Background: In recent research, verbal working memory has been proposed as a primary area of deficit for children with language impairment (LI), and therefore a source of more sensitive assessment measures. In addition, research on non-linguistic tasks has suggested that children with LI may have deficits that extend beyond the linguistic domain. These dual lines of research can be connected to the drive for non-biased assessment tasks for children with diverse language learning experiences (such as bilingual children), because linguistic tasks may be intrinsically biased against such children.

Aims: To investigate the usefulness of a working-memory task with a minimal linguistic load as a potential screening tool for children with primary LI.

Methods & Procedures: Three groups of children aged 7;10-13;11 participated in the study: monolingual English-speaking children with LI, typical monolingual English-speaking children, and typical Spanish-English bilingual children with several years of English experience. Performance of the three groups on the Counting Span task was explored through group comparisons and likelihood ratios.

Outcomes & Results: Children with LI had significantly lower Counting Span Set score accuracy than either typical group, with the monolingual and bilingual groups showing equivalent task performance. However, results from likelihood ratios showed that Counting Span accuracy did not have compelling diagnostic power when the entire age group was considered. Children with less than 67% accuracy were only three times more likely to be from the LI group than from either of the two typical groups. In addition to language status (typical or LI), chronological age was a key contributor to Counting Span performance in this age group. Likelihood ratios for children aged 10;0 and above showed that children with less than 67% accuracy were nine times more likely to come from the LI group than the typical groups.

Conclusions: This intermediate diagnostic ability suggests that performance on the Counting Span alone is insufficient to identify LI among linguistically diverse learners. However, Counting Span may be a very useful component of a larger LI assessment battery, particularly for children over the age of 10;0.