Applying Grice’s Maxims and Speech Act Theory to Analyze Pragmatics in Chatgpt and Gemini AI Dialogues

Authors : Thoriqi Firdaus, Lisda Yasin Bakari, Diva Salsabila Octarianto, Rifwanda Khoirunnisa, Dwi Retno Malini, Mukarromah Mukarromah

Pages : 34-53

Doi:10.63673/Lotus.1631745

View : 90 | Download : 42

Publication Date : 2025-06-24

Article Type : Research Paper

Abstract :The issue of applying pragmatic principles in conversations generated by ChatGPT and Gemini is crucial for the investigation. Grice’s Maxims and Speech Act Theory provide a fundamental framework for guiding conversations to become more structured, relevant, and purposeful. The study aims to provide a more comprehensive understanding of the application of all of Grice’s Maxims and Speech Act Theory in conversations generated by AI models such as ChatGPT and Gemini. The research employs a mixed-methods approach. The data is quantitatively analyzed using a T-test, while the validity and reliability of the study are ensured through data triangulation. The findings of the study reveal that both AI models, ChatGPT and Gemini AI, demonstrate exceptional ability in applying Grice’s Maxims, with an average score of 1.00 for both models, indicating high consistency in maintaining relevance, quantity, quality, and manner in the conversation. Both models also show similar results in applying Speech Act Theory, with an average score of 0.75, although there are challenges in understanding the conversation\\\'s more nuanced intentions or impacts. Grice’s Maxims prove more effective in explaining conversational dynamics, while Speech Act Theory still requires further development. Comparative analysis results indicate that both AI models, ChatGPT and Gemini AI, perform almost identically in applying Grice’s Maxims and Speech Act Theory. For Grice’s Maxims, the average score between the two models reaches 1.00, with no significant differences found (p-value = 0.560), signifying high consistency in applying pragmatic principles. For Speech Act Theory, although there are slight differences in effect size (Cohen’s d = 0.444 for Task 1 and 0.224 for Task 2), these differences are not large enough to be considered significant, indicating that both models have an almost identical understanding of It theory.
Keywords : Grice, Konuşma Eylemi, Chatgpt, Gemini

ORIGINAL ARTICLE URL