9/30/2014. Outline LANGUAGE ASSESSMENT IN BILINGUAL CHILDREN: CHALLENGES AND OPTIONS. Challenges: A Dynamic View of Language

Transcription

1 2 Outline LANGUAGE ASSESSMENT IN BILINGUAL CHILDREN: CHALLENGES AND OPTIONS Maria Kapantzoglou, Ph.D., CCC-SLP Portland State University Challenges When to Assess How to Assess Languages Measures Norm Referenced Criterion Referenced Language Sample Analysis Dynamic Assessment 3 4 Challenges: A Dynamic View of Language Challenges: Misclassification in Bilinguals* Language Level (y) L1 L2 Under-identification Low referral rate Deficits attributed to learning a 2 nd language Early intervention is delayed Over-identification Over-representation in Special Ed. Low diagnostic accuracy Inadequate referrals Lack of bilingual staff Financial, social-emotional, and educational costs Time (x) * Restrepo & Silverman, 2001; Goldstein, 2004; Sampson & Lesaux, Challenges: Language Proficiency vs. Language Abilities Possible scenarios Diagnosis? Low English High Spanish High English Low Spanish High English High Spanish Low English Low Spanish Challenges: Which Language(s) to Assess Ruling in language impairment Interpreters Trained interpreters Other professionals Community members Family members Tips for working with interpreters ASHA rpret.htm 1

2 Challenges: Comparing Monolinguals with Bilinguals Monolinguals L1 Bilinguals L1 & L2 Monolinguals L2 7 Spanish of Monolinguals with TLD and PLI: Grammatical Errors* Subtest TD Mean (SD) Articles 85.60% (10.10) Clitics 87.80% (10.90) Subjunctive 86.70% (12.30) Derivationa l Morphemes 73.30% (12.30) *Morgan, Restrepo, & Auza, 2009 SLI Mean (SD) 54.40% (29.30) 41.10% (26.20) 40.00% (34.60) 47.80% (21.70) 8 Partial η².30*.60**.48*.37* Spanish of Monolinguals with TLD and PLI: Grammatical Errors* 9 Spanish of Monolinguals vs. Bilinguals with TLD and PLI * 10 Same grammatical structures sensitive to PLI No differences between TLD bilinguals and monolinguals with PLI 60% of bilinguals with TLD misdiagnosed when using cut scores from monolingual sample *Morgan, Restrepo, & Auza, 2009 *Morgan, Restrepo, & Auza, 2012 English of Bilinguals vs. English of Monolinguals with PLI* English as a second language Tense marking difficulties Third person singular verb Use of copula and auxiliary difficulties More omission than commission errors Amount of exposure to English? 11 When is Action Needed? * When What Teacher concern Watch and See ESL teacher concern Family concern Assess Risk factors 12 Do TLD bilinguals look like monolinguals with PLI? Assessed in English subtractive environments *Paradis, 2005; Morgan, Restrepo, & Auza, 2012 *Kohnert, 2013; Paul,

3 13 14 How To Assess: Measures for identifying PLI in Bilinguals* Standardized norm-referenced tests E.g. CELF- 4 Spanish, PLS-5 Spanish Language sample analysis E.g., Grammaticality, lexical diversity Dynamic assessment E.g., Word learning, narratives Cognitive / processing measures E.g., Non-word repetition, rapid automatic naming *(Guttierez-Clellen & Simon-Cereijido, 2009; Kapantzoglou et al., 2010; 2012; Kohnert, Windsor & Pham, 2009 ; Restrepo, 1998; Restrepo and Silverman, 2001; Windsor et al., 2010; Wiig et al., 2006) How To Assess: Looking for Converging Evidence Dynamic Assessment Parent Interview Informed Decision Cognitive/Language Based Processing Language Samples Standardized Normed Referenced tests Measures Directory of Speech-Language Pathology Assessment Instruments When to use them Sufficient validity evidence Participant matches the standardization sample Satisfactory sensitive and specificity Recent norms Does my participant fit the characteristics of the norming sample? What if s/he doesn t? 1. Are test sensitivity and specificity satisfactory? 1. Is the test recent enough? Does my participant fit the characteristics of the norming sample? Age Language Language proficiency Socio-economic level 17 What if my participant does not fit the characteristics of the standardization sample? Bias: factors unrelated to the target construct influence the results Sources of bias: subgroup, item characteristics, test administration, scoring, and/or score interpretation 18 Children with same level in target skills, should obtain same scores regardless of group membership 3

