Upload
others
View
5
Download
0
Embed Size (px)
Citation preview
Differentiating Between English L2 Children with Typical and Impaired
Language Development
Johanne Paradis and Tamara Sorenson Duncan University of Alberta
Clinical Markers in English SLI Clinical marker
performance in domain differentiates TD from SLI Non-word repetition / phonological working memory Tense marking morphology Separate and genetic/inherited components for both Children with SLI acquiring English as L2 should also
show vulnerabilities on working memory and tense morpheme tasks
Bishop et al., 2006; Conti-Ramsden, 2003; Gathercole, 2006; Rice & Wexler, 1996; Rice, 2007;Tager-Flusberg & Copper, 1999
Tense Morphology in L2
Marker of SLI in simultaneous French-English and Spanish-English bilinguals
Past tense particular marker of SLI in Spanish-English sequential bilinguals (80 months of exposure)
Gutiérrez-Clellen et al. (2008); Jacobson & Schwartz (2005); Paradis et al. (2003); Paradis (2005); Paradis et al., (2008)
Tense Morphology in L2
Marker of SLI in simultaneous French-English and Spanish-English bilinguals
Past tense particular marker of SLI in Spanish-English sequential bilinguals (80 months of exposure)
Overlap in difficulties with tense morphology between monolingual SLI and L2 English at earlier stages (10 months)
Could tense be a clinical marker in L2 at early stages?
Gutiérrez-Clellen et al. (2008); Jacobson & Schwartz (2005); Paradis et al. (2003); Paradis (2005); Paradis et al., (2008)
Non-Word Repetition in an L2
Phonological working memory = basic language learning mechanism
Basic functional mechanisms less sensitive to incomplete acquisition?
Girbau & Schwartz, 2008; Gutiérrez-Clellen et al. (2004, 2006); Kohnert & Windsor (2004)
Non-Word Repetition in an L2
Phonological working memory = basic language learning mechanism
Basic functional mechanisms less sensitive to incomplete acquisition?
Successful non-word repetition requires mastery of segmental and suprasegmental phonology in L2
Could phonological working memory be a clinical marker in L2 at early stages?
Girbau & Schwartz, 2008; Gutiérrez-Clellen et al. (2004, 2006); Kohnert & Windsor (2004)
Mother Tongue
English
French
Non-official languages
Linguistic Diversity among Canadian Children
Children 0-14 – mother tongue – Statistics Canada 2006
Spanish
Arabic Punjabi
Urdu
Cantonese
Mandarin
Non-Official Languages
Linguistic Diversity among English L2 Children
Children 0-14 years – language spoken most often at home – Statistics Canada 2006
Importance of L2 Clinical Markers Best practice for identification of language impairment in
bilinguals: Assessment in both languages (ASHA and CASLPA)
Direct examination of L1 of English L2 children often impossible in diverse contexts
English L2 children take up to 5 years to catch up to monolinguals Too long to wait for assessments
Uncovering potential clinical markers in early L2 English relevant to assessment practices in diverse contexts
Method
Participants
Children from newcomer families in Edmonton and Toronto, Canada
Children 5-7 years old Exposure to English: 1-3 years
Early sequential bilinguals L1 exclusive or predominant language of home at birth
L1 backgrounds: Mandarin/Cantonese, Arabic, Urdu/Punjabi/Hindi, Spanish 4 Vietnamese & 1 Somali in SLI group
81 TD children and 24 children with SLI
Criteria for SLI
Referred by SLPs Receiving speech
therapy services and/or attending special kindergarten programs
Assessment by SLP required for services/programs
No autism No hearing impairment No acquired
neurological damage No severe intellectual
disability (nonverbal IQ above 80)
Not primary speech impairment
Measures CTOPP (Comprehensive Test of Phonological Processing, Wagner et al., 1999)
Phonological Memory Composite: non-word repetition and memory for digits
TEGI (Test of Early Grammatical Impairment, Rice & Wexler, 2001)
Elicited Grammar Composite: probes for 3rd Sing –s, PAST –ed, BE and DO
ALDeQ (Alberta Language and Development Questionnaire)
Early milestones / current L1 skills / behaviour / family history
ALEQ (Alberta Language Environment Questionnaire)
Maternal education, language use in the home, richness of English environment (contact with English via media and friends)
CMMS (Columbia Mental Maturity Scales, Burgemeister et al., 1972)
Non-verbal IQ
