Enhancing University EFL Learners' Writing Performance: The Role of AI-Enhanced Goal-Setting, Feedback and Social Norm Interventions
Saved in:
| Title: | Enhancing University EFL Learners' Writing Performance: The Role of AI-Enhanced Goal-Setting, Feedback and Social Norm Interventions |
|---|---|
| Language: | English |
| Authors: | Yan Wang (ORCID |
| Source: | Journal of Computer Assisted Learning. 2026 42(1). |
| Availability: | Wiley. Available from: John Wiley & Sons, Inc. 111 River Street, Hoboken, NJ 07030. Tel: 800-835-6770; e-mail: cs-journals@wiley.com; Web site: https://www.wiley.com/en-us |
| Peer Reviewed: | Y |
| Page Count: | 30 |
| Publication Date: | 2026 |
| Document Type: | Journal Articles Reports - Research |
| Education Level: | Higher Education Postsecondary Education |
| Descriptors: | College Students, Second Language Learning, English (Second Language), Writing (Composition), Writing Achievement, Goal Orientation, Feedback (Response), Artificial Intelligence, Technology Uses in Education, Self Efficacy, Writing Attitudes, Academic Language, Writing Instruction, Instructional Effectiveness |
| DOI: | 10.1002/jcal.70179 |
| ISSN: | 0266-4909 1365-2729 |
| Abstract: | Background: English as a Foreign Language (EFL) learners often struggle to develop robust academic writing, especially in online settings with limited feedback and interaction. Existing AI writing tools show promise but are often tested in brief, isolated interventions, so little is known about which AI-supported strategies are most effective. Objectives: In a four-arm randomised controlled trial (N = 383), three AI-enhanced nudges (goal-setting, feedback, social norm) were compared with an active control across 18 × 90-min sessions (total dose = 27 h). Posttest group differences were large (Welch's F (3, 188.454) = 1447.280, p < 0.001, [eta-squared] = 0.895), with the goal-setting nudge yielding the highest writing performance (M = 20.24/25, 81.0% of maximum). Methods: University EFL students were randomly assigned to three AI-enhanced nudge conditions (goal-setting, feedback, social norm) or an active control in a pretest-posttest design. Interventions drew on social cognitive theory (SCT) and nonlinear dynamic language learning theory (NDLLT) and were delivered over 18 × 90-min sessions. Writing performance, motivation and self-efficacy were assessed with validated instruments, complemented by interviews and reflective essays. Results and Conclusions: Goal-setting nudges produced the largest gains in writing performance, motivation and self-efficacy, with feedback and social norm nudges also outperforming the active control. Effect sizes were large: goal-setting versus control [delta]M = 11.01, 95% CI [10.51, 11.52] for performance; [delta]M = 3.50, 95% CI [3.16, 3.83] for motivation; [delta]M = 3.22, 95% CI [2.95, 3.49] for self-efficacy. Students described AI-enhanced goal-setting as making tasks more manageable and progress more visible, while AI feedback and peer comparisons supported revision and engagement. Overall, the findings suggest that carefully designed, instructor-mediated AI nudges can substantially enhance EFL academic writing when used to supplement, rather than replace, human teaching. Centring on three research questions, RQ1 tested differential effects of the three nudges versus control on writing performance, motivation and self-efficacy; RQ2 examined learners' perceived effectiveness and experiences across modalities; RQ3 explored the correspondence between quantitative performance gains and qualitative learner insights. |
| Abstractor: | As Provided |
| Entry Date: | 2026 |
| Accession Number: | EJ1495861 |
| Database: | ERIC |
| Abstract: | Background: English as a Foreign Language (EFL) learners often struggle to develop robust academic writing, especially in online settings with limited feedback and interaction. Existing AI writing tools show promise but are often tested in brief, isolated interventions, so little is known about which AI-supported strategies are most effective. Objectives: In a four-arm randomised controlled trial (N = 383), three AI-enhanced nudges (goal-setting, feedback, social norm) were compared with an active control across 18 × 90-min sessions (total dose = 27 h). Posttest group differences were large (Welch's F (3, 188.454) = 1447.280, p < 0.001, [eta-squared] = 0.895), with the goal-setting nudge yielding the highest writing performance (M = 20.24/25, 81.0% of maximum). Methods: University EFL students were randomly assigned to three AI-enhanced nudge conditions (goal-setting, feedback, social norm) or an active control in a pretest-posttest design. Interventions drew on social cognitive theory (SCT) and nonlinear dynamic language learning theory (NDLLT) and were delivered over 18 × 90-min sessions. Writing performance, motivation and self-efficacy were assessed with validated instruments, complemented by interviews and reflective essays. Results and Conclusions: Goal-setting nudges produced the largest gains in writing performance, motivation and self-efficacy, with feedback and social norm nudges also outperforming the active control. Effect sizes were large: goal-setting versus control [delta]M = 11.01, 95% CI [10.51, 11.52] for performance; [delta]M = 3.50, 95% CI [3.16, 3.83] for motivation; [delta]M = 3.22, 95% CI [2.95, 3.49] for self-efficacy. Students described AI-enhanced goal-setting as making tasks more manageable and progress more visible, while AI feedback and peer comparisons supported revision and engagement. Overall, the findings suggest that carefully designed, instructor-mediated AI nudges can substantially enhance EFL academic writing when used to supplement, rather than replace, human teaching. Centring on three research questions, RQ1 tested differential effects of the three nudges versus control on writing performance, motivation and self-efficacy; RQ2 examined learners' perceived effectiveness and experiences across modalities; RQ3 explored the correspondence between quantitative performance gains and qualitative learner insights. |
|---|---|
| ISSN: | 0266-4909 1365-2729 |
| DOI: | 10.1002/jcal.70179 |
