Introduction
Moodisafundamentalandtraditionalconceptusedinthedescriptionofhuman languageandintheoreticalinquiriesintothenatureoflanguage.Becauseitis fundamentalandtraditional,onemightthinkthatthetheoryofmoodisan importanttopicwithinlinguistics.Yetmoodassuch,initsmostgeneralsense,is notatopicofresearch,andthereareseveralreasonsforthis.Oneisthattheconcept ofmoodissofundamentalthatdifferentscholarlytraditionshavemoldedtheir understandingofwhatmoodistofittheirconceptionofthenatureoflanguage. Weseeitusedindifferentwaysinphilosophy,formalsemanticsandpragmatics, anddescriptive/typologicallinguistics.Andevensettingasidebroaddifferencesin approachtothestudyoflanguage,moodisadifficulttopicbecausealllinguists agreethatitrelatestoatleastthetwodifferentgrammaticalphenomena: verbal mood,thecategorywhichincludesindicativesandsubjunctives,and sentence mood,thecategorywhichincludesdeclaratives,interrogatives,andimperatives. Ithasgenerallynotseemedlikelythatageneraltheoryofmoodincorporating bothofthesecategorieswouldbepossible,andso,inmanyinstances,linguiststake caretoexplainthattheyaretalkingaboutonekindofmoodandnottheother.It willemergeaswegoforward,however,thatsimilarideasandpuzzlesrecuraswe trytounderstandthetwokindsofmood.Theymightnotbeasdifferentaswe oftenassume.
Inwritingthisbook,Ihavetwoprimarygoals.Thefirstistoprovidescholars interestedinthesemanticsandpragmaticsofmoodwithadeepersourceof backgroundthanisfoundinanysinglepreviouswork.Indoingthis,Idrawonas muchoftheliteratureonmoodasIamable,butIthinkandwritefromwithinthe formalsemanticstradition.Andsecond,Iseektoformulateandadvancenewideas aboutthesemanticsandpragmaticsofmood.AsIsurvey,organize,andexplain theliterature,Iwillsometimessketchoutnovelanalyses.Insomecases,Idothis asamodelfordiscussionwithoutentirelyendorsingtheanalysis,whileinothers, Iadvocateforacertainhypothesis.
Itmustnotgowithoutsayingthatthereisagreatdealofinsightfulresearch onmoodwhichwillnotbementionedinthisbook.Thefieldsoflinguisticsand philosophyaretoolargeandtheconceptsofmoodaretoouseful—theyleadto suchavastliteraturethateventhesubsetofitthatisworthstudyingtodayisso extensivethatnosinglebookcanincorporatearesponsiblediscussionofitall. Whatisincludedhereistheresultofmybestjudgmentaboutwhatcombination ofliterature,explanation,andnewideascanbestfurtherthegoalsofthebook.
ThereisagreatdealofadditionalimportantscholarshiponmoodwhichIwould recommend.Ihopethatthisbookcanhelpreadersmakebetteruseofit.
Sometermsandconventions. Beforebeginningthesubstantivediscussion, Iwouldliketospecifysomeofmyterminologyandconventions.
1.Notationalconventions.Inmentioningaword,phrase,orsentence:
(a)Theuseofsmallcapsindicatesatechnicaltermwhoseprecisemeaning isbeingdiscussedintheimmediatecontext;inmanycases,thetermis actuallydefinedthere.
(b)Theuseof boldface merelyservestodrawthereader’sattentiontoa particularwordorwords.
(c)Theuseof italics indicatesthatthematerialisbeingreferredtoasa linguisticobject.Exception:Withinnumberedexamplesandfigures, suchmaterialwillnotbeitalicized,unlessthereisothermaterialwithin thenumberedexamplewhichisnotbeingreferredtoasalinguistic object.
(d)Theuseof‘singlequotes’indicatesthatthematerialstandsforthe linguisticobjectsinanumberoflanguages,undertheassumptionthat allofthoseobjectsaresimilarenoughacrosstherelevantlanguagestobe spokenofasiftheyarethesame.Example:“InmostRomancelanguages, ‘believe’selectstheindicative.”Singlequotesarealsousedforglossesand paraphrases.
(e)Theuseof“doublequotes”indicatesthatthequotedmaterialshouldbe understoodasspokenorwrittenbysomeone.Doublequotesarealso usedinthetraditionalwayas“scarequotes,”indicatingthatthequoted materialshouldnotnecessarilybetakenliterally.
(f) [Brackets] indicatesomeaspectofthecontextinwhichanexampleisto beunderstood.
2.Iusethetermlogicalformintwoways:
(a)Whencapitalized,as“LogicalForm,”itreferstoaleveloflinguisticrepresentationatwhichcertainsemanticproperties,suchasscope,areexplicitly represented.LogicalFormcanbealevelofsyntaxoranindependentlevel betweensyntaxandsemantics.
(b)Whennotcapitalized,as“logicalform,”itreferstoarepresentation whichdisplaysorelucidatesimportantabstractpropertiesofmeaning. Thelogicalformusuallyhassomelogic-likeproperties,suchasasimple vocabularyandanunintuitivesyntax.UnlikeLogicalForm,alogicalform isnotunderstoodtobealeveloflinguisticrepresentation.
