EFL Students' Paraphrasing Challenges in Literature Reviews: Investigation of Quillbot's Mitigating Mechanisms

Abstract

This study investigated the paraphrasing challenges faced by EFL students in writing literature reviews for research proposals and examined the effectiveness of QuillBot in addressing these difficulties. Employing an explanatory sequential mixed-methods design, the research first utilized a questionnaire (n=25) to identify common paraphrasing difficulties across four dimensions, followed by semi-structured interviews (n=6) to explore students' experiences with QuillBot. Quantitative findings revealed a high overall level of paraphrasing difficulty (mean=2.58), with paraphrasing strategy emerging as the most challenging dimension (mean=2.69), followed by language use (mean=2.60), structure aspect (mean=2.57), and content aspect (mean=2.48). Specifically, students struggled most with modifying words using synonyms while preserving meaning (mean=2.80) and changing word order in sentences (mean=2.76). Qualitative analysis identified four primary mechanisms through which QuillBot addresses these challenges: Synonym Selection and Vocabulary Enhancement, Sentence Structure Transformation, Sentence Simplification and Condensation, and Mode Selection and Strategy Integration. These mechanisms directly correspond to students' specific difficulties, suggesting that AI-assisted writing tools can provide targeted support for academic writing development. While QuillBot effectively addresses immediate paraphrasing challenges, the study acknowledges the importance of balanced implementation that fosters independent writing skills. This research contributes to understanding how AI tools can mediate the cognitive and linguistic demands of paraphrasing in EFL contexts, with implications for technology-enhanced writing pedagogy in higher education.