Download: Henderson et al. (2018)
This study investigates the use of narrative dynamic assessment in decreasing overidentification of students from American Indian communities as language impaired.
Background: Previous research has shown the overidentification of American Indian (AI) populations as having language impairment (LI), with AI being the largest cultural group in special education within the U.S. Overidentification may be due to cultural differences, such as AI practices of children listening to, rather than rehearsing, narratives. Linguistic differences observed in AI communities, such as the use of nonlinear narratives and gender-neutral pronouns, may also contribute to this. The CELF-4 and other norm-referenced, standardized measures are no valid assessments of language in AI populations. Two previous studies indicate that dynamic assessment (DA) of narratives may be a valid and efficient method for assessing language in AI children.
Purpose: The authors of the current study investigated the accuracy of the Predictive Early Assessment of Reading and Language (PEARL) pretest and posttest cut off scores in classifying Navajo preschool children as LI or typically developing (TD). Specific focus was placed on whether different subtests, especially the modifiability score (a measure of learning behaviors), produced improved sensitivity and specificity.
Methodology and Participants: Ninety Navajo preschoolers, ranging from 4;0 to 5;11 years old, participated in the study. The children were identified as LI or TD, with 45 children in both groups, based on teacher report, parent report, presence or absence of an Individualized Education Program (IEP), MLU calculated from a narrative sample, and performance on the CELF Preschool-2. The preschoolers participated in a PEARL pretest, mediation period, and posttest. A modifiability scale was used which measured the number of prompting required, child’s confidence, number of child disruptions, and rate of task execution.
Conclusion: As opposed to using the PEARL’s recommended pretest cutoff score of 10 or greater for TD, researchers found a pretest cutoff score of 6 or less for LI children and 7 or more for TD children was closest to meeting sensitivity and specificity, with 98% sensitivity and 73% specificity for this population of Navajo preschoolers. Meanwhile, a posttest cutoff score of 9 or greater for TD revealed sufficient sensitivity and specificity, with both meeting 89%. The authors concluded that cut off scores should be adjusted according to the population being assessed. As has been seen in other studies, modifiability scale scores were excellent in differentiating LI from TD children, with 100% accuracy. When considering single subtest accuracy, story grammar was the most effective in determining LI or TD (84% sensitivity, 87% specificity).
Relevance to the field: There is currently an urgent need to find valid assessments for minority populations, especially AI children. The result of using invalid and culturally inappropriate norm-referenced, standardized testing has resulted in a population that has been grossly overrepresented in special education in the U.S. Current standardized assessments are insufficient in accurately classifying AI children as LI or TD. Dynamic assessment (test-teach-retest) has shown promise in accurately identifying CLD children with LI. The PEARL, appears to be a valid and efficient assessment tool for this population with modified cut off scores and including a modifiability rating. Future research should continue to investigate DA of narratives and modifiability ratings with other CLD demographics.
Henderson, D. E., Restrepo, M. A., & Aiken, L. S. (2018). Dynamic Assessment of Narratives Among Navajo Preschoolers. Journal of Speech, Language, and Hearing Research, 61(10), 2547-2560. doi: 10.1044/2018_JSLHR-L-17-0313