M.A. Theses in Linguistics at the University of North Dakota (Abstracts)
Miller, Dustin D. 2008
Production of Bangla Stops by Native English Speakers Learning Bangla: An Acoustic
Analysis
Differences in the phonetic and phonological systems of Bangla and English result in negative transfer in the Bangla stop productions of native English speakers. The phonetic realizations of Voice and Aspiration and their interactions with each other are the key factors in this. A production study was carried out focusing on sixteen of the twenty Bangla stops that are distinguished by a four-way voice/aspiration contrast at four different places of articulation, providing a contrastive acoustic analysis of the pronunciation of L1 and L2 adult speakers. Data containing these stops in an intervocalic environment in word-initial, word-medial, and word-final positions was elicited by digital recording from twelve native Bangla speakers and twelve native English speakers. The data from the L1 speakers was analyzed to investigate production characteristics related to the following acoustic variables: vowel voicing onset time, closure duration, closure voicing, preceding vowel duration, and duration of aspiration noise. The data from the L2 speakers was then analyzed using the same variables.
The primary acoustic correlates of Voice and Aspiration in Bangla were found to be closure voicing and vowel voicing onset time, respectively, and the interaction of these two variables made a clear distinction between the four stop classes of Bangla: voiceless unaspirated, voiceless aspirated, voiced unaspirated, and voiced aspirated. Evidence was found supporting the work of various researchers who have suggested that a [breathy voice] feature is not necessary for a phonological description of the Indo-Aryan languages. The stop productions of the native English speakers indicated a conceptual awareness of the four stop classes, but it was also clear that they lacked a native-like control of the Voice and Aspiration features and their specific interactions with each other. The degree to which the L2 productions of the four stop classes were different from those of the L1 was directly correlated to each class’s similarity to English phonological patterns, providing evidence of certain predictable aspects of L1 transfer. In order to fully apply the results of this study in a pronunciation acquisition context, perceptual studies will need to be done to identify the salience of these acoustic variables for both L1 and L2 speakers. Perceptual studies involving L1 speakers may also give a greater understanding to the ongoing discussion on the best phonological description of the four-way stop systems of the Indo-Aryan languages.
Download thesis from http://arts-sciences.und.edu/summer-institute-of-linguistics/theses/_files/docs/2008-miller-dustin-d.pdf