Upload
others
View
7
Download
0
Embed Size (px)
Citation preview
ReadingComprehensionandInferences:Comparisonof
LearningDisabledandSecondLanguageSpeakers
LucieGodardUniversitéduQuébecàMontréal
Readingcomprehension! Kintsch et Van Dijck (1978) and Kintsch (1983) present
a model where, while reading, learners process information triggering the elaboration of mental representations essential for comprehension.
! Kintsch (1983) proposed two phases: construction and integration.
! For construction, readers choose information, activate schemas, and store pertinent information in the long-term memory.
! During integration, they prune non-essential information.
! Finally, readers build new networks for the information from the text being read.
ReadingmodelofKintschandVanDijckfromBlancetBrouillet(2003:70)
SurfaceForm
Textbase
SituaNonModel
Images
ReaderknowledgecontribuNonmicrostructure macrostructure
FROMTEXTNONTEXTUAL
Fletcher(1994);vanDijk&Kintch(1983);Zwaan&Radvansky(1998)
C-Imodel(Kinstsh,1998)
! 2levelsofrepresentaNon– TextBase– SituaNonmodel
! 2steps:ConstrucNon-IntegraNon
C-Imodel(Kintch,1998)
! Construction : Through schema activation, readers add information wich are not in the text.
! Integration: Readers desactivate inappropriate constructions by inhibiting irrelevant material and improving relevant elements.
! That is, during construction-integration, readers make inferences
Context:Genderdifferencesinreadingcomprehension
! Gapbetweenboysangirlsgetslargerasthegradelevelincreases– accordingtoalongitudinalevaluaNonofNaNonalAssessmentofEducaNonalProgram(Klecker,2006)
– accordingtoPan-CanadianAssessmentProgram
Differencesbetweengender
! AmongallcountryparNcipaNngofOECD,Pisa(2009)showedthatgirlsoutperformedboysinreadingliteracy.ItisthesameinCanada
! So,theCouncilofMinistersofEducaNon,CanadadecidedtosearchwhatcouldexplainthisdifferenceandChuyandNitulescu(2013)conductaresearchforthem.
! TheyusedPisa(2009)«datasettoinvesNgateandisolatethefactorscontribuNngtothegendergapinCanada»
ChuyandNitulescu(2013)research
• Pisa’sassessmentaskedstudentmanyquesNonsaboutstrategiestheyused
• Theyfound:• «Meta-cogni+vestrategies:girlsweremoreawareofthemosteffec+vemeta-cogni+onstrategiescomparedtoboys…….Thefemaleadvantagewaspar+cularlylargefortheindexofsummarizingstrategies»
• «ResultsofSimpleLinearRegressionModels:summarizingstrategywhichexplained16percentofthevaria+oninreadingscoresalone»1
1.AssessmentMaiers!No.5,2013–CMECp.7
ChuyandNitulescu(2013)research
! «ifCanadianboyswereasawareofeffec+vesummarizingstrategiesasgirlsare,theirreadingscorewouldincreaseby15points(seeOECD,2010b,tableIII.3.4)»
! ForexplainingthegendergaptheymadeOaxaca-BlinderdecomposiNon– Theyfoundenjoymentofreadingisimportantforexplainingthegap,butdoesgirlenjoybeierreadingbecausetheirreadingskillsarebeier?
ChuyandNitulescu(2013)research
! «Besidesenjoymentofreading,tworeadingstrategiesshowedsignificantandimportantcontribu+onstothegenderdifferencesinreading:controlandsummarizing.– Controlisacogni+vestrategyfocusingonunderstandingatask’spurposeanditsmainconcepts,
– Summarizingisameta-cogni+vestrategyreflec+nganawarenessofthemostefficientwaystocondenseinforma+on».1
1.AssessmentMaiers!No.5,2013–CMECp.7
Oakhill,CainetYuill(1998)
Goodreaders• GoodtextrepresentaNon• Longtermmemoryeasyaccess
Weakreaders• CogniNveoverload
TheoriNcalframework! READING COMPREHENSION
– Depends on linguistic knowledge related to: • Syntax • Text structure • Vocabulary - deep knowledge of a
word • Prior knowledge
! Wordknowledgeiscentralinthesystemsinvolvedinreadingcomprehension(PerfeqandStafura,2014)
! But, beyond these linguistics facts, are there other factors at play?
Knowledgeforaword(NaNon2001)
Underlayingknowledge
Form
speaking
wriNng
morphems
meaning
Formandmeaning
Concepts
AssociaNons
Use
Grammar
CollocaNons
Constraint
Andinferencing! InferencingisgoingbeyondtheexplicitinformaNoninthetext.
