Accentedness, comprehensibility and intelligibility of L2 speech: a replication and extended study

This thesis examines the relationship between the accentedness, comprehensibility and intelligibility of Second Language (L2) speech, and is based on the replication of an empirical study carried out in the 1990s by Munro and Derwing (1995a). Following the methodology of the original study, ratings were collected from native English listeners for the accentedness, comprehensibility, and intelligibility of spontaneous L2 English utterances of mother tongue Arabic speakers. These ratings were then correlated with each other to ascertain what bearing a foreign accent has on the comprehensibility and intelligibility of speech. The three listener ratings were also correlated to error counts for phonetic and phonemic errors, and non-native intonation, in an attempt to establish which error types affect an utterance most in terms of its accentedness, comprehensibility and intelligibility.

The original study resulted in two major findings: firstly, accentedness and intelligibility were found to be orthogonal; and secondly, non-native intonation was found to be more highly correlated with problems of comprehensibility than were phonetic and phonemic errors. These two findings were re-examined in the context of the present study. The results suggest that if language is treated as a complex system, whose behaviour is based on the interaction of all of its parts, as Dynamic Systems Theory (De Bot, 2008) proposes, then i) accentedness does in fact affect the intelligibility of L2 speech, and ii) both segmental and suprasegmental issues contribute to problems of comprehensibility. The main finding of this study is that whether speech is comprehensible and intelligible is based on the outcome of the interactions of its parts. It concludes that to examine individual linguistic aspects of an utterance as isolated phenomena is ineffective.