表題番号：2019C-617
日付：2020/03/27
研究課題Automatically noticing pronunciation errors in Japanese learners of French
| 研究者所属（当時） | 資格 | 氏名 | |
|---|---|---|---|
| （代表者） | 国際学術院 国際教養学部 | 教授 | ドゥテ シルヴァン マッチュウ ジュリアン |
| （連携研究者） | IRIT, University of Toulouse III (France) | Associate Professor | Thomas Pellegrini |
| （連携研究者） | Archean Labs (France) | Researcher (Dr) | Maxime Le Coz |
| （連携研究者） | Archean Labs (France) | Researcher (Dr) | Lionel Fontan |
| （連携研究者） | TUFS (Japon) | PhD Candidate | Corentin Barcat |
- 研究成果概要
- The objective of this research base creation was to build on the data collected in our previous projects (Grant-in-Aid for Scientific Research (B) n°23320121 and n°15H03227) with the ultimate aim of automatically noticing pronunciation errors in Japanese learners of French. Our overall aim is to explore how automatic speech processing systems, relying on classifiers and preexisting lexical database, can use minimally annotated non-native speech corpora to notice pronunciation errors in Japanese learners of French for pedagogical purposes (e.g. computer-assisted pronunciation training tools). In 2019-2020, two tasks were achieved: 1) we used the PFC corpus (native) and the CLIJAF 1 corpus (non-native) for the preparation of native and non-native data for first-screen trial with automatic detection systems (native vs non-native); 2) we used the CLIJAF 2 corpus (non-native) and the LEXIQUE 3 (native) database to input phonetic transcription of inaccurate phonological productions by Japanese learners in spontaneous conversation in French for future testing.