Production of Japanese durational contrasts by native and nonnative speakers

Abstract

Non-native speakers’ production of Japanese durational contrasts, and the development of their abilities, are examined in relation to native speakers’ production. Two native speakers and two beginning learners produced 2–4 mora words contrasting in either consonant or vowel length. The learners also produced the same set of words after 4 weeks of perceptual training [Y. Hirata, CLS34 Phonetics Panel (forthcoming)]. Acoustic analyses of the native speaker productions indicate durational systematicity not only in the consonant pairs, as in M. Han [Phonetica 49, 102–127 (1992)], but also in the vowel pairs: the average ratio of short to long vowels was 1:1.8–2.0 regardless of context and speaker. Furthermore, a linear relationship was found between the number of morae and word duration, extending Port et al. [J. Acoust. Soc. Am. 81, 1574–1585 (1987)]. The learners’ results before training indicate an inability to produce the durational contrasts, e.g., subject A’s average durational ratio for the vowel pairs was 1:1.2. After training, the ratio became closer to that of the native speakers, viz., 1:1.6–1.7, while subject B’s did not show a linguistically significant change. Subject A’s results contrast even with those of Han’s advanced subjects, and illuminate the upper limit for L2 learners with training.