Clitic pronouns and past participle agreement in … pronouns and past participle agreement in...

Clitic pronouns and past participle agreement in ItalianinthreehearingimpairedbilingualsItalian/LIS

FrancescaVolpato

The aimof this study is to assess the comprehension and to elicit theproductionofcliticpronounsinsentenceswithleft-dislocation,inthreeadulthearing impaired bilinguals Italian/LIS and in four hearing speakers ofItalian.TheoccurrenceofcliticpronounsandthepresenceofpastparticipleagreementinItalianareinvestigatedthroughsentencecompletionandgram-maticalityjudgementtasks.Resultssuggestthat,inspiteoflateexposuretothe oral language due to hearing impairment, the linguistic competence ofthehearingimpairedindividualsisquiteintactandthesyntacticinformationiscorrectlyprojectedinthesyntactictree.Becauseofaveryselectiveimpair-mentinthemorphologyofcliticpronouns,onlyaqualitativeanalysisoftheerrors is possible. The non-standard forms are discussed within the frame-workoftherecentresearchonϕ-featuresfromasyntactic,morpho-phonologi-calandpsycholinguisticpointofview.Thestatisticallysignificantbetterper-formanceonpluralcliticpronounsasopposedtosingularonesconfirmsthefindingsofpreviousstudiesongenderandnumberfeaturesandprovesonceagainthatnumberinformationisretrievedearlierthangenderinformation*.

1. Introduction

This paper investigates the linguistic competence in Italianof three adult hearing impaired bilinguals (Italian/Italian SignLanguage-LIS) through sentence completion and grammaticalityjudgementtasks,assessingthecomprehensionandelicitingthepro-duction of third person accusative clitic pronouns in sentences withleftdislocation.

Thiswork originateswithin the frameworkof generativegram-mar from much existing literature on language acquisition, both inItalianandother languages, andon the implications thatadelayedaccesstotheinputduetohearingimpairmentmaycausetothisproc-ess. The analysis of responses shows that the performance on cliticpronounsvariesaccordingtothepresenceofsimpleorcompoundtens-es,inwhichpastparticipleagreementisalsorequired,andaccordingtothedifferentcombinationofgenderandnumberfeatures(confirm-ingtheconclusiondrawnbyDeVincenzi&DiDomenico1999).

Rivista di Linguistica20.2(2008),pp.309-345 (ricevuto nel luglio 2008)

Francesca Volpato

310

Thepaperisorganizedasfollows.Sections2and3presentsomedata on the acquisition of clitic pronouns in individuals showingtypicalandatypicallanguagedevelopmentindifferentRomancelan-guagesanddemonstratethatobjectcliticsaresometimesproblematicandacquiredwithsomedelaywhencomparedwithsubjectcliticsordeterminers. Section 4 outlines the linguistic background account-ingfortheoccurrenceofthird-personobjectcliticsandforagreementphenomenaderivingfromtheirpresenceinleft-dislocationsentences.Section5focusesontheexperiment,presentingthesubjectsinvolved,the materials used and the results achieved in the different tasks,also showing a classification of the type of errors made by the par-ticipants. Section 6 discusses the collected data within the currentresearchongenderandnumberfeatures.

2. Typical and atypical language acquisition

First language acquisition is a fast and seemingly effortlessprocess.Everychildisbornequippedwithalanguagefaculty,whichrequires being stimulated very early in one’s life in order to allowhim/hertoattaintheknowledgeofhis/herownlanguage.Ifthisfac-ultyisstimulatedtoolate,languageacquisitionisdelayed,leadingtosevereconsequencesforthedevelopmentoflinguisticabilities.

Deafnessisasensoryimpairmentwhichleavesthelanguagefac-ulty intact,butwhichseriously limitstheperson’sabilitytoacquireanoral languagenaturally.Hearingimpairedpeoplecannotdevelopanorallanguageaseasilyastheirhearingpeersbecausetheydonothavedirectaccesstothelinguisticinput.

Many variables might influence the language development inhearingimpairedpeople.Inanycasethehearingimpairmententailsadelayedacquisitionandconsequentlyadifficulty in thedomainofgrammaticalmorphologyandintheuseofsomefunctionalelements,suchasdeterminers,auxiliaries,prepositionsandpronouns(Volterra& Bates 1989, De Villiers J.G etal. 1994, Fabbretti etal. 1998,Franchi 1998, Tuller 2000, Ajello etal. 2001, Volterra etal. 2001,Tuller & Jakubowicz 2004, Fabbretti 2006, Fabbretti etal. 2006,Delage & Tuller 2007). These alone have no intrinsic meaning, buttheirpresenceisfundamentalinderivingthemeaningofasentence.

However,itisquitepossibleforhearingimpairedsubjectstofacethisarduoustask,aswasthecaseinastudyconductedontwoorally-trained hearing impaired adults (Volpato 2002). This study investi-gated the production and comprehension of clitic pronouns in left-

Clitic pronouns and past participle agreement in Italian

311

dislocation sentences and showed that the linguistic impairment ofthesesubjectsisverylimitedandtheircompetenceinItalianisquiteintact.Thehearingdeficit,whichhindered themfromacquiring theItalianlanguagenaturally,hasnotinfluencedtheirlearningprocess,and it is certain that the longperiod of rehabilitation they enduredhasallowedthemtogainagoodknowledgeofItalian.

Thesameconsiderationsarevalidforthethreehearing-impairedbilinguals Italian/LISstudied in thispaper. I showthat these threeindividualshave indeedattainedaverygood competence in Italian,withaveryselectiveimpairmentinthemorphologyofcliticpronounswhichissignificantforthetheoryofϕ-features.

3. The development of clitic pronouns across different acquisition modes

Themorpho-syntacticimpairmentintheuseofcliticpronounsisnotasurprisingphenomenon.Indeed,itiswell-knownfromliteraturethattheseelementsarecomplexforhearingimpairedpeople,fortypi-cally-developing individuals and for other categories of individualsshowingatypicallanguagedevelopment.

Chesi (2000) & (2006) investigated the production of hearing-impairedItalianchildrenandfoundthattheyomittedagreatdealoffunctionalelements(articles,clitics,prepositions,auxiliaries).1WhatChesi noticed in his study was a widely attested tendency to avoidaccusative clitic pronouns, by adopting a strategy that involved thesubstitutionofthecliticpronounwithaDP(1a)ortheomissionalto-getherofthecliticandofthedirectobjectcomplement(1b):

(1) a.Poiiltoposcappaearrivailgattoperacchiappareiltopo… ‘thenthemouserunsawayandarrivesthecattocatchthemouse…’ TargeT:Poiiltoposcappaearrivailgattoperacchiapparlo. thenthemouserunsawayandarrivesthecattocatch-it:Cl

‘thenthemouserunsawayandthecatarrivestocatchit’ b.…lapadrona(ø)haaperto(ø),hapresoilgatto. …‘themistresshasopenedø,hascaughtthecat’ TargeT:…‘lapadrona(lo) haaperto(ilfrigorifero)ehapresoilgatto.’ …themistress(it:Cl)hasopened(thefridge)andhascaughtthecat ‘…themistressopenedit/thefridgeandcaughtthecat.’

Thesedifficultiesare,however,notpeculiartohearingimpairedpeople.Theproductionofaccusativecliticshasalsobeeninvestigatedin typical first and second language acquisition (Pirvulescu 2006,

Francesca Volpato

312

Hamann & Belletti 2006, Wexler etal. 2002) as well as in atypicallanguage development, such as SLI children (Leonard etal. 1992,Bortolini etal. 2006, Jakubowicz etal. 1998), across different lan-guages(Italian,French,SpanishandCatalan).Incomparingtheuseofdeterminers,andnominativeandobjectcliticsbyalltheseindivid-uals, thesestudiesrevealedthat,whiledeterminersandnominativeclitics are accurately produced, object clitics are more problematicand are mostly avoided, resulting again in ungrammatical omissionorinfelicitousrepetitionofafulllexicalDP.

For Spanish and Catalan, Wexler etal. (2002) conducted anexperimentontypically-developingchildreninordertocheckwhetherthey produce or omit clitics and whether they produce the correctclitic forms inboth languages.Resultsshowthat inboth languages,theproducedcliticsareplacedcorrectlybeforefiniteverbsandafternon-finiteverbs.Withrespecttotherateofcliticomission,theyfoundthat Catalan-speaking children omit clitics significantly more fre-quently than Spanish-speaking children and that the omission rateis above all very high at early stages of language acquisition. Thenomissionsbegintodecreasebytheageofthreeandcompletelydisap-pearbytheageoffour.

