius.17)Thesunmovesroundtheearth.18)Godisgood.19)Murderiswrong.20)Dreamsareanexpressionofsubconsciousfearsanddesires.Weknowwhattheappropriatemethodofverificationfor16)wouldbe;withthehelpofathermometerwecouldproduceexperimentalevidencetoshowthatitistrue.Similarly,weknowwheretolooktofindevidencerelevantto17);modernastronomyinfacttellsusthatitisfalse.Thatisnottosaythatstatementssuchasthisaremeaningless.Example17)canbesubjecttoaprocessofverificationjustas16)can;itisjustthatthisprocesswillshowthat17)isfalse.Forexamples18),19)and20),ontheother\n82Truthandrealityhand,thereisnomethodofverificationavailable.Noonehasbeenabletosuggestexperimentseithertoprovethetruthortoestablishthefalsehoodofstatementssuchasthese.Attitudestothemarebasedonspeculation,argumentand,ofcourse,belief.Forlogicalpositiviststhesestatementsaresimplymeaningless,preciselybecausewecan’tproduceobservableevidencetodemonstratethattheyareeithertrueorfalse.Atheoryoftruthbasedonverificationisthereforeatypeofcorrespondencetheory;tobetrue,astatementmustcorrespondwiththefacts.However,verificationspecifiesafurthernecessaryconditionfortruth.Thefactstowhichthestatementcorrespondsmustbeavailableto(someformof)observation.Inthisway,verificationoffersameansofdistinguishingbetweenstatementswhicharemeaningful,andthereforescientificallyvalid,andthosewhicharemeaningless.Butthelogicalpositivistssawverificationasmorethanthis.Theysawitasamethodofdescribingthemeaningofindividualstatements.Forthem,themeaningofastatementisdefinedbythemethodofverificationappropriateforit.Inperformingtheexperimentnecessarytoverify16),forinstance,youwillinfactbedemonstratingitsmeaning.OnememberoftheViennaCirclewhoproducedparticularlyinfluentialworkonmeaningwasRudolphCarnap,whosedefinitionofpropositionalityweconsideredinChapter2.Carnaparguesthatmanyoftheapparentproblemsofphilosophycaninfactbereducedtoproblemsofexpression.Inparticular,hearguesthatproblemsofmetaphysicsare‘pseudo-problems’,andthattheycanberevealedassuchbyasuitablyrigorousanalysisofthelanguageinwhichtheyareexpressed.Ineffect,metaphysicalstatementsturnouttobemeaningless.Carnapisanxioustomakeclearthatindismissingthestatementsofmetaphysics,membersoftheViennaCirclearearguingnotagainsttheircontent,aboutwhichtheyinfacthavenothingtosay,butagainsttheirform,againstthelanguageinwhichtheyareexpressed.InTheLogicalSyntaxofLanguage,publishedinEnglandin1937,Carnapcomplainsthat:WhenweoftheViennaCirclecriticize,inaccordancewithouranti-metaphysicalview,certainsentencesofmetaphysics(suchas:‘ThereisaGod’)orofmetaphysicalepistemology(suchas:Theexternalworldisreal’)weareinterpretedbythemajorityofouropponentsasdenyingthoseobject-sentencesandconsequentlyaffirmingothers(suchas:ThereisnoGod’or:Theexternalworldisnotreal’,etc.).Thesemisunderstandingsarealwaysoccurringinspiteofthefactthatwehavealreadyexplainedthemmanytimes…andareconstantlypointingoutthatwearenottalkingaboutthe(supposititious)facts,butaboutthe(supposititious)sentences.(309–10)LikeothermembersoftheViennaCircle,Carnapwasinterestedinthe\nTruthandreality83analysisofthelanguageofscience.HisaiminTheLogicalSyntaxofLanguageistoestablishaformallanguage,oratleastthepropertiesofaformallanguage,inwhichtheresultsofanalysisinmathematicsandphysicscanbeexactlyexpressed.Aformallanguageisoneinwhichcategoriesofsymbols,andrulesbywhichexpressionscanbeconstructedfromthesesymbols,arespecified,butinwhichnoreferenceneedbemadetothemeaningofsymbolsorexpressions.Carnaparguesthatsuchartificiallanguagesaremoreappropriatetoscientificdiscoursethananynaturallanguage.Heseesnaturallanguageasimpreciseandimperfect;‘theunsystematicandlogicallyimperfectstructureofthenaturalword-languages’(2)makesitimpracticaltotrytoformulatetheirrulesandstructuresprecisely.Significantly,though,Carnapdoesn’tseehisdiscussionofformal,artificiallanguagesasbeingnecessarilyirrelevanttothestudyofnaturallanguage;the‘generalcharacter’ofthesyntacticrulesheisstudyingmightbeapplicabletoit.Hecomparestheanalysisofformallanguagestotheapproachofthephysicist,whoisinterestedinthelawsofnaturebutdoesn’tattempttoaccountdirectlyfornaturalphenomena.Carnapcontinues:Inthefirstplacethephysicistrelateshislawstothesimplestofconstructedforms;toathinstraightlever,toasimplependulum,topunctiformmasses,etc.Then,withthehelpofthelawsrelatingtotheseconstructedforms,heislaterinapositiontoanalyzeintosuitableelementsthecomplicatedbehaviourofrealbodies,andthustocontrolthem.(8)Thematerialwhichphysicistsworkwithmayseemfragmentaryandthecontextofthelaboratoryartificial,buttheirfindingscanneverthelessberelevanttothoseinterestedinexplainingnaturallyoccurringphysicalphenomena.Inthesamewayanartificiallyconstructed,andthereforefullydescribed,languagemayappeartohavelittleincommonwithanaturallanguageinallitscomplexity,butcanofferawayofexplainingitsregularities.Morerecentsyntactictheories,developedinlinguistics,have,ofcourse,beenmoreoptimisticaboutthepossibilityofdescribingnaturallanguage.Eventhese,however,generallyallowsomedegreeofsimplification,oridealisationofthelanguage.Carnap’scomparisonwiththeisolatedexperimentsofphysicistssuggestsonewayinwhichbothphilosophersandlinguistsmightbeconsideredjustifiedinthesesimplifications.Asthelogicalpositivistspublishedandcirculatedtheirwork,itbecameinfluentialbeyondtheconfinesofViennaandindeedofcontinentalEurope.BertrandRussell,forinstance,acknowledgedtheinfluencewhichthelogicalpositivistshadonhiswork.AnotherBritishphilosopherparticularlyinterestedintheseideaswasAlfredAyer,whohadbecomefamiliarwith\n84TruthandrealitythemwhilespendingsometimeinViennainthe1930s.AftertheSecondWorldWarAyer,whoismostcommonlyreferredtoas‘A.J.Ayer’,becameaprofessorattheUniversityofLondonandlateratOxford,wherehehadbeenanundergraduate.Butitwasin1936,attheageoftwenty-six,thathewroteoneofhismostsuccessfulbooks,Language,TruthandLogic.Inthisshortbook,AyerintroducedanEnglish-speakingreadershiptotheideasoflogicalpositivism,aradicallynewapproachinthecontextofOxfordphilosophyatthattime.HeacknowledgeshisgreatestintellectualdebtasbeingtotheViennaCircle,andparticularlytoRudolphCarnap.Ayeroffersaparticularversionofthecriterionofverifiabilitytodefinemeaningfulness.Thiscriterion,sometimesknownastheverificationprinciple,formsthecentralideaoflogicalpositivism,andhadbeenstatedinvariousformsbymembersoftheViennaCircle.Ayerdiffersfromtheseinthatheseesthemethodofverificationasbeingonlythecriterionofmeaningfulness,ratherthanitselfbeingthemeaningofanexpression.Hethereforedescribeshisprincipleas‘weaker’thanthatoftheViennaCircle,andexpressesitasfollows:Wesaythatasentenceisfactuallysignificanttoanygivenpersonif,andonlyif,heknowshowtoverifythepropositionwhichitpurportstoexpress—thatis,ifheknowswhatobservationswouldleadhim,undercertainconditions,toacceptthepropositionasbeingtrue,orrejectitasbeingfalse.(48)Manyobjectionshavebeenraisedtotheimplicationsoftheverificationprinciple,inallitsvariousforms.Forinstance,ithasbeensuggestedthatit’simpossibletofindanysuremethodofverificationforstatementsofeventsinthepast,yetitdoesn’tseemreasonabletorejectallsuchstatementsasmeaningless.Anotherobjectionconcernsthenatureoftheprincipleitself,forwhichnomethodofverificationcanbesuggested.Hence,theargumentgoes,ifweaccepttheprincipleasvalidwemustimmediatelydismissitasmeaningless.Ayerhimselfrevisedhisopinionofthevalueofverificationlaterinhiscareer,forinstanceinTheCentralQuestionsofPhilosophy,firstpublishedin1973.However,heretainedanessentiallyempiricistapproachtoknowledgeandmeaning.Hemaintainedthatitmustatleastbepossibletoidentifythestateofaffairswhichwouldmakeapropositiontrueorfalse,eveniftherelevantfactorsarenotavailabletoinspection.Ofthismoremodestclaimhecommentedinthelaterbook:TheonlyobjectiontothisviewwhichIcanseeisthatitisnotveryilluminating’(30).Inattemptingtodevelopaworkableverificationprinciple,thelogicalpositivistsweretryingtoproducea‘correspondence’accountoftruthsufficientlyrigorousforthepurposesofscientificdiscussion.Theyarguedthatitisn’tenoughtosaythatatruestatementisonewhichcorresponds\nTruthandreality85withthefacts;forthepurposesofscienceitmustbepossibletospecifywhatobservablephenomena,orwhatsensedata,wouldjustifythescientistinacceptingastatementastrue,orrejectingitasfalse.Theproblemswithsuchanapproach,whichA.J.Ayeracknowledgedinhislaterwork,wereperhapslargelytodowiththefactthatitwasjusttoorigoroustobeworkableoutsideofalaboratory;therearemanystatementswhichwewouldwanttoacceptasperfectlymeaningful,forwhichproofbyobservationissimplynotpossible.Verificationistaccountsarealsolimited,alongwithothercorrespondencetheories,tospecifyingconditionsforthetruthofdeclarativestatementsandthepropositionstheyexpress.Theycansaynothingaboutthemanyotherwaysinwhichlanguageisused.Wewillreturntothislimitationlaterinthischapter.First,wewilllookatanothertwentieth-centuryattempttoproducearigorousaccountoftruth,thistimeonewhichfocusesontruthconditionsthemselves,ratherthanontheabilitytodetermineempiricallywhethertheseconditionshold.Theseaccountsaregenerallyknownas‘truththeories’or,intheirlaterversions‘truththeoriesofmeaning’.TruththeoriesThenamesmostcloselyassociatedwithaccountsofthistypearethoseofAlfredTarskiandDonaldDavidson.Tarskiwasalogician,andworkedattheUniversityofWarsawatmuchthesametimeasthelogicalpositivistsweredevelopingtheirideasinVienna.HetoowasforcedtomovebyeventsinEurope,leavingPolandin1939.AfewyearslaterhetookupapositionattheUniversityofBerkeleyinCalifornia,wherehecontinuedworkonhisaccountoftruth.Davidsonlaterbecameaprofessoratthesameuniversity,buildingonanddevelopingTarski’sideas.In1944AlfredTarskipublishedanarticleentitled‘Thesemanticconceptionoftruthandthefoundationsofsemantics’.Hisaiminthisarticleistodevelopafullytruth-conditionaltheoryoflanguage.However,Tarskiisn’tattemptingtoproducesuchatheoryforanynaturallanguage,suchasEnglish.Heisinterestedonlyinlanguageswhichcouldbethoroughlydefinedintermsoftruth;hearguesthattheseareallartificial,ortechnicallanguages.LikeCarnap,Tarskirejectsnaturallanguageasbeingtoovagueandtooimprecisetobecapableoffulltheoreticalexplanation.Tarskilabelsthelanguageunderscrutinythe‘objectlanguage’.InmostofhiswritingstheobjectlanguageisinfactafragmentofEnglish,butit’simportanttobearinmindthatit’snotEnglishitself,oranyothernaturallanguage,whichheisdescribing.Tarskiattemptstoprovideanaccountofanyobjectlanguageconsistentwiththe‘classical’modeloftruth.Suchamodeldefinesallstatementsaseither‘true’or‘false’,suchthat:Thetruthofasentenceconsistsinitsagreementwith(orcorrespondenceto)reality’(62).7However,heseesthisnotionof‘correspondence’astooimprecisetogiveasatisfactoryaccountofany\n86Truthandrealityobjectlanguage;hewantstorefineitbymakingexplicittheconditionsfortruthofeverysentence.TheaccountoftruthwhichTarskiproposesisonewhichentailsaseriesofstatements,eachofwhichgivesthetruthconditionsforasentenceintheobjectlanguage.ThesestatementsarepresentedasaseriesofT-sentences,oneforeachsentenceoftheobjectlanguage.EachT-sentencetakesasentenceintheobjectlanguageandstatesthoseconditionsunderwhichitwouldbetrue.Ineffect,then,Tarski’saccountpairseachsentenceintheobjectlanguagewithafurthersentence.HeexplainsthatnoindividualexampleofaT-sentenceisitselfadequateasadefinitionoftruth.EachT-sentenceis,rather,apartialdefinitionoftruth,accountingforoneparticularsentenceintheobjectlanguage.Acompletedefinitionoftruth,ontheotherhand,‘hastobe,inacertainsense,alogicalconjunctionofallthesepartialdefinitions’(63).Tarskiintroducesaversionoftheformulationofthecorrespondenceaccountwhichweconsideredearlierinthischapter.EveryT-sentencetakestheform:21)Xistrueiffp.Inthisformula,‘X’standsforanysentenceintheobjectlanguage,and‘p’forapropositionstatingtheconditionsinwhichitistrue.TheexamplewhichTarskiusestoillustratethisisgivenin22).Notethatinthiscase‘X’isinfactasentenceofEnglish,butitneedn’tbe.22)Thesentence‘snowiswhite’istrueiffsnowiswhite.Thismayatfirstsightappearratherdisappointingasapartialdefinitionoftruth.Itlooksdistinctlyuninformative.ButthisT-sentencedoesn’tinfactdowhatitmightappearto;itdoesn’trepeatitself,orsaythesamethingtwice.Tounderstandthis,weneedtonotethat‘snowiswhite’intheantecedentiscontainedininvertedcommas,andthentorememberwhatwelearntinChapter2aboutthedistinctionbetweenuseandmention.Infact,theexpression‘snowiswhite’isusedonlyintheconsequentof22).Wecouldsaythatitisusedastheconsequentof22).Intheantecedentitisonlymentioned,ornamed;aparticularsentenceintheobjectlanguageisidentified,sothatsomethingcanbesaidaboutit.ThisdistinctionwillperhapsbeclearerifweconsideraT-sentenceexpressedinEnglishbutconcernedwithanobjectlanguageotherthan(afragmentof)English.WecouldprovideaT-sentencefortheFrenchsentence‘Laneigeestblanche’asfollows:23)Thesentence‘Laneigeestblanche’istrueiffsnowiswhite.RememberalsothatTarski’saccountoftruthdoesn’tconsistjustinspecifyingthataT-sentencemusthavetheformin21).Itcruciallyincludes\nTruthandreality87hisclaimthatasetofT-sentences,onecorrespondingtoeachsentenceintheobjectlanguage,isasufficientdefinitionoftruthforthatlanguage.Nowaslinguistsweknowthatallnaturallanguagesareinfinite.Itwouldbeimpossibletolistallthesentencesofanaturallanguage,soitfollowsthatitmustbeimpossibletocomeupwithacompletesetofT-sentences.ItmightseemtobequiteaproblemforTarskithat,accordingtohisdefinition,anyaccountoftruthforanaturallanguagemustalwaysbeincomplete,orpartial.ButrememberthatTarskidoesn’tclaimtobewritingaboutnaturallanguage.Hewasamathematicianandalogician,notalinguist,andhisinterestlayinwhathedescribesas‘theoretical’ratherthan‘descriptive’semantics:inprovidinganaccountoftruthforanartificiallanguageratherthanexplainingnaturallanguage.Hesawhimself,asdidsomeofthemembersoftheViennaCircle,ascontributingtothedevelopmentofarestricted,fullyspecifiedlanguagewhichwouldservemorepreciselythannaturallanguageasamediumforscientificdiscourse.Alsolikethelogicalpositivists,hesawnaturallanguagesasnecessarilyvagueandinexact,andthereforenotamenabletoscientificstudy,atleastwithoutfirstbeingneatenedandtidiedup.Inmanyways,then,Tarski’saimsandmethodswereverydifferentfromthoseofmodernlinguists,anditwouldn’tbefairtocriticisehisworkfornotfittingtherequirementsofadifferentdisciplineatadifferenttime.Itis,however,interestingtocompareTarski’sworkwiththatofDonaldDavidson,particularlyDavidson’s1967article‘Truthandmeaning’.UnlikeTarski,Davidsonisinterestedinnaturallanguage.Inthetwenty-threeyearsbetweenthepublicationofTarski’s‘Thesemanticconceptionoftruth’andDavidson’sTruthandmeaning’,linguisticshadstartedtotakeoffasaseparateacademicdiscipline,andnaturallanguagehadbecomealegitimatefieldofstudyinitsownright.Andsomephilosophers,too,hadbeguntotakemoreofaninterestinnaturallanguageandeveninactualusage.In‘Truthandmeaning’,DavidsonisquiteoutspokeninhisdisagreementwithTarskiastothepurposeofanaccountbasedontruth,claimingforinstancethat:‘ThetaskofatheoryofmeaningasIconceiveitisnottochange,improveorreformalanguage,buttodescribeandunderstandit’(98).8AlsounlikeTarski,DavidsonseesthelistofT-sentencesasitselfprovidingadescriptionnotjustoftruth,butofmeaning.Inotherwords,forDavidsonatheoryoftruthisnotjustanecessarypreliminaryfor,butisinfactadequateinitselfas,atheoryofmeaning.Hismainclaim,then,isthatafullaccountofmeaningforanaturallanguagecanbeprovidedbyatheoryoftruthconditionsforitssentences.DavidsonidentifiestwoparticularproblemswithTarski’saccountwhichcallforittobemodified.Thefirstwasn’taproblemforTarskibecauseofhisdifferentgoals;it’stheproblemwhichwehavealreadynotedinconnectionwiththenatureofnaturallanguage.Thenumberofsentencesinanylanguageisinfinite,sotheywouldneedtobematchedwithaninfinitenumberofT-sentences,makingacompleteaccountoftruthforany\n88Truthandrealitylanguageunobtainable.ThesecondproblemwhichDavidsonidentifiesistodowiththelogicalnatureoftheoperatoriff,whichformspartofeveryT-sentence.AswesawinChapter2,thelogicaloperatorsareconcernedjustwithmappingtruth-valuesontofurthertruth-values.Unliketheexpressionswhicharesometimesdescribedastheir‘naturallanguagecounterparts’,theydon’tsuggestanyfurtherrelationshipbetweenthevariableswhichtheyconnect.Wesawinparticularthat‘materialcondition’,includingiff,requiresnoconnection,intermsofsubjectmatterorrelevance,betweenantecedentandconsequent.ThishastheunfortunateresultthatanyT-sentenceinwhichtheconsequentandtheantecedent‘matchup’intermsoftruth-valuemustbeacceptable,regardlessofwhatsentencesoftheobjectlanguageitisbeingusedtodescribe.SoaswellasTarski’s22)above,24)belowmustbecountedasanacceptableT-sentence.24)‘Snowiswhite’istrueiffgrassisgreen.Thepropositionthatgrassisgreenhasthesametruth-valueasthepropositionthatsnowiswhite;theyarebothtrue.InthepurelylogicalformulaofaT-sentence,therefore,itoughttobepossibletosubstituteonefortheother;24)shouldbeasvalidas22),asshouldanyT-sentencewhichcombines‘snowiswhite’withatrueproposition.AnditshouldbepossibletoprovideT-sentencesforallthetruesentencesofalanguagebycombiningeachwithanytrueproposition,orindeedallwiththesametrueproposition.This,clearly,isnotanacceptablestateofaffairswhentryingtodescribethemeaningofanaturallanguagebymeansofasetofT-sentences.TheT-sentencein24)maybelogicallyvalid,butittellsusverylittleaboutEnglish.AsDavidsonhimselfexpressesit,‘anytwosentenceshavethesamereferenceiftheyhavethesametruth-value.Andifthemeaningofasentenceiswhatitrefersto,allsentencesalikeintruth-valuemustbesynonymous—anintolerableresult’(93).ThesolutionwhichDavidsonproposestobothoftheseproblemsistosupplementthelistofT-sentenceswithanaxiomatictheory.Axiomscanbedefinedasprinciplesorrules,muchliketherulesofagrammar,which,whentakentogether,aresufficienttoproduceaseriesofformulae,ortheorems.Inthecaseofagrammar,ofcourse,theseformulaearethesentencesofalanguage.InthecaseofDavidson’saxioms,however,thetheoremsaretheT-sentencesforthelanguage.Hisaxiomaticsystem,inproducingalltheT-sentencesofalanguage,providesanaccountofthemeaningofthelanguage.Thetypesofaxiomwhichhissystemrequiresareaxiomsofreference,ofsatisfaction,andofconnection.Wecanillustratethese,verybriefly,byconsideringwhatwewouldneedtoconstructaT-sentenceforexample25):25)Fireishot.Inordertospecifyauniquemeaningforthissentence,onewhich\nTruthandreality89distinguishesitfromallothertruesentencesofthelanguage,weneedwaysofspecifyingthemeaningofeachofitsparts.Firstly,weneedtoexplainthemeaningofthesubject,somethingwecandobyusingamaximofreference.Thesetakethegeneralform‘“x”referstox’.Again,wehaveherethementionofanexpressioninthesubjectandtheuseofthatexpressioninthepredicate.Sotheaxiomofreferenceweneedhereis:26)‘Fire’referstofire.Nextweneedawayofidentifyingandexplainingthepredicatein25),andofdistinguishingitfromallotherpossiblepredicates.Weneedan‘axiomofsatisfaction’.Thisexplainsthemeaningofapredicateintermsofthoseentitiesofwhichitistrue,ortowhichitcancorrectlybeapplied.Anentity‘satisfies’apredicateifthecombinationoftheentityandthepredicategivesatruestatement.Againdistinguishingbetweenmentionanduse,weget:27)‘Hot’issatisfiedbyhotthings.Finally,weneedtoidentifythewayinwhich‘fire’and‘hot’areconnected.The‘connectionaxiom’forisspecifiesthat:28)éaisbùistrueiffwhatéaùreferstosatisfiesébù.Thesethreeaxioms,takentogetherandappliedtotheEnglishsentenceFireishot,giveusasoutputthetheorem,orT-sentence:29)‘Fireishot’istrueifffireishot.Inthisway,DavidsonavoidsbothoftheproblemsidentifiedinTarski’saccount.TodescribeanynaturallanguagethesetofT-sentencesmustbeinfinite,buttheycannowbederivedfromafinitenumberofaxioms,justastheinfinitesentencesofanaturallanguagecanbederivedfromtheapplicationofafinitenumberofgrammaticalrules.Theseaxioms,togetherwithasystemoflogicalconstantssuchasthosewestudiedinChapter2,andagrammarofwell-formedness,aresufficienttoproduceaninfinitesetofT-sentences.TheaxiomsalsoensurethatnotjustanytruepropositioncanbepairedwithatruesentencetogiveanaccurateT-sentence.Expression29)aboveistheonlyT-sentencewhichcanbederivedfortheEnglishsentence‘fireishot’usingDavidson’saxioms.DavidsonhasappliedhisaxiomaticaccountinpracticetoonlyatinyfragmentofEnglish,butmaintainsthatafullaccountofthistypewouldinprinciplebepossibleforanynaturallanguage.However,heacknowledgescertainothercomplicationsforatruth-theoreticaccountofnaturallanguage.Forinstance,anysuchaccountisnecessarilyfocusedonlanguageasameansof‘statingfacts’ratherthanasawayof,forinstance,asking\n90Truthandrealityquestionsorissuingorders.Ingrammaticalterms,wecouldsaythattruththeoreticaccountsofmeaningarefocusedondeclaratives,attheexpenseofinterrogativesandimperatives.9Forexample,givenasuitablyelaborateaxiomatictheory,wemightbeabletoprovidethetruthconditionsfor30),butwecouldn’tstateconditionsforthetruthof31)or32),bothequallyacceptablesentencesofEnglish.30)Youwillresignyourposition.31)Willyouresignyourposition?32)Resignyourposition!Davidsonaddressesthisprobleminalaterarticle,publishedin1979,called‘Moodsandperformatives’.Heexplainsthatanyadequateaccountofmeaningmustbeabletoaccountforthedifferencesbetweenexamplessuchas30)–32),whichhedefinesasdifferencesin‘mood’,whilestillcapturingthefactthattheyall‘haveacommonelement’(15)ofmeaning.Heisawareoftheimplicationsofthisforatruththeoryofmeaning,statingthatifitcan’texplainmood‘thentruththeoryisinadequateasageneraltheoryoflanguage’(15).Theproblemisthattruth-functionaloperators,theonlytypeofelementsofmeaningallowedbysuchastheory,can’tbeusedtogiveanexplanationofmood.ThesolutionwhichDavidsonsuggestsisthatnon-declarativescanbeseenascomposedoftwoseparateparts.Oneofthesepartsisconcernedwiththe‘commonelement’ofmeaning,andtheotherwithmood.Inotherwords,themeaningofallusesofnon-declarativesmustbecapableofbeingdescribedintermsofadeclarativesentenceanda‘mood-setter’.Thesetwopartsmustbeseenasbeingproducedsimultaneouslywhenasentenceisused,but,Davidsonstresses,astotallyindependentofeachother.Forinstance,ourinterrogativeandimperativeexamplein31)and32)abovecouldbeexpressedintermsofthedeclarative30)andanindependentmood-setter,asin33)and34)respectively.33)Mynextutteranceisaquestion.Youwillresignyourpost.34)Mynextutteranceisimperative.Youwillresignyourpost.Davidsonexplainsthatboththese‘parts’aretruth-conditional,therebyfulfillingtherequirementsofatruththeory.But,beingtotallyindependent,theyaren’tcombinedusinganytruth-functionaloperator,explainingwhyexamplessuchas31)and32)can’tbeassignedtruth-values.InDavidson’sownterms,‘Eachofthetwoutteranceshasatruthvalue,butthecombinedutteranceisnottheutteranceofaconjunction,andsodoesnothaveatruthvalue’(20).Inthisway,Davidsonclaims,it’spossibletoretainatruth-theoreticaccountofmeaningwhiletakingintoaccountusesoflanguageotherthanthoseconcernedwithstatementsoffact.Anothermajorcomplicationwhichnaturallanguageraisesforanytruth-\nTruthandreality91basedtheoryofmeaningisthatwhatcountsas‘true’willoftendependonindividualcontextsofutterance.Onlyasmallproportionofusesofnaturallanguageinvolvestatementsofthe‘snowiswhite’,‘fireishot’nature.Davidsoncommentsin‘Truthandmeaning’that‘thesamesentencemayatonetimeorinonemouthbetrue,andatanothertimeorinanothermouthbefalse’(100).Thisabilityofsentencestobetrueincertaincircumstancesbutfalseinotherswasthemotivatingfactorbehindthedevelopmentofanothertypeofaccountofmeaning,thatofferedby‘possibleworldsemantics’,towhichweturnnext.PossibleworldsWearebynowfamiliarwiththeideathatonepopularwayofexplaining‘truth’istosaythatapropositioncanbedescribedastrue,orasafact,ifitcorrespondswithreality.Wehaven’tbeentoomuchconcernedwiththenatureofthis‘reality’,buthaveseenthatitcanbeequatedwiththewaytheworld,ortheuniverse,actuallyis.10Nowofcourserealitymighthavebeenotherthanitactuallyis.Consider,forinstance,thewayinwhichweall,fromtimetotime,indulgeinimagininghowthingsmighthaveworkedoutifaparticulardecisionhadbeenmadedifferently,oraparticularturnofeventshadnotoccurred.Inotherwords,weareallcapableofimaginingdifferentpossibleversionsofreality,whichareoftencloselyrelatedtohowthingsactuallyare,butdifferinsomeparticularrespect.Onewaytodescribeourimaginedstateof‘whatmighthavebeen’istosaythatitisinfactreality,butrealityinanotherpossibleworld.Possibleworldsemanticsisbasedonthissimplefactthatthingsmightbeotherthantheyare,andonwhatfollowsfromthis:namelythat,indifferentversionsofreality,differentpropositionsmustbetrueandfalse.Theseideashave,asweshallsee,beendevelopedandworkedonbyanumberofphilosophersofthetwentiethcentury.ButtheoriginofpossibleworldsemanticscanbefoundintheworkofLeibniz.AswesawinChapter1,Leibnizwasfamiliarwithmanydisciplines,includingtheology,anditishistheologywhichinformshisaccountofpossibleworlds.ForLeibniz,differentpossibleworldsexistinthemindofGod;inhiswords,differentversionsofrealityalloriginate‘fromtheSupremeReason’(NewEssaysonHumanUnderstanding,227).Aversionofreality,orpossibleworld,consistsofeverythingwhichexistsandhasexisted.11Thenumberofpossibleworldsisinfinite,butonlyoneofthemisrealised,orcreated,andagainthisisdependentonGod:Andeventhoughoneshouldfillalltimesandallplaces,itstillremainstruethatonemighthavefilledthemininnumerableways,andthatthereisaninfinitudeofpossibleworldsamongwhichGodmustneedhavechosenthebest,sincehedoesnothingwithoutactinginaccordancewithsupremereason.(Theodicy,8,128)\n92TruthandrealityAccordingtoLeibniz,becauseGodisgood,theworldwhichheactuallycreatesis,necessarily,thebestpossibleworld.‘Best’inthiscontextcanbethoughtofassomethingsimilartothesimplest.Thebestpossibleworldisthatwhichcontainsthemaximumnumberofstatesofaffairsbuttheminimumofindividuallaws,orthemaximumofeffectsfortheminimumofcauses.Inmorerecentphilosophy,Leibniz’saccountofpossibleworldshasbeengivenasemantic,ratherthanatheologicalapplication.Ifthetruthofasentencedependsoncorrespondencetotheworld,itisargued,thenamultitudeofdifferentworldsmustmeanthatasentencecanbetrueinsomewhileitisfalseinothers.Soonewayofdescribingthepropositionexpressedbyasentencewouldbetospecifyalltheworldsinwhichitwastrue.Let’stakeTarski’sexamplesnowiswhite.Wecouldofferanaccountofthecircumstancesinwhichthisistruebylistingallthepossibleworldsinwhichitcorrespondswithreality,whilerulingoutallthoseworldsinwhichsnowis,forinstance,redorblueorblack.Thepropositionexpressedbyasentenceis,onthisaccount,describednotasaseriesoftruthconditionsbutasasetofthoseworlds,andonlythoseworlds,inwhichitistrue.Thenotionof‘possibleworlds’hasbeenusedbyanumberofphilosophersindifferentways.Someseethemasaconvenientfigurativeormetaphoricalwayofthinkingaboutcomplexissues.TheAmericanphilosopherDavidLewis,however,givesthemamoreconcretedefinition.Accordingtohisaccount,everypossibleworld,ourworldincluded,isdefinedbytheentitieswhichexistinthatworldandthefactswhicharetrueofthoseentities.Ourworldisgenerallydescribedinthiscontextas‘theactualworld’,butistobethoughtofasnodifferentinstatusfromalltheotherpossibleworlds.Otherworldsdifferfromit‘notinkindbutinwhatgoesoninthem’(85).Worldsotherthantheactualarenot‘imaginary’worlds;theyareworldsjustlikeours,butinwhichthingsaredifferent,toagreaterorlesserextent.So,forinstance,anotherpossibleworldmightbeexactlythesameasoursexceptthatinitsnowisblue.12Yetanotherworldmighthavebluesnowandcoldfirebutbelikeoursinallotherrespects.Thesamegoesforentities.Therearepossibleworldswhicharejustlikeours,butpopulatedbyaslightlydifferentselectionofpeople,orofdogs,orindeedofants,thanexistintheactualworld.Thiswayoflookingatthingscanrequirequitealeapofimagination.Wenormallythinkofitasaplainfact,assimplytrue,that,forinstance,snowiswhite.Lookingatthematterintermsofpossibleworlds,however,weneedtoconsiderthatitismerelypossiblethatsnowiswhite,andthatthisjusthappenstobesointheactualworld.Toputthisanotherway,theworldweinhabitisjustoneofthoseworldsinwhichsnowiswhite,andsoonformanyotherfactswetakeforgranted.Possibleworldsemantics,then,isconcernednotjustwithwhatis,butwithwhatispossible,and,aswewillsee,withwhatisimpossible.Possibilityandimpossibilityarewhatareknownasmodalconcepts,andbelongtoatypeofpropositionallogic\nTruthandreality93whichwehaven’tconsideredyet.Wewillneedtofindoutsomethingaboutmodality,therefore,beforewecanunderstandsomeoftheimplicationsofpossibleworldsemantics.ModalityThestudyofmodalitydatesbacktotheclassicallogicofancientGreece.Thefollowingpassage,fromAristotle’sdeInterpretione,givesadefinitionofwhatitisforapropositiontobepossible:Itisnecessaryfortheretobeornottobeasea-battletomorrow;butitisnotnecessaryforasea-battletotakeplacetomorrow,norforonenottotakeplace—thoughitisnecessaryforonetotakeplaceornottotakeplace.So,sincestatementsaretrueaccordingtohowtheactualthingsare,itisclearthatwhereverthesearesuchastoallowofcontrariesaschancehasit,thesamenecessarilyholdsforthecontradictoriesalso.(19a,30–5)Ifaproposition,suchasthepropositionthatthesea-fightwilltakeplacetomorrow,ispossible,itisneithernecessarilytruenornecessarilyfalse.Aristotleisalsoassertingthatifitcanbesaidthatsomethingispossiblythecase,forinstancethatthefightwillpossiblytakeplacetomorrow,itcanequallywellbesaidthatitispossiblynotthecase,thatthefightwillpossiblynottakeplacetomorrow.Fromthiswehavededucedthefollowingstatementsaboutmodality:35)Ifsomethingispossiblythecaseitisnotnecessarilynotthecase.36)Ifsomethingispossiblynotthecase,itisnotnecessarilythecase.Wecanthinkofseveralothersituations.So,forinstanceifitisnecessarythatthefightwillnottakeplace,weareineffectsayingthatitisnotpossiblethatitwilltakeplace.Similarly,ifitisnecessarythatitwilltakeplacetomorrow,itisnotpossiblethatitwillnottakeplace.Wecanaddthefollowingobservationstoourlist:37)Ifsomethingisnecessarilynotthecase,itisimpossiblethatitisthecase.38)Ifsomethingisnecessarilythecase,itisimpossiblethatitisnotthecase.Sofarwehavestuckwiththerathercumbersomeexpressions‘itispossiblethat…’and‘itisnecessarythat…’.Innaturallanguageweinfacthaveanumberofdifferentwaysofexpressingthesemodalities.Toexpress\n94Truthandrealitypossibility,forinstance,wehavealltheoptionsbelow,aswellasmanyothers.39)Itispossiblethatthefightwilltakeplacetomorrow.40)Thefightmaytakeplacetomorrow.41)Maybethefightwilltakeplacetomorrow.42)Thefightwillpossiblytakeplacetomorrow.Informallogic,thereisonlyonelogicaloperatortocoverallthesenaturallanguageexpressions.Thesymbolforpossibilityis
,orsometimesM.Ifwespecifythatpisthepropositionthatthefightwilltakeplacetomorrow,thenwecanrepresentallof39)–42)aboveas
p.Similarly,thesymbolfornecessityis,orL;pmeansthatitisnecessarythatthefightwilltakeplacetomorrow.Usinglogicalnotation,wecanrephraseourfourfindingsaboveasexpressionsoflogicasfollows:43)
p®~~p44)
~p®~p45)~p®~
p46)p®~
