Harnessing Implicit Learning to Support the Discovery of Second Language Phoneme Patterns in Adult Learners

Statistical learning studies have focused primarily on artificial languages, despite having an overall goal of providing insight into how we learn natural languages. Here, we investigate the impact of statistical learning on processing phonemic regularities within a second language in adult learners. Participants passively listened to Italian (L2 exposure group) or English (control group) podcasts for three weeks and completed a word rating task with ERP recorded before and after this listening period. Behaviourally, the L2 exposure group showed a nonsignificant trend towards increased sensitivity to phonotactic probability over the three weeks. At the ERP level, only the L2 group showed a significant change in nonword processing from session 1 to session 2, with significantly greater ERP negativity from 300-800ms to nonwords compared to words at session 2. These results provide preliminary indications that statistical learning may be leveraged to learn phonemic regularities in natural language, extending artificial language research.