The results of the study carried out on spontaneous speech inFrench first language acquisition by Pirvulescu (2006), and the lin-guisticresearchconductedbyHamann&Belletti(2006)onearlyL1andL2,andadultL2acquisitionofFrench,showasimilardelay inthe acquisition of object clitics. These two studies investigate theacquisitionofobjectcomplementsincomparisontosubjectpronounsandrevealthatintypicallydevelopingmonolinguals,complementcli-ticsbegintoappearaboutfourtosixmonthslaterthansubjectclitics.ThistimespanislongerinearlyandadultL2speakers.Inallcases,themostcommonstrategyusedtoovercomethedifficultiesderivingfromthenecessitytoplacethedirectobjectisagaintheuseofafulllexicalDP.However,assoonastypically-developingchildrenbegintoproduceobjectclitics,theyplacethemcorrectly,asopposedtoL2andbilingualchildren,whotendtomisplacethem.2

Hamann & Belletti (2006) also focussed their attention on theacquisition of complement clitics in French-speaking SLI children.Consistent with the results achieved by Jakubowicz etal. (1998) inthe same languagepathology, this study revealed thatSLI childrenshow lower accuracy than typically-developing children in the useofobjectcliticpronouns,comparedwithdefinitedeterminersorsub-jectpronouns.ForItalian,Leonardetal.(1992)andBortolini(2006)comparedSLIchildrenandtypically-developingchildrenintheuseof


313

bothcliticpronounsanddeterminers.Althoughbothelementsprovedto be difficult for SLI children, determiners were nonetheless moreaccuratethancliticpronouns.

Theabsenceorvery rareuseof complement cliticsmaypersistforseveralyearsinSLIchildren,thetimespanfortheiracquisitionbeingevenlongerthanfortypically-developingones.

Although in thispaperweare concernedwithatypical languageacquisitiondue tohearing impairment,all thesestudies revealapat-ternofdevelopmentcommontoallacquisitionmodes.Thedifficultyandthedelay in theacquisitionof thirdpersonaccusative clitics thatallindividualsinthevariousRomancelanguages(includingItalian)showarepresumablyduetothe‘reduced’statusofthesefunctionalelements,both from a syntactic, phonological and morphological point of view(Cardinaletti&Starke1999),and to thecomplexsyntacticmovementtheyundergointheirwayuptothetopofthesyntactictreewhencom-paredtothatofweakandstrongpronouns.Thepresentstudyisonlyconcernedwiththeoccurrenceofpre-verbaldirectobjectclitics,whichoccupya functionalhead in thehighpart of the clausal structurebyattachingontotheverbthatcarriesthefeaturesoffinitemorphology.

In the next section I present some linguistic proposals thataccountfortheoccurrenceofaccusativecliticpronounswithinclausestructure.

4. Linguistic background

In this section I outline the main properties of the sentencesadministered in the experimental task, as well as issues on thesestructures raised by linguists within the framework of GenerativeGrammar.

4.1 The analysis of Clitic Left-Dislocation (CLLD)

TheItalian languagemakes itpossible for theobject tobe left-dislocated,thatis,itoccupiesapositionattheverybeginningofthesentence,intheleftperipheryofit,andaresumptivecliticisplacedbeforethefiniteverb:

(2) a.Illattei,ilbambinoloibeveognimattina. The:m.sgmilk:m.sg,thechildit:m.sg drinkseverymorning ‘Thechilddrinksiteverymorning(themilk).’

Francesca Volpato

314

b.Iladrii,ilpoliziottoliihaarrestatii. The:m.plthieves:m.pl,thepoliceman,them:m.pl hasarrested:m.pl

‘Thepolicemanarrestedthem(thethieves).’

Such doubling structure, in which both a full DP and a cliticpronoun appear in the sentence, is called Clitic Left Dislocation–CLLD– (Benincà 1988/2001, Cinque 1990) and is largely attestedinItalian.Inleft-dislocationsentences,cliticpronounsreferbackandare co-indexed with an antecedent (DP) previously given in the lin-guisticorextralinguisticcontext.Thecliticpronounandtheanteced-entsharethesameϕ-features.

Two types of analysis of CLLD have been put forward in theliterature.Accordingtoabase-generationhypothesis(Cinque1990),thedislocatedelementandthecliticaregenerated intheirsurfaceposition. Other studies suggest instead that these elements attainthe surface position by movement (Cinque 1977, Cecchetto 1999,Torrego1992,Belletti2005andothers).Thehypothesisadopted inthispaperisthemovementanalysisofCLLDsuggestedbyTorrego(1992)andCecchetto (1999).3Becauseof thepresenceof twoargu-ments in the direct object position, they assume that in doublingstructures likeCLLDsentences, the leftdislocatedelement isbasegeneratedintheVPinthe‘double’positionwithintheso-called‘bigDP’ argument. The clitic is thus inserted from the lexicon as theheadofthisspecialDP,whichhoststhedoubleXPinitsSpecposi-tion.

(3) VP

V BIGDP

DP X’

Doubleobject CL …

InCLLD, the topicalizedconstituentmovesandendsup in thespecifier position of a Topic projection in the left periphery of theclause(Rizzi1997),andtheresumptiveobjectcliticpronouncliticizesontotheverb.


315

4.2. Past participle agreement

In configurations with compound tenses, like those in (2b) and(4),thereisovertagreementinϕ-featuresbetweenthecliticandthepastparticiple:

(4) a.Lihaarrestati. (he) them:m.plhasarrested:m.pl

‘Hehasarrestedthem.’ b.Lehaarrestate. (he) them:f.plhasarrested:f.pl

‘Hehasarrestedthem.’

Kayne(1989)suggeststhattheϕ-featuresofgenderandnumberappearingonthepastparticipleshouldbetheresultofanagreementrelation between the past participle and the moved element (nounphraseorclitic).Thesuppositionderivingfromthisassumptionisthatthere is anAgrprojectionabovepastparticiple and that themovedconstituent triggersagreement in itspassingthrough itsSpeconthewaytoitsfinallandingsite.ThisAgrprojectionwasidentifiedwiththeAgrOprojectionresponsibleforaccusativeCasecheckinginChomsky(1993).Friedemann&Siloni (1997)provideevidence that theprojec-tionassociatedwithaccusativeCase(AgrOP)istobekeptdistinctfromtheprojectionwherenumberandgender features ofpastparticiplesarechecked(labelledAgrPstPrtP),andsuggestthefollowingstructure:

(5)

Francesca Volpato

316

Thisstatementderivesfromsomeobservationsonpassive,unac-cusative and transitive verbs. Passives and unaccusatives (unliketransitive verbs) are unable to assign accusative Case and do notrealize an external argument. Consequently they lack the AgrOPprojection. If accusative Case assignment and past participle agree-ment were one and the same process, they would always co-occur.Onthecontrary,checkingofaccusativeCaseoccursindependentlyofthepresenceorlackofapastparticipialmorphology,andpastparti-cipleagreementiscompletelyindependentofthepresenceofaccusa-tive Case checking. This is confirmed by the obligatory agreementbetween past participle and grammatical subject with unaccusativeverbsandinpassivesentences(6a-b)orbythelackofpastparticipleagreementwithtransitiveverbstheobjectofwhichdoesnotundergocliticization(7):

(6) a.Mariaèarrivata. Maria:f.sghasarrived:f.sg

‘Mariahasarrived.’ b.LafinestrasaràapertadaMaria. Thewindow:f.sg willbeopened:f.sgbyMaria ‘ThewindowwillbeopenedbyMaria.’

(7) Giannihaapertolaporta. Giannihasopened:m.sg (default)thedoor:f.sg

‘Giannihasopenedthedoor.’

Inconclusion,cliticpronounscheckaccusativeCaseinAgrOandpastparticipleagreementinAgrPstPrtP4.Thederivationofcliticpro-nounswithsimpleandcompoundtenses isdifferentiatedas follows.Atwo-stepmovementisinvolvedinthederivationofcliticpronounswith simple tenses (Sportiche 1989; Belletti 1999; Cardinaletti &Starke1999),whereonlysyntacticeffectsarevisible:

(8) a.pro compro lemele (i) buy the:f.pl.apples:f.pl. ‘Ibuytheapples.’

b.[AgrSP pro Le compro[AgrOP__[VP__]]]

(i) them:f.pl.buy ‘Ibuythem.’


317

A three-step movement is instead required in the derivationof compound tense sentences displaying past participle agreement(Belletti 1999; Friedemann & Siloni 1997; Kayne 1991), where thederivationofcliticpronounshasvisibleeffectsbothfromasyntacticandamorphologicalpointofviewandisthereforemorecomplex:

(9) a.pro Ho comprato lemele. (i) have bought the:f.pl.apples:f.pl. ‘Ihaveboughttheapples.’

b.[AgrSP pro Le ho[AgrOP___[AgrPstPrtP__comprate[VP__]]]]

(i) them:f.pl. have bought:f.pl. ‘Ihaveboughtthem.’

5. The experiment

In this section, I describe the experiment I have conducted onthree hearing-impaired adults. I present the subjects involved, thematerials and methods of administration, the results and the mainerrorsmadebytheparticipants.

5.1. The participants

Two small groups of subjects participated in the experiment: agroup composed of three adult hearing impaired bilinguals Italian/LISandacontrolgroupcomposedoffourItalian-speakingadults.

The participants belonging to the hearing impaired group arelistedinthefollowingtable:5

Table 1.Hearingimpairedparticipants.

SUBJECTS AGE LOSS(dB)S1 26 70S2 25 100S3 30 100

Noneof thehearing impaired subjects claimed towearhearingaidsatthetimeoftheexperiment,eventhoughtheyusedthemwhentheywerechildren.