! Forinferencing,weshouldacNvateourpriorknowledge,linkinformaNon
Toinfer
! WemustfirsthaveagoodlexicalrepresentaNon(Perfeq2007)– PhonologicalrepresentaNon,– OrthographicrepresentaNon– SemanNcinformaNon
! Whilereading,lexicalrepresentaNonisacNvateddependingonthecontext– Whenreadingawellknownword=richsemanNcrepresentaNon
Context
! Poorreadersunderperformedinmakinginferencesinreadingcomprehensiontests(CainetOakhill1999).
! Inferencesarerelatedtopriorknowledge;themorebackgroundknowledgelinkedtothetextcontentthatanindividualhas,theeasieritisforhim/hertomakeinferences
! ANon-naNvelearners’lackofvocabularyaffectsreadingcomprehension
! Both,Non-NaNve(NN)andLearningDisabiliNes(LD)studentsdemonstratereadingcomprehensionproblems
! Thelinksbetweentheabilitytoinferandreadingcomprehensionweredemonstrated(YuilletOakhill1991,CainetOakhill1999)
L2andinferencing
! Weakreadershavedifficultymakinginferencestounderstandatext(CainetOakhill1999).
! ThequalityoftheL2lexicalrepresentaNoninfluencestheirabilitytomakeinferences(Cain2010)
Typesofinferencesnounanimityabouttherange
! Text-connec?ngorCoherenceinferences(Bowe-Crane&Snowling,2005)(Cain&Oakhill(1999)
! Gap-fillingorElaboraNvesinferences(Bowe-Crane&Snowling,2005)(Cain&Oakhill(1999)
! Localinferences:(coherence)(KyleneBeers,2003)! Globalinferences:covertthewholetext(KyleneBeers,
2003)
! Whilereading(online)(Grasseretal1994;Longetal1996)! Awerreading(offline)(Grasseretal1994;Longetal1996)
ResearchQuesNon
! AredifficulNesinmakinginferencesthesameforNNandLDstudents?
! DoesBoysandgirlshavesamescores?
Method! Students:(n=580)(grade3tograde6)aiendingFrench
schoolsinMontréalaera.– 386L1-NormalAchiever– 152L2-NormalAchiever– 42L1-LearningDisabiliNes
! Test– NarraNvetext– InferencesquesNons
! Procedure– GrouptaskintheirclassroomwithotherlinguisNctasks-textavalaibleforansweringquesNons.
B.Readingcomprehensiontask
! GroupnarraNvetextreading! InferencingquesNons
C.Othertasks
! C-Test! WriNngproducNon! Understandingsynonymstask! ComprehensionofidiomaNcphrasetask
CorrelaNonsbetweenreadingcomprehensionandothertasks
! Externvaliditywithothertasks:highcorrelaNonswithreadingcomprehension:– C-test(r=0,341atp=.000),– WriNngproducNon(r=0,443atp=.000),- Understandingsynonymstask(r=0,500atp=.000),- ComprehensionofidiomaNcphrasetask(r=0,450atp=.000);
Results
! NormalL1>NormalL2>LearningDisabiliNes• NormalL1andLD(F(1,427=39,975p=0,000)• NormalL1andNormalL2(F1,517=16,358p=0,00)
020406080100120140
3rd 4th 5th 6th
N-L1N-L2LD
ANOVAshowedsignificantdifferencesbetweengradesF(4,579)=44,643p=0,000
QuesNonsclassificaNon! AccordingtoCain&Oakhill(1999),wegrouped
inferencesin2types:– Gap-fillinginferences:integraNonofgeneralknowledgewithinformaNonprovidedinthetext.(BakerandStein’s1981terminology):Q1-Q2-Q3
– Text-connecNnginferences:mappinganinstanceofanounspecificnountoalaterspecificreferent.(BakerandStein’s1981terminology)Q5-Q6b-Q7b-Q8-Q9–Q10-Q11
! LiieralinformaNon– Answersisexplicitlyinthetext-Q4-Q6a-Q7a-Q12
! SummarizingabiliNes– FindaNtleforthetext-Q13
TextconnecNngbygrade-
F(3,579)=7,919p=0,000
TextConnecNng-L1(notclear)
F(1,579)=6,073p<0,014MeanL1=115,45andMeanL2=98,9
L1>L2
Text-connecNng-learningdisabiliNes(32subjects)
F(1,579)=12,958p=0,000
LD students:
Grade N
3 14
4 5
5 3
6 10
Fewstudents
N>LD
TextConnecNng-Gender
F(2,579)=6,855p<0,001
Girls>Boys
Gapfillingbygrade
F(3,579)=15,464p=0,000
GapfillingbyL1
F(1,579)=12,364p=0,000
L2>L1
DifferencesforLearningdisabiliNes
F(1,579)=25,233p=0,000
N>LD
Differencesbygender
F(2,579)=3,327p<0,037
Gilrs>Boys
Literalbygrade
F(3,579)=11,756p=0,000
LiteralL1
F(1,579)=9,706p<0,002
L2>L1
LiteralLearningDisabiliNes
F(1,579)=25,742p=0,000
