Bilingual first language acquisition at the interf

Bilingual®rstlanguageacquisitionattheinterfacebetweensyntaxandpragmatics*AAFKEHULKUniversityofAmsterdamNATASCHAMUÈLLERUniversityofHamburgThispaperhasasitsstartingpointtheassumptionthatinacquiringtwolanguagesfrombirth,bilingualchildrenseparatetheirgrammarsfromveryearlyon.Thisdoesnot,however,excludecross-linguisticin¯uence±thepossiblein¯uenceofonelanguageontheother.Themainfocusofthepaperisontheacquisitionofsyntaxinagenerativeframework.Wearguethatcross-linguisticin¯uencecanoccurif(1)aninterfacelevelbetweentwomodulesofgrammarisinvolved,and(2)thetwolanguagesoverlapatthesurfacelevel.Weshowthatbothconditionsholdforobjectdrop,butnotforrootin®nitives.Rootin®nitivessatisfythe®rstcondition:theyinvolvetheinterfacebetweensyntaxandpragmatics.However,theydonotsatisfythesecondcondition.Therefore,weexpectcross-linguisticin¯uencetooccuronlyinthedomainofobjectdropandnotinthedomainofrootin®nitives.ComparingthedevelopmentofthetwophenomenainabilingualDutch±FrenchandaGerman±Italianchildtothedevelopmentinmonolingualchildren,weshowthatthispredictionisborneoutbyourdata.Moreover,thiscon®rmsthehypothesisthatcross-linguisticin¯uenceisduetolanguageinternalfactorsandnottolanguageexternalfactorssuchaslanguagedominance:theperiodsduringwhichweobservein¯uenceinthedomainofobjectdropandnon-in¯uenceinthedomainofrootin®nitivesareidentical.Inthelastdecade,moststudiesofbilinguallanguageacquisitionhaveshownthatbilingualchildrenareabletodifferentiatebetweentheirtwolanguagesfromearlyon(Genesee,1989;Meisel,1989;DeHouwer,1990;MuÈller,1993;Gawlitzek-MaiwaldandTracy,1996;Hulk,1997).Thesestudiescriticizedanearlierviewofbilingualacquisition,namelythatchildrenwhoareexposedtotwolanguagesfrombirthnecessarilystartoutwithoneunitarylanguagesystem(VolterraandTaeschner,1978;Taeschner,1983,forexample).Thehypothesisofaninitialunitarylanguagesystemwasbasedontheoccurrenceofwordsorconstructionsofonelanguageintheother:itwasclaimedthatsuchmixedutterancesshowthatchildrendonotalwaysseparatetheirtwolanguages.However,Lanza(1992),andothers,showedthatformalaspectsoflanguagemixingbybilingualtwo-year-oldsdonotindicatethechild'slackoflanguagedifferentiation,butareasignofthewayinwhichthechildrendifferentiatetheirlanguageuseinacontextuallysensitiveway.Recently,theseparatelanguagehypothesishasbeenfurtherre®ned(Gawlitzek-MaiwaldandTracy,1996;Hulk,1997;DoÈpke,1998;MuÈller,1998;MuÈller,HulkandJakubowicz,1999).Thedevelop-mentoftwolanguagesinabilingualchildmaybelargelyautonomous,butthisdoesnotexcludethepossibilitythattherecanbein¯uencefromonelanguageontheother.Insuchcaseswewillspeakof``cross-linguisticin¯uence''.Cross-linguisticin¯uencethenisnottobetakenasmixingorfusion,butitcouldshowupasfacilitation/acceleration,delay,ortransfer(ParadisandGenesee,1995;GeneseeandParadis,1997).Instudyingbilinguallanguageacqui-sition,weobservetheemergenceofgrammarsoftwolanguagesatthemomentofcreation,whentheyareinclosecontactwitheachother.AssuggestedbyMacWhinney(1987),thebilingualchildmayattempttomakeshortcutsandallowstrategiesfromonelanguageintotheother.Plausibly,such``shortcuts''aretakenwhenthechildhastocopewithproble-maticinput.Theinterestingquestionisto®ndoutwhatthisproblematicinputis,i.e.whichpartsofthegrammararesensitivetosuchcross-linguisticin¯u-enceandwhythisshouldbeso.InthisarticleourstartingpointwillbethatAddressforcorrespondenceAafkeHulk,TaalkundeRomaanseTalenUVA,Spuistraat134,1012VBAmsterdamE-mail:[e-mail],UniversitaetHamburg,RomanischesSeminar,vonMelle-Park6,D20146HamburgE-mail:[e-mail]:LanguageandCognition3(3),2000,227±244#2000CambridgeUniversityPress227*WewouldliketothankFredGeneseeandtwoanonymousreviewersfortheirinvaluableinputtothispaper,andEricaThriftforcorrectingourEnglish.Anyremainingerrorsareours.bilingualchildrenseparatetheirgrammarsfromveryearlyon,butthatacquiringtwolanguagessimulta-neouslyisnotexactlythesameasacquiringeachlanguageseparately.Ourmainfocuswillbeontheacquisitionofsyntaxinagenerativeframework.Assumingthatwithbilingualchildrenaswithmono-lingualchildrenlanguageacquisitionisconstrainedbytheprinciplesofUniversalGrammar,weexplorethehypothesisthatcross-linguisticin¯uenceismorelikelytooccurinexactlythoseareaswhicharealsoproblematic±albeittoalesserextent±formono-lingualchildren.Recently,Platzack(1999)haspro-posedthatitmaybepossibletosingleoutaparticulardomainoftheclause,i.e.,theC-domain,causingproblemsindifferenttypesoflanguageacquisition.Asiswellknown,veryearly®rstlan-guage(L1)learners,childrenwithSpeci®cLanguageImpairment(SLI),adultsecondlanguage(L2)lear-ners,andpatientswithBroca'saphasiausenon-target-likesyntax.Withrespecttoclausalstructure,theselearnersshareanon-automatized,non-target-likesyntaxatthehigheststructurallevel,theC-domain.Grammaticalpropertieslikeverbsecond,complementizers,andtopicalizationaretypicalex-amplesofsyntacticphenomenarelatedtotheC-domain,theleft-peripheryofthesentence,whereaspre/postposition,theorderbetweenmainverbandobject,betweenobjectandadverbialsareaccountedforinotherdomains.Moreover,thereis