Francesca Volpato

318

All of them were diagnosed as hearing impaired at birth andtheyacquiredtheItalianSignLanguage(LIS)astheirfirstlanguage.They were born to hearing impaired parents and mainly attendedspecialschoolsforthedeaf.SpeechtherapyandlinguistictrainingofItalianspokenlanguagewerecarriedoutduringschooltimebytheirschoolteachers.TheactivitiesthroughwhichtheyweretrainedtotheItalian language were mainly dictations and reading tasks. At themoment,allofthemareteachersofLISattheuniversityorinspecialschoolsforLISlearners.AlthoughtheyhavebeenlivinginthenorthofItalyformanyyears,theywerebornandspentsomeyearsintheSouthofItaly.

Thehearinggroupwascomposedoffourparticipants:

Table 2.Hearingparticipants.

SUBJECTS AGEVC 26LM 24ST 24LT 26

They were selected on the basis of age and above all of schooleducation(thelengthofschoolingwasatleast13years).Allofthemlive inPreganziol (Treviso) inNorthernItaly.Somehearingpartici-pantshabituallyusethedialectspokeninthisareaathomeandwiththeirfriends.

5.2. Procedure

Each participant was assessed individually and sentences wereadministeredindifferentsessionsofabout30minuteseach.Alltaskswerepresentedvisually,onseparatestripsofpaper,inordertoavoiderrorsderivingfromincorrectlip-reading.6

5.3. Materials

The experiment can be divided into two distinct parts. Theformerpartconsistsofsomepre-experimentaltasks–namelydelayedrepetition and grammar comprehension tasks– and aims at deter-miningwhethertheparticipantscouldhavedifficultiesinprocessinglongsentencesandperformingcomplextasks,aswellasinassigningθ-roles to sentence constituents. The latter part includes the maintasksandisconcernedwiththeoccurrenceofthirdpersonaccusative


319

cliticsinleftdislocationsentences,andwiththecorrectagreementingenderandnumberbetweencliticsandtheirantecedentsaswellasbetweencliticsandpastparticiples.

Thepre-experimentalpartwascomposedofthefollowingtasks:• Delayed repetition task, consisting in the repetition of completelyunrelatedseriesofwords(10a)andrepetitionofsentences(10b).Thetask involving words included series from two, three, four and sixstimuli. The task involving sentence repetition included items pre-senting typical word order (Subject-Verb-Object), passive sentences,sentenceswithunaccusativeverbsselectingdativecomplementsandcliticleftdislocationsentences.

(10) a. SPECCHIO LATO SASSI

Mirror Side Stones

b. ILCANEMORDEL’UOMO

Thedogbitestheman

• A picture matching task, which is included in the B.A.D.A. test(Miceli etal. 1994). This is a visual grammar comprehension test,investigatingthecomprehensionofdifferentkindsofsentences,andtesting morphological comprehension (e.g.Gli uomini sono inseguiti dalle donne‘themenarechasedbythewomen’)andreversibility(e.g.La bambina accarezza la nonna ‘thechild:fcaressesthegrandmoth-er’)inactiveandpassivestructures.

Themainpartoftheexperimentwascomposedofthefollowingtasks:•Anelicitedproductiontask(sentencecompletiontask–SC),whichconsistedof sentences thathad tobe filled inwith thecorrect thirdperson accusative clitic and the correct form of the verb given inbrackets(113gapsintotal).Somesentencesrequiredverbsinsimpletenses (presentand imperfect)andsomeothers in compound tenses(passatoprossimo).7Somesentenceswerepresentedwiththealreadyinsertedclitic,agreeingingenderandnumberwiththeantecedent,inordertodeterminewhethertheparticipantscouldbesensitivetothepresenceofthisfunctionalelementandwhetherthecliticcouldthenhelptotriggerthepastparticipleagreementincompoundtensesen-tences.Theitemstobecompletedwithboththecliticandtheagree-ingverbinvolvedsentenceselicitingthemasculinesingularclitic lo, sentenceselicitingthe femininesingularclitic la, sentenceseliciting

Francesca Volpato

320

themasculinepluralcliticliand sentenceselicitingthefeminineplu-ralcliticle.

Thetaskincludedsentencessuchasthefollowingexamples:

(11) a.Tuetuofratello,laluce______(accendere),perchélastanzaeraalbuio.

Youandyourbrother, the light:f.sg ____ (to turnon),because theroomwasdark.

‘Youandyourbrother(turnedon)thelight,becausetheroomwasdark.’

b.Illadro,ipoliziotti__________(arrestare)ierisera. Thethief:m.sg,thepolicemen:m.pl_________(toarrest)lastnight. ‘Thepolicemenarrestedthethieflastnight.’ c. Icarabinieri,illadro,lo______(arrestare). thepolicemen:m.pl,thethief:m.sg,him________(toarrest). ‘Thepolicemenarrestedthethief.’

• A grammaticality judgement task (GJ), which involved givinga judgement of correctness or incorrectness to a series of items.Participantswerepresentedwith80itemscontainingagainthefourdifferent accusative clitic pronouns, the masculine singular clitic lo,the femininesingular clitic la, themasculineplural clitic liand thefemininepluralclitic le.Theungrammaticalconstructionscontainedagreementerrorsbetweenthesubjectandtheauxiliaryverb,betweenthe clitic pronoun and the object and, in the sentences with a com-poundtense,betweenthecliticpronounandthepastparticiple.Someexamplesareillustratedin(12):

(12) a.*L’alunno,ilibri,lihannochiusi,dopoaverfinitodistudiare. thestudent,the:m.plbooks:m.pl,them:Cl.m.haveclosed,afterhefinished

studying ‘Thestudenthasclosedthebooks,afterhefinishedstudying.’

b.Lamela,miofratello,l’hamangiata. The:f.sgapple:f.sg,mybrother,it:Cl.f.haseaten ‘Mybrotherhaseatentheapple.’


321

5.4. Results

In this section, an account of the errors made by the partici-pantswillbegivenanddiscussedseparately for each typeof task.Theanalysisofthedatacollectedinthevariousexperimentaltasksprovedthattheerrorsmadebythethreehearingimpairedpartici-pantsarequantitatively fewandconsequentlyparticularattentionwill be paid above all to the qualitative analysis of their perform-ance.

Data resulting from sentence completion and grammatical-ity judgement tasks were analysed by using two statistical tests.The Mann Whitney test was necessary to compare the performancebetweenthetwogroups,andtheWilcoxontestwasusedtodeterminewhether a significant difference could exist in the occurrence of thefourobjectcliticforms.Resultswerecalculatedseparatelyforthetwomaintasks.

5.4.1. Performance in the pre-experimental tasksWewillstartouranalysisbyexaminingtheresultsinthediffer-

enttasksofthepre-experimentalsession.Inthesentencerepetitiontask,allparticipantsperformedatceil-

ing(Table3):In the word repetition task and in the picture matching task,

thehearingimpairedandthehearingsubjectsshowedthesamepat-ternofperformance,withapercentageofcorrectresponsesbetween85%and87% in the former (Table4),andabove95% in the latter(Table5):

TheMannWhitneytestwascarriedouttoanalysedatainthesetasks, and results did not show any significant difference betweenthetwogroupsofparticipants:wordrepetitiontaskU=5p=.724andpicturematchingtaskU=3p=.262.Thelowperformanceinthewordrepetitiontaskwasaboveallduetotherepetitionofsomeseriesofsixwords,whichwassimilarlydifficultforbothgroups.

5.4.2. Performance in the sentence completion taskByconsideringthegeneralperformanceinthesentencecomple-

tiontask,namelythesentences inwhich justverbalagreementwasrequired and those in which both clitic insertion and verbal agree-mentwererequired,itwasfoundthattherewasnosignificantdiffer-encebetweenthetwogroups:U=4p=.476.

Francesca Volpato

322

Ta

ble

3.N

um

ber

and

perc

enta

geo

fco

rrec

tan

swer

sin

th

ese

nte

nce

rep

etit

ion

tas

k.

Ta

ble

4.N

um

ber

and

perc

enta

geo

fco

rrec

tan

swer

sin

th

ew

ord

repe

titi

ont

ask

Ta

ble

5.N

um

ber

and

perc

enta

geo

fco

rrec

tan

swer

sin

th

epi

ctu

rem

atch

ing

task

.


323

However, since the aim of this research in the production taskwas to investigate the correct agreement between the clitic and itsantecedent in both simple and compound tense sentences, and theagreement between the clitic and the past participle in compoundtense sentences, thepercentageof correct insertionof the cliticpro-nounsolelywascalculated.Onthewhole,aswecanseefromTable6,thehearinggroupexhibitedahigherpercentageofcorrectresponsesthanthehearingimpairedgroup(97.79%ofcorrectnessfortheformer–SD1.53and94.69%forthelatter–SD2.34):

However, theMannWhitneytestdidnotrevealanysignificantdifferencebetweenthetwogroups:U=1p=.074.

The analysis aimed at checking the correct agreement betweenthe clitic pronoun and its antecedent gives two different resultsdependingonwhetherasimpleoracompoundtensewasrequiredinthesentences.

