∗ Tense, Time, and Adverbs in Italian Sign Language SandroZucchi Universita` degli StudidiMilano August13, 2005 Abstract InItalianSignLanguage(LIS),whenpastorfuturetimeadverbsarepresent, thesignsforverbsexhibitthesame manualconfigurationswhetherthesen- tencereportsapasteventorafutureevent. Factsofthiskind,alsoobserved forAmericanSignLanguage(ASL)andothersignlanguages,haveledsome authors(Friedman1975,amongothers)toconcludethattheselanguages,on aparwithspokenlanguageslikeChinese,lackgrammaticaltense. Neidleet al. (2000)andJacobowitzandStokoe(1988)have challenged thisviewfor ASL and have argued thatASL sentences contain tense markers. I present somedatashowingthatLISverbsinflectfortense. Iargue,moreover,thatthe apparentlackoftense inflectionwhenLIS pastandfuturetimeadverbs are presentisduetothefactthattheseadverbsshiftthespeechtimeandthatLIS pastandfuturetensesare absolutetenses. Iprovideaformalaccountofthe LIStense system basedonthese assumptions. The accountisimplemented inHeim’s(1997)analysisoftense. 1 The starting point ConsiderthefollowingsentencesofItalianSignLanguage(LIS):1 ∗The data presentedin this paper were collectedfrom native signersof LIS from the Napoli- Salerno area in the South of Italy. I thank Giuseppe Amorini and Giammarco Eletto, my main informants, for providing the data and for discussingthem with me. Special thanks to Marilina Cortazziforhelpinunderstandingthedata.IthankIlariaFrana,DanielaFinizio,andDanielDonato, whoparticipatedincollectingthedataandhelpedmetothinkaboutsomeoftheissuesIdiscusshere. VersionsofthispaperwerepresentedatGoingRomance2002andattheWorkshoponIntensionality at ESSLI2002, I thanktheir audiencesfor discussion. Finally, I thankthe studentsof my course onsignlanguageandlinguisticsattheUniversityofSalernoforprovidinganattentiveaudienceon whichIcouldtrysomeofmyideas. 1I follow the standardpracticeof usingcapitalizedwordsto representsigns. All the example sentencesinthepaperarefromLIS,unlessindicatedotherwise. 1 (1) GIANNIHOUSEBUY “Gianniisbuyingahouse” (2) TIME-AGOGIANNIHOUSEBUY “SometimeagoGianniboughtahouse” (3) TOMORROWGIANNIHOUSEBUY “TomorrowGianniwillbuyahouse” One difference betweentheseLIS sentencesand theirEnglishtranslationsisthat, while the English verbs are inflected for tense, in LIS the sign for the verb ap- pears in its citational form. Similar facts have also been observed for American SignLanguage(ASL)andhaveledsomeauthors2 toconcludethatthesesignlan- guages,onaparwithspokenlanguageslikeChinese,lackgrammaticaltense. Nei- dle,Kegl,MacLaughlin,BahanandLee(2000)andJacobowitzandStokoe(1988) have challenged this view for ASL and have argued that ASL sentences contain tensemarkers. Inparticular,Neidleetal. claimthatASLhasasetoflexicaltense markers located in the head positionof TP that differ, both morphologicallyand distributionally,from related timeadverbs; Jacobowitzand Stokoeargue thatcer- tain movement features of ASL verbs encode tense inflection. In this paper, I’ll presentsomedatashowingthatLISverbsarealsoinflectedfortense,andI’llargue that,contrarytowhat(1)-(3) suggest,tenseinflectiondoesnotworkdifferentlyin LISandinspokenlanguageslikeItalian. In section 2, I’ll describe how temporal information is encoded in LIS and I’llshowthatLIS verbsinflectfortense,thoughtenseinflectionandtimeadverbs interact differently in LIS and in spoken languages like Italian and English. In section3,I’llpresentaninformalsketchofhowtheinterpretationoftenseandtime adverbsworksinaparadigmatictense-inflectedlanguage: Italian. Insection4,I’ll describetheintuitiveideasonwhichmyaccountoftheinteractionoftenseandtime adverbs in LIS is based. Section 5 containsthe official proposal: formal account of the LIS tense system and of itsinteractionwith time adverbs based on Heim’s (1997)analysisoftense. Section6discussessome relateddatafrom LIS. Section 7tacklessomeissuesleftopenbytheanalysisandleadstoarefineddescriptionof thetypologyofLISadverbs. Section8containssomefinalconsiderations. 