Joining-in-type CALL system using dialogue agent

2021 Fiscal Year Final Research Report

• Project/Area Number: 19K00927
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 02100:Foreign language education-related
• Research Institution: Doshisha University
• Principal Investigator: Yamamoto Seiichi 同志社大学, 研究開発推進機構, 嘱託研究員 (20374100)
• Co-Investigator(Kenkyū-buntansha): 加藤 恒夫 同志社大学, 理工学部, 教授 (60607258)
• Project Period (FY): 2019-04-01 – 2022-03-31
• Keywords: CALL / implicit learning / interactive alignment

Outline of Final Research Achievements

Automatic speech recognizer (ASR) is indispensable for dialogue-based computer-assisted-language learning (DB-CALL) system to give learners adequate corrective feedback. However, ASR of the second language learners is still a challenge, because their speech contains not just pronouncing, lexical, grammatical errors, but is sometimes totally disordered. This research proposed a novel DB-CALL system using two conversational agents, one as a teacher and the other as an advanced learner. The system is designed to simulate multiparty conversation, expecting implicit learning and enhancement of predictability of learners' utterance through an alignment similar to "interactive alignment" observed in human-human conversation. The experiment demonstrated that repetitive queries of specific grammatical expressions consistently improved the correct use of the expressions than simple single query, and also suggests that its retention effect is higher than conventional repeating training.

Free Research Field

音声言語処理

Academic Significance and Societal Importance of the Research Achievements

第二言語での音声会話では、限られた時間内に発話内容とその表現形式及び使用語彙の選択を行って発声することが求められ、これを実現するための発話行為の手続き知識化が必要となる。従来この過程を強化する訓練として教師の発話をrepeatingする方法が主に使用されている。本研究により適切な表現形式で行われる模範的な会話を聴取し，その表現形式を借用して応答を行う質問応答による訓練が手続き知識化に効果的であることが示された．本結果は深い処理水準を経た記憶は、浅い処理水準のみを経た記憶より定着し易いと実験心理学の知見とも合致しており，効果的な外国語の発話訓練を提供する．