Investigating ChatGPT’s Impact on Developing the Tertiary-Level English Department Students’ EFL Writing Skills

Despite the growing body of research involving examining teachers’ and students’ experiences of AI integration in EFL writing pedagogy in Bangladesh, relatively little is known about English department students’ perceptions of ChatGPT’s influence in major domains of EFL writing. To bridge this gap, our research investigates tertiary-level English department students’ perceptions of ChatGPT’s impact in developing their EFL writing skills within the Bangladeshi context through employing a mixed-method research design, comprising both quantitative and qualitative data. Quantitative data were collected from 74 undergraduate and graduate students at a public university in Bangladesh using a structured five-point Likert-scale questionnaire and subsequently, qualitative insights were gathered to supplement the survey data through three focus group discussions (FGDs). Our study explored ChatGPT’s perceived role across four key domains of EFL writing: motivation and engagement, grammatical correction, writing essays and creating references. Findings revealed that ChatGPT positively influenced students’ motivation through providing immediate and non-judgmental feedback, facilitating idea generation, content organization and context-specific writing. Also, ChatGPT facilitated the pre-writing and foundational stages of essay writing. Students exhibited more caution regarding ChatGPT’s role as a grammar correction and referencing tool. Despite its usefulness in explaining grammar rules and providing language-related feedback, students found specialized grammar correction tools to be more reliable than ChatGPT. Likewise, ChatGPT emerged as an unreliable referencing tool due to producing misleading citation information. Our study offers key implications for educators, stakeholders and policy- makers to integrate ChatGPT more ethically and effectively in EFL writing while preserving independent reasoning and judicious evaluation. Spectrum, Volume 19, June 2024: 135-149