Work

Generative AI feedback use and English writing self-regulation among Chinese non-English majors: a two-wave study of feedback literacy and writing anxiety

Year: 2026
Type: article
Study design: Observational Study
Abstract: Introduction Generative AI chatbots now give students instant commentary on their writing, yet little is known about how engaging with that commentary relates to the self-regulatory capacities on which writing development depends. This study examined a chain mediation model in which generative AI feedback use is associated with writing self-regulation through feedback literacy and, in turn, second language writing anxiety. Methods Non-English-major undergraduates at three universities in Henan Province, China, completed two surveys 12 weeks apart in the 2026 spring semester (wave 1 N = 650; 520 retained, 80.0%). Wave 1 assessed generative AI feedback use and feedback literacy; wave 2 assessed writing anxiety and writing self-regulation. Serial mediation analysis used 5,000 bias-corrected bootstrap resamples, with item-level validity checks and re-estimations without item parcels added as sensitivity analyses. Results Generative AI feedback use was positively associated with later writing self-regulation [ c = 0.265, 95% CI (0.190, 0.339)]. The total indirect effect was 0.128 [0.094, 0.167], or 48.3% of the total association: 0.086 [0.053, 0.122] via feedback literacy, 0.024 [0.008, 0.049] via writing anxiety, and 0.018 [0.009, 0.031] for the serial route. The direct association remained significant (c′ = 0.137). Focal paths kept their signs and significance when item parcels were removed and when only feedback-seeking items defined the predictor. Discussion These estimates describe associations rather than effects of an intervention. With no wave-1 baselines of anxiety or self-regulation and with two within-wave orderings resting on theory, the design cannot establish temporal sequence or within-person change. The pattern is consistent with a theoretically proposed ordering; the pedagogical suggestions are recommendations only, not demonstrated outcomes. The generative AI feedback use scale is a preliminary study-specific measure.
Language: English
Open Access status: gold
Estimated APC paid: $3,801
OpenAlex ID: W7235899437
FWCI:
Cites: 56
Cited by:
Related to: 10
Parts: 1