~pPossibleworldsemanticsoffersanotherwayoflookingatrelationshipssuchasthese.Infact,itallowsustomakethesamegeneralisationwithoutusingtheconceptsofnecessaryandpossible,concentratinginsteadonstatementsinvolvingsomeandall.Theseterms,aswesawinChapter2,aredealtwithinconventionalpredicatelogic,andcanthereforebedefinedintermsoftruth-values.Now,aswehaveseen,apossiblepropositioncanbeseenasonewhichistrueinsome,butnotall,possibleworlds.Itispossiblethatsnowiswhiteand,infact,snowactuallyiswhiteinourworld;theactualworldisamemberofthesetofworldsinwhichitistrue.Now,weknowfrom46)abovethatanecessarypropositionisonewhichisnotpossiblynottrue,sowecandefineanecessarypropositionasonewhichistrueinallpossibleworlds.Rememberthatthemostclear-cutofnecessarytruthsareanalyticsentencessuchas47)and48).Wecansaythattherearenologicallypossibleworldsinwhich47)and48)arenottrue.Thesetofworldsinwhichtheyaretrueisthesetofallpossibleworlds.47)Whiteliliesarelilies.48)Allthoroughbredracehorsesarehorses.Similarly,apropositionwhichisimpossible,suchasonewhichislogicallycontradictory,willnotbetrueinanyworld.Thesetofworldsinwhich,forinstance,49)isliterallytrue,isempty.\nTruthandreality9549)Thisisawhitelilyanditisnotalily.ThepotentialofpossibleworldsemanticsforexplainingmodallogicisamajorinterestofSaulKripke,theAmericanphilosopherwhoseworkonnameswestudiedinChapter1.Hesuggeststhefoundationsforsuchanaccountinanarticlepublishedin1959,whenhewasonlynineteenyearsold:‘Acompletenesstheoremformodallogic’.Thisisadifficultandverytechnicalarticle,butcontainsattheoutsetaclearstatementofKripke’sviewoftherelevanceofpossibleworlds,whichhealsoreferstoas‘conceivableworlds’.Heexplainsthat‘inmodallogic,wewishtoknownotonlyabouttherealworldbutaboutotherconceivableworlds;P[aproposition]maybetrueintherealworldbutfalseinsomeimaginableone’(2–3).Kripkeexplainsthattheevaluationofapropositionaseithertrueorfalserelatestooneparticularworld;ingeneralifwesaythatsomethingistruewemeanthatitholdsintheactualworld.Tospeakofnecessarytruths,however,wemustrefertotheentiresetofpossibleworlds:‘apropositionBisevaluatedastruewhenandonlywhenBholdsinallconceivableworlds’(3).Kripkesuggeststhatsuchanapproachassuresthat‘atleastacertainnon-trivialportionofthesemanticsofmodalityisavailabletoanextensionalistlogician’(3).Onceitistranslatedintopossibleworldsemantics,modalitycanbeaddressedbyalogicianconcernedonlywiththeextensionsofsentences—theirtruthorfalsity—andwithtruth-functionaloperators.Thisisbecausetreatingmodalityintermsofpossibleworldsemanticsmeansexplainingitasarelationshipbetweenthevaluesoftruthandfalsityandthesetofworlds.Wehaveestablishedthatpossibleworldsemanticsoffersusawayofdiscussingmodalitywithinthelimitsoftruth-functionallogic.Butsomemodalsentencesposefurtherproblemsforextensionallogic,andtothesealso,possibleworldsemanticsoffersasolution.TheseproblemsaresimilartothoseweconsideredinChapter1whenwelookedatopaquecontexts.Rememberthatopaque,orintensionalcontextsaresentencesinwhichsubstitutionofoneexpressionbyanotherwithwhichitisextensionallyequivalentdoesn’tnecessarilyresultinthetruth-valueofthewholeremainingunchanged.ItisgenerallyacceptedthattheauthorofGreatExpectationsandCharlesDickensareextensionallyequivalent,orrefertothesameindividual.But,aswesaw,substitutingoneexpressionfortheotherinapairsuchas50)and51)doesn’tnecessarilygiveustwoextensionallyequivalentsentences.It’squitepossible,forinstance,that50)mightbetruewhile51)isfalse.50)PipbelievesthattheauthorofGreatExpectationsisagenius.51)PipbelievesthatCharlesDickensisagenius.\n96TruthandrealityTheseparticularexamplesillustratetheproblematicopaquecontextsassociatedwithpropositionalattitudes,butthesametypeofproblemisraisedbymodalexpressions.Considerthepairofexamples52)and53)inwhich,again,oneexpressionissubstitutedforanotherwithwhichitisextensionallyequivalent.52)It’spossiblethatCharlesDickensisnottheauthorofGreatExpectations.53)It’spossiblethatCharlesDickensisnotCharlesDickens.Farfromthissubstitutionresultinginextensionallyequivalentsentences,itseemstobethecasethat52)and53)musthavedifferenttruth-values.Wecanprobablyagreethat52)istrue.It’spossiblethatCharlesDickensdidn’twritethenovelafterall;wecanimaginethesortofdocumentaryevidencewhichmightconvinceusthatsomeoneelsewroteit.Butwecertainlywouldn’twanttosaythat53)isalsotrue;indeedit’salogicalimpossibility.Itisn’tpossiblethatthepersonreferredtoasCharlesDickensisnotthepersonreferredtoasCharlesDickens.Anaccountoftheseopaquecontextsintermsofpossibleworldswouldgoasfollows.Indiscussinganopaquecontext,aswesawinChapter1,weneedtoconsidernottheextensions,ordenotationsofitsparts,buttheirintensions.Sowemustconsidernotthereferenceofthesubordinateclause,butitssense.TakingtheexampleoftheclauseCharlesDickensisnottheauthorofGreatExpectations,weneedtotakeaccountnotofitstruth-valueintheactualworld,butofthepropositionitexpresses.Now,apropositioncanbedescribedasasetofpossibleworlds,asthesetofallthepossibleworldsinwhichitistrue.Soweneedtotakeintoaccountthesetofpossibleworldsinwhichthesubordinateclauseistrue.Totakethemodalcontextsin52)and53)tobeginwith,wecansaythatthepropositionthatCharlesDickensistheauthorofGreatExpectationsistrueinacertainsetofpossibleworlds,asetwhichhappenstoincludetheactualworld.ThereisanothersetofpossibleworldsinwhichCharlesDickensisnottheauthorofGreatExpectations.Nowtosaythatapropositionistrueinonesetofworldsandfalseinanothersetissimplytosaythatitispossiblytrue.Wehaveestablishedthatthereisasetofpossibleworldsinwhich‘CharlesDickensisnottheauthorofGreatExpectations’istrue,hence52)istrue.However,thepropositionthat‘CharlesDickensisnotCharlesDickens’isnottrueinanypossibleworld.Thesetofpossibleworldsinwhichitistrueisempty.Example53),whichstatesthatthissetofpossibleworldsisnotempty,isthereforefalse.Asimilaraccountcaninfactbeappliedtotheexamplesinvolvingpropositionalattitudes.Theycanbethoughtofnotasarelationshipbetweenthesubjectandthereferenceoftheproposition,atruth-valueintheactualworld,butasarelationshipbetweenthesubjectandthesense\nTruthandreality97oftheproposition,asetofpossibleworlds.Thebeliefdescribedin50)isPip’sbeliefinaworldinwhichtheauthorofGreatExpectationsisagenius.ThisisnottosaythatPipnecessarilybelievesinaworldinwhichCharlesDickensisagenius,becauseaswehaveseen,CharlesDickensisnottheauthorofthenovelineverypossibleworldbutonlyinacertainsetofpossibleworlds,asetwhichhappenstoincludetheactualworld.CounterfactualsWecouldsummarisetheapplicationswehaveconsideredsofarbysayingthatpossibleworldsemanticsoffersawayofdescribing‘unreal’situations,orstatesofaffairsotherthanthoseoftheactualworldweinhabit.Wewouldn’tneedtotalkaboutpossibleworldsifweonlyeverwantedtomakestatementsoffactaboutreality.Butoncewestarttalkingaboutwhatcouldpossiblybethecase,oraboutwhatpeoplebelieveorimaginetobethecase,wearetalkingaboutthingsbeingotherthantheyactuallyare.Thereisafurthertypeofexpressionusedtodescribe‘unreal’situations.Rememberthatwhenwefirstencounteredthenotionofpossibleworlds,weconsideredthewayinwhichwesometimesimagine‘howthingsmighthavebeen’,or‘whatcouldhavehappened’.Suchthoughtsareoftenexpressedinthefollowingtypeofconstruction:54)IfonlyIhadn’tmissedthetrainIwouldhavebeenintimetomeethim.55)IfJohnwereherenowhewouldbemakinguslaugh.56)Ifweweremeanttoflywewouldhavebeenbornwithwings.Suchexamplesmightatfirstappeartobestraightforwardconditionalsofthefamiliarif…thentype.Butaproblemariseswhenwetrytomakedecisionsabouttheirtruth-values.Ineachcasewefindthatwecan’tcalculatethetruth-valueofthewholebydeterminingthetruth-valuesoftheantecedentandtheconsequent,aswehavedonebefore.Thisisbecauseneitherpartreferstoanactualsituation,andsoneithercanbetestedintermsofitscorrespondencetoreality.Examplessuchasthis,concernedastheyarewith‘unreal’situations,generallyoccurinthesubjunctivecase(ifJohnwerehereratherthanifJohnwashere)whichservestoindicatetheirdistancefromreality.Theuseofthesubjunctivegenerallyindicatesthattheantecedentis,orisbelievedtobe,false.Wewouldn’texpecttohearsomeonesay55),forinstance,inasituationinwhichJohnisactuallypresent.Nowrememberthatinthecaseofmaterialimplication,thelogicalequivalentofif…thenwhichweconsideredinChapter2,iftheantecedentisfalsethentheconditionalasawholemustbetrue;everythingfollowsfromafalseantecedent.Butweneedtomakeanexceptionforcaseslikethese.We\n98Truthandrealitymightbehappytoaccept54)–56)astrue,butifweallowedallsuchexamplestobetruejustbecausetheirantecedentsarefalse,wewouldalsohavetoacceptanexampleaspreposterousas57)astrue.57)Iftherewerenoclockswewouldallliveforever.Thewayinwhichthisconclusionhasbeenavoidedisbymakingaseparatecategoryforthese‘unreal’conditionals.Theyareknownascounterfactuals,orascounterfactualconditionals.Theyhaveplayedanimportantroleinthedevelopmentofpossibleworldsemantics,andarethesubjectofashortbookbythephilosopherDavidLewis,aprofessoratPrincetonUniversity,whosedefinitionsofpossibleworldsweconsideredearlier.InCounterfactuals,publishedin1973,Lewisarguesthat,‘althoughvague’,counterfactualconditionalscanbeassignedtruthconditions,andthatitisthereforenotnecessarytodismissthemallas‘unreal’andsimplylogicallytrue.Thisisbecausediscussionofwhat‘might’havehappenedisdiscussionofwhatdidinfacthappeninadifferentpossibleworld.Sotouseexample54)istosaythatthereisanotherpossibleworldinwhichIcaughtthetrainandIwastheretomeethim;itjustsohappensthatintheworldI’minImissedthetrainandwasn’ttheretomeethim.Lewisintroduceshisstudywiththeexample:58)Ifkangarooshadnotails,theywouldtoppleover.Heusesthesymbolwhichhasbecomeconventionalforcounterfactualconditionals,andwhichdistinguishesthemfromnormalconditionals,‘•®’.Soifpisthepropositionthatkangarooshavenotails,andqisthepropositionthatkangaroostoppleover,wecanrepresent58)as:59)p•®qLewissuggeststhatthemeaningof58)aboveis:‘inanypossiblestateofaffairsinwhichkangarooshavenotails,andwhichresemblesouractualstateofaffairsasmuchaskangarooshavingnotailspermitsitto,thekangaroostoppleover’(1).Aswehaveseen,tosaythatsomething‘mighthavebeen’thecaseistosaythatitactuallyisthecaseinsomeotherpossibleworld.So,Lewisissuggesting,wecandecideonthetruth-valueofacounterfactualsuchas58)bylookingatthepossibleworld,orsetofworlds,inwhichtheantecedentistrue:theworldsinwhichkangarooshavenotails.However,itisn’tjustanyoftheseworldswhicharerelevant.Weneedtoconsiderthoseworldswhichdifferenoughfromourworldtomaketheantecedenttrue,butwhichareinotherrespectssimilartoours.Thereisapossibleworldinwhichkangarooshavenotailsandtheyfallover.There\nTruthandreality99isanotherpossibleworldinwhichkangarooshavenotails,butthisiscompensatedforbythefactthatallanimateobjectsareheldupbybigstringshangingdownfromclouds,sothatkangaroosdon’tfallover.Thisisapossibleworld,butit’squitedifferentfromourworld,certainlymoredifferentthantheworldinwhichthekangaroosjustfallover.Wecansummarisethetruthconditionsforacounterfactualconditionalasfollows.60)p•®qistrueiffsomeworldwherepandqarethecaseismoresimilartoourworldthananycasewherepand~qarethecase.Substitutingthepropositionsaboutkangaroosforpandq,weseethat60)makes58)true;theworldinwhichtheyfalloverismoreliketheactualworldthantheworldinwhichtheydon’t.Italsooffersusawayofexplainingwhywewanttoreject57)above.Aworldinwhichtherewerenoclocksandeveryonelivedforeverwouldbelesslikeourworldthanaworldinwhichtherewerenoclocksbutpeoplewerestillmortal.Possibleworldsemantics,anditsapplicationstovarious‘problem’examples,getsmuchmorecomplicatedthanthebasicsofthetheorywhichwehaveconsideredhere.Forinstance,thenotionof‘similarity’betweenworldshasbeendiscussedinmuchmoredetailthantheloosedefinitionwehaveused,andcriteriahavebeensuggestedforcalculatingthe‘relativecloseness’ofdifferentpossibleworlds,andthedegreeof‘accessibility’whichholdsbetweenthem.Someintroductionstothiswork,andsomeoftheprimaryphilosophicaltexts,arementionedinthe‘furtherreading’sectionattheendofthischapter.Beforeweleavethetopicofpossibleworldsemantics,weshouldpausetotakeaccountofoneparticularapproachtothestudyoflanguagewhichdevelopedwithinthisframework:MontagueGrammar.MontagueGrammarMontagueGrammartakesitsnamefromthephilosopherRichardMontague.Hiscontributiontosemantics,bothofformalisedandofnaturallanguages,isasignificantone,althoughhediedattheearlyageofforty-onein1971,eightyearsafterbecomingProfessorofPhilosophyattheUniversityofCalifornia,LosAngeles.Hisaimwastoproducea‘UniversalGrammar’.Montagueusedthistermtorefertoasystemofsyntaxandsemanticscapableofdescribingbothformalisedlogicallanguageandnaturallanguage.Thatisnottosaythathewantedto‘purify’naturallanguageor‘makeitmorelogical’.Buthedidwanttoestablishanaccountoflanguagebasedonaformaltheoryconsistentwithlogic,ratherthanonintuitionandapproximation.Hesawthedescriptionofnaturallanguageasalegitimatefieldofstudy,butasonebelongingtomathematics.Inthissense,MontagueGrammarisbasedonidealsthatdatebackatleasttoLeibniz,andarealsofamiliartousfromtheworkofFregeandofRussell.\n100TruthandrealityAnexplicitstatementofthispositioncanbefoundinFormalPhilosophy,aselectionofMontague’spaperspublishedposthumouslyin1974.Montague’sownaccountofthisviewgoesasfollows:Thereisinmyopinionnoimportanttheoreticaldifferencebetweennaturallanguageandtheartificiallanguagesoflogicians;indeed,Iconsideritpossibletocomprehendthesyntaxandsemanticsofbothkindsoflanguageswithinasinglenaturalandmathematicallyprecisetheory.(222)However,headmitsthatitispossiblethatsuchasystemis,inpractice,unobtainable.MontagueGrammar,andthemodallogiconwhichitdepends,isnotoriouslydifficultforstudentsoflanguagewhodon’thaveagoodgraspofmathematics,chieflybecauseitisbasedonformalsettheory.Montague’slogicdrawsontherelationshipsbetweenentities,predicatesandbeliefswithinandacrosspossibleworlds.LikeDonaldDavidson,Montaguesawthechiefgoalofstudyinganylanguageasbeingtoprovideanaccountofthetruthconditionsforallthesentencesofthatlanguage.Inotherwords,hetoowasinterestedinproducinganaccountofmeaningbasednotjustontruth-values,butontheindividualpropositionsexpressedbythesentencesofalanguage.UnlikeDavidson,however,Montaguesawthetruthconditionswhichhisaccountwastoprovideasbeingnecessarilydefinedoverpossibleworlds.Heclaimedthatitmustbefeasibletoofferadefinitionoftheconditionsunderwhichanygivensentencewillbetrueinanypossibleworld.Initsvariousforms,then,possibleworldsemanticsoffersanexplanationofcertain‘problem’examples.Italsosuggestsamoregeneralapproachtothetaskofdefiningthemeaningofalanguageintermsofthetruthconditionsofitssentences.InitsincarnationintheworkofRichardMontague,atleast,ithasbeenappliedtonatural,aswellasformallanguages,indicatingthatnaturallanguagecanberecognisedwithinthisframeworkasalegitimate,andvalid,topicofstudy.Inthenextchapterwewillconsidersomeverydifferentapproachestothestudyofnaturallanguage.ThesedifferfromatheorysuchasMontague’sbothindistinguishingsharplybetweennaturalandformallanguagesand,crucially,inrejectingtheideathatthemeaningofnaturallanguagecanbedescribedbest,oratall,intermsoftruthconditions.Butfirstwewillconcludethischapterbylookingatthecontributionwhichtruth-basedaccountsofmeaninghavethemselvesmadetopresent-daylinguistics.Theirinfluencecanbemoststronglyfeltintheareaofformalsemantics,wherediscussionsofsentence-meaningoftenmakereferencetothenotionoftruth-conditionality.\nTruthandreality101LinguisticsemanticsThroughoutthischapter,wehavebeenconcernedwithvariousattemptstoexplainthemeaningsofsentencesintermsofthosefactorswhichservetomakethemtrueorfalse.Theseaccountshaveall,inonewayoranother,offeredexplanationsofhowlanguagerelatestotheworld,andattemptedtoexplaintruthintermsofarelationshiptoreality.Thissameinterestisapparentintheareaofpresent-daylinguisticswhichdealswithliteral-orsentence-meaning,namelysemantics.Thereis,however,quiteacleardistinctionbetweenphilosophicalandlinguisticsemantics.Theindividualworkswehavebeenconcernedwithinthischapter,disparateastheyareinotherways,mightallbesaidtobeconcernedwiththenatureofmeaningitself,andwiththequestionsofhowmeaningispossibleandhowitisbestdescribed.Theyaregenerallyconcernednotwithofferingatheoryofmeaningforanyparticularlanguage,butwithconsideringwhatsuchatheorywouldneedtobelike.ThistypeofapproachiswellexemplifiedbytheideasofCarnap,andinthetheoriesputforwardbyTarskiandlaterDavidson.Suchaccountsimplythat,inordertounderstandtheproblemssurroundingthenotionofmeaning,it’snecessarytoconsiderthebasicformwhichatheoryofmeaningforanyparticularlanguagewouldtake.It’snotthatthisisnecessarilythoughttobeachievableinpractice,butthattheprocessissaidtoilluminatetheproblems,andidentifytheessentialconceptsinvolved.AsthephilosopherMichaelDummettsuggestsatthestartofhisessay‘Whatisatheoryofmeaning?(I)’inhisbookTheSeasofLanguage,publishedin1993:Accordingtoonewell-knownview,thebestmethodofformulatingthephilosophicalproblemssurroundingtheconceptofmeaningandrelatednotionsisbyaskingwhatformshouldbetakenbywhatiscalled‘atheoryofmeaning’foranyoneentirelanguage;thatis,adetailedspecificationofthemeaningsofallthewordsandsentence-formingoperationsofthelanguage,yieldingaspecificationofthemeaningofeveryexpressionandsentenceofthelanguage.Itisnotthattheconstructionofatheoryofmeaning,inthissense,foranyonelanguageisviewedasapracticalproject;butitisthoughtthat,whenoncewecanenunciatethegeneralprinciplesinaccordancewithwhichsuchaconstructionshouldbecarriedout,weshallhavearrivedatasolutionoftheproblemsconcerningmeaningbywhichphilosophersareperplexed.(1)Linguisticsemantics,ontheotherhand,mightbedescribedasbeingconcernedwiththepracticalconstructionofaccountsofmeaningforparticularlanguages.Itsaimisnotsomuchtoshedlightonthetypesof\n102Truthandrealityproblemsinvolvedindefiningmeaning,astoengagewiththoseproblems.Itsmethodstendtobemoreempiricalthanthoseofsomephilosophicalapproaches,inthattherawdataofsemanticaccountsinlinguisticsgenerallycomefromthefactsofthelanguageinquestion.Inthetraditionofgenerativegrammar,atleast,semanticsislargelyconcernedwiththerelationshipbetweenmeaningandstructure.Semanticanalysesinthistraditionareoftenusedtodemonstratehowsentencesofalanguagerelatetoeachother.Topicsrelevanttothisarediscussedindetailinsomeoftheintroductoryworksonsemanticslistedinthe‘furtherreading’sectionattheendofthischapter.Onesuchtopicistherelationshipbetweentwosentenceswhich‘meanthesame’,arelationshipwhichissimilartothatofsynonymity,whichweconsideredinChapter1inrelationtoword-meaning.Thefollowingpairofexamplesmightbesaidtobeparaphrasesofeachother:61)BertrandRussellwroteHumanKnowledge.62)HumanKnowledgewaswrittenbyBertrandRussell.Wemightsaythat61)and62)havethesamemeaningbecausetheydescribethesameeventorstateofaffairs.Thesamesetoffactorswouldmakethembothtrueorbothfalse;inourworldwewouldwanttolabelthemboth‘true’.Generativegrammarianswouldclaimthattheyareinfactboththesameatthelevelof‘deepstructure’,thelevelatwhichmeaningisdetermined.Theywouldbeabletoexplaintoushowthepassivesentencein62)wasderivedfromthesamedeepstructureasitsactivecounterpartin61).Tosaythatthesetwosentencesmeanthesamebecausetheyarebothtrueinexactlythesamesituationsis,ofcourse,tosaythattheyhavethesametruthconditions.Thesamecouldbesaidforthefollowingpair:63)JohnStuartMillwroteASystemofLogic.64)ItwasJohnStuartMillwhowroteASystemofLogic.Inthiscasealso,thetwodiffernotintermsoftheirtruthconditions,butintermsoftheir‘surfacestructure’.Example64)containswhatisknownasacleftconstruction,aconstructionofthegeneralform‘itwasxwho…’,while63)isanon-cleft.Again,agenerativegrammariancouldexplainhow63)and64)arebothproducedfromthesamedeepstructureusingdifferentgrammaticalrules.Buttosaythatthetwopairsofexampleswehavejustconsideredaretruth-conditionallyequivalentisnot,ofcourse,totellthewholestory.Wemightnotbeabletodistinguishbetween61)and62)intermsofthesituationtheydescribe,butthatsamesituationcertainlyseemstobepresentedindifferentwaysinthetwoexamples.Similarly,thecleftandthenon-cleftsentencesmaybeidenticalintermsoftheconditionswhichmakethemtrue,butthey\nTruthandreality103wouldseemtobeappropriatetoratherdifferentcontexts.Forinstance,while63)presentsafactaboutthephilosopherinquestion,64)wouldseemappropriateonlyiftheauthorshipofASystemofLogicwereasubjectofdisagreement,oratleastunderdiscussion.Wehaveseensomeoftheproblemswhichanentirelyextensionalaccountofmeaning,dependentonassigningthepropertiesof‘true’and‘false’toeachsentence,encounters.TheseweretheproblemswhichDavidsonwasconfrontinginsupplementinganextensionalaccountwithanaxiomaticsystem,sothatsentencescouldbedefinedascorrespondingtoparticularstatesofaffairs,andnotjusttotruth-values.Buttheaboveexamplesseemtosuggestthatanyaccountwhichconsidersmeaningintermsoftruthconditions,orofthepropositionexpressedbyasentence,willalsobemissingsomesubtlebutimportantaspectsofmeaning.Onesolutionistosupplementatruth-conditionalaccountofmeaningwithapresuppositionalone.WesawattheendofChapter2thatoneaccountofpresuppositiondefinesitintermsofsentencesemantics.Suchanaccountwouldexplainthatalthough63)and64)aretruth-conditionallyequivalent,theydifferintermsofwhatispresupposedandwhatasserted.Principally,itispresupposedin64)thatsomeonewroteASystemofLogic,whatis‘atissue’iswhothiswas.Anotheraccountofthoseaspectsofmeaningwhichcan’tbeexplainedintermsoftruthconditionsistosaythat,sincetheyseemtorelatetocontextualconsiderationsof‘appropriateness’,theymustbeexplainedaspragmaticinnature.Therelationship,andalsotheappropriatedivisions,betweensemanticsandpragmatics,havebeenandcontinuetobethesubjectofmuchdebate.Averygeneraldefinitionmightspecifythatsemanticsisconcernedwiththestudyofmeaninginisolation,andpragmaticswithmeaningincontext.Anotherwayoflookingatthis,aswesawinconnectionwithdiscussionsofpresuppositioninthelastchapter,istocontrastlinguisticmeaning,thefocusofsemantics,withspeakermeaning,thefocusofpragmatics.Thedevelopmentofpragmaticsasaseparatedisciplinewithinlinguistics,anditsoriginsintwentieth-centuryphilosophy,arethesubjectsofthenextchapter.FurtherreadingTruthconditionsKant’sideas,andthedistinctionbetweenaprioriandsyntheticknowledge,arediscussedinAyer(1971)Language,TruthandLogic,particularlyinChapter4.AbriefbutclearaccountofKantandhisviewsonaprioriknowledgeisofferedinChapter8ofBertrandRussell’s(1980)TheProblemsofPhilosophy.Russelloffersmorediscussionthanwehavehadhereofthenotionandnatureof‘thingsinthemselves’,andengagesinacriticismofKant’ssolution,andoftheconnectionhedrawsbetweentruthandthought.\n104TruthandrealityVerificationBothofthebooksbyAyermentionedinthissectionaresuccinct,clearandveryreadable.Theywerewrittenwithinterested‘lay’peopleaswellasprofessionalphilosophersinmind.Inparticular,Ayer’sownintroductiontothe1946editionofLanguageTruthandLogic,andChapter2ofTheCentralQuestionsofPhilosophy(1973)dealwiththeformulationandproblemsoftheprincipleofverification.Verification,andtheworkoflogicalpositivismingeneral,isconsideredinDevittandSterelny(1987)LanguageandReality,Chapter11.TruththeoriesTarski’struththeoryofmeaningissetoutinmostofitsdetailinhis(1944)articleThesemanticconceptionoftruthandthefoundationsofsemantics’.Davidson’sversionissummarisedinhis(1967)article‘Truthandmeaning’;furtherdiscussionoftheimplicationsofaTarski-styleaccount,andDavidson’sdiscussionofnon-indicatives,canbefoundinhis(1979)‘Moodsandperformatives’.ThetruththeoriesofmeaningofTarskiandDavidsonarediscussedinR.Martin(1987)TheMeaningofLanguage,Chapter22.MartinrelatesDavidson’saccounttodiscussionsoflanguageandthemind,andof‘mentalese’,whichwewillconsiderinChapter5.Evnine(1991)DonaldDavidsonisabook-lengthstudyoftheworkofDavidson,whichpresentshisaccountoftruthandmeaninginthecontextofhisphilosophyasawhole.Chapter5inparticularisrelevanttothoseaspectsofDavidson’sworkdiscussedhere.Evnineconsiderstheproblemsoftryingtoapplyatheorydevelopedforaformallanguagetonaturallanguage.HealsooffersanaccountofTarski’stheory.PossibleworldsTheconnectionbetweenmodalityandpossibleworldsisdiscussedbyR.Martin(1987)TheMeaningofLanguage,Chapter15,whorelatesmodalityandextensionality.Stainton(1996)PhilosophicalPerspectivesonLanguageconsiderspossibleworldsemanticswithparticularreferencetoopaquemodalcontexts,inChapter4.McCawley(1981)EverythingthatLinguistsHaveAlwaysWantedtoKnowaboutLogicdiscussesmodallogicinChapter10,lookinginparticularatdifferenttypesofnecessity.InChapter11hegoesontoconsidertheapplicationsofpossibleworlds,lookinginparticularatLewis’accountofcounterfactualsandnotionsofaccessibilityandtherelativeclosenessofworlds.Bothchaptersareheavygoing,buttheearlysectionsofeacharequiteaccessible.McCawleyalsogivesadetailedtreatmentofMontagueGrammar,whichisnotdiscussedhereinanydetail,inChapter13.Cann\nTruthandreality105(1993)FormalSemanticsiswrittenwithintheframeworkofMontagueGrammar,andoffersagoodintroductiontoit.LinguisticsemanticsUsefulintroductionstolinguisticsemanticsincludeLeech(1981)Semantics,andHurfordandHeasley(1983)Semantics:ACourseBook.Thisincludesanintroductiontologic,anddetaileddiscussionofword-meaning,topicsintroducedhereinChapters2and1respectively.AlsousefulisHofmann(1993)RealmsofMeaning.Kempson(1977)SemanticTheoryisamoredifficultread,butitincludesdiscussionofsomeconceptsfromphilosophicalsemantics,suchasTarski’sconceptionoftruth.Shealsolooksatsometopicswhichbelongmoreobviouslyinlinguisticsemantics,suchastherelationshipbetweensyntaxandsemantics,andtheroleofdeepstructure.Frawley(1992)LinguisticSemanticsalsooffersanintroductiontomanyofthetopicsconsideredinthischapter.Inhisintroductionheoutlineswhatheseesassomeofthefundamentaldifferencesbetweenphilosophicalandlinguisticsemantics.Frawleyconcentratesonmethodinexplainingthisdifference;heseesphilosophyasfundamentallydeductive,linguisticsasinductiveandempirical.Lyons(1995)LinguisticSemanticsisacomprehensiveintroductiontothesubject.HisearlierSemantics(1977)intwovolumes,isamoredetailedtreatment,whichincludesdiscussionoftherelationshipbetweenformalandlinguisticsemantics,andoftherelationshipbetweensemanticsandgenerativegrammar.Katz(1972)SemanticTheorystillranksasoneofthemajorattemptstocombinesemanticsandgenerativegrammarintoacoherentdescriptionofnaturallanguage.\n4SpeakersandhearersIntroductionInourstudyoftruth-conditionalaccountsofmeaninginthelastchapter,weencounteredwhatcouldbesummarisedastwodifferentcategoriesofproblem.Thefirstcategoryincludesallthoseproblemswhichareraisedbyindividualtheoriesofthiskind,andwhichcanbestberesolvedbymodification,ordevelopmentofthosetheories.Davidson’saxiomaticaccountofmeaning,forinstance,wasdesignedtoaddresstheproblemsthatTarski’struththeoryencountered:namelythatitcouldneitherdistinguishbetweendifferent‘true’sentences,norgiveafulldescriptionofaninfinitelanguage.Thesecondcategoryofproblemsfortruth-conditionalaccounts,however,can’tberesolvedbyanyamountof‘repairwork’totheaccountsthemselves.Thesearetheproblemsraisedbyevidencethatmeaningisdeterminedasmuchbyhowlanguageisusedasbysemanticcontent.Inresponsetoproblemsofthistype,variousdifferentapproacheshavebeentakentothetaskofdescribingandexplainingmeaninginphilosophyand,latterly,inlinguistics.Theemphasisonlanguageinuseindicatesthatthephilosophersandlinguistsengagedintheseapproacheshavebeenconcernedwithlanguageasaneverydaymeansofcommunicationand,therefore,withnaturallanguage.ManyofthephilosopherswhoseworkweconsideredinChapter3wereinterestedinlanguageasaformalsystem,asatypeoflogic,orasamodeofexpressionforscientificdiscovery.Forthephilosopherswhoseworkwewillbeconsideringinthischapter,however,languageisasystemwhich,primarily,isusedbyhumanbeingstointeractwitheachother.Thecanonicaluseoflanguageinsuchaccountsisonewhereaspeakerproducesanutterance,inaparticularcontext,whichisinterpretedbyahearer.Noneofthesehighlightedtermswouldhaveaplaceinatruth-conditionalaccountofmeaning.Theyemphasisetheimportanceofawholevarietyoffactorswhichmightbelabelledextra-linguistic.Thatistosaythatmeaningisdeterminednotjustbytheruleswhichexplainalanguage,andrelateittotheworld,butalsobyawholevarietyoffactorswhichcouldn’tbedescribedasapartofthelanguagesystem.106\nSpeakersandhearers107Asourfirstexampleofthistypeofaccountofmeaning,wewillstartthischapterbylookingattheworkofLudwigWittgenstein.Wittgensteinisoftenseenasthefounderofthephilosophicalinterestinlanguageusewhichbeganaroundthemiddleofthetwentiethcentury.Hisphilosophicaldevelopmentisaninterestingonefromourcurrentperspective.Inhisearlywork,hewasconcernedwiththeformalpropertiesoftruth-conditionallogic.Laterinhiscareerhemovedawayfromthisapproach,concentratinginsteadonthevarietyoffunctionswhichlanguagecanbeusedtoperform,andcoiningthephrase‘meaningisuse’.ThissloganmightremindusoftheaccountofpresuppositionproposedbyPeterStrawson,whichweconsideredinChapter2.Inhisarticle‘Onreferring’,Strawsonemphasisedtheimportanceofuseandcontext,suggestingthatitisindividualspeakerswhodothingslikereferringandmentioning,notthelanguageitself.Thisapproachtothediscussionofmeaningingeneral,andreferringexpressionsinparticular,broughthimintoconflictwithBertrandRussell.Thetwophilosophersdisagreedaboutwhethertheuseofeverydayspeech,andtheattitudesandintentionsofspeakers,arerelevantfocusesofphilosophicalanalysis.Strawson’spositioninthisdisagreement,aswesaw,wasrepresentativeofagroupofphilosophersworkingatOxfordinthemiddleofthetwentiethcentury,engagedinthestudyofwhatbecameknownasordinarylanguagephilosophy.Ordinarylanguagephilosophyissometimesalsoreferredtoaslinguisticphilosophy,althoughwewillusetheformer,morewidespreadterm.Itsmostprominentproponents,apartfromStrawsonhimself,includedJ.L.Austin,JohnSearleandPaulGrice,anditistheirworkwhichwillbethemainsubjectofthischapter.Theterm‘ordinarylanguagephilosophy’isnotintendedtoimply,asissometimessupposed,thattheywereinterestedinstudyingvernacularorcasual,asopposedtostandardorformal,usesoflanguage.Thisisamuchmorerecentdevelopmentwithinlinguistics.Rather,itreferstotheirinterestinapproachingphilosophicalproblemsbymeansofananalysisoftheordinaryusesofthelanguageinwhichtheyareexpressed.Thisinturnledtoaninterestinnaturallanguageasalegitimatefieldofstudyinitsownright.Theyrejectedtheideathatlanguagecould,orshould,beexplainedsolelyintermsoflogicalrelations.Inparticular,theyrejectedtheideathataccountingformeaningwassimplyamatterofdeterminingconditionsfortruth.Thefollowingtypesofexamplehavebeenusedasevidencethatanytruth-conditionalaccountofmeaningmustbeinadequatefordescribingactualusage:1)Itisrathercoldinhere.2)Ibetyouthatitwillsnowbeforetheendoftheday.3)Idowishyouwouldshutthewindow.\n108SpeakersandhearersThesecouldallbedescribedinformallinguistictermsasdeclarativesentences.Itwouldthereforebepossibletodeterminethecircumstanceswhichwouldmakeeachofthemtrue,andtopresentthesecircumstances,intheformofaseriesoftruthconditions,asthemeaningofeachsentence.Thismightseemtoworkwellenoughforexample1);ifweknowtheconditionswhichwouldmake1)truewemightbeabletoclaimthatweknowthemeaningof1).1Butitdoesn’tworkaswellfortheothertwoexamples.Wemightbeabletodescribethestateofaffairswhichwouldmake2)strictly‘true’,buttoofferthisasanaccountofthemeaningof2)wouldsomehowbeto‘missthepoint’.Suchanaccountwouldmissthefactthat2)wouldmostlikelybeusednottodescribethisstateofaffairsbuttodosomethingelseentirely.Itmightbeusedactuallytomakethebet.Similarly,knowingwhatwouldmake3)true,therelevantstateof‘wishing’whichwouldhavetobeinplace,doesn’texplainhowinmanysituations3)wouldbeunderstoodasarequest,orperhapsanorder,toshutthewindow.Infact,incertaincircumstances,wecanimagine1)alsobeingunderstoodasarequesttoshutthewindow,suggestingthatatruth-conditionalaccountofmeaningmaynotalwaysbeappropriateevenforsuchanapparentlystraightforwardexample.Theassumptionthattheprimaryuseoflanguageistoproducestatementsoffacts,ordescriptionsofreality,issometimesknownasthedescriptivefallacy.ThistermwasfirstusedbyJ.L.Austin,whosuggestedthat,insteadofconcentratingonlanguageasameansofdescription,itwouldbemoreusefultodescribelanguageasameansof‘doingthings’.Laterinthischapterwewilllookathisanalysisofexamplessuchas1)–3)asbeingdifferenttypesofactions,orspeechacts,whichspeakerscancarryout.However,labellingdifferentutterancesasdifferenttypesofactiondoesn’tsolvealltheproblemsraisedbytruth-conditionalaccountsofmeaning.Evenrestrictingourselvestoexamplesinwhichsomedescriptionofrealityismade,wecanthinkofmanyinstancesinwhichsimplyknowingthetruthconditionsdoesn’tseemadequatetothetaskofexplainingmeaning.Imaginethatwehavearrangedtomeetafriendatacertaintimeand,asisourcustom,weshowupjustasexpected.Ourfriendcommentswithapproval:4)Icanalwaysrelyonyoutobeontime.Wecanexplainthissituationquiteadequatelyintermsoftruth.Wecanmakesenseof4),weknowwhatcircumstanceswouldmakeittrue,andwecanconfirmthatthesecircumstancesaremet.Butnowimagineadifferentscenario.Wemakethesamearrangementwithourfriendthefollowingweekbut,becauseofaseriesofmishaps,arrivehalfanhourlate.Ourfriendwho,asluckwouldhaveit,hashadtowaitinthepouringrain,greetsuswithexactlythesameremark,althoughdeliveredinaratherdifferenttoneofvoice.Atruththeoryofmeaningisn’tmuchhelphere.In\nSpeakersandhearers109fact,allitcandoistellusthatwhatourfriendhassaidisfalse.Butwecan’thelpthinkingthatthissomehowisn’tthe‘point’,andthatifweweretorespondwithacheerful‘No,youcan’t,I’mhalfanhourlateactually’,wewouldonlymakethingsworse.Thepoint,ofcourse,isthatalthoughourfriendhas,literally,saidthatweareontime,thatisn’tquitewhatshemeant.Rather,shewasbeingsarcastic,andusingapatentlyfalsestatementtoconveyherdispleasure.Sarcasmoffersonetypeofexampleinwhichwhatpeople‘say’andwhatthey‘mean’canbetwoverydifferentmatters.Buttherearemanyother,moresubtleexamplestobefoundinwhichasimilarprocessseemstobetakingplace.Theseareoftenmostapparentwhenwelookatanutteranceinthecontextoftheconversationinwhichitoccurs,forinstancebyconsideringitastheresponsetoaparticularstatement.Imaginethefollowingexchangetakingplaceinacrowdedroom:5)AMrsXisanoldbag.BTheweatherhasbeenquitedelightfulthissummer,hasn’tit?TheresponsewhichBmakesisapparentlyabouttheweather,butinthiscontextwe’dprobablywanttosaythatshemeanssomethingentirelydifferent.Withoutfurtherinformation,inparticularwithoutfurtherdetailsofthecontext,wearen’treallyinapositiontobesureexactlywhatshemeant,butherutterancemightwellbebesttakenassomesortofhintthatitwouldbewisetoselectanothertopicofconversationhastily.Perhaps,unlikeA,BisawarethataparticularfriendofMrsXiswithinearshot.Weknowthatthereisprobablyadiscrepancybetweentheliteralandtheimpliedmeaninginthisexample,notbecauseofanyaspectofB’sresponseinitself,orbecauseofanylinguisticrule,butbecauseofthecontextinwhichitoccurs.Inotherwords,weknowthatBprobablydoesn’tintendherutterancetobeinterpretedliterallypreciselybecausesheisofferingitasaresponsetoA’sunfortunateremark.ThisexamplewasoriginallysuggestedbyanotherOxfordphilosopherofordinarylanguage,PaulGrice,toillustratewhathesawasanimportantaspectofmeaninginuse:thedistinctionbetweenwhatpeople‘say’andwhatthey‘implicate’.Heexplainedthisexampleasinvolvingatypeofconversationalimplicature,anelementofutterancemeaningwhichcanoftenbeverydifferentfromtheliteralmeaningofthesentenceuttered.Wewillbelookingatsomemoreexamplesofconversationalimplicaturelaterinthischapter,andconsideringthewaysinwhichGriceexplainsthemwithreferencetohisco-operativeprincipleofconversationalinteraction.ThisbriefintroductiontoGrice’sworkhasraisedtwoissueswhichmakeitparticularlyrelevanttopragmatics,thebranchoflinguisticswithwhichwewillconcludethischapter.First,Grice’saccountisaconversationalone.Hisexampleispresentedwithashortconversationalcontext,andtheactofinterpretinganutterancesuchasB’sin5)aboveisonewhichmustbe\n110Speakersandhearersperformedbyaconversationalpartner.Thisfocusrelatestothesecondpartofthetitleforthischapter.Languageuseinvolvesnotjustspeakerswhoproduceutterances,butalsohearers,whoarethemselvespartoftheconversationalcontext.Forlanguagetooperateeffectivelyasameansofcommunication,theroleofthehearerisascentralasthatofthespeaker.Aswewillsee,thehearer’staskofinterpretationisanimportantfocusinpresent-daypragmatics.Second,Grice’sdistinctionbetweenliteralandimplicatedmeaningdrawsattentiontothepossibilitythattherearedifferenttypes,orlevels,ofmeaning,apossibilitywhichhasbeenmuchdiscussed,andalsohotlydisputed,inpresent-daylinguistics.Itsuggeststhatwhileaccountsofmeaningbasedonlinguisticanalysisandonsemanticrulesmaybeabletotellusallweneedtoknowaboutsentence-meaning,themeaningofanyparticularutteranceofasentenceincontextwilldependonawholehostofother,non-linguisticfactors.Semanticscantellus,atbest,onlypartofwhatweneedtoknowaboututterance-meaning.Strawsonsumsupanimportantattitudeofordinarylanguagephilosophywhenheclaims,in‘Onreferring’,thatThecontextofutteranceisofanimportancewhichitisalmostimpossibletoexaggerate’(230,emphasisadded).Itisimportanttobearinmind,then,thatnotalltheaccountswewillbeconsideringinthischapterneedbeseenasalternativestosemanticaccounts,offeredtoreplaceinadequate,truth-conditionaltheoriesofmeaning.Rather,someofthemarebestseenascomplementstosuchaccounts.Theysuggestthatthemeaningspecifiedbylinguisticruleisonlypartofthestory,andthatsuchrulesneedtobesupplementedbyaconsiderationofother,non-linguisticfactors.ButwewillbeginwithaconsiderationoftheworkofLudwigWittgensteinwho,asweshallsee,cametobelievethatmeaningcouldbest,andperhapsonly,beexplainedintermsofuse.Wittgenstein’sphilosophyoflanguageLudwigWittgenstein’swritingstouchonmanyofthetopicswehavebeenconcernedwith,buthisworkwasbynomeansconfinedtothephilosophyoflanguage.Healsowroteinfluentiallyonthephilosophyofmathematics,ofthemind,andonthenatureofphilosophicalinvestigationitself.In1922,theEnglisheditionofTractatusLogico-Philosophicus(generallyreferredtobyitsabbreviatedtitle,Tractatus)waspublished,inwhichhesetouthisviewsonlogicandlanguage.Itwasaremarkablefirstbookintermsofitscomplexityanditssubsequentimportance,allthemoresointhatitwaswrittenwhileWittgensteinfoughtinthetrenchesintheFirstWorldWarintheAustrianarmy,andsubsequentlywhilehewasdetainedinaprisonerofwarcamp.Evensettingasidehistimeinthearmy,Wittgenstein’scareerwasanunusualandeclecticone.HewasborninViennain1889,andwhenhe\nSpeakersandhearers111movedtoEnglandin1908hefirststudiedengineeringatManchester,beforemovingtoCambridgeandtakingupphilosophy.Then,afterpublishingTractatus,WittgensteinabruptlyleftCambridgeandturnedhisattentiontovariousotheroccupations,includingtimeasaschoolteacherandasagardener.Thismoveissometimesinterpretedasagrandiosestatementtotheeffectthathehadsaidallhehadtosayonthesubject.ButafterseveralyearsWittgensteindidreturntophilosophy,andtoCambridge,wherehecontinuedworkonthethemeshehadbeguninTractatus,althoughhislaterworkdifferedinmanyrespectsfromhisearlier.Hebecameaprofessorofphilosophyin1939,butimmediatelylefttoserveasamedicalorderlyintheSecondWorldWar,whichmeantthatineffecthetookupthepostin1945,justtwoyearsbeforeheretiredandsixyearsbeforehisdeath.TractatuswastheonlybookwhichWittgensteinpublishedduringhislifetime.Hisotherworks,allpublishedposthumously,includetheaccountofhislaterphilosophyknownasPhilosophicalInvestigationspublishedin1953,andvariouscollectednotesandlectures.Thestyleofmuchofhiswritingissometimesdescribedasenigmaticor‘aphoristic’,inthathewritesinaseriesofstatementsorassertions,oftenleavingthereadertofillintheconnectionsbetweenthem.InTractatus,henumberedthesestatements,startingeachonanewline.Aswithclassicalwriters,itisthecustomtorefertotheworkbymeansofthesenumbers,ratherthanbypagenumberinanyspecificedition.Wittgenstein’soriginalphilosophicalinterest,growingoutofhisstudiesinengineering,wasinthephilosophyofmathematics.HisearlyinfluencesincludeBertrandRussell,whosepupilhewasduringhisbriefstudiesatCambridgebeforetheFirstWorldWar.Inturn,RussellwashimselfinfluencedbyTractatus,andparticularlybytheclaimthatofferinganaccountofthoughtinvolvesofferinganaccountoflanguage,aclaimwhichbecamecrucialinthesubsequentdevelopmentofanalyticphilosophy.Wittgenstein’searlyworkalsohadanimpactontheViennaCircle,whoreadTractatusatseveraloftheirmeetings,someofwhichWittgensteinhimselfisknowntohaveattended.Again,hisemphasisonthecentralityoflanguagewasperhapsthemostinfluentialpartofhiswork.Wehavealreadyseenthat,inthelogicalpositivismoftheViennaCircle,discussionofproblems,or‘pseudo-problems’,ofphilosophyoftenfocusonthelanguageinwhichtheyareexpressed.Wittgenstein’slaterworkwasnolessimportant.Itcanbeseenastheprecursorofthegrowinginterest,inthemiddleofthetwentiethcentury,instudyinghowlanguageisactuallyused,ratherthanfocusingonspecificlogicalstructures,oridealised‘perfect’languages.Tractatusstartswiththecharacteristicallyboldandchallengingstatementthat‘Theworldisthetotalityoffacts,notofthings’(1.1).Theworldinwhichweliveischaracterisednot,oratleastnotsolely,bytheobjectswhichitcontains;itischaracterisedbythewaysinwhichthese\n112Speakersandhearersobjectsareconstitutedandrelated.Thefacts,ofcourse,determinewhattruestatementscanbemadeabouttheworld.WecanseefromthisthatWittgensteinbeginsessentiallywithacorrespondenceaccountoftruthandmeaning;thefactsoftheworlddeterminewhatistrue.Heexplicitlystatesthat:Tounderstandapropositionmeanstoknowwhatisthecaseifitistrue’(4.024).ForWittgenstein,apropositionisthepresentationofathought.Hethereforeequateshistaskinstudyinglanguagewiththatofstudyingthought;languageisessentiallytheexpressionofthought.Thethoughtswhichareexpressedbythepropositionsof‘factuallanguage’,languageusedtoconveyinformation,hedescribesas‘pictures’.Nowofcourseonlythosepictureswhichrepresent,orcorrespondwith,reality,canbedescribedas‘true’,butallpicturesatleastrepresentapossiblestateofaffairs.HerewecanseeaconnectionbetweenWittgenstein’saccountoflanguageandtheideaof‘possibleworlds’,whichwasdevelopedlaterinthecentury.Indeed,ithasbeensuggestedthatpossibleworldsemanticscanbeseenasageneralisationofWittgenstein’sframework,inwhichthespecificnotionof‘possibleworlds’isaddedtohisnotionsofthingsandtruth-values.2Wittgenstein’sowntreatmentofpossiblestatesofaffairsisaveryspecificone.Forhim,allpossiblestatesofaffairsarecontainedwithintheobjectsoftheactualworld.Possiblestatesmaydiffertovaryingdegreesfromourown,butallhaveacommonformdeterminedbytheobjects.Withinthisconstriction,wecanthinkofanumberofdifferentstatesofaffairs,and‘Whatisthinkableisalsopossible’(3.02).Whatisthinkablecanbeexpressedinaproposition.The‘perceptiblesign’ofthatproposition,inotherwordsaspokenorwrittensentence,is‘aprojectionofthepossiblestateofaffairs’.Hisaccount,then,isatruth-conditionalone,inwhichmeaningisdefinedintermsofthesituationwhichwouldmakeaproposition,andthereforethesentencewhichexpressesit,true.Butitisanaccountconcernedwithunderstandingtheregularitiesofnaturallanguage,ratherthanwithconstructinganidealone.Theprocessoflogicalanalysisisoneofdiscoveringthelogicalformofasentence,whatevergrammaticalformitmaytake.OnewayoflookingatthedifferencebetweenWittgenstein’sideasinTractatusandthoseinhislaterwork,particularlyinPhilosophicalInvestigations,isintermsofhisaccountoflanguage,orratherhisviewofwhatanaccountoflanguageentails.Inparticular,inthelaterworkhedispenseswiththeideathatspeechorwritingconsistsprimarilyinproducing‘perceptiblesigns’ofpropositions.Tractatusoffersauniformaccountofthenatureoflanguage,definedintermsofpropositions,anditsroleinexpressingthought.Inhislaterwork,Wittgensteinexplicitlyrejectsthepossibilityofgivingageneralaccountofpropositions,orindeedoflanguage.Insteadofbeingaunifiedphenomenon,languageisseenasacollectionofactivitieswhicharealldifferent.Wittgensteinfamouslyreferstotheseactivitiesas‘languagegames’;justastherearemanydifferenttypes\nSpeakersandhearers113ofgame,allofwhichcanbecharacterisedas‘games’butallofwhicharedifferent,sotherearemanydifferenttypesofuseoflanguage,allofwhichcanequallyvalidlybedescribedas‘language’.Wittgensteinexplainsthat:‘Heretheterm“language-game”ismeanttobringintoprominencethefactthatthespeakingoflanguageispartofanactivity,orofaformoflife’(23,originalemphasis).3Thereare‘countless’differentlanguagegames,becausetherearecountless,andeverchanging,waysinwhichpeopleuselanguage.Wittgensteinsuggestsjustafewofthesegames,whichinclude,forinstance,‘givingorders’,‘reportinganevent’,‘makingupastory’,‘makingajoke’and‘translatingfromonelanguagetoanother’.Thenatureofthelanguagegamecanbeascertainedonlybyobservinguse;thereisnothing‘hiddenbeneaththesurface’tobediscoveredbyanalysis.Thetaskofthephilosopher,therefore,istoobserveanddescribetheselanguagegames,nottosubjectcertain,selected,structurestologicalanalysis.Wittgensteincommentsonthephilosophicaltradition,includinghisownearlierwork:Itisinterestingtocomparethemultiplicityofthetoolsinlanguageandofthewaystheyareused,themultiplicityofkindsofwordandsentence,withwhatlogicianshavesaidaboutthestructureoflanguage.(IncludingtheauthoroftheTractatusLogico-Philosophicus.)(23)Wittgensteinalsopresentsinhislaterworkamarkedlydifferentaccountoftheroleofnames,ortherelationshipbetweenwordsandobjects,theissuewhichweconsideredinChapter1.InTractatusheseemstoenvisageastraightforwarddenotationalrelationshipbetweenwordsandobjects.Hespecifiesthat:‘Thenamemeanstheobject.Theobjectisitsmeaning’(3.203).InPhilosophicalInvestigations,however,herejectstheideathattherecanbeanystraightforwardrelationshipbetweenwordsandobjects.Individualwordscan’tbeassociatedwithanyonemeaning,butrathershouldbethoughtofastoolswhichcanbeusedforavarietyofpurposes,andwhicharebestdefinedinrelationtothosepurposes.Therefore,‘foralargeclassofcases—thoughnotforall—inwhichweemploytheword“meaning”itcanbedefinedthus:themeaningofawordisitsuseinthelanguage’(43).Thereareanumberofdifferentwaysinwhichanyonewordmaybeused.Ifweconsiderthesedifferentusesasawayofidentifyingmeaning,wewillfindthatthereisn’t‘somethingthatiscommontoall,butsimilarities,relationships,andawholeseriesofthematthat’(66).Forthisreason,it’sbesttoconsideranygivenwordashavingnotone,fixedmeaning,but‘afamilyofmeanings’(77),whicharerelatedbyaseriesof‘familyresemblances’.Justasdifferentmembersofafamilymaynotallshareevenonecharacteristicincommon,butmaybesaidtodisplayacertainnetworkofsimilarities,sothemeaningsofawordmaybeidentifiableasaset,butnotdefinedbyanyonefeature.\n114SpeakersandhearersInhislaterwork,then,Wittgensteinemphasisestheimportanceoflanguageasitisactuallyused.Untilthispoint,philosophy,includinghisown,hadlargelyignoredmanytypesoflanguageuse.Suchstraightforwardusesasaskingquestionsandissuingcommandshadlargelybeendismissedfromphilosophicalstudybecausetheycouldn’tbeexplainedintruth-conditionalterms.Wittgensteinalsorejectstheideathatanywordoranysentencehasanidentifiablemeaning,anideafundamentaltotheworkofphilosopherssuchasFrege.Rather,heseeswordsandsentencesasbeingcharacterisedbytheindividualandoftenverydifferentusestheycanbeputto.Finally,weshouldpayattentiontoastatementinPhilosophicalInvestigationsaboutthephilosophicalmethoditself.Wittgensteinremarksthat:‘WhenItalkaboutlanguage(words,sentences,etc.)Imustspeakthelanguageofeveryday’(120).Hesees‘everyday’languageuseasappropriatenotjustasatopicofstudy,butasameansforexpressingthatstudy.ThiswastobeacentraltenetinthedevelopmentofordinarylanguagephilosophyintheyearsimmediatelyfollowingWittgenstein’sdeath.OrdinarylanguagephilosophyOneofthemostinfluentialfiguresinthisandinotherareasofBritishphilosophyofthetimewasGilbertRyle.Bornin1900,RylespenthisacademiccareeratOxford,becomingProfessorofPhilosophy.In1953hepublishedanarticlecalled‘Ordinarylanguage’whichcaninmanywaysbeseenassettingtheagendafortheapproachtothephilosophyoflanguagetakeninOxfordinthemiddlepartofthetwentiethcentury.Assuch,itperhapsmadepossiblethedevelopmentoftheideaswewillbelookingatinthischapter,whichhaveinturnbeenimportanttothedevelopmentofmodernlinguistics.Rylepointsoutthat,inemphasisingtheimportanceofordinarylanguage,philosopherswere,atleastinpart,appealingtoaconsiderationofthe‘stock’usesofwords,asopposedtoanyunusualor‘non-standard’uses.Philosophywastotreatthewayinwhichspeakersordinarilyuselanguageasavalidareaofstudy.Butinadditionitwastotreatthewordsusedinphilosophyashavingtheir‘ordinary’or‘stock’meanings.4Rylesuggeststhatideallyitshouldn’tbenecessarytoexplainsuchordinaryusesofwordsalthough,hecommentswryly:inphilosophicaldebatesoneissometimesrequiredtodoit,sinceone’sfellow-philosophersareatsuchpainstopretendthattheycannotthinkwhatitsstockuseis—adifficultywhich,ofcourse,theyforgetallaboutwhentheyareteachingchildrenorforeignershowtouseit,andwhentheyareconsultingdictionaries.(110)\nSpeakersandhearers115In‘Ordinarylanguage’,Ryleemphasiseswhatwastobeanothercentraltenetofhisschoolofphilosophy,theideathatitistheuseofawordwhichshouldbeconsideredinanydiscussion,nottheworditself.Herejectstheideathatwordsinthemselves‘have’meanings,thatsomeentity,ameaning,isassociatedwitheachword.Rather,themeaning,orvalue,ofawordisdeterminedonlybyhowitcanbeused.Inalaterarticlepublishedin1957,‘Thetheoryofmeaning’,hearguesthatitisridiculoustosaythattheman,Hillary,isthemeaningofthephrasethefirstmantostandonthetopofMtEverest.Hearguesthatthisisimpossible;‘meaningsarenotbornanddonotdieandtheyneverwearboots’(134).InthispaperheacknowledgestheobviousdebtwhichtheseideasowetoWittgenstein,andparticularlytoPhilosophicalInvestigations.Inthisbook,Ryleclaims,Wittgensteinrealisedthat:‘theuseofanexpression,ortheconceptitexpresses,istheroleitisemployedtoperform,notanythingorpersonoreventforwhichitmightbesupposedtostand’(144).Thisaccountisclearlyatoddswithsomeofthoseweconsideredinearlierchapters.Remember,however,thatphilosopherssuchasFrege,RussellandCarnapwereprincipallyconcernedwiththephilosophyofmathematics,andattemptedtoexplainlanguageintermsofthelogicalregularitiestheyfoundinthatsubject.Suchearlierphilosopherswerenot,ofcourse,unawarethatlanguagewasusedinlesspreciseways,buttheyregardedsuchusesasperipheral,or‘imperfect’,andthereforenotsuitablefocusesofstudy.InChapter3wesawhowCarnapwaseagertoproducea‘perfect’languagewhichwouldobeylogicalrulesinawayinwhichordinarylanguagewouldnot.Andin‘Onreferring’,StrawsonattackswhatheseesasRussell’smisplacedattemptsto‘purifylanguage’.Aspartofourconsiderationof‘conversationalimplicature’laterinthischapter,wewilllookatsomeofthespecificproblemswhichnaturallanguagepresentsforaccountsbasedonmathematicallogic.TheordinarylanguageapproachtophilosophyflourishedforadecadeormoreatOxford,andhasbeenveryimportantinthesubsequentdevelopmentofvariousbranchesofphilosophyandlinguisticsinBritainandAmerica.Butitwasnotuniversallyaccepted.BertrandRussell,forinstance,wasstillverymuchamajorfigureinBritishphilosophyinthe1950s,andwasnotatallimpressedbythisnewmovement.HehadbeenanenthusiasticchampionofWittgenstein’searlywork,beinginstrumentalingettingTractatuspublishedinEnglish.ButhemadelittlesecretofhisdisappointmentinWittgenstein’slaterwork,seeingPhilosophicalInvestigations,inparticular,asanabandonmentofhisearlierrigourinfavourofdismissingcomplicatedmattersoflogicas‘pseudoproblems’.Russellcontinuedtoproduceworkwhichwasconcernedmainlywithmathematicalandlogicalproblemsofmeaningandknowledge,andonlysecondarilywithlanguage.WesawinChapter2,inhisresponsetoStrawson,thathewashighlycriticalof‘linguistic’approachestosuchmatters.\n116SpeakersandhearersThephilosophersworkingonordinarylanguageatOxfordinthemiddlepartofthetwentiethcenturyallsharedtheideawhichRussellfoundsoobjectionable:thatnaturallanguage,togetherwithitsuseineverydaysituations,wasalegitimatefieldofstudyinitsownright.Thisideaderivedfromtheiremphasisonthe‘everyday’usesofwordsastheappropriatetoolsforclarifying,andpossiblysolving,thephilosophicalproblemswhichareexpressedusingthem.Thisisnottosaythatthereisonlyonelineofargumentwhichcanbeidentifiedasrepresentingordinarylanguagephilosophy,orthatitsproponentswereallinagreementoneverymatter.Rather,ordinarylanguagephilosophyrepresentsaparticularframeworkwithinwhichmanydifferenttopicswerediscussed.Thetwotopicswhichhavebeenmostinfluentialinlinguistics,andwhichwillthereforebethefocusofourattentionintheremainderofthischapter,arespeechactsandimplicature.Theworkwhichweshallbeconsideringcanallbeseenasbelongingwithintheframeworkofordinarylanguagephilosophy,despitethemanydifferencesinemphasisandinterpretationwhichweshallencounter.SpeechactsIntheintroductiontothischapterweconsideredthe‘descriptivefallacy’,identifiedbyAustin,aleadingphilosopherofordinarylanguage.Austin’sfullnamewasJohnLangshawAustin,andheisgenerallyreferredtoinphilosophicaldiscussionas‘J.L.Austin’,toavoidconfusionwithanother,earlier,JohnAustin,whowroteonmoralandlegalphilosophyintheearlynineteenthcentury.J.L.AustinestablishedareputationforhisworkatOxfordaftertheSecondWorldWar,andbecameProfessorofMoralPhilosophytherein1952,eightyearsbeforehisdeathattheageofforty-nine.Inhiswork,hemadeparticularclaimsaboutthewaysinwhich,orthepurposesforwhich,peopleuselanguage.ItisinterestingthatevenStrawson,whoaswehaveseen,insistedin‘Onreferring’onthesignificanceofcontextandtheinapplicabilityofmathematicallogictonaturallanguage,seemedtotakeitforgrantedinthissamepaperthat‘Oneofthemainpurposesforwhichweuselanguageisthepurposeofstatingfactsaboutthingsandpersonsandevents’(229).Austin’sclaiminidentifyingthe‘descriptivefallacy’wasthat,onthecontrary,statingfactscouldaccountforonlyaveryfewofthemanyusesoflanguage.LittleofAustin’sinfluentialworkinthisareawaspublishedduringhislifetime;hisideasweremadepublicalmostentirelythroughhislecturesandtalks.In1955hegavetheannualWilliamJameslecturesatHarvardUniversity,andafterhisdeathhisnotesfromthesewereeditedandpublishedin1962asabook,HowtoDoThingswithWords.Thetitleisasuccinctstatementofhismaintheme.Austinwasinterestedinthemanythingswhichpeopledowithlanguage,suchasaskingquestions,issuingorders,makingrequestsandofferinginvitations.Anaccountofmeaninginterestedonlyintruthandfalsitycouldhavelittletosayaboutthese.\nSpeakersandhearers117HowtoDoThingswithWordsbeginswithaclearstatementoftheimportanceofthesenon-truth-conditionalusesoflanguagetoanyaccountofmeaning.OnthefirstpageAustinalsointroducesanotherofhiscentralobservations:thatthesedifferentusesarenotnecessarily,orevenusually,distinguishedbygrammaticalform.Considerthefollowing,allofwhichcould,inappropriatecircumstances,beinterpretedasorderstoopenawindow:6)Openthewindow.7)Couldyouopenthewindow?8)I’dbegratefulifyouwouldopenthewindow.9)Shallweletinsomefreshair?10)It’sgettingratherstuffyinhere.Agrammarianwouldnoticethatonly6)isanimperative,thegrammaticalformconventionallyassociatedwithgivingorders;7)and9)areinterrogatives,usuallyassumedtobeusedtoaskquestions;8)and10)aredeclaratives,andassuchmightbeexpectedtobeusedtomakestatementsoffact.Buttheycanallbeusedtodothesamething:toasksomeonetoopenawindow.Austindrewadistinctionbetweenthemeaningofanexpressionanditsfunction.Themeaninghesawasroughlyequivalenttotheconventionalmeaningofthesentence,includingtheappropriatesenseandreference.Thefunction,ontheotherhand,wasentirelydependentonthesituationinwhichtheexpressionwasusedand,crucially,theintentionwithwhichitwasused.Whetherornotexamplessuchas6)–10)havethefunctionofissuinganorderwilldependatleast,althoughnotexclusively,onwhetherthespeakerintendsthemtohavethisfunction.Aswewillseeinthenextsection,thesuccessoftheseexamplesasordersdependsonthatintentionofthespeakerbeingrecognised.TheusesoflanguagewhichAustindiscussedcametobeknownas‘speechacts’.ThiswasanexpressionwhichAustinhimselfusedonlyinconnectionwiththelaterpartofhisHarvardlectures,whichwewillconsiderbelow.However,itsumsuponeofhiscentralobservations.Speechcanbeanalysedfromthepointofviewnotjustofwhatinformationitcommunicates,butofwhatactsitperforms.Austinoriginallyconsideredthepossibilityofdistinguishingaparticulartypeofsuchacts,whichhelabelledperformatives.Inhislaterworkhecametoseethatsuchactswerenotasclearlydistinguishedashehadthought,andheeventuallyabandonedtheidea.Butwewilllookfirstatthecharacteristicsandpropertiesofperformatives,becausetheyillustratethenatureofAustin’sinsights,andareasignificantstageinthedevelopmentofhisideas.PerformativesThe‘performativehypothesis’,asthisstageinAustin’sthoughtcametobelabelled,holdsthatperformativescanbeclearlydistinguishedfromthoseusesoflanguagewhichareintendedsimplytoconveyinformation.Austin\n118Speakersandhearerscalledtheselatterusesconstatives.Itisappropriatetoaskofconstativeswhethertheyaretrueorfalse,butthesamecan’tbeaskedofperformatives.Instead,theyareeitherappropriateorinappropriateinaparticularcontext.InAustin’stermstheycanbeeither‘happy’or‘unhappy’.‘Happy’speechactsaresometimesknownasfelicitous,andthefactorswhicharenecessarytomakethemsoareknownasfelicityconditions.Austinnoticedthat,bytheveryactofsayingcertainwords,intheappropriatecontextandwiththeappropriateintentions,speakerscanperformtheveryacttheyareapparentlydescribing.5Hismostfamousexamplesare‘ritualisedperformatives’ofthefollowingkind:11)InamethisshiptheBlackPig.12)Isentenceyoutofiveyearshardlabour.13)Iwill.[inthecontextofaweddingceremony]Theuseoftheseexpressionscaninitselfbringaboutcertainstatesofaffairs:shipscanbenamed,convictssentenced,couplesmarried.Austinoriginallysuggestedthatcertaingrammaticalpropertiesarenecessarytoperformatives,namelythattheymustalwaysbeinthefirstperson,andthattheycanonlybeinthesimple,nottheprogressive,presenttense.Sowhile11)–13)canallbeusedtoperformactions,14)–15)canonlybeusedtodescribewhatactionisbeingperformed.14)IamnamingthisshiptheBlackPig.15)Hesentencesyoutofiveyearshardlabour.Now,utteringexpressionssuchas11)–13)doesnot,ofcourse,alwaysbringaboutanyparticularsituation.Theymustbeutteredbysomeonewithaspecificauthoritytoperformthataction,whileintheappropriatecircumstances.AsAustinnoticed,anexamplelike11)wouldn’tworkifutteredbyaprotestorwhohadjustgrabbedthechampagnebottleoutofthehandoftheinviteddignitary,and13)wouldn’tworkifutteredbysomeonewhowasalreadymarried.Theidentityofthespeakerandthedetailsofthesituationmakeupthefelicityconditionsfortheperformative.Toutter13)happilyyouneedatleast,underEnglishlaw,tobeunmarried,tobeinthepresenceofanotherunmarriedpersonofoppositesexwhomyouintendtomarry,andtobeinthepresenceofaregistrarorpriestandtwowitnesses.Therearemanyperformativeswhichhavealesslimitedusethanthosein11)–13).Austinoriginallyconsideredthepossibilityofgoingthroughadictionaryandlistingalltheverbswhichcouldbeusedperformatively.Herearejustafewexamples:\nSpeakersandhearers11916)Ibetyou£5thatJohnwillbelateforthelectureagain.17)Ipromisenottoplagiariseyouressay.18)Idareyoutostandupandsingthenationalanthembackwards.19)Iwarnyounottocomehereagain.Onthisaccount,theverbwarncanbeclassifiedasperformativebecauseyoucanwarnsomeonejustbyuttering19).Theverbinsult,however,isnotbeaperformativeverbbecauseyoucan’teffectivelyinsultsomeonebysaying‘Iinsultyou’.Althoughthecircumstancesinwhichthesespeechactscanbeperformedarelessrestricted,thereareneverthelesscertainfelicityconditionsattachedtotheiruse.Soyoucan’tactuallybringaboutabetbyuttering16)unlessthepersonyouareaddressingtakesyouuponit.Similarly,youdon’tgenuinelypromiseifyouutter17)insincerely.Again,thenotionofintentionissignificanthere;17)onlyworksasapromiseifthespeakerintendstouseittopromise,andifthehearerrecognisesandacceptsthatintention.Asmentionedabove,inthecourseofthelectureswhichwerepublishedinHowtoDoThingswithWords,Austinfoundproblemswith,andfinallyabandoned,theperformativehypothesis,althoughmanyofitsinsightsremainsignificanttohisworkandtolaterworkonspeechacts.ThevariousproblemswhichAustinidentifiedcanberelatedtohisowndistinctionbetweentruthconditionsandfelicityconditions.Consideringthetypesoflogicalrelationswhichcanexistbetweenstatements,henoticedthattherearemorewaysinwhichaconstativecanbeinappropriate,orinAustin’sterm‘outrageous’,thansimplybeingfalse.Forinstance,makingastatementimpliesbeliefinthatstatement.Tosay‘thecatisonthemat’whenyoudon’tbelievethatthecatisonthematisinsincere,andthereforeunhappyinthesamewaythataninsincerepromiseisunhappy.Similarly,Austinconsideredpresupposition.Tosay‘allJack’schildrenarebald’whenJackinfacthasnochildrenisnotfalse.WesawinChapter2thatitsuffersfromwhathasbeendescribedas‘presuppositionfailure’.Austindefinedthisasatypeofunhappiness,similartothatwhichariseswhen‘Inamethisship…’isutteredwithouttheappropriateinstitutionalsetting.Conversely,Austinnoticedthattosaythataperformativeutterancesuchas‘Iapologise’ishappyistomakeaseriesofclaimsabouttruth,mostnoticeablythat‘thestatementthatIamapologisingistrue’andthat‘thestatementthatcertain[felicity]conditionsobtainmustbetrue’(53).Thereareotherexamplesofperformativesforwhichitseemsappropriatetospeakevenmoredirectlyoftruthandfalsity.Forinstance,aperformativesuchas‘Iwarnyouthatthebullisabouttocharge’,utteredinacontextwherethebullisnotabouttocharge,wouldn’tbedescribedasafailuretowarn,butasafalseormistakenwarning.Austinalsodrewattentiontoaclassofperformativeswhichbegin‘Istatethat…’.Thesecanbeusedtomakestatementsoffact,andassuchcanbesaidtobetrue\n120Speakersandhearersorfalse.Austinclaimedthatitisfalsetosay‘IstatethatJohnisrunning’ifJohnisnotinfactrunning.ThisparticularclaimwaslatercriticisedbyJohnSearleinhis1968article‘Austinonlocutionaryandillocutionaryacts’.SearlearguesthatAustinishereconfusing‘statement-objects’with‘statement-acts’.Intheexampleabove,thestatement-objectisthe‘thingstated’:thepropositionthatJohnisrunning.Thestatement-actisthestatingitself:theactofuttering‘IstatethatJohnisrunning’.Statement-objects,then,arepropositionsandcanthereforebetrueorfalse,whereasstatement-actsareindividualactsofstating,andassuchcan’thavetruth-values.Nevertheless,Austin’srealisationthatconstativescouldinfactbediscussedintermsoffelicityorinfelicity,andperformativesintermsoftruthorfalsity,isgenerallytakentovalidatehisclaimthatthedistinctioncouldn’tbemaintained.Hedevelopedadifferentwayofthinkingaboutspeechacts,whichdependednotonadistinctionbetweenconstativesandperformatives,butonidentifyingthreelevelsofactswhichareperformedwhenanutteranceisproduced.Thesehecalledthelocutionary,theillocutionaryandtheperlocutionaryacts.IllocutionaryforceInhislaterwork,Austinreturnedtohiscritiqueofthe‘descriptivefallacy’.Heexplainedthatphilosophershaveingeneralbeenconcernedonlywiththeactofsayingsomething,ignoringthemuchmoresignificantactperformedinsayingsomething.Theseactshelabelledlocutionaryandillocutionary,respectively.ThelocutionaryactiscloselyrelatedtowhatAustindescribedas‘meaning’inhisearlierwork.Itistheactofutteringasequenceofwords,togetherwiththeirliteralmeaning,includingtheappropriatesenseandreference.Attheleveloftheillocutionaryact,theintentionofthespeakerbecomesrelevant.Toknowwhatillocutionaryacthasbeenperformedweneedtoknowwhatitisthatthespeakerintendstoachieveorbringaboutbyproducingthisutterance.Thisissometimesknownastheillocutionaryforceofanutterance.Finally,theperlocutionaryactdependsnotjustonthespeakerbutonthehearer.Itisconcernedwiththeresultorconsequenceoftheutterancehavingbeenproduced.InAustin’sterms,itiswhatwebringaboutbysayingsomething.Austinillustratesthesedistinctionswithadiscussionofthepossibleusesofanutteranceof‘shoother’.Examples20),21)and22)respectively,representsomeofthepossiblelocutionary,illocutionaryandperlocutionaryactsinvolved.20)Hesaidtome‘Shoother!’meaningby‘shoot’shootandreferringby‘her’toher.21)Heurged(oradvised,ordered,etc.)metoshoother.\nSpeakersandhearers12122)Hepersuadedmetoshoother.Theperlocutionaryactin22)isonewhichcoincideswiththespeaker’sintention,butAustinnotedthatthisneedn’talwaysbethecase.Thehearermay‘missthepoint’andfailtorespondasexpected.Orhemayunderstandtheintentionbutfailtocomplywithit,forinstancebyproducingafurtherutterance,suchasoneobjectingtothefactthatthefirstutterancewasperformed.Austindescribedillocutionaryactsasbeingdefinedbyconvention,inthattheyarerelatedtotheformofwordsused.Perlocutionaryacts,ontheotherhand,hesawasunpredictableandcontext-dependent.Austinsawthenotionofillocutionaryforceasa‘generaltheory’ofmeaning,whichcouldexplaintheproblemshehadidentifiedforthedistinctionbetweenconstativesandperformatives.Bothtypesofutterancecanbesaidtohaveillocutionaryforce,since‘tostateiseverybitasmuchtoperformanillocutionaryactas,say,towarnortopronounce’(134).Tosay‘Istatethathedidnotdoit’,inwhichtheillocutionaryforceofstatingisexplicitlyidentified,is,Austinclaims,toperformthesameactasusingthe‘primary’form,‘hedidnotdoit’.SearleonspeechactsInthelastlectureinHowtoDoThingswithWords,Austinsetouttoclassifydifferenttypesofutterance,accordingtotheirillocutionaryforce.ItwasthistaskwhichwassubsequentlytakenupbyJohnSearlealthough,aswewillsee,hesuggestedvariousmodificationstoAustin’saccount.SearlewaswellplacedtoevaluateanddevelopAustin’sideas.AsastudentatOxfordhewastaughtbybothAustinandStrawsonduringthe1950s.HecompletedaD.Phil,thesisonsenseandreferencein1959,andlaterbecameProfessorofPhilosophyattheUniversityofCaliforniaatBerkeley.HecanbeseenasanimportantfigureinlinkingthepredominantlyEnglish-basedworkofthephilosophersofordinarylanguageinthemiddlepartofthetwentiethcentury,withworkonthephilosophyofmindandinthegrowingdisciplineoflinguisticsinAmericainthelaterpartofthatcentury.IntheprefacetoSpeechActs,abookpublishedin1969,herecordshisthankstobothJ.L.AustinandNoamChomsky.InSpeechActs,andinvariouspublishedarticlesonthesametopic,SearledevelopsAustin’sworkbyenumeratingthedifferentclassesofillocutionaryact,byelaboratingthenatureofthefelicityconditionsfortheseclasses,andbyrelatingtheaccountofspeechactstowiderissuesinthephilosophyoflanguage.Hisdiscussionofsomeoftheseissuesbeginstosoundlikethequestionswhichwerelatertobeaskedinpragmatics.Forinstance,nearthebeginningofSpeechActsheasks:‘WhatistherelationbetweenwhatImeanwhenIsaysomethingandwhatitmeanswhether\n122Speakersandhearersanyonesaysitornot?’(3).ThisbookalsoincludessomeparticularlybluntexpressionsofwhatSearlesawastheessentialrelationshipbetweenmeaninganduse.Forinstance,borrowingtheby-thenpopularcomparisonbetweenlanguagesandgames,hesuggeststhatstudyinglanguagewithoutuseislikestudyingtherulesofbaseballwithoutconsideringitasagame.TheworkforwhichSearleisprobablybestknown,atleastinlinguistics,ishisaccountofindirectspeechacts(oftenabbreviatedtoISAs).Inhis1975articleofthatname,SearledescribesISAsas‘casesinwhichoneillocutionaryactcanbeperformedindirectlybywayofperforminganother’(60).Heisconsideringexamplessuchasthefollowing,commonenoughinordinarylanguage,butpotentiallyproblematicforanaccountofspeechactsintermsoftheillocutionaryactperformed:23)Canyoupassthejam?24)I’dbegratefulifyouwouldkeepquiet.25)Wouldyoubewillingtohelpme?Intheseexamples,itisn’tpossibletodefineasingleillocutionaryact,evenifweareawareofthespeaker’sintentions.Although23)canbeunderstoodasanenquiryaboutthehearer’sabilities,itcanalsobeunderstood,andisinfactmostlikelytobeintended,asarequest.Searledefinedthismostlikelyintentionastheprimaryillocutionoftheutterance,whichisderivedfromthesecondaryillocution,inthiscasethatofquestion,withreferencetothefelicityconditionsforeachact.Inthecaseofquestions,oneofthefelicityconditionswhichSearleidentifiesisthatthespeakermustbegenuinelyignorantastotheanswer.Inmostcasesinwhich23)isuseditwillbeunlikelythatthespeakerreallycan’tassesswhetherthehearerisabletopassthejam.However,thehearer’sabilitytopassthejamisoneofthefelicityconditionsformakingarequestthathedoso.6Example23)willthereforebeinterpretednotasaquestionbutasarequest.Searledescribestheprimaryillocutionaryforceas‘conventional’.However,hearguesthatitisn’tappropriatetodescribeexamplessuchas23)ashavingthesingleillocutionaryforceof‘request’,ortoarguethatthey‘haveanimperativeforceaspartoftheirmeaning’(67).Itwouldbequiteacceptableforthehearertorespondtobothprimaryandsecondaryillocutionaryacts,forinstancebysaying‘yes’andpassingthejam.Itissignificantthat,whileitwouldbepossibletorespondtojustthesecondary,derivedillocutionaryforce,bypassingthejaminsilence,itwouldseeminappropriatetorespondonlytotheprimaryforce.Inotherwords,itwouldseematleastsarcastic,andprobablybeinterpretedasdeliberaterudeness,simplytoanswer‘yes’.Similarexplanationscanbefoundfor24)and25).Anutteranceof24)isunlikelytofulfilthefelicityconditionforanassertion,namelythatitconveysinformationwhichisnewtothehearer;itisthereforelikelytobeinterpretedasarequestforsilence.And25)islikelytobeunderstoodnotas\nSpeakersandhearers123asimplequestion,butasanindirectrequestforhelp.Searlenotices,almostinpassing,that‘Indirectives,politenessisthechiefmotivationforindirectness’(64),apointwhichwastobetakenuplaterinpragmatics,aswewillseeattheendofthischapter.Inlookingatspeechacts,wehavebeenreferringtotheimportanceofspeakerintentionindefiningmeaning,andindeterminingtheeffectofanutterance.Thisconceptisanimportantoneinordinarylanguagephilosophy,andalsoinpresentdaylinguistics,andit’sworthtakingsometimetoconsiderhowitfitsintothebroaderphilosophicalframework.Indoingso,wewilllookatsomeimportantworkinthisareabyPaulGrice,whoseaccountof‘conversationalimplicature’,alsodevelopedwithinordinarylanguagephilosophy,willbeournextmaintopic.MeaningandintentionTwobasicbeliefsaboutlanguagehavebeenmoreorlessimplicitintheworkofmanyofthewriterswehaveconsidered,fromclassicaltimesrightuptothetwentiethcentury.Thefirstofthesebeliefsisthatthesignificanceofwordsisarbitraryandconventional.Thesecondisthat,whenitisusedasameansofcommunication,languageoperatesinmuchthesamewayasanyotherarbitrarycode;athoughtinthespeaker’smindisencodedintolanguage,whichisthendecodedbythehearer,sothatthethoughtendsupinhismind.Thecommunicationwillbesuccessfultotheextentthatthethoughtinthemindofthespeakertobeginwithandthethoughtinthemindofthehearertofinishwitharethesame.Thefirstofthesebeliefswasputforwardbymanyofthephilosophersinterestedinmeaningwhomwehavestudiedinearlierchapters,forinstancePlato,LockeandMill.ItwaschallengedbyLeibniz,whoattemptedtoprovidereasonsfortheformswhichwordstake.Butit’sgenerallyagreedthatitisn’tappropriatetoask‘why’aparticularwordhasaparticularmeaning.OneofthebestknownstatementsofthisistheoneofferedbytheSwisslinguistFerdinanddeSaussure,whoisregardedasoneofthefoundersofStructuralism,andwhoseworkwewillconsiderincontextinthefinalchapter.Inhis1915CoursdeLinguistiqueGenerate(translatedintoEnglishasCourseinGeneralLinguisticsin1960),hediscussestherelationshipbetweena‘signifier’,forourpurposesaword,andthe‘signified’,theconceptwhichformsitsmeaning.Saussurenotesthatit’sgenerallyacceptedthat‘thebondbetweenthesignifierandthesignifiedisarbitrary’(67),andgoesontoadvanceinsupportthefactthatdifferentlanguageshavedifferentwordsforthesameconcept:Theideaof‘sister’isnotlinkedbyanyinnerrelationshiptothesuccessionofsoundss-ö-rwhichservesasitssignifierinFrench;thatitcouldberepresentedequallybyjustanyothersequenceisprovedby\n124Speakersandhearersdifferencesamonglanguagesandbytheveryexistenceofdifferentlanguages:thesignified‘ox’hasasitssignifierb-ö-fononesideoftheborderando-k-s(Ochs)ontheother.