Article
Reference
Mild-to-moderate hearing loss and language impairment: how are they linked?
TULLER, Laurice, DELAGE, Hélène
TULLER, Laurice, DELAGE, Hélène. Mild-to-moderate hearing loss and language impairment:
how are they linked? Lingua , 2014, vol. 139, p. 80-101
DOI : 10.1016/j.lingua.2013.10.009
Available at:
http://archive-ouverte.unige.ch/unige:150358
Disclaimer: layout of this document may differ from the published version.
1 / 1
Mild-to-moderate hearing loss and language impairment:
How are they linked?
Laurice Tuller
a,* , Hélène Delage
baUniversitéFrançois-RabelaisdeTours,UMRInserm930,3ruedesTanneurs,37041ToursCedex01,France
bUniversitédeGenève,LaboratoiredePsycholinguistiqueExpérimentale,40boulevardduPontd’Arve,1205Genève, Switzerland
Received18October2011;receivedinrevisedform19September2013;accepted8October2013 Availableonline15November2013
Abstract
Whatisthenatureofthelinkbetweenprelingualmild-to-moderatehearingloss(MMHL)andimpairedlanguageinchildrenand adolescents?Althoughthescientificliteratureissparse,itisclearthatmanyexperienceconsiderabledifficultyacquiringlanguage,and thatthisdifficultyisnotlimitedtophoneticform.WereportonaseriesofstudieswehaveconductedinvolvinganumberofFrench- speakingchildrenandadolescents(N>80)aged6--16,allofwhomhavebilateralsensorineural,prelingualhearingloss.Usingavariety ofmethodologiestotestawiderangeoflanguageskills,wehavefoundthatmorphosyntacticdevelopmentinchildrenwithMMHLis highlylikelytobeimpaired,thatitisoftenveryseverelyimpaired,andthatdifficultiescontinueintoadolescence.Comparisonswithother contextsofatypicalacquisitionofFrench,bothwithpathology(SLI,epilepsy)andwithoutpathology(secondlanguageacquisition)show thataspectsofmorphosyntaxwhicharesubjecttodifficultyarenotspecifictothecontextofhearingloss.Althoughthereissomeevidence forcorrelationswithdegreeofhearingloss,thesearenotregularanddonotpredictmorphosyntacticperformance:thelinkbetween hearinglossandlanguageimpairmentisindirect.Itissuggestedthatthislinkmightbemediatedbyworkingmemoryandauditory attention,whichcouldthusbeexploredasaplausibleavenueforfindinganexplanationfortheheterogeneityinlanguageperformance observedinindividualswithMMHL.
©2013ElsevierB.V.Allrightsreserved.
Keywords:Mild-to-moderatehearingloss;Acquisition;Morphosyntax;French
1. Mild-to-moderatehearinglossasamodelofatypicallanguageacquisition
Mild-to-moderatehearingloss(MMHL)correspondstoanaveragehearingloss(asmeasuredbypuretoneaverage, PTA),intherangeof21--70decibels(21--40dBformild HL,41--70dBformoderateHL).Thisdegree ofcongenital, sensorineuralHLismuchmorefrequentinchildrenthansevereandprofoundHL.So,forexample,Russetal.(2003) foundthat75%ofagroupof134Australianchildrenfittedwithhearingaidsbetweentheageof0and6,hadmildor moderateHL(respectively42%and33%),comparedtosevereandprofoundHL(16%and9%).1Sincenotallchildren withmildHLusehearingaids,2thismeansthatthetotalproportionofchildrenwithmildHLisnecessarilybiggerthanthe www.elsevier.com/locate/lingua Availableonlineatwww.sciencedirect.com
ScienceDirect
Lingua139(2014)80--101
*Correspondingauthor.
E-mailaddress:tuller@univ-tours.fr(L.Tuller).
1SeealsoFortnumetal.(2002),whofound53%moderatehearinglossinagroupof17,160Britishchildrenages3--18withhearinglossover 40dB.
2InthecaseofHLlessthan30dB,ifthisHLisconsiderednottohavenegativeimpactonlanguagedevelopment,itispossiblethatnohearing aidsareproposedatall.
0024-3841/$--seefrontmatter©2013ElsevierB.V.Allrightsreserved.
http://dx.doi.org/10.1016/j.lingua.2013.10.009
figurereportedinthisstudy.Theadvent,technicalimprovement,andgeneralizationofcochlearimplantsinchildrenwith profound(90--120dB),andevensevere(70--90dB)HL,entailsthatthenumber,andtheproportionofchildrenfunctioning withthislevelofHLisnowevengreater.HearinggainsduetocochlearimplantationresultinPTAsof30--45dB(Verbist, 2010),andthustolevelsequivalenttomildormoderateHL.3Thequestionoflanguagedevelopmentinthecontextof MMHLthereforecurrentlytakesonnewsignificance,intermsofthenumberofchildreninvolved,butalsointermsofthe relevanceof thispopulationas acontrol groupinstudies evaluatingchildrenwithcochlearimplants. Understanding languagedevelopmentinthiscontextthushasbroaderimplicationsandclinicalapplications.
AnimportantfactaboutMMHListhatitisfrequentlydetectedrelativelylate:aroundtheageoffiveincountrieswhere neo-natalhearingscreeningisnotyetuniversal,asinFrance(Delage,2008;DelageandTuller,2007,2010;Tullerand Jakubowicz,2004).Furthermore,neonatalscreening(whichusesbrainstemERPs)onlydetectsHLhigherthan40dB (Govaertsetal.,2002),and thuscannotidentifychildrenwithmild HL.4Onceagain, acertainparallelcanbemade betweenlanguagedevelopmentinthecontextofacochlearimplantandinthecontextofMMHL:theageofHLdetectionin childrenwithMMHLisakintothevariableofageofcochlear implant.Inbothofthesecases,prior toaidedhearing (whetherbyhearingaids,inthecaseofMMHL,orbycochlearimplant),languageinputisdegraded.LatedetectionofHL inchildrenwithMMHLthusmeansthatmanyofthesechildrenexperienceseveralyearsofdegradedlanguageinput, years which fall squarely within the temporalwindow typically regarded to constitutea criticalperiod for language acquisition(BishopandMogford,1993;BortfeldandWhiteburst,2001;Newportetal.,2002).Studiesoftheclassical contextforcriticalperiodeffects,thatofsecondlanguageacquisition(seeJohnsonandNewport,1989,1991),have shownthatlanguageacquisitionaftertheageof6--7doesnotresemblenativeacquisitioninitsoutcome,particularlyin formalaspectsoflanguage(phonology,morphosyntax),thoughtheinterpretationoftheseresultsisnotunanimous(see Abrahamsson and Hyltenstam,2009,forreview). Studiesofchildrenwithcochlear implantshavealsofound strong correlationsbetweenlanguageperformanceandage ofimplant, highlightingtheexistence ofcriticalperiodsfor both maturationoflanguage(NicholasandGeers,2004;Tomblinetal.,2005)andmaturationofaudition(Sharmaetal.,2002, 2005).Theliteraturesupportingtheexistenceofacriticalperiodforlanguageacquisitionincludesstudiesofdeafchildren withearlyexposuretosignlanguagecomparedtothosehavinghadlateexposure.Thesehavealsoshownstrongcritical periodeffects,forbothfirstlanguagesignlanguageacquisitionandforsecondlanguageacquisitionofanorallanguage (Newport,1990;MayberryandLock,2003).ChildrenwithMMHLaregenerallynotexposedtosignlanguage,andtheydo display(some)spontaneousorallanguagedevelopment(doubtlessoneofthecausesforthefrequentlateHLdetectionin thispopulation).MMHLthusconstitutesadifferentconstellationforexplorationofpotentialcriticalperiodeffects:earlyoral languageinputisclearlymoreaccessiblethanitistoseverelyorprofoundlydeafchildrenwholearnsignlanguagelate and/orwhoundergocochlearimplantation.Ontheotherhand,latesigninputisentirelyaccessibletoadeafchild,whereas late aidedauditory inputthroughhearing aidsisnotentirelyaccessible (evenifconventionalhearing aidshavealso undergoneconsiderabletechnicalimprovements).Inotherwords,bothearlytimingoflanguagedevelopmentandlong- lasting quality of inputare alteredin MMHL. Whatis thenthe combined effect ofearly degraded inputfollowed by continuedlessthanoptimalinput(evenafterhearingaidfitting)onlanguagedevelopment?Thisconstellationsuggests thataninterestingcomparisongroupwouldbethecaseofsecondlanguageacquisition,theclassicalcontextforstudyof criticalperiodeffects,inwhichinputstartslate(similaritythenintimingoflanguageexposure),butisentirelyaccessible.
