One of the key factors that determine speech intelligibility under challenging conditions is the difference between the accents of the talker and listener. For example, normal-hearing listeners can be accurate at recognizing a wide range of accents in quiet, but in noise they are much poorer (e.g., 20 percentage points less accurate) if they try to understand native (L1) or non-native (L2) accented speech that does not closely match their own accent. The aim of this PhD research is to provide a more detailed account of this talker-listener interaction in order to establish the underlying factors involved in L1 and L2 speech communication in noise for normal-hearing populations. Study 1 examined the effects of L2 proficiency on the L1-L2 accent interaction in noise, with Study 2 investigating the contribution of acoustic similarity to accent intelligibility. Study 3 examined L1 listeners’ adaptation processes to unfamiliar accents in noise. Finally, Study 4 took a cross-linguistic approach and investigated how language experience and accent similarity affect the talker-listener accent interaction in noise across languages. Overall, the results revealed that several factors contribute strongly to the L1-L2 accent interaction in noise, with the emerging findings contributing to our general understanding of speech in noise perception. For instance, acoustic similarity in the accents of the talkers and the listeners accounted for a great amount of the variance in intelligibility. Linguistic background and L2 experience were also shown to play a major role in the interaction, shaping the listeners’ accent processing patterns in their L1 and L2, as well as general speech-in-noise processes, with bilingual and highly proficient L2 listeners showing facilitation effects for speech processing in both their languages. Finally, the selective tuning processes found for standard accents in English were not replicated for French, indicating that accent processing varies across languages.
展开▼