(67–8)Thesecondbasicbeliefinasensefollowsfromthis;iflanguageconsistsofaseriesof‘signs’whicharbitrarilystandforideas,thenitmustbepossibleto‘encode’athoughtbytranslatingitintoaseriesofsigns.Inmostdiscussionsthesesignsareidentifiedassounds,which,whenreceivedbyaspeakerofthesamelanguagecanbe‘decoded’backintothesamethought.Morerecentphilosophersandlinguistshaveretainedthe‘arbitrarysign’beliefbutrejectedthe‘code’,oratleastthe‘simplecode’modelofcommunication.ThisapproachcanbefoundintheworkofPaulGrice.Gricewasamemberofthegroupofordinarylanguagephilosophers,althoughhehimselfwasuneasyaboutthesuggestionthatthisrepresentedasingleschoolofthought.HewasactiveinOxfordatthesametimeasAustinand,likeSearle,helaterbecameProfessorofPhilosophyattheUniversityofCalifornia,Berkeley.Hecontinuedwritingonvariousareasofphilosophyalmostuntilhisdeathin1988.Griceisnowbestknownforhisworkon‘conversationalimplicature’,whichwewilllookatnext.Butin1957,tenyearsbeforehedeliveredtheseriesoflecturesinwhichthisworkwasoutlined,hepublishedanarticlecalledsimply‘Meaning’.Gricebeginsthisbycomparingsomeofthewaysinwhichthewordmeanisusedinordinarylanguage.Herearesomeofhisexamples:26)Thosespotsmeanmeasles.27)Thosethreeringsonthebell(ofthebus)meanthatthebusisfull.28)Thatremark‘Smithcouldn’tgetonwithouthistroubleandstrife’,meantthatSmithfoundhiswifeindispensable.Griceobservesthattherearetwodifferenttypesof‘meaning’involvedhere,onerepresentedby26),andtheotherby27)and28).Thefirsttypehedescribesas‘naturalmeaning’.Thespotscorrespondtomeaslesbecauseoftheirphysicalnature;theyaresymptomsofthedisease.Itisnotthecasethatanyonemeansanythingbythespots.Andfurther,if26)istruethenwearenecessarilyfacedwithacaseofmeasles.Examples27)and28),however,representwhatissometimesknownas‘conventionalmeaning’.Thethreerings,andthequotedutterance,havethemeaningascribedtothemnotbecauseoftheirphysicalnature,butbecauseoftheirplaceinapre-arrangedsystem.Inthecaseofthethreeringsthissystemisaverysimpleone,acodeagreedbetweentheconductorandthedriver.In28),however,thesystemisaverycomplicatedone.Themeaningoftheutterancereliesontheindividualmeaningsofallthewordsitcontains;ineffectitreliesontheEnglishlanguage.Inthesecasessomeone(thebus\nSpeakersandhearers125conductor,theproduceroftheoriginalremark)meanssomethingbytheringsandtheutterance.Andit’spossiblefor27)and28)tobetrueinthemselveswhilethese‘meanings’areincorrect.Thebusmaystillhaveroomonitbuttheconductorincorrectlybelievesittobefull;Smithandhiswifemayhavebeenseparatedformanyyears,butthespeakerbeingreportedwastryingtohidethisfact.Griceproposestocallthetypeofmeaningrepresentedby27)and28)‘non-naturalmeaning’,or‘meaning’forshort.Insuchcasesit’sNNappropriatetosaythatsomeone(A)meantsomethingbyanutterance(x).Gricethereforeintroducesanotionofintentiontothistypeofmeaning;thespeakerproducesanutterancewithaparticularintentioninmind,whethertoinform(asisgenerallythecasewithdeclaratives)ortoproducearesult(aswithimperatives).Butitisn’tjustthespeaker’sintentionwhichisimportant,butalsotherecognitionofthisintentionbythehearer.Gricesuggeststhatifheleavessomethingprominentinaroomforhiswifetofind,forinstanceapieceofchinabrokenbyhisdaughter,hemaywelldosowiththeintentionofimpartingsomeinformationtohiswife,forinstancethataparticularaccidenthastakenplace.However,hiswifewillunderstandthisinformationwhetherornotsherecognisesthisasanactofcommunication;thebrokenpiecesareinthemselvesanaturalindicationthatanaccidenthastakenplace.Thiswon’tbeacaseofmeaningNNbecausetheintention,althoughpresent,wasn’tnecessarytothecommunication.GriceoffersthedefinitionthattosayAmeantNNsomethingbyxistosaythat‘Autteredxwiththeintentionofinducingabeliefbymeansofrecognitionofthisintention’(219).7Thesamedistinctionbetweennaturalandnon-naturalmeaningcanbemadewithrespecttocommunicationintendedtocauseaparticularactiononthepartofanother,ratherthansimplytoimpartinformation.Gricecomparesthecaseof‘apolicemenwhostopsacarbystandinginitsway’withthatof‘apolicemanwhostopsacarbywaving’(220).Inbothcasesthepolicemanintendstomakethecarstop,butonlyinthesecondcaseisitnecessaryforthedrivertorecognisethisintentioninorderforthecommunicationtobesuccessful.Inthefirstcasetheactofstandingintheroadisinitselfsufficienttostopthecar.Onlythesecondcaseisanexampleofmeaning.Withthistypeofmeaning,‘toaskwhatAmeantistoaskforNNaspecificationoftheintendedeffect’(220).Grice’selaborationoftheroleofintentioninutteranceproductionandcomprehensionhadanimportantinfluenceonhisownlaterwork,andonmuchoftheworkwhichwastobeproducedinthisareainlinguistics.Forinstance,itexplainsonewayinwhichanaccountofcommunicationwhichisdependentonlyonspeakerandhearersharingacommoncode,isinadequatetoexplainwhatgoesonininstancesoflanguageuse.Thehearerneedstobeablenotjustto‘decode’thespeaker’sutterance,butalsotobeabletoassesstheintentionwithwhichsheproducedit.This,togetherwithothertypesofinformationderivedfromcontext,needstobeaddedtowhat\n126Speakersandhearerscanbeunderstoodfromencodedmeaning.Inotherwords,the‘code’modelofcommunicationneedstobecomplementedbyan‘inferential’model.Inthenextsection,wewillconsiderhowGrice’sapproachtomeaninginformedhisownaccountoflanguageuse,andinparticularhowheexplainedtheseadditional,‘inferential’aspectsofmeaning.ConversationalimplicatureLikeAustin,GricepresentedsomeofhismostinfluentialideasaboutlanguagewhenhewasinvitedtogivetheWilliamJameslecturesatHarvard.Grice’slecturesweregivenin1967,andwereentitled‘Logicandconversation’.Someoftheideaspresentedtherewerepublishedinanarticleofthesamenamein1975,andthelectureswerelaterpublished,togetherwithselectionsfromGrice’sotherwritings,inStudiesintheWayofWordsin1989.‘Logicandconversation’beginswithadiscussionofthedifferencesbetweenthelogicaloperatorsandtheirapparentcounterpartsinnaturallanguage.Gricerejectsthe‘formalist’approachtothisproblem:theclaimthatnaturallanguageisimperfectandthatonlyanidealisedformoflanguageisworthstudying.Healsorejectsthe‘informalist’account:theclaimthatnaturallanguageshouldbestudiedinitsunsimplified,illogicalform.Grice’sreasonforrejectingbothapproachesisthattheirbasicpremiseisflawed,andthatnaturallanguageexpressionsdon’tdivergefromtheirlogicalcounterpartsasmuchasissupposedineither.Hesuggeststhatthedifferencesininterpretationcanbeexplainednotintermsoftheactualmeaningsofthenaturallanguageexpressions,butintermsoftheusetowhichtheyaregenerallyput.Significantly,heclaimsthatthismistakenassumption‘arisesfrominadequateattentiontothenatureandimportanceoftheconditionsgoverningconversation’(24).BeforeweconsiderGrice’ssuggestionastowhattheseconditionsare,wewilllookatsomeoftheapparentdiscrepanciesbetweenexpressionsoflogicandofnaturallanguage.Insodoing,wewillbereferringbacktoourinvestigationoftheoperatorsofpropositionallogicinChapter2.NaturallanguageandlogicItisn’thardtounderstandwhythoseinterestedinthelogicofmathematicsmightbetemptedtorelatethisinteresttolanguage.Thereareanumberofwordsandexpressionsinnaturallanguagewhichappeartohaveequivalentsinlogic.Themostobviousandfrequentlycitedoftheseisthewordand.Ifaskedtodefinethemeaningofand,wemightbeginbysuggestingthat,likelogicalco-ordination,itjoinstogethertwopropositions,expressedbythetwoconjuncts.Andifwewereinclinedtodefinemeaningintermsoftruth,wemightgoontosay\nSpeakersandhearers127thatitsmeaningistruth-functional.Joiningtwotrueconjunctsgivesatruestatement,whilejoiningonefalseandonetrue,ortwofalseconjuncts,givesafalseone.Thisisthestandarddefinitionforlogicalco-ordination,theonewhichappearsintruthtables.Itcanbeillustratedwithexamplessuchasthefollowing:29)WellingtonwontheBattleofWaterlooand1968wasaleapyear.30)WilliamtheConquerorinvadedEnglandin1066andHenryVIIIhadsevenwives.31)HenryVIIIhadsevenwivesandWilliamtheConquerorinvadedEnglandin1066.32)NapoleonwontheBattleofWaterlooand1979wasaleapyear.Assumingthatwecanagreeonourfacts,wecanprobablyagreethat29)istrue,becauseitistheco-ordinationoftwotrueconjuncts,whereas30),31)and32)areallfalsebecausetheyeachcontainatleastonefalseconjunct.Noticefromacomparisonof30)and31)thatforsuchanaccountofandtheorderoftheconjunctsisnotsignificant;itdoesn’taffectwhatwewanttosayabouttruthandfalsity.Itisalsosignificantthatallwearesayingabout29)isthatitistrue;wehaven’tmadeanyclaimsaboutitsoundinginanysense‘natural’:thesortofthingwemightimaginesaying,orthatwemightbeabletothinkofanappropriatecontextfor.Thisispreciselytheproblemwithapplyingsuchlogicaldefinitionstoexpressionsinnaturallanguage.Althoughtrue,29)soundsrather‘odd’,becausewecan’tfindanysignificantlinkbetweenthetwoconjuncts.Inotherwords,althoughweknowwhatthesentencemeans,wecan’timaginewhatanyonemight‘mean’bysayingit.Incontrast,33)seemsmuchmorenatural:33)TheLoneRangerjumpedontohishorseandrodeoffintothesunset.Wecanimaginesaying33),becausewecanseeaconnectionbetweenthetwoconjuncts.Butnoticethatsomethingelsehaschangedtoo.Nowitdoesseemtomatterwhichordertheconjunctsarein;the‘?’infrontof34)isaconventionalwayofshowingthatitsoundsdistinctly‘odd’eventhough,ifbotheventsdidinfacthappen,itshouldbenodifferentintruth-valuefrom33).34)?TheLoneRangerrodeoffintothesunsetandjumpedontohishorse.8Thisoddityarisesbecauseandinthiscaseseemstomeanmorethanjustitslogicalequivalent.Itseemstomeanthattheeventsdescribedbothtookplaceplustheytookplaceintheorderinwhichtheyarepresented.It’snotjusttheorderofeventswhichweseemtobeabletounderstandfromusesofthewordand.Examples35)and36)illustratetwofurther\n128Speakersandhearersuses,bothofwhichseemperfectlynatural,andalsoseemtohavesome‘extra’meaningattached.35)Theinebriatedlecturerfellofftheplatformandshehadtobetakentohospital.36)Johnwalkedintotheroomandheturnedonthetelevision.Again,wewouldprobablyunderstandthatthetwoeventsdescribedin35)occurredintheorderpresented,butthereisalsoanotionof‘causality’associatedwiththisuseofand.Weunderstandnotjustthattheeventstookplaceintheorderpresented,butthatthesecondeventoccurredasaresultofthefirst.Similarly,ininterpreting36)itcouldbeclaimedthataswellasorderingthereisan‘identityofplace’;weunderstandthatJohnturnedonthetelevisionwhilehewasintheroom.Noticethatthevariousparaphrasesforand,whetheritbe‘andthen’,‘andasaresult’,or‘andinthatplace’,allincludesimpleco-ordination.Itisalmostimpossibletoparaphraseanyofthesemeaningsofandwithoutstarting‘and…’.Soalltheusesofandsharethisaspartoftheirmeaning,buttheyallhavedifferent,extrameaningsaswell.Beforeweconsiderhowwemightaccountforallthesevariations,wewilllookattwofurtheraspectsofnaturallanguagewhichmightappeartohavelogicalequivalents,butwhichpresentproblemssimilartothosewehaveencounteredwithand.Thesearedisjunctionandnegation.Considerthewordor.Youmightwanttoclaimthatatleastyouknowwhatthismeansand,crucially,howitdiffersfromand.Ifthefollowingsignweredisplayedinacafeteria,youmightreasonablyexpecttobechallengedforhelpingyourselftobothsoupandapudding.37)Lunchincludessouporapudding.However,thereisanotheruseoforwhichdoesseemtoallowforboththeapparentalternatives,andthistooisusuallyeasilyinterpreted.Ifitwerethecasethatyouwereoversixty-fiveandalsoonalowincome,youwouldbeunlikelytobedeterredfromaskingforaconcessionifyoureadthefollowingnotice:38)Concessionsareavailabletothosewhoareover65orareonalowincome.Toglossthisuseoforwewouldprobablysaythattheappropriateconditionsaremetifatleastoneofthealternativesisthecase.Andthisisclosetothelogicalrelationshipof‘disjunction’,therelationshiptowhichnaturallanguageorissometimescompared.Rememberthatinthetruthtablefordisjunction,theexpressionasawholeisfalseonlyifbothdisjunctsarefalse.Inallothercasesitistrue.Thisrelatestowhatissometimes\nSpeakersandhearers129describedas‘inclusiveor’.Incontrast,usesoforsuchasthatin37)areknownas‘exclusiveor’.Equatingorwithlogicaldisjunctiondoesn’tallowustoexplainsuchuses.Thereisonefurtherproblempresentedbywordssuchasandandor.Theextra,non-logicalmeaningsassociatedwithvarioususesofthemaredefeasible.Thatis,it’spossibletousetheexpressionandthentodenythenon-logicalmeaning.Incontrast,youcan’tusetheexpressionandthendenypartofitslogicalmeaningwithoutsayingsomethingwhichisnonsensical,orcontradictory.Example39)belowisquiteacceptable,while40)isjustacontradiction.39)Itookoffmyshoesandgotintobed,butnotinthatorder.40)!Itookoffmyshoesandgotintobed,butIdidn’tdoboththosethings.9Thesameseemstoholdforthedifferentmeaningsofor.Youcandenythe‘exclusive’senseofor,asin41),butitdoesn’tworktotrydenyingthe‘logicaldisjunction’sense,asin42).41)Youcanhavecreamoricecreamwithyourpudding—infactyoucanhaveboth.42)!Youcanhavecreamoricecreamwithyourpudding—infactyoucan’thaveeither.Similarproblemsareposedbynegation.Inthecaseoflogicalnegation,wemightsaythatifastatementistruethenitsnegationmustbefalse,andvice-versa.Therearevariousexpressionsinnaturallanguagewhichmightbethoughtofascorrespondingtologicalnegation,themostobviousbeingnot,aswellasitisnotthecasethat,itisnottruethat,andtheprefixun-onadjectives.If43)istruethen44)mustbefalse,andthesameappliesto45)and46).43)FredericklivesinCanterbury.44)FrederickdoesnotliveinCanterbury.45)Tabathaishappy.46)Tabathaisunhappy.Now,iftheseexpressionsweresimplyequivalenttologicalnegation,weshouldfindthatusingtwosuchexpressionswouldcanceleachotherout,givingustheequivalentofasimplepositivestatement,justaswearetaughtinmathematicsthat‘twominusesequalaplus’.However,ifweweretohearorreadexample47),wewouldn’tbeinclinedtothinkthatitwassimplyequivalentto48):47)Ifoundyourlecturenotunhelpful.48)Ifoundyourlecturehelpful.\n130SpeakersandhearersWeneedtobeabletoexplainwhythesetwodon’tseemtomeanthesame:why,inshort,47)seemstobearathermore‘reserved’formofpraisethan48).Therearefurtherproblemsforanaccountofmeaninginnaturallanguagedependentonlogicalrelationsoftruthandfalsity,thistimeoneswhichdon’tdependonthesupposedequivalencebetweenlogicalfunctionsandindividualwords.Tautologiesareexpressionswhich,logicdictates,arenecessarilytruebecauseoftheirinternalproperties.Astatementofidentitybetweentwoobjectswhicharethesamemightbesaidtobenecessarilytrue,aswouldamathematicalequationbewhichstatedthat‘x=x’.Butyetagainthesituationinnaturallanguageseemstoberathermorecomplicatedthanalogicaldefinitionallows.ConsiderGrice’sexampleofalogicaltautology:49)Wariswar.Logicallythisisanecessarytruth,andifmeaningweredefinedintermsoftruthrelationsthereshouldbenothingmorewhichwecouldsayaboutit.Butwecanimagineasituationinwhichsomeonemightusethisexpressiontomakesuggestionswithwhichwemightwanttodisagree.Forinstance,wemightwanttoarguethat,evenduringwartime,certainrulesmustapply,orcertainhumanrightsmustbeupheld.Somethingsimilarcanbesaidaboutcontradictions.Itispossibletofindexamplesinnaturallanguageuseinwhichastatementwhichisnecessarilyfalsecanneverthelessbeusedtomakeasignificantcomment,somethingwithwhichyoumightwanttoagree.Mostpeoplewouldagreethat50)and51)arebothperfectlyacceptable,andcanbeusedmeaningfully,eventhoughtheyarelogicallycontradictory.50)I’mnothisgirlfriend—he’smyboyfriend!51)I’msorrytobeleavingandI’mnotsorrytobeleaving.RememberthatGricemaintainedthatnaturallanguagecanbeseenascloselyrelatedtoitslogicalcounterparts,providedthatsufficientattentionispaidtotheconventionsofconversationalexchange,thecontextinwhichnaturallanguageismostfrequentlyused.Hethereforesethimselfthetaskofoutliningtheseconventions,insuchawaythattheywouldbeabletoexplaintheapparentdivergencesbetweenlogicandnaturallanguage.Itishisversionoftheseconventionswhichwenowneedtoinvestigate.Theco-operativeprincipleItisGrice’sinterestintheregularitiesofconversationwhichformsthebasisofhisaccountofmeaninginuse.In‘Logicandconversation’hedefines\nSpeakersandhearers131conversationasanessentiallyinteractiveandco-operativeprocess.Hesumsuptheconditionswhichgovernitinhis‘co-operativeprinciple’,whichhestatesasfollows:52)‘Makeyourconversationalcontributionsuchasisrequired,atthestageatwhichitoccurs,bytheacceptedpurposeordirectionofthetalkexchangeinwhichyouareengaged.’Unfortunatelythisprincipleissometimesreadasa‘rule’forproperconversationalbehaviour,andGriceissometimescriticisedfortryingtoregulatesomethingwhichshouldbeafreeandcreativeprocess.Butit’sclearfrom‘Logicandconversation’,andfromotherwritings,thathedoesn’tseehimselfaslayingdownrulesforcorrectbehaviour,orastryingtoimposeorderonconversationalpractice.Rather,heisattemptingtoaccountforhowpeopleactuallybehave,toexplainhowconversationworks.Theco-operativeprincipleisintendedtobedescriptive,describingwhatpeopledowhentheyengageinconversation,ratherthanprescriptive,layingdownlawsforhowtheyoughttobehave.Inexplainingconversation,Griceisdrawingonhisinterestintherelationshipbetweenintentionandmeaning.Hisaccountofconversationisbasedontheassumption,nowcommonplaceinpragmatics,thatlinguisticinteractionisgenerallysuccessfulbecausespeakersintendtocommunicate,andbecausehearersrecognisethisintention.Peopleingeneralbehaveco-operativelyinconversationbecauseingeneraltheywanttobeunderstood.Theco-operativeprincipleisanormofsuccessfulconversationalbehaviour,butitsfullsignificancetointerpretationonlyemergeswhenitisapparentlynotbeingfollowed.Griceidentifiesfourmainareasinwhichconversationalpartnersco-operate,andpresentstheseasfourmaximsofconversation.Thesemaximstogetherdescribehowspeakersfollowtheco-operativeprinciple,andtheyarethemselvesdividedupintofurthersubmaxims.Thefourmainmaximsrelatetoquantity,quality,relationandmanner,andwewillbrieflyconsiderthesefourcategoriesbeforeexaminingtheireffectsonmeaninginconversation.Themaximofquantityisconcernedwiththeamountofinformationwhichissuppliedinanyconversationalcontribution.Itissubdividedintotwosubmaxims,whichGriceexpressesasfollows:53)‘Makeyourcontributionasinformativeasisrequired(forthecurrentpurposesoftheexchange).’54)‘Donotmakeyourcontributionmoreinformativethanisrequired.’Thesetwosubmaximsdescribethedelicatebalancewhichmakesacontributionappropriate.Itisn’tco-operativetowithhold(relevant)information,butnorisitco-operativetoprovideanexcessof(irrelevant)\n132Speakersandhearersinformation.Itissignificantthatit’salmostimpossibletoexplainthemaximofquantitywithoutmakingsomereferencetorelevancewhich,aswewillsee,isdefinedbyaseparatemaxim.ThedistinctionsandrelationshipsbetweenthemaximsareoneoftheacknowledgedproblemswithGrice’saccount;hementionsithimself,andithasfrequentlybeendiscussedincommentariesonhiswork.Themaximofqualityisalsoconcernedwiththeinformationprovidedbyaspeaker,thistimewiththequality,orveracity,ofthatinformation.Again,Gricesuggeststwosubmaxims:55)‘Donotsaywhatyoubelievetobefalse.’56)‘Donotsaythatforwhichyoulackadequateevidence.’Itisuncooperativetoprovideinformationwhichyoudon’tyourselfbelieve,orwhichyouhavenoparticularlyconcretereasontobelieve.Themaximofrelationistheleastdefinedand,byGrice’sownadmission,themostproblematicofthemaxims.Hedefinesitsimplyasfollows:57)‘Berelevant.’Griceadmittedthatmoreworkneededtobedonetoelaboratethis,worktowhichhehimselfneverreturned.Themaximofmanner,unliketheotherthreemaxims,relatesnottotheinformationconveyedinaconversationalcontribution,buttohowitisconveyed.Griceoffersfourseparatesubmaximsrelatingtomanner:58)‘Avoidobscurityofexpression.’59)‘Avoidambiguity.’60)‘Bebrief.’61)‘Beorderly.’Hesuggeststhat‘onemightneedothers’(27).Thesefourmaxims,then,canbeseenasdescribingtheformoffullyco-operative,fullycommunicativeconversation.Theyalso,significantly,offeranexplanationofhowitisthatliteralmeaningsooftendiffersfromthemeaningintendedbyaspeaker.Todescribethisadditional,intendedmeaning,Gricesuggeststheterm‘implicature’,andheproposestoinvestigatethedistinctionbetween‘whatissaid’and‘whatisimplicatedbythesayingofwhatissaid’.Indoingso,GriceisacknowledgingasimilardistinctiontothatofwhichAustinwasawarewhenhediscussedthedifferencebetweenmeaningandfunction.ButGricegoesfurtherthanAustin;aswellasidentifyingvariousdifferencesbetweenlinguistic-andspeaker-meaning,heattemptstoprovideanexplicitaccountoftheprincipleswhichaccountforthesedifferences.Hisaimistoshowhowwhat\nSpeakersandhearers133isimplicatedisderived,asaresultofco-operationbetweenspeakerandhearer,fromwhatissaid.Griceseeshis‘whatissaid’asasomewhatelaborateformofliteralmeaning.Hedescribeswhatissaidasbeing‘closelyrelatedtotheconventionalmeaningofthewords(thesentence)uttered’(25),butasincluding,atleast,referenceassignmentandanynecessarydisambiguation.Itisfromthisthat‘whatisimplicated’isderived.Griceallowsthat‘insomecasestheconventionalmeaningofthewordsusedwilldeterminewhatisimplicated’(25).Heillustratesthispointwiththefollowingexample:62)HeisanEnglishman;heis,therefore,brave.GriceclaimsthatallthatisstrictlysaidhereisthatthesubjectpossessesthepropertiesbothofbeingEnglishandofbeingbrave.Theideathatthetwoarecausallylinked,thatbraveryfollowsfromEnglishness,isanimplicature.Example62)wouldthereforenotbestrictlyfalseifthiscausalrelationdidnothold,althoughitwouldbeamisleadingwayofpresentingthetwoproperties.Gricelabelsthistypeofphenomenonconventionalimplicature.However,itisconversationalimplicatureswhicharedependentonthemaximsofconversation,andonwhichwewillconcentratehere.Weshallbeginbyreturningtosomeoftheexamplesweconsideredearlier,whichseemedtoillustrateamismatchbetweenformallogicandnaturallanguage.WeshalllookathowGrice’saccountmightexplainthese.WehaveseenthatGricedescribesthediscrepanciesbetweenlogicandnaturallanguageasapparentratherthanreal.Soforhimthewordandinexamplessuchas63)and64)literallymeanssimplythatthetwoeventstookplace.Thisis‘whatissaid’.63)TheLoneRangerjumpedontohishorseandrodeoffintothesunset.64)Theinebriatedlecturerfellofftheplatformandshehadtobetakentohospital.However,inpresentingtheinformationinthisway,thespeakerimplicatessomeadditionalmeaning.Specifically,ifthehearerassumesthatthespeakerisbeingco-operative,hewillassumethattheordersinwhichtheeventsareexplainedtohimareinsomewaysignificant,andthereforethatthereisalinkofsequencein63)andofsequenceandprobablyalsocausalityin64).Itwouldbeinappropriateandmisleading,althoughnotstrictlyfalse,foraspeakertoproduceanutteranceof64)todescribetwoeventswhichwerecompletelyunrelated.Wecouldsaythattheseimplicaturesfollowfromanassumptionthatthemaximofmannerisbeingobserved,andparticularlythesubmaximto‘beorderly’.Similarly,wecouldsaythattheinclusivemeaningoforiswhatisactuallysaidinthefollowing:\n134Speakersandhearers65)Lunchincludessouporapudding.Strictlyspeaking,itwouldn’tbefalsetousethisinasituationinwhichyoucouldhavebothsoupandpudding.However,ifthespeakerwereinapositiontogivethisadditionalinformationsheought,followingthemaximofquantity,todoso.Thehearerassumesthatthespeakerisprovidingalltheinformationsheisableto,andthereforethattheuseoforherecanbetakentoimplicateanexclusivemeaning.Ineachoftheexampleswehaveconsideredsofar,theimplicaturearisesfromtheuseofaparticularword,andfromthehearer’sknowledgeoftheliteralmeaningofthiswordtogetherwithhisassumptionthatthespeakerisbehavingco-operatively.Suchexamplesarenotdependentonanyparticularcontextofutterance,andaredescribedbyGriceasgeneralisedconversationalimplicatures.Aswesawearlier,suchaspectsofmeaningcanbecancelledwithoutcontradiction,forinstancein66):66)Youcanhavecreamoricecreamwithyourpudding—infactyoucanhaveboth.Gricedistinguishedthistypeofimplicaturefromthosewhichhedescribedasparticularisedconversationalimplicatures.Thesearedependentoncontext,andarethereforeparticulartoaspecificuse;removedfromthatcontextthewordsarenotinanywayassociatedwiththeimplicatures.Grice’sexamplesofthisare67)and68):67)(Aisstandingbyanobviouslyimmobilizedcar)AIamoutofpetrol.BThereisagarageroundthecorner.Here,AwouldbejustifiedinassumingthatthegaragementionedbyBsellspetrol.Bwouldn’tbefollowingthemaximofrelationifherremarkaboutthegaragewerenotrelatedinthiswaytoA’ssituation.68)ASmithdoesn’tseemtohaveagirlfriendthesedays.BHehasbeenpayingalotofvisitstoNewYorklately.Again,themaximofrelationisinvolvedhere.AllthatBactuallysaysisthatSmithhasbeengoingtoNewYorkalot,butsheimplicatesthatSmithmayhaveagirlfriendthere.Ifnosuchimplicaturewereintended,B’sremarkwouldbeuncooperative,becauseirrelevant,asaresponsetoA’sleadingcomment.10Intheseexamples,theco-operativeprinciplecanbeusedtoexplainhowaresponsetoaquestionorcommentcanbeinterpretedasrelevant.ButGricedoesn’tassumethatthepictureisalwaysthissimple,orthatallconversationalexchangesareasstraightforwardastheseexamples.Healsoconsidersexamplesinwhichitislessclearthataparticularspeakerisbeing\nSpeakersandhearers135fullyco-operative.Thesetoo,heargues,canbeexplainedbymeansoftheco-operativeprinciple,butonlyifitisassumedthatspeakersmaysometimeschooseto‘flout’themaximsintheprocessofconversation.FloutingthemaximsIn‘Logicandconversation’,Griceallowsthatspeakerssometimesproduceconversationalcontributionswhichappeartobeextremelyuncooperative.Heconsidershowitisthathearersneverthelessmanagetointerprettheseascontributionstothecurrentconversation.Suchexamplesareexplainedintermsofthehearer’scontinuedassumptionthatthespeakerisbeingco-operative,despiteappearancestothecontrary.Gricedescribesexampleswhereconversationalmaximsareblatantlydisregarded,withnoapparentintentiontodeceive,asexamplesof‘flouting’amaxim.Thespeakermightbesaidtobeexploitingaparticularmaximforcommunicativeeffect.Gricenotesthat‘Thissituationisonethatcharacteristicallygivesrisetoaconversationalimplicature’(30).BelowaresomeofGrice’sexamplesofthisphenomenon.Ineachcasethemaximisinfringedonlyatthelevelofwhatissaid.Atthelevelofwhatisimplicatedthesecanbeinterpretedasco-operativecontributions.69)Wariswar.Rememberthatwediscussedthisearlierasanexampleofablatanttautology,whichcanneverthelessbeusedtoconveysignificantideas.InGrice’saccount,suchexamplesare‘totallynoninformative’atthelevelofwhatissaid,failingtoimpartanyinformation.Assuch,theyclearlyfloutthefirstsubmaximofquantity.However,ifthehearermaintainstheassumptionthatthespeakerisbeingco-operative,hewillassumethatsomeinformationisprovidedatthelevelofwhatisimplicated.Exactlywhatthisinformationiswillofcoursedependoncontext.Itcouldbethatthespeakerissuggestingthat,onceastateofwarhasbeendeclared,noparticularrulescanbesaidtoapply.Orshemaybesuggestingthattherelationshipbetweentwopeopleunderdiscussionisaveryhostileone,andthatthisexplainstheirbehaviour.Therearecertainother‘figuresofspeech’whichGriceexplainsintermsofthedistinctionbetween‘whatissaid’and‘whatisimplicated’,notablyironyandmetaphor.Hisexamplesofthesetwophenomenaareasfollows:70)[InasituationwhereXhasbetrayedasecretofthespeaker.]Xisafinefriend.71)YouarethegininmyMartini.\n136SpeakersandhearersAtthelevelofwhatissaidwecouldsaythatboththeseexamplesaresimplyfalse.Theyaresaidinapparentdisregardofthemaximofquality.Butontheassumptionthatthespeakerisn’tsimplystatingfalsehoodsuncooperatively,theseexamplescanbereanalysedasconveyingsomethingsignificant,andsomethingpossiblytrue,atthelevelofwhatisimplicated.Inthecaseoftheironic70),Gricesuggeststhatthe‘mostobviouslyrelatedproposition’isthecontradictionofwhatissaid,andthisistakentobewhatisimplicated.Thisisnotthecaseformetaphorssuchas71),wherethecontradictionofwhatissaidisanuninformativetruism.Rather,thespeakeristakentobeimplicatingthattherearecertainfeaturesofgininMartiniwhichcanbeattributedtothesubject.Remembertheexampleoftheabruptchangeoftopic,usedtoindicatethatwhathadjustbeensaidwasinsomewayinappropriate.Griceexplainsthisasanexampleofafloutingofthemaximofrelation,amaximwhichhesuggestsisperhapsnotfloutedascommonlyastheothersare.Heasksustoconsidertheexchange,repeatedhere,astakingplace‘atagenteelteaparty’(35).72)AMrsXisanoldbag.BTheweatherhasbeenquitedelightfulthissummer,hasn’tit?BproducesaresponsewhichisclearlyandblatantlynotrelevanttowhatAhasjustsaid.OntheassumptionthatBisbeingco-operative,andnotmerelyproducingrandomcomments,AunderstandsthatBisimplicating,althoughnotofcoursesaying,thatthisisanunsuitablecomment.Conversationalimplicatureswhicharisefromanapparentdisregardforthemaximofmannerare,ofcourse,dependentontheformofexpressionwhichthespeakerchooses.Griceinfactsuggeststhatthisdistinguishesthemfromotherimplicatures,becauseitmeansthattheycanbe‘detached’;theywon’t,orwon’tnecessarily,ariseifthespeakerrephrasesherconversationalcontribution.Toillustratethistypeofimplicature,wecanreturntoanotheroftheproblematicexampleswhichweconsideredearlierinthischapter,thatofdoublenegation.Rememberwesawthat,ifnegationinnaturallanguageweretocorrespondexactlytologicalnegation,73)shouldbeindistinguishablefrom74)intermsofmeaning.Yetsomeonewhosaid73)wouldnotbetakentohavepaidthesamecomplimentassomeonewhosaid74).73)Ifoundyourlecturenotunhelpful.74)Ifoundyourlecturehelpful.Bychoosing73)ratherthanthesimpler,andlogicallyequivalent74),thespeakermust,ifsheisbeingco-operative,beintendingsomeparticularmeaningtobeassociatedwiththismodeofexpression.Specifically,sheisapparentlyfloutingthesubmaximofmannerwhichstates‘bebrief’;74)is,\nSpeakersandhearers137quitesimply,ashorterformofexpressionthan73).Sheistakentoimplicatethatthebriefer,andthereforeapparentlymoreco-operative74)isnotappropriateinthecontext.Thereforesheistakentoimplicate,althoughnotstrictlytosay,thatthelecturewasnotentirelyhelpful.ThetwopapersinwhichGricedevelopshisaccountofconversationalimplicature,‘Logicandconversation’and‘Furthernotesonlogicandconversation’,bothreadas‘workinprogress’.Heacknowledgesthatsomeofhisdefinitionsneedclarification,forinstancethatof‘whatissaid’,andthatfurtherthoughtneedstobegiventoissuessuchastherelationshipsbetweenthevariousmaxims.Thiswasworktowhichheneverreallyreturned,althoughaswewillseeinthenextsection,ithasbeentakenupbyothers.However,Grice’sworkhasbeenhugelyinfluential,withitsfocusonconversationasaninteractive,negotiatedtask,anditsattempttocharacterisealimitednumberofprinciplestoexplainawidevarietyofcommunicativeeffects.Thisinfluencecanbeseenparticularlyinlinguisticpragmatics,whichwewillconsiderbrieflytoconcludethischapter.PragmaticsThephilosophicalaccountsofmeaningwehaveconsideredinthischapterdifferinanumberofimportantways.Buttheyareallbasedontheideathatwhatiscommunicatedbetweenspeakerandhearercanbefullyexplainedonlywithreferencetocontext,anideawhichhasbeenfundamentalinthedevelopmentofpragmatics.Pragmaticsissometimesdefinedasbeingthestudyof‘languageinuse’or‘languageincontext’.Itisconcernedcentrallywithmeaningassomethingwhichiscommunicatedbetweenusersofalanguage,andonlyindirectlywithmeaningassomethingwhichisspecifiedbylinguisticrule.Pragmaticsisoftencontrastedwithsemantics,thestudyofpurelylinguisticmeaning.Semanticsissometimesdescribedasbeingconcernedwiththemeaningofsentences,entitieswhichhavenophysicalformbutaredefinedbythegrammarofthelanguageinquestion.Pragmatics,ontheotherhand,isconcernedwithutterances,specificusesofsentencesinspeechorwritingonparticularoccasions.ThisdistinctioncanclearlyberelatedtoStrawson’sdiscussionofthedifferencebetweena‘sentence’anda‘useofasentence’,andAustin’sargumentthatcertainconditionsofcontextmustbeinplaceforanact,oruseoflanguage,tobe‘happy’.Austin’sworkinthisareahashadaprofoundandcontinuinginfluenceonpragmatics.Therearestilllinguistsworkingwithinabroadly‘speechact’framework,inwhichutterancesarediscussedintermsoftheirillocutionaryforce.Inparticular,therehasbeenalotofinterestinthenotionofindirectspeechacts.Thediscussionofthesehasbeendeveloped,alonglinessuggestedbySearle,intermsof‘politeness’.Inpolitenesstheory,indirectnessisdescribedasameansofindicatingdeferenceandminimising\n138Speakersandhearersimposition,orofbalancingthedemandsof‘face’oftheparticipantsinaconversation.Theuseofdirectandindirectformsofaddresshasbeenstudiedasonewayinwhichthebalanceofpowerisestablished,maintainedorchallengedinvarioustypesofinstitutionalandinformaldiscourse.Theconceptofspeechactshasalsobeenemployedinstudiesoflanguageacquisition.Ithasbeensuggested,forinstance,thatanunderstandingofsomeofthedifferentfunctionsofspeechactsisessentialtotheacquisitionoflanguage,andmustinfactbeattainedbyachildbeforegrammaticalcompetencecandevelop.Therehasbeenonenotableattemptinlinguisticstotackletheproblem,ofwhichAustinwasaware,ofthevarietyofgrammaticalformswhichperformativescantake.RememberthatAustinwaseventuallyforcedtoabandonthenotionthatitwouldbepossibletodrawupalistofperformativeverbs,whenherealisedthatmanyutterances,includingdeclarationsoffact,couldbeseenas‘implicitperformatives’.Aswellasanexplicitperformativesuchas75),a‘hint’suchas76),andevenastatementsuchas77)couldbedescribedasperformative.75)IbetyouthatNewcastlewillwintheFAcup.76)Iwouldn’ttroublethevicechancellorrightnowifIwereyou.77)ThetrainnowatplatformthreeisforDoverPriory.Austinnoticedthatitwouldbepossibletoparaphraseanimplicitperformativesuchas76)as‘I(hereby)warnyounottotroublethevicechancellorrightnow’,andastatementsuchas77)as‘I(hereby)informyouthatthetrainnowatplatformthreeisforDoverPriory’.Austinhimselfmerelynotedthediscrepancieswhicharepossiblebetweentheformofanutteranceandthespeechactitcanbeusedtoperform.However,withintheframeworkofgenerativesemanticsinthe1970s,anattemptwasmadetoexplainimplicitperformativesatthelevelofdeepstructure.Usingtheconventionalabbreviations‘Vp’for‘verbphrase’and‘S’forsentence,the‘performativeanalysis’,or‘performativehypothesis’,stipulatedthateverysentencesharedthefollowingformatthe‘highest’levelofdeepstructure:78)I(hereby)Vpyou(that)S.Inotherwords,the‘underlying’formof76)and77)wouldbe79)and80),respectively:79)I(hereby)warnyou(that)youdon’ttroublethevicechancellorrightnow.80)I(hereby)tellyou(that)thetrainnowonplatformthreeisforDoverPriory.Inafinaltransformationbeforesurfacestructure,these‘performative’parts\nSpeakersandhearers139couldthenbe(optionally)deleted,explaininghow76)and77)couldbeperformativeinmeaningwithoutbeingperformativeinform.Generativesemanticistssupportedthisanalysiswithcomplexsyntacticarguments,claimingthatitcapturedanumberofimportantgeneralisationswhichcouldn’totherwisebeexplained.Italsofittedinwiththeirgeneralapproachtomeaning,whichwasthatallaspectsof‘surface’meaningshouldbeexplainedintermsofgrammaticalstructureatsomelevel,andthereforebelinguistic,orsemanticinnature.Theirclaimwasthatillocutionaryforce,ratherthanbeingdependentoncontextandonindividualspeakers,couldbeexplainedintermsofsemantics.However,inthelate1970sand1980s,thegrowinginterestinpragmaticsasseparatefrombutrelatedtosemantics,ledtoaccountsofthistype,basedentirelyonform,beinglargelyabandonedinfavourofthosewhichconcentrateontheinteractionofformandcontext.LinguistshavealsodrawnonGrice’sdistinctionbetweenwhatissaidandwhatisimplicated,andhisideathatthereshouldbecertainprinciplestoconnectthetwo.