Another,obvious,butimportantfactaboutchildrenwithMMHListhat,allotherthingsbeingequal,theyhaveanormal languageacquisitiondevice, meaningthat,puttingasideforthe momentthequestion ofco-morbidity,anylanguage impairmentfoundinthispopulationwouldbea consequenceoffactorsexternaltoUniversalGrammarandthusthis populationconstitutesaparticularlyinterestingcomparisongroupforchildrenwithSpecificLanguageImpairment(SLI), forwhichproposalsaboutthesourceoflanguageimpairmentincludethesuggestionthatlanguagecompetenceisaltered inthesechildren(vanderLelyetal.,2011).Suchcomparativestudiesaresofarratherscarce,thoughresultsobtainedso farpointtoremarkablesimilarities.
Summarizing,languagedevelopmentinchildrenwithMMHLposesinterestingquestionsforthestudyoflanguage acquisition,andconstitutesaparticularlypertinentmodelforcomparativestudyoflanguageacquisitionindeafchildren withcochlearimplants,but,moregenerally,ofatypicallanguageacquisitioninchildren.
PreviousstudiesfocusedonlanguageperformanceinchildrenwithMMHLhaveshownsignificantratesoflanguage impairment, especiallyin theareas ofphonologyand morphosyntax, althoughall ofthesestudies havepointedout considerableinter-subjectvariability(seeforexampleGilbertsonandKamhi,1995;Briscoeetal.,2001;Norburyetal., 2001,2002;TullerandJakubowicz,2004;Hanssonetal.,2004,2007;DelageandTuller,2007).Table1listspublished studies which have presented results onoral language development in children and adolescents with mild and/or
3ChildrenwithMMHLarenotCIcandidatessinceconventionalhearingaidsresultinaidedhearingat/belowlevelobtainedbyCI,and,of course,destructionofa(partially)functioningcochlearaisesclearethicalquestions.
4Currenttechnologyentailsanunacceptablenumberoffalsepositivesiflevelsbelow40dBareused.
moderateHL(seealsoDelageand Tuller,2007for a detailedreview).5Itisimmediatelyapparenthowrarestudies focusingspecificallyonindividualswithMMHLare.Indeed,eveninthestudieslistedbelow,someincludedchildrenwith moresevereHL(e.g.Borgetal.,2007).Moreover,thenumberofparticipantsinmanyofthesestudiesissmall,andvery fewincludelongitudinaldata.Whenagerangeisaddedtothepicture,somecleargapsemerge:childrenwithMMHLhave beenstudied(Briscoeetal.,2001;Norburyetal.,2001,2002,etc.),andsohaveadolescents(DelageandTuller,2007), butnolongitudinalstudiesbridgingtheseagegroupshavebeenpublished,andthusimportantquestionsremainopen abouthowlanguagedevelopsduringtheperiodcoveringbothchildhoodandadolescence.Forexample,howdoesthe oftennoticedinter-subjectvariabilityforlanguageevolvewithage?Movingtothelanguagedomainsexploredinprevious workonlanguageinchildrenwithMMHL,Table1pointstothefactthatstudiestargetingspecificlinguisticvariablesare ratherlimited.Moststudieshaveusedgloballanguagemeasures,anditisthusknownthattherearelanguagedifficulties Table1
Publishedstudiesonorallanguagedevelopmentreportingresultswhichspecificallytargetchildrenwithmildand/ormoderateHL.
References Nchildren
(language)
Age HL Linguisticdomain(method) Otherchildren
Briscoeetal.(2001), Norburyetal.(2001,2002)
19(English) 5--10 20--70 Morphosyntax(sentencecomprehension, sentencerepetition,elicitedproductionofverbal inflection),phonology(non-wordrepetition, phonologicalawarenessanddiscrimination), vocabulary(receptive,expressive).
SLI
Borgetal.(2007)a 81(Swedish) 4--6 21--60 Morphosyntax(sentencecomprehension, sentenceproduction),phonology(phoneme discrimination,rhymematching,rhyme construction),vocabulary(receptive,expressive).
Ø
DelageandTuller(2007), Tulleretal.(2011)
19(French) 11--15 27--69 OralLanguageBattery(receptivevocabulary, sentencecompetition,sentencecomprehension, wordrepetition),morphosyntax(elicited productionofpronominalclitics).
SLI,Epilepsy
GilbertsonandKamhi(1995) 20(English) 7--10 20--65 OralLanguageBattery(articulation, comprehension,expressivelanguage), phonology(nonwordrepetition),vocabulary (expressiveandreceptive,novelwordlearning).
Ø
HallidayandBishop(2005) 22(English) 6--13 21--67 Morphosyntax(sentencecomprehension), phonology(non-wordrepetition),vocabulary (receptiveandexpressive).
Ø
HallidayandBishop(2006) 16(English) 8--14 24--70 Phonology(non-wordrepetition;wordnaming).
Hanssonetal.(2004) 18(Swedish) 9--13 30--57 Morphosyntax(sentencecomprehension), phonology(non-wordrepetition)vocabulary (expressive,novelwordlearning),.
SLI
Hanssonetal.(2007), SahlénandHansson(2006)
12(Swedish) 5--8 30--69 Morphosyntax(sentencecomprehension), phonology(non-wordrepetition)vocabulary (expressive,novelwordlearning),.
SLI
Kiese-HimmelandReeh(2006) 16(German) 2--4 22--66 ExpressiveVocabulary.Longitudinalstudy. Ø McGuckianandHenry(2007) 10(English) 6--8 41--70 Morphosyntax(elicitedandspontaneous
productionofgrammaticalmorphemes)
Ø Moelleretal.(2010) 4(English) 2--7 26--54 Orallanguagebattery(expressiveandreceptive),
vocabulary(expressive),speechproduction (articulationtest,spontaneousproductiondata), morphologyandsyntax(spontaneousproduction data).Longitudinalstudy.