In compound tense sentences, where the participants had toinsertboththecliticandthepastparticipialverb,theaveragelevelofaccuracyisabove90%forbothgroups,aswecanseefromthefollow-ingtable:

Even though the hearing group achieved higher scores thanthehearingimpairedgroup(HG:99.39%–SD1.22–HIG:93.45%– SD 6.27), results showed no significant difference between thetwogroups:U=2.5p=0.172.Theerrorsidentifiedareerrorsofsub-stitution of the correct morphology with the unmarked form, butthecliticalwayscorrectlyagreedwiththepastparticipleinallsen-tences.8

Theperformance in simple tense sentences showedaveryhighrate of success for both groups (98,09% – SD 0.82 for the hearinggroupand99,54%–SD0.79forthehearingimpairedone).9

Thistableshowsthatthedifferencebetweenthepercentagesofthetwogroups isquitesmall.Surprisingly, the impairedgroupper-formed almost at ceiling and obtained higher percentages of correctresponsesthanthehearingone.

5.4.3. Performance in the grammaticality judgement taskA similarly unexpected result occurred in the grammatical-

ity judgement task where the performance of the hearing impairedparticipantswassignificantlybetterthanthatof thehearinggroup:U=.500p=.048.Withinthehearingimpairedgroup,twoparticipantsperformedatceilinginthistask,whilethethirdmadeonlytwoerrorswithpluralforms.

Francesca Volpato

324

Ta

ble

6.N

um

ber

and

perc

enta

geo

fco

rrec

tcl

itic

pla

cem

ent

int

he

sen

ten

cec

ompl

etio

nt

ask

.

Ta

ble

7.

Per

cen

tage

of

corr

ect

agre

emen

tbe

twee

nt

he

clit

ics

and

thei

ran

tece

den

tsi

nt

he

com

pou

nd

ten

ses

ente

nce

sin

th

ese

nte

nce

com

plet

ion

tas

k.


325

Ta

ble

8.

Per

cen

tage

of

corr

ect

agre

emen

tbe

twee

nt

he

clit

ics

and

thei

ran

tece

den

tsi

ns

impl

ete

nse

sen

ten

ces

int

he

sen

-te

nce

com

plet

ion

tas

k.

Ta

ble

9.P

erce

nta

geo

fco

rrec

tan

swer

sin

th

egr

amm

atic

alit

yju

dgem

ent

task

.

Francesca Volpato

326

The lower performance of the hearing group compared withthe hearing impaired one is mainly concerned with the incorrectjudgement of those items containing a plural subject and the thirdperson singular form of the auxiliary.10 Some hearing participantshave judgedascorrectasentence like (13),whereamismatcherrorbetweenthepluralsubjectandthesingularauxiliaryispresent:

(13) *Igiardinieri,glialberi,lihapotati. Thegardeners,thetrees,them:Clhaspruned ‘Thegardenershaveprunedthetrees.’

Thisphenomenonisprobablyrelatedtothefactthatthesehear-ingparticipantsuse theVenetandialectas their first languageandare influenced by it when speaking Italian. In the dialect, the verbhas the same form in the singular and the plural, and this formresemblesthesingularformofItalian:

(14) Igiardinieri,ialberi,(i)ligapotai. Thegardeners,thetrees,(Subj.Cl)them:Clhaspruned

ThecorrectItaliansentenceis:

(15) Igiardinieri,glialberi,lihannopotati. Thegardeners,thetrees,them:Clhavepruned

5.4.4. Performance in the different combinations of gender and number features

A non-parametric related sample test (Wilcoxon) was carriedout in order to compare the use of gender and number features ofclitic pronouns in the sentence completion and in the grammatical-ity judgement tasks, to determine whether participants could bemoresensitivetothemanipulationofsentencescontainingaspecificgenderand/ornumberfeature.Thefollowingcomparisonswerethuscarriedout,firstconsideringtheoppositionbetweensingleclitics(16)andthentheoppositionbetweenpairsofclitics(17)separatelyinthetwodifferenttasks:

(16) • masculinesingularvs.femininesingular(lovs.la) • femininesingularvs.feminineplural(lavs.le) • masculinepluralvs.feminineplural(livs.le) • masculinesingularvs.masculineplural(lovs.li),


327

(17) • masculinevs.feminine(lo+livs.la+le) • singularvs.plural(la+lovs.li+le)

Figure 1 shows the percentage of correct answers of the wholegroupforeachcliticpronounandseparatelyforeachofthetwomaintasks(sentencecompletiontaskandgrammaticalityjudgementtask).Intheelicitedproductiontask,pluralcliticsliandlehavethehigh-estrateofaccuracy.Thesituation is inverted inthegrammaticalityjudgement task, where items with singular forms lo and la show ahigherrateofsuccesswithrespecttothepluralonesliandle.

Figure 1.Percentageofcorrectresponsespercliticpronounforeachofthetwomaintasksconsideringallparticipants.

In the grammaticality judgement task, the greatest number oferrorswithpluralcliticsconcernsthemismatchbetweenthesubjectand the auxiliary (see section 5.4.3. example (13)). For this reason,onlyresultsonthesentencecompletiontaskwillbeconsideredfromhereon.

By examining the accuracy of single clitics within the wholegroup (hearing and hearing impaired groups) using the Wilcoxontest, a significant difference is attested in the elicited production ofaccusativecliticsregardingtheoppositionbetweenfemininesingular(la)andfeminineplural(le):Z=-2.032p=0.042.Inthegrammaticalityjudgementtask,nosignificantdifferenceisnoticed.

Francesca Volpato

328

Figure 2.Numberandgendercliticoppositioninthetwomaintaskswithinthe hearing-impairedgroup.

Inthesentencecompletiontask,inthoseitemsrequiringacom-poundtense,theoppositionbetweenpairsofcliticswasalsoanalyzedtodeterminewhethersomefeatures(genderand/ornumber)couldbeperformedbetterthanothers.Byconsideringthewholegroupofpar-ticipants,nosignificantdifferencewasattestedinthediscriminationbetween singular and plural and between masculine and femininefeatures.Asfarasnumberfeaturesareconcerned,thelackofsignifi-canceisduetothefactthatthehearinggroupachievedhigherscoresonsingularthanonpluralfeatures,whilethehearingimpairedgroupshowedthereversedpattern(seefigure3):

Figure 3.Oppositionbetweensingularandpluralfeaturesineachgroupofparticipants.

Byconsideringonlythehearingimpairedgroup,asignificantdif-ferenceisattestedinthediscriminationbetweenpluralandsingular(li-le>la-lo):Z=–2.032p=0.042.

To summarize all results, a between-group analysis revealedthat in the sentence completion task, no significance is attestedbetweenthetwogroupsofparticipants,while inthegrammaticality


329

judgement task the hearing impaired group performed significantlybetterthanthecontrolgroup.Interestingfindingshaveemergedfroma within-group analysis, in which the performance in the differentcombinationsofgenderandnumberfeaturesinthesentencecomple-tiontaskwasconsidered.Contrarytowhatonecouldexpect,resultsdemonstrate a clear predominance of correctness rate in sentencesrequiringpluralfeatures,andsometimesspecificallyfemininepluralfeatures.

Before looking at the linguistic theories that could help usexplainsuchbehaviour,letusexaminemorecloselythetypeoferrorsthatthetwogroupsofparticipantsmade.

5.5. The relevant errors

5.5.1. The hearing impaired groupThe performance of the three hearing impaired participants

showedanoverallhighlevelofaccuracyintheresponsestheyprovid-edandthenumberoferrorsmadebythemisverylow,provingthatimpairmentisveryselective.Sinceerrorsarequantitativelyfew,themostinterestingobservationsandgeneralizationscanbedrawnfromtheanalysisofthedataobtainedinthesentencecompletiontaskandthis section attempts to give an explanation for the non standardformsproducedbythesubjects.

Theerrorsproducedbyparticipantscanbedividedintotwomaintypologies:

• omissionofthecliticpronounanduseoftheunmarkedformofthepastparticiple11

• substitution of the correct gender and number morphologywiththeunmarkedone,i.e.masculinesingular12

The hypothesis explaining the first type of errors would be theretrievalofapost-verbalnullobjectpronoun(pro) insteadofacliticpronoun.Theexistenceof thisnullobjectpronounhasalreadybeenprovenforItalianbyRizzi(1986)insentenceslike(18),where‘pro’islicensedbyVandisassignedanarbitraryinterpretation:

(18) Unmedicoseriovisitaproinudii

Aseriousdoctorvisitsproiunclothed:m.pli

Thenullobject retrievalhypothesishasalsobeenproposedbyTuller(2000)forFrenchhearingimpairedindividuals.Althoughthepresenceofthisnullobjectpronounisillicitinthecorpusproducedby the subjects she tested, the French spoken language does not

Francesca Volpato

330

excludetheoccurrenceofthiselementinsomespecificcontexts(forexample, inheraccount,whencorrespondingtosufficientlysalienttopics).

IntheCLLDsentencesadministeredinthesentencecompletiontask,theparticipantsomitthecliticbecausetheypresumablyretrievetheweakpronominalelementinthecanonicalobjectposition,whichis not correct in the corpus they produce, but which is nonethelesslicit intheparticularstructureof Italianseen in (18).13 (In (19)andthroughout,theunderlinedwordsarethoseproducedbythepartici-pants):

(19) *La lezionei, il professore aveva spiegato proi due volte, perché iragazzinonavevanocapitoniente.

