LLM-Mediated Feedback in ESP Writing: Assessment, Learner Engagement, and Affective Response

Chaves-Yuste, Beatriz; Calle Martínez, Cristina

doi:10.22055/rals.2026.52587.3872

LLM-Mediated Feedback in ESP Writing: Assessment, Learner Engagement, and Affective Response

Document Type : Research Article

Authors

Beatriz Chaves-Yuste ¹

Cristina Calle Martínez ²

¹ Department of Applied Linguistics to Science and Technology, Escuela Técnica Superior de Ingenieros de Minas y Energía, Universidad Politécnica de Madrid, Madrid, Spain

² Department of English Studies: Linguistics and Literature, Facultad de Comercio y Turismo, Universidad Complutense de Madrid, Madrid

10.22055/rals.2026.52587.3872

Abstract

This study examines LLM-mediated, teacher-revised feedback in English for Specific Purposes writing. A pre-experimental repeated-measures design was implemented with 63 Tourism undergraduates, who completed two comparable writing tasks: one assessed and commented on by the teacher, and another supported by ChatGPT and subsequently revised by the teacher. Grades and questionnaire data on feedback quality, perceived learning, future use, affective engagement, supportive tone, and writing anxiety reduction were analysed using Wilcoxon signed-rank tests. Results showed significantly higher grades in the AI-assisted condition. Given the sequential design, this difference is interpreted cautiously. However, both feedback sources were perceived as largely comparable, while teacher feedback scored higher in supportive tone and anxiety reduction.

Keywords

Generative AI

ESP Writing

LLM-Mediated Feedback

Teacher Feedback.