‘Griceanpragmatics’hasbeenusedasaframeworkforinvestigatingthelanguageofliterature,ofthecourtroom,andofjokes,amongmanyotherapplications.Grice’sbasic,rathersketchyaccountofconversationalimplicaturehasbeendevelopedbylinguistsinvariousways.Forinstance,LaurenceHornhassuggestedthatthemaximofquantitycanexplainwhyitisthatvarioustermsinEnglish,whichcanbearrangedsemanticallyon‘scalesofinformativeness’,seemtoimplicatethedenialofothertermsonthesamescale.StephenLevinsonhassuggestedthatGrice’smaximscanbereducedtoadistinctionbetweenso-called‘Q’and‘I’maxims.Perhapsthemostsignificantnewworkinthisareasince‘Logicandconversation’hasbeenthedevelopmentof‘relevancetheory’byDanSperberandDeirdreWilson,andothers.ProponentsofrelevancetheoryseethemselvesasbuildingonGrice’sinsightsaboutcommunication,whileabandoningthenotionthattherearecertain‘maxims’whichspeakers‘follow’.TheyreassessGrice’sproblematicmaximofrelationasageneralprincipleofhumancognition,the‘principleofrelevance’.Humanbeings,theyclaim,arenaturallyinclinedtosearchforthemostrelevantinterpretationofanystimulustheyarepresentedwith.Asaresultspeakersconstruct,andhearersinterpret,utterancesinsuchawayastomaximisetheirrelevanceintheirparticularcontext.SperberandWilsonareexplicitintheirclaimthattheiraccountofcommunicationdependsontheinteractionof‘code’with‘inference’.However,somelinguistsworkingintheareaoflanguageusehavecriticisedtheverynatureoftheapproachestakenbyAustinandbyGrice.Theircriticismshavelargelybeenaimedatthemannerinwhichthephilosophersdescribeandexplaincommunicativeprocesses,andraisesomeinterestingquestionsaboutwhatitiswearelookingforinatheoryoflanguageuse.Basically,ithasbeenarguedthatAustinandGriceboth\n140Speakersandhearersconcentrateexclusivelyonshort,constructedexamplesofoneortwo‘utterances’.Theirworkdoesn’ttakeaccountof,andisinfactinadequatetoexplain,whatgoesoninrealexchangesinvolvingactuallanguageusers.Speechacttheoryinparticularhasbeenseenasunabletoexplainlongerstretchesofdiscourse.Ithasbeenpointedoutthatseveralturnsintheconversationmaymakeupasinglespeechact.Austiniscriticisedforwriting,despitehisinsistenceonthesignificanceofcontext,asifutteranceswereproducedmoreorlessinisolation.Similarly,Gricehasbeencriticisedforconcentratingonshortexchanges,andforprovidingonlyabriefstatement,orasingleprecedingutterance,ascontext.Linguistsworkingwithreallanguagedata,forinstancewithrecordingsandtranscriptionsofactuallyoccurringconversations,havesuggestedthatthecontextofanyutterancewillalmostalwaysbetoocomplextoanalyseordescribefully.TherearetwodistinctresponsestotheworkofAustinandGriceinthelightofthisconclusion.Oneistoaccepttheiraccountsassimplificationswhicharenecessaryinordertobeabletosayanythingsystematicaboutlanguageuse.Thisapproach,adoptedin‘discourseanalysis’,isratherlikethatofthescientistwho,aswesawinChapter3fromCarnap’sanalogy,studiescontrolledexperimentsinordertotrytounderstandthenaturalworld.ThealternativeapproachistorejectaccountssuchasAustin’sandGrice’sasunjustifiablesimplifications,whichcantellusnothingaboutlanguageuseinthe‘realworld’.‘Conversationalanalysis’,whichfollowsthisschoolofthought,describeslanguageuseasaseriesofinteractionalsequences,whichcanonlybeanalysedasawhole.Suchapproachessuggestratherdifferentanswerstothequestionofwhatisrequiredfromanaccountoflanguageuse,althoughtheybothusereallanguagedataastheirbenchmarkfortestingsuchaccounts.Thefirstsuggeststhatatheoryoflanguageuseisvalidtotheextentthatitadequatelymodelsnaturallanguageexchanges,givenasuitablyrestrictedrangeofcontextualfactors.Thesecondsuggeststhatthemostsuccessfulapproachisonewhichexplainsthewaysinwhichactual,individualexchangesarestructuredandsequenced.Yetanotheranswertothequestionisthatputforwardbypsycholinguistsworkingonlanguageprocessing.Griceprefiguredlaterworkinlinguisticsbyemphasisingtheimportanceofthehearer’sinterpretationtotheconstructionofmeaning.Morerecentworkhasfocusedonthisinterpretationasaseriesofcognitiveprocesses,whichcantellussomethingaboutthewaysinwhichlanguageisinterpreted,andalsoperhapssomethingabouthowthemindworks.Thisrelatestoagrowinginterest,inlinguisticsandinotherareasofstudy,inthenatureandoperationofthemind.Thestudyofthemind,andinparticularofhowitrelatestolanguage,isthesubjectofthenextchapter.\nSpeakersandhearers141FurtherreadingWittgensteinAgoodselectionfromWittgenstein’swritings,includingextractsfrombothTractatusandthelaterwork,canbefoundinKenny(1994)TheWittgensteinReader.ManywritershaveofferedinterpretationsofWittgenstein’swork.OneoftheseistheOxfordphilosopherMichaelDummett,whohasdiscussedanddevelopedWittgenstein’sworkinhiswriting.Anumberofhispapersarecollectedinhis(1993)bookTheSeasofLanguage(thetitleitselfisaquotationfromPhilosophicalInvestigations).Inparticular,hediscussesthelaterWittgenstein’snotionof‘languagegames’intheessay‘Languageandtruth’.In‘Whatdoestheappealtousedoforthetheoryofmeaning?’,heconsiderssomeofthewaysinwhichWittgensteinagreeswith,anddiffersfrom,Frege.Wittgenstein’sapparentinsistenceinhislaterwork,thatmeaningcanbeexplainedonlyintermsofuse,prefiguresadebateinlinguisticsastowhetheranythingatallcanbesaidaboutmeaningawayfromcontext.Thisissueisnotdiscussedhere,butseeforinstanceHarris(1996a)TheLanguageConnection:PhilosophyandLinguistics.OrdinarylanguagephilosophyThetwoarticlesbyGilbertRylereferredtoherearebothveryreadable.Assuggested,‘Ordinarylanguage’canbeseenalmostasamanifestoforthisparticularschoolofphilosophy.Itconsistsofaseriesofdefinitionsofkeytermssuchasordinary,use,usage,etc.‘Thetheoryofmeaning’presentsasurveyofthephilosophicalbackground,lookingparticularlyatMill,FregeandWittgenstein.BothofthesearticlesbyRylearereprintedinCaton(1963)PhilosophyandOrdinaryLanguage,ausefulcollectionofessaysinthefield,includingsomelesswell-knownones.ReferencestoRyle’sessaysinthischapteraretothisedition.PaulGrice’saccountanddefenceofordinarylanguagephilosophy,‘PostwarOxfordphilosophy’waswrittenin1958andappearsinhis(1989)StudiesintheWayofWords.AcontemporarycritiqueofthewholeapproachtophilosophyisErnestGellner(1979)WordsandThings.SubtitledAnExaminationof,andanAttackon,LinguisticPhilosophy,thisamountsattimestoapolemicalcriticism,particularlyoftheworkofWittgensteinandAustin.ItissupportedinaforewordbyBertrandRussell.SpeechactsAustin’s1955HarvardlectureswerepublishedposthumouslyasHowtodoThingswithWords(1962).Thisisreadableandlively,andaccessibleeven\n142Speakersandhearersasanintroductiontothesubject.Theessentialsofhisaccountofperformatives,andofhisrealisationthattheycouldnotbeclearlydistinguishedfromstatements,areinhispaper‘Performativeutterances’(1961).Searle’scritiqueanddevelopmentofAustin’sideas,andthewiderimplicationsheseesinthemforthephilosophyoflanguage,aresetoutinSpeechActs(1969).Themainpointsofhisaccountarealsopresentedin‘Whatisaspeechact?’(1965)and‘Indirectspeechacts’(1975).AnotherinterestingcommentaryonAustin’swork,whichreconsidersperformativesinthelightofworkongrammaticaltransformations,andthereforetiesinwithlaterworkontheperformativehypothesis,isVendler(1972)ResCogitans.MeaningandintentionClark(1987)PrinciplesofSemioticsisasuccinctsurveyofworkonnaturalandnon-naturalmeaning.Inparticular,Chapter2tracesitsoriginsinclassicalphilosophy’sdiscussionof‘signs’.Thisaccountalsofillsinsomeofthegapswehaveleft,bylookingatthephilosophicaltraditionoftheMiddleAges,andcontinuesuptothebehaviourismofthetwentiethcentury,whichwewillconsiderinthefinalchapter.Chapter4isconcernedwiththeroleofsignsincommunication,andincludesadiscussionofGrice’swork.ConversationalimplicatureThediscrepancybetweenlogicandnaturallanguageisdiscussedinAllwoodetal.(1979)LogicinLinguisticsand,ingreaterdetailinMcCawley(1981)EverythingthatLinguistsHaveAlwaysWantedtoKnowaboutLogic.Inhischapteron‘Speechactsandimplicatures’,McCawleyexplainshowGrice’sco-operativeprincipleattemptstoaccountforthesediscrepancies.MostofthesalientpointsofGrice’saccountofconversationalimplicaturearesetoutsuccinctlyinhisarticle‘Logicandconversation’(1975).Theyaredevelopedin‘Furthernotesonlogicandconversation’(1978).Thereisagreatdealofexplanatoryandcriticalworkavailableonthissubject.Muchofthishasbeenproducedbylinguistsworkinginpragmatics,anditisthereforelistedinthenextsection.PragmaticsTherearealargenumberofintroductorybooksonpragmaticsavailable,mostofwhichincludesomediscussionoftheworkofAustin,Searle,andGrice,andtheircontributionstolinguistics.Someofthemostpopularwithstudentsoflinguisticsare:Lyons(1981)Language,Meaningand\nSpeakersandhearers143Context,Leech(1983)PrinciplesofPragmatics,Levinson(1983)PragmaticsandThomas(1995)MeaninginInteraction.Bates(1976)LanguageandContext:TheAcquisitionofPragmatics,suggeststhattheacquisitionofspeechactsisnecessarilypriortothatofgrammar.The‘performativehypothesis’isproposedinanumberofworksfromthe1970s,notablySadock’s(1974)TowardsaLinguisticTheoryofSpeechActs.Themainargumentsfromthis,andarefutationofthem,areofferedinChapter5ofLevinson’s(1983)Pragmatics.ThemajorworkonthedevelopmentofthestudyofindirectspeechactsintoanaccountofpolitenessinconversationalinteractionisBrownandLevinson’s(1978)article‘Universalsinlanguageuseage:politenessphenomena’.LarryHorn’sdevelopmentofGriceanimplicaturesofquantityintoscalarimplicatureshasnotbeenpublishedinbookform;itappearedinhis(1972)Ph.D.thesis‘OnthesemanticpropertiesofthelogicaloperatorsinEnglish’.ThefundamentalaspectsofthisarediscussedinChapter3ofLevinson(1983)Pragmatics.Horn’sstudyoftheinteractionofnegationwiththistypeofimplicatureisinhisarticle‘Metalinguisticnegationandpragmaticambiguity’(1985),anextendedversionofwhichappearsinhis(1989)ANaturalHistoryofNegation.SperberandWilson’scritiqueofGrice,andtheiralternativeaccountofcommunicationintermsofasingleprincipleofrelevance,issetoutinRelevance(1995).AusefulcommentaryonthisisBlakemore(1994)UnderstandingUtterances.Thereisahugeliteratureontheempiricalstudyofreallanguagedata.GoodintroductorytextsincludeBrownandYule(1989)DiscourseAnalysisandSchiffrin(1984)ApproachestoDiscourse.Bothincludediscussionofthedifferentmethodsofpragmatics,discourseanalysisandconversationalanalysis.ThelatterworkoffersadetailedanalysisofsomenaturallyoccurringlanguagedatausingGriceanimplicature,andtherebyhighlightssomeoftheproblemswhichsuchdatapose.\n5LanguageandmindIntroductionInturningourattentiontotherelationshipbetweenlanguageandmindweare,inonesense,embarkingonawholenewtopic.Wewillbecomparingaccountswhichdescribelanguageasatypeofknowledgewiththosewhichdescribeitasatypeofbehaviour,andwewillbelookingatwhattheseaccountssuggestabouthowlanguageislearnt,usedandunderstood.Inanothersense,however,wehavebeenconcernedwithlanguageandmindallalong.So,forinstance,inChapter1whenwelookedatwhathasbeensaidabouttherelationshipbetweenwordsandobjects,weconsideredtheideathatwordsstandnotforobjectsthemselvesbutfor‘mentalimages’oftheobjectsinthemind.AndinChapter4weexaminedaccountsofcommunicationwhichdependonthe‘intention’inthemindofthespeaker,andtherecognitionofthatintentionbythehearer.Thediscussionofmanyaspectsoflanguagesoonerorlaterinvolvesadiscussionofthemindorofmentalprocesses,becauselanguageissointimatelyboundupwiththewayweunderstand,describe,andaffecttheworldaroundus.Wewillbeconcentratingontwomainissueshere.1Principally,wewillbelookingatthewaysinwhichdescriptionsoflanguagehaveeitherdependedon,oralternativelyruledout,therelevanceofpropertiesofthehumanmind.Inotherwordswewillbecomparingworkwhichseeslanguageassolely,oratleastmostimportantly,amentalphenomenon,withworkwhichexcludesallreferencetothemindfromthedescriptionoflanguage.Thesecondmainissueconcernsthenatureoftheprocessesinvolvedintheacquisitionoflanguagebychildren.Aswewillsee,thisiscloselyrelatedtothemoregeneralissueoftherelationshipbetweenlanguageandmind.Inessence,itcentresonquestionsofwhetherchildrengothroughaprocessofacquiringcertaintypesofknowledge,oroflearningcertainformsofbehaviour.WewillbeginwiththeSwisslinguistFerdinanddeSaussure,whoseworkweconsideredbrieflyinChapter4.WehavealreadyseenthatSaussuredescribedtherelationshipbetweenanobjectanditsnameasarbitraryandconventional,andinthischapterwewilllookinmoredetail144\nLanguageandmind145atSaussure’sdefinitionsofobjectsandnames.Wewillalsoconsiderwhathesaidabouttherelationshipsbetweenwords.Briefly,hearguedthattheserelationshipsare,incontrast,notarbitrarybuthighlystructured,andthatitisthesestructureswhichmakeupthelanguagewespeak.Saussureclaimedthatweneedtolearnalanguagebeforewecanentertainfullydevelopedthoughts.Soourlanguagedeterminesthewayourideasabouttheworldarestructured,whetherornotweexpresstheseideastootherpeople.Wewillseethatsomelaterlinguistsarguedthatspeakersofdifferentlanguagesactuallyperceivetheworldaroundthemdifferently.Thewordsandformsofexpressionwelearndetermine,butalsolimit,thewaysinwhichwecandescribethethingsandeventsweseetootherpeople,andalsotoourselves.Ourinvestigationswillthenbringusbacktoatopicwhichweconsideredinthefirstchapter:theinnatepropertiesofthehumanmind.Wehaveseenthatdiscussionofwhatcanbesaidtobeinnateinthehumanminddatesrightbacktoclassicaltimes.Thediscussioncentresonthequestionofhowmuchofourunderstanding,andindeedknowledge,isinherited,andhowmuchmustbelearntfromourenvironment.Todescribesomethingasinnateistoclaimthatitisanessentialpropertyofhumannature;todescribesomethingaslearntistoclaimthatitisdependentonexperience.RememberthatPlato’saccountofIdealForms,whichareknownbeforeanyexperienceofactualobjects,canbeseenasanearlymodelofinnateknowledge.Aristotle,ontheotherhand,tooktheviewthatknowledgeoftheworldisderiveddirectlyfromtheimpressionswereceivefromoursenses.Thesetwoapproachesleadtoverydifferentaccountsofwhatweare‘bornwith’.Oneconsequenceofinnatismisthatanewbornchildmustbedescribedashavingcertaintypesofknowledge(forPlato,knowledgeoftheIdealForms)alreadyinplace.Incontrast,Aristotle’sversionisoftentakentoimplythatthemindofthenewbornchildisatabularasa.ThisLatinphrasetranslatesas‘blanktablet’or‘cleanslate’;thechildisbornwithoutknowledge,whichmustbe‘written’intoitsmindbyexperience.Bothofthesetheoriesofmindhavefeaturedinmorerecentdiscussionsofhowitisthatchildrenlearn,oracquire,thelanguagebeingspokenaroundthem.Wewilllookatsomeverydifferentaccountsoftheprocessoflanguageacquisition,andconsiderhowtheystemfromthesetwopicturesofthenewbornmind.WesawinChapter1thattheideasofbothPlatoandAristotlere-appearedintheworkofphilosophersoflatercenturies.Plato’sinterestininnateknowledgeisreflectedintheworkofrationalistssuchasLeibniz.Rationalistsarguedthatdatafromthesensesisnotenoughtogiveknowledgeoftheworld;theremustbeacertaindispositionofmindwhichdeterminestheformofknowledgewhichcanbeacquired.Inalatercentury,Kantclaimedthatcertaintypesofknowledgeareapriori;theydon’tderivefrom,butratherprecede,experience.Ontheotherhand,\n146LanguageandmindAristotle’sinfluencecanbeseeninparticularintheworkofempiricistssuchasLockeandBerkeley.Wehaveseenthat,whileLockeallowedthatsomebasiccognitivecapacitymustbeinnate,hemaintainedthatallknowledgemustbederivedfromexperience.Thetermempiricisthasalsobeenapplied,inaslightlydifferentbutrelatedsense,toanapproachtothestudyoflanguagewhichdevelopedintheearlyandmiddlepartofthetwentiethcentury.Althoughmuchcriticised,thisapproachhasbeenveryinfluential,andisstillcurrentinvariousforms.Itisalsogenerallyacknowledged,evenbythosewhodisagreewithit,ashavingbeenveryimportantintheprocessofestablishinglinguisticsasaseparatebranchofacademicstudy.Interestininnatepropertiesofthemindcertainlydidn’tdieout,andindeeditwasarevivaloftheseideaswhichledtothedevelopmentofanimportantalternativetoempiricisminthemiddleofthetwentiethcentury.Wewillbeconsideringthisrevivallater.ButbeforethatwewillreturntoSaussure,andlookathowhisideasinfluencedthestudyoflanguageearlierinthecentury.SignsandstructuresFerdinanddeSaussurewasoneofthefirstpeopletobaseanacademiccareeronthestudyoflanguageforitsownsake.BorninSwitzerlandin1857,hebecameaprofessorattheUniversityofGenevain1906.Hisinterestsincludedthedetaileddescriptionofindividuallanguages,aswellasthehistoryandnatureoflanguageitself.Hisworkwasextremelyinfluentialinthesubsequentdevelopmentoflinguistics,buthepublishedverylittleduringhislifetime.Afterhisdeathin1913,anumberofhisstudentscompiledsomeofthelectureshehadgivenatGenevafromtheirnotesandfromhisowndrafts.ThesewerepublishedasCoursdeLinguistiqueGénéralein1916,whichwastranslatedintoEnglishandpublishedasCourseinGeneralLinguisticsin1960.Intheselectures,Saussurerejectstheextensionalaccountofmeaning:theidea,whichweconsideredinChapter1,thatlanguageconsistsofaseriesofnamesreferringdirectlytoobjects.Hearguesthatlanguagecanbesaidtoconsistofaseriesof‘signs’inwhichtwodifferententitiesareunited,butthat‘thelinguisticsignunites,notathingandaname,butaconceptandasound-image’(66).Saussurestressesthatboththeseentities,andthelinkbetweenthem,existprimarilyinthemindofthespeakersofthelanguage.The‘concept’,likethe‘ideas’inideationalaccountsofmeaning,canbethoughtofasamentalimageofsomeactualobject(Saussure’sexamplesareoftheconceptsof‘horse’and‘tree’).The‘sound-image’isnottheactualsoundsformedwhenyoupronouncethewordsassociatedwiththesethings,butthe‘psychologicalimprintofthesound’(66).Thisissomethingyouareawareofifyoujustrecitethewordssilentlytoyourself.Thestudyoflanguage,then,isthestudyofoneaspectofthehumanmind:aseriesofmentalsoundimages,andtheconceptswehavelearnttoassociatewiththem.\nLanguageandmind147Saussureadoptsthetermssignifiedforaconceptandsignifierforthesound-imageassociatedwithit.Thesetwotogetherformasign,thebasiclinguisticunit.RememberthatwesawinChapter4thatSaussuredescribestherelationshipbetweensignifierandsignified,therelationshipwhichformsasignofthelanguage,asarbitrary.Thereisnorationalornaturallinkbetweenthetwopartsofthesign;theyareunitedonlybytheconventionsofindividuallanguages,anddifferwithoutlimitbetweenlanguages.Buttherelationshipwithinthesignisn’ttheonlytypeofrelationshipwhichthelinguistneedstotakeintoaccount.Therearealsoawholeseriesofrelationshipsbetweenthesignsofalanguage.Thesearecentraltothenaturenotjustofthelanguageasawhole,butofeachindividualsign,sinceeachsignissignificantonlyinhowitrelatestoand,crucially,differsfrom,theothersignsinthelanguage.InSaussure’sownwords,signsfunction‘notthroughtheirintrinsicvalue,butthroughtheirrelativeposition’(118);tofunctionefficientlyasignmustberecognisablydistinctfromallothersignsinthesystem.LanguageforSaussure,then,isamentalstructure.Butthisisonlypartofhisdefinition.Noaccountoflanguageiscomplete,heargues,withoutreferencetoitsfunctioninasociety;‘languageneverexistsapartfromthesocialfact…itssocialnatureisoneofitsinnercharacteristics’(77).Languagebecomesameansofcommunicationfortheindividualwhenasound-image,relatedwithinasigntoaconcept,isrealisedinactualsound.Thesoundmustthenbereceivedbyanotherspeakerofthesamelanguage,whoisabletorecognisethesoundimageandthereforetoarriveatthesamementalconcept.SoSaussurepresentsamodelofcommunicationofatypeweconsideredinChapter4;itisdependentontheencodinganddecodingofideas.Hecallsthepsychologicalcomponentofcommunication,thesystemofsigns,the‘language’(Saussure’soriginalFrenchterm,widelyusedinthisspecialsense,islangue).Thisissharedbyacommunityofspeakersbecausetheyhaveallacquired,‘notexactlyofcourse,butapproximately’(13)thesamesetofrelationshipsbetweensignifiedandsignifier.Thesystemwhichconstitutesthelangueiscontrastedwiththeprocessofproducingsoundsinordertocommunicate,whichSaussurelabels‘speaking’(orparole)andwhich‘isalwaysindividual’(13).Speaking,then,canonlybeperformedbyindividuals,deliberatelyandwithpurpose.Language,ontheotherhand,‘isnotcompleteinanyspeaker;itexistsperfectlyonlywithinacollectivity’(14).Alanguageisdependentonthecommunityofspeakerswhoshareit.Saussuresawlanguageasdifferingincomplexity,butnotinbasicnature,fromallsortsofothersystemsofsignssharedbygroupsofpeople.Hewasinterestedinthestudyofhumanuseofsignsingeneral,orthedisciplineofsemiology,ashelabelledit.Themethodhesuggestedforthestudyofallsystemsofsigns,likethatforlanguage,wasoneoflookingforthestructuralrelationshipsbetweenthesigns.Hisapproachwasthereforelabelledstructuralism,atermwhichcametobeappliedtoamajorschoolof\n148Languageandmindthoughtofthetwentiethcentury.Saussure’sgeneralinterestinsemiologyhadsomeimportantconsequencesforhisviewoflanguagewhich,aswewillseelaterinthischapter,distinguishitfromsomealternativeviewsfromlaterinthesamecentury.Inparticular,Saussuredescribedlanguagesasnotsharinganyparticularcharacteristics,otherthantheirlevelofcomplexity,whicharenotalsopresentinothersocialsystems.Asaresulttheycandifferfromeachotheralmostwithoutlimit.Oneaspectofstructuralismwhichbecameparticularlysignificantinlinguisticsderivedfromtheideathatthesignificance,orvalue,ofeverysign,isdependentonitsplaceinthegeneralstructure.Putsimply,themeaningofanindividualworddependsprimarilynotonsomethingexternaltothelanguagesystem,butonthesystemitself.Thesystemdeterminesthevaluesofitssigns,sotheindividual‘concepts’whichformpartofthesesignsarebroughtintobeingbythelanguage.Inthisway,thoughtitselfisactuallydependentonlanguage.Peoplehaveaccesstocertainconcepts,andthereforeseetheworldincertainways,becauseofthestructureofthelanguagetheyhavelearnt.Saussurearguesthattheconceptswithwhichwearefamiliarcan’tactuallyexistindependentlyof,orpriorto,ourlanguage:‘ifwordsstoodforpre-existingconcepts,theywouldallhaveexactequivalentsinmeaningfromonelanguagetothenext;butthisisnottrue’(116).ThisideawassubsequentlytakenupbyanumberofAmericanlinguists,mostnotablyEdwardSapirandBenjaminLeeWhorf.LinguisticdeterminismStructuralismissometimesdescribedashavingdevelopedintwodirections,verydifferentfromeachotherinfocusandmethod,butbothdrawingonSaussure’sideas.InEurope,interestbroadenedfromthestudyofsystemsofsignstothestudyoftheiruse,the‘texts’theyareusedtocreate,andtheeffectswhichthesetextshave.Semiotics,asitbecameknown,wasconcernedlargelywiththeanalysisandcriticismofliterarytexts,butalsomoregenerallywithstructuresoftheworldandofsociety.InAmerica,thefocuswasmorenarrowlylinguistic.Thistraditionwasalso,confusingly,labelledsemiotics,atermcoinedindependentlybyCharlesMorris,whoseworkweconsideredintheintroductiontothisbook.Itwasconcernedwiththedetailedstudyofthestructuresofindividuallanguages:theirsystemsofsyntax,phonologyandmeanings.Modernlinguisticscanbeseenasdeveloping,atleastinpart,fromthisbranchofstructuralism.This‘American’semioticsofthemiddlepartofthetwentiethcenturyincludedLeonardBloomfield’sempiricalstudiesoflanguage,whichwewillconsiderinthenextsection.ItalsoincludedthetheorydevelopedbySapirandWhorfwhichbecameknownaslinguisticdeterminism.ThiscanbasicallybeseenasadevelopmentofSaussure’sideasaboutthe\nLanguageandmind149relationshipbetweenlanguageandthought.RememberthatSaussurearguedthatlanguagesdifferintermsbothofthesignifierstheycontainandalsooftheideassignified.Sincetheactualconceptswhichareavailableforspeakerstoencodeandcommunicatearedeterminedbytheparticularlanguagetheyspeak,thoughtitselfmustbedependentonlanguage.Thetheoryoflinguisticdeterminismstatesthatthewayinwhichanindividualthinks,andindeedperceivestheworld,isdependentonhisorherlanguage;welearntoseetheworldinacertainwayaspartoftheprocessoflearningourlanguage.EdwardSapirwasbornin1884andheldpostsattheUniversityofChicago,andthenatYale,wherehebecameProfessorofAnthropologyin1931.HisanthropologicalinterestswerechieflyfocusedonNativeAmericantribes;hestudiedtheircultures,theirtraditionsand,principally,theirlanguages.Hewasimpressedbytheevidencethat,althoughthelanguageshestudiedoftendifferedinmanysignificantwaysfromeachother,andfromEnglish,allshowedsimilarlevelsofcomplexityandpowersofexpression.Thisobservationmaynotseemparticularlyremarkablenow,butatthetimeinwhichSapirwaswriting(hismajorwork,thebookLanguage,waspublishedin1921)itwasasignificantclaimtomake.Sapirisfollowingthecustomofhistimewhenhewritesof‘thelanguageofthesavage’(22),andhedoesn’thesitatetosuggestthatmoresophisticationandcivilisationistobefoundina‘higherculture’.Buthegoesontoarguethatsuchdifferencesinculturearen’treflectedinthecomplexityofthelanguage:‘bothsimpleandcomplextypesoflanguageofanindefinitenumberofvarietiesmaybefoundspokenatanydesiredlevelofculturaladvance’(219).ForSapir,then,differencesinculturearenotreflectedindifferencesinlanguage.Hedoes,however,maintainthathumanlanguagesdifferinremarkableways,intermsoftheconceptstheycontainandthereforethethoughtstheycanbeusedtoexpress.LikeSaussurebeforehim,hearguesthatknowledgeofalanguage,includingtheparticularconceptsitcanbeusedtosignify,isanecessaryprerequisitetothought.Insupportofthis,hearguesthattheintroduction,oratleastthefullunderstanding,ofanewconceptisdependentontherebeingawordtoexpressit:Thebirthofanewconceptisinvariablyforeshadowedbyamoreorlessstrainedorextendeduseofoldlinguisticmaterial….Assoonasthewordisathand,weinstinctivelyfeel,withsomethingofasighofrelief,thattheconceptisoursforthehandling.(17)Sapir’sideasabouttherelationshipbetweenlanguageandthoughtwereofparticularinteresttohisstudentBenjaminLeeWhorf.WhenWhorfenrolledonthecoursein‘AmericanIndianlinguistics’whichSapirsetuponarrivingatYale,hewasalreadydeeplyengagedinresearchofhisown\n150Languageandmindinthisarea.Hislinguisticworkwas,andremained,entirelyamatterofhisowninterest,pursuedasanendinitself.Heneversoughtacademicpositionsortitles,andinfactturneddownanumberofuniversitypostsofferedtohim.Allthetimethathewasresearchingandpublishinghisprolificandhighlyoriginalworkinlinguistics,hewasworkingfull-timeasaninspectorforafireinsurancecompany.Indeed,thereissomesuggestionthattheexhaustioncausedbythis‘doublelife’mayhavecontributedtohisearlydeathin1941attheageofforty-four.2Whorfpublishedanumberofarticlesduringhislifetime,buthisworkisnowbestknownfromLanguage,ThoughtandReality,aselectionofhiswritingspublishedposthumouslyin1956.SapirencouragedWhorftostudytheNativeAmericanlanguageHopi,andthisworkwastoproveparticularlyimportantinthedevelopmentofhisideas.WhorfnoticedthatthemeansavailablefortalkingabouteventsinHopiseemedtobecompletelydifferentfromthoseinEnglish.Inparticular,‘theHopilanguageisseentocontainnowords,grammaticalforms,constructionsorexpressionsthatreferdirectlytowhatwecall“time”’(57).This,heargues,doesn’treflectaweaknessinthelanguage,butratherillustratesthepointthat‘itispossibletohavedescriptionsoftheuniverse,allequallyvalid,thatdonotcontainourfamiliarcontrastsoftimeandspace’(58).Insteadofthesecontrasts,Hopihasanalternativedistinctionbetweenmanifested(roughly:allthatisorhasbeenaccessibletothesenses)andmanifesting(roughly:themental,includingthefuture),adistinctionwhichallowsforarichandcomplexdescriptionofreality,withouttheneedforgrammaticalformssuchasverbtenses.Whorfalsolookedatthewaysinwhichdifferentlanguagesdivideuptheworldbymeansofvocabulary.Famously,henotedthatEnglishhasoneword,snow,whichdoesthesamejobasanumberofdifferentwordsinEskimo:ToanEskimo,thisall-inclusivewordwouldbealmostunthinkable;hewouldsaythatfallingsnow,slushysnow,andsoon,aresensuouslyandoperationallydifferent,differentthingstocontendwith;heusesdifferentwordsforthemandforotherkindsofsnow.(216)Whorf’scontentionisthatdifferencesbetweenlanguagesofthesetypesactuallyresultindifferentwaysofperceivingandunderstandingtheworld.SpeakersofHopithinkabouteventsdifferentlyfromspeakersofEnglishbecausethelanguagetheyhavelearntdescribesthemdifferently.Similarly,Eskimosperceivewhatwewouldthinkofasvarioustypesofsnowascompletelyseparateentities,becausetheyhavelearnttocallthembydifferentnames.ItmaybethatwewouldfailaltogethertoseesomeofthedistincttypesofentityobvioustotheEskimo,becausewehavenotlearnttolabelthemdifferently.Thegeneraltheorybecameknownasthe‘principle\nLanguageandmind151oflinguisticrelativity’,orsometimesthe‘Sapir-Whorfhypothesis’.Itisbasedonaconceptoflinguisticdeterminism;thelanguagepeoplespeakissaidtodeterminehowtheyseetheworldaroundthem.Instructuralistterms,thestructureofalanguage(includingtherelationshipsbetweenitssigns)limitsthewaysinwhichitsspeakersunderstandrealityasstructured.Languageisanimportantfocusofstudypreciselybecauseitcanexplainworldviewsandsystemsofbelief.Whorfsumsthisupbyclaimingthattherealconcernoflinguisticsis‘tolightupthethickdarknessofthelanguage,andtherebyofmuchofthethought,theculture,andtheoutlookuponlifeofagivencommunity’(73).TheSapir-Whorfhypothesisexplainsthestructureofmindasbeing,atleastinpart,dependentonthestructureoflanguage.Thelanguagewelearndeterminesthewaywethinkabouttheworldandtherefore,tosomeextent,thewayourmindswork.Theirhypothesisisalsoanaccountwhich,likeSaussure’s,highlightsthedifferencesbetweenindividuallanguages.Assuchithasn’tfittedwellwithsomemorerecentdevelopmentsinlinguisticswhich,asweshallseelaterinthischapter,haveemphasisedthebasicandremarkablesimilaritiesbetweenlanguages.Inaddition,somedoubthasbeencastonthevalidityofWhorf’smorespecificclaims.Forinstance,hisassessmentoftheHopiconceptoftimehasbeenshowntobeunfounded.Andithasbeensuggestedthathesignificantlyoverestimatedthenumberofwordsfor‘snow’inEskimo,whileatthesametimeoverlookingthevarietyofexpressionalsofoundinEnglish(thinkofsleet,blizzard,avalanche,slush,etc.).Nevertheless,theSapir-Whorfhypothesiswasinitstimeanimportantdevelopmentinlinguistics,andhasremainedinfluentialinworkontheconnectionbetweenlanguageandthought.Inthenextsectionweturnourattentiontoanotherbranchofmid-twentieth-centurylinguistics,againonewhichhaditsoriginsinSaussure’sstructuralism,butthistimeonewhichemphasisedtheobservablephenomenaoflanguageuseoveranyputativerelationshipbetweenlanguageandthemind.EmpiricismPhilosopherssuchasLockeandBerkeleybecameknownas‘empiricists’becausetheyemphasisedtheimportanceofexperience,andoftheevidenceofthesenses.Theyexplainedallknowledgeasderivednotfrominbornunderstanding,butinresponsetowhatwemightcallthe‘data’availabletotheindividual.The‘empiricist’approachtothestudyoflanguagewhichdevelopedinthetwentiethcenturyalsoemphasisedtheevidenceofferedbydata.Dataofactuallanguageusewasheldtobetheonlylegitimatefocusofattentioninthestudyoflanguage.Thesenewempiricists,aswewillsee,rejectedmetaphysicalspeculationabout‘mentalstates’,andevenaboutabstract‘meaning’,infavourofconcentratingontheobservablephenomenaoflanguageuse:spokenandwrittenutterances.OnepioneerofthisnewstyleofempiricismwastheAmericanlinguistandprofessorattheUniversityofChicago,LeonardBloomfield.