The:f.sg lesson:f.sg, the teacher had explained:m.sg.(default) pro twice,becausethestudentshadnotunderstoodanything.

‘Theteacherexplainedthelessontwice,becausethestudentsdidnotunderstandanything.’

ThesesentenceshavethesamestructureasthoseadministeredintheexperimentcarriedoutinVolpato(2002),wherethesamekindoferrorhasalsobeenmadebythetwohearingimpairedparticipantsFRandGR14.

Itshouldbenotedthattheunmarkedformofthepastparticiplein(19)(spiegato‘explained:m.sg’)excludesanotherpotentialanalysis,namelythatthecliticpronounissimplynon-overt.Ifacliticpronounwerepresentinthesyntacticderivation,wecouldexpecttheagreeingformofthepastparticiple(spiegata‘explained:f.sg’).ThiscorrespondswithCardinaletti’s(2002)conclusionsthatnullcliticpronounsdonotexist.

Turningourattentionnowtothesubstitutionerror,wefindthatitusuallyoccurredwhenacompoundtenseandthefemininesingularclitic were required. In this case, the hearing impaired participantsuse the default morphology on the past participle, as the followingexampleshows:

(20) *Lacenaperibambini,lamammal’hapreparatoprimadiandarealristorante.

The:f.sg.dinner:f.sg.forthechildren,the:f.sgmother:f.sgit:Clhasprepared:m.sg,beforegoingtotherestaurant.

‘The mother prepared dinner for the children, before going to therestaurant.’


331

This sentence is ungrammatical because the past participledoesnotcorrectlyagreeingenderwiththeleftdislocatedconstitu-ent (target: La cena (…) la mamma l’ha preparata ‘The:f.sg. din-ner:f.sg. (…), the:f.sgmother:f.sg it:Clhasprepared:f.sg’).Lookingonlyattheclitic,itisnotclearwhetheritisinthemasculineorthefeminine form, because with compound tenses, the gender featureisnot overtlymarkedon the cliticpronounand the elided form (l)is correct both for the masculine and the feminine singular (l’hapreparato/l’hapreparata).However,sinceitisevidentfromtheourearlieranalysis that the testedparticipantsalwaysmakethecliticcorrectlyagreewiththepastparticiple,Iassumethat inthiscase,thecliticandthepastparticiplealsosharethesamegenderfeature(masculine singularo). Hence it follows that the feminine singularformhasbeenreplacedbytheunmarkedone (masculinesingular).ItisworthpointingoutthatthissubstitutionoccursmainlywhenasingularfeminineNPispresent(itoccursonlyoncewiththemascu-lineplural).15

Theaspectthatiscommontobothtypesoferrors–omissionandsubstitution–isthatboththeobject‘pro’in(19)andtheoccurrenceoftheunspecifiedcliticlin(20)alwaystriggerthedefaultpastpartici-pleagreement(o).

5.5.2. The control groupBy turning our attention to the performance of the control

subjects, it is worth pointing out that the sentences they producedalso include some errors and non-standard forms. The type of erroroccurringintheirperformanceandwhichalsooccurs inthehearingimpairedgroupistheomissionofthecliticinaverysmallamountofitems.

Apart from the omission errors like those described above,there are non-standard forms that occur only in some sentencesproducedby twoof thehearingparticipants. In these cases,as farasgenderandnumberfeaturesareconcerned,theclitictheychooseisundoubtedlycorrect(lafemininesingular),butthefullformtheyuseisnotacceptableinthestandardlanguage,sinceonlytheelidedformofthisclitic isgrammatical.16Thiskindoferror is illustratedin(21):

(21) *Lavaligia,ilfacchinodell’hotellahaportataincamera. thesuitcase:f.sg.,theporterof-thehotelit:Cl.f hasbroughtintotheroom ‘Thehotelporterbroughtthesuitcaseintotheroom.’

Francesca Volpato

332

Inthestandardlanguage,thecorrectformofthissentencewouldbeasin(22),wherethegenderfeatureisnotovertlymarkedonthecliticpronoun:

(22) Lavaligia,ilfacchinodell’hotell’haportataincamera.

Whatisworthobservingisthatthiserroroccursmainlyinthoseparticipants who habitually speak the Venetan dialect during dailyinteractionswith familyand friends.17Since sentences like (21) arenotacceptedinStandardItalian,itisthuspossiblethatcontrolsub-jectsareusingaformofhypercorrectionthathasbeeninfluencedbytheVenetandialect,wherethefiniteformoftheauxiliaryverb‘have’hasadifferentphoneticandphonologicalcontentwithrespecttothatoftheItalianform.IntheVenetandialect,thethirdpersonindicativepresent auxiliary is realized as ga, thus forcing the presence of thenon-elidedfemininesingularcliticform:

(23) Avaisa,elfachin(...)laga portada...

ThefollowingparadigmsarefoundinItalianandinthedialect:

(24)

STANDARDITALIAN VENETANDIALECTloha/l’ha loga

l’ha lagaliha ligaleha ligaleha lega

Table10summarizesthedifferenttypesoferrorsmadebybothgroupsofparticipantsinthesentencecompletiontask:


333

Table 10.Typeoferrorsinthesentencecompletiontaskforbothgroups.Type of errors example group

Omission

Useofunmarkedp.p.morphology

*Lalezione,ilprofessoreaveva spiegatoduevolteThe.f.sg.lesson.f.sg.,theteacherhadexplained.m.sg.twice

Hearingimpairedgroup/

Hearinggroup

Substitution

Useofunmarkedmorphology

*Lacenaperibambini,lamammal’ha preparatoThe.f.sg.dinner.f.sg.forthechildren,the.f.sg. mother.f.sg.it.clhasprepared.m.sg.

Hearingimpairedgroup

Hypercorrection(influenceofVenetandialect)Useofthefullfeminineform

*Lavaligia,ilfacchinodell’hotella ha portataincameraThesuitcase.f.sg.theporterof-thehotelit.cl.f.hasbroughtintotheroom

Hearinggroup

Toaccountforthesubstitutionerrors,thenextsectionwilloffersomeexplanationsonhowgenderandnumberfeaturesarerepresent-edandprocessedinclausestructure.

6. Discussion: gender and number features on clitic pronouns

The results and the error typologies explained in the previ-ous sections reveal that the level of accuracy of the three hearingimpairedindividualsontheproposedstructuresisquitehigh.

Thecorrectproductionandplacementofmost clitics,mainly insimple tense sentences, is clear evidence that left dislocation itselfdoes not represent a problem for hearing impaired individuals and,despite someerrors,all thenodesof the syntactic treeare correctlyprojected in thegrammarof theseparticipants.Becauseofawidelyattestedpresenceofcliticpronounsinmostsentences,thereasonforomissionandsubstitutionerrorsmustbefoundelsewhere.

Asalreadyexplainedinsection4,CLLDinvolvesthepreposingofaconstituentthatislinkedtoitsargument/canonicalposition.TheleftdislocatedXP,pronouncedwithoutanystressandassociatedwithalreadyknown information, co-occurswitha coreferential clitic pro-nounwithwhichitsharesnumberandgenderfeatures.

Although limited innumber,mosterrorsproducedbythehear-ing impaired participants are found in compound tense sentences,where past participle agreement in gender and number with cliticpronouns is also required. The difficulties that the subjects encoun-teredwithgenderandnumbermarkersleadustodiscusstheerrorspresentinthecorpuswithintheframeworkofthecurrentresearchonϕ-features.

Francesca Volpato

334

What is most relevant to the aim of this discussion is that thedata collected during the experiment showed a significant differ-enceintheoccurrenceanduseofthefouraccusativecliticpronouns,provingabetterperformanceonthosemarkedwithplural (number)featuresas opposed to singular ones.This evidence on thedifferentnatureandbehaviourofcliticpronounsthuscontributestostrength-ensomesyntactic,morpho-phonologicalandpsycholinguisticpropos-alsthataccountforthetriggeringforceofnumberongender.Theseproposalswillformthetopicsfordiscussioninthenextsections.

6.1. The syntactic proposal

Onthebasisofthedataandtheresultsobtainedfromtheexperi-ment,itisclearthattheparticipants’performancevariesaccordingtospecific features displayed by clitic pronouns. We now discuss howthese features are represented syntactically. Harris (1991), Kayne(2000)andFerrari(2005)suggestthatthirdpersonaccusativecliticsarecomplexelementsthatcanbesplitintotwodistinctmorphemes:theconsonantalmorphemelwithoutspecificcharacterizationforgen-derand/ornumber,andtherelevantgenderandnumbermarker(l+o,l+a, l+i, l+e).Althoughthese featuresappear tobehave in thesameway,thereseemstobeanunderlyingasymmetryinthewaynumberandgenderarespecifiedonthemorphemel.

