ABSTRACT
This chapter presents the key results, implications, and limitations of this study and suggestions for future research. The system-centric evaluation showed relatively satisfactory flagging precision and suggestion precision but low recall of Grammarly in providing feedback on correctness issues in research papers. The user-centric evaluation returned results indicating that revision with Grammarly could effectively improve the linguistic accuracy of research papers, based on both the error-correction success ratios and the regression analysis results. Meanwhile, the engagement study found that students demonstrated in revision active behavioral, cognitive, and emotional engagement, but little social engagement, with Grammarly use. Their engagement profiles were related to various individual factors (i.e., independence awareness, genre awareness, and research publication experience) and contextual factors (i.e., error types, automated writing evaluation tool design, and disciplinary conventions). The present study has theoretical, methodological, and practical implications for the use of Grammarly by academic writing course instructors and students and for the improvement of automated writing evaluation tools. Limitations are discussed regarding the sample and the stimulated recall data collection. Suggestions for future research are provided.
