Abstract
The study examines the efficiency of Artificial Intelligence (AI)-driven paraphrasing tools in preserving the originality of the source text from a lexicon semantics perspective, particularly for educational purposes. Drawing on Taylor’s (2017) linguistic framework, we conduct a comparative qualitative analysis of paraphrased output generated by three popular AI-driven paraphrasing tools: QuillBot, Scribber, and Semrush. The corpus consists of 100 abstracts belonging to two disciplines: hard science and soft science. To ensure consistency in the evaluation process, a questionnaire was distributed to linguistic experts (e.g., PhD holders) to examine the lexical semantics and coherence using a predefined academic writing rubric. The findings show that while the three AI paraphrasing tools provided a moderate ability to maintain the lexical semantics of original source texts, QuillBot and Scribbr excel in preserving the lexical semantics in hard science abstracts, whereas Semrush performs better in paraphrasing the soft science abstracts. In addition, QuillBot scores the highest in providing coherence and clarity. In contrast, Semursh scores the lowest due to its adherence to maintaining the grammatical structure over the originality of the source text. This study provides a practical educational guide for educators, students, and language learners on the challenges and opportunities of using AI-driven paraphrasing tools to improve the human learning process in academic writing.
Presenters
Rima Jamil MalkawiStudent, PhD Candidate in Linguistics and Translation, University of Sharjah, United Arab Emirates
Details
Presentation Type
Paper Presentation in a Themed Session
Theme
KEYWORDS
Lexical Semantics, Artificial Intelligence, Paraphrasing Tools, Education, Human Learning
