Presentation From Error Correction to Writer Development: EFL Student Writing Before and After AI more

This presentation compares student writing produced before the availability of generative AI tools with writing produced in the current AI-integrated classroom, drawing on samples from the same intermediate-level English writing course at a Japanese university. The study examines two comparable cohorts: approximately 30 students in 2022 (pre-AI) and in 2025 (post-AI). Both groups used the same textbook, writing tasks, and instructional approach. In 2025, “AI support” referred primarily to guided use of ChatGPT for language refinement with clear classroom policies on responsible use. A comparison of selected writing samples shows a reduction in grammatical error frequency per 100 words and improved readability scores in the AI-supported cohort. Before AI integration, instructors devoted substantial time to correcting surface-level errors, limiting attention to organization, argumentation, and writer voice. With AI-assisted drafts, grammatical accuracy improved sufficiently to allow more sustained focus on higher-order concerns, although challenges in coherence and rhetorical effectiveness remain. By presenting anonymized samples from both periods, this study demonstrates how AI reshapes instructional priorities. Rather than replacing writing instruction, AI mediates it, enabling a shift from error correction toward content development and rhetorical awareness. Attendees will gain practical strategies for designing AI-integrated writing classes that prioritize strengthening ideas and structure over sentence-level correction.