The Effects of Rater-Specific Characteristics on the Rating of Foreign Accent

Sahar Almohareb

Major Professor: steven h. weinberger, PhD, Department of English

Committee Members: Harim Kwon, Douglas Wulf

Online Location, https://gmu.webex.com/meet/weinberg
July 22, 2020, 10:00 AM to 01:00 PM

Abstract:

This dissertation examines how rater-specific characteristics affect the rating of foreign accent in English. Previous studies have focused on the effects of speaker-specific characteristics on the rating of foreign accents (e.g., Munro & Derwing, 1995; Flege et al., 1999; Yeni-Komshian, et al., 2000; Piske, et al., 2001). However, recent studies (e.g., Kang, 2012; Schoonmaker-Gates, 2012; Weber & Pollman, 2010; Hayes-Harb et al., 2008; Huang & Jun, 2015) have shown that rater-specific characteristics may also influence the rating of accented speech, but the effects of these rater-specific characteristics are not well understood. Specifically, there are conflicting results regarding the effects of linguistic training of the raters (trained vs. non-trained), the raters’ nativeness status (native vs. nonnative), and the nonnative rater-speaker shared L1 status (shared L1 vs. not shared L1) on the rating of foreign accent. Additionally, the relationship between the nonnative raters' degree of accentedness and their ratings of other’s accented speech has not been previously explored.

Accordingly, the primary focus of this dissertation is to investigate how these rater-specific factors affect the accent rating. In particular, it examines whether trained and non-trained native raters differ in their rating behaviors, whether native and nonnative raters rate accented speech differently, whether nonnative raters rate speech from speakers sharing the same L1 background with them differently from speakers from a different L1 background, and whether nonnative raters’ length of residence in an English-speaking country, self-reported L1 use, and own degree of foreign accent influence their foreign accent ratings.

In an online foreign accent rating experiment, trained and non-trained native English raters and naïve nonnative raters from Arabic and Mandarin L1 backgrounds rated the degree of foreign accentedness of 150 short English phrases extracted from the Speech Accent Archive (Weinberger, 2019) from native and nonnative speakers from Arabic and Mandarin L1 backgrounds. Raters were also recorded reading the Stella passage for the assessment of nonnative raters' foreign accent by trained raters.

The results show that native English raters with linguistic training did not differ from non-trained native raters in their ratings of the degree of foreign accent and that their ratings were strongly correlated. As for the difference between native and nonnative raters, the results show that they were different in their rating behaviors. Overall, native raters always assigned lower accent ratings to all the speech samples than nonnative raters. However, when looking at the rater-speaker shared L1 status, an interesting pattern emerged. The results show that nonnative raters rated nonnative speech samples from a different L1 background to have more foreign accent. In particular, while native raters and Arabic raters rated Arabic-accented samples to have similar degrees of foreign accent, Mandarin raters rated Arabic-accented samples significantly differently from English and Arabic raters. Similarly, while native raters and Mandarin raters rated Mandarin-accented samples to have comparable degrees of foreign accent, Arabic raters rated the Mandarin-accented samples significantly differently from English and Mandarin raters. Additionally, the results show no correlations between nonnative raters’ length of residence, self-reported L1 use, and own degree of foreign accent and their ratings of other’s accents.

These findings suggest that rater-specific characteristics, such as nativeness status and rater-speaker shared L1 status, can influence the rating of foreign accent, pointing to the complex nature of accent rating and the factors affecting it. This dissertation provides important implications for research on foreign accent rating, spoken language assessment, and language pedagogy. It also offers a foundation for further investigation on the rater-specific characteristics influencing foreign accent assessment, which in turn contributes to our understanding of speech perception more generally.