\n152LanguageandmindBloomfieldAswehaveseen,SapirandWhorfarguedthatifwedescribealanguageweareinevitablydescribingsomethingofthemindsofthepeoplewhospeakit.LeonardBloomfield,however,maintainedthatindescribinglanguagewecan’tclaimtobesayinganythingatallaboutmind.Throughouthiscareer,hewaskeentopromotethescientificstudyoflanguageasafieldofenquiryinitsownright.By‘scientific’hemeantthat,likeanynaturalscience,thestudyoflanguageshouldbebasedonrigorousobservationandanalysis.AtthestartofhisbookLanguage,publishedin1933,helamentsthat:Theeffectsoflanguageareremarkable,andincludemuchofwhatdistinguishesmanfromtheanimals,butlanguagehasnoplaceinoureducationalprogramorinthespeculationofourphilosophers’(3).Thecontemporarytraditionoflanguageteachingwasprescriptive,withanemphasisonprovidingrulesandcorrectmodelsofhowlanguageshouldbeused.Bloomfield’saimwastopromotetheteachingofwhatisnowknownasdescriptivelinguistics,inwhichlinguistsrecordanddescribelanguageasitisactuallyused.Bloomfieldsetsouthisagendaforthescientificstudyoflanguagesuccinctlywhenheclaimsthat:Theonlyusefulgeneralizationsaboutlanguageareinductivegeneralizations’(20).Bythishemeansthattheonlylegitimatemethodistoobservelanguagedataanddrawconclusionsfromtheseobservations.Theinductivescientificmethodisonewhichlooksatarangeoftheavailableempiricalevidence,andarrivesatgeneralconclusionswhicharelicensedbythese.Incontrast,thedeductivemethodtakesageneralidea,orhypothesis,asitsstartingpoint,andlookstothedatatoeitherconfirmorfalsifythis.Aswewillseelater,bothmethodshavebeen,andcontinuetobe,usedinlinguistics.Bloomfield,then,allowsonlygeneralisationsdirectlylicensedbyobservationoflanguagedataandofhumanbehaviour.Suchgeneralisationscanonlybeaboutlanguageandbehaviour;it’snotlegitimatetousesuchdataasevidenceformentalstates,whicharen’taccessibletoempiricalstudy,Bloomfieldexplicitlyrejectsthementalisticapproachtolanguage,anapproachwhichmaintainsthatafullaccountofthedatacanbegivenonlyifreferenceismadetothe‘underlying’factsofmind.InBloomfield’sownterms,amentalisticaccountoflanguagewouldexplainitintermsof‘theinterferenceofsomenon-physicalfactor,aspiritorwillormind…thatispresentineveryhumanbeing(32,originalemphasis).Incontrast,Bloomfieldadvocatesamaterialisticaccountofhumanbehaviour,includinglinguisticbehaviour.3Suchanaccountdescribestheobservablefactsofbehaviour,andexplainsthemwithoutreferencetomentalstates.So,forinstance,heoffersamaterialisticaccountofthemeaningofanexpressionwhenheclaimsthatitismadeupof‘thesituationinwhichthespeakeruttersitandtheresponsewhichitcallsforthinthehearer’(139).Indiscussingsemantics,Bloomfieldargues,itisn’tlegitimate\nLanguageandmind153totalkabout‘underlyingmeaning’or‘linguisticknowledge’.Rather,thelinguistmustcollectinformationfrommanyobservationsofutteranceswhichshareaphoneticform,andgeneralisefromtheseobservationstothecharacteristiccontextsandconsequencesofthatform.Totakeaverysimpleexample,ifthelinguistobservesthewordapplebeingutteredonanumberofdifferentoccasions,eachtimeinthepresenceofapieceoffruitwhich‘presentscertainrelativelydefinablecharacteristics’(141),thelinguistmayconcludethatappleisthenameofthatclassoffruitinthelanguageunderobservation.Indevelopingthisaccountofmeaning,Bloomfieldwasinfactapplyingthecontemporaryschoolofthoughtknownasbehaviourismtothestudyoflanguage.ThishaddevelopedinAmericanpsychologysometwentyyearsbeforethepublicationofLanguage,althoughitsimplicationsforlinguisticsarenowperhapsbestknownfromthe1957bookVerbalBehaviourbytheAmericanpsychologistB.F.Skinner.Behaviourismwasbothempiricalandmaterialisticinspirit.Itsmethodofpsychologicalinvestigationwastoobserveindividualinstancesofbehaviour.Theexplanationsitofferedforthesewerenotintermsofmentalstatesbutofthephysicalconditionswhichcausedthebehaviour,andtheresultswhichfollowedfromit.Behaviourcouldbeseenintermsofaseriesofresponsestoobservablestimuli.Certainresponsesleadtofurtherstimuliwhichservetoreinforcetheresponse,withtheconsequencethattheorganismismoreinclinedtoproducethatsameresponsethenexttimethestimuliispresented.Thetermorganismisanimportantoneinthiscontext.Behaviourismdidn’trelyonanyexclusivelyhumancharacteristicsinitsexplanations,andinfactbecamenotoriousforextrapolatingfromthebehaviourofanimalsincontrolledexperimentstothatofhumanbeingsinsocialsettings.Bloomfield’saccountoflanguageisbehaviouristbecauseheexplainsthemeaningoflinguisticexpressionsintermsofthesituations,orstimuli,whichcharacteristicallycausethemtobeuttered.Themeaningofwaterisdefinedbythefactthatitischaracteristicallyutteredwhensomeoneisthirsty.Ifanutteranceissuitablyreinforced(ifthespeakerishandedaglassofwater)thespeakerwillbelikelytoproduceaphoneticallysimilarutterancethenexttime.This,ofcourse,hasimportantimplicationsforthewayinwhichlanguageislearnt,implicationswhichBloomfieldoutlinesasfollows.Thechildbeginsbyproducingsoundsinimitationoftheadultsaround.Iftheadultsrecogniseasoundasbeingsimilartooneoftheirspeechsoundstheyrewardthechild(forinstancebyhandingoveradollinresponseto‘da’).Thisrewardservestoreinforcethechild’sbehaviour,makingitmorelikelytoberepeatedinthefuture.Inthisway,‘thechild’sspeechisperfectedbyitsresults’(30).Theimplicationsofbehaviourismforlanguage,itsrelationshiptothemind,andthewayitislearntbychildren,werealsoconsidered,andweremorefullydeveloped,bythephilosopherWillardVanOrmanQuine.\n154LanguageandmindQuineW.V.O.Quine,asheisgenerallyreferredto,isanempiricistinhisgeneralphilosophicaloutlook,notjustinhisapproachtolanguage.HebecameProfessorofPhilosophyatHarvardUniversityin1948,andpublishedextensivelyduringhislongcareerthere.Butbeforethat,intheearly1930s,hespenttimeinEuropeandwasheavilyinfluencedbythephilosophyoftheViennaCircle,especiallythatofRudolphCarnap.Inhisearlyworkhedevelopedwhathasbecomeknownasa‘scepticism’aboutmeaning.Hearguedthattherearenomeaningswhichwordsorexpressionsinsomesense‘have’independentlyofanyoccasionofuse.Thisideaisn’tnewtous;wehavealreadyencounteredversionsofitinthelaterworkofWittgenstein,andinBloomfield.AnditisconsistentwithQuine’sempiricalapproach;thebestwaytoestablishwhatawordorsentencemeansistoobserveasmanyinstancesaspossibleofhowitisused.Quine’sscepticismaboutmeaningisapparentin‘Twodogmasofempiricism’,anarticlefirstpublishedin1951,andincludedinhis1961bookFromaLogicalPointofView.Inthis,heconcedesthatlinguisticexpressionsare‘meaningful’,inthattheymaybeusedtorefertoobjects,andmaybejudgedtobesynonymousornotwithotherexpressions.But,heclaims,itsimplyisn’tnecessarytopositanotionof‘meaning’,whichwemightthinkofasbeingequivalenttoFrege’s‘sense’,toexplainthis.Quinerejectstheideaofmeaningassomethingwhichexistsinthemindinbetweenawordandanobject;‘meaningsthemselves,asobscureintermediaryentities,maywellbeabandoned’(22).Wecouldsummarisethispositionbysayingthatheallowsthatlinguisticexpressionsmayhaveextensional,butnotintensional,meanings.Aswellasascepticalandextensionalapproachtomeaning,Quineadoptsaformofholisminhisearlywork.Aholisticaccount,asthenamesuggests,isonewhichexplainsanycomplexentityasawhole.Itstipulatesthatthesignificanceofthewholeismorethanmerelythesumofitsparts,andthatit’snotpossiblefullytoexplainthenatureofanyindividualpartwithoutlookingatthewholetowhichitbelongs.SoaccordingtoQuine’s‘semanticholism’,anyexpressionwithinalanguagesystemcanonlybefullyexplained,ordefined,inrelationtotheotherexpressionsinthatsystem.WecanrecogniseinitsomethingsimilartoSaussure’saccountoflangue,thelanguagesystem.RememberthatSaussureexplainedthesignificance,orvalue,ofeverysigninthelanguageasbeinginterrelatedwith,anddependenton,thatofeveryothersigninthesamesystem.ThisstructuralistaccountofmeaningisinkeepingwithQuine’sscepticismaboutmeaningingeneral.Meaningisn’t‘determinate’,orassignedtoanylinguisticexpressionbyvirtueofitsform.Rather,itisdependentontheuseoftheexpressionandontheuseofeveryotherexpressioninthelanguage.Thisapproachhasprofoundconsequencesforthenatureoflanguage,andfortherelationshipbetweenlanguageandmind,whichwecanconsiderinrelationtosomeofQuine’slaterwork.\nLanguageandmind155In1960,QuinepublishedabookcalledWordandObject,whichhededicatedtoRudolphCarnap,andinwhichhesetouthisapproachtosemantics.Aswemightexpect,theapproachisthoroughlyempiricalinnature;itcombinestheextensionalandholisticoutlookdevelopedinhisearlierworkwithabehaviouristaccountoflanguageofthetypeproposedbyBloomfieldandbySkinner.ThismixtureisnowheremoreapparentthaninQuine’sdiscussionofradicaltranslation.Quineaskshisreadertoengageinthefollowingmentalexercise,intheinterestofdiscoveringsomethingofthenatureoflanguage,andofitsrelationshipstothespeakerandthespeaker’sperceptionoftheworld.Imaginethatyoufindyourselfamongaremotetribeofpeople,whoselanguageyouareinterestedinunderstanding,butwithwhomyouhavenoobviouswayofcommunicating.Yourtaskistoobservethenatives’behaviour,untilyouareabletosupplyatranslationinyourownlanguageforeverywordorsentenceintheirs.Youareengagedin‘radical’translationbecauseyoucan’trelyonanysimilaritybetweenthetribe’slanguageandyours,asyoumightwithagroupofpeoplespeakingFrenchorSpanish.YouhavenoreasontobelievethatanywordsintheirlanguagewillsoundsimilartowordsinEnglish,thattheywillusegesturesandbodylanguageinthesamewayasyoudo,oreventhattheywillshareyourperceptionoftheworld.Quine’sdescriptionoftheprocessesyouwouldneedtogothroughislengthyandcomplex.Wewillconsiderjusttheverybeginningofit,andthenskiptotheconclusionswhichhedrawsfromit.ThewholeprocessisdescribedindetailinWordandObject,andisdiscussedinsomeoftheworkslistedinthe‘furtherreading’sectionofthischapter.Inyourtaskasfieldlinguist,youaremostlikelytobeginbytryingtofindtranslationsforwordswhichare‘keyedtopresenteventsthatareconspicuoustothelinguistandhisinformant’(29).Outwalkingwithanativeoneday,youbothseearabbit,andthenativesays‘Gavagai’.YouspeculatethatthiscanbetranslatedintoEnglishas‘Rabbit’.4Butofcourseyoucan’tyetbesureaboutthis;thenativemighthavebeenreferringtothecolouroftherabbit,orthewayitwasmoving,orjustusingawordequivalentto‘animal’.Yournextmovemightbetoexperiment,inordertodiscoverwhatotherobjectswillleadthenativetoassentto‘Gavagai’.Youutter‘Gavagai?’,inwhatyouhopewillberecognisedasatoneofenquiry,inthepresenceofvariousrabittyandnon-rabittyobjects,andnotewhatresponseyouget.Youcan’trelyongestureorfacialexpression(forallyouknow,shakingtheheadmaymean‘yes’forthesepeople),butyounoticethatsometimeswhenyousay‘Gavagai?’yougettheresponse‘Evet’,andsometimesyougettheresponse‘Yok’.Thismightseemlikesomesortofprogress,butyouquicklyspotanotherproblem.Youmayspeculatethatthesetwocanbetranslatedas‘Yes’and‘No’,butyouhavenowayofknowingwhichiswhich.Elicitingan‘Evet’inthepresenceof\n156Languageandmindarabbitisnoguaranteethatyoucantranslateitas‘Yes’,sincethiswouldrelyonyourown,unconfirmedspeculationthat‘Gavagai’canbetranslatedas‘Rabbit’.Infact,yourtaskofradicaltranslationbecomesharderthelongeryoustaywiththetribe,asyourealisethateveryhypothesisyouformaboutmeaningisrelatedto,andhaspotentialeffectsfor,yourtranslationofeveryotherwordorsentenceyouencounter.WecanclearlyrecogniseinthisQuine’s‘semanticholism’;ourdecisionsaboutthemeaningofeachexpressionofthelanguagewearestudyingdependonthosewemakeaboutallotherexpressions.ButQuinedrawsafurther,relatedconclusionfromthis,whichhasimportantimplicationsfortherelationshipbetweenlanguageandmind.Ifyouperseverelongenough,youmayeventuallycomeupwithtranslationsforeveryexpression,whichareconsistentwitheachotherandtogethergiveyouapictureofthelanguageyouarestudying.Supposethatyouthenmeetanotherlinguistwho,unknowntoyou,hasalsobeenstayingwiththetribeandhasalsojustcompiledanaccountofthelanguagebyradicaltranslation.Youcomparenotes,andfindthat,althoughbothofyouhavereachedsystemsoftranslationwhicharecoherentandwhichallowyoutoconversefreelywiththenatives,thesystemsoftranslationwhichyouhavedevelopeddiffermarkedlyfromeachother.Hereyouhavetwooptions.Eitheryoucouldeachinsistonthesuperiorityofyourownsystem,andattempttodiscoverwhohasmadethe‘right’translation.Oryoucouldfindthissituationperfectlysatisfactory;youbothhavesystemswhichwork,sothereisnoreasontobelievethatonemustbe‘right’andtheother‘wrong’,oreventhatonemustbebetterthantheother.Quinesuggeststhesecondoftheseresponses.ThefactthatyouhaveeachcomeupwithadifferentsystemoftranslationisanalmostinevitableconsequenceofwhatQuinecallsthe‘principleofindeterminacyoftranslation’(27).Aswehavealreadyseen,hearguesthatwordsdon’thavedeterminatemeaning.Nowhesuggeststhatthereisnosingle,determinate,wayofexplainingthesemanticsofalanguage,oroftranslatingitintoanotherlanguage.Allthatneedstobeexplainedbyanadequateaccountofanylanguageiswhatcanbeobservedbysomeoneinterestedinthebehaviourofitsspeakers.Anytwoaccountswhicharebothsuccessfulinexplainingthisbehaviourcanbeseenas‘empiricallyequivalent’(78),aclaimwhich,Quinenotes,isinconflictwith‘thealmostuniversalbeliefthattheobjectivereferencesoftermsinradicallydifferentlanguagescanbeobjectivelycompared’(79).Quineisreiteratinghisclaimthatmeaningisnotsomethingwhichwordsandexpressionscanbesaidto‘have’inanyconcretesense.Rather,it’snecessarytomakesubjective,empiricaldecisionsabouthoweachexpressionisused.Thecorrectprocessoftranslationisnotoneofmatchingupexpressionsfromthetwolanguageswhich‘havethesamemeaning’,butofobservingsimilaritiesinverbalresponsestovariousstimuli.\nLanguageandmind157Thisis,ofcourse,abehaviouristaccount.Languageisseenasaseriesofobservableresponsestostimuli;thelinguist’staskistopredictaccuratelywhatutteranceswillbepromotedbywhatstimuli,forinstancebytheappearanceofarabbit.Itisalsobehaviouristinthatitseestheobservablebehaviourasallthereistolanguage;thereisnothing‘underlying’thebehaviourwhichneedstobeexplained.Ifyouandyourfellowlinguisthavebothdeducedexplanationsofthebehaviouryouhaveobserved,youhavebothproducedsatisfactoryaccountsofthelanguage,becausethebehaviourisallthereistoexplain.Thishasimportantimplicationsforthenatureoflanguage,andforwhatitmeansto‘know’alanguage,whichweshallconsiderlaterinthischapter.ThemoralaboutthenatureoflanguagewhichQuinedrawsfromtheexerciseofradicaltranslationhasimplications,too,forthechild’staskinlearningafirstlanguage.RememberthatinLanguage,Bloomfielddescribedaprocessofstimulus,responseandreinforcementwhichgraduallyleadstothechild’suseoflanguagebecomingmoreandmorelikethatoftheadultsaround.Similarly,inachapterofWordandObjectcalled‘Theontogenesisofreference’,Quinedescribesaprocesswherebyachildinitiallyproducesrandomutterances,whichare‘selectivelyreinforced’.Subsequently,‘thecreaturetendstorepeattherewardedactwhenstimulirecurthatchancedtobepresentattheoriginalperformance’(81).ForQuine,whatwecalla‘language’isdefinedasaseriesoftendenciestoreinforcecertaintypesofverbalbehaviourinagroupofpeople.Languagethereforeiscentrallydefinedintermsofthepracticesofacommunity;hebeginsWordandObjectwithaclearstatementtothiseffect:Languageisasocialart.Inacquiringitwehavetodependentirelyonintersubjectivelyavailablecuesastowhattosayandwhen.Hencethereisnojustificationforcollatinglinguisticmeanings,unlessintermsofmen’sdispositionstorespondovertlytosociallyobservablestimulations.(ix)ForQuine,alanguageexistsonlyinthepracticesofthosewhouseit,andacquiringthelanguageinvolvesagradualprocessofobservingthosepractices.Inasensethen,thetaskofchildreninacquiringalanguageisverysimilartothatofthefieldlinguistswehavejustconsidered;theytoohavenopreconceivedideasaboutwhatsoundsmightrelatetowhatobjects,butmustdeducethisfromobservation.Justlikethefieldlinguists,childrenwillarriveatasatisfactoryversionofthelanguageinsofarastheyareabletoproducetheappropriatebehaviourintheappropriatesituation,andeventuallytoreinforcetheappropriatebehaviourinothers.Thereisnothing‘beyond’thedispositionstobehaviourwhichtheyneedtoacquire.ItwastheseaspectsofQuine’saccountoflanguageacquisition,andtheirimplicationsfortherelationshipbetweenlanguageandmind,whichwere\n158LanguageandmindthefocusofacritiqueofWordandObjectbythephilosopherandlinguistNoamChomsky.Wewilllookatthisnext,sinceitillustratesmanyofthefundamentalsofwhatcametobeknownasthe‘Chomskyanrevolution’.TheInnatenessHypothesisIn1969,whenChomskypublishedhisresponsetoWordandObject,itwouldhardlyhavecomeasasurprisethathedidn’ttakeaveryfavourableviewofQuine’slinguistics.IfQuineisbestdescribedasanempiricalphilosopher,belongingtothetraditionofLockeandBerkeley,thenChomskycanbeclassifiedasarationalist,drawingontheworkofLeibniz,ofKant,andultimatelyofPlato.Sincethelate1950s,Chomskyhadbeenpublishingbooksandarticleswhichhadestablishedhisreputationasaninnovativethinkeraboutthelanguageandthemind.Amonglinguists,hehadattractedadmirersandcriticsinperhapsequalnumbers,buthisimpactonlinguisticswassuchthatnoonehadbeenabletoignorehim.NoamChomskywasborninPhiladelphiain1928.Attheageoftwenty-sevenhetookuparesearchpostattheMassachusettsInstituteofTechnology,whereheeventuallybecameaprofessor.Chomsky’scontributiontolinguisticshasbeenwide-rangingandprofound,andwewillbeassessingonlypartofithere.However,linguisticsisjustoneofthesubjectsinwhichhehasdistinguishedhimself,andmaynotbetheoneforwhichheisbestknown.Probablymorepeopleareawareofhiswritingsinpoliticsandcurrentaffairs.HehasbeenastaunchcampaignerforhumanrightsandcriticofAmericanforeignpolicy,andindeedspentsometimeinjailafterprotestingagainsttheVietnamWarinthe1960s.5Chomsky’s1969responsetoWordandObjectwasanarticleentitled‘Quine’sempiricalassumptions’.Inthis,hecriticisesboththedetailsofQuine’saccountoflanguageacquisition,andtheverynatureoftheaccountitself.Hiscriticismsofthedetailsarelargelyrelatedtowaysinwhichhefindstheaccountunworkable,orunabletoexplaintheapparentfactsoflanguageacquisition.ThereasonswhyheopposesthenatureofQuine’saccountingeneralwillbecomeapparentasweinvestigatethementalistandinnatist,asopposedtoempiricalandbehaviourist,accountwhichChomskyproposes.Chomsky’smainargumentagainstanypurelyempiricalaccountoflanguageacquisitioncanbesummarisedasfollows.Considertheknowledgechildrenmustbuildupinordertobeabletousealanguage;itisverycomplex,andveryspecific.Thenconsiderthetypeofevidencewhichisavailabletodevelopingchildren;itisfragmentaryandunreliable,arandomcollectionofutterancesincludingsomewhichagrammarianwouldclassas‘errors’.Putthesetwotogetherand,Chomskyargues,youcan’thelpbutreachtheconclusionthattheavailabledataaren’tsufficienttoaccountforthederivedknowledge.Thisobservationledhimtodiscussthe\nLanguageandmind159‘povertyofinput’availableintheprocessoflanguageacquisition.Inparticular,thenumberofsentenceswhichspeakersofalanguagearecapableofunderstandingorusingisinfinite,butinacquiringthatlanguagetheywillactuallyencounteronlyaverysmallnumberofthem.Iflearningalanguageinvolvedsimplytheprocessofbeingconditionedtorespondinappropriateways,itcouldonlyeverresultinafinitenumberof‘setresponses’tosentencesencounteredmanytimesbefore.InhiscriticismofQuine,then,Chomskywouldappeartohavesethimselfthetaskofexplaininghowchildrenreachthestagewheretheycanunderstand,anduse,aninfinityofsentences,veryfewofwhichtheywilleverhaveencounteredbefore.Hehad,infact,beendevelopingjustsuchanexplanationoveranumberofyears,invariouspublications.WehavealreadyseenthatChomskycanbeplacedintherationalisttradition,theschoolofthoughtwhichexplainsthewaysinwhichweperceiveandunderstandtheworldintermsofinnatepropertiesofthehumanmind.Hisaccountoflanguageacquisitionhasbecomeknownasthe‘InnatenessHypothesis’.Itisalsosometimesdescribedasbeingnativist,ageneraltermappliedtotheorieswhichrelyoninnateorinbornideasorknowledge.Chomsky’sInnatenessHypothesis(commonlyabbreviatedtoIH)isbasedontheclaimthatthereisaspecialisedandindependent‘languagefaculty’.Thisisamentalcapacity,commontoallhumanbeings,whichallowsustoacquireanduselanguage,andwhichservesnootherpurpose.Itisthespecialisednatureofthelanguagefaculty,andtheclaimthatitcausesthenaturalandeffortlessdevelopmentoflanguage,whichmaketheIHdistinctive.ForQuine,tospeakof‘acquiring’alanguageisjustanalternativewayofspeakingabout‘learning’it;welearnlanguageusingmentalcapacitiessimilartothoseweusetolearnavastnumberofotherskills.ButforChomskythetwoarenotequivalent.Children‘acquire’languageastheydevelop,buttheydon’t‘learn’itastheymight,forinstance,learntorideabicycleoruseaknifeandfork.Theideathatthereisamentalfacultyemployedinacquiringlanguage,whateverindividualhumanlanguagethatmightbe,isnotoriginaltoChomsky.Ashehimselfacknowledgesinhis1966bookCartesianLinguistics,itwasonefocusofseventeenth-centuryrationalistphilosophy.Therewasageneralinterestwithinthistraditioninthecontributionmadebythehumanmindtothetypeofknowledgeitcanacquire.Thisinturnledtoaninterestindiscoveringinnateprinciplesofmind,includingthoserelatingtolanguage.AsChomskyexplains,‘Byattributingsuchprinciplestothemind,asaninnateproperty,itbecamepossibletoaccountforthequiteobviousfactthatthespeakerofalanguageknowsagreatdealthathehasnotlearned’(60).However,Chomskyappliedthisideatotheobservableproblemsoflanguageacquisition,andsuggestedadefiniteformforthisfaculty,aswewillseelaterinthischapter.Chomsky’sproposalhasimportantimplicationsforthenatureoflanguage,whichwewillexplorelater,andalso,ofcourse,forthehuman\n160Languageandmindmind.This,infact,wasChomsky’smainfocusofinterestindevelopingtheIH.Althoughheiswellknownforhisworkonlanguageacquisition,hehasn’tspentmuchtimeontheactualdevelopmentalstageschildrengothrough.6Rather,hisinterestliesinconsideringwhatthehumanmindmustbelike,giventhelanguagesitisabletoacquireanduse.Themindofanadultwhohassuccessfullyacquiredalanguagecanbeconsideredashavingreachedacertain‘steadystate’;Chomsky’sinterestisinconsideringthisand,byextension,the‘initialstate’ofthehumanmind,orthepropertieswhichmustbeinnatetoitinorderforthesteadystatetobeobtainable.Inhisownwords,thetaskhehassethimselfisto‘deviseahypothesisaboutinitialstructurerichenoughtoaccountforthefactthataspecificgrammarisacquired,undergivenconditionsofaccesstodata’.7ThisquotationrevealssomethingofChomsky’sgeneralphilosophicalmethod,anissuewhichheaddressesdirectlyin‘Quine’sempiricalassumptions’.There,herespondstoQuine’scriticismthat,inpositingtheIH,hewasgoingbeyondwhatwasstrictlylicensedbyexperience,anddiscussingtheexistenceofmentalentitiesforwhichhehadnoconcreteevidence.Chomskyarguesthatthismethodisinfacttheonewhichmustbeadoptedifinterestingphilosophicaldiscussionistobepossible.Serioushypotheses,heargues,willalways‘gobeyond’theempiricalevidenceavailable:‘Iftheydidnot,theywouldbewithoutinterest’(66).Inotherwords,whereasBloomfieldadvocatedastrictlyinductivemethodinlinguistics,Chomsky’smethodisadeductiveoneofforminghypotheses.Theroleofthedataoflanguageistotestoutthesehypotheses;thedatamightsupportthehypothesesoralternativelycausethemtobeamended,orevenabandoned.Theprocessofhypothesisformation,heargues,allowslinguiststodomorethandescribeandanalyse;itallowsthemtousetheevidencewhichisavailabletoinspection,suchastheobservabledataoflanguageuse,tothinkaboutphenomenawhicharenotavailable,suchastheunderlyingmentalstatesandprocesses.Aswehaveseen,oneofChomsky’smainargumentsinfavourofaninnatelanguagefacultyisthecomparisonbetweenthecomplexlanguagesystemandtheapparentlyinadequateevidenceonwhichitisbuilt.Heclaimsthat,despitethepovertyoftheinput,childrenacquirelanguagerapidly,withoutmuchapparenteffort,andinanorderandtimescalewhichisremarkablystandardacrossallhumanlanguages.Itisfurtherclaimedthatthereislittleevidenceofdifferencesinorderandspeedofacquisitionbetweenchildrenofwidelydifferingintelligence.Sochildrencan’tbeusinggeneralintelligenceandcognitivecapacitiesto‘figureout’thelanguage.Rather,Chomskydescribestheknowledgeofalanguageas‘growing’inthemindinanappropriateenvironment,attheappropriatestageofdevelopment.8Anyaccountoflanguagewhichisbasedoninnatepropertiesofmindreliesonanassumptionoftheuniversalityoflanguage.Atitsmostbasic,thisuniversalityisreflectedinthefactthatallhumanbeings,regardlessof\nLanguageandmind161raceorgeographicallocation,havelanguage,apropertysharedbynootherspecies.9Despitehugedifferencesbetweenculturesintermsofsocialarrangements,beliefsandcustoms,nomutehumantribehaseverbeenfoundand,asEdwardSapirpointedoutinthe1920s,allhumanlanguagesarealikeintermsofcomplexityandpowerofexpression.However,strongerclaimsofuniversalityhavebeenmade.WehaveseenthatSaussureandBloomfield,forinstance,describehumanlanguagesascapableofdifferingfromeachotherwithoutlimit.Chomsky,ontheotherhand,arguesthatthereareregularities,or‘languageuniversals’;allhumanlanguagessharecertaingeneralrulesaboutwhatwordclassesandwhatstructuresare,andarenot,possible.It’simportanttobearinmindthat,intalkingaboutlanguageuniversals,linguistssuchasChomskyaren’tconcernedwiththeindividualwords,orthevocabularylistswemightfindourselveslearningaspartofaforeignlanguageclass.Indiscussinglanguage,Chomskyisconcernedwithgrammar,withtherulesofsentenceformation.Thereisnosuggestionthatthereisinnateknowledgeoftheactualwordsofanylanguage.Afterall,anewbornchildisequallywellequippedtolearnanyhumanlanguage.Rather,languageuniversalsaresaidtoexistamongthegeneralgrammaticalruleswhichdeterminehowlanguagesarestructured.Theexistenceoflanguageuniversalsisacontroversialclaim,aswewillsee,butonenecessarytotheIH.Iftherewerenorestraintsonwhatformshumanlanguagescouldtake,therecouldbenosinglefacultyspecificallytailoredtoacquiringanylanguage.Ifthelanguagefacultyisinnate,itmustbeuniversal,justasthephysicalorgansofthehumanbodyareuniversal.Beforeweconsiderthenatureoftheinnate‘languagefaculty’whichChomskyclaimsisabletoaccountforsuchuniversals,wewilllookfairlybrieflyatsomeoftheargumentswhichphilosophicalcontemporariesofChomskyhaveusedagainstitsveryexistence.Wewillconsidersomeoftheseparateargumentsputforwardbylinguistsinthefinalsectionofthischapter.10Anti-nativismPhilosophicaloppositiontoChomsky’sideascamefromanumberofsides,notjustfromthebehaviouristtraditionrepresentedbyQuine.TwonamesparticularlyassociatedwiththisoppositionarethoseofHilaryPutnam,anear-contemporaryofChomsky,andNelsonGoodman,twentyyearsolder.BothAmericanphilosopherswereawardedprofessorshipsatHarvard;Goodmanin1967andPutnamin1976.Putnam’sobjectionstotheIHaresummedupinanarticleentitled‘The“InnatenessHypothesis”andexplanatorymodelsinlinguistics’,firstpublishedin1967.HesummariseswhatheseesasthemainclaimsmadeintheIH,characterisingitas‘essentiallyandirreparablyvague’(293,originalemphasis).11Mostofhis\n162Languageandmindcriticisms,however,arebasedaroundthecentralnotionthattheIHissimplynotnecessarytoaccountforthefactsoflanguageacquisitionanduse;fortheIHtobeworthseriousconsideration,itmustnotjustbecompatiblewiththerelevantlinguisticfacts,butactuallyoffertheonlywayofexplainingatleastsomeofthem,somethingforwhichPutnamfindsnoevidence.Insupportofhisclaims,Putnamspeculatesabouttheprobablepropertiesofa‘Martian’languageusedbybeingswithhuman-likeintelligence.Hearguesthatanysuchlanguagewouldhavetohaveagrammar—itwouldhavetobe‘builtupbyrecursiverulesfromalimitedstockofbasicforms’—orelseitcouldn’tbe‘practicallyinfinite’(294).WewouldhavetosaythattheMartianssharedsomegeneralcognitivecapacitywithhumans,butnotthattheyalsosharedaninnatelanguagefaculty.Anyorganismcapableofusinglanguagemust,tosomeextent,possessinnatementalcapacities:Howcouldsomethingwithnoinnateintellectualequipmentlearnanything?Tobesure,human“innateintellectualequipment”isrelevanttolanguagelearning;ifthismeansthatsuchparametersasmemoryspanandmemorycapacityplayacrucialrole.ButwhatrankBehavioristissupposedtohaveeverdeniedthis?(295,originalemphasis)Thisraisesanimportantissueinrelationtothecompetingtheoriesoflanguageandmindweareconsidering.Itissometimestemptingtoconstruethedebateasastraightforwardoppositionbetween‘innatist’and‘non-innatist’accounts,thatis,betweenseeinglanguageacquisitionasdeterminedentirelyby‘humannature’ontheonehand,andby‘experience’ontheother.However,bothtypesofaccountdependtosomeextentonbothnatureandexperience;theyjustplacedifferentemphasisontheirrelativeimportance.ForaninnatistsuchasChomsky,therightsortofexperience,orinput,isnecessary,butonlyinordertotriggerthenaturalprocessoflanguagedevelopment.ForphilosopherssuchasPutnam,andindeedQuine,experienceisofprimaryimportance,andisactedonbyverygeneralinnatelearningprinciples.Aswehaveseen,thesearegeneralenoughtobeappliedtolanguageaswellasotherskillsandbehaviours.PutnamisequallyscepticalabouttheclaimthattheIHisjustifiedby‘universal’wordclassessuchasnoun,verbandadjective.Hesuggeststhatsuchapparentsimilaritiescanbemoresimplyexplainedbythefactthatalllanguagesincludemeansforidentifyingobjects,andforsayingthingsaboutthoseobjects.Hearguesthatmanyofthe‘universal’featuresofhumanlanguagecouldequallywell,andmoresimply,beexplainedbyalllanguagesbeing‘descendedfromasingleoriginallanguage’(296).Putnamfurthersuggeststhattheapparenteaseandrapiditywithwhichchildrenlearntheirfirstlanguageisnogreatsurpriseifyouconsidertheamountoftimethey\nLanguageandmind163haveattheirdisposaltoobserveandstudyit.Itturnsouttobenomoreremarkablethanthatpeopleshouldbeabletolearntosolvedifficultproblems,ortoplaychess.Chomsky,heargues,isignoringtherealissue:themuchmoresignificantquestionofhowitisthathumanbeingslearningeneral.Thus,‘invoking“Innateness”onlypostponestheproblemoflearning;itdoesnotsolveit’(298).NelsonGoodmanhasalsoraisedquestionsaboutthenecessityofaninnatelanguagefaculty.InhisbookOfMindandOtherMatters,publishedin1984butdrawingonworkfromearlierdecades,hesetsoutsuccinctlysomeofthereasonswhyhefindsclaimsaboutlanguageuniversalsdubious.Hearguesthatanycollectionoflanguages,orindeedofanytypeofsystem,will,coincidentally,havecertainfeaturesincommon.Theclaimthatthefeatureswhichhavebeenidentifiedascommontoalllanguagesarenatural‘languageuniversals’can’tbesupported;itis‘agratuitousoneimmunetofeasibleexperimentaltest’(16).Heintroducesafurtherargumentagainsttheexistenceoflanguageuniversals,whichgoesasfollows:Thelinguisticuniversalsusuallycited,suchassubject-predicateform,seemtobefeaturesoftranslationsintoourlanguageratherthanofthelanguagetranslated’(16).Inotherwords,Goodmanissuggestingthatlinguistsareabletoclaimthatthereareregularitieswhichholdacrossalllanguagesonlybecausetheyhavebeenableto‘impose’certainfeaturesontolanguageswhichmayberadicallydifferentfromtheirown,usuallyfromEnglish.Theexistenceofso-called‘languageuniversals’,heargues,isneitherasremarkablenoraswidespreadasChomskyclaims.ThelanguagefacultyAswehaveseen,theinnatisthypothesisisstrikinglydifferentfromother,contemporaryexplanationsoflanguageandtheprocessesbywhichitislearnt.IfwecompareChomsky’saccountofaninnatelanguagefacultywithQuine’sbehaviouristmodelofstimulus,responseandreinforcement,andwithPutnam’sinterestingeneralcognitivecapacityandlearningmechanisms,wemightfindourselveswonderinghowthethreephilosopherscouldbewritingaboutthesamephenomenon.Andperhapsthereisasenseinwhichtheyarenot.Thatis,theirviewsonlanguage,andthereforeontheprocessesinvolvedinacquiringlanguage,aresodifferent,thattheyaren’teventryingtoaccountforthesamesetoffacts.IncomparingChomsky’stheorywiththoseofhiscontemporaries,wearen’tjustconsideringwhichbestaccountsforthefactsoflanguage;wearelookingatcompletelydifferentversionsofthosefacts.Indrawingonthephilosophicaltraditionconcernedwithinnatepropertiesofthemind,Chomskywascommittedtoamentalistaccountoflanguage.Inotherwords,afulldescriptionoflanguageforhimwouldinvolveanaccountnotjustoftheobservablebehavioursassociatedwithit,butalsoofthementalprocess‘behind’these.RememberthatforQuinethe\n164Languageandmind‘indeterminacyoftranslation’wasevidencethatthereisnoobjectivetruth,orunderlyingrules,tobediscoveredinstudyinglanguage,andthereforethatnosingledescriptionofanylanguagecouldbesaidtobethe‘right’one.Allthatthelinguistcould,orindeedshould,dowastoobserveandexplainthelinguisticpracticesofacommunityofspeakers.ForChomsky,ontheotherhand,languageisnotjustgovernedbyasetofunderlyingrules;itactuallyconsistsofthoserules.Thechild’staskinacquiringalanguageisnotoneoflearningthecorrectdispositionstobehave,butofmasteringtherulesofthelanguage.Thelinguist’staskisnotoneofdescribingthebehaviourofagroup,butofaccuratelymodellingtheknowledgeofeachindividual.Chomsky’sbeliefsaboutthenatureofthisknowledgearesignificantinhisresponsetoPutnam’scriticisms.Putnamarguedthat,giventhenatureofcommunicationandofintelligence,it’snogreatsurprisethatlanguagesexhibitcertainsimilarities.Itwouldbemoreusefultoexplaintheseintermsofgenerallearningprocessesthanofspecificinnatecapacity.Chomsky’sreplytothesecriticismsfirstappearedin1969aspartofalongarticle,butwerepublishedin1981asanextractentitled‘ReplytoPutnam’.ThisbeginswithastatementtotheeffectthatPutnam’sargumentsarebasedonanincorrectviewoflanguage:Specifically,heenormouslyunderestimates,andinpartmisdescribes,therichnessofstructure,theparticularanddetailedpropertiesofgrammaticalformandorganisationthatmustbeaccountedforbya‘languageacquisitionmodel’,thatareacquiredbythenormalspeaker-hearerandthatappeartobeuniformamongspeakersandalsoacrosslanguages.