THE STUDY OF AFFIXATION IN CHATGPT VENTING SESSION

Abstract

This research analyzes the affixation and reduplication patterns found in the Indonesian language responses generated by ChatGPT. The research focuses on identifying types of affixation and affixation errors in informal venting sessions, and how they vary across different emotional contexts (happy and sad), with a comparison between free (non-premium) and paid (premium) versions of ChatGPT. The study also includes formal language examples as a contrastive point. This research aims to describe the morphological behavior of an AI-generated text, especially in relation to affix use in the Indonesian language. This study uses a qualitative descriptive method with a structural approach. The data source is the ChatGPT platform, particularly its output in informal venting sessions. The data were collected through observation, documentation, and note-taking techniques. The analysis method used is the distributional method with immediate constituent analysis (ICA) and morphological transformation techniques. The validity of the data is ensured through rechecking techniques to confirm consistency and reliability of findings. The results indicate that ChatGPT demonstrates sensitivity to linguistic context and user emotion through its use of affixation and reduplication, adapting its language based on tone and membership type. The free version shows more informal and empathetic affix forms, while the premium version displays more structured and grammatically accurate usage. Ultimately, this research concludes that ChatGPT does not merely generate linguistically correct outputs, but also reflects pragmatic adaptability aligned with emotional and conversational settings, even though occasional nonstandard forms still appear in informal interactions.