Linguistic Choices and Relevance of Essays Generated Through AI Tools: A Corpus-Assisted Analysis
Abstract
This study aims to identify AI-based linguistic choices in various text types within the academic genre and to investigate the relevance of AI-generated text in performing the communicative function across informative and persuasive texts, employing a corpus-assisted mixed-methods research paradigm. The corpus for this study comprises AI-generated essays, classified into informative and persuasive text types within the academic genre. It uses Lexical Specificity by Hyland (2009) as a theoretical framework to synthesise the findings. The study's findings reveal that although AI generates genre-compatible content, a significant linguistic deviation from norms has been found in both the selected text types of the academic genre. However, AI tools generate more relevant content in informative writing than in persuasive writing, as evidenced by text-type-compatible linguistic choices. Thus, the study concludes on the reliability of the AI-generated content in informative writing within the academic domain. The study's findings could help developers enhance the outputs of such tools in the future, enabling them to be utilised by students in the academic domain, particularly in linguistics.
Keywords: Artificial Intelligence, Corpus Linguistics, Linguistic Choices, Academics, Genre Analysis