(300)Chomskyemphasisesthesignificanceofthewayinwhichsentencesinnaturallanguagesareformed,somethingwhichhesuggestsPutnameitheroverlooksorisunawareof.Phrasestructurerulesgeneratebasestructuresonwhichtransformationalrulesoperatetogivesurfacestructures.Theseprocessescanbetraced,andindeedshowremarkablesimilarities,acrossallnaturallanguages.Thenatureoftheseprocesses,orstructure-dependentoperations,are‘ofapeculiarsortthathaveneverbeenstudiedoutsideoflinguistics,inparticularnotinanybranchofmathematicswithwhichIamfamiliar’(302).Linguisticknowledge,then,isofaverycomplex,veryspecifictype,andcouldhardlybeseenasarisingnaturallyandunsurprisinglyfromgeneralcognitivecapacity.Therelevantmentalstateofaspeakerwhohasacquiredalanguageisbestdescribedasafullgenerativegrammarforthelanguage.Indevisingasuitablegenerativegrammar,thelinguististhereforemodellingpartofthehumanmind.Aswehaveseen,theinitialstateofthehumanmindmustbecomplexenoughtoenableanychildtoacquireanyhumanlanguage,butspecific\nLanguageandmind165enoughtobecompatibleonlywiththeformcommontoallsuchlanguages.Chomskyconceivesofitascontainingaseriesofrules,orprinciples,whicharesharedbyalllanguages.Thechild’staskinacquiringalanguage,then,istodiscovercertainpropertieswhichareuniquetothelanguageinquestion.Toaidthisprocess,theinitialstatealsocontainsanumberofvariables,orparameters,whichwillbegivendifferentvaluesdependingonthelanguagetowhichthechildisexposed.Theseallowforthe‘finetuning’whichdeterminesthatthechildacquiresoneparticularlanguage.Inthisway,Chomskyoffersasolutiontothelogicalproblemoflanguageacquisition.Thedataavailabletochildrenwouldn’tbeadequateiftheyhadtolearneverythingaboutthelanguagefromscratch.Butitissufficienttoallowthemtocarryoutthetaskofdetermininghowcertainparameters,ofwhichtheyhaveinnateknowledge,aresetforthelanguageinquestion.Theprinciplesandparameters,then,makeuptheinnateinitialstate.Theyarecompatiblewitheveryhumanlanguage,andexplaintheeaseofacquisitionforanyindividuallanguage.ChomskydescribesthemastheUniversalGrammar.Hismaininterest,aswehaveseen,isnotintheprocessoflanguagedevelopment,butinthefinalorsteadystateoftheadultmind:inthequestionofwhatitisthatpeople‘know’whentheyhavesuccessfullybeenthroughtheseprocesses.WewillconcludeourinvestigationoftheInnatenessHypothesiswiththisquestion.KnowledgeanduseoflanguageWehaveseenthatmentalistandempiricalaccountsviewlanguage,andthereforethetaskofexplaininglanguage,verydifferently.Asaresulttheydifferintermsofthescopeofwhattheyattempttoexplain.So,unlikeanempiricalaccountsuchasQuine’s,Chomsky’smentalistaccountdoesn’tattempttoexplaincompletelytheobservablebehaviouroflanguageuse.Chomskyisquiteexplicit,inmuchofhiswork,thathisaimisnottoaccountforeverythingwhichgoesonincommunication:thatheisconcernedonlywiththosementalphenomena,thegenerativerulesofgrammar,whichforhimmakeuphumanlanguage.Thesecontributetotheobservablephenomenaoflanguageusedincommunication,butareonlyonetypeofanumberofdisparatefactorsinvolved.Youmayhavespottedaprobleminthelastparagraph,centringonthetermlanguageitself.Foranempiricist,languagereferssimplytowhatyoucanobserveactualspeakersandhearersdoing.ButChomskyclaimsthat,whatevertheobservablebehaviourmaybe,languageisactuallysomethinginthemindofthespeakersandhearers.Thispresentsaproblemofterminology,whicharisesfromtryingtousethesametermtorefertoboththeobservableandthementalphenomena,andexplainswhyChomskyfounditnecessarytocomeupwithtwonewterms.Inhisearlywork,Chomskydistinguishesbetweencompetenceandperformance.Competenceisthesetofknowledgewhichformsthesteady\n166Languageandmindstateofthespeaker.Itismodelledbyagenerativegrammar,capableofproducingallandonlythegrammaticalsentencesofthelanguage.Performance,ontheotherhand,ismadeupofthemuchmorediversephenomenawhichactuallyoccurinlanguageuse.Itisdependentoncompetence,butalsodrawsonarangeofotherpersonalandcontextualfactors.Forinstance,inAspectsoftheTheoryofSyntax,publishedin1965,Chomskyarguesthat‘Tostudylinguisticperformance,wemustconsidertheinteractionofavarietyoffactors,ofwhichtheunderlyingcompetenceofthespeaker-hearerisonlyone’(4).Theotherrelevantfactorsincludephenomenaasdiverseasmemory,tirednessandemotionalstate.12Inlaterwork,forinstanceinKnowledgeofLanguage,publishedin1986,ChomskyintroducesthedistinctionbetweenI-languageandE-language.TheI-,orInternalised-languageisthestateofmindofthespeaker,whichhasalwaysbeenhismainfocusofinterest.TheE-,orExternalised-languageistheobservablebehaviour.Hedescribesthechangeofemphasisbroughtaboutbygenerativegrammar:theshiftinfocuswasfromthestudyofE-languagetothestudyofI-language,fromthestudyoflanguageregardedasanexternalisedobjecttothestudyofthesystemofknowledgeoflanguageattainedandinternallyrepresentedinthemind/brain.(24)Chomskygoessofarastoclaimthat,ingenerativegrammar,theconceptofE-language‘appearstoplaynoroleinthetheoryoflanguage’(26).Hisclaimisthatlanguageisaninternalstateofmindwhichonlysecondarily,andalmostcoincidentally,isusedinproducingatypeofbehaviourwhichservesincommunication.Thisparticularlyextremeviewhasattractedalotofcriticism,butitisn’tonewhichheseemstoadhereto,ortoexpresssocontroversially,insomeofhisotherwritings.Inanycase,wecanseehowdifferentChomsky’sapproachisfromaccountssuchasthoseofferedbyQuine,orindeedbySaussure,inwhichlanguageisdependentonitsuseincommunication.Generativegrammarattemptstomodel,andsoexplain,whatitisthatspeakersknowwhenthey‘knowalanguage’.Itdoesn’tattempttomodelwhathappenswhenthosespeakersproceedtocommunicatewitheachother.Inbothphilosophyandlinguistics,workhascontinued,withinbroadly‘innatist’and‘non-innatist’traditions,ontherelationshipbetweenlanguageandthemind.TheAmericanphilosopherJerryFodor,forinstance,canperhapsbeseenasgoingtotheoppositeextremefromSapirandWhorf,whoarguedthatmodesofthoughtaredependenton,andacquiredwith,individuallanguages.Hehasproposednotjustaninnatelanguagefaculty,butaninnatelanguageormentalese,whichisthemediumofthought,andisthereforenecessarilypriortotheacquisitionofanyparticularlanguage.Theconnectionbetweenlanguageandmindhas,more\nLanguageandmind167generally,remainedacentraltopicinthevariousdirectionsinwhichlinguisticshasdevelopedandcontinuestodevelop,andwewillconcludewithabrieflookatsomeofthese.EmpiricistandmentalistlinguisticsTherelationshipbetweenlanguageandmindhascontinuedtobeanenduringtopicofinterestinlinguistics.Thishasinevitablymeantthatsomelinguistshavebeenworkinginareascloselyrelatedtopsychology.Theintersectionofinterestsbetweenlinguisticsandpsychologyhasbecomethefocusofthebranchofmodernlinguisticsknownaspsycholinguistics.Thiscanbroadlybedefinedasthestudyofwhatgoesoninpeople’smindswhentheyproduceandinterpretspokenandwrittenlanguage:whenlanguageisstoredandaccessed,processedandinterpreted.Mentalprocessesare,ofcourse,difficulttoobserveandrecordforthepurposesofanalysis.Languageprocessinginparticulartendstotakeplacesoquicklyandwithsolittleconsciouseffortthatitisimpossibletostudyitbyaskingspeakerswhattheyare‘doing’whentheyuselanguage.Researchersinthisareaofpsycholinguisticshavethereforedevisedvariousexperimentsandotherformsofobservationtoserveasindirectevidenceoftheprocessesinvolved.Forinstance,theyhavestudied‘slipsofthetongue’,arguingthatthemistakesinwordchoicewhichpeoplemakeinspeechcanilluminatethewaysinwhichwordsarearrangedandstoredinthe‘mentallexicon’.Itis,ofcourse,possibletoseetheinfluenceofmentalistapproachestolanguageinthistypeofstudy,concernedasitiswiththementalprocesses‘behind’theobservabledataoflanguageuse.Thedifferentapproachestothestudyoflanguageofferedbyempiricismandmentalismaretobefoundindiscussionsofothertopicsinpsycholinguistics,andindeedinlinguisticsingeneral.Theempiricallinguisticmethodisparticularlyapparentintheareaofconversationanalysiswhich,aswesawattheendofthelastchapter,emphasisestheobservationandanalysisofactualdataoflanguageuse.Empiricismisalsotheframeworkforthedisciplineofcorpuslinguistics,whichemergedwiththegrowthincapacityandavailabilityofcomputers.Largesamplesoflanguagedataarecollectedinmachine-readableformatforanalysisbycomputer,oftenstatisticalanalysis.Theemphasisisontheauthenticityof‘reallife’data,asopposedtothedataof‘inventedsentences’andintuition.Ithasbeenclaimedthatacorpusshouldideallybesufficientlylargetobefullyrepresentativeofthelanguage;itshouldcontain,intheformofexamples,everythingwhichthelinguistcouldneedtoknowaboutthelanguage.Notsurprisingly,Chomskyhasbeenanopponentofthemethodsandclaimsofcorpuslinguisticssinceitsoriginsinthelate1950sandearly1960s.Hismentalistapproachentailsthatthefactsaboutalanguagecan\n168Languageandmindneverbecontainedinalistofexamples,howeverlarge.Inhisownmorerecentterms,allthiscouldevergiveyouissomeexamplesofE-language.Theactualfactsoflanguage,theI-language,existonlyinthemindsofindividualspeakers,andcanbeaccessedonlybymeansofthoseminds.Everythingyouneedtoknowaboutalanguageispresentinthemindofaspeaker,andcanbeaccessedeitherbyquestioninginformants,orevenbyaprocessofintuitiononthepartofthelinguist.Acorpuswon’taddanythingtothis,andmayinfactcontainexampleswhichdon’tbelongtothelanguageunderinvestigation.Itmaycontainslipsandother‘performanceerrors’.Aswehaveseen,indrawingadistinctionbetweenthe‘real’factsoflanguageandthedistractionsofperformance,Chomskyseesthetaskoflinguistictheoryasbeingtoaccountforjustpartofwhatgoesonincommunication.Somemorerecentapproaches,incontrast,haveseenthetaskofthelinguistasoneofexplainingcommunicationinallitsaspects.Linguistsworkingwithinthisframeworkseektodoawaywithwhattheyseeastheartificialdistinctionbetween‘linguistic’and‘contextual’.Inotherwords,theydon’trecognisetheexistenceoflinguisticknowledgewhichcanbedistinguishedfromtheobservablebehaviouroflanguageuse.Onesuchaccounthasbecomeknownasintegrationistlinguistics,developedbytheOxfordlinguistRoyHarris,andothers.Thisattemptstointegrateallaspectsofcommunicationintooneaccount;meaningisnotsomethingwhichindividualpartsoflanguage‘have’,itissomethingwhichonlyexists,andcanonlybeanalysed,intermsofwhatindividualspeakersdoinparticularcircumstances,andwhateffectsthishas.Chomsky’srigorouslymentalistapproachtolinguistics,then,isnomorewidelyacceptednowthanitwaswhenitwasfirstofferedasachallengetoempiricismandbehaviourism.However,itsinfluenceisstillstronginsomeareasoflinguistics.Forinstance,psycholinguistshavebuiltonhisworkonthenatureoflinguisticknowledgeandonlanguageacquisition.Inparticular,clinicallinguists,interestedinlanguagedisordersandtheevidencetheyofferaboutthenatureoflanguage,havestudiedpatientswithspecificmentalimpairments,whetherpresentfrombirthorresultingfromstrokeorothertrauma.Thefindingsofsomeofthesestudieshavebeenclaimedasevidencethatlinguisticabilityisquiteseparatefrommanyothercognitivecapacities,includingothertypesofcognitionusedincommunication.Thishasinturnbeenofferedinsupportofamentalistaccountoflanguage;theapparentindependenceoflinguisticabilitycanonlybeexplainedbytheexistenceofaparticular,separatementalfacultyconcernedwithlanguage.WorkwithinthementalisttraditionintheareaoflanguageacquisitionhasbranchedoutfromtheearlyemphasisonsyntaxanditsrelationshiptoUniversalGrammar;itnowincludesstudiesofchildren’sdevelopmentinphonology,semanticsandmorphology,withaninterestinassessingwhetherinnateuniversalprinciplescanbefoundintheseareas.Ithasbeen\nLanguageandmind169claimedthat,ingeneral,deliberatecorrectingofgrammardoesn’ttakeplace,andisn’teffectivewhenitdoes;thechildwillproducethecorrectformsandstructuresonlywhendevelopmentallyreadytodoso.Thesuggestionthatthereisno‘negativeevidence’availabletothechild—thatthechildisingeneralnottoldwhatisincorrect—hasbeenusedtoarguethatlanguageisn’t,andcan’tbe,taught.Otherapproachestolanguageacquisitionhaveconcentratedonthesocialandinteractionalnatureoflanguageuseratherthanonthecomplexityofthementalstatesinvolved.Asaresult,theyhaveemphasisedtheplaceoflanguageinthechild’sgeneraldevelopment,focusing,forinstance,onhowthechilddevelopsboth‘corelinguistic’andmoregeneral‘pragmatic’abilities.Linguistsinthisfieldarguethatchildrenacquirenotaseriesofmoreorlessabstractmentalstructures,butarichsystemforcommunicationwiththosearoundthem.Inparticular,thelinguistCatherineSnowhaschallengedChomsky’scentralclaimaboutthepovertyofinput.RememberthatChomskyarguesthattheremustbeaninnatelanguagefacultybecausechildrencouldneverarriveatthecomplexknowledgeofalanguagejustonthebasisofthefragmentaryevidencetheyreceive.Snowhasarguedthat,onthecontrary,adultstendtoadoptaparticularstyleoftalkingwhentheyareinteractingwithchildrenwhichmakesiteasyforthemtolearnthelanguage.Thisstyle,whichhasbecomeknownas‘ChildDirectedSpeech’(CDSforshort)issimplifiedtomeetthechild’slevel,andisslowandcarefulandfullofrepetitions.Snowclaimsthatevidencefromanempiricalstudyofhowadultsspeaktochildrenisenoughtoremovetheapparentmysterybehindtheprocessoflanguageacquisition.13Wehavelookedatideasabouttherelationshipbetweenlanguageandmindputforwardintheworkofsometwentieth-centuryphilosophers,andseenhowtheyhavecontributedtothedevelopmentoflinguisticsasaseparateacademicdiscipline.Theseideasformedpartofagrowinginterestwithinphilosophyinstudyinglanguageinitsownright.Asweknow,philosophersfromearliercenturieswhodiscussedlanguagewererarelyinterestedinitforitsownsake.Theirworkonlanguagedevelopedoutofinterestsinareassuchasknowledge,science,orlogic.Duringthetwentiethcentury,however,somephilosophersbegantoconcentrateonthebestwaystodescribeandexplainlanguages,aswellasthewaysinwhichlanguageingeneralcouldberelatedtoothertypesofhumanactivity.Thisisnottosaythatlinguisticsoriginatedjustinthisbranchofphilosophy,orindeedjustinphilosophyatall.Aswehaveseen,workwhichprovedimportantinthedevelopmentoflinguisticscamenotjustfromphilosophybutalsofrompsychology,anthropology,andotherdisciplines.Writersfromallthesedifferentbackgroundshaveconfrontedquestionsofhowlanguagerelatestomind,ofwhetheritinfluencesorisinfluencedbythought,oftheextenttowhichpeoplecanbesaidto‘share’alanguage.Somehaveseenlanguageprincipallyasatypeofbehaviour,gearedtowards\n170Languageandmindcommunication;othershaveseenitasatypeofknowledge,existingprimarilyasamentalstateintheindividual.Buttheyhaveallbeeninterestedinthestudyoflanguagebecauseofthepotentialanswersitofferstosuchquestions,andthereforetheinsightsitaffordsintohumannatureitself.Interestinthesequestionswasfundamentalintheestablishmentoflinguisticsduringthetwentiethcentury,andtheyremainsomeofthemostdifficult,butintriguingfocusesofenquiryforpresent-daylinguists.FurtherreadingSignsandstructuresSaussure’sCourseinGeneralLinguistics(1960)isawide-rangingdiscussionofmanyoftheissueswhichweretobecomecentraltomodernlinguistics.HedealswiththenotionofthelinguisticsigninPartI,‘Generalprinciples’.Whorf’sversionoflinguisticrelativityissetoutinthepapersinLanguage,ThoughtandReality(1956).Theintroductiontothis,byJohnB.Carroll,givesasketchofWhorf’sunusualcareer,anddrawsinterestingparallelsbetweenhisprofessionalworkandhisinterestinlinguistics.Linguisticrelativityingeneral,andtheSapir-Whorfhypothesisinparticular,isdiscussedinChapter10ofDevittandSterelny(1987)LanguageandReality.Someofthecriticismswehaveconsidered,includingspecificrejectionsofWhorf’sclaims,aresummarisedinPinker(1995)TheLanguageInstinct,Chapter3.EmpiricismThetwomajorempiricalworksonlanguagewhichwehaveconsideredinthischapterareBloomfield(1935)LanguageandQuine(1960)WordandObject.WealsomadereferencetoSkinner(1957)VerbalBehavior.ThemostfamouscritiqueofSkinner’sworkonlanguageisChomsky(1959)‘AreviewofB.F.Skinner’sVerbalBehavior’.HereChomskycriticisesbehaviourismasspeculativeandunscientific,andtakesissuewiththeideathatanimalexperimentsareunproblematicallyapplicabletohumanbehaviour.Healsointroducessomeoftheideasthatweretoprovecentraltohisownaccount,suchastheeaseandregularitywithwhichchildrenacquirelanguage,theabilityofspeakersofalanguagetoconstructandinterpretnovelsentences,andthecomplexandstructurednatureoflanguage.Quine’sworkdoesn’tmakeparticularlyeasyreading,buthasbeenthesubjectofmanymoreaccessiblecommentaries.R.Martin(1987)TheMeaningofLanguageincludesadiscussionofQuine’s‘radicaltranslation’inChapter6.Radicaltranslationisalsodiscussed,andcontrastedwithChomsky’stheoriesofinnateness,inStainton(1996)PhilosophicalPerspectivesonLanguage,Chapter8.\nLanguageandmind171TheInnatenessHypothesisChomskydevelopedthetheorywhichhasbecomeknownastheInnatenessHypothesisoverseveraldecadesandmanydifferentpublications.In1966hepublishedCartesianLinguistics,whichoutlinestheessentialsofanaccountoflanguageacquisitionintermsofinnatepropertiesofthemind,andtracesitshistoryinrationalistschoolsofthought.Hisinterestinstudyingthehumanmind,andinhypothesisingfromtheevidenceofthesteadystatetotheprobableformoftheinitialstate,issetoutinhis(1980)RulesandRepresentations,particularlyChapter5.Heoffersadetailedaccountofwhathecalls‘Plato’sproblem’,thelogicalproblemoflanguageacquisitioninhis(1986)KnowledgeofLanguage,Chapter3.InthisbookhealsogatherstogetherhisideasaboutI-languageandE-language,whicharepresentedinChapter2.AspectsoftheTheoryofSyntax(1965)containsadetailedaccountofthenatureandfunctionofthelanguagefaculty.MuchhasbeenwrittenabouttheimpactofChomsky’sworkonideasaboutthenatureoflanguageandtheprocessesbywhichitisacquired.DevittandSterelny(1987)LanguageandReality,Chapter8,discussChomsky’saccountoflinguisticcompetence,anditsrelationshiptohistheoriesofinnateness.Pinker(1995)TheLanguageInstinctisapopularandaccessibleintroductiontotheinnatenesshypothesisanditsimplications,althoughhisviewsandapproachdifferinplacesfromChomsky’s.JerryFodor’sideasaboutaninnate‘languageofthought’whichisnecessarilypriortotheacquisitionofanynaturallanguage,togetherwithsomediscussionofChomsky’swork,canbefoundinhis(1976)bookTheLanguageofThought.EmpiricalandmentalistlinguisticsApopularandveryreadableintroductiontopsycholinguisticsingeneral,withparticularemphasisonlanguageacquisition,isAitchison(1989)TheArticulateMammal.Thestorage,retrievalandrecognitionofwordsarediscussedinanotherofAitchison’sbooks,her(1987)WordsintheMind.OfparticularinterestisChapter2,inwhichshediscussestheevidencepresentedbyslipsofthetongueandothersources.ManytextbooksandintroductionstolinguisticsincludeoutlinesoftheconsequencesofChomsky’slinguistics,eveniftheyarenotcreditedtoChomsky,ordirectlylinkedtohiswork.See,forinstance,thefirstchaptersofR.Martin(1987)TheMeaningofLanguage,andO’Gradyetal.(1997)ContemporaryLinguistics.SmithandWilson’s(1990)ModernLinguisticsisentirelydevoted,asthesubtitleexplains,to‘theresultsofChomsky’srevolution’.\n172LanguageandmindAgoodintroductiontocorpuslinguistics,whichdiscussestheempiricistapproachtolanguagewhichunderpinsit,isMcEneryandWilson(1996)CorpusLinguistics.Chapter1inparticulardiscussestheframeworkwithinwhichcorpuslinguisticsdeveloped,anddistinguishesitfromthementalistapproachtolinguistics.Harris’‘integrationist’accountoflanguage,andhiscritiquesofmanyofthephilosopherswehaveconsideredhere,canbefoundinhis(1996a)TheLanguageConnectionand(1996b)Signs,LanguageandCommunication.Anextendedstudyofonecaseofcognitiveimpairment,presentedinsupportofaninnatistaccountoflinguisticknowledge,isSmithandTsimpli(1995)TheMindofaSavant.Goodluck(1991)LanguageAcquisitionisafairlytechnicalbutaccessibleintroductiontothesubjectfromaninnatistperspective.Incontrast,asmentionedinthe‘furtherreading’sectionofthelastchapter,Bates(1976)LanguageandContextconsiderstherelationshipbetweenthedevelopmentoflanguageandthedevelopmentofthechild’spragmaticabilities,includingtheappropriateuseofspeechacts.FletcherandGarman(1986)LanguageAcquisitionisausefulcollectionofessays,includingworkonanumberofdifferenttopicsandfromanumberofpointsofview.Inthis,CatherineSnow’sarticle‘Conversationswithchildren’,describesandillustratesCDS,andassessesthechallengeitposestotheInnatenessHypothesis.\nGlossaryThefirstoccurrenceofthesetermsishighlightedinboldtypeinthemaintext.AnalyticAnanalyticsentenceislogicallyanecessarytruthbecausethepredicateiscontainedwithintheconceptofthesubject;itistruebyvirtueofitsintrinsicproperties,regardlessofcontext.Oftencontrastedwithsyntheticsentences.AposterioriAdescriptionofknowledgewhichisderivedfrom,andlogicallyfollows,experienceofthesenses:knowledgewhichisempirical.Oftencontrastedwithaprioriknowledge.AprioriAdescriptionofknowledgewhichisnotderivedfromexperience,butisindependentofandpriortoobservationofreality.Usuallydiscussedinoppositiontoaposterioriknowledge.BehaviourismApsychologicalschoolofthoughtwhichconcentratesondescribingandexplainingobservablebehaviourratherthanpostulatingunobservablementalstates.Appliedtolanguagestudy,behaviourismentailsthatthemeaningofanyindividualutterancecanbedefinedonlyintermsoftheeffectsithas,andthatlanguagelearningmustbeachievedintermsofaseriesofstimuli,responsesandreinforcements.Oftencontrastedwithmentalism.ConditionalAlogicalexpressionequivalentto‘ifp,thenq’,whichistrueexceptwhenpistrueandqfalse.Expressedinpropositionallogicasp®q.ConjunctionAtermoflogicforthejoiningtogetheroftwopropositions,symbolisedasp?qintheformp&q.Thepropositionspandqareinthiscontextdescribedasconjuncts.Logicalconjunctionistruth-functional;joiningtwotrueconjunctsgivesatruestatement,whilejoiningonefalseandonetrue,ortwofalseconjuncts,givesafalseone.ConnotationTheconnotationofawordorphraseisthesetofpropertieswhichdeterminewhatisreferredto;itisdependentnotonexternalrealitybutontheindividualwordorwords.Oftencontrastedwithdenotation.173\n174GlossaryConstativeTypeofspeechactidentifiedbyJ.L.Austin.Anutteranceusedtomakeastatementoffact,whichcanbejudgedtobeeither‘true’or‘false’.Oftencontrastedwithperformativespeechacts.ContingentAsentenceislogicallycontingentifitisneithernecessarilyfalsenornecessarilytrue,butdependsonthenatureofexternalrealitytobeassignedatruth-value.Acontingentlyfalsesentencecouldbetrueindifferentcircumstances;acontingentlytruesentencecouldbefalse.ContradictionTwoexpressions,oneofwhichentailsapropositionpandtheotherofwhichentailsthenegationofparesaidtobecontradictory,becausetheycannotbothbetrueorbothfalse.Asingleexpressioncanbedescribedasacontradiction,orasself-contradictory,ifitentailsbothpandthenegationofp,andisthereforenecessarilyfalse.CorrespondenceAccordingtoacorrespondencetheoryoftruth,astatementistruejustincaseitcorrespondstoreality,ortothewaytheworldactuallyis.DeductiveThemethodofscientificinvestigationwhichstartsfromageneralhypothesis,andusestheavailabledatatoconfirm,tomodifyorcompletelytofalsifythis.Itisoftencontrastedwiththeinductivemethod;bothareusedinmodernlinguistics.DenotationThedenotationofawordorphraseistheobjectorobjectstowhichitrefers.Thetermisoftenrestrictedtotherelationshipbetweenasingularterm(generallyapropernameordefinitedescription)anditsreferent.Itcanbecontrastedwiththeconnotation.DisjunctionAtermoflogicsimilartonaturallanguageor,whichlinkstwopropositions,describedinthiscontextasdisjuncts,intheformp?q.Theexpressionasawholeisfalseonlyifbothdisjunctsarefalse;inallothercasesitistrue.E-LanguageAtermusedinthelaterworkofChomskytodescribetheobservableperformanceofspokenandwrittenutterances.ItiscontrastedwithI-Language,therealfocusofinterestinthestudyoflinguisticknowledge.EmpiricalEmpiricalknowledgeisthatbasedonexperience,oronevidenceoftheworldprovidedbythesenses.Anempiricalmodeofstudyisonewhichinvolvesperceptionandanalysisofdata.EntailmentIfpropositionpentailspropositionq,thenqcanbesaidto‘followlogically’fromp,arelationshipwhichisgenerallyrepresentedasp?q.Ifpistruethenqmustbetrue,butifpisfalsewecansaynothingaboutthetruthvalueofq.ExtensionTheextensionofareferringexpressionistheobjectorpropertyreferredto.Theextensionof‘thepresentPrimeMinisterofBritain’istheactualpersonwhocurrentlyholdsthatoffice.Extensionalityissometimesalsoappliedtosentences;theextensionofasentenceissimplyitstruth-value.Oftencontrastedwithintension.HolismTheideathatanycomplexwholeconsistsofmorethanthesumofitsparts,andthereforethatnosinglepartcanbefullyexplainedwithout\nGlossary175referencetothewholetowhichitbelongs.Semanticholismclaimsthatnoindividualexpressionofalanguagecanbedefinedinisolation;itsmeaningdependsonthemeaningofeveryotherexpressioninthelanguage.IdealFormAconceptfromthephilosophyofPlato.Wordsproperlyrefertoidealorperfectversions,whetherofobjects(‘man’,‘dog’)orofqualities(‘justice’,‘beauty’).Actualobjectsandqualitieswhichweencounterintheworldare‘namedafter’theseidealsbecausetheysharecertainpropertieswiththem,althoughtheycanneverattainthesamelevelofperfection.Platonicidealformsarerealbutabstract.IdeationalTheoryAnaccountofmeaningwhichclaimsthat,sincewecanhavedirectaccessonlytoourideasofobjects,andnottotheobjectsthemselves,ourwordsmustrefertoourideas.SuchaccountscanbetracedbacktotheworkofAristotle,andarefoundin,forinstance,theworkofJohnLocke.IffUsedinlogic,anabbreviationfortherelationship‘ifandonlyif.I-LanguageAterminthelaterworkofChomskytodescribethementalstatewhichformsthespeaker’slinguisticcompetence,orknowledgeofalanguage.ContrastedwithE-Language.IllocutionaryactThesecondofthethreeactsidentifiedinthelaterworkofJ.L.Austinasmakingupaspeechact.Atthislevelitisnecessarytotakeaccountoftheintentionofthespeakerinproducingtheutterance,ortheillocutionary‘force’oftheutterance.InductiveThescientificmethod,usedforinstanceinempiricallinguistics,whichinvolvesobservingtheavailableempiricalevidence,andconsideringwhatgeneralconclusionsarelicensedbythese.Contrastedwiththedeductivemethod.InnatenessHypothesisTheterm‘innatist’isappliedtoanyaccountofthemindfromPlatoonwardswhichpositscertaininnate,orgeneticallyinherited,mentalfacultiesorformsofknowledge.TheparticularhypothesiswasdevelopedbyChomskyandothersinthemid-twentiethcentury;itscentralclaimisthatthereisauniquefacultywhichgovernstheacquisitionandknowledgeoflanguage.IntensionTheintensionofawordorphraseisthesetofpropertieswhichdetermineitscorrectapplication.Whiletheextensionof‘thetallestmanintheworld’iswhicheverindividualfitsthatdescriptionatanygiventime,itsintensionisthepropertyofbeingthetallestmanintheworld.Whenthetermisappliedtosentences,theintensionissaidtobethethoughtorpropositionexpressed,ratherthantheactualtruth-value.LocutionaryactThefirstofthethreelevelsofactwhichconstitutethedefinitionofspeechactsinthelaterworkofJ.L.Austin.Theactofutteringasequenceofwords,togetherwithanidentificationofwhattheyliterallymean,includingtheappropriatesenseandreference.\n176GlossaryLogicalpositivismAbranchofpositivistphilosophywhichdevelopedinEuropeintheearlypartofthetwentiethcentury,andisparticularlyassociatedwiththephilosophersoftheViennaCircle.Likepositivismgenerally,itisanessentiallyempiricalapproach.Thepropositionswhichitdescribesasmeaningfularethosewhichexpressanalytictruths,thoseoflogicandmathematics,andthosewhichcanbesubjecttoaprocessofverification.Allotherpropositions,notablythoseofmetaphysics,religionandaesthetics,aretreatedassenseless,ormeaningless.MentalismTheschoolofthoughtwhichstipulatesthatacompleteaccountofanyobservablebehaviourmustincludereferencetotheunderlyingmentalprocesses.Mentalistaccountsoflanguagehaveconcentratedonunderstandingandexplainingaspeaker’sknowledgeof,orcompetencein,alanguage,ratherthantheactualperformanceofindividualutterances.MetaphysicsAtermappliedtoanyformofstudywhichgoesbeyondthedescriptionofmaterialreality,andconsiderstheexistenceofnon-physicalentities,orrealitywhichliesoutsideofnormalsensoryexperience.Thevalidityofmetaphysicsasatypeofphilosophicalinvestigationhasbeenchallengedby,amongothers,empiricistsandlogicalpositivists.ModalityInlogic,expressionscontainingmodalityareconcernedwithstatementsofnecessityandpossibility.Necessarypropositions(?p),possiblepropositions(?p)andactualpropositions(p)canbedefinedintermsofeachotherandofnegation.NativismAnothername(besidesinnatism)forthetheorythatsomeformsofknowledgeorofthoughtareinnate,andhencethatnotallknowledgeandunderstandingisderiveddirectlyfromexperience.NegationAlogicaloperator,similartotheuseof‘not’inordinarylanguage,whichmapsatruepropositionontoafalseoneandviceversa.Thelogicalsymbolforthenegationofpis~p.OpaquecontextsAlsodescribedasintensionalcontexts.Formulaeinwhichsubstitutionofoneexpressionbyanotherwithwhichitisextensionallyequivalentdoesnotnecessarilyresultinthetruth-valueofthewholeremainingunchanged.Modalsentencesandprepositionalattitudesaretypesofopaquecontext.OrdinarylanguagephilosophyAschoolofthoughtwhichflourishedinOxfordinthemid-twentiethcentury.Thephilosophersinvolvedwereallinterestedinnaturallanguage,andparticularlyinlanguageuse,asalegitimatefieldofstudyinitsownright.Theyrejectedtheideathatlinguisticmeaningcouldbeadequatelydefinedintermsoflogicalrelations,andoftruth-conditionalsemantics.PerformativeAtypeofspeechactidentifiedbyJ.L.Austininanearlyphaseofhisworkinthisarea.Theveryactofutteringaperformativecan,inappropriatecircumstances,countasaperformanceoftheactapparentlybeingdescribed,andbringaboutcertainstatesofaffairs.Performatives\nGlossary177aren’tappropriatelylabelled‘true’or‘false’,butcanonlybe‘felicitous’or‘infelicitous’.PerlocutionaryactThelastofthethreeactsidentifiedbyJ.L.Austinasmakingupaspeechact.Itisconcernedwiththeresultorconsequenceoftheutterancehavingbeenproduced;inAustin’sterms,itiswhatthespeakerbringsaboutbysayingsomething.PossibleworldsAconceptintroducedintoformalsemanticstodealwithvariousproblemsofreference,andusedinonedefinitionofmodality.Possibleworldsemanticsrecognisesthatthingscouldbeotherthantheyinfactare,andpositsaseriesofworlds,orstatesofaffairs,otherthanwhatisactuallythecase.PresuppositionOriginallyusedinworkbyFregeandStrawsontodescribealogicalrelationshipbetweenpropositions,thetermhassincebeenusedinlinguisticstodescribe,variously,relationshipsbetweensentences,words,utterancesandspeakers.Inthecaseoflogicalpresuppositionbetweentwopropositions,whereppresupposesq,thetruthofqisgenerallytakentobeanecessaryconditionforthetruthorfalsityofp.Ifqisfalse,pissaidtobeneithertruenorfalse,makingthisrelationshipinconsistentwithclassical,two-valuedlogic.PropositionMostcommonlyinrecentphilosophyoflanguage,thistermisusedtorefertoanon-linguisticunitofmeaning.Apropositioncanbetheobjectofthought,belief,etc.Itistheintensionofadeclarativesentence.PropositionalattitudeAsentencedescribingsomeone’sbelieving,desiring,doubting,etc.,thatacertainpropositionisthecase,isdescribedasbeingconcernedwithapropositionalattitude.Suchsentencesareonetypeofopaquecontext.QuantifierInlogic,anoperatorusedtomakestatementsoversetsofentities.Themostcommonquantifiersaretheexistentialanduniversalquantifiers.Theexistentialquantifierisrepresentedbythesymbol?.(?x)canbereadas‘thereisatleastoneentityx’.Theuniversalquantifierisrepresentedby?.(?x),canbereadas‘forallentitiesx’,‘ofeveryx’.ScepticismThebeliefthatnoknowledgecanbecertainandthatwecanatbestmakejudgementsofprobability.Theviewwasparticularlyassociatedwiththeempiricalapproachtoknowledge;wehaveaccessonlytooursenseexperienceandcan’tbesurethatthisrelatestoanyexternalreality.SyntheticAdescriptionofasentenceinwhichthepredicateislogicallyindependentofthesubject.Thetruthorfalsityofsuchsentencescanthereforeonlybeascertainedwithreferencetoexternalreality.Contrastedwithanalytic.TautologyAstatementwhichwillalwayshavethetruth-valueof‘true’;anecessarytruth.TruthconditionThesemanticsofasentenceareoftenequatedwithitstruth\n178Glossaryconditions.Thesespecifythestateofaffairswhichwouldhavetoholdforthesentencetobetrue.Truth-functionAnylogicalconnectivewhichcombinestwoormorepropositionstoformacompoundproposition,inwhichthetruthofthecompoundisdependentonthetruthofthesimplepropositions,issaidtobetruth-functional.Theconnectivesinvolvedareknownasoperators.T-sentenceAsentencestatingthetruth-conditionsforasentenceinanylanguagebeingstudied(‘objectlanguage’)andhavingtheform:ThesentenceSistrueinLiffP’.Tarskiclaimedthatyouwouldhaveanadequateaccountoftruthforalanguage(L)ifyoucouldstipulateaT-sentenceforeachsentenceinL.UniversalGrammarAseriesofruleswhichaccountforlinguisticuniversals,andtherebyplaceconstraintsontheformwhichalanguagecantake.IntheworkofChomskyandothers,UniversalGrammar(UG)issaidtoexplainthestrikingsimilaritiesbetweenallhumanlanguages.Itissaidtobeinnateinallhumanminds,andtoexplaintheeaseandtheuniversalregularityoflanguageacquisition.VerificationAmethodologyadoptedbythelogicalpositivistsforidentifyingmeaningfulpropositions.Tobemeaningful,anypropositionwhichisnoteitheranalyticorbelongingtomathematicsorlogicmustbecapableofbeingverified;theremustbeaprocessbywhichitstruthcanbeassessedbymeansofempiricalobservation.