Ø
TullerandJakubowicz(2004) 20(French) 6--13 36--65 Morphosyntax(elicitedproductionofpronominal cliticsandverbalinflection)
SLI Wakeetal.(2006) 55(English) 7--11 16--40 Orallanguagebattery,phonology(non-word
repetition,phonemeawarenessand discrimination).
Ø
a ThisstudyalsoincludesthefollowinggroupsofchildrenwithsensorineuralHL:unilateral,0--20dB(minimalHL),and61--80dB(amixof moderateandsevereHL),aswellasgroupsofchildrenwithconductiveHL.
5Thereareseveralstudieswhichincludechildrenwithmildand/ormoderateHLinlargergroupsofchildrenwithalldegreesofHL(see,for example,Blameyetal.,2001;Davisetal.,1986;Stelmachowiczetal.,2004).
in this population, but little is known about their nature and in particular about the common denominator behind morphosyntacticpropertiesthat aredifficultforchildrenandadolescents withMMHL.Finally,exploringthe natureof impairmententailscomparingthelanguagedifficultiesinthispopulationwiththoseobservedinotherstudiedpopulations, andinparticularinSLI.AsTable1shows,comparativestudiesarefew(and,inthedomainofmorphosyntax,theyare restrictedtocomparisonswithSLIandepilepsyinourgroup(TullerandJakubowicz,2004;DelageandTuller,2007;Tuller etal.,2011)andwithSLIbyBishop,Norburyandcolleagues(Briscoeetal.,2001;Norburyetal.,2001,2002).Theexisting MMHL-SLIcomparativestudieshavefoundcomparable(low)performanceonsomelanguagemeasures(e.g.non-word repetition),andquantitativeinter-groupdifferences,withtheMMHLgroupoutscoringtheSLIgroup,forotherareas(e.g., morphosyntactic measures). However, each of these studies has underlined the high variability in language skills observedinchildrenwithMMHL,withsomechildrenhavinglanguageimpairmentassevereasthatfoundinchildrenwith SLI,andothersnolanguageimpairmentatall.Anothertypeofcomparativestudy,toourknowledge,issofarentirely absentfromtheliteratureonMMHL,despiteitspotentialrelevance:thecaseoflatelanguageexposureintheabsenceof impairedcapacityforlanguage,whichisfoundinsecondlanguageacquisition.
The general goal of this paper is to present data addressing these (little studied) questions about language developmentinchildrenandadolescentswithMMHL.WewilldrawondifferentstudieswehaveconductedonFrench- speakingchildrenand adolescentswithMMHL(altogether,theseinclude 82participants aged6--16, withprelingual, bilateral,sensorineuralHL,andwithPTAsrangingfrom27to69dB).Weproposetopresentherebothsummariesof resultsstemmingfromearlierstudies(DelageandTuller,2007;Scheidnesetal.,2009;TullerandJakubowicz,2004;
Tulleretal.,2011),andnewresults,includingresultsofanalysisofspontaneouslanguagedataandresultsoflongitudinal dataina(different)groupof296-to11-year-oldchildrenwithMMHL.Giventhelimitednumberofstudiesoflanguage developmentinchildrenwithMMHL,empiricalcoverageforconclusionsisanimportantconcern.Ourspecificgoalisto contributetobetterunderstandingoftheprevalence,thenatureandthespecificityofmorphosyntacticimpairmentinthis population,and,inparticular,oftheproblemofthecommonlyfoundhighinter-subjectvariability.Toachievethesegoals, wehaveevaluatedspecificaspectsofFrenchwhichwebelieveareparticularlyindicativeoflanguageimpairment.A growing body of literature has presented results which support that hypothesis that similarities between language propertieswhichtakelongertoemergeinyoungtypicallydeveloping(TD) childrenandthosewhichcauseparticular difficulties inchildrendevelopinglanguageinunusualcircumstances (suchas SLI)are duetothe complexityofthe linguisticcomputationunderlyingtheseproperties(Jakubowicz,2005,2011;JakubowiczandTuller,2008;Prévostetal., 2010; Jakubowiczand Strik,2008; Soares,2006). Inthesestudies,complexityhasbeenmeasuredbyreference to linguisticoperations:numberofinstancesofsyntacticmovementoperations,asoriginallyproposedbyJakubowicz),orof agreementoperations,morphologicaland phonologicalclustering,and,moregenerally,level ofsyntacticembedding (Hamannetal.,2007).Wewillpursuethislineofinvestigationheretoexplorethenatureofmorphosyntacticdifficulties experiencedbychildrenwithMMHL.
Section2,afterpresentinggeneralmeasuresofmorphosyntaxderivedfromstandardizedlanguageassessmenttools, focusesondifferentmeasuresofcomputationallycomplexaspectsofmorphosyntaxinFrench:productionofaccusative clitics, whichhasbeenargued to bea clinicalmarker oflanguage impairmentinFrench, and thenperformanceon syntacticstructuresentailingdifferentdegreesoflinguisticcomputationalcomplexity(passives,relatives,andsyntactic embeddingingeneral).Inkeepingwiththepointsoutlinedabove,wewillbecomparinglanguagedevelopmentinchildren withMMHLtoTDchildrenandadolescents,butalsotoindividualsdevelopinglanguageinotheratypicalsituations,with pathology(SLIandsevere-to-profoundHL)andwithout(secondlanguageacquisition).
Section3addressestheinter-subjectvariabilityfoundinthestudiespresentedinSection2,whichmirrorinthisrespect otherstudiesoflanguageinchildrenwithMMHL.WefocusonthevariablesofdegreeofHLandage.Agebringsupthe questionoftheevolutionoflanguageperformanceovertime,whichweaddressviaresultsofalongitudinalstudy.Wewill alsobrieflymentionthevariablesofnonverballevel,socioeconomicstatus,ageofHLdetectionandageoffittingwith hearingaids.Regardingthelattertwovariables,itisimportanttomakeclearattheoutsetthatthesevariablesarevery difficulttostudyinthecontextofMMHL.Firstofall,sincethelatterlogicallyandpracticallyfollowstheformer,wecanrefer toasinglevariable,althoughitshouldbenotedthatsomechildrenwithmildHLdonotwearhearingaidsatall.Although thisvariablehasbeenshowntobeimportantinthecaseofmoreseverehearingloss,withsignificantcorrelationstypically foundbetween agesofdetection/fittingand languageperformance,itsrelevancetothecaseofMMHLhasnotbeen shown.Thereasonsforthisaresimple.GreaterHLtendstobediagnosedearlierthanlesssevereHL,asitismore apparenttoparentsandotherpersonsassociatingwiththechild.Atthesametime,milderhearinglosscanbeexpectedto have less severe linguistic repercussions,particularly in childrenwith mild HL whohave never worn hearing aids.
Explorationofthelengthofaidedexposuretolanguageandlanguageperformancethuscannotreallybecarriedoutinthe caseofMMHL.Discussionofthisvariablewillthusnecessarilybelimited,thoughdirectlyaddressedinSection3.
AgeneraldiscussionispresentedinSection4,whichreturnstothebasicquestionofthenatureofthelinkbetween MMHLandlanguageimpairment,andsuggeststhatthislinkmaybetheresultofinterfacebetweenlanguageandareasof linguisticperformancesystems.
