As generative artificial intelligence (GenAI) becomes increasingly embedded in writing assessment, teachers are asked not only to use new tools but also to negotiate what counts as valid judgment, meaningful feedback, and professional authority in AI-mediated workflows. This study examines teacher agency in GenAI-mediated writing assessment through a duoethnographic inquiry situated in Vietnamese higher education English as a foreign language (EFL) writing courses. Drawing on four months of reflective writing, dialogic response threads, assessment artifacts, and prompt records, we analyze how two teacher-researchers encountered and interpreted friction across AI-assisted scoring, feedback drafting, and platform-based assessment workflows. Conceptually, the study defines teacher agency as an ecological and temporal achievement enacted through situated judgment, disrupted by sociotechnical constraints, and reimagined through teachers' efforts to reconfigure their roles in relation to AI output. The findings show that agency was enacted through rubric-based interpretation, disrupted when AI outputs narrowed assessment evidence or repositioned teachers as confirmers of machine judgment, and reimagined through re-voicing, contestation, and ethical mediation. The study argues that GenAI did not simply reduce assessment labor, but it redistributed it into new forms of criterial verification, feedback calibration, data-sensitive decision-making, and workflow resistance.
Generic large language models (LLMs), including ChatGPT, have recently been applied in second language writing instruction due to their capability to provide immediate feedback on students' writing. However, these models still exhibit limitations, including 'hallucinations' and overcorrection. In response, we developed Dr. Write, a domain-specific AI-powered writing feedback system trained on the world's largest corpus of German writing by Chinese learners. This study outlines the pedagogical rationale behind the design of Dr. Write and empirically evaluates its effectiveness through a quasi-experimental design. A total of 124 L2 German learners were assigned to three feedback conditions: a domain-specific LLM (Dr. Write), a generic LLM (Qwen), and teacher feedback. The results showed that both GAI feedback conditions outperformed teacher feedback in learners' cognitive and behavioural engagement with feedback, but not in affective engagement. Dr. Write was also associated with significant gains in writing self-efficacy, particularly in self-regulatory efficacy and performance self-efficacy, as well as in clause-level writing accuracy. Qualitative findings further suggest that its guided, proficiency-aligned feedback fostered active revision, self-monitoring, and reflection. The study highlights the pedagogical potential of domain-specific LLMs for L2 writing and offers a transferable framework for future research on AI-supported feedback in educational settings.
L2 grit and growth mindset (GM) are increasingly linked to engagement in language education, yet prior work has largely assumed linear effects and focused on English learning. This study examined whether the alignment between L2 grit and GM, along with their potential nonlinear joint effects, predicts engagement among Thai learners of Chinese in a Languages Other Than English (LOTE) context. Survey data from 325 Chinese-as-an-additional-language (CAL) learners in Thailand were analyzed using polynomial regression with response surface analysis (RSA) in R to model congruence, incongruence, and interaction effects. A congruence hypothesis was not supported: results revealed that optimal CAL engagement did not occur simply when L2 grit and GM were aligned. Instead, CAL engagement increased most sharply as both constructs rose together. Effects were asymmetrical: GM exhibited a largely linear relationship with CAL engagement, whereas L2 grit showed a curvilinear pattern, with diminishing returns at higher levels, suggesting that additional grit may not always correspond to proportionally higher engagement. CAL engagement under the low-low configuration remained moderate, consistent with coping or 'survival-mode' engagement under contextual pressure. Educators should emphasize the quality of learners' persistence over mere quantity, guiding them to employ adaptive strategies, set flexible goals, and make timely adjustments, while also harnessing the synergistic effects of L2 grit and GM to sustain CAL learners' engagement. Extending RSA to an underexplored LOTE context, this study reveals that the relationship between L2 grit, GM, and CAL engagement is configuration-dependent, with evidence pointing to nuanced nonlinear effects.
