Integrating Llms Into Efl Academic Writing: Perceptions, Reported Practices, And Pedagogical Priorities In Algerian Higher Education
Auteurs : Ladjal Rym . Messekher Hayat .
Résumé
In Algerian EFL higher education, Large Language Models (LLMs) have expanded academic writing assistance from discrete tasks to the generation of extended texts. Yet little is known about how students and faculty understand and use these tools, and what support and pedagogical priorities should guide their integration. This study examined how EFL students and faculty perceive and report LLM use in academic writing through a cross-sectional convergent mixed-methods design, administering a parallel-instrument survey to 165 respondents (104 students, 61 faculty). Items were organised around four conceptual domains: human agency and responsible use, prompting, cognition and perceived academic effects, and transparency and classroom practices. Quantitative analysis was combined with thematic coding of 590 substantive open-ended responses, yielding 40 final codes. Students recorded their lowest mean on self-regulation (M = 3.21), the only scale item not differing significantly from the midpoint, while faculty recorded their highest mean on cognitive over-reliance (M = 4.74). Students' strongest endorsement concerned authorship and responsible use. Informal prompting experience, reported by 86.5% of students, was not significantly associated with self-reported autonomy or critical-thinking confidence. Faculty who had personally tested prompts reported greater concern about cognitive over-reliance than those who had not. Faculty reported relying more on oral examination than on detection software when checking student work. Both populations converged on structured instruction as a leading pedagogical priority. Implications address embedded prompting pedagogy, oral accountability that assesses process over fluency, differentiated professional development, and participatory institutional norm-setting.
Mots clés
Large language models ; EFL academic writing; student and faculty perspectives; Prompting ; Human agency ; Responsible AI use.