Much linguistic literature is devoted to the description of theway gender and number features are encoded and represented inthe syntactic tree. Ritter (1995) and Di Domenico (1997) postulatethat gender information is encoded differently from number infor-mationandconsequentlyitisrepresentedstructurallyinadifferentmanner.ForlanguagessuchasHebrew,Ritter(1995)proposesthatgenderisafeatureofN(25b),becauseitappearsonthenounstematall representational levels,while inRomance languages, genderispartofthenumberphrase(25a),asthefollowingrepresentationsshow:


335

(25)

DiDomenico(1997)offersaslightlydifferentstructuralanalysisfor Italian, by distinguishing between the representation of gram-matical gender and that of semantic gender. Grammatical genderispartof the lexical entryof thenoun (as in (25b)),while semanticgenderisprojectedtogetherwithnumberunderNumP(asin(25a)).Both authors assume that number projects its own syntactic head.Thenounphrasehasapluraldenotationonlywhenthenumbersuf-fixheadsadistinctfunctionalprojection.Asfaraspronounsarecon-cerned,sincetheylacktheNPprojection,genderisspecifiedonandprojectedwithnumber:

(26)

Insummary,numberisasyntactichead,separatelyrepresentedinthelexiconandautonomouslyprojectedinsyntax,whilegenderisprojected in syntax either with number (see (25a) and (26)) or withthenoun(see(25b)).

Francesca Volpato

336

ThesyntacticanalysisoftheItaliannominalsystemsuggestedby Ferrari (2005) strongly corroborates the structure proposed byRitter(1995),byconvincinglyarguingthatpluralfeaturesarereal-izedbyMergeofafurtherprojection(NumP)intotheclausestruc-ture. In Ferrari’s proposal, Number is projected only in the pluraland not with singular features. The assumption that the numberprojectionispresentonlyinthepluralhasstrongconsequencesforthe representation of gender features.However, this issuewillnotbe addressed in this paper but will be further developed in futurework.

6.2. The morpho-phonological proposal

Gender and number markers in third person accusative cliticsare suppressed under certain conditions. Let us examine the datainTable11concerningtheoccurrenceofcliticpronounswithsimpletenses:

Table 11.Formofcliticpronounwithsimpletenses.

LO LAloamo *l’amo laamo *l’amoloami *l’ami laami *l’amiloama *l’ama laama *l’amaloamiamo *l’amiamo laamiamo *l’amiamoloamate *l’amate laamate *l’amateloamano *l’amano laamano *l’amano

LI LEliamo *l’amo leamo *l’amoliami *l’ami leami *l’amiliama *l’ama leama *l’amaliamiamo *l’amiamo leamiamo *l’amiamoliamate *l’amate leamate *l’amateliamano *l’amano leamano *l’amano

The data on lo and la are from the Venetan variety of Italian.Theelidedformsofloandlaareungrammatical.OtherItalianvarie-tiesallowthatinsimpletenses,thesingularmasculineandfemininemarkerscanbesuppressedinfrontofverbsbeginningwithavowel.In other words, overt markedness is not obligatory with singularforms, and sometimes the morpheme l can appear instead of thebimorphemicloandla.Garrapa(2007)claimsthatdorsalword-finalvowelsarenotlikelytodropiftheyarenotrecoverablefromthecon-


337

text,asinla amo ‘Iloveher’, lo amo ‘Ilovehim’,althoughtheelidedformmightalsobeacceptable.Incliticsspecifiedforpluralfeatures,coronalvowels/i/and/e/areinsteadalwayspreserved.18

Things are slightly different with compound tenses, in whichbothanauxiliaryandthepastparticipialverbappear,andthereforegender and number features are marked both on the clitic pronounandonthepastparticiplemorphology.Theoccurrenceofobjectcliticsisillustratedinthefollowingtable:

Table 12.Formofcliticpronounswithcompoundtenses.

LO LA*lohovisto l’hovisto *lahovista l’hovistalohaivisto l’haivisto *lahaivista l’haivistalohavisto l’havisto *lahavista l’havistaloabbiamovisto l’abbiamovisto *laabbiamovista l’abbiamovistaloavetevisto l’avetevisto *laavetevista l’avetevistalohannovisto l’hannovisto *lahannovista l’hannovista

LI LElihovisti *l’hovisti lehoviste *l’hovistelihaivisti *l’haivisti lehaiviste *l’haivistelihavisti *l’havisti lehaviste *l’havisteliabbiamovisti *l’abbiamovisti leabbiamoviste *l’abbiamovisteliavetevisti *l’avetevisti leaveteviste *l’avetevistelihannovisti *l’hannovisti lehannoviste *l’hannoviste

In compound tenses, accusative clitic pronouns behave differ-entlyfromeachotheraccordingtothefeaturestheypossess.Table12shows that incliticsmarkedas [+plural],ϕ-featuresarealwaysvis-ibleonthemorphemel,whilesingularcliticsundergovoweldeletion.Inparticular,cliticsspecifiedfor[feminine]gender,areungrammati-cal inthenon-elidedform,whilemasculinearegrammaticalbothintheelidedandinthenon-elidedform.

Onthebasisofdatacollectedfrominformantsspeakinganorth-ernvarietyofStandardItalian(spokeninLecco),Garrapa(2007)pro-posestheVowelDeletionScalepresentedin(27):

(27) a>o>i>e19

This implicational scale is applicable to accusative clitic pro-nouns both in simple and compound tenses. In order to try to gen-

Francesca Volpato

338

eralize on thisphenomenon, it ispossible to infer thatplural cliticsalways appear with gender and number features overtly markedon the morpheme l, while for singular clitics the occurrence variesaccordingtodifferentfactors.MorphemesthatrealizetheNumheadcannotbeelided.Thisthusseemstorepresentthetriggeringelement.Theaccurateperformancewithpluralcliticsthatwehaveobservedinthehearingimpairedindividualsseemstobeduetothismorpho-pho-nologicalvisibility.

6.3. The psycholinguistic proposal

The syntactic and the morpho-phonological analyses, whichaccountforthespecificationofgenderandnumbermarkersoncliticpronouns, have strong consequences for the development of a psy-cholinguisticproposalrelatedtotheprocessingofthesefeatures.

As we have seen in 6.1., the presence of number specifica-tion [+plural] involves the activation of a further syntactic projec-tion (NumP) with respect to singular markers (Ferrari 2005). It isworth remembering that economy principles tend to lead speakersto perform the least costly operations, thus forcing them to choosethe structure containing unmarked forms, not involving the NumPprojection.20Althoughthechoiceoftheunmarkedformisindeedthestrategypreferredbythehearingimpairedindividualsinsubstitutionerrors, it is evident that in thedifferent combinationsof ϕ-features,theparticipantsnonethelessshowedahigherlevelofaccuracyonplu-ralratherthanonsingularforms.

According to the approaches developed by Ritter (1995) and DiDomenico (1997), the different linguistic representation of the twofeaturesisrelevantbecause,sincegenderisnotaheadwithitsownfunctional projection, it is not visible to the syntactic parser. Thisphenomenon has strong consequences on the interpretation of ele-ments containing both features. From a psycholinguistic point ofview,byassumingamodular theoryof languageprocessing,genderinformation is used by the parser later than number informationsince, according to Kimball (1973) and Frazier (1985), the syntacticprocessorusesonlytheinformationrelevanttothesyntacticlevelofanalysis, inordertoassignastructuralrepresentationtothe input.Therefore, at this initial processing stage only those features thatprojectautonomouslyinsyntaxandthataremorevisiblefromamor-pho-phonologicalpointofviewwillbeeasilyrecognizedandcomputedbytheparser.Hence,sinceonlynumberfeaturesheadtheirownsyn-tacticprojection,onlynumberinformationwouldbeusedtoestablish


339

a co-reference between the noun and the pronoun. Gender informa-tion,althoughovertlymarkedonthenoun,shouldnotbeavailableatthesameextentasnumberinformation.21

Thevalidityofthisapproachhasbeenconfirmedbythefindingsof an experiment runbyDeVincenzi&DiDomenico (1999) explor-ingtheprocessingofgenderandnumberfeaturesintheretrievalofpronounantecedentsinItalian.Thisexperimentprovesthatnumberfeaturesconstitutethemorphologicalinformationthatisfirstlyusedtoselectapronounantecedent.Incontrast,genderinformationdoesnothave the same reactivation effect as thatderiving fromnumberinformation.

The conclusion drawn by De Vincenzi & Di Domenico (1999)is confirmed by the analysis of the data gathered from the presentexperiment on the three hearing impaired individuals. Indeed thisexperiment supports the hypothesis of the difference in the use ofgender and number information on clitic pronouns, clearly provingthatnumberinformationisretrievedearlierthangenderinformationand thus justifying the better performance in sentences containingcliticsspecifiedforplural featurescomparedwiththosespecifiedforsingularones.

7. Conclusion

In thispaperwehaveanalysed the competenceof thirdpersonaccusative clitic pronouns in CLLD by hearing impaired individu-als. The presence of high scores in the performance of the hearingimpaired group (sometimes higher than those of the control group)leadstotheconclusionthattheimpairmentisverylimitedandthatthelinguisticcompetenceisquiteintact.Theresultsarenonethelessusefultoshedlightonsomelinguisticaspectsthatconfirmpreviouslinguisticfindingsconcerningϕ-features.

