Abstract:

Statistical learning (SL), which was first addressed in the seminal study on speech segmentation of infants by Saffran et al. (1996), is a process of detecting the statistical regularities such as transitional probability in continuous flow of stimuli. Previous studies have proven the general existence of SL, and in recent years close attention has been placed on its specificity and its impact on other cognitive activities, especially revealing the cognitive neural mechanisms of SL and its interaction with language by exploring the process and the specificity of SL. According to the multimodal data from brain and behavior measures, future studies should seek more behavioral and neural indexes to evaluate the performance of SL, to explore the dynamic changes in neural activities of different types of SL and to construct the connection between neural correlates and behavioral performance, which will help to have an in-depth understanding of SL. Based on previous discoveries on the interaction between SL and language, future studies could determine whether SL is an effective intervention to improve language acquisition and how it works in the improvement, through exploring the effect of music SL training on second language learning of adult learners.

Key words: statistical learning, transitional probability, specificity, language, multimodal data