2. Morphosyntacticimpairment
ThissectionreportsonmorphosyntacticperformanceinseveralgroupsofFrench-speakingchildrenandadolescents withMMHL.First,wewillpresentresultsindicativeofgeneralperformance,asmeasuredbystandardizedtests.Wewill subsequentlyfocusonparticularsyntacticstructures,reportingonresultsusingavarietyofdifferentkindsofmeasures:
elicited production of accusative clitics, elicited production of subject versus object relative clauses, grammaticality judgmentsonmorphosyntacticproperties(passives,relativesclauses,personandnumberagreementonpronouns,and verbraisinginsimplesentences),andfrequencyandtypeofsyntacticembeddinginspontaneouslanguagesamples.
ClinicianswhoworkwithchildrenwithMMHLoftendescribethesechildrenasgivingtheillusionofnormality.Wehavesought tousedifferenttypesofmeasuresinanefforttounderstandwhatmightbebehindthisobservation.Do(someof)these childrenavoidstructureswhichcausethemdifficulties?Finally,ineachofthesestudies,wehavesoughttocomparethe morphosyntaxofchildrenandadolescentswithMMHLtothatoftheirTDpairs,butalso,crucially,tothatofindividuals acquiringFrenchasasecondlanguageoracquiringFrenchwithSLI.Inthissection,wewillfocusentirelyoncross-sectional results,reservingourlongitudinalstudyforSection3,whichdirectlyaddressesthequestionofinter-subjectvariability.
2.1. Generalperformance
How do children with MMHL fare on standardized language tests designed to assess children for language impairment?Toanswerthisquestion,wepresentresultsfromourstudyof32childrenwithMMHLaged6;1to11;11 (M=9;2,SD=1;9).Hearingthresholds(PTA)inthisgroupofchildrenrangedfrom29to69decibels.TheageofHL detectionvariedbetween0;2and8;8yearold(M=4;9,SD=1;9),ageofHLfittingbetween0;5and9;9yearold(M=5;6, SD=2;3).ThreechildrenwithmildHL(29--35dB)didnotwearhearingaids.Itisworthnotingthatthesechildrenwerenot recruitedfrom caseloads ofspeech-language pathologists,since our goalwas tostudya group ofchildrenwhose languagewastrulyrepresentativeofthatofentirepopulation ofchildrenwithMMHL.Wethereforerecruitedthrough systematicconsultationofalldossiersofchildrendiagnosedwithbilateral,prelingual,sensorineuralmildormoderate hearing lossat a regional pediatric hospitalotorhinolaryngology service. Ofthe 40families contacted, 2 refused to participateinthestudy,and6wereexcludedaftertestingbecauseofclearcomplicatingfactors(onechildwithcerebral palsy,andtheotherswithglobaldevelopmentaldelayand/orabsenceofregularschooling).The32childrenwithMMHL retainedforstudywerecomparedto10childrenwithSLIaged8;0to11;11(M=10;1,SD=1;4).6Therewasnosignificant differenceinagebetweenthetwogroups(U=120.5,p=0.2).7Allparticipantswereadministeredacomputerizedoral languagebattery(Khomsietal.,2007).Inthisbattery,expressivegrammarisevaluatedusingasentencecompletion task:Thechildlistenstoasentencecorrespondingtoapictureonacomputerscreenandthenhastocompletetheendof thesentencetofitwiththepicture(e.g.,Ici,lechienboitsonlait;làleschiens...boiventleurlait‘Herethedogisdrinking hismilk;There,thedogs...aredrinkingtheirmilk’).Thistestassessesabilitytoproduceavarietyofspecificgrammatical morphemes:nominal,adjectivalandverbalinflexion,irregularplurals,prepositions,passives,andpronominalclitics.The morphosyntacticcomprehensiontest consists ofa taskassessingability tojudgethe well-formednessofsentences presented orally with a related picture. The child has to decide whether the sentence heard with the picture is grammaticallycorrect,bygivingayesornoanswer(e.g.,apictureofseveralbirdssittingonatreebranchisshownand thechildhearsthefollowing(ungrammaticalsentence):*Lesoiseauxestsurlabranche‘Thebirdsisonthebranch’).
AlthoughthechildrenwithMMHLscoredsignificantlybetterthanthechildrenwithSLIoneachofthesetwotasks(for expressive grammar, U=45, p<001; for receptive grammar, U=65.5, p<01), their mean scores showed that morphosyntax,astestedbysuchgeneralmeasures,isimpairedinthesechildren:ThemeanSDwas 1.1forexpressive grammar( 3fortheSLIgroup)and 1.4forreceptivegrammar( 4fortheSLIgroup).Behindthesegroupmeans,there washighinter-subjectvariability:whileimpairmentcanbesevere,notallchildrenareseverelyimpaired,andsomearenot impairedatall.Toillustratethispoint,considerFig.1,whichpresentsindividualresultsofthechildrenwithMMHLforthe sentencecompletiontask:16childrenobtainedlowscores(asdefinedbyaSDbelow 1)and16childrenhadcompletely normalones(i.e.≥ 1SD).Fig.1alsoshowsthatmanyofthelow-scoringchildrenobtainedverylowscores:37%ofthem hadSDsbetween 1.65and 6.Overall,50%ofthechildrendisplayedsomedegreeoflanguageimpairment,arate whichistypicalforstudiesofchildrenwithMMHL(seeforexampleGilbertsonandKamhi,1995;Briscoeetal.,2001or Hanssonetal.,2007).
Thesetwomorphosyntactictasksweretwoofthesixtasksoftheorallanguagebatteryallparticipantswereadministered.
Theothertaskstestedphonology(awordrepetitiontask),expressivevocabulary(apicturenamingtask),lexicaljudgment
6Thiscomparativestudywaspartofthesecondauthor’s(unpublished)doctoraldissertation(Delage,2008).
7Notethat,duedothenon-normaldistributionofthedata,ouranalyseswereconductedwithnon-parametrictests,withMann--Whitneytestfor intergroupcomparisons,WilcoxontestforintragroupcomparisonsandfinallywithSpearman’srankcorrelation.
(aword--picturematchingtask)andsentencecomprehension(asentence--picturematchingtask).Interestingly,ifwefocus exclusivelyonthechildrenwithMMHLwhohadlowglobalresultsontheentirelanguagebattery(the16childrenhavingat least 2scores< 1.65SDamong the 6 scoresin the battery),childrenwe havedubbedas ‘‘MMHLwith Language Impairment(MMHL-LI)’’inFig.2,nostatisticaldifferenceswerefoundbetweentheMMHLandSLIgroupsforeitherofthetwo morphosyntactictasks(forexpressivegrammar,U=43andp=0.08orforreceptivegrammar,U=57.5andp=0.33),as illustratedbyFig.2.Inotherwords,asubgroupofchildrenwithMMHLhadperformancequantitativelyverysimilartothatof childrenwithSLI.Theimportantpointhereisthat,whereastheprevalenceofSLIisestimatedtobearound7%(Tomblinetal., 1997),theprevalenceoflanguageimpairmentintheMMHLgroupherereaches50%andthuscouldnotbetheresultofa simplephenomenonofcomorbidity(sincenomorethan7%ofdeafchildrenshouldalsohaveSLI).
Summarizing, standardized language measures of morphosyntax identified approximately 50% of this group of childrenwithMMHLashavingimpairedmorphosyntax.Thisresult,includingthisparticularrateofimpairment,coincides withthoseofpreviousstudies,includingourown.Alsoconsistentwithpreviousstudiesisthefindingthat,asagroup,the childrenwithMMHLwithover-allimpairedlanguageresembledchildrenwithSLI,asagroup,intermsoftheseverityof theirmorphosyntacticimpairment.
