Does speech production in L2 require access to phonological representations?

Abstract

Following the theory of direct realism [1, 2, 3], non-native (L2) speakers should be able to imitate a stimulus without requiring the access to L2 phonological representations. In line with theories of working memory [4, 5], however, they should encounter difficulties in imitating a stimulus with L2 phonological structure at the point once phonetic information decayed and therefore phonological representations are required. In order to test the validity of these claims, the current study investigates L2 speakers' ability to immitate L2 segmental length contrasts in an immediate vs. delayed imitation paradigm. In the immediate imitation condition, participants should be able to make use of phonetic information taken from the acoustic echo of the stimuli. In the delayed imitation condition, however, phonological representations were required after the decay of phonetic information. The results show that L2 speakers' performance differed from that of native (L1) speakers in the immediate imitation condition, suggesting that phonological representations had been already activated in the immediate condition. L2 speech production may inevitably require phonological representations. The claim made by the direct realist view was not supported in this study.

Cognitive Factors in Thai-Naïve Mandarin Speakers’ Imitation of Thai Lexical Tones

“…The difference scores from the target stimuli in the low memory load condition suggest that participants did not imitate the phonetic details of the target tones correctly. Previous research on imitation of consonant length also showed little effect of memory load [24]. The less than expected effect of memory load could be because when waiting for imitation in the long interval condition, participants rehearsed internally, which reduced the decay of phonetic details perceived from the stimuli, thus diminishing the difference between two memory load conditions.…”