Participant Sample Characteristics
AGE MOE ALDeQ CMMS TD 71 24 .79 106 SLI 72 25 .48 98
p > .05 p > .05 p > .001 p > .01
Results
Between-Group Comparisons
CTOPP: TD > SLI (p < .001) d = .75
TEGI: TD > SLI (p < .001) d = 1.05
Regression Model: CTOPP
Coefficients Estimate Standard error t-value
Intercept 69.097 5.752 12.012*** TD or SLI -8.931 -2.860 -3.123**
IQ 0.479 0.158 3.028** TEGI 10.011 4.420 2.264*
R2 = .31, F (3,92) = 13.76**
Did not improve fit of model: AGE, MOE, Home Language, English Richness, L1
Regression Model: CTOPP
Coefficients Estimate Standard error t-value
Intercept 69.097 5.752 12.012*** TD or SLI -8.931 -2.860 -3.123**
IQ 0.479 0.158 3.028** TEGI 10.011 4.420 2.264*
R2 = .31, F (3,92) = 13.76**
Did not improve fit of model: AGE, MOE, Home Language, English Richness, L1
Regression Model: CTOPP
Coefficients Estimate Standard error t-value
Intercept 69.097 5.752 12.012*** TD or SLI -8.931 -2.860 -3.123**
IQ 0.479 0.158 3.028** TEGI 10.011 4.420 2.264*
R2 = .31, F (3,92) = 13.76**
Did not improve fit of model: AGE, MOE, Home Language, English Richness, L1
Regression Model: CTOPP
Coefficients Estimate Standard error t-value
Intercept 69.097 5.752 12.012*** TD or SLI -8.931 -2.860 -3.123**
IQ 0.479 0.158 3.028** TEGI 10.011 4.420 2.264*
R2 = .31, F (3,92) = 13.76**
Did not improve fit of model: AGE, MOE, Home Language, English Richness, L1
Regression Model: TEGI
Coefficients Estimate Standard error t-value
Intercept -1.4139 0.4033 -3.506*** TD or SLI -0.1983 0.0824 -2.407*
IQ 0.0119 0.0046 2.581* AGE 0.0144 0.0046 3.538***
ENG RICH 0.0088 0.0030 2.974** HOME LANG 0.1005 0.0385 2.612*
R2 = .35, F(5,90) = 9.649**
Did not improve fit of model: MOE, CTOPP, L1
Regression Model: TEGI
Coefficients Estimate Standard error t-value
Intercept -1.4139 0.4033 -3.506*** TD or SLI -0.1983 0.0824 -2.407*
IQ 0.0119 0.0046 2.581* AGE 0.0144 0.0046 3.538***
ENG RICH 0.0088 0.0030 2.974** HOME LANG 0.1005 0.0385 2.612*
R2 = .35, F(5,90) = 9.649**
Did not improve fit of model: MOE, CTOPP, L1
Regression Model: TEGI
Coefficients Estimate Standard error t-value
Intercept -1.4139 0.4033 -3.506*** TD or SLI -0.1983 0.0824 -2.407*
IQ 0.0119 0.0046 2.581* AGE 0.0144 0.0046 3.538***
ENG RICH 0.0088 0.0030 2.974** HOME LANG 0.1005 0.0385 2.612*
R2 = .35, F(5,90) = 9.649**
Did not improve fit of model: MOE, CTOPP, L1
Regression Model: TEGI
Coefficients Estimate Standard error t-value
Intercept -1.4139 0.4033 -3.506*** TD or SLI -0.1983 0.0824 -2.407*
IQ 0.0119 0.0046 2.581* AGE 0.0144 0.0046 3.538***
ENG RICH 0.0088 0.0030 2.974** HOME LANG 0.1005 0.0385 2.612*
R2 = .35, F(5,90) = 9.649**
Did not improve fit of model: MOE, CTOPP, L1
Linear Discriminant Function Analysis
Model = ALDeQ (.950) + TEGI (.346) + CTOPP (.214)
X2 = 104.9, p < .001 Canonical correlation = .803
Specificity
TD SLI
Sensitivity
SLI TD
Linear Discriminant Function Analysis
Model = ALDeQ (.950) + TEGI (.346) + CTOPP (.214)
X2 = 104.9, p < .001 Canonical correlation = .803
Specificity
TD SLI
Sensitivity
SLI TD
Discussion
Summary Significant differences between TD and SLI groups for
CTOPP and TEGI at 2 years exposure to English CTOPP predictors internal (Group, IQ) and TEGI
TEGI association = phonology = cluster reduction?
TEGI predictors internal (Group, IQ, AGE) and external (home language and English Richness) Tense = more language specific, so external factors matter
more?
Combined with L1 development information (ALDeQ), CTOPP and TEGI show very good-to-excellent discrimination of SLI and TD
Conclusions Clinical markers in a language are the same if it is
acquired as an L1 or an L2 Even though input conditions for L2 acquisition can be different Similar findings for subj-verb agreement in Dutch L1 and L2
with SLI (Orgassa, 2009)
Differentiation between TD and SLI among L2 learners emerges as early as 2 years of exposure
But, L2 phonological working memory and tense accuracy not sufficient to discriminate SLI L1 development information needs to be included
Many Thanks to Kristyn Emmerzael, Ruiting
Jia, Dorothy Pawlina Pinto, Emily Yiu, Tatiana Zdorenko
Edmonton Public School Board and Toronto Catholic District School Board
Multicultural Health Brokers Cooperative
Alberta Heritage Foundation for Medical Research, Social Sciences and Humanities Research Council of Canada, and the Canadian Language and Literacy Research Network