3.Whenusingapronounanaphorictoanantecedentwhichcouldreferto amaleorfemaleindividual,Iroughlyalternategenders.Example:“Ifa philosopherreadsthis,shemightdisagree”(nexttime“he”).Occasionally, however,itwillbebettertouse“heorshe,”“s/he,”orsingular“they.”
Indifficultcases,IwillapplytheseconventionsinthewayIfeelismosthelpful.
1.1Whatdowestudywhenwestudymood?
1.1.1Conceptualpreliminaries
Oneofthemostfundamentalissuestobeaddressedbysemantictheoryconcerns thekindofmeaningsthatsimpledeclarativesentenceshave,andwecanjudge anyapproachbyhowwellithelpsusunderstandsuchbasicpropertiesas:they canbe(andtypicallyare)eithertrueorfalse;theycanbeusedtosharefactual information;theyareappropriateforrepresentingthecontentsofthoughtand someothercognitivestates;andtheystandinlogicalrelationstooneanotherlike entailmentandsynonymy.Considerthedeclarativesentence(1a).Ithasallofthese properties:
(1)a.Benisholdingabird.
b.Thisistruebecause:BenisholdingJohnFei.
c.Noahtoldme:“Benisholdingabird.”AndnowIknowwhatBenis holding.
d.NoahthinksthatBenisholdingabird.
e. Benisholdingabird. Entails: Benisincloseproximitytoabird.
Moreover,semanticiststhinktheyhaveagood(thoughnotyetperfect)understandingofhoweachofthesepropertiesshouldbeexplained,anunderstanding whichisbasedonthetheoreticalconstructofaproposition.
Informalsemantics,ourtheoriesofsentencemeaningmostcommonlywork withtheideathatpropositionscanbedefinedintermsofpossibleworlds.A possibleworldisawaythingscouldbe,completethroughspaceandtime,an alternativehistoryoftheuniverse.Ourownuniverse-historycanbereferredto asthe“actualworld”or“realworld.”Forthepurposesoflinguisticsemantics, letusassumethatwehaveasetofpossibleworldsconceivablebyhumans.By thisImeanthatanydifferenceinhowthingscouldbewhichahumancould recognize,imagine,ordescribecorrespondstoadifferencebetweenpossible worldsinthesetofallworlds W .Forexample,ifItellyouthatBenhasa cockatiel,youcanimaginethatitisgrey,orthatitiswhite.Therefore, W shouldcontainatleastonepossibleworldinwhichitisgreyandatleastone inwhichitiswhite.Thesetofworldsconceivablebyhumansseemssuretobe adequatefordoingnaturallanguagesemantics,ifanytheorybasedonpossible worldsis.
Apropositioninpossibleworldssemanticsisasubsetof W .Forexample,the meaningof(1a)is,orcanbecharacterizedintermsof,theset(2):
(2) {w :Benisholdingabirdin w}
(=thesetofworldsinwhichBenisholdingabird.)
Thisdefinitionofpropositionhelpstoexplainthepropertiessummarizedin(1). Forexample,(1b)amountstotheclaimthattheactualworldisoneoftheworlds intheset.Thoughtheconceptionofpropositionsassetsofpossibleworldsisnot
4introduction
withoutitsproblems,wecanignorethemfornow.Ourgoalistouseittodevelop ausefulwayofthinkingaboutmood.
Anotherfeatureoflanguageweneedtounderstandifwearetogetahandle ontheconceptofmoodismodality.Modalityis“thelinguisticphenomenon wherebygrammarallowsonetosaythingsabout,oronthebasisof,situations whichneednotbereal”(Portner2009,p.1).Thenoun Ben in(1a)isnotmodal;it isusedtorefertoaparticular,actualperson.Incontrast,theauxiliary should in(3) ismodal:
(3)Benshouldputdownthebird.
ThissentencesaysthatsituationsinwhichBenputsdownthebirdareinsome respectbetterthan,orpreferableto,situationsinwhichhedoesnot.Sinceheeither willorwillnotputdownthebird,someofthesesituationswillneverbereal,and sotheword should countsasmodal.
1.1.2Thegeneralconceptofmood
Withthesepreliminariesinplace,wecanattemptadefinitionofmood:
Moodisanaspectoflinguisticformwhichindicateshowapropositionisused intheexpressionofmodalmeaning.
Aswestartout,it’sacceptabletobevagueaboutwhatwemeanbythe“use”ofa proposition,becauselinguistshaveemployedtheterm“mood”inmanydifferent ways.Letusconsidertwoexamples:
1.Inmanylanguages,subordinateclauses(andsometimesrootclausesaswell) comeindifferentforms,knownasverbalmoods,suchasindicativemood andsubjunctivemood.Theseclausescanbeusedtohelprepresentvarious cognitivestatesandmentalevents,suchasbeliefs,desires,anddreams.When theyareusedinasubordinateclausetohelptalkaboutadesire(asin(4)), theytypicallytakethesubjunctiveform;incontrast,whentheyareusedto describeadream((5)),theytypicallytakeindicativeform.
(4) Pierre Pierre veut wants que that Marie Marie soit is.subj heureuse. happy (French) ‘PierrewantsMarietobehappy.’
(5) J’airêvé Idreamed qu’il thathe était was.indic président. president ‘Idreamedthathewaspresident.’
