Related Experiment Videos
Effects of feedback literacy scripts and a second-rater mechanism on EAP writing revision in generative AI-supported
1Foreign Language Teaching Department, Hainan Vocational University of Science and Technology, Haikou, Hainan, China.
Introduction:
As generative artificial intelligence increasingly enters university English for Academic Purposes (EAP) writing instruction, a key question is how AI-generated feedback can be translated into effective revision. Existing studies have mainly examined generative AI as a tool for feedback delivery, writing support, and instructional efficiency, whereas the links among feedback uptake, revision depth, and delayed retention remain insufficiently explored.
Methods:
To address this gap, this study developed a large language model-driven classroom simulation framework and adopted an exploratory 2 × 2 design. Two treatment variables were examined: whether a feedback literacy script was provided and whether an AI second-review mechanism was introduced. The study simulated 100 B2-level writers across different experimental conditions and tracked writing scores, effective feedback uptake, revision depth, and retention performance across Task 1, Task 2, and delayed Task 3.
Results:
The results revealed four main patterns. First, the feedback literacy script produced a clear positive effect during the initial revision stage. In Task 1, the scripted group gained 5.75 more points on average than the non-scripted group, and the regression coefficient was 7.09, with an exact permutation test yielding p = 0.0286. Second, this effect weakened in Task 2. Third, the script did not substantially increase effective feedback uptake, but it redirected revision toward the argument level; argument-level revision was positively associated with score gain (r = 0.755, p = 0.0304). Fourth, the second-review mechanism did not produce a stable benefit.
Discussion:
Overall, the findings suggest that improvement in AI-supported EAP revision depends less on the addition of more feedback sources than on learners' ability to process feedback through structured decision-making. Feedback literacy scripts may therefore be more useful as cognitive and procedural scaffolds for revision than as simple prompts for adopting AI-generated feedback.
Related Concept Videos
Non-equilibrium in the Cell
Sources of Self-Esteem II: Performance Feedback
Self-Efficacy
Negative and Positive Feedback
The Sense of Self: Reflected Self-Appraisal and Social Comparison