[PS-2.12] The (missing) link between statistical learning and expressive grammar knowledge: A meta-analysis

Lammertink, I. 1 , Collée, E. 1 , Boersma, P. 1 , Wijnen, F. 2 & Rispens, J. 1

1 University of Amsterdam, Amsterdam Center for Language and Communication
2 Utrecht University, Utrecht Institute of Linguistics OTS

While learning language, children unconsciously detect and process structural regularities that reflect (morpho)syntactic rules of language. This mechanism is referred to as statistical learning (SL). A link between SL and children?s knowledge of (morpho)syntactic rules (i.e. grammar) has been proposed. While intuitively such a link may be evident, most studies that investigate this relationship report on small (or nonexistent) correlations between SL and grammatical knowledge in children with and without Developmental Language Disorder (DLD). This observation, together with the recently raised concern on the existence of a publication bias in the SL literature, motivated us to conduct a meta-analysis. A qualitative summary over 16 unique effect sizes, that collectively examined the correlation between serial reaction time task performance and expressive grammatical knowledge (e.g. sentence recall and sentence formulation) in 160 children with DLD and 254 typically developing children between 6 and 12 years, provides no evidence for (or against) such link (Fischer z = 0.0124, 95% CI [-0.0900; 0.115], p = 0.54). Nor does it provide evidence that the strength of the relationship differs (or does not differ) between children with and without DLD (p = 0.51). We will discuss the theoretical, methodological and clinical implications of this outcome.