Neural evidence for linguistic statistical learning is independent of rhythmic and cognitive abilities in neurotypical adults

Abstract

Statistical Learning (SL) is an essential mechanism for speech segmentation. Individual differences in SL ability are associated with language acquisition. For instance, better SL correlated with a larger vocabulary size and impaired SL was found in populations with language impairments. The aim of the current study was to contribute to uncovering the underpinnings of individual differences in auditory SL for word segmentation. We hypothesized that individuals with better musical – specifically rhythmic – abilities would show better SL. Participants (N = 106) were exposed to an artificial language consisting of trisyllabic nonsense words. Electroencephalography (EEG) measures of neural entrainment to the auditory signal allow online assessment of SL. The current study used this method to measure individual SL performance during exposure. To assess individual differences, we linked the neural measure of SL to a battery of tests measuring rhythmic, musical, and cognitive abilities, as well as vocabulary size. We replicated earlier work, finding both online (neural) and offline (behavioral) evidence of SL in our sample. In contrast to our expectations regarding individual differences, we found evidence for the null hypothesis regarding correlations between the tests of rhythmic ability and the neural measurement of SL. Exploratory analyses concerning working memory remained inconclusive, while exploratory analyses regarding vocabulary size yielded moderate evidence for a small correlation with the neural measure of SL. Overall, our results suggest that linguistic SL is largely independent from abilities in other cognitive domains, including rhythmic processing and musical abilities, as measured within a sample of healthy, typically developed adults.