2 Time in LIS ThereareatleastfourwaysofconveyingtemporalinformationinLIS: (i) bymeansoftimeadverbs, 2SeeFriedman(1975)forASLandPizzuto,Cameracanna,CorazzaandVolterra(1995)forLIS. 2 (ii) bymeansofsuprasegmentalfeaturesco-occurringwiththeverb, (iii) bymeansoflexicalmarkerslikeDONE3andMUST, (iv) bymeansofcontext. Sentences (2)-(3) in the previous section are examples of (i). Sentences (4)-(6) belowinstantiate(ii): shoulderstraight (4) GIANNIHOUSE BUY “Gianniisbuyingahouse” shoulderbackward (5) GIANNIHOUSE BUY “Gianniboughtahouse” shoulderforward (6) GIANNIHOUSE BUY “Gianniwillbuyahouse” In these sentences, the positionof the shoulderwhilethe verb is beingsignedin- dicatesthattheactionispresent,pastor future: iftheshoulderisalignedwiththe restofthebody,theactionisclaimedtobetakingplaceatthetimeofutterance;if theshoulderistiltedbackward,theactionisclaimed totakeplacebeforethetime of utterance; if the shoulder is tilted forward, the action is claimed to take place afterthetimeofutterance.4 Sentences(7)-(8)areinstancesof(iii): (7) GIANNIHOUSEBUYDONE “Giannihasboughtahouse” (8) GIANNIHOUSEBUYMUST “Gianniwillbuyahouse” The sign DONE in (7) indicates that the action is completed before the time of utterance and the sign MUST in (8) indicates that the action takesplace after the timeofutterance.5 3ThissignistraditionallyglossedinItalianasFATTO.Here,I’llkeeptheEnglishglosssinceits meaningismoretransparentforEnglishreaders. 4Here,Ishouldmentionthatthisuseoftheshouldertoconveytemporalinformationismainly foundinthevarietyofItalianSignLanguageusedintheSouthofItaly. 5Theuseoflexicallycontentfulelementstoperformgrammaticalfunctionsiscommoninsign languagesandinCreoles. Onthispoint, seeFischer(1978),FischerandGough(1999),andMeir (1999). 3 Finally,temporalinformationmaybegatheredfromthecontext. Forexample, in discourse (9) below the first sentence specifies that the action of going to the moviesoccurredyesterdayandthefollowingsentenceisunderstoodasdescribing a past actionas well, althoughit lacks an overt marker indicatingthat the time is past: (9) YESTERDAYGIANNIMOVIE-THEATERGOTHEREMARIA HIMMEET “YesterdayGianniwenttothemovie-theater. Mariamethimthere” Data(4)-(6) suggestthattheview thatLIS lacksgrammatical tense shouldbe reconsidered. The positionof the shoulderco-occurring withthesign of theverb seems to play the same role as tense inflection on the verb in spoken languages likeItalianandEnglish,whatchangesissimplyhowgrammaticaltenseismarked: by means of suffixes or stem modification in Italian and English and by means of a suprasegmental element, the position of the shoulder, in LIS. On the basis of these considerations, I will assume that the shoulder position while the verb is signed in LIS is a way of inflecting the verb for tense, and, from now on, I’ll glossthesuprasegmentalfeature ontheverb in(4)-(6) aspresent,pastandfuture, respectively: pres (4) GIANNIHOUSEBUY “Gianniisbuyingahouse” past (5) GIANNIHOUSEBUY “Gianniboughtahouse” fut (6) GIANNIHOUSEBUY “Gianniwillbuyahouse” Evenifweregarddata(4)-(6)asinstancesoftenseinflection,andthusasevidence that LIS is unlike Chinese and more like Italian and English with respect to the markingofgrammaticaltense,thereisstillaconspicuousdifferenceinhowtenses andtimeadverbsinteractinLISandinspokenlanguageslikeItalianandEnglish. In Italian and English, past and future tense inflection co-occurs with past and futuretimeadverbs. InthecorrespondingLISsentencescontainingpastandfuture timeadverbs,ontheotherhand,pastandfutureinflectionontheverbisabsent: in (2) and(3) belowtheshoulderpositionisstraight,and notbackwardandforward respectively,asitshouldbeiftheLISverbBUYwereinflectedforpastandfuture tense. 