Effect of the consonant context on the corner vowel produced by native and Chinese speakers: based on AESOP corpora.

Abstract

This study investigated the differences in the consonant context on the four corner vowels (/i/, /u/, /æ/, and /ɑ/) between native and Chinese speakers using the AESOP-ILAS 1 corpus dataset. Ninety-six test subjects with manually adjusted data comprising of 12 native speakers (the control group) and 84 non-native speakers (the targeted group), with 58 Chinese speakers (even gender distribution), 22 Taiwanese speakers (quasi-equal gender distribution), and 4 Hakka-speaking women, were chosen for the analysis. By adopting continuous speech initially, a general comparison of the vowel space for the native and the non-native speakers is presented. Next, the vowel space of the liquid and glide sonorants in the onset and the coda position are compared. The nasal sonorants are compared in their subsections. Finally, the study analyzed the obstructive sounds and compared the vowel spaces of different pronunciation sites (alveolar, labial, posterior alveolar, and palatal sounds). Since male speakers typically having longer vocal tracts, their vowels may be more centralized or lower in pronunciation compared to females. In order to comprehensively understand the spatial distribution of vowels, it is valuable to further analyze whether there are differences in vowel production patterns between male and female speakers in the native and non-native language groups. Compared to native speakers, Chinese English learners have a larger range of vowel spaces, which may be due to the fact that the corpus is collected from sentences rather than isolated words. Chinese learners exhibit lower F1 and F2 values on corner vowels, with particularly significant differences between /ɑ/ and /u/. These differences are influenced by adjacent phonemes, such as the/w/sound at the beginning and end of syllables. The study suggest that improving the stress distribution of Chinese learners in sentences will significantly enhance their pronunciation level.