VietTESOL International Convention 2026

Automated writing feedback is increasingly used in second-language writing, but less is known about how explanation-based feedback shapes learners’ revision decisions and sense of ownership. This explanatory mixed-methods study examined Vietnamese tertiary EFL students’ engagement with ProWritingAid in an English for Specific Purposes writing course. Sixty-eight second-year English majors completed three argumentative essays over eight weeks. Data included pre-post surveys, coded draft-revision comparisons, reflection notes, and interviews with 18 volunteers. Descriptive scale means increased for both trust (3.11 to 3.84) and authorship perception (3.22 to 3.97). Scale-level paired-samples t-tests showed significant within-participant increases for both trust, t(67) = -25.50, p < .001, d = -3.10, and authorship perception, t(67) = -21.00, p < .001, d = -2.55. Supplementary item-level analyses showed significant increases across all trust and authorship items. Across 2,040 feedback-response instances, students most often modified suggestions (56.0%), followed by direct acceptance (29.0%) and rejection (15.0%). Qualitative analysis identified critical filtering, empowered revision, and ethical caution. The findings suggest that explanation-based automated writing feedback was associated with selective, inspectable revision practices rather than simple compliance. The study contributes to writing research by showing how learners negotiate trust, agency, and authorship through observable revision decisions in AI-mediated L2 writing.