58 Thai pre-service teachers chose their own resources and wrote handwritten notes in a six-week flipped writing course: most endorsed preparation and writing readiness, yet 73% struggled with vocabulary and grammar, 66% with judging multiple sources, and 53% with GenAI dependence
Synopsis
In a six-week flipped EFL writing course, 58 second-year student teachers at a Thai university independently selected resources such as textbooks, websites, educational videos, and GenAI tools and synthesized their learning through handwritten notebook summaries; questionnaire responses showed generally positive perceptions of learning preparation, information management, learning responsibility, and writing readiness, while focus group discussions with 30 students revealed four interconnected constraints—limited linguistic and background knowledge, uncertainty without immediate teacher guidance, difficulty evaluating and synthesizing multiple resources, and challenges in negotiating GenAI use—which students navigated through increased effort, planning and self-regulation, use of multiple
Interpretation
The study shifted the pre-class component of a flipped classroom from teacher-provided materials to learner-directed resource selection with handwritten notebook synthesis, and students rated preparation, information management, learning responsibility, and writing readiness positively overall. Prior flipped classroom research has largely examined teacher-curated videos, quizzes, and interactive slides; this study gave learners the choice of pre-class resources and documented their experiences within a structured self-access environment. All 58 participants completed a seven-dimension five-point Likert questionnaire with content validity IOC from .67 to 1.00 and internal consistency α = .81; the highest-rated item in every dimension was above the scale midpoint.
Students reported four interconnected constraints: limited linguistic and background knowledge (22/30, 73%), uncertainty without immediate teacher guidance (18/30, 60%), difficulty evaluating and synthesizing multiple resources (20/30, 66%), and challenges in negotiating GenAI use including dependence and verification (16/30, 53%). The study places GenAI use and resource evaluation side by side as core constraints on self-access learning, rather than treating AI only as a convenient tool. Based on six focus group discussions with 30 volunteers (five per group, about 50 minutes each), analyzed through inductive thematic analysis that distinguished enacted strategies from future-oriented intentions.
Students navigated these constraints with four strategies: increasing learning effort (enacted 14/30, 46%; intended 12/30, 40%), planning and self-regulation (enacted 13/30, 43%; intended 17/30, 56%), using multiple learning resources (20/30, 66%, with 13 explicitly cross-checking sources), and strategic help-seeking (22/30, 73%). The study separates actions students reported actually taking from intentions they only expressed, showing that some self-regulatory behavior remained at the planning stage. The coding framework classified statements as experienced, enacted, intended, or ambiguous, with frequencies based on the 30 focus group participants and themes that are not mutually exclusive.
The study proposes that GenAI expands rather than replaces learner autonomy, with students showing emerging AI literacy in clarifying concepts, explaining grammar, and generating examples, but at varying levels, as some still relied on AI outputs without sufficient verification. It positions AI literacy alongside self-regulated learning as a necessary condition for autonomy in digital environments, rather than focusing only on access to technology. Questionnaire items on accepting GenAI information without checking (M = 3.00, SD = 1.24) and relying on GenAI instead of understanding independently (M = 2.95, SD = 1.20) were moderate, and some focus group participants described over-reliance on ChatGPT.
Perspective
The study applies to a six-week sentence-level writing flipped course for second-year English Education students at a Thai public university, and speaks to EFL learners still developing independent learning capacity and to self-access language learning practitioners. It indicates that a semi-structured design can offer structured opportunities for guided autonomy by letting learners choose resources, plan preparation, and reflect while the teacher retains goal setting and classroom guidance; for practitioners, the implication is that self-access design should combine clear instructional goals with learner choice, explicitly develop self-regulated learning strategies such as goal setting, comprehension monitoring, resource evaluation, and reflection, and embed GenAI literacy so that students verify AI-generated information, compare sources, and treat AI as support for understanding rather than a substitute.
This is a descriptive mixed-methods study in a single course at a single institution, with no control group and no longitudinal measure of autonomy over six weeks, so what is reported is student perceptions and retrospective accounts rather than evidence of gains in autonomy or writing ability. The questionnaire and focus groups were administered in English; although explanations and clarification were provided, language comprehension may still have shaped responses. In focus groups, participants could hear one another, so complete confidentiality within groups could not be guaranteed. Frequencies are based on the 30 focus group participants and are not mutually exclusive, so they do not sum to a total. In addition, the loaded text is an incomplete scope, with Figure 1 and some table content not fully presented, so the summary of the intervention guide and some quantitative details is limited to what the main text and appendices state.