Journal of Second Language Pronunciation - Volume 2, Issue 2, 2016

The current study investigated the effect of listener status (native, nonnative) and language background (French, Mandarin) on global ratings of second language speech. Twenty-six nonnative English listeners representing the two language backgrounds (n = 13 each) rated the comprehensibility and accentedness of 40 French speakers of English. These same speakers were previously rated by native listeners and coded for 19 linguistic measures of speech (e.g., segmental errors, word stress errors, grammar accuracy) in Trofimovich and Isaacs (2012). Analyses indicated no difference in global ratings between nonnative and native listeners, or between the two nonnative listener groups. Similarly, no major differences in the linguistic dimensions associated with each group’s ratings existed. However, analyses of verbal reports for a subset of nonnative listeners (n = 5 per group) demonstrated that each group attributed their ratings to somewhat different linguistic cues.

Perceptual training can lead to improvements in production which generalize to new words and talkers. This study investigated the effects of perceptual training on productions in larger discourse contexts of continuous speech, and additionally examined whether training generalized to a new syllable structure and across grammatical domains. Participants included Korean L2 learners of English in a pretest-training-posttest design. An experimental group completed perceptual training on singleton coda palatals, and a control group completed training on an unrelated target. Results indicated that perceptual training on singleton coda palatals was successful in significantly improving learners’ productions in continuous speech. Learners were able to generalize production improvements to a new syllable structure (simple vs. complex coda), but not across grammatical domains (-ed morphemes). These findings provide further support for the use of perceptual training in pronunciation classrooms, but demonstrate some limitations to its generalizability.

Processing lexical tones is known to be difficult for non-native speakers from various language backgrounds. Perceptual accuracy of six Mandarin tone contrasts (T1–T2, T1–T3, T1–T4, T2–T3, T2–T4, T3–T4) of two groups of Japanese listeners (learners and non-learners) and a control group of Mandarin listeners were compared in Mandarin learning experience, phonetic context, and speaker gender. Both Japanese groups perceived T2–T3 poorly, but the learner group was significantly better than the non-learner group for this and the T1–T2 contrasts. The learners’ advantage was observed across various phonetic contexts (initial consonants and tone bearing vowels), suggesting that their tone perception was more stable and resistant to speaker variation. In regard to speaker gender, both Japanese groups perceived the T2–T3 contrast more poorly, and T1–T2 and T1–T4 contrasts more successfully, when produced by female speakers.