Reducing Bias in Language Assessment

Abstract

One potential solution to the problem of eliminating bias in language assessment is to identify valid measures that are not affected by subjects' prior knowledge or experience. In this study, 156 randomly selected school-age boys (31% majority; 69% minority) participated in three "processing-dependent" language measures, designed to minimize the contributions of prior knowledge on performance, and one traditional "knowledge-dependent" language test. As expected, minority subjects obtained significantly lower scores than majority participants on the knowledge-dependent test, but the groups did not differ on any of the processingdependent measures. These results suggest that processing-dependent measures hold considerable promise for distinguishing between children with language disorders, whose poor language performance reflects fundamental psycholinguistic deficits, and children with language differences attributable to differing experiential backgrounds.