Humanity in AI-assisted second language writing: Agency, ethics, and pedagogical design
Organizer
Cong Zhang, Zhejiang University, China
Presenters
Cong Zhang, Zhejiang University, China
Yachao Sun, Duke Kunshan University, China
Jian Xu, Sichuan International Studies University, China
Xiaodong Zhang, Beijing Foreign Studies University, China
Yabing Wang, Guangdong University of Foreign Studies, China
The rapid integration of generative artificial intelligence (GenAI) into second language (L2) writing has fundamentally reshaped how texts are produced, evaluated, and taught. While much public and scholarly discourse has centered on what AI can do—its efficiency, accuracy, and scalability—less attention has been paid to what AI means for the human dimensions of writing: agency, voice, judgment, responsibility, and ethical action. This colloquium brings these human concerns to the foreground by examining AI-assisted second language writing not as a technical problem to be managed, but as a pedagogical, ethical, and relational practice to be designed.
Across diverse contexts—university and secondary education, EFL and EAL settings, teacher and learner perspectives—this colloquium advances a shared premise: humanity in AI-assisted writing does not reside in rejecting technology, nor in uncritical adoption, but in how humans actively mediate, negotiate, and reassert meaning-making in AI-rich environments. The five presentations collectively challenge instrumental and surveillance-driven approaches to AI (e.g., detection-first regimes, grammar-only feedback, automation of judgment) and instead highlight the roles of teachers as ethical designers, learners as agentive decision-makers, and pedagogy as a site for cultivating voice, responsibility, and critical engagement.
Methodologically, the colloquium brings together qualitative, mixed-methods, and action research approaches, offering both empirical insight and pedagogical innovation. Conceptually, it reframes AI as neither replacement nor authority, but as a socio-technical participant whose influence must be interpreted, constrained, and reshaped through human values and professional judgment.
Together, these studies argue that the future of AI-assisted second language writing depends not on better tools alone, but on sustaining humanity through ethical design, learner agency, meaningful feedback engagement, and pedagogies that prioritize voice, purpose, and learning over automation and control.
Presentation 1: Teachers as Ethical Designers: Reclaiming Pedagogical Agency in AI-Assisted Second Language Writing
Cong Zhang, Zhejiang University, China
The rapid integration of generative AI into second language (L2) writing instruction has intensified long-standing questions about teacher agency, professional responsibility, and the ethical foundations of pedagogy. While much existing research emphasizes what AI tools can do for writing development, far less attention has been paid to how teachers actively design, regulate, and ethically position AI within classroom practice. This study reconceptualizes L2 writing teachers not as passive adopters of AI technologies, but as ethical designers who exercise pedagogical judgment in shaping human–AI interactions.
Drawing on qualitative data from in-depth interviews, instructional artifacts, and reflective narratives of L2 writing teachers, the study examines how teachers negotiate ethical concerns such as authorship, learning legitimacy, fairness, and student dependence when integrating AI into writing tasks and feedback practices. The analysis foregrounds teachers’ design decisions, including when and how AI use is permitted, how AI-assisted tasks are framed, and how AI-generated outputs are repositioned as resources for learning rather than substitutes for writing.
Findings show that teachers’ agency is enacted through ongoing ethical design work that reasserts human judgment, pedagogical values, and relational responsibility in AI-mediated writing environments. Rather than relinquishing authority to AI systems, teachers strategically position themselves as curators, interpreters, and moral anchors of writing instruction. This work reshapes teachers’ professional identities and highlights humanity as an irreducible dimension of AI-assisted pedagogy.
By theorizing teachers as ethical designers, this study contributes to emerging scholarship on humanity in AI-assisted writing and offers a framework for understanding teacher agency beyond instrumental adoption. Pedagogical implications are discussed for sustaining human-centered L2 writing instruction in increasingly AI-rich educational contexts.
Presentation 2: Beyond Detection in GenAI-Mediated EAL Writing Education
Yachao Sun, Duke Kunshan University, China
As generative artificial intelligence (GenAI) becomes embedded in English as an additional language (EAL) writing education, institutional responses have increasingly emphasized detection and policing. This study critically examines how such detection-oriented regimes shape students’ writing practices and, more importantly, how they reconfigure the human dimensions of learning and assessment.
