The Acquisition of Segmental Timing by Children in a Japanese Immersion Program

Tetsuo Harada


Many previous studies have claimed that the earlier children are exposed to a second language in a naturalistic setting, the more likely they are to acquire native-like pronunciation (e.g., Flege, 1992; Oyama, 1976; Patkowski, 1994; Scovel, 1988). However, very little acoustic research has been done to give us any clue as to how early exposure in an immersion program might affect the acquisition of second language speech.


This cross-sectional and longitudinal study acoustically analyzed the production of voice onset time (/p, t, k/) and geminates (/pp, tt, kk/) in Japanese by English-speaking children in grades 1, 3, and 5 (N=19) in a Japanese immersion program. In addition, 52 native speakers of Japanese rated the contrast between single and geminate consonants produced by the immersion children.


The results show that the immersion children's Japanese VOT was significantly longer (58 to 67 ms) than monolingual Japanese children's VOT (28 to 30 ms) . This suggests that Japanese VOT for the immersion children is influenced by their English VOT (85 to 89 ms), which is longer than Japanese VOT. However, interestingly, it has been found that although they have not reached the native speakers' norm with respect to VOT, they are found to distinguish clearly their Japanese VOT values for /p, t, k/ from those of their native English counterparts (58 to 67 ms vs 85 to 89 ms) .


For single and geminate consonants, Han (1992) found that the duration of L1 English speakers' Japanese geminates is shorter than that of Japanese mono1inguals. However, my results show that both the immersion children's geminates and their singletons were longer than those of the Japanese monolinguals. In addition, the immersion children's mean ratio for singletons to geminates was smaller (1.4 to 1.6) than that of the monolinguals (1.9 to 2.3).


The accent ratings by the Japanese native speakers suggest that the immersion children retain a noticeable L1 accent and there is no statistical difference in the scores of accentedness across the grade levels. Moreover, the degree of perceived accentedness for all the immersion children correlated fairly highly with their ratio of singletons to geminates (r = 0.773, p < 0.0001) .  This suggests that the ratio of singletons to geminates is a good measure of the acquisition of the phonemic contrast between them.