Detail of contribution
Auteur: Becky TAYLOR
Co-Auteur(s): Sam HELLMUTH, University of York
Titre:
The effect of L1 background on perception of varying phonetic correlates of word accent
Abstract/Résumé: English speakers have been shown to have difficulty mastering word-accent in a second language (Kijak 2009, Taylor 2011) and thus appear to be ‘stress deaf’, to borrow the term of Dupoux & Peperkamp (DP, Dupoux et al 1997, 2001, 2008). DP attribute ‘stress deafness’ to the degree of unpredictability in the word-accent system of a particular language (Peperkamp et al 2010), but as English stress is only partially predictable English learners should not be ‘stress deaf’. Kijak (2009) attributes her English listeners’ apparent stress deafness to the low functional load of stress in English (relatively few minimal pairs), but also suggests that differing phonetic correlates of stress between L1 and L2 may affect a listener’s ability to perceive the position of L2 stress. In this paper we demonstrate an L1 background effect on listeners’ perception of differing phonetic realisations of word accent, through two perception experiments. Expt 1 uses DP’s word sequence recall test, designed to elicit responses that tap into abstract phonological processing, rather than a surface acoustic memory strategy. Three groups of participants were exposed to stimuli in which word accent was realised with 1) melodic cues only (Japanese-like), 2) melodic and dynamic cues (Spanish-like) or 3) melodic and dynamic cues plus unstressed vowel reduction (Dutch-like). 38 British English listeners participated in the study; each was assigned pseudo-randomly to a group and heard only one type of stimulus. The results show a clear difference in degree of ‘stress deafness’: group 3) display no ‘stress deafness’ but group 1) display a high degree of stress deafness, similar to that observed by DP for French listeners; there is considerable speaker variability in group 2). These results suggest that ‘stress deafness’, in British English listeners, is directly related to which phonetic correlates are used to mark the position of word accent. In Expt 2 the same three groups performed an AX task with the same type of stimulus as in Experiment 1. DP found French listeners were able to discriminate accent acoustically in an AX task and thus defined ‘stress deafness’ as an inability to represent and store word-level stress in working memory. Our listeners display varying degrees of difficulty in discriminating word accent: good discrimination in group 3) vs. no discrimination in group 1), and individual variation in group 2). English listeners thus appear to have difficulty in representing and storing word accent because they are unable to hear the position of word accent; the difficulty disappears when presented with stimuli which display phonetic correlates of stress which match those observed in their L1.
Co-Auteur(s): Sam HELLMUTH, University of York
Titre:
The effect of L1 background on perception of varying phonetic correlates of word accent
Abstract/Résumé: English speakers have been shown to have difficulty mastering word-accent in a second language (Kijak 2009, Taylor 2011) and thus appear to be ‘stress deaf’, to borrow the term of Dupoux & Peperkamp (DP, Dupoux et al 1997, 2001, 2008). DP attribute ‘stress deafness’ to the degree of unpredictability in the word-accent system of a particular language (Peperkamp et al 2010), but as English stress is only partially predictable English learners should not be ‘stress deaf’. Kijak (2009) attributes her English listeners’ apparent stress deafness to the low functional load of stress in English (relatively few minimal pairs), but also suggests that differing phonetic correlates of stress between L1 and L2 may affect a listener’s ability to perceive the position of L2 stress. In this paper we demonstrate an L1 background effect on listeners’ perception of differing phonetic realisations of word accent, through two perception experiments. Expt 1 uses DP’s word sequence recall test, designed to elicit responses that tap into abstract phonological processing, rather than a surface acoustic memory strategy. Three groups of participants were exposed to stimuli in which word accent was realised with 1) melodic cues only (Japanese-like), 2) melodic and dynamic cues (Spanish-like) or 3) melodic and dynamic cues plus unstressed vowel reduction (Dutch-like). 38 British English listeners participated in the study; each was assigned pseudo-randomly to a group and heard only one type of stimulus. The results show a clear difference in degree of ‘stress deafness’: group 3) display no ‘stress deafness’ but group 1) display a high degree of stress deafness, similar to that observed by DP for French listeners; there is considerable speaker variability in group 2). These results suggest that ‘stress deafness’, in British English listeners, is directly related to which phonetic correlates are used to mark the position of word accent. In Expt 2 the same three groups performed an AX task with the same type of stimulus as in Experiment 1. DP found French listeners were able to discriminate accent acoustically in an AX task and thus defined ‘stress deafness’ as an inability to represent and store word-level stress in working memory. Our listeners display varying degrees of difficulty in discriminating word accent: good discrimination in group 3) vs. no discrimination in group 1), and individual variation in group 2). English listeners thus appear to have difficulty in representing and storing word accent because they are unable to hear the position of word accent; the difficulty disappears when presented with stimuli which display phonetic correlates of stress which match those observed in their L1.