The production and perception of English consonant sequences by Japanese‐speaking learners of English.

Abstract

This study reports the second part of a study investigating modifications of nonnative consonant sequences among Japanese speakers. A previous study (Sperbeck, 2009) that measured categorial discrimination demonstrated that some contrastive CC(C)V versus C\schwa\C(C)V sequences were hard for Japanese listeners (71% correct overall). Specifically, perceptual discrimination of voiced stops plus liquids contexts was poor for Japanese listeners (67% correct). The current study explored difficulties in production and how production correlated with perception. Nonsense words were constructed as the stimuli. They were of the form /CC(C)ani/ and /C\schwa/C(C)ani/, where CC(C) combinations were /sp, sk, pl, kl, bl, gl, spl, skl/. A delayed imitation task was used to assess production. Participants heard a native speaker's productions (e.g., Say blani now) twice, produced the target word in isolation (e.g., blani), and then produced it in the carrier sentence (e.g., I said blani now). Two phonetically trained native English speakers perceptually transcribed the productions. Results showed that the overall percent correct was 64% among Japanese speakers. There was a significant correlation between perception and production performance (rho = + 0.78. p < 0.01). However, the major error type was vowel deletion, rather than vowel epenthesis in producing the C\schwa\CV tokens.