Undercertaincircumstances,theymaytakeadifferentform,suchasinfinitive.Forourpurposes here,it’shelpfultopretendthereareonlytwopossibilities.
whatdowestudywhenwestudymood?5
Wecallthiskindofmood“verbalmood”becauseitisfrequently(thoughnot always)markedinagrammaticalsenseontheverbalheadoftheclause;for example,theverb soit isasubjunctiveformof‘be’inFrench.
2.Inalllanguages,rootsentenceshavevariousfunctions,includingeasy-tointuitoneslikedirectingsomebodytodosomething(asin(6a)),requesting information((6b)),andprovidinginformation((6c)).
(6)a.Pickupabird!
b.IsBenholdingabird?
c.Benisholdingabird.
Eachofthesefunctionsisassociatedinanintuitivelyclearway(whichis nonethelesshardtodefine)withparticularmembersofaparadigmofforms knownassentencemoods.Themostprominentsentencemoodsarethe threejustillustrated:theimperative,interrogative,anddeclarative.
Eachofthesetwoimportantconceptsofmoodrealizesthegeneraldescriptionof moodindicatinghowasentence’spropositionistobeused:verbalmoodtellsus somethingabouthowitistobeused,withinthecompositionalcomputationof meaning,todescribeanindividual’smentallife,whilesentencemoodindicates howitistobeused,inamulti-partyexchange,toachievespecifiedcommunicative functions.
Givenanunderstandingofmoodliketheonejustdeveloped,whatwoulda linguistictheoryofmoodlooklike?Wecouldspelloutatoo-simpletheoryof verbalmoodlikethis:Assumethatverbslike want and dream,verbswhichtake sententialcomplementsandtalkaboutsomeaspectoftheirsubject’smentallife, aremodalwordswhichexpressarelationbetweentwoarguments,anindividual andaproposition.Theirlogicalformcanberepresentedas V (x, p),forexample wants(Pierre, {w : Marieishappy w}).Moodchoiceisdeterminedbythefollowing principle:
(7)Iftherelationexpressedbytheverbconcernsapreferenceabouthowthe futurewillbe,theclausewhichdenotesthepropositionargumentofthis relationshouldbeinthesubjunctivemood.Otherwise,itshouldbeinthe indicativemood.
Giventhisprinciple,ifyouwanttosaysomethingwhichmeans wants(Pierre, {w :Marieishappy w}),theverb‘be’willbesubjunctive,butifyouwanttosay somethingwhichmeans dreams(I, {w :Pierrewaspresident w}),itwillbeindicative. Nothingsemanticchangesbetweenthetwocases,otherthanthemainrelation, wants or dreams.
Andwecouldspelloutatoo-simpletheoryofsentencemoodlikethis:Assume thatthefunctionofarootsentenceusedindialogueistoadjustthespeaker’sand hearer’ssharedassumptions.Forexample,sometimesthespeakermaywantto createasharedassumptionthatonewaythefuturecouldbeispreferabletoanother, perhapsthreateningsomesortofpunishmentupontheaddresseeifthepreferred
futuredoesnotcomeabout.Wecandescribethisasthespeakerdirectingthe addresseetodosomething.Thereisaparticulargrammaticalform,theimperative, forasentenceusedwiththispurpose,asin(6a).Thisreasoningsuggeststhe followingprinciple:
(8)Ifasentenceisusedtocreateanassumptionthatitispreferablethatthe propositionexpressedbythesentencecomestobetrue(asopposedtofalse), itshouldbeexpressedusingimperativemood.
Therewouldbesimilarprinciplesforinterrogativesentences(WillBenputdown thebird? createsanassumptionthattheaddresseewillhelpthespeakerknowsomething)andindicativesentences(Benwillputdownthebird createsanassumption thatacertainfactholds).Nothingwoulddifferamongthesecasesintermsofthe propositioninvolved,buttherewouldbecrucialdifferencesinthegoalwhichthe speakeraimstoachievebyusingasentencewhichdenotesthatproposition.
Thesepicturesofsubjunctiveandimperativeclauseshavebeenpresentedto helpconveyanunderstandingoftheideabehindtheinformaldefinitionofmood. Thisconceptionofmoodwillbeveryimportantinthebook,asitservesasan unarticulatedintuitionbehindtheactualpracticeslinguistshaveindescribing phenomenaas“mood,”andbecauseitplaysaroleinmanyattemptstoprovide concretesemanticorpragmaticanalysesofmoodforms.However,thespecific statementsaboutthesubjunctiveandimperativeabovearenottobetakenas seriousproposals.Besidesoversimplifyingtherelevantphenomena,theymake manyassumptionswhichcouldturnouttobewrong:forexample,theyassumethat allsentenceshavepropositionsastheirbasicmeaningsandthatthevariousmoods havenoeffectthemselvesonthemeaningofagivensentence.Allsuchassumptions mustbecarefullyevaluatedaspartofanyseriousinvestigationofmood.Much oftheworkinthisbookistoexaminesomeofthephenomenawhichmeetthe characterizationofmoodbasedontheconceptof“modaluse”andtoconsider varioustheoriesofthem.Inthenextsubsection,Igiveapreviewofwherethisway ofthinkingaboutmoodwillleadus.
