Speaker
Description
Generative artificial intelligence (GenAI) increasingly enables second language (L2) writers to obtain ideas, translations, paraphrases, and feedback through a single interface, raising questions about how teachers can preserve meaningful engagement with writing. This article revisits evidence from four specialised digital writing tools—Sketch Engine, Google Translate, QuillBot, and Grammarly—to compare teachers’ recommended practices with students’ observed practices across planning, composing, and revising. Data comprise lesson plans and interviews from 10 IELTS writing teachers, screen recordings and interviews from 20 Vietnamese university students, and independently rated IELTS Academic Writing Task 2 essays. Teachers emphasised planning-stage use of Sketch Engine (70%) and Google Translate (90%), but only 10% of students used Sketch Engine at that stage and none used Google Translate. During composing, 60% of students translated words or phrases, compared with 10% of teachers recommending this use; an additional 10% of students translated complete drafts. Revising showed closer correspondence for Grammarly (teachers 90%; students 95%), QuillBot (50%; 45%), and Sketch Engine (20%; 20%). Nevertheless, similar tool choices did not consistently entail similar evaluation of output. Higher-scoring students more often checked contextual fit, compared alternatives, and selectively rejected unsuitable suggestions. The article distinguishes alignment in stage, function, and evaluation and considers how this framework can inform GenAI writing pedagogy. Although GenAI was not investigated, the findings suggest a rationale for designing question-led assistance that supports planning, linguistic decision-making, and reflective revision without replacing students’ authorship.
Keywords: L2 writing, generative AI, process writing, teacher–student alignment, cognitive engagement, AI literacy, feedback evaluation