Using screen recordings, written drafts, and interviews with five Chinese EAL university students, the study identifies four recurrent GenAI functions in students’ writing: brainstorming, structuring, occasional drafting, and revising. However, students’ engagement with these functions was consistently calibrated against perceived risks of being flagged by AI-detection tools. To mitigate these risks, participants shuttled between ChatGPT, Grammarly, and self-paraphrasing strategies to “humanize” AI-generated output. In doing so, they often reallocated effort away from idea development and argumentation toward surface-level revision and textual disguise.
These findings problematize a detection-first approach in three ways. First, such regimes encourage strategic but pedagogically hollow manipulation of text. Second, they blur distinctions between originality and plagiarism by foregrounding algorithmic judgment over writer intention and learning processes. Third, they risk shifting writing assessment toward product-centric criteria that marginalize voice, creativity, and rhetorical quality.
The study argues for moving beyond detection toward critical AI literacy in EAL writing education. Rather than policing AI use, pedagogy should support learners in making informed, reflective decisions about when, why, and how to engage with GenAI. Prioritizing voice, argument quality, and learning-oriented writing practices is essential for sustaining humanity in AI-mediated EAL instruction.
Presentation 3: Understanding Agency in AI Use and Engagement with AI Feedback among University L2 Learners
Jian Xu, Sichuan International Studies University, China
This mixed-methods study examines second language (L2) learners’ agency in AI use and their engagement with AI-generated feedback in university learning contexts. While AI feedback is often framed as efficient and supportive, less is known about how learners exercise agency in interacting with it and how different dimensions of agency relate to engagement.
The study draws on questionnaire data from 1,166 L2 learners and semi-structured interviews with six focal participants. Quantitative analyses (descriptive statistics, repeated-measures ANOVA, and structural equation modeling) reveal significant variation across agency dimensions. Learners reported the highest levels of agency in actions (e.g., deciding whether to use AI), followed by abilities and mentalities. In terms of engagement, cognitive and behavioral engagement were rated higher than affective engagement. Actions significantly predicted both cognitive and behavioral engagement, while mentality predicted affective and cognitive engagement; perceived abilities showed no significant predictive power.
Qualitative findings provide depth to these patterns. Learners described exercising agency by selectively adopting AI feedback, preserving decision-making authority, and viewing AI as a collaborator rather than an authority. Importantly, participants reported that AI created an emotionally safe environment that reduced anxiety and encouraged sustained practice, particularly when AI use was framed as supportive rather than evaluative.
Together, the findings suggest that agency in AI-assisted L2 learning is not monolithic but multidimensional and relational. Supporting humanity in AI-mediated feedback requires pedagogical designs that strengthen learners’ sense of control, reflection, and responsibility. Implications are discussed for fostering learner agency and meaningful engagement with AI feedback in L2 writing instruction.
Presentation 4: Optimizing ChatGPT-Assisted Writing for Secondary L2 Students’ Academic Transition: A Systemic Functional Linguistics Approach
Xiaodong Zhang, Beijing Foreign Studies University, China
Generative AI tools such as ChatGPT are increasingly present in second language (L2) classrooms, yet their role in supporting secondary students’ transition to academic writing remains underexplored. While ChatGPT can assist with grammar and vocabulary, its default support often emphasizes surface-level language forms, offering limited guidance for the meaning-making demands of advanced academic writing.
This paper argues that integrating systemic functional linguistics (SFL) with ChatGPT can better support secondary L2 students’ academic transition. Grounded in Halliday’s (1993) SFL framework, the paper conceptualizes writing as a process of meaning construction shaped by the interaction of ideational, interpersonal, and textual resources. From this perspective, effective academic writing requires more than accuracy; it requires control over how content, stance, and organization work together in context.
The paper reviews literature on SFL-informed L2 writing pedagogy and examines how ChatGPT’s affordances can be reoriented through an SFL lens. It then proposes practical strategies and illustrative cases in which teachers guide students to use ChatGPT not only to refine language forms but also to analyze genre, develop arguments, and make purposeful rhetorical choices.
By aligning ChatGPT’s technological capabilities with SFL’s theoretical insights, the paper advances a human-centered framework for AI-assisted writing instruction. This approach positions AI as a mediational resource rather than an authority, enabling secondary L2 students to develop the conceptual and linguistic foundations necessary for successful academic writing in higher education.
