The processing and evaluation of fluency in native and non-native speech

Abstract

Disfluency is a common characteristic of spontaneously produced speech. Disfluencies (e.g., silent pauses, filled pauses [uh’s and uhm’s], corrections, repetitions, etc.) occur in both native and non-native speech. There appears to be an apparent contradiction between claims from the evaluative and cognitive approach to fluency. On the one hand, the evaluative approach shows that non-native disfluencies have a negative effect on listeners’ subjective fluency impressions. On the other hand, the cognitive approach reports beneficial effects of native disfluencies on cognitive processes involved in speech comprehension, such as prediction and attention. This dissertation aims to resolve this apparent contradiction by adopting both the evaluative and cognitive approach to the study of native and non-native fluency. With respect to the evaluation of fluency, it is found that pause and speed characteristics contribute most to subjective fluency judgments, and that there is no difference in the way listeners weigh the fluency characteristics of native and non-native speech. Regarding the processing of fluency, it is found that listeners use uhm’s to predict what the next word will be, but only when listening to a native speaker. Finally, it is found that both native and non-native uhm’s elicit heightened attention to the following linguistic content. In the concluding chapter of this dissertation, an integrative account of native and non-native fluency perception is proposed ‒ informative to both language testing practice and cognitive psycholinguists. Because disfluencies have been argued to arise from speech production difficulty, these symptoms of speaker trouble (i.e., individual disfluencies in the speech signal) negatively affect listeners’ impressions of the speaker’s fluency level. Nonetheless, listeners can, in a very clever way, make use of the disfluent character of spontaneous speech for comprehension. Disfluencies may elicit anticipation of - and heightened attention to - subsequent linguistic content. Taking the studies together, this dissertation testifies to the notion that speech performance matters: communication through spoken language does not only depend on what is said, but also on how it is said and by whom.