2.2. AclinicalmarkeroflanguageimpairmentinFrench:accusativecliticproduction
Standardizedtestsofthetypepresentedintheprecedingsectiongiveageneralpictureofthedegreeofimpairment.
However,theyarenotdesignedtoexplorewhichareasofmorphosyntaxcausedifficultiesandwhichonesdonot.Tasks focusedonspecificlanguagepropertiesallowforsuchexplorationandthusholdthepromiseoffurtheringunderstanding of thenatureoflanguageimpairmentin childrenwithMMHL.In otherwords, wewouldliketoknow whichareas of morphosyntaxareaffectedandwhetherthesecanbedirectlylinkedtoHL.
Fig.2. Expressiveandreceptivegrammar:MeanZ-scoresobtainedbychildrenwithMMHLwithlanguageimpairment(MMHL-LI)andbychildren withSLI.
Fig.1. Expressivegrammar:standarddeviationsobtainedbychildrenwithMMHLcomparedtomeanSDobtainedbychildrenwithSLI.
TullerandJakubowicz(2004)exploredthequestionofwhetherperceptuallimitationsresultingfromHLleadtodifficulty withgrammaticalelementswhosesurfaceformislesssalient.Todothis,severalsuchgrammaticalitemswerecompared, and,inparticular,definitedeterminersandthirdpersonaccusativeclitics(seeTable2).Theseparticularmorphemeswere chosen since, although their phonological form is identical, they differ in terms of the complexity of the linguistic computationrequiredtoderivethem.Notably,whiledefinitedeterminersappearinthecanonicalpositionfordeterminers inFrench(atthebeginningoftheDP),accusativecliticsdonotappearpostverballyinthecanonicaldirectobjectposition, butratherpreverbally(seeTulleretal.,2011,fordetaileddiscussionofthecomplexpropertiesofthirdpersonaccusative clitics,whichalsoincludethefactthattheygenerallyco-occurwithanominativecliticinspokenFrenchandthattheyare optionalunderspecificdiscourseconditionsinspokenFrench).
Elicitedproductionofdeterminerswascomparedtothatofthirdpersonaccusativecliticsin20childrenwithMMHL aged6;0to13;9comparedto10childrenwithSLIaged6;0to8;6andto36youngtypically-developing3-,4-,and6-year- olds(12childrenperagegroup).ThechildrenwithMMHLweredividedintotwoagegroups:10childrenaged6--8and10 childrenaged9--13.Anexperimentalprobewasusedinwhichchildrenwereshownapictureandaskedaquestion,asin (1),elicitingadefinitedeterminer,and(2),elicitinganaccusativeclitic.
(1) [pictureofMinniesprinklingagirlwithagardenhose]
Experimenter:QuiaroseMinnie?‘WhoissprinklingMinnie?’ Expectedresponse:lafille‘thegirl’
(2) [pictureofagirlaimingawaterpistolatMinnie]
Experimenter:QuefaitlafilleàMinnie?‘What’sthegirldoingtoMinnie?’
Expectedresponse:Ellelavise‘She’saimingather’
AsillustratedinFig.3,accusativecliticswerelesswellmasteredthandefinitedeterminersbychildrenwithMMHL,by children with SLI, and by 3-year-old control children. Performance in all groups of children reached ceiling for determiners,forwhichproductionrateswereabove95%,whereasaccusativecliticswereproducedatmuchlower ratesbychildrenwithMMHL(40%inyoungerchildrenand68%inolderchildren),byTD3-and4-year-olds(44%and 79%,respectively),andbychildrenwithSLI(only29%).TullerandJakubowicz(2004)concludedthattheseresults showedthatperceptualsalienceisnotthe(sole)factorwhichdeterminesmorphosyntacticperformanceinchildrenwith MMHL(orwithSLI),sinceonlyoneoftwophonologicallyidenticalmorphemeswithlowsalienceisasourceofdifficulty andsuggestedthereforethattheincreasedcomputationalcomplexityinvolvedinoneofthem(accusativeclitics)isakey factorinaccountingfordifficulty,inlearners,bothtypicalandatypical(withandwithouthearingloss).8Asmentioned above,ithasbeenarguedthataccusativecliticproductionisarobustmarkerofmorphosyntacticdifficultiesinchildren acquiringFrench.Notonlydodifficultiespersistintoadolescence(seeTulleretal.,2011),theyarefoundinthecontextof verydifferentpathologies,inMMHLandSLI,butalsoinbenignchildhoodepilepsy(seeTulleretal.,2011;Monjauze, 2007),andinchildrenacquiringFrenchasasecondlanguageinCanadianimmersionschools(ParadisandCrago, 2003;Paradis,2004).GiventhecircumstancesoflanguageacquisitioninchildrenwithMMHL,andinparticularthefact thattimingissuesareinvolved(accessibleinputbeginninglateformanyofthesechildren,andremaininglessthan optimal), we hypothesized inSection 1 that comparison with TD children learning a second language (L2) after early childhood would be appropriateinthatlanguage input forthese children alsostarts late. Is performance in thesethreedifferentgroups,childL2learners,childrenwithMMHL,andchildrenwithSLI,comparableforthismarkerof morphosyntacticdifficultyinFrench?
Table2
Definitedeterminersanddirectobjectpronouns(AccusativeClitics)inFrench.
Definitedeterminers Directobjectpronouns(accusativeclitics)
Singular Plural Singular Plural
Masculine Feminine Masculine Feminine
le la les le la les
legarcon theboy
lafille thegirl
lesfilles thegirls
Maxlevoit Maxhimsees
Maxlavoit Maxhersees
Maxlesvoit Maxthemsees
8TullerandJakubowicz(2004)notethatnosignificantdifferenceswerefoundbetweentheage-matchedchildrenwithMMHLandwithSLI (foraccusativecliticproduction,U=39.5,p=.43),thoughtheolderchildrenwithMMHLoutperformedtheyoungerchildrenwithMMHL(U=24, p<.05).WeaddressageeffectsdirectlyinSection3.2.
Webringtogetherhereresultsallowingforawideinter-groupcomparisonofelicitedproductionofaccusativeclitics:
The32childrenwithMMHL(ages6--11)presentedabove inSection2.1 werecompared to3 groupsofTDchildren (6-year-olds,8-year-olds,and11-year-olds),to18childrenwithSLI(ages6--12),andto29firstlanguageBritishEnglish- speaking childrenacquiring L2 French (ages 6--12), naturally,after familyimmigration to France.9Accusative clitic productionwastested,butalsoproductionoftwoothertypesofpronominalclitics,nominativeandreflexive.Nominative cliticsdo notinvolvemovement to anon-canonical position,and,reflexivecliticshave alocal antecedent,whereas accusativecliticsinvolvebothappearanceinanon-canonical(preverbal)positionandtheirantecedentisnecessarilynon- local (seeJakubowiczetal.,1998; Tulleretal.,2011).Giventhesedifferencesinthe computationofthesedifferent elements,weexpectedtofindthataccusativecliticswouldbelesswellproducedthantheothertwotypesofclitics.To elicittheseclitics,wedevelopedaProductionProbeforPronounClitics(PPPC),whichelicitsproductionofthirdperson nominative,reflexive,andaccusativeclitics,10byrequiringaresponsetoaquestionabouta drawingappearingona computerscreen,asin(3)and(4).
