Author Information
Harumi Kashiwagi, Kobe University, JapanMin Kang, Kobe University, Japan
Abstract
This study examined how a three-step learning cycle—(1) independent English production, (2) reference to AI-generated samples, and (3) self-correction—affected university students’ psychological barriers and “noticing” in AI-generated image description tasks. A total of 129 university students were divided into two groups (n = 65, n = 64) and performed two identical tasks while referring to AI-generated sample descriptions based on two different educational frameworks. Participants rated the AI-generated samples on a 5-point Likert scale across vocabulary, grammar/syntax, and organization, followed by an overall survey. Key findings revealed that: First, over 90% of participants perceived the cycle positively, as the subsequent revision process lowered anxiety about mistakes and fostered psychological security. Second, both groups assigned usefulness ratings of 4.4 or higher on average across all dimensions. While the ratings for vocabulary and grammar/syntax showed no significant differences between the groups, a significant difference was observed in one task, in which structures explicitly describing spatial relationships (e.g., “in the middle,” “to the left,” or “on the right side”) were rated as particularly helpful. Furthermore, over 90% reported a sense of improvement from revising their own work using AI-generated samples as a point of reference rather than merely viewing model answers. Additionally, 60–70% identified AI-generated similar images and texts as effective support resources. These results suggest that sample English sentences generated by generative AI can serve as triggers to promote autonomous awareness in learners, while also providing design principles for psychologically supportive AI-assisted language learning environments.
Paper Information
Conference: ECE2026Stream: Design
Published: Wednesday, September 23, 2026
This paper is part of the ECE2026 Conference Proceedings (View)
Full Paper
View / Download the full paper in a new tab/window
To cite this article:
Kashiwagi H., & Kang M. (2026) English Production Learning Cycle Using Generative AI in Image Description Tasks: Focusing on Lowering Psychological Barriers and Promoting Autonomous Noticing ISSN: 2188-1162 The European Conference on Education 2026: Official Conference Proceedings (pp. 961-974) https://doi.org/10.22492/issn.2188-1162.2026.73
To link to this article: https://doi.org/10.22492/issn.2188-1162.2026.73
Comments
Powered by WP LinkPress