Production of Japanese Suprasegmentals by Taiwanese Learners: Effects of Language Proficiency on Durational Contrast and Fluency

• Main Authors: Wan-TingLai, 賴琬婷
• Other Authors: Hua-Li Jian
• Format: Others
• Language: en_US
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/37410293138281904686

碩士 === 國立成功大學 === 外國語文學系碩博士班 === 98 === Japanese had long been classified as mora-timed language (Grabe ＆ Low, 2002; Ramus, Nespor ＆ Mehler, 1999). Studies concerning Japanese mora-timing were very prevalent (Beckman, 1982; Port, Dalby, ＆ O’Dell, 1987; Kubozono, 1989, and Aoyama, 1999). In this study, six Japanese suprasegmental features were discussed: vowel length, sokuon length, speech rate, pause frequency, pause duration, and pause distribution among four groups with different Japanese language proficiency. These suprasegmental elements could be factors which may contribute to foreign accent and different speech fluency. The data were collected by interviewing with total thirty-two subjects. These were eight native Japanese speakers, eight Taiwanese students who had passed Japanese Language Proficiency Test (JLPT) level 1, eight for JLPT level 2, and eight for JLPT level 3. This study was composed of three experiments concerning temporal patterns. For vowel length distinction, the designed materials were ten minimal pairs of real Japanese words which were all disyllabic. Each pair contained short and long vowels and was composed of two or three moras (Hirata, 2004). For sokuon length distinction, the materials were eight minimal pairs of real Japanese words. For speech fluency, the materials were four types of Japanese sentences excerpted from five Japanese conversations. For each sentence type, three sentences were chosen. Finally, all the sound segmentation was conducted by PRAAT software. The statistical software R project was used for data entry and analysis. Statistical procedures employed included one-way ANOVA and Tukey post hoc tests. The results of the statistical analysis revealed significant effect for proficiency. In terms of vowel and sokuon length, the speech production of JLPT L1 ＆ L2 was closer to natives more than JLPT L3. Regarding the speech fluency, the natives spoke much faster than non-natives and had much less number of pauses and shorter of pause duration. In regard to pause distribution, the group of JLPT L3 often located pauses at wrong locations in syntactic structures more than JLPT L1 ＆ L2. Overall, compared with participants with low linguistic proficiency, the results indicated that the performance of participants with high linguistic proficiency was more accurately and was more similar to natives. The influencing factors such as language transfer including overproduction and underproduction (Odlin, 1989) would be discussed. Finally, by examining the result of this study, we proposed suggestions for Japanese learning and teaching.