The Use of Dynamic Assessment for the Diagnosis of Language Disorders in Bilingual Children: A Meta-Analysis
American Journal of Speech-Language Pathology
Orellana, C. I., Wada, R., et al. (2019).
American Journal of Speech-Language Pathology, 28(3), 1298-1317.
This meta-analysis investigates the diagnostic accuracy of dynamic assessment to determine the presence of a language disorder in bilingual children.
Not stated
Up to August 2017
Randomized controlled trials; quasi-experimental designs; single-subject experimental designs
6
<div>All included studies investigated dynamic assessment (DA) that consisted of a test-teach-retest format. Results indicated two key patterns regarding the effectiveness of DA in determining a language impairment (LI) from a language difference/typically-developing (TD) language in bilingual children:</div>
<ul>
<li>"Our review revealed that bilingual children with LIs performed significantly more poorly than their TD controls on the language measures at both pretest and post-test and that comparisons of group differences on language change from pretest to post-test were inconclusive" (pp. 1314-1315).</li>
<li>"In three of the five studies that examined modifiability [i.e., the child’s ability to benefit from instruction], TD children earned significantly higher scores on clinician ratings of modifiability than children with an LI" (p. 1315). Group differences, however, were not significant across studies.</li>
</ul>
<div>Results indicated the diagnostic accuracy for dynamic assessment (DA) to be suggestive due to a lack of gold standard for evaluating bilingual children for a language disorder. DA was indicated to be more accurate in distinguishing children with typically-developing language (i.e., specificity) than in distinguishing children with a language disorder (i.e., sensitivity). DA’s high specificity ranged from 0.80 to 0.96 and sensitivity ranged from 0.77 to 1.00. "The high diagnostic value of 7.34 was below the optimal LR+ of 10, suggesting that there was approximately 45% probability that a child with a positive score has a disorder compared to 50% probability if the LR+ value was 10 (the clinically informative cutoff)" (p. 1315).</div>