A Corpus-Assisted Comparative Analysis of Adversative Conjunctions in Student and ChatGPT Essays: A Rhizomatic Perspective

Abstract

Employing a Deleuzian-Guattarian rhizomatic framing of multiplicity, this study had a dual purpose: to investigate the occurrence frequencies of four adversative conjunctions (but, however, while, and yet) in student-written discursive essays (SWDEs, n = 50) and in ChatGPT-generated discursive essays (ChatGPT-GDEs, n = 50) as analysed by AntConc; and to extrapolate the Deleuzian-Guattarian multiplicity in these two essay sets as displayed by the use of these four adversative conjunctions in the paragraphs of the said essay sets. These two essay sets were selected, in the case of the former set, and generated in the case of the latter set, through convenience sampling. Student essays were written by first year university students, who were registered in an undergraduate English language module in the first semester of 2023 at a university in South Africa. Then, the study used AntConc (version 4.3.0) to analyse the two essay sets. In the main, the study found that the usage of the first three adversative conjunctions (but, however, and while) was almost evenly spread across SWDEs, a feature that indicated that in this essay set, students very nearly displayed a consistent adversative variation in contrasting ideas or in expressing concession or opposition to such ideas. On the contrary, in ChatGPT-GDEs, while was excessively and very nearly monopolistically employed by ChatGPT in contrasting ideas or in expressing concession or opposition to such ideas in this set of essays as compared to the other three adversative conjunctions. Finally, cases of multiplicities were almost nearly evenly distributed across the first three adversative conjunctions in SWDEs, whereas they were inordinately and almost monopolistically concentrated in one adversative conjunction (while) in ChatGPT-GDEs.