This study compares the effects of teacher-provided feedback (TF) and ChatGPT-generated feedback (CGF) on English for Academic Purposes (EAP) writing revision outcomes and learner engagement. Forty engineering-major university students, randomly assigned into two EAP classes, were provided with either delayed TF or CGF based on a standardised rubric for revision after completing their first draft. Both the quantitative improvements in essay revisions and the qualitative aspects of student engagement, motivation, and emotional responses to the feedback were investigated. Results show that students who received TF achieved statistically greater improvement in their second draft and reported a more positive affective response in the questionnaire, whereas the CGF group improved less in their revisions and reported less-positive experiences. Despite that, the statistical analysis shows that the correlation between students’ improvement and their affective responses was not significant. This research contributes to the growing exploration of the correlation between written corrective feedback (WCF) and learner attitudes, highlights the importance of learners’ perceptions in feedback, and suggests that teacher feedback remains more effective in promoting both writing quality and learner engagement.