slcn slides for distr - ru.nl · on mul‐cyclicity in early asl and libras. boston university...
TRANSCRIPT
BibibiProject SLCN3–Stockholm
15June2010 1
Annota&onofChildLanguageCorpora:Acomparisonoftwomethodswithspecial
emphasisonbimodalbilingualdata
DianeLillo‐MarAn&DebbieChenPichlerSignLinguisAcsCorporaNetwork
Workshop3:AnnotaAonStockholm,Sweden14‐16June2010
1Ahandouttoaccompanythistalkisavailableat:
hPp://web.me.com/dianelillomarAn/DLM/PresentaAons.html
Acknowledgments• Collaborators: RoniceMüllerdeQuadrosand
JulieHochgesang• Warmthanksto:– bimodalbilingualchildrenandtheirfamilies– researchassistants
• Financialsupportfrom:– AwardNumberR01DC009263fromtheNaAonalInsAtuteonDeafnessandOtherCommunicaAonDisorders.ThecontentissolelytheresponsibilityoftheauthorsanddoesnotnecessarilyrepresenttheofficialviewsoftheNIDCDortheNIH.
– TheGallaudetResearchInsAtute.– CNPq(BrazilianNaAonalCouncilofTechnologicalandScienAficDevelopment)Grant#200031/2009‐0and#470111/2007‐0.
2
Longitudinalstudiesofchildlanguage(childlanguagecorpora)
• AddressawidevarietyofresearchquesAons• Eachdatasetcanbeminedinmanyways
• Complementsexperimental/cross‐secAonalstudynicely
3
Challengesofconduc&ngchildlongitudinalstudies
• Balancechild’scomfortzoneandneedforarepresentaAvesampleoflanguage
• RequiresrealcreaAvitytocoaxarichandvariedsampleoutofchild– InvestinAme,gettoknowchildandfamily,learnwhatgetsthemtalking/signing
– Thinkingonyourfeettofollowthechild’sleadandexpandonwhatthechildsays
4
Collabora&vestory‐telling• Ben051(2;07)
5
Themoviehasbeendeletedfromthedistribu3onfile.
Challengesofconduc&ngchildlongitudinalstudies
• Letchilddowhatshewants,yetmakesurethatcondiAonsaremaximizedforlatertranscribability– MonitorambientlighAngandsound
– Filmchildinroomswithoutplacestohideortoomuchoff‐cameraspace
6
BibibiProject SLCN3–Stockholm
15June2010 2
Datacollec&oninthedark
7
• SAL002(1;08)
Themoviehasbeendeletedfromthedistribu3onfile.
Technologicaltools• JIL019(2;02)
8
Themoviehasbeendeletedfromthedistribu3onfile.
Drawbacksoflongitudinalspontaneouscorpora
• MacWhinney’s(2001)three‐headedmonsterofcorpustranscripAon:– Lackofstandardformat+rapidproliferaAonofalternaAveformats
– Indeterminancy• Difficulttodeterminewhatwasreallysaid/signed
– Tedium• Highlylabor‐intensive,conAnuallysubjecttorevisionandexpansion
9
CHILDES:ChildLanguageDataExchangeSystem
• Startedintheearly1980’sbyBrianMacWhinneyandCatherineSnow(withothers)
• Goal:tosharechildlanguagedata• Method:– Developcomputersomwareforstoringandsearching
– DesignconvenAonscompaAblewiththesomwareandteachtheseconvenAons
– Convinceresearchers(over100)todonatetheirdata
– Makethedatafreelyavailableontheinternet10
CHILDES–MainPoints
• Threemaincomponents:– CLAN–ComputerizedLanguageAnalysis
