マイケル バー   Michael BARR   マイケル・バー    所属   京都外国語短期大学  キャリア英語科    職種   准教授
発表年月日	2024/08/26
発表テーマ	Chat-GPT for writing analysis: How can it assist the university student (and professor)?
会議名	EUROCALL 2024: CALL for Humanity
主催者	EUROCALL - The European Association for Computer Assisted Language Learning
学会区分	国際学会
発表形式	口頭（一般）
単独共同区分	共同
国際共著	国際共著
国名	スロバキア
開催地名	Trnava, Slovakia
開催期間	2024/08/26～2024/08/28
発表者・共同発表者	Barr, Michael and Dennisson, Jeanette
概要	The halls of higher education are full of discussions, concerns, and anxieties about how ChatGPT is going to upend teaching and learning methodologies for writing. On one side are professors who wish to ban the use of all translation and generative technologies, and on the other, those who recognize its inherent potential to benefit the teaching and self-directed learning of writing in higher education. One thing is for certain: ChatGPT cannot be removed from the university community, and despite anyone’s objections or opinions, it will be used for writing support and analysis by both instructors and students. How, then, can we most effectively harness and manage its potential in writing education? When aligned with designated frameworks for writing evaluation, such as the rubrics for band scores connected to the IELTS test, a ChatGPT-driven assessment of writing informs the teaching process, and enables instructors to devote their valuable time and energy to tasks and techniques which most profoundly affect outcomes. With the aim to test the feasibility of ChatGPT as a writing assessment tool, this study analyzed the English academic writing samples of Japanese undergraduates in medical and humanities programs by professors at three universities in Japan. A corpus of over 300,000 words of student writing activities, drafts, and short essays were collected, anonymized, and organized into sets which could then be evaluated. Each professor used the corpus to develop ChatGPT prompts which enable the evaluation of lexical complexity, grammar, and other aspects of student writing. These results will lead to informed teaching and self-directed learning in the classroom. Students who have access to customized and specific feedback of their written work are more motivated and informed about improving their essay writing capabilities. Therefore, as educators, our aim is to equip students with the most up-to-date and appropriate tools they need to access writing feedback. This analysis should be of interest to those who believe ChatGPT can automate and facilitate teaching of second language writing skills which, in the past, has been cumbersome and time-consuming to accomplish.