N>LD
Literalbygender
F(1,579)=4,849p<0,008
Gilrs>Boys
Discussion
0
10
20
30
40
50
60
3rd 4th 5th 6th
text connectingliteralgap filling
TextconnecNng<literal<gapfilling
L1vsL2
0
10
20
30
40
50
60
3rdL1
3rdL2
4thL1
4thL2
5thL1
5thL2
6thL1
6thL2
text connectingliteralgap filling
Forboth:textconnecNng<literal<gapfilling
NormalachievervsLD
0
10
20
30
40
50
60
3rdN
3rdLD
4thN
4thLD
5thN
5thLD
6thN
6thLD
text connectingliteralgap filling
Ingrade3andgrade4LDaredifferentfromNormalachieverfewstudentscouldexplainthedifferences
GenderG=girlsB=boys
0
10
20
30
40
50
60
3rdG
3rdB
4thG
4thB
5thG
5thB
6thG
6thB
text connectingliteralgap filling
TextconnecNng<literal<gapfillingGilrs>Boys
Differencesbygender
0
20
40
60
80
100
120
140
3rd 4th 5th 6th Global
girls
boys
N=581
F(2,579)=7,366p<0,001
Discussion! Daneman1988andPerfeq1994proposelexicalknowledgeisrelatedtoreadingabilityintwodifferentways:– Richness– Speedofaccess
! SpeedofaccessseemstobemoreimportantwithourpopulaNon:L2>L1>LD
Discussion
Presentstudy! TextconnecNng<literal<gap
filling ! FantasynarraNve! Lessskilledreaderspooreratall
typesofinferencing! Poorcomprehendersalwaysthe
weakest
! Strategychoicemightbeplayingakeyrole
Cain&Oakhill(1999)! Gap-filling<literal<text-
connecNng! RealisNcnarraNve! Lessskilledreaderspooreratall
typesofinferencing! Poorcomprehenderswere
weakeringap-fillingthantheothergroups
! Strategychoicemightbeplayingakeyrole
Discussion
! Cain&Oakhill(1999)showedpoorcomprehendersimprovedtheirtext-connecNnginferenceswhentheycouldlookbackatthetext.Inthepresentstudy,asinYull&Oakhill(1988),evenwhenthetextpresentless-skilledcomprehenderswerepoorer.
! Bower-Crane&Snowling(2005)havedemonstratedthat“differentreadingteststapdifferenttypesofinferencingskills”
Discussion
! Ourfindingssuggestthatgenderdifferencesaresignificant
! OurfindingsaddtoexistantevidencethatboyshavemoredifficulNeswithreadingcomprehension(PISA,2009,PIRLS,2011)
ImplicaNons
! Training– Alltypesofinferencing:gap-filling,literal,text-connecNngwithdifferentkindofnarraNvetexts.
– SystemaNcworktobuildlexicalknowledgeinL2andLDchildren
! Research– NeedmoreexploraNonofdifferencesbetweengirlsandboys.
ImplicaNons
! Teach:– gapfilling1st,– literal2nd,– textconnecNng3rd
! LikeCainandOakhill(1999),weobserveddifficulNesinmakinginferencesforLD.Traininginmakinginferencescouldbehelpfullforthem.
Explicitteaching
! Explicitteachingofreadingcomprehensionstrategy:sizeeffect=1.18(Bissonneie,Richard&Gauthier2010)mega-analysis
Pearson et Gallagher (1983) The instruction or reading comprehension
Traininginferences! McGeeandJohnson(2003)didinferencetrainingbythesesteps:– Lexicaltraining:explainedmeaningofspecificwords– QuesNongeneraNon:pupilsgeneratedtheirownquesNonsfromthetext
– PredicNon:studentsguessedthemissingsegmentsbyinference
! Theyconcludethisisaready-madeinferencetrainingbecausetheyhadagreatsucces(student6-10yimproveover17monthsinNealeAnalysisofreadingAbilitytest)
Traininginferences
! Elbro&Iversen(2013)trained16Grade6classes,236parNcipantsfor8lessonsof30minutesintext-filling,explicitteaching:– 1.pre-filledorganizers– 2.whenstudentsfamilar,theyfilledinthebox– 3.last2sessions,theyreadthetextandansweredinferencesquesNonswithoutthesupportoforganizers
• Theyfound:1)theyimprovedabilitytomakegap-fillinginferences:effectsize0.92;• 2)Trainingwasassociatedwithasignificantadvanceinreadingcomprehension
Conclusion
! LimitaNons:– fewLDsubjects
! NormalL1>L2>LD– ThespeedofaccesstovocabularyforLDandthelackofstrategiescouldexplaindifferences,ascouldtheabilitytousegeneralknowledgetointerpretatext
! TextconnecNng<Literal<Gapfillingforallstudents– LDarepooreronalltypes
.
