The acquisition of language is a daunting task for infants. Research on this process provides an opportunity to address central issues related to the early development of human cognition. This application focuses on statistical language learning mechanisms that detect linguistic units by tracking input patterns of sounds. Recent research suggests that infant learners possess statistical learning mechanisms that may play an important role in language acquisition. Still unknown is the way in which these mechanisms interact to derive linguistic structure, given multiple possibilities. The proposed research addresses these issues by posing the following questions: (1) Can infants perform multiple analyses of the same set of input subsequently, so that the output of one set of statistical computations is the input to the next analysis, and simultaneously, such that multiple levels are processed at once? What factors determine which analysis emerges as the output of the learning process? (2) Do similar constraints on learning, with respect to acoustic and statistical structure, govern statistical language learning by adults, children, and infants? (3) Are the statistical learning mechanisms investigated in (1)-(2) tailored specifically for language learning, or can they operate on stimuli drawn from other domains? These questions will be addressed using previously developed laboratory learning paradigms which permit careful manipulation of the input. The answers to these questions will inform an emerging theoretical framework, constrained statistical learning, intended to elucidate the study of language acquisition and other pressing issues in human learning and development.
Showing the most recent 10 out of 31 publications
