Perceiving the fluency of native and non-native speakers

Within the language testing practice, the fluency level of test-takers is commonly assessed by human raters. The process through which raters come to their conclusions has long been subject of research. One approach to this issue has been the correlational analysis of acoustic measures and subjective judgments: which disfluencies (pauses, fillers, corrections) play a large role in fluency assessment and which do not? This approach has mainly been concerned with non-native speech, since native speakers are commonly considered to be ‘fluent’ in their mother tongue. But disfluencies also occur in native speech. Therefore, the focus of this presentation lies on the perception of fluency in non-native and native speech. The first set of experiments involved the analysis of non-native fluency perception. Acoustic measurements of non-native speech were compared against subjective fluency ratings. It was found that fluency raters largely depend on the acoustics of the speech signal, mostly on pauses and speed. Subsequently, the perceptual salience of pauses and speed of speech was evaluated. It was hypothesized that pauses and speed of speech may be easy to perceive in the speech signal and therefore play a large role in fluency ratings. The results showed that perceptual saliency alone could not account for why fluency raters largely depend on pause and speed characteristics. The second set of experiments compared native and non-native fluency using phonetic manipulations. Altering the pause or speed characteristics of the speech signal had strong effects on fluency ratings, but these effects did not differ across native and non-native speech. The observations from these empirical studies lead us to conclude that fluency ratings are largely dependent on the acoustics of the speech signal, that pause and speed characteristics are the main contributors to fluency judgments and that these contributions are similar across native and non-native fluency perception