4 (2) TIME-AGOGIANNIHOUSEBUY “SometimeagoGianniboughtahouse” (3) TOMORROWGIANNIHOUSEBUY “TomorrowGianniwillbuyahouse” Indeed, if we add past and future inflection to the verb in (2)-(3), the resulting sentencesareanomalous: past (10) *TIME-AGOGIANNIHOUSE BUY “SometimeagoGianniboughtahouse” fut (11) *TOMORROWGIANNIHOUSEBUY “TomorrowGianniwillbuyahouse” Thisistheproblemwefacethen: thatofexplainingwhyverbalinflectionandpast and future time adverbs interact differently in LIS and in spoken languages like ItalianandEnglish. Noticethatthequalificationregardingpastandfuturetimeadverbsisimportant here. Thelackofpastandfutureinflectionontheverbdoesnotarisewithalltypes of time adverbs in LIS. Indeed, with adverbs that in Italian and English fail to locate the event time relative to the utterance time, past and future inflectionsare possible. For example, theEnglishadverb today,by itself, doesnot locateevents in the past, present, or future with respect to the utterance time, as shownby the factthatthisadverbcanco-occurwithpast,future,andpresenttenses: (12) a. TodayJohnboughtahouse b. TodayJohnisbuyingahouse c. TodayJohnwillbuyahouse AdverbsofthiskindinLIS,likeinEnglish,canco-occurwithdifferenttenses: past (13) a. TODAYGIANNIHOUSEBUY pres b. TODAYGIANNIHOUSEBUY past c. TODAYGIANNIHOUSEBUY Finally,beforeIgoonwiththeplot,Ishouldpointoutthatthelackofpastand future tenseinflectionswithpastandfuture timeadverbs isno exclusiveproperty of sign languages. Some spoken languages behave in a way similar to LIS in this respect. England (1983) reports that in Mam, a Mayan language spoken in Guatemala and Mexico, past tense particles are obligatorilyomitted when a past 5 timeadverbispresent: (14) o chin jaw tz’aq-a past1st-abs. direct. slip-1st “Islipped” (15) eew chin jaw tz’aq-a yesterday1st-abs. direct. slip-1st “Islippedyesterday” Comrie(1985)reportsthatsimilar,thoughlessstrict,co-occurrencerestrictionsare foundinJamaicanandotherCreolelanguages. In the next section, I’ll prepare the ground for my account by providing an intuitivepictureoftheinteractionoftenseandtimeadverbsinItalian(I’llbeusing ItalianratherthanEnglishastheparadigmatictense-inflectedspokenlanguagefor reasonsthatwillbeclear later). Insections4-5, I’llcome back totheLISfactsin (2)-(3). Thedatain(13)willbeaddressedlateroninsection7. 3 Tense and time adverbs in Italian AccordingtoReichenbach(1947),naturallanguagetensesexpressrelationsamong times. In his theory, the time parameters at stake are three: the speech point, the eventpoint,andthereference point. Thereference pointplaysaroleindescribing the behavior of perfect tenses in English, while for the simple tenses the refer- ence pointcoincides with the event point.6 Since in the followingdiscussionthe reference point will not play a role, I’ll assume for simplicitythat tenses express relationsbetweentwotemporalparameters: thespeechpointsandtheeventpoint e. If we make this assumption, one might describe Italian (as well as English) simpletensesinthisway: thesimplepresenttenserequiresthespeechtimetoco- incidewiththeeventtime,thesimplepasttenserequirestheeventtimetoprecede the speech time, and the simple future tense requires the event time to followthe speechtime. Simplepast: e<s Simplepresent: e=s Simplefuture: e>s 6Thisisnotquitetrueforthesimplefuturetense,forwhichReichenbachalsoallowsaninterpre- tationwherethereferencepointcoincideswiththespeechpoint. 