The Acquisition of L2 Speech Sounds in Early Language Learning : Linguistic and Cognitive-Psychological Aspects

本研究从语音学和心理语言学的角度探讨了沉浸式学习中第二语言（L2）语音的习得，并对Flege（1995）提出的语音学习模型（SLM）进行了验证。具体来说，该研究考察了在日语沉浸式课程中，许多内容科目都是用日语教授的，说英语的儿童是否能够成功地获得分段计时(即：日语的语音开始时间（VOT）和关闭时间（单胎和双胎），并且在退出程序后即使L2输入减少也能保持这种能力。本研究收集了15名日语浸入式课程的儿童、5名浸入式课程的教师、10名日语单语者、10名英语单语者和10名小学退出日语浸入式课程的英语大学生的数据（N=50）。被调查者被要求在一个句子框架中重复几个目标词，其中首字母为/p、t、k/，首字母为/p、t、k/，中间字母为/pp、tt、kk/。测量初始停搏点的VOT和内侧停搏点闭合时间。结果表明：(1)浸没型儿童的单音和双音虽然明显长于日语单语儿童，但获得了对比；(2)浸没型儿童的日语无音顿音的VOT值明显长于日语单语儿童，但其VOT值明显短于英语单语儿童。研究结果表明：3)双语儿童可能具有不同于母语者的语音类别，这可能是二语者区分母语和二语相似语音的心理策略，这将支持Flege的SLM理论。此外，虽然结果不是决定性的，但浸入式课程的毕业生在退出课程后似乎保留了他们的语音能力。

This study investigates the acquisition of second language(L2) speech in an immersion program from both phonetic and psycholinguistic points of view, and also tests the Speech Learning Model(SLM) hypothesized by Flege(1995). Specifically, the study examines whether or not English-speaking children in a Japanese immersion program, in which many content subjects are taught in Japanese, can successfully acquire segmental timing(i.e., voice onset time(VOT) and closure duration fur singletons and geminates) in Japanese, and can retain this ability despite the decrease of L2 input after they have exited the program. Data were collected from 15 children in a Japanese immersion program, 5 immersion teachers in the program, 10 Japanese monolinguals, 10 English monolinguals, and 10 English-speaking university students who exited a Japanese immersion program in elementary school(N=50). The informants were asked to repeat several target words with initial/p, t, k/for VOT, and medial/p, t, k/ and/pp, tt, kk/ for singletons and geminates in a sentence frame. Both VOT of the initial stops and closure duration of the medial stops were measured. The results show that 1) although both their singletons and geminates were significantly longer than those of the Japanese monolinguals, the immersion children acquired the contrast, and that 2) although they produced Japanese voiceless stops with significantly longer VOT values than the monolingual Japanese children, they produced them with significantly shorter VOT values than their English VOT. The findings suggest that 3) the bilingual children may have phonetic categories different from Ll speakers', which can be a L2 speaker's psychological strategy for differentiating between similar sounds in Ll and L2, and this will support Flege's SLM. Also, though the results are not conclusive, the immersion graduates seem to have retained their phonetic ability after they exited the program.