Language acquisition by Japanese speakers: Explaining the why, how, and when of adult learners' segmental success

This work grapples with several important questions in the field of second language acquisition: Why are the results of adult language acquisition not as uniform as those of younger learners? How do underlying structures constrain and explain success in perceiving and producing second-language segments? When do adult learners succeed---what factors contribute to perception and production success? A comprehensive review of the Critical Period for phonology in the first chapter is followed by three studies which explore the reasons for differential success in perceiving and producing segments in an L2.;The first study examines predictions made by Brown's (1997) theory of phonological interference and Flege's (1995) Speech Learning Model in explaining the results of a perception study conducted with 30 Japanese learners of Russian in 3 skill levels. Eight Russian segments are examined: /∫, 3, l, r, ts, f, mj, pj/. Japanese learners of all levels poorly differentiated /r-l/ and /∫-c/ contrasts, while beginners also had problems with palatalized segments. The Featural Model as elaborated here (based on Brown) accurately predicted the problems encountered by learners, while Flege's theory could not. These results provide evidence that learners' underlying representations are constrained by the features contained in their L1 phonemic inventory.;A companion production study examined the same learners and sounds as the perception study. Accuracy scores were overall quite low (65--70%), even for native controls. However, Japanese performed more poorly than controls on the /r-l/, /∫-c/, and palatalized contrasts. I propose two hypotheses concerning Feature Geometry architecture and phonemic correspondence to account for improved performance by advanced learners.;Additionally, a Deliberate Articulators Hypothesis was proposed to account for discrepancies between perception and production abilities. This hypothesis states that production consists of underlying representations plus conscious articulation. Hierarchies of articulation difficulty are proposed.;The third study replicated Flege, Takagi & Mann (1995), who stated that experienced Japanese (EJ) speakers with a length of residence (LOR) of 12 years or more could perform at NS levels on /r-l/ production tasks. The present study failed to replicate these findings, in fact finding that LOR correlated negatively with production success.