Date of Award
Doctor of Philosophy (PhD)
Communication Sciences and Disorders
A meaningful sentence loaded with appropriate phonemic and syllabic forms was synthesized as a "standard" stimulus, and 60 "accented" versions of the sentence were made to stimulate varying degrees of a moderate and a strong Spanish accent by manipulating the following Spanish cues singly and in combination: (1) fundamental frequency, (2) voice onset time for syllable-initial voiceless stops, (3) Duration of medial stressed vowels, (4) F1, F2 and F3 for full vowels, and (5) F1, F2, and F3 for reduced vowels. Two tapes for each level of accent were prepared on which 30 accented stimulus sentences were each paired with the standard sentence in four randomized sequences. Forty-two English speakers rated how different each accented sentence was from the standard sentence on a 10-point scale; they also gave a confidence rating on a 5-point scale for each item. It was demonstrated that synthesized sentences can be reliably rated for cue modifications indicative of a moderate Spanish accent in English. Statistical analysis revealed that an increase in the number of cues (from 1 to 2, to 3, to 4, to 5) resulted in the perception of increased accentedness in both the moderate- and strong-accent condition. In addition, subjects' confidence in their judgments increased along with an increase in number of cues. A factor analysis showed that the suprasegmental cue, fundamental frequency (intonation), was the most perceptually prominent cue signalling a moderate Spanish accent in English. The segmental cue, stressed vowel quality, was the next prominent cue. The presence of these cues also resulted in an increase in the subjects' confidence in their ratings of stimuli. The two strongest accent-bearing cues signalling the strong accent were both segmental, stressed vowel quality and VOT, but the strong-accent data was determined to be generally unreliable, possibly because of errors in its generation.
Rekart, Deborah Marie, "Evaluation of Foreign Accent Using Synthetic Speech (Perception)." (1985). LSU Historical Dissertations and Theses. 4152.