\nNotes1Wordsandthings1Asweshallsee,notallphilosophershaveacceptedthatpropernamessuchasthesehaveintensionsatall,butwewillassumeforthetimebeingthattheycanbetreatedjustthesameasdescriptions,suchasthechapintheConservativeclub.2ItistraditionaltogivereferencestoPlato’sworknotbypagenumberintheparticulareditionused,butbymeansofthepagenumbertothe1587editionpublishedinParis.Inthis,thepageswerefurtherdividedintofiveparts,indicatedbylowercaseletters.Inmostmoderneditionsthepagenumberingandletteringaregiven,providingauniformwayofreferringtoPlato’swork.TheeditionusedhereisJowett(1964).3Again,therearespecialconventionsforreferringuniformlytoAristotle’sworks,ratherthanrelyingontheindividualeditionused.Themethodusedhereisperhapsthesimplestoftheseconventions.WorkssuchasdeInterpretationearedividedintoshortchapters,andthenumberofthechapterisgivenasthereference.TheeditionusedhereisBarnes(1984).4ReferencestoAnEssayConcerningHumanUnderstandingaregivenbybook,chapterandnumberedparagraph.5ReferencetoPrinciplesofHumanKnowledgeisbyparagraphnumber.Theintroductionandthemaintextarenumberedseparately.6LeibnizwroteallhisbookseitherinLatinor,asinthiscase,inFrench.7AtthetimeinwhichMillwaswriting,ofcourse,thewordmanwasuseduncontroversiallyinthegenericsenseof‘memberofthehumanrace’or‘person’.ItisstrikingforamodernreadertofindMillusingMaryasanexampleofanindividual‘man’,butitmayhavebeenadeliberatelyinclusivegestureonhispart.Millhasbeendescribedasanearlyfeminist,andarguedinprintthatmenandwomenshouldhaveequalfreedoms.8Morerecently,forinstanceinMartinich(1996),Frege’sarticlehasappearedas‘Onsenseanddenotation’.Thisisquiteahelpfulterminology,makingitclearthatBedeutungcorrespondstosomethinglikewhatwehavebeendescribingasdenotation,butitsuseisnotwidespreadindiscussionsofFrege’swork.9Frege,orhistranslator,ishereusingthetermdesignateinthewayinwhichwehavebeenusingdenote.Somewritersdistinguishbetweendesignateforasingulartermanddenoteforageneralterm,butitisalsocommonpracticetousedenoteforboth,aswearedoinghere.10Thethreepairsoftermsarenotinfactexactlyequivalent,butweneednotworryhereabouttheprecisedifferencesbetweenthem.ThesearediscussedindetailinCarnap(1957)esp.124–9.179\n180Notes11Theratherneutralterm‘convey’isusedheredeliberately;wearepostponingdiscussionofthenatureofthisrelationshiptothenextchapter.Russellseesitasarelationofentailment;others,suchasPeterStrawson,havedescribeditasoneofpresupposition.12LikeFregeandRussellbeforehim,Kripkedefineshisownparticularuseoftheterm‘name’.Hespecifiesthatheisusingitjustfor‘propernames’:John,Hull,England,etc.13PagereferencesaretothereprintedarticleinMartinich(1996).14Notethatinthiscasecontextsinvolvingdirectquotationwouldnotbeappropriatetests;ifGussaid‘I’mabitofalad’,thenitiscertainlynottruethatGussaid‘I’mabitofaboy’,butthistellsusnothingaboutthemeaningsofladandboy.Prepositionalattitudesprovideamoresatisfactorytest.2Propositionsandlogic1Forpresentpurposes,weshallignoretheproblemthatitmayinfactneverbepossibletoproduceanexacttranslationfromonelanguagetoanother.2Wemightclaimthatwecanimaginesayingthis,orhearingitsaid,insuchawaythatwewouldwanttoarguethatitwastrue,oratleastnotnecessarilyfalse.Forinstance,inaparticularcontextitmightbeusedtomeansomethinglike‘Aliceisonlytenbutbehaveslikeanadult’.ThisissomethingwillreturntoinChapter4.3Itiscustomarytocitejustthepropertybeingascribedbythepredicate(‘threebedrooms’,‘rotten’)astheinstantiationofthevariable.‘Grammatical’wordssuchashas,is,etc.,tendtobeomittedonthegroundsthattheyarerequiredsimplybythegrammarofthelanguageanddon’taddanythingtothelogicalstructureoftheproposition.4Thereisacomplicationwhichweneedn’tworrytoomuchabouthere.It’snotactuallythecase,ofcourse,thateveryentityhasafather,althoughitiscertainlythecasethateveryentitywhichishumanhasafather.Tobeaccurateweshouldperhapsaddthistoourlogicalformula,byintroducingthepredicate‘H’for‘human’andspecifyingthatifanentityishumanthenithasafather:(?y)(Hy?((?x)(Fxy)).5InFrege’sexamplethepresuppositionisactuallyoneofexistenceinthepast.Mostexamplesofexistentialpresuppositionswhichhavebeendiscussedintheliteratureinvolvepresentexistence.Thedistinctionisduetothetenseoftheverb,nottoanythingconcerningthenameitself,andsoneednotconcernushere.6PagereferencesaretoGrice(1989).7Forafairlyextensivelistofthese,seeLevinson(1983)181–4.3Truthandreality1Theterms‘theworld’and‘reality’areherebeingusedinterchangeably.Thisreflectsgeneralphilosophicalandlinguisticusage.ScepticismofthetypeweconsideredinChapter1aside,itisgenerallyassumedthatwhatwecanseearoundusmakesuptheworldinwhichwelive,andcountsaswhatweknowofreality.Similarly,theterm‘universe’isoftenusedinthesameway.2Thatisnot,ofcourse,tosaythatandexamplelike2)can’tbeusedsuccessfullytoconveyanideawhichisnotnecessarilyfalse.Amemberofthebank’sexecutiveboardmightuse2)tosuggest,forinstance,thatBraeburn-Twinsett’srecentdomesticproblemshavemadehimunfittocontinueinhisjob,becauseheisnolongerabletoconcentratesufficiently.Thisisn’tafactaboutthesemantic,\nNotes181truth-conditionalmeaningofthesentence,butaboutthespecificsignificanceofanutteranceofitincontext.Wewillconsiderapproacheswhichhavebeensuggestedtothistypeofprobleminthenextchapter.3Insayingthis,wehavetoassumethatanysituationinwhichweconsiderthetruth-valueof6)isoneinwhichthewordsthirty-fourandtwentymeanthesameasweunderstandthemtomean.Itisofcoursepossiblethatinsomesciencefictionworldthesewordsmightmean‘eighteen’and‘ninety-five’respectively.Inthatcase,tosay6)wouldbetosaysomethingfalse,butwewouldn’tbemakingajudgementoftruth-valueaboutthesameproposition.4Thewordcorrespondisperhapsnotasself-explanatoryasitmayatfirstseem.Indeed,oneofthecriticismswhichhasbeenlevelledatthistypeofaccountisthatitisbynomeansclearwhatitwouldmeanforapropositiontocorrespondwithastateofaffairs.5ItiscustomarytogivereferencestotheoriginalversionoftheCritiqueofPureReason,whateverindividualtranslationisused.Pagenumbersareprefacedwith‘A’forthefirsteditionand‘B’forthesecondedition.Inthisparticularcase,thepassagequotedappearsinthesecondbutnotinthefirstedition.6Itisinfactthisdoctrinewhichgainedthemthename‘logicalpositivists’.Theyweredistinguishedfromthepositivistsofthenineteenthcenturybytheirinclusionofmathematicalandlogicalstatementsasmeaningful.7PagereferencesaretothereprintedversionofTarski’sarticleinMartinich(1996).8Pagereferencesareto‘Truthandmeaning’,inMartinich(1996).9Thereis,ofcourse,animportantdistinctiontobedrawnbetweensentencetypesandtheusestowhichtheycanbeput.Asjustoneexample,declarativesdon’talwaysstatefacts;theycanbeusedtogiveorders,issuethreats,elicitinformationandmanyotheruses.ThisdistinctionisoneofwhichDavidsonisaware,andwhichheinfactdrawsin‘Moodsandperformances’.Forthetimebeingwewilluseinterrogativesandimperativesasourexamplesoflanguageusedtodosomethingotherthanstatefacts,butwewillfocusonthisimportantdistinctioninthenextchapter.10fcourse,eventhisseeminglyinnocuousdefinitionisitselfcontroversial,becauseofthesortofobjectionswhichhavebeenputforward,whichweconsideredinChapter1,totheideathatwecaneverbeconfidentaboutthenatureof‘howthingsactuallyare’,apartfromourownperceptionofit.11Leibniz’sownaccountthereforeimpliesthat,asweconsideredabove,theterm‘world’caninthiscontextbeusedinterchangeablywith‘universe’.12LewisarguesinCounterfactualsthatthistypeofdifferencebetweenworldswouldneverinfactbepossible.TranslatingLewis’argumenttoourpresentcase,wemightsaythatitwouldbeimpossiblefortwoworldstodifferjustinthecolourofsnow;bluesnowwouldalsoimplyadifferenttypeofwaterfromthatwhichwehave,orperhapsdifferentlawsofphysics.Weneednotworrytoomuchaboutthishere.4Speakersandhearers1Incomingupwithatruth-conditionalaccountof1),wewouldhavetoconfronttheproblempresentedbyhere.Itsmeaning,theactualplacereferredto,isentirelydependentoncontext.Thisisanexampleofwhatisknownasdeixis:anexpressionwhichservesto‘pointat’,orsingleout,someaspectofcontext.Deixishasbeenextensivelystudiedinlinguistics,andisanotherfactorwhichhasbeen\n182Notesusedtoargueagainsttheadequacyofapurelytruth-conditionalaccountofmeaning,butisn’tsomethingwewillbelookingathere.2See,forinstance,RichmondThomason’sintroductiontoRichardMontague’sFormalPhilosophy,41.3ReferencetoPhilosophicalInvestigationsisbyparagraphnumber.4TheexamplesofwordswhichRylegivesincludecause,infinitesimalandremorse.5Throughoutthischapter,wewillbeadoptingtheconventionwhichhasnowbecomefairlystandardpracticeinpragmatics,ofreferringtothespeakerofanyexampleassheandthehearerashe.Thissystemhastheadvantageofavoidingthetypeofconfusionwhichcanoccurwhenonegenericpronounisusedtorefertoboth.Italsoservestogivewomena‘voice’indiscussionsoflanguageuse;somecriticismsofaccountssuchasAustin’sandGrice’shavefocusedontheirimplicitassumptionthatparticipantsininteractionare,typically,male.6Searleinfactusestheterm‘preparatoryconditions’inplaceofAustin’s‘felicityconditions’.7PagereferencesforallofthearticlesbyGricecitedinthischapteraretoGrice(1989).8Thisexampleisused,toillustratethispoint,byStephenLevinsoninhisbookPragmatics(1983).9Anexclamationmarkatthebeginningofanexamplesentenceisaconventionalwayofindicatingthat,althoughitmaybegrammaticallycorrect,itisinsomeway‘unacceptable’intermsofmeaning:forinstance(asinthisexample)thatitislogicallycontradictory.10Afairlycommonreactiontothisexampleistosuggest,asanalternativeexplanation,thatBmaybehintingthatSmithdoesn’thaveagirlfriendbecausehehasbeengoingtoNewYorkalot.Inotherwordstheimplicatureintendedmaybethatheistoobusyformuchofasociallifeatthemoment.Thisisn’tapossibilitywhichGriceseemstorecognise,butitisneverthelessonewhichcanalsobeexplainedwithreferencetotheco-operativeprinciple,andparticularlythemaximofrelevance.Thealternativeinterpretationsservetohighlightthewayinwhichparticularlisedconversationalimplicaturesarenotattachedtotheactualformofwordsused,butdependentonvariousfactorsinthewidercontext.5Languageandmind1Wewon’t,forinstance,belookingatwhathasbeenwrittenabouthowweprocessandunderstandlanguagewhenweencounteritinspeechorwriting.Issuessuchasthisare,however,highlyimportantinmodernlinguisticsandpsycholinguistics,andreferencetosomeintroductoryworkcanbefoundinthe‘furtherreading’sectionofthischapter.2See,forinstance,JohnCarroll’sintroductiontoLanguage,ThoughtandReality,22.3Bloomfieldhimselffavoursthetermmechanistic,butmaterialisticismorewidelyused,especiallyincontrasttomentalistic.4Or,accordingtoQuine,‘Lo,arabbit’.ThecapitallettershereindicatethattheseutterancesareofwhatQuinecalls‘onewordsentences’,asopposedtosimply‘words’.Wewillobservethenotation,butneedn’tworrytoomuchaboutthedistinction.\nNotes1835Chomskyhasalwaysmaintainedthathisworkonpoliticsandthatonlinguisticsaretwocompletelydifferentinterestsbut,perhapsinevitably,peoplehavetriedtofindconnectionsbetweenthetwo.Perhapsthemostsuccessfuloftheseattemptsistoseethemasbothdependentontheideaofa‘sharedhumanity’.Inhislinguisticsthismanifestsitselfinhisinterestinaninnate,humanlanguagefaculty.Inhispoliticsitisevidentinhiscampaigningforequalityandhumanrights.6WorkinthisareahasbeendonebyotherswithintheframeworkChomskyestablished.Seethe‘furtherreading’sectionattheendofthischapterforsomereferencestothiswork.7From‘ReplytoPutnam’(1969)302.8Theinnatistaccountincludesa‘criticalperiodforlanguageacquisition’,aperiodinthechild’sdevelopmentwhichnaturallyendsaroundtheonsetofpuberty.This,itisclaimed,explainsthefairlywidelyacceptedfactthattheabilitytoacquirelanguagesdeterioratesataboutthistime.9Theparticularissueofwhetherlanguagecanbesaidtobeuniquetohumanshasbeenthesubjectoflengthyandheateddebate.Foradiscussionoftheextenttowhichnon-humananimalscanbesaidto‘havelanguage’,seeChapter16ofO’Gradyetal.(1997)ContemporaryLinguistics.Variousattemptshavebeenmadetoteachanimals,particularlyapes,to‘uselanguage’,andvaryingdegreesofsuccessclaimed.Foranoverviewofthisdebate,andadiscussionofitsimplications,seeJeanAitchison(1989)TheArticulateMammal,Chapter2.10OtherargumentshavebeenputforwardinsupportoftheIHwhichwehavenotconsideredhere.Forinstance,itappearsthatwhen,inparticularcircumstances,twogroupsofmonoglotspeakersfindthattheyneedathird,commonlanguage,afullycomplexandrule-governedlanguage,knownasaCreole,appearsbythesecondgeneration.SeePinker(1995)TheLanguageInstinct,Chapter2,foranoutlineofthesignificanceofCreolestotheIH.11PagereferencesaretothearticlereprintedinBlock(1981)292–9.12ChomskynotesthathistermscanbelikenedtoSaussure’slangueandparole,butarenotidentical.Structuralistgrammarisamatteroflisting,notgenerating,sentences.13SomesubsequentdiscussionsofCDShaveinfactpresentedtheoppositepointofview.Specifically,theyhaveclaimedthatCDSisspecifictocertain,Westerncultures;itisn’tpractisedatallinmanyothercultures.\nReferencesAitchison,Jean(1987)WordsintheMind,Oxford:Blackwell.——(1989)TheArticulateMammal,3rdedn,London:UnwinHyman[1stedn1976].Allwood,Jens,Andersson,Lars-GunnarandDahl,Osten(1979)LogicinLinguistics,Cambridge:CambridgeUniversityPress.Austin,JohnLangshaw(1961)‘Performativeutterances’,inJ.L.Austin(1979)PhilosophicalPapers,3rdedn,Oxford:OxfordUniversityPress.ReprintedinA.P.Martinich(ed.)(1996)ThePhilosophyofLanguage,3rdedn,Oxford:OxfordUniversityPress[1stedn1985]120–9.——(1962)HowtoDoThingswithWords,Oxford:OxfordUniversityPress.Ayer,Alfred(1971)Language,TruthandLogic,Harmonsworth:Pelican[1stednGollancz1936].——(1976)TheCentralQuestionsofPhilosophy,Harmondsworth:Pelican[1stednWeidenfeldandNicolson1973].Barnes,Jonathan(ed.)(1984)TheCompleteWorksofAristotle,2vols,Princeton:PrincetonUnivesityPress.Bates,Elizabeth(1976)LanguageandContext:TheAcquisitionofPragmatics,NewYork:AcademicPress.Berkeley,George(1988)PrinciplesofHumanKnowledge,Harmondsworth:Penguin[1stednDublin1710].Blakemore,Diane(1994)UnderstandingUtterances,Oxford:Blackwell.Block,Ned(ed.)(1981)ReadingsinthePhilosophyofPsychology,Vol.2,London:Methuen.Bloomfield,Leonard(1935)Language,London:GeorgeAllen&Unwin[1stAmericanedn1933].Brown,PenelopeandLevinson,Stephen(1978)‘Universalsinlanguageuseage:politenessphenomena’,inE.Goody(ed.)QuestionsandPoliteness:StrategiesinSocialInteraction,Cambridge:CambridgeUniversityPress,56–311.Brown,GillianandYule,George(1989)DiscourseAnalysis,Cambridge:CambridgeUniversityPress.Burton-Roberts,Noel(1989)TheLimitstoDebate,Cambridge:CambridgeUniversityPress.Cann,Ronnie(1993)FormalSemantics,Cambridge:CambridgeUniversityPress.Carnap,Rudolf(1937)TheLogicalSyntaxofLanguage,trans.AmetheSmeaton,London:KeganPaul[1stGermanedn1934].——(1956)MeaningandNecessity,Chicago:UniversityofChicagoPress[1stedn1947].Caton,Charles(ed.)(1963)PhilosophyandOrdinaryLanguage,Urbana:UniversityofIllinoisPress.Chomsky,Noam(1957)SyntacticStructures,TheHague:Mouton.184\nReferences185——(1959)‘AreviewofB.F.Skinner’sVerbalBehavior’,Language,35,26–58.ReprintedinJerryFodorandJerroldKatz(eds)(1964)TheStructureofLanguage,NewJersey:Prentice-Hall,547–78.——(1965)AspectsoftheTheoryofSyntax,Cambridge:MITPress.——(1966)CartesianLinguistics,NewYork:Harper&Row.——(1969)‘Quine’sempiricalassumptions’,inDonaldDavidsonandJaakkoHintikka(eds)WordsandObjections,Dordrecht:Reidel,53–68.——(1972)StudiesonSemanticsinGenerativeGrammar,TheHague:Mouton.——(1980)RulesandRepresentations,Oxford:Blackwell.——(1981)‘ReplytoPutnam’,inN.Block(ed.)(1981)ReadingsinPhilosophyofPsychology,Vol.2,London:Methuen,300–4.——(1986)KnowledgeofLanguage,NewYork:Praeger.Clarke,D.S.Jr(1987)PrinciplesofSemiotics,London:RoutledgeandKeganPaul.Davidson,Donald(1967)‘Truthandmeaning’,Synthese,17,304–23.ReprintedinDavidson(1984)InquiriesintoTruthandInterpretation,Oxford:ClarendonPress,17–36.ReprintedinA.P.Martinich(ed.)(1996)ThePhilosophyofLanguage,3rdedn,Oxford:OxfordUniversityPress[1stedn1985]92–103.——(1979)‘Moodsandperformatives’,inAvishaiMargalit(ed.)MeaningandUse,Dordrecht:Reidel,9–20.——(1984)InquiriesintoTruthandInterpretation,Oxford:ClarendonPress.Derrida,Jacques(1988)LimitedInc,Evanston:NorthwesternUniversityPress.Devitt,MichaelandSterelny,Kim(1987)LanguageandReality,Oxford:Blackwell.Donnellan,Keith(1966)‘Referenceanddefinitedescriptions’,PhilosophicalReview,75.Dummett,Michael(1993)TheSeasofLanguage,Oxford:OxfordUniversityPress.Eagleton,Terry(1983)LiteraryTheory,Oxford:Blackwell.Evans,Gareth(1982)TheVarietiesofReference,Oxford:OxfordUniversityPress.Evnine,Simon(1991)DonaldDavidson,Cambridge:PolityPress.Fletcher,PaulandGarman,Michael(eds)(1986)LanguageAcquisition,2ndedn,Cambridge:CambridgeUniversityPress[1stedn1979].Fodor,Jerry(1976)TheLanguageofThought,Hassocks:TheHarvesterPress[1stAmericanedn1975].Frawley,William(1992)LinguisticSemantics,NewJersey:LawrenceErlbaumAssociates.Frege,Gottlob(1892)‘Onsenseandmeaning’,inPeterGeachandMaxBlack(eds)(1980)TranslationsfromthePhilosophicalWritingsofGottlobFrege,Oxford:Blackwell[1stedn1952]56–78.ReprintedinA.P.Martinich(ed.)(1996)ThePhilosophyofLanguage,3rdedn,Oxford:OxfordUniversityPress[1stedn1985]as‘Onsenseandnominatum’,186–98.Gazdar,Gerald(1979)Pragmatics,NewYork:AcademicPress.Geach,PeterandBlack,Max(eds)(1980)TranslationsfromthePhilosophicalWritingsofGottlobFrege,Oxford:Blackwell[1stedn1952].Gellner,Ernest(1979)WordsandThings:AnExaminationof,andanAttackon,LinguisticPhilosophy,London:RoutledgeandKeganPaul[1stedn1959].Glock,Hans-Johann(ed.)(1997)TheRiseofAnalyticPhilosophy,Oxford:Blackwell.Goodluck,Helen(1991)LanguageAcquisition,Oxford:Blackwell.Goodman,Nelson(1984)OfMindandOtherMatters,CambridgeMA:HarvardUniversityPress.\n186ReferencesGrice,Paul(1957)‘Meaning’,ThePhilosophicalReview,66.ReprintedinGrice(1989)StudiesintheWayofWords,CambridgeMA:HarvardUniversityPress,213–23.——(1958)‘PostwarOxfordphilosophy’,inGrice(1989)StudiesintheWayofWords,CambridgeMA:HarvardUniversityPress,171–180.——(1975)‘Logicandconversation’,inP.ColeandJ.L.Morgan(eds)SyntaxandSemanticsvol.9:Pragmatics,NewYork:AcademicPress.ReprintedinGrice(1989)StudiesintheWayofWords,CambridgeMA:HarvardUniversityPress,22–40.ReprintedinA.P.Martinich(ed.)(1996)ThePhilosophyofLanguage,3rdedn,Oxford:OxfordUniversityPress[1stedn1985]156–67.——(1978)‘Furthernotesonlogicandconversation’,inP.ColeandJ.L.Morgan(eds)SyntaxandSemanticsvol.9:Pragmatics,NewYork:AcademicPress.ReprintedinGrice(1989)StudiesintheWayofWords,CambridgeMA:HarvardUniversityPress,41–57.——(1981)‘Presuppositionandconversationalimplicature’,inP.Cole(ed.)RadicalPragmatics,NewYork:AcademicPress:183–99.ReprintedinGrice(1989)StudiesintheWayofWords,CambridgeMA:HarvardUniversityPress,269–82.——(1989)StudiesintheWayofWords,CambridgeMA:HarvardUniversityPress.Guttenplan,Samuel(1986)TheLanguagesofLogic,Oxford:Blackwell.Hacking,Ian(1975)WhyDoesLanguageMattertoPhilosophy?,Cambridge:CambridgeUniversityPress.Haegeman,Liliane(1994)IntroductiontoGovernmentandBindingTheory,2ndedn,Oxford:Blackwell.Harris,Roy(1996a)TheLanguageConnection:PhilosophyandLinguistics,Bristol:Thoemmes.——(1996b)Signs,LanguageandCommunication,London:Routledge.Hofmann,Thomas(1993)RealmsofMeaning,London:Longman.Horn,Laurence(1972)‘OnthesemanticpropertiesofthelogicaloperatorsinEnglish’,Ph.D.thesis,UniversityofCalifornia,LosAngeles.——(1985)‘Metalinguisticnegationandpragmaticambiguity’,Language,61,121–74.——(1989)ANaturalHistoryofNegation,Chicago:UniversityofChicagoPress.Hurford,JamesandHeasley,Brendan(1983)Semantics:ACourseBook,Cambridge:CambridgeUniversityPress.Jowett,Benjamin(1964)TheDialoguesofPlato,4vols,4thedn,Oxford:ClarendonPress[1stedn1871].Kant,Immanuel(1929)CritiqueofPureReason,trans.NormanKempSmith,London:Macmilian[1stGermanedn1781].Katz,Jerrold(1972)SemanticTheory,NewYork:HarperandRow.Kempson,Ruth(1977)SemanticTheory,Cambridge:CambridgeUniversityPress.Kenny,Anthony(ed.)(1994)TheWittgensteinReader,Oxford:Blackwell.Kripke,Saul(1959)‘Acompletenesstheoreminmodallogic’,TheJournalofSymbolicLogic,24,1,1–14.——(1972)‘Namingandnecessity’,inD.DavidsonandG.Hartmen(eds)SemanticsofNaturalLanguage,Dordrecht:Reidel.ReprintedinA.P.Martinich(ed.)(1996)ThePhilosophyofLanguage,3rdedn,Oxford:OxfordUniversityPress[1stedn1985]255–70.Leech,Geoffrey(1981)Semantics,2ndedn,Harmondsworth:Peguin[1stedn1974].——(1983)PrinciplesofPragmatics,NewYork:Longman.\nReferences187Leibniz,GottfriedWilhelmvon(1951)Theodicy,trans.E.M.Huggard,London:Routledge[1stLatinedn1710].——(1981)NewEssaysonHumanUnderstanding,trans.PeterRemnantandJonathanBennett,Cambridge:CambridgeUniversityPress[1stFrenchedn1765].Levinson,Stephen(1983)Pragmatics,Cambridge:CambridgeUniversityPress.Lewis,David(1973)Counterfactuals,Oxford:Blackwell.Linsky,Leonard(ed.)(1952)SemanticsandthePhilosophyofLanguage,Illinois:UniversityofIllinoisPress.Locke,John(1993)AnEssayConcerningHumanUnderstanding,London:Everyman[1stedn1690].Lyons,John(1977)Semantics(2vols)Cambridge:CambridgeUniversityPress.——(1981)Language,MeaningandContext,London:Fontana.——(1995)LinguisticSemantics,Cambridge:CambridgeUniversityPress.Martin,John(1987)ElementsofFormalSemantics,London:AcademicPress.Martin,Robert(1987)TheMeaningofLanguage,CambridgeMA:MITPress.Martinich,A.P.(ed.)(1996)ThePhilosophyofLanguage,3rdedn,Oxford:OxfordUniversityPress[1stedn1985].McEnery,TonyandWilson,Andrew(1996)CorpusLinguistics,Edinburgh:EdinburghUniversityPress.McCawley,James(1981)EverythingthatLinguistsHaveAlwaysWantedtoKnowaboutLogic,Oxford:Blackwell.Mill,JohnStuart(1867)ASystemofLogic,London:Longman.Montague,Richard(1974)FormalPhilosophy,NewHaven:YaleUniversityPress.Morris,Charles(1938)‘Foundationsofthetheoryofsigns’,inO.Neurath,R.CarnapandC.Morris(eds)InternationalEncyclopaediaofUnifiedScience,Chicago:UniversityofChicagoPress:77–138.O’Grady,William,Dobrovolsky,MichaelandKatamba,Francis(eds)(1997)ContemporaryLinguistics,2ndedn,London:Longman[1stedn1987].Pinker,Steven(1995)TheLanguageInstinct,Harmondsworth:Penguin[1stAmericanedn1994].Putnam,Hilary(1967)‘The“InnatenessHypothesis”andexplanatorymodelsinlinguistics’,Synthesis,17,12–22.ReprintedinN.Block(ed.)(1981)ReadingsinthePhilosophyofPsychology,Vol.2,London:Methuen,292–9.Quine,WillardVanOrman(1956)‘Quantifiersandprepositionalattitudes’,TheJournalofPhilosophy,53,177–87.ReprintedinA.P.Martinich(ed.)(1996)ThePhilosophyofLanguage,3rdedn,Oxford:OxfordUniversityPress[1stedn1985]330–6.——(1960)WordandObject,CambridgeMA:MITPress.——(1961)FromaLogicalPointofView,CambridgeMA:HarvardUniversityPress[1stedn1953].Radford,Andrew(1988)TransformationalGrammar,Cambridge:CambridgeUniversityPress.Russell,Bertrand(1905)‘Ondenoting’,Mind,14,479–99.——(1919)IntroductiontoMathematicalPhilosophy,London:GeorgeAllen&Unwin.——(1940)AnInquiryintoMeaningandTruth,London:GeorgeAllen&;Unwin.——(1957)‘MrStrawsononreferring’,Mind,66,385–9.——(1978)Autobiography,London:GeorgeAllen&Unwin[1stedn1967].——(1980)TheProblemsofPhilosophy,Oxford:OxfordUniversityPress[1stedn1912].——(1992)HumanKnowledge,London:Routledge[1stedn1948].\n188ReferencesRyle,Gilbert(1953)‘Ordinarylanguage’,ThePhilosophicalReview,LXII,167–86.ReprintedinCharlesCaton(1963)PhilosophyandOrdinaryLanguage,Urbana:UniversityofIllinoisPress.——(1957)‘Thetheoryofmeaning’,inC.A.Mace(ed.)BritishPhilosophyintheMid-Century,London:GeorgeAllen&Unwin,239–64.ReprintedinCharlesCaton(ed.)(1963)PhilosophyandOrdinaryLanguage,Urbana:UniversityofIllinoisPress.Sadock,Jerold(1974)TowardsaLinguisticTheoryofSpeechActs,NewYork:AcademicPress.Sapir,Edward(1970)Language,London:Hart-Davis[1stednHarcourt,Brace&World1921].Saussure,Ferdinandde(1960)CourseinGeneralLinguistics,trans.WadeBaskin,London:PeterOwen[1stFrenchedn1915].Schiffrin,Deborah(1984)ApproachestoDiscourse,Oxford:Blackwell.Searle,John(1965)‘Whatisaspeechact?’,inMaxBlack(ed.)PhilosophyinAmerica,London:GeorgeAllen&Unwin,221–39.ReprintedinJ.Searle(ed.)(1971)ThePhilosophyofLanguage,Oxford:OxfordUniversityPress,39–53.ReprintedinA.P.Martinich(ed.)(1996)ThePhilosophyofLanguage,3rdedn,Oxford:OxfordUniversityPress[1stedn1985]130–40.——(1968)‘Austinonlocutionaryandillocutionaryacts’,PhilosophicalReview,57,405–24.——(1969)SpeechActs,Cambridge:CambridgeUniversityPress.——(1975)‘Indirectspeechacts’,inP.ColeandJ.L.Morgan(eds)SyntaxandSemanticsvol.9:Pragmatics,NewYork:AcademicPress,59–82.ReprintedinJohnSearle(1979)ExpressionandMeaning,Cambridge:CambridgeUniversityPress,30–57.ReprintedinA.P.Martinich(ed.)(1996)ThePhilosophyofLanguage,3rdedn,Oxford:OxfordUniversityPress[1stedn1985]168–81.——(1977)‘Reiteratingthedifference:areplytoDerrida’,Glyph,1,198–208.Seldon,Raman(1989)PractisingTheoryandReadingLiterature,London:HarvesterWheatsheaf.Skinner,BurrhusFrederick(1957)VerbalBehavior,NewYork:Appleton-Century-Crofts.Smith,NeilandTsimpli,lanthi-Maria(1995)TheMindofaSavant,Oxford:Blackwell.Smith,NeilandWilson,Deirdre(1990)ModernLinguistics:TheResultsofChomsky’sRevolution,Harmondsworth:Penguin[1stednPelican1979].Sperber,DanandWilson,Deirdre(1995)Relevance,2ndedn,Oxford:Blackwell[1stedn1986].Stainton,Robert(1996)PhilosophicalPerspectivesonLanguage,Ontario:Broad-view.Stalnaker,Robert(1974)‘Pragmaticpresuppositions’,inM.MunitzandP.Unger(eds)SemanticsandPhilosophy,NewYork:NewYorkUniversityPress,197–213.Strawson,Peter(1950)‘Onreferring’,Mind,59,320–44.ReprintedinA.P.Martinich(ed.)(1996)ThePhilosophyofLanguage,3rdedn,Oxford:OxfordUniversityPress[1stedn1985]215–30.——(1952)IntroductiontoLogicalTheory,London:Methuen.Tarski,Alfred(1944)Thesemanticconceptionoftruthandthefoundationsofsemantics’,PhilosophyandPhenomenologicalResearch,4,341–74.ReprintedinA.P.Martinich(ed.)(1996)ThePhilosophyofLanguage,3rdedn,Oxford:OxfordUniversityPress[1stedn1985]61–84.ReprintedinL.Linsky(ed.)(1952)SemanticsandthePhilosophyofLanguage,Illinois:UniversityofIllinoisPress,13–47.\nReferences189Thomas,Jenny(1995)MeaninginInteraction,London:Longman.vanderSandt,Rob(1988)ContextandPresupposition,London:CroomHelm.Vendler,Zeno(1972)ResCogitans,Ithaca:CornellUniversityPress.Whorf,BenjaminLee(1956)Language,ThoughtandReality,CambridgeMA:MITPress.Wilson,Deirdre(1975)PresuppositionsandNon-Truth-ConditionalSemantics,London:AcademicPress.Wittgenstein,Ludwig(1922)TractatusLogico-Philosophicus,trans.C.K.Ogden,London:RoutledgeandKeganPaul.——(1958)PhilosophicalInvestigations,trans.G.E.M.Anscombe,Oxford:Blackwell[1stedn1953].\nIndexanalyticphilosophy2,6,7,32,111conversationalimplicatureseeanalyticsentence73,76–9,81,94,173implicatureaposterioriknowledge77,173co-operativeprinciple109,130–5,142,aprioriknowledge76–80,103,145,182n10173cornerquotes46–7Aristotle14,16–17,20,75,93,145,corpuslinguistics167–8,171179n3correspondencetheory75–6,82,84–6,Austin,J.L.2,107,108,116–21,124,91–2,112,174,181n4126,132,137–8,139–40,141–2,counterfactuals97–9,104182n5axiomatictheory88–9,103,106Davidson,Donald85,87–91,100,Ayer,Alfred70,83–5,103,104101,103,104,106deductivemethod152,160,174behaviourism6,142,153,155,157,definitedescription32–4,36,38,158,161,163,168,170,17358–60,61,63,64,70Berkeley,George19–22,39,146,151,denotation10–13,26–7,28–9,31,33,15857,58–61,96,174,179n9Bloomfield,Leonard148,151–3,154,Derrida,Jacques2–3,7155,157,160,161,170descriptivefallacy108,116,120determinism148–51,170Carnap,Rudolf39,45–6,82–3,84,directreference10,15,23,28,33,34,85,101,115,154,155,179n1040,146CDS(ChildDirectedSpeech)169,172,discourseanalysis5,140,143183n13disjunction47,50–1,128–9,174Chomsky,Noam4,7,67,71,121,Donnellan,Keith38158–66,167–9,170,171,182n5,Dummet,Michael7,10,141183n6,183n12competence/performance165–6E-languageseeI-languagecondition47,51–2,56,88,97,173empiricism6,17,17–22,23,77,79,conjunction47,48–9,50,51,55,80,145–6,151,158,174;in126–8,173linguistics164,151–8,165,167–70,connotation11,26,28–9,31,57,173171constative118,119,120–1,173–4Enlightenment17–18,22,77continentalphilosophy2–3,7,81entailment6,42–3,55–6,59–60,61,contingenttruth74,78,17462,63,174,180n11contradiction49,73,78,94,129,130,extension11–13,15,31,37,39,43–4,174,182n972,73,75,95,174conventionalimplicatureseeimplicatureconversationalanalysis140,143,167190\nIndex191felicityconditions118,119,120,121,Leibniz,Gottfried22–5,91–2,99,123,122,186n6145,158,179n6,181n11;seealsoFodor,Jerry166,171LawofSubstitutionFrege,Gottlob2,27–32,34,35,36,Lewis,David92,98,104,181n1239–40,43,44,47,57–9,61,67,99,linguistics1,2,3–5,6,7,8,9,11,32,114,115,141,154,179n9,180n1236–8,41,43,63,67–9,83,87,100,101–3,105,106,107,110,114,Gellner,Ernest141121,125,138–9,141,142,146,generalideas10,19–20,38148,151,152,158,166,167–70,generativegrammar4,102,105,164,181n1;descriptivevs.prescriptive165,166131,152,160Goodman,Nelson161,163Linsky,Leonard24,39,179n7Grice,Paul65–7,107,109–10,Locke,John18–20,22,23,25,39,77,123–37,139–40,141,142,182n5,80,123,146,151,158182n10locutionaryact120,175logicalpositivism80–85,87,104,111,Harris,Roy39,141,168,172175–6,181n6holism154,156,174materialism20;inlinguistics152,153IdealForm14–16,18,145,175mathematics5,27,32,46,61,78–9,ideationaltheory10,17,38,146,17581,83,99,110,111,115,126,iff52,88,175129,130IH(InnatenessHypothesis)158–67,mentalese104,166170–2,175,183n10mentalism6,152,158,163,165,I-language/E-language116,167–8,171,167–70,176174,175metaphor135–6illocutionaryact120–1,122,175metaphysics15–16,26,80,81,82,immaterialism20151,176implicature6,116,182n10;Mill,JohnStuart23,25–9,31,32,35,conventional133;conversational65,39,123,141109,115,123,124,126–37,139,modality93–7,100,104,176142MontagueGrammar99–100,104indirectspeechact122,137,143mood90inductivemethod152,160,175,Morris,Charles4,7,148innateknowledge14–15,18,145,159;seealsoIHnativism6,159,176;seealsoinnatisminnatism145,162;seealsoIHnaturallanguage3,5,37,50,51,52,intension11–12,31,39,43,44–5,71,54,57,60,61,62,67,83,85,87,75,96,175,179n189,93,99–100,104,105,106,107,intensionalcontext12,30–1,37,95–7;112,115,126–30,142,164seealsoopaquecontextnegation42,43,47,49–50,57,59–60,intention107,117,118,119,120,68,128,129–30,136–7,143,176121,122,123–6,131,142,144non-naturalmeaning(meaning)125,NNirony135–6142Kant,Immanuel77–80,103,145,158opaquecontext12–13,40,95–7,104,Kripke,Saul34–6,39–40,95,180n12176ordinarylanguagephilosophy2,6,60,languageacquisition6,138,143,144,65,107,110,114–16,121,123,145,153,157,158–65,168–9,124,141,176183n8languagefaculty159,161,163–5,171performative117–20,121,138,languagegames122–3,141141,176LawofSubstitution30\n192Indexperformativehypothesis138–9,168;lexical6,37;seealsopossible142,143worldsperlocutionaryact120–1,177semiotics148phonology4,5,148,168sense(Fregean)27–31,32,33,35,Plato14–16,17,18,123,145,158,44–5,57,61,96,154179n2Skinner,B.F.153,155,170possibleworlds74,91–100,104,sociolinguistics5112,177Socrates14,15,16pragmatics5,6,63–5,69,71,103,speechacts108,116–23,137–8,109–10,121,123,131,137–40,141–2,143;theory3,140142–3Strawson,Peter2,60–3,64–5,66,67,predicate14,53,54,73,77,79,8970,107,110,115,116,121,137,presupposition6,43,56–8,60,61–9,180n1170–1,103,107,119,177structuralism38,123,147–8,151,propositionalattitude31,37,183n1296–7,177synonymy37psycholinguistics140,167,168,171,syntax4,41,83,99,105,148,168182n1syntheticsentence76–80,81,103,177Putnam,Hilary161–3,164Tarski,Alfred85–7,101,104,105,quantifiers53,59,177;existential10653–4,59–60;universal53–4,59–60tautology73,130,135,177Quine,W.V.O.40,153–8,159,160,thought:relationshiptolanguage5,7,161,162,163,165,166,170,111,112,123–4,145,148–51,166,182n4169,171transcendence78rationalism23,77,145,158,159,171translation38,45,180n1;radicalreference(Fregean)27–31,32,35,36,155–7,17044,57,61,73truth-conditions:definitionof73,75–6,relevancetheory139,143177–8Russell,Bertrand2,4,31–6,38,39–truthtables48,49,50,51,5240,43,52,58–60,61,62–3,66,67,truthvalue:definitionof42–576,83,99,103,107,111,115,T-sentence86–9,178141,180n11,180n12Ryle,Gilbert114–15,141,182n4UniversalGrammar165,168,178use/mention13,23–5,30,39,86,89Sapir,Edward,148–50,152,161,166Sapir-Whorfhypothesisseedeterminismverification74,80–5,104,178Saussure,Ferdinandde38,123,144–5,ViennaCircle81–4,87,111,154;see146–9,151,154,161,166,170,alsologicalpositivism183n12scepticism21–2,77,177,180n1;Whorf,BenjaminLee148–51,152,semantic154166,170Searle,John3,7,39,70,107,120,Wittgenstein,Ludwig4,7,76,107,121–3,124,137,142,182n6110–14,115,141,154semantics4,41,70,71,75,99–100,101–3,105,110,137,139,155,