Generative AI and the Reshaping of Second Language Writing: A Cognitive Process Perspective

Addressing how Generative AI (GenAI) reshapes the cognitive act of second language (L2) writing, this conceptual article draws on recent empirical literature (2023–2026). It grounds its analysis in the Cognitive Process Theory of Writing (CPTW), chosen because its decomposition of planning, translating, and reviewing under working-memory limits illuminates how cognitive labor is redistributed between learner and machine. Through CPTW, three key insights emerge. First, planning undergoes an externalization of productive struggle: offloading idea generation causes learners to bypass essential rhetorical problem framing. Second, translating experiences a voice constraint: model optimization toward conventionality flattens non-standard L2 syntax and homogenizes authorial voice. Third, reviewing suffers a reviewing bifurcation, splitting evaluation into goal-level appraisal and opaque decision-level monitoring. Theoretically, GenAI transforms writing into a co-generative sociotechnical practice requiring GenAI feedback literacy. Pedagogically, educators should adopt process-oriented assessments, using pre-prompting framing logs, annotated drafts, and prompt histories to evaluate cognitive engagement over polished final products.