The Performance on Relative Clauses in Children With and Without Developmental Language Disorder: A Meta-Analysis

This study attempts to quantify the magnitude of difference in relative clause (RC) performance (comprehension and production) between children with developmental language disorder (DLD) and their typically developing (TD) peers matched for age across nine languages by conducting a three-level meta-analysis. It takes sampling variance (Level 1), within-study variance (Level 2), and between-study variance (Level 3) into consideration.

A systematic literature search identified 25 studies investigating the performance on RCs in children with and without DLD. Data from identified studies were carefully selected and converted to Hedges's g effect sizes. Detection of outliers led to the removal of two extreme values, leaving 24 studies and 67 effect sizes for meta-analysis.

A significant overall combined effect size (g = -1.024, 95% CI [-1.350, -0.699], p < .001) was obtained, confirming the sensitivity of RCs in differentiating children with and without DLD. The moderator analysis provided no evidence that RC type or language impacted the overall effect size. However, it showed that task type was a significant moderator. Specifically, the difference in the performance on RC tasks between children with DLD and TD children was not statistically significant in discourse elicitation, syntactic priming, and sentence judgment tasks, while their performance on the sentence-picture matching task, the character-picture matching task, the preference elicitation task, the picture description task, and the sentence repetition task showed significance.

Across the included languages, children with DLD show markedly poorer performance than their TD peers on RC tasks for both subject and object RCs. Moderator analyses detected no significant variation by RC type or language but indicated that task type significantly modulates the magnitude of the DLD-TD difference.

https://doi.org/10.23641/asha.33977932.