Abstract
Empirical studies have been rare in the realization of suprasegments of Mandarin speakers’ English, and acoustic correlates of English lexical stress seemed to be the main concern for the few existing ones. In this study, the L2 Intonation Learning theory (LILt; Mennen, 2015) was introduced, which proposed that in terms of suprasegmental learning, phonological categories should be acquired prior to phonetic implementation, as found in segmental learning. To test the hypothesis, investigation of pitch accents and boundary tones in Taiwan EFL learners’ production has been conducted.
Sixteen Taiwan EFL learners’ read-aloud recordings were randomly selected from a test-based learner corpus in Taiwan. Half of them received the highest score in the test (high-proficiency L2 learners, or HL2), while the other half did not pass the exam (low-proficiency L2 learners, or LL2). Native speakers of American English and Taiwan Mandarin speakers were also recruited and recorded following the same instruction. Materials were two short English passages containing 165 words in total. The English Tone and Break Indices System was used for data labelling (Beckman & Ayers, 1997).
Results showed that in terms of boundary tones, both HL2 and LL2 speakers showed a similar preference as native speakers in the assignment of boundaries and choices of tone types when boundaries coincide with punctuation marks. However, when a boundary was covert, i.e., with no additional cue from punctuation, HL2 still behaved like native English speakers, while LL2 were less consistent in both tonal assignment and tonal choices. For pitch accents, HL2 learners also behaved more like native speakers in the number of pitch accents assigned to content and function words. However, for measures like declination and accent density, significant difference was found between natives and nonnatives, which indicated that these measures may be more difficult to learn. Although minor differences were found in some measures, Taiwan EFL learners showed better performance in phonological categorization than phonetic implementation. Finally, the LILt (2015) was further modified based on these findings. It is believed that the development of the phonological and phonetic aspects of L2 prosody should be co-occurring, rather than sequential.
研究貢獻
本研究將以往教學者憑經驗描述之EFL超音段現象，透過界限聲調及音高重音等聲學測量，進行實證資料之標記及分析，並與現行之理論架構進行比較。研究對象為臺灣英語學習者，因中、英分屬不同韻律系統，透過觀察不同程度學習者的英語產出，可歸納兩語言間是否出現交互作用。此外，本研究所使用的語料來自標準化測驗，所有評分人員均依據相同評分規範接受訓練，於評分期間亦隨時接受校準，故在學習者英語流利度的判定上具一定之公信力。研究發現未來可作為機器學習的訓練材料，以創建適合模型、精準預測臺灣英語學習者之韻律特色。
---
References
Beckman, M. E., & Ayers, G. (1997). Guidelines for ToBI labelling. The OSU Research Foundation, 3, 30. (Version 3.0 ed.).
Mennen, I. (2015). Beyond segments: Beyond segments: towards an L2 intonation learning theory (LILt). In Delais-Roussarie, E., Avanzi, M., & Herment, S. (Eds.) Prosody and languages in contact: L2 acquisition, attrition, languages in multilingual situations (pp. 171-188). Berlin: Springer.