Distributional learning of vowel categories in infants and adults

Distributional learning is learning from simple exposure to the environment, without receiving explicit instruction or feedback. This thesis examines to what extent this basic form of learning contributes to learning the vowels of a language, both in infancy, when the mother tongue must be acquired, and in adulthood, when new languages can be learned. The results are based on neurophysiological (i.e., ERP or event-related potential) and behavioural experiments, and on an extensive literature review of possible neural correlates. The dissertation provides a number of insights: (1) 2-to-3-month old infants already learn from a distributional training. This means that distributional learning can contribute to the development of language-specific speech perception, which infants start showing from 6 months of life onwards; (2) Adults can also learn from a distributional training. They probably learn to use increasingly subtle cues to identify the vowels of the new language. However, the capacity for distributional learning is smaller than in infants; (3) The observed distributional training effects (both those in this dissertation and those obtained in other studies) may not be due to the number of peaks in the training distributions. Future research should determine whether a prominent other candidate factor, namely the dispersion of the distributions, is responsible for the effects instead.