Recent work by Brown (2000) argues that difficulty in acquiring novel phoneme contrasts in a second language (L2) is due to the feature inventory of the first language (L1), and not the L1 phonemic inventory. This paper presents a case study examining the acquisition of length contrasts in Japanese by a native speaker of English. We examine production data, looking at the duration of geminate and single consonants and long and short vowels, as well as the spectral quality of long and short vowels. We find that a significant length contrast was produced for all consonants and vowels (as indicated by two-tailed t tests, where p < 0.001), although the length contrast was not implemented in a native-like fashion. Furthermore, the spectral analysis indicates that although the subject is not on the whole substituting English vowels into her Japanese, she did maintain a contrast between long and short vowels in terms of vowel quality. These results support the view that we can differentiate between the phonological representation and phonetic implementation for any given segment. The results also indicate that L1 features can be re-deployed to build appropriate phonological representations for novel L2 segments.
Proceedings of the 6th Generative Approaches to Second Language Acquisition Conference (GASLA 2002): L2 Links
edited by Juana M. Liceras, Helmut Zobl, and Helen Goodluck
Table of contents