Evaluating prosodic similarity as a means towards L2 teacher’s prosodic control training

Olivier Nocaudie, Corine Astésano


Studies on professional impersonators and average speakers has underlined that speech imitation proficiency varies across speakers. Imitation in speech supposes that a speaker succeeds in reproducing specific features of the perceived speech. Because of the inherent variability of human speech behaviors, the question lies open whether different speakers can accurately imitate phonetic features, and more specifically, prosodic patterns. This exploratory study proposes to test imitation of F0 contours of four sentences originally pronounced by a female speaker, thereafter by four naïve listeners undertaking three different tasks: mere repetition, imitation, and exaggeration of the perceived sentences. On the one hand, imitated sentences and models were time-warped and objective comparisons were performed using two (dis)similarity measures reported in the literature. On the other hand, a panel of 15 listeners had to perceptually evaluate the same set of sentences during an AX similarity judgement task. Similarity scores were used to build multiple rankings in order to observe the correlation between these two tests’ rankings and to evaluate prosodic imitation proficiency across speakers/listeners. This research has implication for L2 phonetic correction using the Verbo-Tonal Method, which requires excellent prosodic awareness and control by the teacher in the production of lexicalized and delexicalized sentences.


DOI: 10.21437/SpeechProsody.2016-6

Cite as

Nocaudie, O., Astésano, C. (2016) Evaluating prosodic similarity as a means towards L2 teacher’s prosodic control training. Proc. Speech Prosody 2016, 26-30.

Bibtex
@inproceedings{Nocaudie+2016,
author={Olivier Nocaudie and Corine Astésano},
title={Evaluating prosodic similarity as a means towards L2 teacher’s prosodic control training},
year=2016,
booktitle={Speech Prosody 2016},
doi={10.21437/SpeechProsody.2016-6},
url={http://dx.doi.org/10.21437/SpeechProsody.2016-6},
pages={26--30}
}