(3) Experimenter:QuefaitThomas? ‘What’sThomasdoing?’ Expectedresponse:Ilsepèse. ‘He’sweighinghimself’
(4) Experimenter:QuefaitMarieaveclechien? ‘What’sMarydoingtothedog?’ Expectedresponse:Ellelelave. ‘She’swashingit’
Table3presentstheresultsobtainedbyeachgrouptested.Accusativecliticswereproducedatlowerratesthanboth nominativeandreflexiveclitics,ineachofthe‘‘atypical’’groups---MMHL,SLI,andL2.Thiswasalsothecaseforthe youngestTDgroup,the6-year-olds,butnotfortheTD8andtheTD11groups,whichdisplayedceilingperformance.Inter- groupcomparisonsrevealedasignificantgroupeffect,withtheMMHLperformingworseonaccusativecliticsthanthe TD8(U=79,p<01)andtheTD11(U=55,p<001),thoughnotthantheTD6(U=163.5,p=0.4),andbetterthantheSLI
Fig.3. Meanproductionratesfordeterminersandaccusativeclitics:MMHL,SLI,TD-3,TD-4,TD-6.
9ThemeanageoffirstexposuretoFrench,uponarrivalinFrancewas7;0(SD1;10andrange4;5to10;9),andthemeanlengthofexposure was2;5(SD1;2).AsubsetoftheL2resultswasreportedinanunpublishedthesis(Haiden,2011).
10Thistaskalsoelicitsfirstpersonclitics,whichfarebetterthanthirdpersonclitics(seeTulleretal.,2011).
(U=182.5,p<05).Interestingly,therewasnosignificantdifferenceforproductionofthisclinicalmarkerbetweenchildren withMMHLandthegroupofL2children(U=441.5,p=0.7).
Insummary,twodifferentgroupsofFrench-speakingchildrenwithMMHL(atotalof52children,aged6--13,meanage 9)displayedthesamespecificdifficultywiththirdpersonaccusativecliticsfoundinyoungerTDchildren,inchildrenthe sameagewithSLI,andinchildrenthesameagelearningFrenchasasecondlanguageafteragefour.Whilechildrenwith MMHL,asagroup,performedbetterthanchildrenwithSLIofthesameage,theirperformancewasquitecomparableto thatoftheL1English/L2Frenchchildren.
2.3. Measuresofcomputationalcomplexity
AccusativecliticsarenotoriouslydifficultfordifferentlearnersofFrench,andthisdifficultyincludeschildrenwithMMHL, aswehavejustseen.Investigationofotherstructuresentailinghighlevelsofcomputationalcomplexitywillallowustoseeto whatextentthecontextofMMHLleadstolanguageimpairment.Wewillreportonresultstestingchildren’sabilitytojudgethe grammaticalityofrelativeclausesandpassives,ontheonehand,andsimplesentencesinvolvingagreementandordinaryV- to-Tenseraising,ontheother.Wethenfocusspecificallyonelicitedproductionofsubjectandobjectrelativeclauses,and turn,finally,totheproductionofdifferenttypesofclausalembeddinginspontaneouslanguagesamples.
2.3.1. Grammaticalityjudgmentsonconstructionsofvaryingcomplexity
Inordertoassessavarietyofstructuresimplyingdifferentdegreesofcomputationalcomplexity,weadministereda grammaticalityjudgmenttestto10adolescents withmoderateHL(50--70dB)compared toadolescentswithsevere (71--89dB)orprofoundHL(92--120dB),totypicallydevelopingadolescentswhowereage-matchedforthedeaflearners andalsotoL2adults.ThisL2populationwasmadeupof19nativespeakersofEnglishwhohadbegunacquiringFrench atameanageof13;11(SD3;3),whosemeannumberofyearsstudyingFrenchwas6;10(SD3;1),andwhowereallin Franceona study-abroadprogram (havingarrivedinFrance between1 and 6monthsprior totesting time).These learnersweredividedintoagroupof10advancedand9intermediatelearners,accordingtotheresultsofClozeand vocabularytests.InformationonparticipantsissummarizedinTable4.
Thegrammaticalityjudgmenttestconsistedof48items(24grammaticaland24ungrammatical),whichappearedona computerscreen,alongwith‘‘yes’’and‘‘no’’buttons;theparticipanthadtoclickaccordingtohis/her judgmentofthe sentence.Awrittentestwasusedinsteadofanoraltestinordertotestgrammaticalityjudgmentforsentenceswithdifferent
Table4
AgeandsexofTDadolescents,L2adults,andadolescentswithHL.
Groups N Sex Meanage Agerange
TDadolescents 51 28M23F 13;1 11;4--15;3
L2adults Advanced 10 1M9F 21;1 19;11--23;2
Intermediate 9 5M4F 21;4 19;9--25;11
AdolescentswithHL ModerateHL 10 8M2F 14;1 11;10--16;4
SevereHL 15 10M5F 14;1 11;1--15;9
ProfoundHL 26 13M13F 13;3 11;0--16;6
Table3
Production(%)ofnominative,reflexiveandaccusativeclitics:MMHL,TD,SLI,andL2.
Group Agerange Nominative Reflexive Accusative
MMHL(N=32) 6;1--11;11(M=9;2) 91%(SD=13) 87%(SD=27) 65%(SD=33) Accusative<nominative(p<001) Accusative<reflexive(p<001) SLI(N=18) 6;6--12;11(M=9;5) 76%(SD=36) 72%(SD=36) 42%(SD=36) Accusative<nominative(p<01)
Accusative<reflexive(p<01) L2(N=29) 6;4--12;7(M=9;5) 85%(SD=21) 72%(SD=35) 61%(SD=35) Accusative<nominative(p<.01)
Accusative<reflexive(p<.05) TD6(N=12) 6;1--6;7(M=6;4) 98%(SD=2) 100% 78%(SD=15) Accusative<nominative(p<01)
Accusative<reflexive(p<01) TD8(N=12) 7;9--8;7(M=8;2) 98%(SD=6) 100% 93%(SD=13)
TD11(N=12) 11;1--11;9(M=11;4) 94%(SD=9) 100% 98%(SD=5)
degreesofcomputationalcomplexitywithouttestingatthesametimeproficiencyinlip-readingintheadolescentswith severeandprofoundHL.Foursyntactic constructionswere evaluated(12 items each):relativeclauses(including subject,object,indirectobjectoradjunctrelativeclauses)inwhichtheungrammaticalsentencesinvolvedtheformofthe relativepronoun,asin(5),passives,asin(6),inwhichtheerrorinvolvedtheabsenceofpar‘by’intheby-phrase,simple sentencesinwhichgenderandnumberagreementonpronounswasmanipulated,asin (7),andsimplesentences thatvariedingrammaticalityaccordingtowhetherthetensedverbwasraisedtotheleftoftheadverbofnegationornot, asin(8).
