Phonetic encoding of coda voicing contrast and its interaction with information structure in L1 and L2 speech

Abstract

This study examines the coda voicing effect in English on production of preceding vowel by comparing L1 and L2 speech. Particular attention is paid to how coda voicing contrast is encoded in temporal vs. spectral dimensions, how the effect interacts with information structure (with varying focus types), and how the use of information structure differs in L1 vs. L2. Both English (native) and Korean (non-native) speakers show a comparable pattern in the temporal dimension (vowel duration), but Korean speakers fail to encode the spectral attribute (vowel F1/F2) due to coda voicing. Furthermore, while native speakers make use of information structure in an efficient way that is accountable by the output-oriented and the system-oriented constraints, non-native speakers do not use information structure as efficiently. Alongside the L1/L2 issues, a ‘long-distance’ coda voicing effect is observed on VOT of the preceding initial stop, which illuminates the nature of the coda voicing effect.