Contribution of Linguistic Variables to Bilingual Listeners' Perception of Degraded English Sentences

Abstract

The present study was designed to investigate what linguistic variables best predict bilingual recognition of acoustically degraded sentences and how to identify bilingual individuals who might have more difficulty than their monolingual counterparts on such tasks. Four hundred English speech-perception-in-noise (SPIN) sentences with high and low context were presented in combinations of noise (signal-to-noise ratio: +6 and 0 dB) and reverberation (reverberation time: 1.2 and 3.6 s) to 10 monolingual and 50 bilingual listeners. A detailed linguistic profile was obtained for bilingual listeners using the Language Experience and Proficiency Questionnaire. Variables per reading in English (age of fluency, proficiency, and preference) emerged as strong predictors of performance across noise, reverberation, and context effects. Via discriminant analyses, bilingual listeners who rated their accent to be perceptible and reported shorter length of immersion in an English-spoken country or school tended to score significantly lower on the SPIN test than monolingual listeners. Bilingual listeners' linguistic background plays a major role in their use of context in degraded English sentences. Rather than conventional variables such as age of acquisition, variables pertaining to reading, proficiency, immersion, and accent severity may be obtained for improved prediction of bilingual performance on the task.