The perception and production of epenthetic vowels in non-native clusters in Japanese : phonetic and phonological influences.
Degree GrantorUniversity of Canterbury
Degree NameMaster of Arts
This thesis investigates the quality of epenthetic vowel that native speakers of Japanese tend to produce and perceive between unfamiliar sequences of consonants. Research on perceptual epenthesis in Japanese has revealed the high back [ɯ] to be the vowel commonly perceived in illicit consonant sequences. However, loanword studies suggest that there are three epenthetic vowels, which reflect phonotactic restrictions on certain consonant + vowel sequences. That is, the quality of epenthetic vowel is predictable from the preceding consonantal environment. In this study, I tested to what extent the response patterns in perceptual and production experiments are consistent with native phonotactics, and how phonetic properties of the listeners’ native language play a role in speech perception. This thesis first investigates the potential influence of the preceding consonant environment on perception and production of illicit consonant clusters. Second, the current study considers the effect of all vowel categories in Japanese, including allophonic variation of the Japanese high vowel [ɯ] — the high vowel undergoes devoicing when it occurs between voiceless obstruents — on the perception of illicit consonant sequences. This study thus integrates perceptual and production experimental work in an investigation of the contextual environments that contribute to predicting the quality of epenthetic vowels in Japanese. In the perception experiment, a same-different AX discrimination task was employed, in order to determine whether native speakers of Japanese are able to tell the difference between licit [VC1VC2V] (C=consonant, V=vowel) and illicit [VCCV] pairs (e.g., [apata]-[apta]) when they listen to pre-recorded pseudo-word stimuli. In each trial, participants were asked to judge whether a pair of stimuli were the same or different. The experiment enabled us to test whether Japanese listeners perceive an illusory vowel between consonants in an illicit sequence and whether the vowel percept differs according to a given phonological environment. The results show that to some extent, the preceding consonant does influence the vowel perceived, yet there is a bias toward perceiving [ɯ] in voiceless consonantal contexts, a result not predicted by the language’s phonotactic patterns. Additionally, it was found that the order that the stimuli were presented to subjects influences epenthesis in perception. Japanese listeners were less accurate in identifying whether members of a pair were same-different with the [aCVCa-aCCa] order than with the [aCCa-aCVCa] order. In the production experiment, a read-aloud task was employed. Speech production data was collected using the same pseudo-words as in the perception experiment though in this experiment the stimuli were presented to subjects orthographically. The results showed that for some preceding environments, the findings are relatively consistent with expectations based on the language's phonotactics, but this was not the case for all contexts. The results also revealed that there was variability across speakers as to which vowels they epenthesized after particular consonants. The current series of studies revealed that the quality of epenthetic vowels was not merely influenced by the phonotactics of the native language in speech perception and production. Instead, other factors interact in a complex way during speech perception and production.