GivespecialaienNontoLD&L2boysteachtheminferencingandsummarizaNon
strategies
THANKSFORYOURATTENTION!
Références! [1]Kintsch,W.&VanDijk,T.A.(1978).Towardamodeloftextcomprehensionand
producNon.PsychologicalReview,85(5),363-394.! [2]Kintsch,W.(1988).TheRoleofKnowledgeinDiscourseComprehensionConstrucNon-IntegraNon
Model.PsychologicalReview,95,163-182.! [3]Perfeq,C.StafuraJ.(2014),WordKnowledgeinaTheoryofReadingComprehension,ScienNficStudy
ofReading,Routledge,vol18,no1,22-33.! [4]NaNonK.(2001),ReadingandLanguageinChildren:ExposinghiddendeficitsPsychologist.14:238-242.! [5]
Perfeq,C.A.(2007).ReadingAbility:LexicalQualitytoComprehension.ScienNficStudiesofReading,11(4),357-383.
! [6]Cain,K.Oakhill,J.V.(1999),InferenceMakingAbilityanditsRelaNontoComprehensionFailureinYoungChildren,ReadingandWriNng:AnInterdisciplinaryJournal11:489–503.
! [7]Cain,K.(2010),ReadingDevelopmentandDifficulNes,BriNshPsychologicalSocietyandBlackwellPublishing
! [8]Yuill,N.,&Oakhill,J.(1991).Children’sProblemsinTextComprehension:AnExperimentalInvesNgaNon.Cambridge:CambridgeUniversityPress.
! [9]Baker,L.Stein,N.(1981),TheDevelopmentofProseComprehensionSkills.InC.M.Santa&B.L.Hayes(eds.),Children’sProseComprehension:ResearchandPracNce;Newark,DE:InternaNonalReadingAssociaNon(pp.7–43).
! [10]Cain,K.Oakhill,J.(2014).ReadingComprehensionandVocabulary:IsVocabularymoreImportantforSomeAspectsofComprehension?.L’Annéepsychologique,114,pp647-662.
!
References[11]Klecker,B.M.(2006).TheGenderGapinNAEPFourth-,Eighth-,andTwelwh-GradeReadingScoresAcrossYears.ReadingImprovement,43(1),50-56.[12]OECD.(2010b).PISA2009Results:LearningtoLearn–StudentEngagement,StrategiesandPracNces(Vol.3).Paris:[13]Chuy,M.,&Nitulescu,R.(2013).PISA2009:ExplainingtheGenderGapinReadingthroughReadingEngagementandApproachestoLearning.Researchpaper.Toronto:CouncilofMinistersofEducaNon,Canada(CMEC)andHumanResourcesandSkillsDevelopmentCanada(HRSDC).[14]Yuill,N.M.etOakhill,J.V.(1988).EffectsofInferenceAwarenessTrainingonPoorReadingComprehension.AppliedCogniNvePsychology,2(1),33-45.[15]Bowyer-Crane,C.,&Snowling,M.J.(2005).AssessingChildren'sInferenceGeneraNon:WhatDoTestsofReadingComprehensionMeasure?.BriNshJournalofEducaNonalPsychology,75(Pt2),189-201.[16]Bissonneie,S.,Richard,M.,Gauthier,C.&Bouchard,C.(2010).Quellessontlesstratégiesd’enseignementefficacesfavorisantlesapprenNssagesfondamentauxauprèsdesélèvesendifficultédeniveauélémentaire?Résultatsd’uneméga-analyse.Revuederechercheappliquéesurl’apprenNssage,3(1),1-35.[17]McGee,A.Johnson,H.(2003)TheEffectofInferenceTrainingonSkilledandLessSkilledComprehenders,EducaNonalPsychology,Vol,23,No.1,2003.[18]Elbro,C.Buch-Iversen,I.(2013)AcNvaNonofBackgroundKnowledgeforInferenceMaking:EffectsonReadingComprehension.ScienNficStudiesofReading,17:435-452,2013.[19]Graham,S.Harris,K.(2010)InstrucNonandAssessmentforStrugglingWritersP.171-172andPerinetGraham2012inMoatetal2012).