FROM AUTHOR TO AI - STYLISTIC ANALYSIS OF CHATGPT'S LITERARY MIMICRY
FROM AUTHOR TO AI - STYLISTIC ANALYSIS OF CHATGPT'S LITERARY MIMICRY

Summary/Abstract: The research proposes two case studies of ChatGPT text generation within the scope of determining if the AI-based program can reproduce canonical literary styles of two Romanian authors. The scientific endeavour aims to analyse the correspondence between the generated texts and the original author's style. Along this objective, the paper proposes a short description of how Large-Language Models operate, within the AI text generation process, in order to describe and to identify the linguistic features instrumentalised by these programs. By this means, the main principles taken into account by the present research are focusing on how these programs are able to distinguish between functional styles, content or text structures, and how accurate ChatGPT can mimic or replicate a given auctorial style. The corpus analysis falls in the stylistic linguistic description and features a compared view between the canonical literary style and the generated style, through linguistic comparisons. The case studies are put through this endeavour in regard solely to their capacity to mimic the author's style and in which degree, exclusively to the chosen texts, ChatGPT may replicate at a linguistic level the original auctorial style. Also, due to the constant influence of literary and cultural studies approaches, the research debates how the author concept gets reinterpreted in the AI era.

• Details
• Contents