생성형 AI 시대 책임 있는 담화의 양상과 형성 조건: 고등학교 영어 독해 수업을 중심으로 = Accountable Talk in the Era of Generative AI: Patterns and Conditions in High School English Reading Classes.

다국어 초록 (Multilingual Abstract)

This qualitative case study with supplementary descriptive quantitative data explored how accountable talk was organized in high school English reading classes using AI-generated paraphrases for comparison. Participants were 287 second-year students in 12 classes at one high school taught by the teacher-researcher over ten weeks. Classroom transcripts served as the primary data, while ClassDojo participation records and assessment scores were used as contextual indicators. Guided by Michaels et al.’s framework of accountability to knowledge, rigorous thinking, and the learning community, the analysis found recurring interactional sequences, in which answer confirmation was extended into requests for textual evidence, explanation, and reasoning. AI-generated paraphrases served not as answer substitutes but as objects for comparison and verification. Participation norms, follow-up questions, and redistributed participation opportunities helped broaden student participation in accountable talk. Assessment data showed modest score changes but were interpreted only descriptively because the study lacked a control group and did not control test difficulty. The findings suggest that English reading instruction can support accountable talk when students are asked to evaluate AI outputs against texts and justify their interpretations and reasoning within a structured discourse community.