Using Nonword Repetition to Identify Developmental Language Disorder in Monolingual and Bilingual Children: A Systematic Review and Meta-Analysis

Purpose: A wealth of studies has assessed the diagnostic value of the nonword repetition task (NWRT) for the detection of developmental language disorder (DLD) in the clinical context of speech and language therapy, first in monolingual children and, more recently, in bilingual children. This review article reviews this literature systematically and conducts a meta-analysis on the discriminative power of this type of task in both populations. Method: Three databases were used to select articles based on keyword combinations, which were then reviewed for relevance and methodological rigor based on internationally recognized checklists. From an initial pool of 488 studies, 46 studies were selected for inclusion in the systematic review, and 35 of these studies could be included in a meta-analysis. Results: Most of the articles report significant discrimination between children with and without DLD in both monolingual and bilingual contexts, and the meta-analysis shows a large mean effect size. Three factors (age of the child, linguistic status, and language specificity of the task) yielded enough quantitative data for further exploration. Subgroups analysis shows variance in effect sizes, but none of the three factors, neither their interactions, were significant in a metaregression. We discuss how other, less explored factors (e.g., nature of the stimuli, scoring methods) could also contribute to differences in results. Sensitivity and specificity analyses reported in 33 studies confirmed that, despite possible effect size differences, the diagnostic accuracy of the NWRT is generally near thresholds considered to be discriminatory. It generally increases when it is combined with other tasks (e.g., parental questionnaire). Conclusions: This review indicates that the NWRT is a promising diagnostic tool to identify children with DLD in monolingual and bilingual contexts with a large mean effect size. However, it seems necessary to choose the precise NWRT materials based on the children's language background and to complement the assessment sessions with other tools in order to ensure diagnosis and to obtain complete language profile of the child.