(5) a. J’aidesamisquiarriventcesoir ‘Ihavefriendsthat[subject]arecomingtonight’ b. *J’aidescopainsqueviennentcesoir ‘Ihavefriendsthat[object]arecomingtonight’ (6) a. L’arbreestcachéparlebâtiment ‘Thetreeishiddenbythebuilding’
b. *Lebâtimentestcachélesarbres ‘Thebuildingishiddenthetrees’
(7) a. Lamoto?Jelagaredansleparking ‘Themotorbike[fem]?Iit[acc-fem]parkintheparkinglot’ b. *Lavoiture?Jelegaredanslegarage ‘Thecar[fem]?Iit[acc-masc]parkinthegarage’
(8) a. Legarçonneveutplusfairesesdevoirs ‘Theboynolongerwantstodohishomework’
b. *Mescopainsnepassortentcesoir ‘Myfriendsnotgoouttonight’
Scheidnesetal.(2009)attributedvaryingdegreesofcomplexitytothesefourconstructions,accordingtowhetherornot their derivationentails Internal Merge(syntactic movement), in accordance withthe Derivational ComplexityMetric (Jakubowicz,2005).Sincerelativeclausesinvolvewh-movementofawh-operatortotheleftperipheryoftherelative clause,andsyntacticpassivesA-movementofthedirectobjecttosubject,thesebothentailahighdegreeofcomplexity.
Ontheotherhand,thesimplesentencestestinggenderandnumberagreementonpronounsandadverbplacementdo notinvolveeitherwh-movementorNP-movement.Theselatterdoentailraisingoftheverbtoinflection(resultinginthe surfacepositionoftheadverbaftertheinflectedverb),butthismovementisrequiredinalltensedclausesinFrench.These considerationstakentogethermeanthatthesimplesentencesinwhichonlynumberandgenderagreementonobject clitics(agreement)orverb-adverborderaremanipulatedarecomputationallylesscomplexthanthesentencescontaining relativeclausesorpassives.
Fig.4presentsthemeanglobalscoresonthegrammaticalityjudgmenttaskforeachofthegroupstested:groupresults wereallabovechance(ascertainedtobeat30/48viaabinomialtest).TheadolescentswithmoderateHLshowedpoorer performance(M38.4/48)thanthe TDadolescents (M43.1/48) andthanthe advancedL2 adults(M44.9/48). They performedbetterthantheadolescentswithprofoundHL(M32/48;U=63.5;p<.01),butaswell/asbadlyasboththe IntermediateL2learners(M38.4/48;U=45;p=1.00)andasadolescentswithsevereHL(39.1/48;U=69;p=0.74).
Interestingly,andwereturntothispointinSection3.1,withinthegroupofadolescentswithmoderateHL,asignificant negativecorrelationwasfoundbetweenthesescoresanddegreeofhearingloss(rs= 0.607).
Comparisonsofitemtypes(Table5)revealedthatnoneofthesub-scoresobtainedbyadolescentswithmoderateHL differed significantly from those of adolescents with severe HL.However, adolescents with moderate HL performed significantlyworsethantheintermediateL2adultsonitemsinwhichnumberandgenderagreementweremanipulatedon accusativecliticsandonrelativeclauses.Infact,allthreegroupsofparticipantswithHLhaddifficultiesinjudgingagreement onaccusativecliticsinsimplesentences:Theyfailedtodetectagreementviolations(7b)significantlymoreoftenthanthe adolescentcontrolsandthanbothgroupsofL2learners.Wesupposethatthismaybeaneffectofthepersistentdifficulty associatedwithaccusativeclitics(seeSection2.2).Thedifficultywithrelativeclauses,theitemsarguedtoinvolvethehighest degreeofcomputationalcomplexity,issharedbyallgroupsofatypicallearners,whoseperformancewassignificantlybelow
Fig.4. Scoresongrammaticalityjudgmenttask:meannumberofcorrectresponses/48.
thatoftheTDadolescents.TheadolescentswithmoderateHLalsoperformedsignificantlybeloweventheintermediateL2 adults.And,likebothgroupsofL2adults,andtheothergroupsofadolescentswithHL,grammaticalityofpassivesentences waslesswelldetectedbyadolescentswithmoderateHLthanintheTDadolescents.Thepatternoflowerscoresforrelative clausesandhigherscoresforverbraisingisparticularlynoteworthy.TheL2andthedeafgroupsalldisplaythissame pattern.11ThisisstrikingbecauseitmeansthattheL2groupshadmoredifficultycorrectlyjudgingthegrammaticalityofthe constructionswhichweremorecomplex,butsimilartotheirL1(relativeclauses)thantheydidjudgingthegrammaticalityof constructionswhichwerelesscomplex,butdifferentfromtheirL1(V-Raising).Scheidnesetal.(2009)arguedthatthisresult wouldseemtoindicatethatL1transferfromEnglishplaysalimitedrole,atleastinthistypeoftask.
Ageneralpicture thusemergesin whichthe computationallymost complexitemswere moredifficult forbothL2 learners and learners with HL, including adolescents with moderate HL. This fundamental similarity between the L2learnersandthelearnerswithHLisallthemorenoteworthygiventhedifferencesinageandquantityandqualityof exposurebetweenthesegroups.
2.3.2. Elicitedproductionofrelativeclauses
Relativeclauseshavebeenwellstudiedinbothtypicalandatypicallanguageacquisition(seeFriedmannetal.,2009,fora review).Thesyntaxofrelativeclausesinvolvesbothclausalembeddingandwh-movementofanoperatorfromthethematic positiontoaclause-initialposition,twosourcesofcomputationalcomplexityundertheapproachbeingadoptedhere.While bothsubject(9)andobject(10)relativeshavethesetwoproperties,objectrelativesinlanguageslikeFrench,inaddition, entailadependencybetweenthethematicpositionoftheobject(thecanonicaldirectobjectposition)andthewh-operator whichislongerthanthatbetweenthesubjectpositionanditsantecedent.AsFriedmannetal.(2009)pointout,thepath betweenthethematicpositionanditsantecedent,crucially,isinterruptedbyanothernounphrase(thesubject),whichthey arguerendersthesyntacticcalculationinobjectrelativesconsiderablymorecomplexthanthatinsubjectrelatives.
(9) Je connais le garçoni[quiitipousse le chien] ‘I know the boy who is pushing the dog’
(10) Je connais le garçoni[0i que le chien pousse t t] ‘I know the boy who the dog is pushing’
Weevaluatedtheproductionofsubjectandobjectrelativeclausesin29ofthe32childrenandadolescentswithMMHL presentedinsection2.1,whentheywereagedfrom7;11to13;11(M=11;2)andcomparedtheirresultstothoseof126- year-oldcontrolchildrenandto8childrenwithSLIaged8;0to11;11(M=9;6).12TheparticipantswithMMHLweredivided intotwoagegroups:14childrenwithMMHLagedfrom7to11(M=9;8)and15adolescentswithMMHLagedfrom11to 13(M=12;6).Weusedthe sametaskasFriedmannandNovrogodsky(2004),foreliciting productionofsentences containingasubjectoranobjectrelativeclause(asin(11)foranobjectrelative)
Table5
Sub-scores on grammaticality judgmenttask items: mean numberof correctresponses/12 (above chance performance=9/12, chance performanceinboldface).