Thecollecteddataconfirmthedifferentstatusofthefourthird-personaccusativecliticpronounsaccordingtothecombinationofgen-derandnumberfeatures.Whengenderfeaturesarenotvisibleonthepronounbecausethevoweliselided,asinthefemininesingular,gen-der features are not accessible. On the other hand, gender featuresare always accessible in the plural. This proves that it is easier toproducethoseformsthestructureofwhichiscomplexfromamorpho-syntacticpointofview,i.e.structuresspecifiedforpluralfeatures([i],[e]).Incontrast,cliticpronounsspecifiedforsingularfeatures([o],[a])aredifficultnotonlyfromamorphologicalandsyntacticpointofview

Francesca Volpato

340

butalsofromaphonologicalone,becausetheymaylackovertrealiza-tion.Furthermore, since these featuresdonotprojectautonomouslyinsyntax,theyarecomputedwithdifficultyandarehardlyvisibletothesyntacticprocessor.Hence,consistentlywithamodulartheoryoflanguage processing, individuals find it easier to produce the struc-turethatissyntacticallymorecomplex,withthegreatestnumberofcheckedvisiblefeatures,becausetheyhavemoreevidenceofit,thusconfirmingatriggeringforceforpluralnumberfeaturesasopposedtosingularones,andthusprovingthatmorphosyntacticandpsycholin-guisticperspectivesconvergeontothesameconclusions.

Address of the Author

Francesca Volpato, Università Ca’ Foscari Venezia, Dipartimento diScienzedelLinguaggio,Dorsoduro1075,30123Venezia

<[email protected]>

Notes

* I would like to thank Anna Cardinaletti for her help in the data analysisandforhercommentsontheearlierversionsofthispaper.IalsowishtothankElisaDiDomenico,FrancaFerrari,StavroulaStavrakaki,VirginiaVolterraandtwoanonymousreviewers for theirsuggestionsandhelpfulcomments.Aspecialthanks to the seven participants, for their availability and collaboration in thedata collection. I am also thankful to Iulia Zegrean and Nicholas Langley forreviewingmyEnglish.1 Although the individualsanalysed inChesi (2000,2006)andVolpato (2002)areall hearing impaired, it is worth considering that the population of deaf people isextremelyheterogeneousandthatmanyfactorsinfluencethedevelopmentoflanguagebytheseindividuals(ageatonsetofhearingloss,degreeofhearingloss,useofhear-ingaids,durationofuseofhearingaids,etc.).Theheterogeneitywithinthispopula-tionmakesitdifficulttocarryoutastraightforwardcomparisonbetweenthesamplesinvolvedinthesestudies.Forthisreason,theexplanationforsuchvariabilityandforthedifferenceinperformancebetweenthegroupsisnotaddressedinthispaper.2 Actually,childrenwholearnasecondlanguagesometimestendtotransfertheparametric settingsof the first language to thegrammarof the second language,givingrisetonon-target-likeperformance.Inthiscase, therefore, the influenceofoneofthetwounderlyinggrammaticalsystemsallowsauniqueanalysisofobjectpronounsforbothlanguagesthusallowingobjectcliticstooccupypositionswhere,inoneofthetwolanguages,onlylexicalDPsorstrongpronounsareusuallyplaced.3 Theaimofthispaperisnottogivesupporttooneortheotherhypothesisandforthisreason,themovementhypothesiswillonlybeadoptedforthesakeofthepresentation.4 In most recent theories (Chomsky 1995), the AgrO projection is no longerassumed, but the projection responsible for past participle agreement is stillpresentinclausestructure.


341

5 According to BIA (Bureau International d’Audiophonologie), four categoriesofhearinglossarerecognized:mild(26to40dB),moderate(41to70dB),severe(71to90dB)andprofound(greaterthan91dB).Inthehearingimpairedgroup,aparticipantwithmoderatehearingimpairment(S1)wasalsoincluded.However,it isworthconsideringthatS1,having70dBhearing loss, isat theupper limitof moderate impairment, very close to the severe one, and that someone withanaudiologicalhearing lossof70dB,withouthearingaids,mayfunctionallybeconsideredasprofoundlydeaf(Bakeretal.2005).Althoughmoststudiesfocusonsevere-to-profounddeafness,somerecentresearchmainlyconcernedwithmild-to-moderatehearing lossprovedthatevenlowdegreesofhearing impairmentmaybe crucialandprevent individuals fromdevelopinga complete linguistic compe-tence(Delage&Tuller2007).6 Although the hearing-impaired participants were tested through the writtenmodality,theproposedtasksaimatinvestigatingthemoregenerallinguisticcom-petenceoftheItalianlanguage,andnotspecificallythecompetenceinthewrittenlanguage.7 Althoughinsomesentencestheproductionofacompoundtensewasexpected,theproductionofadifferenttensewasalsoconsideredcorrect.8 Foramoredetailedaccountofthistypeoferror,seesection5.5.1.9 Someerrorsareduetotheuseofanincorrecttenseinthemainclause.Sincetheaimofthisstudyismainlytheanalysisofthecorrectinsertionandagreementof the clitic pronounwith its antecedentandwith thepastparticiple, the tenseerrors(consecutio temporum)arenotconsideredandaccountedforinthispaper.Iincludedtheresultshereforcompleteness.10 Sincethepresentanalysisfocusesontheuseofgenderandnumbermorpholo-gy,thepersonmismatchbetweenthesubjectandthefiniteverbisnotconsidered,butresultshavehoweverbeenincludedforanoverallaccountoftheperformance.11 Omission errors are rarely attested in simple tense sentences, the omissionrateishigherinthoserequiringacompoundtense.12 FollowingPretto(1985),Iassumethat,inItalian,masculineistheunmarkeddefaultgenderandfeminineisthemarkedgender.13 Ananonymousreviewerhaspointedoutthattheseleft-dislocationsentencesmighthavebeenmentallyrehearsedbytheparticipantasassociatedwithacon-trastivefocusintonation:(i) la leZione,ilprofessoreavevaspiegatoduevolte The leCTure,theprofessorhadexplainedtwiceInthisway,itwouldbeincorrecttoconsiderthemasungrammatical.Actually,sentenceslike(i)mightbeacceptableinparticularcontexts,suchascontras-tive focus.However, in ‘out-of-the-blue’ contexts the contrastive intonationhas tobeexcluded,andforthisreason,sentenceslike(19)havebeenconsideredasincorrect.14 Inthepresentexperimentalinvestigation,theomissionofthecliticpronounisalsoattestedinsomesentencesfilledinbythehearinggroup.15 Thesame tendencyhasbeen identifiedbyCiprianietal. (1993) in typically-developingchildrenatearlystagesoflanguageacquisition.Fromthe24thtothe31stmonthofage,whenchildrenbegintousecliticsproductively,theysubstitutesome formswithsomeothers in thepreverbalposition (rarely in thepostverbalposition).Themostcommonerrortypologiesarethoseconcerninggenderfeatures(loinsteadofla)andnumberfeature(lo insteadofli).16 Foramoredetailedmorpho-phonologicalaccountexplainingtheoccurrenceofthedifferentcliticforms,seesection6.2.17 The same error is also performed by one of the control subjects (CTRL2) inVolpato (2002).Thisparticipantaswell speaksavariety of theVenetandialectandusesithabitually.

Francesca Volpato

342

18 InsomelinguisticvarietiesofcentralandsouthernItaly,the‘reduced’formofthepluralcliticslincompoundtensessometimesoccurs(Garrapatoappearaandb).ThesubjectsofmyexperimentallspeakNorthernvarieties,inwhichthedataintables1and2hold.19 ‘>’meansmorefrequentlyelided.20 Theexpression‘economyprinciples’isintendedhereinwidesense,assimplic-ityofrepresentation,onthebasisofwhichpeopleareledtochoosethesimplestnumeration.21 To support thedifferentdegreeof salience thatgenderandnumber featureshave, the implicational relations to one another is predicted by the FeatureHierarchyproposedbyCarminati(2004):

(i) (Person)>Number>Gender

showingthatnumberplaysagreaterroleinthedisambiguationofpronounsthangender.Thisevidencethussupportsallproposalsthatassignthesefeaturesadif-ferentgrammaticalstatusinthecoresyntax.

Bibliographical References

ajello Roberto, Giovanna maroTTa, Laura maZZoni & FloridaniColai 2001. MorphosyntacticfragilityinthespokenandwrittenItalianofthedeaf.InfavaElisabetta(ed.).Clinical Linguistic.Vol.1.101-117.

Baker Anne, Beppie van den Bogaerde and Woll BenCie 2005. Methodsandprocedures in sign language acquisition studies. Sign Language & Linguistics8.7-60.

BelleTTi Adriana1999.Italian/Romanceclitics:Structureandderivation.InVanRiemsdijk1999.543-579.

BelleTTi Adriana2005.ExtendeddoublingandtheVPperiphery.Probus 17.1-35.

BeninCà Paola1988/2001.L’ordinedeglielementidellafraseelecostruzionimarcate. In renZi Lorenzo (ed.). Grande grammatica di consultazione.Bologna:IlMulino.Vol.1.115-225.

BorTolini Umberta, Barbara arfé, M. Cristina Caselli, Luisa degasperi,Patriciadeevy&LaurenceB.leonard 2006.Clinicalmarkers forspe-cific language impairment in Italian: the contribution of clitics andnon-wordrepetition.International Journal of Language Communication Disorders41.695-712.

CardinaleTTi Anna & Michal sTarke 1999. The typology of structural defi-ciency:Acasestudyofthethreeclassesofpronouns.InVanRiemsdijk1999.145-233.