1.2Mainfindingsaboutthenatureofmood
Theliteratureonmoodisextensive,contradictory,andattimesconfusing. Nevertheless,Ihopewewon’tendupinamuddle.Thediscussionwillleadnotonly tolotsofquestions,buttoafewlinguisticallysignificantconclusionsaswell.Here attheoutset,Iwouldliketohighlightthework’smostbasic,broadlyconceptual findings:
1.Whileverbalmoodandsentencemoodaredistinctbothintermsof morphosyntaxandintermsofmeaning,theyarecloselyrelated.They arerelatedbecauseoftightparallelsbetweenthemodalsemanticsof sentence-embeddingconstructions,whichdetermineverbalmoods,andthe
mainfindingsaboutthenatureofmood7 communicativefunctionsservedbyrootsentencesinconversation,interms ofwhichsentencemoodsarecharacterized.
2.Ataformallevel,manytheoriesofverbalmoodandsentencemoodcanbe expressedwithinaunifiedframework.Thisframeworkstatesthatbothkinds ofmoodencodetheinteractionbetweenaclause’smeaningandacontextually givenpartiallyorderedsetofworlds(theposw).Theposwmaybe derivedfromeitherthegrammaticalorthediscoursecontext.Thoughthe posw-frameworkisclearlynotthelastwordontherelationbetweenverbal moodandsentencemood,itcanpointthewaytowardsamoregeneraland explanatorytheoryofthetwokindsofmoodandtherelationbetweenthem.
Ontheassumptionthatsometypesofverbalmoodandsometypesof sentencemoodcanbeanalyzedintermsofasingletheoreticalsystem(be ittheposworsomethingelse),Ilabelthebroaderlinguisticcategorywhich includesthembothcoremood.
3.Thereisarangeofformsandconstructionsacrosslanguageswhoserelation tocoremoodisnotcurrentlywellunderstood.Amongtheseareevidentials, thekindsofformswhichElliott(2000)describesasmarking“realitystatus,” andnumerousotherspecificmoodsandmodesusedinthedescription oflanguagestypologicallydifferentfromthosewhichhavebeenthefocus ofresearchonverbalmood.Wedonotyethaveagoodenoughunderstandingofalltheformswithlabelslike“realis,”“irrealis,”“conjunct,”“inferential,”“hearsay,”“conjecture,”“potential,”“desiderative,”“assertive,”and “contingent”tosayhowtheyarerelatedtocoremood.
Ithinkthatcoremoodcoversmorethanwhatlinguiststypicallythinkofasmood, ratherthanless,andIdon’tknowofanylinguisticformswhichatonceshould clearlybeclassifiedasmood,yetalsoclearlynotascoremood.However,sometypes ofelements,forexampleevidentials,whichIthinkcouldreasonablybethought ofascoremood,arenottypicallythoughtaboutinthatway,andsotheymay exemplifynon-coremood.Moresignificantly,fewphenomenawhichhavebeen describedintermsoftheconceptsofrealitystatushavebeenanalyzedinaprecise enoughwayforittobeclearwhattheirrelationtocoremoodis.Table1.1outlines
Table.Semanticclassificationsformoodandmodality
Modality
Coremood Non-coremood(Therestofmodality)
VerbalSentenceOthercore moodmoodmood indicative, subjunctive, imperative, declarative, evidentials?evidentials?epistemic,priority, deontic,dynamic certain infinitives interrogativerealitystatus?realitystatus?modals;modal adjectives,adverbs
thiswayofthinkingabouttherelationshipsamongvariouskindsofmoodand betweenmoodandmodality.
1.3Backgroundonmodality
Asdescribedabove,forthepurposesofthisbookIdefinemodalityasthelinguistic phenomenonwherebygrammarallowsonetosaythingsabout,oronthebasisof, situationswhichneednotbereal.Since,onthisconception,moodfitswithinthis broadercategoryofmodality,itwillbeessentialtohaveanunderstandingofthe semanticsofmodality.InthissectionIwillprovideanoverviewofsomekeyideas inthesemantictheoryofmodality,drawingonthemoreextensivediscussionby Portner(2009).
1.3.1Classificationsofmodality
Webeginwithtwowaysofclassifyingmodalexpressions:bythelinguisticlevelat whichtheyareexpressedandbythemeaningstheyconvey.
Classificationsofmodalitybyleveloflinguisticorganization. Modalitymay beclassifiedaccordingtotheleveloflinguisticorganizationonwhichitoperates. Specifically,it’susefultodistinguishsubsentential,sentential,anddiscourse modality.
1.Subsententialmodalityoperatesbelowthelevelofpropositionexpressedbya completesentence.Itincludesmodaladjectives,modalnouns,propositional attitudeverbs(verbswhichtakeanargumentwhichexpressesaproposition), verbalmood,andinfinitives,amongothertypes.
(9)(a)A possible solutiontothisproblemistocalltherecalcitrantreviewer.
(b)The probability ofsuccessislow.
(c)I think/hope/regret thatshearrivedontime.
(d)Ihope tobe happy.
Ofcourse,manyoftheseelementsalsoaffectthemeaningofthecomplete sentence,buttheydosoviathemeaningofsomesmallerconstituent.For example,thesubjectphrasesin(9a–b)havenounphrase-typemeanings whichhavebeenbuiltupusingthemodalconceptsexpressedby possible and probability,andthepredicatein(9c)denotesaproperty,likeotherpredicates do,butthispropertyinvolvesconsiderationofnot-necessarily-realsituations whichareimportantinthespeaker’smentallife.
