Key points are not available for this paper at this time.
Purpose The rapid development of generative AI (GenAI) technologies, such as large language models and AI-based content generation tools, has drawn increasing attention in ELT due to their potential for creating adaptable and cost-effective resources. This study investigates the potential of GenAI for developing comprehensive ELT instructional materials.Methodology Adopting a design-based research approach, the study was conducted over two iterative cycles with 639 and 577 first-year university students enrolled in a remote English course. Intact program-based groups were assigned to control and treatment conditions. Quantitative pre- and post-test data were analyzed using ANCOVA to control for initial proficiency, and qualitative data were gathered from interviews with 4 instructors in Cycle 1 and 5 instructors in Cycle 2.Findings The analyses indicated that AI-generated materials did not yield practically significant differences in student learning outcomes. In Cycle 1, although the difference was statistically significant (p = .005), the effect size was negligible (partial η² = .009), and in Cycle 2, there was no significant difference (p = .715, partial η² ≈ .000). These results and interviews suggest that AI-generated materials maintain learning outcomes, while offering clear advantages in efficiency, customizability, and flexibility. Overall, despite certain limitations, GenAI offers considerable opportunities for customized ELT resources.Originality/Value Unlike previous studies that focused on a single type of AI-generated content, this study explores the integration of text, visuals, and audio in experimental ELT material development. It advances multimodal instructional design in ELT, positioning GenAI as a co-designer that supports teachers in creating learning experiences.
Aguş et al. (Tue,) studied this question.