Author Information
Ronald William Marbun, Tokyo Denki University, JapanMakoto Shishido, Tokyo Denki University, Japan
Abstract
Narratives can be powerful tools for teaching English as a Second Language (ESL), but they must be carefully tailored to specific proficiency levels to be effective. This paper evaluates the capability of large language models (LLMs) to generate level-specific narratives and introduces a novel prompting method designed to enhance narrative generation for distinct educational levels. The method leverages English profiling frameworks, such as the Commoerbs, nouns) based on target levels. Usinn European Framework of Reference (CEFR), by introducing and restricting word forms (e.g., vg the Flesch-Kincaid and Dale-Chall readability frameworks, the study adopts a quantitative approach to assess efficacy. Narratives were generated for two proficiency levels: 7th grade (elementary) and 10th grade (intermediate). To evaluate the efficiency of the proposed method, it was compared to the widely used Instruction and Role-based Zero-Shot Prompting approach. Additionally, the study examined its performance when integrated with complementary techniques, such as the Tree of Thought method. Results demonstrate that the novel method improves narrative performance by 63% for elementary levels and 20% for intermediate levels. While the Tree of Thought method did not enhance efficiency, it contributed to a better balance of difficult word usage.
Paper Information
Conference: ECE2025Stream: Educational Research
This paper is part of the ECE2025 Conference Proceedings (View)
Full Paper
View / Download the full paper in a new tab/window
To cite this article:
Marbun R., & Shishido M. (2025) Enhancing Narrative Generation in ESL: Tailored Prompting for Proficiency-Specific Learning ISSN: 2188-1162 The European Conference on Education 2025: Official Conference Proceedings (pp. 145-155) https://doi.org/10.22492/issn.2188-1162.2025.13
To link to this article: https://doi.org/10.22492/issn.2188-1162.2025.13
Comments
Powered by WP LinkPress