While learners desire to acquire so comprehensible pronunciations as to make themselves understood smoothly, acquisition often becomes difficult because, outside of classrooms, it is not rare that learners can hardly find chances to talk in the target language. Even when they talk to native speakers, they may receive only lenient or superficial suggestions from native speakers. How can learners know native speakers' honest perception on their utterances? In this paper, shadowing is introduced not to learners but to native listeners, who are asked to shadow learners' utterances. Since shadowing is as simultaneous repetition as possible, it is expected that native listeners' perceived comprehensibility can be measured objectively as smoothness of natives' shadowings. Experiments show that 1) shadowers' subjective assessment of learners' speech and that of their shadowings are highly correlated and that 2) the former is more correlated with the GOP scores of natives' shadowings than that of learners' speech. These results indicate that it is valid to regard comprehensible pronunciation as shadowable pronunciation.
Cite as: Inoue, Y., Kabashima, S., Saito, D., Minematsu, N., Kanamura, K., Yamauchi, Y. (2018) A Study of Objective Measurement of Comprehensibility through Native Speakers' Shadowing of Learners' Utterances. Proc. Interspeech 2018, 1651-1655, doi: 10.21437/Interspeech.2018-1860
@inproceedings{inoue18_interspeech,
author={Yusuke Inoue and Suguru Kabashima and Daisuke Saito and Nobuaki Minematsu and Kumi Kanamura and Yutaka Yamauchi},
title={{A Study of Objective Measurement of Comprehensibility through Native Speakers' Shadowing of Learners' Utterances}},
year=2018,
booktitle={Proc. Interspeech 2018},
pages={1651--1655},
doi={10.21437/Interspeech.2018-1860}
}