Stringer, LM; (2015) Accent intelligibility across native and non-native accent pairings:investigating links with electrophysiological measures of word recognition. Doctoral thesis , UCL (University College London).

Preview

PDF Stringer_MPhil_2015.pdf Available under License : See the attached licence file. Download (3MB)

Abstract

The intelligibility of accented speech in noise depends on the interaction of the accents of the talker and the listener. However, it is not yet clear how this influence arises. Accent familiarity is commonly proposed to be a major contributor to accent intelligibility, but recent evidence suggests that the similarity between talker and listener accents may also be able to account for accent intelligibility across talker-listener pairings. In addition, differences in accent intelligibility are also often only found in the presence of other adverse conditions, so it is not clear if the talker-listener pairing also influences speech processing in quiet conditions. This research had two main aims; to further investigate the relationship between accent similarity and intelligibility, and to use online EEG methods to explore the possible presence of talker-listener pairing related differences on speech perception in quiet conditions. English and Spanish listeners listened to Standard Southern British English (SSBE), Glaswegian English (GE) and Spanish-accented English (SpE) in a speech-in-noise recognition task, and also completed an event-related potential (ERP) task to elicit the PMN and N400 responses. Accent similarity was measured using the ACCDIST metric. Results showed the same (or extremely similar) patterns in accent intelligibility and accent similarity for both listener groups, giving further support to the hypothesis that accent similarity can contribute to the level of intelligibility of an accent within a talker-listener pairing. ERP data also suggest that speech processing in quiet is influenced by the talker-listener pairing. The PMN, which relates to phonological processing, seems particularly dependent on a match between talker and listener accent, but the more semantic N400 showed some flexibility in the ability to process accented speech.

Download activity - last month

Export as JSONXMLCSV

Download activity - last 12 months

Export as XMLCSVJSON

Downloads by country - last 12 months

Export as JSONXMLCSV

Archive Staff Only

View Item