Learning to discriminate English /r/ and /l/ in adulthood: Behavioral and modeling studies

I describe a body of work undertaken to explore the effect of experience on the perception of speech sounds. The work is undertaken within the context of my overall theoretical perspective, in which language is viewed as reflecting the influence of graded rather than categorical constraints and in which experience gradually shapes the way we respond to the sounds we experience, including sounds in our native language and sounds in foreign languages. Focusing on the case of Japanese adults learning the contrast between English /r/ and /l/, I present a neural network simulation model that captures how experience shapes perception during native language exposure during childhood. I go on to describe predictions from the model about the effects of different training regimes on learning the English /r/-/l/ contrast after acquiring natural Japanese spoken language categories. Then, I present the results of several experiments, partially supporting the model’s predictions but also demonstrating some limitations Japanese adults face in learning the English /r/-/l/ contrast that are not fully captured by the model. Overall, the work suggests that perceptual change is possible in adulthood, with some limitations that remain to be addressed in future work.