TD11--15 Adv.L2 Inter.L2 ModerateHL SevereHL ProfoundHL
V-Raising M/12 10.7 11.3 11.0 ns 10.6 10.4 9.1
SD 0.8 1.6 1.3 0.8 1.1 1.8
Agreement M/12 10.8 10.9 10.9 * 9.1 9.6 7.6
SD 2 1.2 1.2 1.9 2.0 2.6
Passives M/12 11 10.7 9.6 ns 10.7 10.5 8.5
SD 1.7 2.7 2.2 1.8 2.0 2.3
Relatives M/12 10 9.5 9.8 * 7.5 8.0 6.5
SD 1.2 1.5 2.6 1.3 2.2 2.0
11Awithin-groupcomparisonshowsthattheL2ersscoredsignificantlyloweronRelativesclausescomparedtoV-raising(Z=2.56,p<0.05) andAgreement(Z=2.10,p<0.01).Thesameinter-constructiondifferencesweresignificantinthedeaf:relativeswereworsethanV-Raising (Z=5.59,p<.0001)andAgreement(Z=3.01,p<.01).
12Thiscomparativestudywaspartofthesecondauthor’sdoctoraldissertation(Delage,2008).
(11) Elicitationofanobjectrelative:
Experimenter:Voiciunéléphantetungarçon‘Hereareanelephantandaboy’ Icil’éléphantpeintlegarçon‘Heretheelephantispaintingtheboy’
Icilegarçonpeintl’éléphant‘Heretheboyispaintingtheelephant’ Celui-là,c’estqueléléphant?‘Whichelephantisthis?’
[whilepointingtotheelephantinthebottompicture]
Expectedanswer:C’estl’éléphantquelegarçonpeint
‘That’stheelephantthattheboyispainting’
Fig.5presentsproductionratesofsubjectandobjectrelativeclausesforeachgrouptested.Despitethepresenceofhigh inter-subjectvariabilityinallgroups,itemergesthatchildrenwithMMHLandchildrenwithSLIproducedsignificantlyfewer subjectrelativeclausesthantheolderparticipantswithMMHL(U=23.5,p<.001)andthecontrols(U=3.5,p<.01).The subjectrelativerateoftheyoungerchildrenwithMMHLdidinfactnotdiffersignificantlyfromthoseofthechildrenwithSLI.
Theobjectrelativeratesaremuchlowerthanthesubjectrelativeratesinallgroupsofchildren.
Thissubject--object relativeasymmetryiswellknowninthe acquisitionliterature(seeFriedmannetal.,2009,for review).Whatisinterestinghereistoseewhatkindsofstructureschildrenproducedinsteadofanobjectrelative.Fig.6 showsthedifferenttypesofresponsesproducedbychildreninthefourgroups.
6-year-oldcontrolchildrenandolderchildrenwithMMHLgavesimilarresponses,producingcausativepassiveswith thereflexiveformsefaire‘makeoneself’andpassivesubjectrelativeclausesasin(12bandc),inlieuofanobjectrelative
Fig.6. Structuresproduced(%)insteadofanobjectrelativeclause:MMHL,SLI,TD-6.
Fig.5.Production(%)ofsubjectandobjectrelativeclauses:MMHL,SLI,TD-6.
(12a).Althoughthis kindofresponse isnot what waselicited in the warm-up, itissemantically, pragmatically and grammaticallycorrect,and,assuch,isexpected.Althoughproductionofanappropriatesubjectrelativeinthiscontext requiresusingapassivestructure,itdoesnotentailaconfigurationinwhichthepathbetweenthethematicpositionandits antecedentisinterruptedbythesubjectnounphrase.Productionofapassivesubjectrelativeconstitutesnotonlythemost appropriatealternativetoanobjectrelative,itisthesyntacticallymostsophisticatedalternative,inthatitinvolvesboth productionofarelativeclauseandapassivestructure.75%ofthenon-objectrelativeanswersproducedbycontrolsand 92%ofthoseproducedbyolderchildrenwithMMHLwereresponsesofthistype.
(12) a. C’estle roi quele garçonpeigne it’s thekingthattheboy combs
‘Itisthekingwhotheboyiscombing’
b. C’estle roi qui se fait peigner parle garçon it’s the kingwho himselfmakesto-combby theboy
‘Itisthekingwhohashimselfcombedbytheboy’
c. C’estle roi qui estpeigné parle garçon it’s thekingwhois combedby theboy
‘It’sthekingwhoiscombedbytheboy’
However,theyoungerchildrenwithMMHL(aged7--11)andthechildrenwithSLIproduced,respectively,only54%and 31% non-object relative responses of this type. Instead, they tended to produce less complex structures: simple sentenceswithoutanyrelativeclauseatall(42%inchildrenwithSLIand18%intheyoungerchildrenwithMMHL),and evenactivesubjectrelatives,whichthusinvolvedthematicrolereversal(5%intheyoungerchildrenwithMMHLand23%
inthe childrenwithSLI).The youngerchildrenwithMMHL alsoproducedadverbialrelativeclauses (22%)usingoù
‘where’,both grammatical, as in (15a),with a direct objectpronoun, and thus nowh-movement of the object, and
ungrammatical,asin(16a),withamissingdirectobjectpronoun.
(13) a. Lelionmouillel’éléphant SimpleSentence
‘Thelionissprinklingtheelephant’
b. C’estl’éléphantquelelionmouille TargetObjectRelative
‘Itistheelephantthatthelionissprinkling’
(14) a. C’estlechatquimordlechien ActiveSubjectRelative
‘It’sthecatthatbitesthedog’
b. C’estlechatquelechienmord TargetObjectRelative
‘It’sthechatthatthedogbites’
(15) a. C’estcelleoùlagirafelalave Where-relativewithpronominaldirectobject
‘Itistheonewherethegiraffeiswashingher’
b. C’estlagirafequelafillelave TargetObjectRelative
‘Itisthegiraffethatthegirliswashing’
(16) a. C’estlegarçonoùlechevalpeigne Where-relativewithoutpronominaldirectobject
‘Itistheboywherethehorseiscombing’
b. C’estlegarçonquelechevalpeigne TargetObjectRelative
‘Itistheboywhothehorseiscombing’
Althoughwhere-relativeswerenot observedin the SLIgroup orinthe TD6 groups,we foundpreciselythis typeof responseinTD5-year-olds,inasimilarelicitedproductiontask(Friedmannetal.,2013),andLabelle(1990)foundthemin elicitedproductioninTD3-to6-year-olds.Thisstrategyalsoallowsthechildtoavoidwh-movementofthedirectobject, therebysimplifyingthederivation.ThefactthatthisparticularavoidancestrategyisnotobservedinolderMMHLchildren, andalsothefactthattheolderchildrenwithMMHLproducedmorepassivesubjectrelativesarebothindicativeofwhat appearstobeanageeffect,withtheolderchildrenwithMMHLhavingmorematurelanguage.13
13Areviewersuggeststhatthiscouldalsomeanalinkwiththelengthofhearingaiduse.AsnotedinSection1,thisvariablemaycorrelatewith degreeofHL.ThishypothesiswasconfirmedinthestudybyDelage(2008),whereitwasfoundthatdegreeofHLinthese32childrenwithMMHL inverselycorrelatedwithageofHLdetection(rs= 0.41,p<.05)andfitting(rs= 0.44,p<.05).Thisresult,inconjunctionwiththefactthatsome childrenwithmildHLhaveneverwornhearingaidsatall,thuscomplicatesanydirectexplorationoflanguageperformanceandlengthofaided hearing.AswewillseeinSection3,nosuchlinkswerefound.