The purpose of this study is to compare the performance of the native speakers of American English to that of the Japanese in the following aspects: (1) the degree of compression effect in production, (2) the discrimination of vowel durations, and (3) the influence of an ISI (interstress interval) and a stressed vowel on the perception of temporal naturalness.
Linguistic materials for the production experiment were English sentences which differed systematically in the number of unstressed syllables that intervened between a target stressed syllable and the next stressed syllable. The compression effect in the stressed vowel was observed in the sentences produced by the American speakers and those by the Japanese good learners of English, while the degree of compression effect was very small for the poor learners. In addition, the overdoing of the compression was observed in the case of the god learners.
In the first perceptual experiment, the discrimination of vowel durations was measured by using LPC analysis-synthesis speech. The duration of a stressed vowel in each of the seven English sentences was varied in g steps from -40ms to +40ms. The discrimination for the lengthening of the vowels was found to be nearly the same between the Americans and the Japanese poor learners, while the latter was more sensitive to the shortening of the vowels than the former. The result of the second experiment showed that the degradation of naturalness due to the lengthening of a stressed vowel lessened in the case of the American listeners, but not in the Japanese, when there was a durational compensation in the following unstressed syllable.