2.Sententialmodalityoperatesatthelevelofthecompleteproposition.Inother words,ifasentencecontainsaconstituentwhichdenotesaproposition,and thenamodalelementcombineswiththistocreateanotherpropositional constituent,wehaveacaseofsententialmodality.Ingrammaticalterms, thismeansthatitistypicallyrealizedabovethelevelofthemainsubject–predicatestructureintheclause,thatis,abovetheS,IP, vP,orotherroughly
backgroundonmodality9 equivalentsyntacticunit.Sententialmodalityaffectstheprimarymeaningof thesentence;thus,inthecaseofarootdeclarativeclause,itaffectsthetruth conditionsofthesentence;inthecaseofarootinterrogative,itaffectswhat informationisliterallyrequested,andsoforth.
Sententialmodalityhascommandedthevastmajorityofscholarlyattentiononmodalitywithinlinguistics.InEnglish,therehasbeenagreatdeal ofworkonthemodalauxiliarieslike can in(10a).Wecanalsoclassifyas expressingsententialmodalitysuchelementsasmodalverbs(bythisImean verbswhicharenotauxiliaries,butwhicharedistinctfromregularlexical verbsaswell,asin(10b)),modaladverbs,andsometenseandaspectforms.
(10)(a)Noah can swim.
(b) Devo must.pres-1sg partire leave domani. tomorrow (Italian;Squartini2004) ‘Imustleavetomorrow.’
(c) Probably,Benwasthere.
(d)She will belate.
(e)Theteam isbuilding abridge.
Insomecases,thedecisionwhethertotreatagivenmodalelementas representingsubsententialorsententialmodalityrequiresacarefulsyntactic andsemanticanalysis.Forexample,mylabeling(10b)assententialisbased ontheassumptionthat dovere (‘must’)issyntacticallydistinctfromregular lexicalverbs,andtheideathattheprogressivein(10e)exemplifiessentential modalityassumesthatitisasentence-leveloperator.Ifeitherofthesepoints turnsoutnottobecorrect,thentheyarenotexamplesofsententialmodality.
3.Discoursemodalityisthedirectcontributionofmodalitytomeaningin discourse.Isay“directcontribution”because,ofcourse,anytypeofmodality ultimatelycontributestothebroaderdiscourse;whatdistinguishesdiscourse modalityisthatitdoesnotdosoviaaneffectontheregularsemanticmeaning expressedbyasentence.Though(10b)maycausethediscoursetocontainthe informationthatthespeakerfavorsfuturesituationsinwhichsheleavesover onesinwhichshedoesnot,thiseffectismediatedbyitsordinarysentencelevelsemantics;hence,itisnotdiscoursemodality.Incontrast,discourse modalityisseparatefromanymodalcontributiontowardsregularsentence meaning.Someplausibleexamplesofdiscoursemodalityareevidentiality, sentencemood,andtheperformativemeaningsofmodalauxiliariesand modalverbs.
(11)(a)Para-sha-n-mi/-si/-chá. rain-prog-3-bpg/rep/conj (CuscoQuechua,Faller2006b) ‘Itisraining.’(Directevidence/Reported/Conjectural.) (b)Leaverightaway!
(c)It might rain.
In(11a),weseethethreeevidentialmarkersinCuscoQuechua,thefirst indicatingthatthespeakerhasdirectevidence(whatFallercalls‘bestpossible
grounds’),thesecondthatheisconveyingareport,andthethirdthatheis makingaconjecture.In(11b),wehaveanimperative.Thesetwoexemplify discoursemodalityontheassumptionthattheevidentialsandimperative formdonotcausethesentencetodenoteamodalizedproposition—for examplethat(11a)doesnotmean‘Ihavedirectevidencethatitisraining’(if itdidmeanthis,itwouldshowsententialmodality),butratherconveysthis meaningwithoutaffectingtheprimaryproposition‘itisraining.’Example (11c)canbeconsidereddiscoursemodalityifweaccepttheclaim(Portner 2008)thatitnotonlymeansthatrainiscompatiblewithourinformation (itssententialmodality),butalsomakesthequestionofwhetheritwillraina topicofconversation(additionaldiscoursemodality).
Theexamplesofeachsubtypeabovearegivenonlyinordertohelpmakeclearwhat shouldfallundereachsubcategory.Itwouldn’tbesurprisingiffurtherresearch showedsomeofthemtobemiscategorizedorevennotmodalatall.Forexample, whilemanythinkthattheprogressiveismodal,thisissomewhatcontroversial (Portner2011a).
Classificationsofmodalitybymeaning. Modalityinvolvessayingsomething about,oronthebasisof,situationswhichneednotbereal,anditisusefulto classifymodalelementsaccordingtothewayinwhichitsmeaningisbasedon thosesituations.Roughlyspeaking,(12)saysthattherelevantsituations(thosein whichthegirlgetstheprize)aregood(perhapsmorespecifically,theyaregoodin relationtothecriteriasetupforthecontest):
(12)Thatgirlshouldgettheprize.