6 The intended result of this characterization is that, for example, sentences (16)- (18) beloware true ifthetime ofthe housebuyingevent precedes, coincides,and follows,respectively,thetimeofutteranceof(16)-(18). (16) Giannicompro` unacasa “Johnboughtahouse” (17) Giannicompraunacasa “Johnisbuyingahouse” (18) Giannicomprera` unacasa “Johnwillbuyahouse” Howdotimeadverbsinteractwithsimpletenses? IntheItaliansentencesin(19)- (20)(andintheirEnglishtranslations),thetimeadverbsspecifytheeventtime. In (20),thetimeadverbtellsusthatthetimeofthehousebuyingeventisincludedin thedaythatfollowsthetimeofutteranceandtheinflectionontheverbredundantly specifies that this time is in the future. In (19), the adverb tells us that the event timeisinthepastandthetensereiteratesthisinformation. (19) TempofaGiannicompro` unacasa “SometimeagoJohnboughtahouse” (20) DomaniGiannicomprera` unacasa “TomorrowJohnwillbuyahouse” ThisinteractionbetweentensesandtimeadverbsisillustratedinFigure1(wheret istheintervalspecifiedbythepasttimeadverb). e s t Figure1: TemporalrelationswithItaliansimplepasttenseandpasttimeadverbs There are several ways in which this preliminary sketch of the workings of tensesandoftheirinteractionwithtime adverbsneedstoberevised. For thetime being, I’ll ignore most of the complex issues that arise in this respect. There is, however, one pointthat needs to be clarified for the purpose of our discussion. I observed that, accordingto the relationalanalysisof tense, present tense requires theeventtimetocoincidewiththespeechtime. Asiswellknown,however,present tensemay alsobeusedtoreporteventsthatprecede thetimeof utterance. Thisis 7 showninthefollowingexamplesfromItalianandEnglish(duetoBertinetto1991 andHornstein1990): (21) Stavamoaspettandoiltreno. All’improvviso,giungetrafelatoEnrico. Ha appenaparlatoconilcapostazioneedicecheilrapidoarrivera` conmolto ritardo. Fucos`ıchedecidemmodiprenderel’espresso. “We were waiting for the train. Suddenly, Enrico arrives in a hurry. He has just spokento the stationmaster and says that the rapidowill arrive withalongdelay. Thatwashowwedecidedtotaketheexpresstrain.” (22) It was 1812, just before the Battle of Borodino. The anticipationof the coming struggle is palpable. Napoleon has just woken. He is getting ready to inspect the troops and see that they are ready for the battle that willdeterminethefateofEurope. How can this use of the present tense be reconciled with the assumption that the simple present identifiesthe event time withthe speech time? The answer isthat the term “speech time” for the temporal parameter s needs tobe qualified. In the simplecase, thistime intervalisidentifiedwiththe timeof utterance, but, ingen- eral, itneed notcoincidewithit. In thehistoricalpresent examples (21)-(22), the speechtimeisshiftedbackwithrespecttothetimeofutteranceandtheeventtime isidentifiedwiththeshiftedspeech time.7 Figure 2 illustratestherelationamong time of utterance, speech time, and event time in the historical present (c is the T timeofutterance). e,s c T Figure2: Temporalrelationsinhistoricalpresentsentences Therelationamongtimeofutterance,speechtime,andeventtimeinItaliansimple pastsentenceswithpasttimeadverbsisnowillustratedinFigure3. 7Theideathatthespeechtimemaybedistinctfromthetimeofutteranceisalreadypresentin Dowty(1982),wherethespeechpointisshiftedinthescopeofcertaintenseoperators. Hornstein (1990)suggeststhatthepossibilityofanchoringthespeechtimetotimesotherthanthetimeofut- terancemayaccountforthehistoricalpresent. InZucchi(2005),Iarguethatthiswayofanalyzing thehistoricalpresentaccountsforitsaspectualanddiscourseproperties. InZucchi(2001),thetem- poralpropertiesofmodifierslikeInthenovelareexplainedbyassumingthatthesemodifiersshift thespeechtimebutnotthetimeofutterance. 