Voice quality differences between American English speakers and Korean learners of English

Abstract

Various features have been examined as markers of foreign accent including segment-level errors, deviation in subsegmental acoustic properties, as well as prosodic patterns such as intonation, rhythm, and speaking rate. Features that characterize voice quality may also be useful but they have been investigated less extensively in the literature. The present study explores the utility of voice quality features by analyzing and comparing speech samples from American English speakers and Korean learners of English documented in the Wildcat Corpus of Native- and Foreign-Accented English. Measures of jitter, shimmer, and harmonics-to-noise ratio (HNR) were extracted from vowels and compared between the two speaker groups. Overall, Korean learners produced English vowels with less jitter and shimmer as well as higher HNRs than English speakers. The pattern was consistently found in both scripted and unscripted speech and was more salient among male speakers than female speakers. Furthermore, the same trend, although weaker, was also found when the Korean and English speakers produced vowels in their respective L1s while reading comparable scripts. However, the Korean speakers showed more shimmer while reading scripts in Korean (L1) than English (L2).