Intheliteratureonmodality,wefindvarioussystemsforclassifyingmodalelements,especiallysententialmodals,alongsuchparameters;inPortner(2009), Ioutlineatop-levelclassificationintoepistemic,priority,anddynamicmodality, withvarioussubtypes:
1.Epistemicmodalityhastodowithwhatcanbeconcludedbasedon someone’sknowledge.
Trueepistemicmodalityisnotcommonlydividedintosubtypes,butin logicandphilosophy,thereisconsiderationofsuchrelatedconceptsas alethicmodality(concerningconceptsoflogicalpossibilityandnecessity)andmetaphysicalmodality(concerningmetaphysicalpossibilityand necessity).Itisnotclearthatnatural languagemodalelementseverexpress theseconcepts,exceptinthecontextoftechnicallogical/philosophicaldiscourse.Historicalmodalityisakindofmetaphysicalmodalitywherethe relevantworldsarethosewithafuturethatiscompatiblewitheverythingthat hasalreadyoccurred,andshouldperhapsbeclassifiedasakindofdynamic modality.
2.Prioritymodalityhastodowithreasonsforassigningpriority,orpreference,toonetypeofsituationoveranother.
backgroundonmodality11
Thereareseveralsubtypes,suchas:deonticmodalityconcernspriority basedonrulesorrightandwrong,buleticmodalityconcernsprioritybased ondesire,andteleologicalmodalityconcernsprioritybasedongoals.We donotassumethatthesesubcategoriesaremutuallyexclusive.
3.Dynamicmodalityhastodowiththepossiblecoursesofeventsinthe world,basedonthefactualcircumstances.
Themostprominentsubtypeisvolitionalmodality.Volitionalmodals concerntheactionsavailabletoavolitionalindividual,withsub-subtypes includingabilitymodality(focusontheindividual’sabilities),opportunitymodality(focusonthecircumstancessurroundingtheindividual),and dispositionalmodality(focusontheindividual’sdispositions).Thereare alsoformsofdynamicmodalitywhicharenottiedtoavolitionalindividual, andIwillcalltheseintrinsicmodality.Asomewhatspecialvarietyof dynamicmodalarethequantificationalmodals,whichseemtoinvolve quantificationoverindividuals.
ExamplesofallofthesesubtypesfromPortner(2009,ch.4)aregivenin(13)–(15):
(13) Epistemic
(a)Atyphoonmayhittheisland.
(b)Marymusthaveagoodreasonforbeinglate.
(14) Priority
(a)Deontic:Therichmustgivemoneytothepoor.
(b)Buletic:Youshouldtrythischocolate.
(c)Teleological:Youcouldaddsomemoresalttothesoup.
(15) Dynamic
(a)Volitional:
(i)Johncanswim.(ability)
(ii)Youcanseetheoceanfromhere.(opportunity)
(iii)Marywilllaughifyoutellherthat.(dispositional)
(b)Intrinsic:
(i)Thecupisbreakable.
(ii)Everyempireeventuallyfalls.(historical)
(c)Quantificational:
(i)Aspidercanbedangerous.(existential)
(ii)Aspiderwillbedangerous.(universal)
FollowingKratzer(1981,1991),priorityanddynamicmodalityareoftengrouped togetherascircumstantialmodality(i.e.modalitywhichmakesreference tofactualcircumstances,ratherthanonlyanindividual’sknowledgeorbeliefs), andfollowingthisterminology,dynamicmodalitycanbecalled“pure”circumstantialmodality(circumstantialmodalitywhereprioritiesdonotplay arole).
Onefeatureoftraditionalclassificationswhichmightbelessthanidealisthe greatdifferenceitimpliesbetweentwousesofwordslike likely, certain,and chance
Accordingtothestandardclassificationabove, likely isambiguousbetweenan epistemicmeaningin(16a)andadynamicmeaningin(16b).
(16)a.Giventheevidence,Bobislikelytobehidinginthebasement. b.Thewaythatstormisdeveloping,itislikelytospawnatornado.
Itisusefultohaveacovertermfortypesofepistemicanddynamicmodalitywhich expresseithersubjectiveorobjectivechance;wecanusethetermpredictive modalityforthisclass.
Iwillrefertothedifferencesinmeaningalongthedimensionsoutlinedin(13)–(16)asdifferencesofjudgmenttype.(Intheliteratureandespeciallyinthespoken jargonofsemantics,theyareoftendescribedas“flavors”ofmodality.)SeePortner (2009)fordiscussionofotherclassificationschemes.Notethatmanyoftheabove examplesinvolveEnglishmodalauxiliaries.Othervarietiesofmodalityarefound inotherlanguages,andinotherconstructionswithinEnglish,butthesehavenot madetheirwaysintothegenerallysharedterminologyofsemanticists.
Yetanotherparameteralongwhichmodalmeaningsvaryisthatofstrength. Scholarswhostudymodalitymakeatleastatwo-waydistinctionbetweenstrong andweakmodals,althoughnoteverylanguagemayactuallyhavemodalsofboth strengths(Deal2011).Both should and may canbedeontic,but(17a)isstronger than(i.e.itentails)(17b):
(17)(a)Heshouldvoteforher.(strong)
(b)Hemayvoteforher.(weak)
Similaroppositionsexistwithineachoftheothersubtypesofmodality,ascanbe seenin(13)–(15).Inthoseexamples, must, should,and will wouldbeclassifiedas strongmodals,whiletheotherswouldbeclassifiedasweak.Strongmodalsare sometimescallednecessitymodals,andweakonespossibilitymodals,onthe groundsthat itisnecessarythat isstrongand itispossiblethat isweak.
