発達言語障害のある子供とない子供における語彙プライミングと反応時間に関するベイジアンメタアナリシスの検討

Children with developmental language disorder (DLD) exhibit linguistic difficulties that may stem from deficits in predictive processing. Several studies report impaired lexical priming in DLD groups compared to age-matched groups with typical language development (TLD). However, other studies report compensatory enhancement of lexical-semantic processing, leading to larger priming effects in DLD.

We used Bayesian meta-analysis to examine lexical priming and overall reaction time (RT) differences in age-matched groups of children with DLD and TLD (k = 19, m = 56, N = 360 DLD, N = 369 TLD; M_age [DLD] = 9.1 years). Moderators were priming type (semantic, form based) and response type (button press, verbal). Results are reported with outliers removed.

Both DLD and TLD groups exhibited lexical priming effects (faster RTs for primed targets; Hedges's g = 0.22 and 0.33, respectively), with similar between-study heterogeneity. There were trivial differences in the effects for semantic versus form-based (repetition/phonological) priming and for tasks requiring button press versus verbal responses. While group differences in priming effects were negligible (slightly larger priming effects in TLD groups, g = 0.11), group differences in overall RTs were of medium magnitude (shorter RTs in TLD groups, g = 0.56).

Children with DLD exhibit slower processing than age-matched peers with TLD but do not show impaired lexical priming. Discrepancies with prior findings may result from small samples and longer RTs contributing to nonsignificant priming effects in DLD groups. By de-emphasizing significance testing, Bayesian meta-analysis offers a more robust framework than frequentist meta-analysis for clinical research with small samples.