– CHAT–CodesfortheHumanAnalysisofTranscripts– Database(33languages)
• AddiAonalcomponents– Groundrules– Guidelinesforcontributors– …
11
CHILDES
hPp://childes.psy.cmu.edu/
• System
• ProgramsandDatabase
• Links• Manuals
• PhonologyandFonts• TeachingwithCHILDES• SpecialPopulaAons• MorphologyandLexicon• Mirrors
• Contact
12
BibibiProject SLCN3–Stockholm
15June2010 3
CHILDES‐Outcomes
• MajorchangeinmanyareasoflanguageacquisiAonresearch– QuanAtaAve,systemaAc,widerrange
• Over3000arAclespublishedbasedonCHILDESdata(asof2008)
• Over1millionhitstowebsite(early2010)
• ConAnuingaddiAonofdata,increasingtypes
13
SampleCHATtranscript@SituaAon:CHIislookingatapicturebookwithMOT*CHI: xxxfour.*CHI: Iseefour[/]Iseefouryyy.%pho: ki:kæ*CHI: one#two#three+…%act: poinAngtopicturebook*MOT: thosearebunnies.*MOT: whatis[//]whatarethebunniesdoing?*CHI: sleeping[?].%pho: sipi%com: Altsheadtooneside,couldbegestureforsleeping*CHI: it’sdarkout[=?darker]%pho: da:kou*MOT: yes,it’sAmeforbed.*CHI: goodnightbunnies!
14
Systemsfornota&onofchildsign
• Mostchildsignresearchersusevariantsofsystemsdevelopedforadultsigning– Baker,vandenBogaerdeandWoll(2005)discussmanyimportantgeneralconsideraAons
– Morgan(2005)–DynamicSpaceTranscripAon– Takkinen(2005)–HamNoSys
– Slobinetal.(2001)–BTS(BerkeleyTranscripAonSystem)
15
Ourgoals
• AdopAonofIDglossing(Johnston1991)– TranscripAonfocusisonannotaAngsignlemmas
• TranscripAoninELAN– TranscriptprovidesconsistentinformaAon,sufficientforcomputerizedsearching– BasictranscripAonavoidsanalysisasmuchaspossible– Analysisbyresearcherslater,usingtranscriptandvideo(unlikeCHILDES,whereanalysisalmostalwaysbasedontranscriptalone)
16
MLSSA:Mul&‐LanguageSignandSpeechAnnota&on
• ConvenAonalizednotaAonandprocedurescrucialformakingtri‐universitycollaboraAonpossible(Bibibiproject–BinaAonalBimodalBilingualstudyoflanguageacquisiAon)
• Specificallydesignedtoaccommodatebimodaldata– speechandsignareannotatedindependently– bimodalismisidenAfiedatanalysislevel
17
MLSSAProceduralconven&ons
• LabmanagertrainstranscribersandassignsandtracksprogressiveaddiAonstotranscripts,recordedinonlinelogsaccessibletoallprojectmembers
• Wetranscribespeechfirst,asitomenhelpsusidenAfyaccompanyingsigns
• Proofing• Coding/Analysis
18
BibibiProject SLCN3–Stockholm
15June2010 4
MLSSANota&onalconven&ons:ComparisonwithCHILDES
• MELISSAadoptsmanyCHILDESconvenAons,butwithslightmodificaAonsdueto:– RequirementsorcapabiliAesofELAN– ConvenAonsspecifictosignlanguageglossing
ASLuPerance: g(hey)SEEFOUR[/]SEEFOURYYY
FreetranslaAon: ‘hey…Iseefour,Iseefour[garbled]’ IX(book)#FOURDOG[?] ‘Therearefourdogsthere’ DOG[+]DV(sit‐in‐a‐line) ‘Thedogsarelyinginaline’
19
UseofMLSSAforresearch
• Sample:BEN_029(2;01),00:00:25–00:00:59• Transcribedandcodedfortwoprojects.