Whenwelookbeyondmodalauxiliaries,itbecomesclearthatstrengthisnot atwo-waydistinction.Inthefollowing,(18a)isstrongerthan(18b),andsoforth downtheline:
(18)(a)Itwillcertainlyrain. (b)Itwillalmostcertainlyrain. (c)Itwillprobablyrain. (d)Thereisareasonablechancethatitwillrain. (e)Itisjustpossiblethatitwillrain.
Itseemsthatmodalstrengthisgradable,anditisnaturaltothinkofthisgradability asbeingsimilartothegradabilityofconcreteproperties,suchasheightandweight. (Gradablemodalityiscurrentlyatopicofmuchstudyinsemantics;seeforexample
Rubinsteinetal.()findthatspeakershavesignificantdifficultiesmakingthedistinction betweenepistemicanddynamicmodalityincertaintexts.
backgroundonmodality13
Portner2009;Yalcin2010;Katzetal.2012;Klecha2014;Lassiter2016.)Asweseein thesefewexamples,variationinstrengthcanarisefromacombinationoflexical choice(certainly isstrongerthan probably)andcompositionalsemantics(almost certainly isweakerthan certainly).Amongstrongmodals,elementsincluding should and ought aresometimescalled“weaknecessitymodals,”becausetheyfeel weakerthanother(“strong”)necessitymodalslike must.(Theterminologyissomewhatconfusinghere,becauseaweaknecessitymodalisstillastrong,ornecessity, modalbyourterminology.Aweaknecessitymodalistheweakersubtypeofstrong modal.)It’snotyetclearwhetherweaknecessitymodalsarelogicallyweakerthan theirstrongcounterparts,orwhethertheydifferinmeaninginsomeotherway (e.g.vonFintelandIatridou2008;vonFintelandGillies2010;Rubinstein2012).
1.3.2Modalityinpossibleworldssemantics
Inordertofullyengagewiththeideasanddebatesintheanalysisofmoodwhich willoccupythebulkofthisbook,itisimportantforreaderstohaveagood understandingofsemantictheoriesofmodality.ThegoalofPortner(2009)was tointroducetheanalysesofmodalitywithinfoursemanticframeworks:modal logic,Kratzer’sorderingsemantics,dynamicsemantics,andcognitivesemantics. Icannotrecapitulatethatdiscussionhere,butinsteadsuggestthatreaderswithout sufficientbackgroundbeginwiththatbookorotherresourcesonthesemantics ofmodality.Inthissection,Iwillreviewsomeofthekeyideasoftheimportant traditionofmodalsemanticsbasedonpossibleworlds.Thistraditionbeganwithin modallogicandhasevolvedintothestandardaccountwithinformal(linguistic) semantics.Itisofthemostrelevanceherebecausethebulkoftheoreticallyinformedworkonmoodtakesplacewithinthisframework.Mygoalwiththis reviewofpossibleworldsanalysisofmodalityistoemphasizeaspectsofthe approachwhichwillproveespeciallyimportantintherestofthebook,toestablish myownterminologicalandnotationalpreferences,andtojogthememoriesof readerswhomayknowaboutthesemanticsofmodality,butnothavethought aboutitinawhile.
Thestandardtheoryofmodalitywithinpossibleworldssemanticsfocuseson sententialmodalconstructions, andaimstoexplaintheirjudgmenttypeand strength.
Judgmenttype. Withinpossibleworldssemantics,judgmenttypeisdetermined bytheparticularsetofpossibleworldsonwhichthemodal’smeaningisbased. Forexample,withtheepistemicexample(13a),thejudgmentbeingmadeisthata
Subsententialanddiscoursemodalityareconsideredimportanttopicstowhichthetheoryshould beextended(andmuchofthisbookconcernssuchattempts),butnotexemplarsofthestandard theoryasitstands.Anexceptiontothisstatementissubsententialmodalitythatgetstreatedasifit weresententialmodality,forexampletheuseofmodaladjectiveswithsententialcomplements: Itis necessary/likely/possiblethatitisraining. Thesetreatmentsgenerallyignorefeaturesofthesubsentential modalconstructionswhicharenotsharedbysententialmodals.
14introduction
typhoonhittingtheislandiscompatiblewiththespeaker’sknowledge.Supposewe identifytheset K ofworldscompatiblewithwhatthespeakerknows;inthatcase, wecanexpressthejudgmentof(13a)bysayingthatsomeoftheworldsin K contain situationsinwhichatyphoonhitstheisland.Inmoreformalterms,modallogic basesthesemanticsofaparticularmodalelementonanaccessibilityrelation betweenworlds R.In(13a),wemightuse Rep :
(19)Foranyworlds w∗ ,and w: Rep (w∗ , w) iffeverythingthespeakerknowsin w∗ alsoholdsin w
Taking w∗ tobetheactualworld,thesetofaccessibleworlds K is {w : Rep (w∗ , w)} Wecandefinethetruthconditionsof(13a)asfollows:
(20) Atyphoonmayhittheisland istrueinaworld w∗ iffthereissomeworld w suchthatboth Rep (w∗ , w) andatyphoonhitstheislandin w.
