Doctoral Dissertation Research:Do individual differences in statistical learning predict linguistic adaptation?

George Mason University, Fairfax VA

Investigators

Abstract

A rich body of research has shown that statistical properties of language, like word or sound frequencies and co-occurrences, provide crucial information used by children to learn their native language. At the same time, there is evidence that adults' language use is influenced by expectations - e.g., about likely upcoming sounds, words, structures - which are dynamic and can be rapidly updated based on the current linguistic environment. This project investigates the idea that these abilities in children and adults reflect the same underlying mechanism: implicit tracking of statistical information. This single mechanism would then be used both during language learning, and continually across the lifespan to adapt to new situations or speakers. This research aims to provide evidence for this connection by showing that an individual's ability to track statistical information when learning a new language is correlated with their ability to adapt their own language under changing conditions. This project will focus first on learning and adaptation of higher-level linguistic structures. The study will test whether an individual's performance on a classic statistical learning task - implicitly learning that the presence of one syllable depends on the presence of another - predicts their ability to rapidly adapt expectations regarding upcoming syntactic structures (e.g., whether a verb is likely to act as a main verb or introduce a relative clause). The study will then look for evidence of the generality of the mechanism in question by testing whether statistical learning of higher-level structures predicts adaptation of lower-level sound category information. The results of the project will have implications for second language research and teaching; if individual differences in statistical learning and adaptation reflect a mechanism active across the lifespan, the kinds of tasks used here could be developed as diagnostic tools for assessment of language ability, and how it changes with age. This project also involves developing new open-source web-based tools for conducting cognitive psychology and psycholinguistics research more quickly and with fewer resources.