20
Exporta&onofanalysis&erstoExcel
21
CurrentandfutureresearchChenPichler,Deborah,Quadros,RoniceMüllerde,&Lillo‐MarAn,Diane(2009).EffectsofBimodalProducAon
onMulA‐CyclicityinEarlyASLandLibras.BostonUniversityConferenceonLanguageDevelopment.Boston,MA;November2009.InJaneChandlee,KaAeFranich,KateIserman,&LaurenKeil(Eds.),ASupplementtotheProceedingsofthe34thBostonUniversityConferenceonLanguageDevelopment,April2010.hPp://www.bu.edu/linguisAcs/BUCLD/supp34.html.
ChenPichler,Deborah,Hochgesang,Julie,Lillo‐MarAn,Diane&Quadros,Ronice(2010).ConvenAonsforSignandSpeechTranscripAoninChildBimodalBilingualCorpora.Language,Interac3onandAcquisi3on1.1,11‐40.SpecialissueguesteditedbyMarie‐AnneSallandreandMarionBlondel.
Lillo‐MarAn,Diane,ChenPichler,Deborah,&Quadros,RoniceMüllerde(2009).BestPracAcesforBuildingaBi‐modalBi‐LingualBi‐NaAonalCorpusofChildLanguage.WorkshoponSignLanguageCorpora:LinguisAcIssues;London,UK;July2009.
Lillo‐MarAn,Diane,Quadros,RoniceMüllerde,Koulidobrova,Helen&ChenPichler,Deborah(2009).BimodalBilingualCross‐LanguageInfluenceinUnexpectedDomains.GeneraAveApproachestoLanguageAcquisiAon;Lisbon,Portugal;September2009.[Proceedingstoappear;CambridgeScholarsPress]
Quadros,RoniceMüllerde,Lillo‐MarAn,Diane,&ChenPichler,Deborah(2010).TwoLanguagesButOneComputaAon:Code‐BlendinginBimodalBilingualDevelopment.TobepresentedattheconferenceonTheoreAcalIssuesinSignLanguageResearch;WestLafayePe,IN;October2010.
Quadros,RoniceMüllerde,Lillo‐MarAn,Diane,Koulidobrova,Helen,&ChenPichler,Deborah(inprogress).ConstraintsonCross‐LanguageInfluence,Code‐Switching,andCode‐Blending.
22
WorkscitedBaker,Anne,vandenBogaerde,Beppie&Woll,Bencie(2005).Methodsandproceduresinsign
languageacquisiAonstudies.SignLanguage&Linguis3cs8:1/2,7–58.Johnston,T.(1991).TranscripAonandGlossingofSignLanguageTexts:ExamplesfromAustralian
SignLanguage.Interna3onalJournalofSignLinguis3cs2:1,3–28.
MacWhinney,B.(2000).TheCHILDESProject:ToolsforAnalyzingTalk.3rdEdi3on.Mahwah,NJ:LawrenceErlbaumAssociates.
MacWhinney,Brian(2001).FromCHILDEStoTalkBank.InAlmgren,M.,Barreña,A.,Ezeizaberrena,M.,Idiazabal,I.,&MacWhinney,B.(Eds.),ResearchonChildLanguageAcquisi3on,17‐34.Somerville,MA:CascadillaPress.
Morgan,Gary(2005).TranscripAonofchildsignlanguage:AfocusonnarraAve.SignLanguage&Linguis3cs8:1/2,117–128.
Slobin,D.I.,HoiAng,N.,Anthony,M.,Biederman,Y.,Kuntze,M.,Lindert,R.,Pyers,J.,Thumann,H.,Weinberg,A.(2001).SignLanguageTranscripAonattheLevelofMeaningComponents:TheBerkeleyTranscripAonSystem(BTS).SignLanguage&Linguis3cs4,63–96.
Takkinen,Ritva(2005).SomeobservaAonsontheuseofHamNoSys(HamburgNotaAonSystemforSignLanguages)inthecontextofthephoneActranscripAonofchildren’ssigning.SignLanguage&Linguis3cs8:1/2,97–116.
23 24