The effects of second language proficiency on speech production in noise

Abstract

Speech production is a dynamic process and talkers produce variance of speech in order to maximize intelligibility and minimize articulatory efforts (cf. hyper- and hypo-articulation model). Prior research has shown that native (L1) talkers modify their speech produced in noise, such as increasing intensity, and changing formant frequency (especially F1 for vowel production). However, it is still unknown to what extent L2 talkers accommodate the environmental need. For example, speaking in noise is less familiar for L2 talkers, especially English as foreign language (EFL) learners. In addition, EFL learners lack interaction in their L2 usage, i.e., their speech may not be successfully modified considering listeners’ facilitation. The current study thus tested this with Japanese EFL learners (high and low L2 proficiency). The target stimuli were English tense/lax vowel production (/i-ɪ/ and /u-ʊ/), which is known that Japanese learners of English tend to assimilate the spectral differences and instead use the durational contrast. Our preliminary results showed that both groups of the learners produced spectral enhancement of F1. However, only the high proficiency group used the temporal modification on /i/, potentially due to their faster speaking rate. These results suggest that different modification strategies were observed based on their L2 proficiency.