Comparing Learners’ Perceptions of AI-Enabled Writing Feedback and Human Teacher Feedback in IELTS Writing Preparation

Abstract

This study investigated International English Language Testing System (IELTS) Writing learners’ perceptions of AI-enabled writing feedback and human teacher feedback in relation to perceived writing improvement, IELTS Writing areas, and trust. Using an explanatory sequential mixed-methods design, questionnaire data were collected from 130 learners at a private IELTS teaching center in Ho Chi Minh City, Vietnam, and follow-up interviews were conducted with eight purposively selected participants. The questionnaire used a paired-source structure, enabling learners to evaluate AI feedback and human teacher feedback across parallel IELTS Writing dimensions. Quantitative data were analyzed in RStudio through reliability analysis, descriptive statistics, paired-samples t-tests, and Cohen’s dz, while interview data were analyzed in QualCoder using hybrid deductive–inductive thematic analysis. The findings showed no significant difference between AI feedback and human teacher feedback in overall perceived writing improvement. However, AI feedback was rated higher for grammar correction, vocabulary improvement, and revision speed, whereas human teacher feedback was rated higher for coherence and cohesion, task response, idea development, and trust. The integrated findings indicate differentiated, task-sensitive perceptions of the two feedback sources and perceived complementarity in how some learners combined them, although the study did not evaluate objective writing outcomes.