CardinaleTTiAnna2002.Againstoptionalandzeroclitics. RightDislocationvs.Marginalization,Studia Linguistica56.29-57.

CarminaTi Maria Nella 2004. Processing reflexes of the Feature Hierarchy(Person > Number > Gender) and implications for linguistic theory.Lingua 115. 259-285.

CeCCheTToCarlo1999.AcomparativeanalysisofleftandrightdislocationinRomance.Studia Linguistica 53.40-67.


343

Chesi Cristiano 2000. Inferenze Strutturali. Analisi sull’uso degli elementi funzionali nelle produzioni verbali dei bambini sordi.Siena:UniversityofSiena.Graduationthesis.

ChesiCristiano2006.Il linguaggio verbale non-standard dei bambini sordi.Roma:EdizioniUniversitarieRomane.

Chomsky Noam 1993. A minimalist program for linguistic theory. MIT Occasional Papers in Linguistics 1.Cambridge,MA:MITPress.

ChomskyNoam1995.TheMinimalistProgram.Cambridge,MA:MITPress.CinqueGuglielmo1977.Themovementnatureof leftdislocation.Linguistic

Inquiry 8.397-412.Cinque Guglielmo 1990. Types of A’-dependencies. Cambridge, MA: MIT

Press.Cipriani Paola, Anna Maria Chilosi, Piero BoTTari, & Lucia pfanner 1993.

L’acquisizione della morfosintassi in Italiano: fasi e processi. Padova:Unipress.

delageHélène&LauriceTuller 2007.LanguageDevelopmentandMild-to-ModerateHearingLoss:DoesLanguageNormalizewithage?Journal of Speech, Language, and Hearing Research 50.1300-1313.

de villiers Jill G., Peter A. de villiers & Esme hoBan 1994. The centralproblem of functional categories in the English syntax of oral deafchildren. In Tager-flusBerg Helen (ed.). Constraints on Language Acquisition: Studies of Atypical Children. Hillsdale, NJ: Erlbaum. 9-47.

de vinCenZi Marica & Elisa di domeniCo 1999. A distinction among ϕ-fea-tures: theroleofgenderandnumber in theretrievalofpronounante-cedents.Rivista di Linguistica 11. 41-74.

di domeniCo Elisa 1997. Per una teoria del genere grammaticale. Padova:Unipress.

faBBreTTi Daniela 2006. La scrittura delle persone sorde. In Fabbretti &Tomasuolo.63-81.

faBBreTTi Daniela, Virginia volTerra & Clotilde ponTeCorvo 2006. Unaricerca sull’italiano scritto da persone sorde adulte. In Fabbretti &Tomasuolo2006.97-116.

faBBreTTi daniela, elena Tomasuolo (eds.) 2006. Scrittura e Sordità.Roma:CarocciEditore.

faBBreTTi Daniela, Virginia volTerra & Clotilde ponTeCorvo 1998. Writtenlanguage abilities in Deaf Italians. Journal of Deaf Studies and Deaf Education3. 231-244.

ferrariFranca2005.A Syntactic Analysis of the Nominal Systems of Italian and Luganda: How Nouns Can Be Formed in the Syntax. New York:NewYorkUniversity.PhDDissertation.

franChi Elisa1998.Nel mondo dei sordi. Un cammino verso il linguaggio. Il ruolo delle categorie funzionali e della logogenia.Venice:UniversitàCa’FoscariVenezia.Graduationthesis.

fraZier Lyn 1985. Modularity and the representational hypothesis. InStephen Berman, Jae-Wdong Choe & Joyce mCdonough (eds.).Proceedings of North Eastern Linguistic Society 15. Amherst, MA:GraduateLinguisticsStudentsAssociation.131-144.

Francesca Volpato

344

friedemannMarkA.&Talsiloni1997.AgrobjisnotAgrparticiple.The Linguistic Review14.69-96.

garrapa Luigia 2007. Vowel Elision in two varieties of spoken Italianis constrained by Morphology. Universität Konstanz: FachbereichSprachwissenschaft.ArbeitspapierNr.121.

garrapa Luigia. To appear (a). Vowel elision in spoken Florentine Italian.In ToCk Daniele & W. Leo WeTZels (eds.). Romance Languages and Linguistic Theory 2006.Amsterdam:JohnBenjamins.

garrapa Luigia. To appear(b). L’elisione vocalica ai confini di parola nelfiorentino parlato e l’interazione fra morfologia, fonologia, frequenzad’uso e variazione di registro.Actes du XXV CILPR (Innnsbruck 2007September3-8).

hamann Cornelia & Adriana BelleTTi 2006. Developmental patterns in theacquisition of complement clitic pronouns: Comparing different acqui-sition modes with an emphasis on French. Rivista di Grammatica Generativa 31. 39-78.

harris James W. 1991. The exponence of gender in Spanish. Linguistic Inquiry 22.27-62.

jakuBoWiCZ Celia, Lea nash, Catherine rigauT & Christophe-Loic gérard1998.DeterminersandcliticpronounsinFrench-speakingchildrenwithSLI.Language Acquisition7.113-160.

kayne Richard 1989. Facets of Romance past participle agreement. InBeninCà Paola (ed.), Dialect Variation and the Theory of Grammar.Dordrecht:Foris.85-103.

kayne Richard 1991. Romance clitics, verb movement and PRO. Linguistic Inquiry 22.647-686.

kayne Richard 2000. Person morphemes and reflexives in Italian, Frenchand related languages. In kayne Richard. Parameters and Universals.Oxford:OxfordUniversityPress.131-162.

kimBallJohnP.1973.Sevenprinciplesofsurfacestructureparsinginnatu-rallanguage.Cognition2. 15-47.

leonard B. Laurence, M. Cristina Caselli, Umberta BorTolini, Karla K.mCgregor&LetiziasaBBadini 1992.Morphologicaldeficits in childrenwithspecificlanguageimpairment:Thestatusoffeaturesintheunder-lyinggrammar.Language Acquisition2.151-179.

miCeli Gabriele, Alessandro laudanna, Cristina Burani & Rita Capasso1994.BADA. Batteria per l’Analisi dei Deficit Afasici.CEPSAG.Roma:UniversitàCattolicaSacroCuore.

pirvulesCu Mihaela 2006. The acquisition of object clitics in French L1:Spontaneousvs.elicitedproduction.InAdrianaBelleTTi,ElisaBennaTi,Cristiano Chesi, Elisa di domeniCo & Ida ferrari (eds.), Language Acquisition and Development. Proceedings of GALA2005. Cambridge:CambridgeScholarsPress.450-462.

preTTo Annamaria 1985. Il genere grammaticale. In Annalisa franChi de Bellis & savoia Leonardo (eds.). Sintassi e Morfologia della lin-gua Italiana d’uso: Teorie e applicazioni descrittive. Atti del XVIICongressoInternazionalediStudi,Urbino1983.Roma:Bulzoni.289-310.


345

riTTerElizabeth1995.Onthesyntacticcategoryofpronounsandagreement.Natural Language & Linguistic Theory 13.405-443.

riZZi Luigi 1986. Null objects in Italian and the Theory of pro. Linguistic Inquiry 17.501-58.

riZZiLuigi1997.Thefinestructureoftheleftperiphery.Inhaegeman Liliane(ed.).Elements of grammar.Dordrecht:KluwerAcademicPress.281-337.

sporTiChe Dominique 1989. Le Mouvement Syntaxique: Constraintes etParametres.Langages 95.35-80.

Torrego Esther 1992. Case and argument structure. Boston: University ofMassachussets.Ms.

Tuller Laurice 2000. Aspects de la morphosyntaxe du français des sourds.Recherches linguistiques de Vincennes 29.143-156.

Tuller Laurice & Celia jakuBoWiCZ 2004. Développement de la morphosyn-taxedufrançaischezdesenfantssourdsmoyens.Le Langage et l’homme39.191-207.

van riemsdijk HenkC.(ed.)1999.Clitics in the Languages of Europe.Berlin:Mouton-DeGruyter.

volpaToFrancesca2002.Clitic Pronouns and Verbal Agreement in Two Deaf Adults: an Experimental Investigation. Venice: Università Ca’ FoscariVenezia.Graduationthesis.

volTerra Virginia & Elizabeth BaTes 1989. Selective impairment of Italiangrammaticalmorphologyinthecongenitallydeaf:acasestudy.Cognitive Neuropsychology 6.273-308.

volTerraVirginia,OlgaCapirCi & MariaCristinaCaselli2001.Whatatypi-cal populations can reveal about language development: the contrastbetween deafness and Williams syndrome. Language and Cognitive Processes16.219-239.

WexlerKenneth,Anna gavarrò &Vincent Torrens2002.FeaturecheckingandobjectcliticomissioninchildCatalanandSpanish.InReinekeBok-Bennema, Bart holleBrandse, Brigitte kampers-manhe & Petrasleeman(eds.). Romance Languages and Linguistic Theory 2002. Amsterdam:JohnBenjamins.253-268.

Clitic pronouns and past participle agreement in … pronouns and past participle agreement in...

Documents

Transcript of Clitic pronouns and past participle agreement in … pronouns and past participle agreement in...