Year: 2025
Type: article
Abstract: Artificial intelligence (AI)-powered large language models, such as ChatGPT, are increasingly utilized in education, particularly for language learning. However, the effectiveness of AI-assisted learning is highly dependent on the quality of user-generated prompts. This study examines how structured instruction in prompt engineering affects English as a Foreign Language (EFL) students' ability to craft effective ChatGPT prompts, their engagement with AI, and their confidence in AI-assisted learning, culminating in the development of skills that will increasingly be in demand in the workplace. Using a three-phase intervention, students first generated self-created prompts, then used teacher-provided prompts based on the Nazari and Saadi framework, and finally received explicit instruction in structured prompt writing. After receiving instruction in prompt engineering, there were significant improvements in student prompt quality. Additionally, students’ confidence in prompt generation significantly increased after instruction. However, teacher-provided prompts alone did not significantly enhance student perceptions of ChatGPT’s usefulness for grammar or vocabulary improvement. These findings underscore the importance of explicit instruction in AI prompt engineering, suggesting that providing structured frameworks enhances student engagement, prompt effectiveness, and confidence in AI-assisted learning. Implications for AI-based language instruction and future research directions are discussed.
Institution: Johannes Kepler University of Linz
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
SDG: Quality education
Open Access status: diamond
OpenAlex ID: W7087106984