4 19 20 Observed Scores Group 1 Group 2 Observed Scores Group 1 Group 2 Underlying Abilities Underlying Abilities Figure 2. No over- or underestimation / no bias Figure 1. Constant over- or under- estimation of abilities for one of the groups What if the standardization sample includes some participants with the same characteristics as my client s? Middle Class Low SES Example: the standard score (SS) for a monolingual child from low SES background on a vocabulary test is 72. Would you say this score is within normal limits knowing that for middle class monolinguals the mean on the test is 100 and the SD is 15? Sample Characteristics Mean SS Middle Class Monoling. Eng % Middle Class Monoling % Low SES Monol. 80% Middle Class Monoling. 20% Low SES Monol. 60% Middle Class Monoling. 40% Low SES Monol % Low SES Monol What is the classification/diagnostic accuracy of the test? Sensitivity: % of children with LI classified as children with PLI Specificity: % of children with TD classified as children with TLD Did the authors use a rigorous diagnostic procedure for the a priori classification of the participants? 24 4

5 25 26 TD TD LI TD LI I don t know 3. Are the norms recent enough? Reduction in prediction using the old equation suggests a change in the sample and a new equation should be used equation Current Year s equation Does my participant fit the characteristics of the norming sample? What if s/he doesn t? 1. Are test sensitivity and specificity satisfactory? 27 Language Sample Analysis: Outline 1. Advantages 2. What is the classification accuracy? 3. What are some common measures? 4. Does it matter how I will elicit a language sample? Is the test recent enough? Language Sample Analysis* Natural testing environment Reflective of child s discourse Culturally appropriate Efficient * e.g., Restrepo, 1998; Gutierrez-Clellen, Restrepo, & Simon-Cereijido, Language Sample Analysis: Classification Accuracy Study Language(s) Ages Measures Classification Accuracy Klee et al., 2004 Cantonese 2-5 MLU, D, Age 98% Klee et al., 2007 Restrepo, 1998 Simon-Cereijido & Gutiérrez-Clellen, 2007 British- and American- English Spanish- English Spanish- English 2-4 MLU, D, Age 83% 5-7 GETU, Parent Report 4 MLUw, Ungram. Utter %, 100% 79%, 100% 5

6 31 32 Language Sample Analysis Why not always used? Not always required Practice needed for efficiency Structure of analysis not provided No absolute cut scores (?) Suggestions Start with small language samples Focus on most relevant estimates initially and then expand analysis; have developmental criteria handy Use as one piece of evidence in a comprehensive assessment Attend training sessions Measures Lexicon NDW = # different words (type) NTW = # total words (token) Type Token Ratio (TTR) = NDW/TNW Morphology Grammaticality Index = #Grammatical Errors / #CUs MLUw = #words / #CUs; Percentage correct for a grammatical structure = = (Total Correct) / (Total of correct + errors) Language Sample Analysis: Measures Syntax Subordination Index = #clauses / #CUs OR #clauses / #TUs Narratives Language Sample Analysis: Measures Example He jumping, because he is happy. He leave, and he cries. Utterances (CUs) 1 CU = Main + Subordinate Clause NTW NDW GE Clauses Phonology Total (. CUs) Language Sample Analysis: Measures Example Utterances (CUs) 1 CU = Main + Subordinate Clause NTW NDW GE Clauses Total (3 CUs) Type-Token Ratio (TTR) = MLUw = Subordination Index = (Un)Grammaticality Index = Language Sample Analysis: Measures Example Percentage Correct for Third Person Singular (-s) Utterances (CUs) He jumping, because he is happy He leave And he cries Total (3 CUs) Correct Error Type of Error Percentage correct = (Total Correct) / (Total of correct + errors) = 6

