Prosodically Induced Acoustic Variation in English, Korean, and Hindi

Abstract

This dissertation presents results from speech production experiments on English, Korean and Hindi that examine factors governing acoustic variation. The focus of this research is the patterns of variation in the acoustic expression of laryngeal phonological contrasts among stop consonants and supralaryngeal contrasts among vowels, and the role of systemic and prosodic factors in governing such variation. These research goals are investigated with English, Korean, and Hindi CV sequences in utterance initial (U) or Intonational Phrase initial (IP) positions, U/IP-final position, and U/IP-medial position, with and without contrastive focal accent. VOT and FO at the onset of the following vowel are measured as the main acoustic correlates of the laryngeal features in stops, in conjunction with vowel duration and the first and second formant frequencies for vowels. Measurements are taken from the speech of six male American English speakers, five male Seoul Korean speakers, and five male Delhi Hindi speakers. Statistical results indicate that the effect of focal accent on acoustic measures is generally more distinctive and consistent than the effect of prosodic boundary, particularly in Korean and English. The distribution of acoustic measures also shows a greater distinction between the contrastive stops and vowels when accented. Variation due to domain boundary position, on the contrary, is not very great and when it occurs it is found to be localized to segments at the edges of prosodic domains. This study finds no evidence that stop consonant or vowel contrasts are enhanced in domain-initial position. Hindi provides an example of the initial strengthening of vowels without enhancement of the preceding consonants, and Korean shows that vowels have a more peripheral distribution in final position. All in all, the accentual effect is rather uniformly observed in acoustic variation across these three languages, whereas the boundary effect defines language-specific patterns of acoustic variation. The correlation between the size of the phonological inventory and the corresponding acoustic variation is not strongly supported with the findings in the present study.