| Summary: | 博士 === 國立清華大學 === 語言學研究所 === 103 === This study investigated how the interlanguage grammars work in the process of L2 prosodic acquisition by examining the L2 Japanese accent assignment by Taiwan Mandarin-speaking learners. We analyzed the effects of linguistic properties (phonological and morphological structures) on the L2 accent assignment by using monomorphemic and compound words. We also studied the difference in L2 accent patterns between the two groups of learners with more or less amount of learning experience.
The results revealed that both phonological and morphological structures affect the L2 accent pattern of Japanese words. Firstly, the findings regarding the influences of phonological structures are; (a) word length exerts effect on the accent assignment in that words with 4 or less syllables tend to be accented on penultimate syllable, while 5-syllable words prefer to put accents on antepenultimate syllable, (b) the role of syllable weight is not so salient as far as accent assignment is concerned, suggesting that Weight-to-Stress principle (Hayes 1995) is not strictly obeyed in L2 accent assignment by TM learners. Secondly, the effects of morphological structure on L2 accentuation were observed in the three points below: (a) N2-initial syllable attracts accents, (b) two morphemes are treated as two different prosodic words (one word has one accent, i.e, Culminativity), and (c) the same interlanguage grammar used in L2 accentuation of monomorphemic words applies in compound accentuation. Third, the (mis-)overgeneralization of L2 accent pattern appeared in the results of advanced learners, although the interlanguage grammar which determines the basic accent pattern remained the same as beginners.
Overall, the results of our study revealed that these accent patterns were determined integrately by the metrical parameters, phonological principles, and phonetic characteristics of the L1. Thus, it was confirmed that L1 transfer occurs even between the L1 and the L2 which are distinct in stress system (i.e., tone language and pitch-accent language). This result contradicts to Archibald’s (1997) claim that L1 transfer of metrical parameters does not occur among the L1 and the L2 which are typologically distinct in regard with stress system. Moreover, the developmental patterns were formed by the (mis-)overgeneralization of accent patterns which frequently occur in Japanese both on the phonetic and phonological levels, resulting in the L2 accent patterns idiosyncratic to the learners.
|
|---|
