ARTIFICIAL INTELLIGENCE TOOLS VERSUS STUDENTS: COMPARISON OF PARAPHRASES OF ACADEMIC TEXTS

Author(s) : Tanveer Hussain Mangnejo and Sadia Irshad

Abstract:

Paraphrasing is an important skill for learners to take part in academic discourse and successful paraphrasing provides solutions to learners of the difficulties they face during the academic writing process (Chen & Liou, 2015). With the advancement of AI paraphrasing tools, learners paraphrase and incorporate that text in their writing without double-checking the validity of the paraphrased text. This article aimed to investigate the differences between students' paraphrases of academic texts and the paraphrases done through different AI tools. Furthermore, this study also explored how corpus can help students evaluate the paraphrases of academic texts. The concurrent mixed methods approach was used. Twenty-five graduate students were selected randomly and five AI tools were chosen purposively. Five passages in English with different levels of complexity were given to the participants to paraphrase. The grammar, coherence, academic vocabulary, and organization of the paraphrases were then assessed through the Independent Samples Test. The triangulation was sought through the corpus-based approach. The independent samples t-test results show statistically significant differences between the three variables except for the academic vocabulary variable. The results of academic vocabulary and grammar variables obtained through the corpus website also support the results obtained through the t-test. While AI technologies are less precise and coherent when paraphrasing English texts, there could be a better sense of grammar and academic vocabulary use. As a result, it is advised that students may use AI tools to benefit from the technology, but they need to be the decision-makers for the relevance, accuracy, preciseness and validity of the paraphrases. 
Keywords: concurrent, paraphrase, AI tools, t-test, corpus