Tomoko Oyama, Anubhav Anubhav, Yuka Akiyama
Abstract In the age of generative artificial intelligence (GenAI), exercising agency has become increasingly critical for second language (L2) writers to make deliberate authorial decisions. The agency enactment is essential particularly for the population of engineering L2 writers, as they need to navigate through increasingly diversified writing styles in their discipline of engineering. Furthermore, research on pedagogical frameworks that foreground agency in today’s GenAI era has been increasingly called for. This paper reports on a two-year longitudinal case study to investigate how an interdisciplinary team of applied linguists and an AI engineering researcher collaborated to jointly design instructional conditions intended to support engineering L2 writers’ agency in GenAI-mediated genre-based research paper writing. Adopting Cultural-Historical Activity Theory as a theoretical lens, we analyzed data from individual interviews, focus group reflections, instructional materials, researcher notes, and pre-instructional meeting transcripts. The findings reveal that the interdisciplinary collaboration initially generated conflicts deriving from differing epistemologies towards teaching L2 writing, misaligned division of labor, and agency-diminishing instructional artifacts. Yet, the tensions across and within those stakeholders’ actions were negotiated over time and led to collective pedagogical transformations, as evidenced in the redesigned instructional artifacts aimed at better informing the learners about sub-disciplinary rhetorical variations, prompt customization, output evaluation, and critical GenAI literacy. Pedagogical implications are discussed on designing instruction to support L2 writers’ agency through informed, reflective, and ethical decisions in GenAI-mediated genre-based L2 writing.