Presentation 5: Enhancing Engagement with GenAI Feedback in EFL Narrative Writing: A Teacher-Led Action Research
Yabing Wang, Guangdong University of Foreign Studies, China
Although GenAI is rapidly reshaping L2 writing feedback, students’ engagement with AI suggestions often remains superficial, characterized by grammar-only edits or uncritical copy-paste revisions. This teacher-led action research investigates how instructional design can foster deeper engagement with GenAI feedback in a college EFL narrative writing course.
Across three iterative cycles in a semester-long module, the instructor implemented a sequence of pedagogical interventions: (1) rubric anchoring that foregrounded two narrative dimensions per task (e.g., plot coherence, showing details) alongside an uptake decision log; (2) a Verify–Select–Rewrite protocol requiring students to evaluate, justify, and paraphrase AI suggestions; and (3) brief teacher-mediated micro-conferences using diagnostic questioning to prompt authorial decision-making.
Data sources included teacher reflective journals, lesson materials, AI interaction excerpts, student drafts and revisions, uptake logs, and stimulated-recall interviews. Framework-guided thematic analysis shows that students gradually increased multi-turn querying, critical evaluation of feedback, and meaningful revision participation. The teacher attributed these improvements to clearer goal alignment and strengthened learner agency, while also identifying tensions related to uneven digital literacy and integrity-oriented assessment practices.
The study highlights the central role of teachers in mediating AI feedback and sustaining humanity in AI-assisted writing. Rather than treating GenAI as an autonomous feedback provider, the findings underscore the importance of teacher-led design in supporting voice, agency, and purposeful revision.
About the Presenters
Cong Zhang is a Professor of Applied Linguistics at the School of International Studies, Zhejiang University, China. She got her PhD in Second Language Studies from Purdue University. Her research interests include second language writing, technology-assisted language learning, and teacher education. Her recent work has appeared in the Journal of Second Language Writing, System, Innovation in Language Learning and Teaching, RELC Journal, European Journal of Education, and the Asia-Pacific Education Researcher, among others.
ORCID: https://orcid.org/0000-0003-2571-0562
Yachao Sun is an assistant professor in the Language and Culture Center at Duke Kunshan University. He got his PhD in Second Language Studies from Purdue University. His research interests include multilingual writing, translingual studies, technology-empowered language education, and multimodal composition. His recent publications have appeared in Journal of Second Language Writing, System, Language Teaching Research, Assessing Writing, among others. ORCID: https://orcid.org/0000-0002-8238-6079
Jian Xu is a Professor at Sichuan International Studies University, China. He obtained his Ph.D. degree from The Chinese University of Hong Kong. His research interests include second language writing and listening comprehension. His work has appeared in several journals, such as Journal of Second Language Writing, Assessing Writing, System, Studies in Second Language Learning and Teaching, Language Teaching Research, RELC Journal, Applied Linguistics Review, Journal of Multilingual and Multicultural Development, Journal of Language, Identity, and Education, and Higher Education Research & Development, among others. ORCID: https://orcid.org/0000-0002-2275-6197
Xiaodong Zhang is a professor at Beijing Foreign Studies University, China. He holds a Ph. D degree in Linguistics from University of Georgia, U.S.A. His work has appeared in several international journals, such as Assessing Writing, Computers & Education, Teaching in High Education, Applied Linguistics Review, Language Teaching Research, Language and Education, and Linguistics and Education. ORCID: https://orcid.org/0000-0001-7216-6542
Yabing Wang is an associate professor at the School of English Education and the Center for Linguistics and Applied Linguistics, Guangdong University of Foreign Studies. She got her PhD in Educational Psychology from the Chinese University of Hong Kong, Hong Kong, China. Her research interests include second language acquisition, individual differences, and emotions. Her recent publications have appeared in System, Language Teaching Research, Assessing Writing, Applied Linguistics Review, Studies in Second Language Learning and Teaching, Journal of Multilingual and Multicultural Development, British Journal of Educational Technology, Asian Journal of Psychiatry, The Asia-Pacific Education Researcher, among others. ORCID: http://orcid.org/0000-0003-3539-2633
Back to 2026 Invited Colloquia