8 e c ,s T t Figure3: Locationofe,s,c withItaliansimplepasttenseandpasttimeadverbs T 4 Tense and time adverbs in LIS Nowthatwehaveseenanintuitivesketchofhowtensesandtimeadverbsinteract inItalian,let’scomebacktotheLISfactsin(2)-(3),(5)-(6): (2) TIME-AGOGIANNIHOUSEBUY “SometimeagoGianniboughtahouse” (3) TOMORROWGIANNIHOUSEBUY “TomorrowGianniwillbuyahouse” past (5) GIANNIHOUSEBUY “Gianniboughtahouse” fut (6) GIANNIHOUSEBUY “Gianniwillbuyahouse” Atfirstblush,thesedataseemtoshowthat,althoughLISverbscanbeinflectedfor tense,tenseinflectionisabsentwhena pastorfuturetimeadverbispresent. Why shouldtherebethisco-occurrencerestrictionontensesandtimeadverbsinLIS? A goodruleof thumb totellsomeone whowantstolearn how theinteraction oftensesandadverbsworksinLISisthis:pastandfuturetenses(i.e.,theshoulder positionsco-occurringwiththesignsoftheverbstoindicatepastandfuturetimes) are used only when the action isn’t already located in the past or in the future by an adverb; if the timespecificationintroducedbythe tenseis redundant,tense is dropped. This intuitive rule leads us to expect that tenses should be dropped in (2)-(3), as it indeed happens. However, effective as it may be for predicting the behavior of tenses in LIS, appeal to redundancy to explain what blocks the occurrence offuture and pastinflectionin(2)-(3) isnotsatisfactory. Redundancy inLISdoesnotyieldanomaly,astheacceptabilityof(23)shows: (23) IN-PASTYEARS-AGOMANYPARISGO “Inthepast,severalyearsago,IwenttoParis” 9 Moreover, appeal to redundancy fails to answer a natural question raised by the LIS data: whyisn’tthe same co-occurrence restrictionalso presentin othertense inflectedlanguageslikeItalianandEnglish? Aswehaveobserved,fromaseman- tic standpoint, tense in (20) is no less redundant than it is in (2)-(3), yet (20) is perfectlyacceptable: (20) DomaniGiannicomprera` unacasa “TomorrowJohnwillbuyahouse” Noticethat,ifindeedweviewthedatain(2)-(3)asshowingthattenseinflection is absent when a past or future time adverb is present, this raises a puzzle for grammaticaltheory. Itisusuallyassumedthattherearetwopossiblewaysinwhich nominativecase isassigned: by tense or byagreement.8 In LIS, however, person agreement showsupinthecomplementofraisingverbswherethesubjectNPhas undergoneraising:9 (24) HESEEMLEAVE DONE 3rdpers. “heseemstohaveleft” Sincein(24)thesubjecthasmovedtoreceivecase, itfollowsthatinLISnomina- tive case cannot be assigned by agreement.10 Thus, the only optionavailable for LIS isthat nominativecase is assignedbytense. But, if tenseis absent in(2)-(3) above,howiscaseassignedtothesubjectthere? Again,appealtoredundancyfails toanswerthisquestion. A possible hypothesis is that LIS, Italian, and English differ in this respect: in LIS, but notinItalian or English,time adverbs somehowfulfillthe functionof tense. Tosee howthisproposalmaybe statedmore precisely,let’sconsidersome factsaboutASL.Neidleetal.,aswementionedabove,haveclaimedthatASLhas a set of lexical tense markers located in the head positionof TP. One example of suchalexicaltensemarkeristhesignFUTUREinASLsentence(25): (25) JOHNFUTURE BUYAHOUSE (ASL) tns “Johnwillbuyahouse” The claim that these lexical items are indeed grammatical tense markers and not time adverbs is supported by the fact that they can be distinguished from time adverbs both from a distributional point of view and from an articulatory point 8SeeChomskyandLasnik(1993)andChomsky(1995) 9Following Bahan(1996) andPadden(1988), I assumethat sign languageverbs may express agreementwiththeirargumentsbyspatialmovement. 10CaseassignmentmayworkdifferentlyinLISandItalian,ifGiorgiandPianesi(1997)areright inclaimingthatcaseisassignedbyagreementinItalian. 10