7 37 38 Language Sample Analysis: Elicitation Type* Books or picture sequences vs. personal narratives Story retell vs. spontaneous language samples Story (re)tell vs. play procedures Language Sample Analysis: Elicitation Type* Were there group differences in D and MLUw in predominately Spanish-speaking preschoolers with and without SLI? If so, were these differences influenced by language sample elicitation type? Spontaneous story telling Story-retell *Morris-Friehe & Sanger, 1992; Klecan-Aker et al., 1987; Kapantzoglou et al., 2010 Kapantzoglou, Fergadiotis, & Restrepo, Language Sample Analysis: Elicitation Type* Language Sample Analysis: Elicitation Type* Kapantzoglou, Fergadiotis, & Restrepo, 2010 Elicitation Type x Group Interaction: F(1, 36) =.39, p =.54, η 2 =.01 Main Effect for Group: F(1, 36) = 11.08, p <.01, η 2 =.24 Main Effect for Elicitation Type: F(1, 36) = 9.40, p <.01, η 2 =.21 Kapantzoglou, Fergadiotis, & Restrepo, 2010 Elicitation Type x Group Interaction: F(1, 38) = 18.79, p<.01, η 2 =.34 Main Effect for Elicitation Type: F(1,38) = 16.69, p<.01, η 2 =.32 Main Effect for Group: F(1, 38) = 1.20, p=.28, η 2 = Language Sample Analysis Elicitation Type* Can MLU and D be used to accurately classify predominantly Spanish-speaking children with and without SLI? Sensitivity = 79% Specificity = 79% Static vs. Dynamic Assessment* Dynamic Assessment Assesses potential for learning Includes assessment + intervention Static / Knowledge Based Tests Assess existing knowledge Include only assessment Assess domain-specific skills + modifiability Assess domain-specific skills *Kapantzoglou, Fergadiotis, & Restrepo, 2010 η 2 =.36, χ 2 (2, N=38) = 15.65, p <.01 *Vygotsky, 1978; Feuerstein, 1979; Campione,

8 43 44 Dynamic Assessment - Modifiability Dynamic Assessment Theoretical Background Child Gains How a child learns Derived from the theories of Vygotsky Social factor in cognitive processes Zone of actual development (ZAD) Actual developmental level Examiner effort & support Zone of proximal development (ZPD) Potential development under adult guidance or in collaboration with more capable peers *Feuerstein, 1979; Lidz, 1991, Peña, 1993 Vygosky, 1934/1962; Why Dynamic Assessment with Bilinguals? Minimizes previous language experience bias Possible outcomes from static assessment L1 high L2 high L1 high L2 low L1 low L2 high L1 low L2 low Classification Accuracy Study (Ukrainetz et al, 2000) (Peña el al, 2001) (Peña el al, 2006) Participants Native American kindergartners (15 TD, 8 WL) African-American Latino-American preschoolers (62 TD, 17 WL) Various cultural/linguistic backgrounds (67 TD, 14 WL) Language Area Assessed Word categorization Vocabulary knowledge (EOWPVT) Story retell Classification Accuracy Sensitivity 75% Specificity 87% Sensitivity 78% Specificity 95% Sensitivity 100% Specificity 100% Predictors Expressive* Receptive Modifiability** Pretest Posttest + Modifiability ** Pretests Posttest + Modifiability ** Dynamic Assessment Example* How well can DA of word-learning skills differentiate predominately Spanish-speaking preschoolers with and without PLI? What Did We Teach? Fote (animal) Kina (toy) *Kapantzoglou, Restrepo & Thompson, Three Known Words Depa (food) Non-word Characteristics: CVCV Spanish phonological rules Early acquired sounds 8

9 49 50 Study Design Phase I Pretest + Play activity + Assessment (Prod. + Identif.) Phase II Play activity + Assessment (Prod. + Identif.) Phase III Play activity + Assessment (Prod. + Identif.) How Did We Teach? Supports for Learning Phonological and semantic support in each Phase 9 presentations 3 imitations Object manipulation Semantic category Object description Function + Iconic Gesture Modifiability How Did We Teach? Mediation Principles* Explicit statement of the purpose (mediation of meaning) & connection of the activities with children s previous experiences (mediation of transcendence) try to remember the names of these toys, the same way you remember the names of objects you have at home. What kinds of objects do you have at home? Do you have toys? What toys do you have? Do you have cars? Barbies? You know their names!... These are the names of your toys How Did We Teach? Mediation Principles* Explicit statement of the goals (mediation of intentionality) Now, we are going to play and learn some new words. Try to remember the names of the objects... Let s see if you remember the names of the objects We are going to play this game one more time to remember their names really well. Remember, you have to pay attention to the names of the objects. *Lidz, 1991 *Lidz, Classification Accuracy Discussion Points η 2 Sensitivity (PLI classified correctly) Specificity (TL classified correctly) Phase 1.39* 76.9% (10/13) 80% (12/15) Phase % (7/13) 80% (12/15) Phase % (7/13) 80% (12/15) Brief teaching sessions more effective for diagnostic purposes Best indicators of child s language skills Word identification Modifiability Which language areas should we assess? In which language should we assess? 9

10 55 Thank you! 10