Thoughthedefinitionsin(19)–(20)showtechnicallyhowanaccessibilityrelation works(andsomightbeusefulforsomepurposes),itisfartoosimple.Forinstance, itassumesthattheonlypersonwhoseknowledgewecareaboutisthespeaker’s andtheonlyknowledgeofhisthatwecareaboutisthatwhichhehasatthepresent moment.Ifthereareexampleswheretheknowledgeofanindividualotherthan thespeaker,oratatimeotherthanthepresent,isrelevant,we’dneedadifferent accessibilityrelation.
Letusexaminesomedatatoseewhocanbethe“knower”andwhatcanbethe “knowingtime”incasesofsententialepistemicmodality.
(21) [Twochildrenarediscussingwhetherthecreaturetheycaughtisanewtora salamander.]
(a)Thismightbeasalamander.
(b)Itmighthavebeenasalamander.
(c)Ryansaidthatitmightbeasalamander.
In(21a),theknoweriseitherthechildspeakingorthetwochildrenjointly,and theknowingtimeisthespeechtime.In(21b),theknowerisagainthespeaker ortwochildrenjointly,whiletheknowingtimecouldbeeitherthespeechtime orsometimeinthepast,forexample,whenthechildrenstillhadthecreaturein theirhands.(Toseethelatterpossibility,considerthecontinuation butitturns outitwasn’t.)In(21c),Ryanistherelevantknower,andtheknowingtimeisthe (past)timeatwhichhespoke.Clearly,thedetailsoftheaccessibilityrelationcan varyfromcasetocase,andyetthisvariationisverymuchlimitedbygrammatical factors.Itwouldbeverydifficultforoneofthechildrentouse(21a)tomakea modalstatementbasedonwhatRyanknew.
Itisanimportantgoalformodalsemanticstocomeupwithanadequatetheory ofthekindofvariationillustratedin(21).Onewaytodothisistoincorporatea contextsituation s intothedefinitionoftheaccessibilityrelation.Theknowerand knowingtimeareextractedfromthecontextsituation,whilethecontextsituation
backgroundonmodality15 itself(asthenamesuggests)isdeterminedbythelinguisticorextralinguistic context,inwaystobedetermined:
(22)Foranycontextsituation s andworlds w∗ , w: Rep (s)(w∗ , w) iffeverything thatthethinkingparticipant(s)of s knowin w∗ atthetimeof s alsoholds in w.
Thereissomeredundancyin(22)becausethesetofaccessibleworldsdependson bothacontextsituation s andaworld w∗ .Butifweassumethateachsituationis onlypartofasingleworld,andthatitonlydeterminesaccessibilityfromtheworld ofwhichitisapart,wecanbasetheaccessibilityrelationonthecontextsituation alone,asfollows:
(23)Foranycontextsituation s andworld w: Rep (s, w) iffeverythingwhichthe thinkingparticipant(orparticipants)of s knowin s alsoholdsin w.
Inrootsentences,itisalwayspossible(andusuallypreferred,ifnotoutright required)for s toincludethespeakerasamongthethinkingparticipants,and forthetimeof s toincludethetimeatwhichthesentenceitselfwasused.Thus, thetypicalinterpretationofarootsentencewith might concernsthespeaker’s knowledgeatthetimeofutterance.Thisiswhatweseein(21a).However,thetime candivergefromthespeechtime,particularlyinthepresenceofperfectaspect (havebeen in(21b)),andit’sevenpossibleforthethinkingparticipantsnotto includethespeaker,forexampleinanarrative.Inembeddedclauses,thesituation s istypicallydeterminedbygrammaticalfactors;in(21c),weseethematrixverb said controllingtheinterpretationofthemodalintheembeddedclausebymaking surethatRyanisthethinkingparticipantof s
Otherjudgmenttypescanbeassociatedwithaccessibilitydefinitionssimilarto (23),suchasthefollowingforadeonticmodalusedtomakeastatementaboutwhat thelawrequires(forexample, Thatguyshouldputacoinintheparkingmeter ):
(24)Foranycontextsituation s andworld w: Rlegal (s, w) iffallofthelawsinforce in s arefullycompliedwithin w
Wedescribeaworldaccessiblebytherelationassociatedwithanepistemicmodal asanepistemicallyaccessibleworld.Similarly,wemaytalkaboutdeonticallyaccessibleworlds,buleticallyaccessibleworlds,andsoforth.Wecan alsouselessesotericlanguagelike“knowledgeworlds”(forepistemicallyaccessible ones)and“desireworlds”(forbuleticallyaccessibleones).
Strength. Differencesinstrengthareanalyzedwithinmodallogicandsemantic systemscloselybasedonmodallogicasadifferencebetweenuniversaland existentialquantification.Thestrengthofagivenmodalisalsoknownasits
Hacquard()arguesthatthemeaningsexpressedbymodalauxiliariesarerelativetoeventsin awayanalogousto().Shealsoassumesthatthejudgmenttypecanbedeterminedfrom s,sothata singlegeneral-purpose R canworkforallflavorsofmodality.