Asian Journal of Arts, Humanities and Social Studies

This article analyzes the language use of AI models Gemini and Claude from a comparative linguistic perspective. In recent years, rapid advances in large language models (LLMs) such as ChatGPT, Gemini, and Claude have driven widespread adoption in education and academic communication, creating new opportunities and challenges for pedagogy. Rather than focusing only on technical performance such as accuracy or speed, it examines vocabulary and grammar, discourse organization, pragmatic strategies, and stylistic features. This study employed a descriptive case-based comparative analysis using identical prompts to generate outputs for comparison. Findings indicate that Gemini tends to generate academic-style text, using abstract vocabulary, complex sentence structures, and logical connectors. Claude, by contrast, favors a semi-conversational tone, simpler language, and step-by-step organization that supports practical application. These differences reflect distinct discourse identities.

In terms of pedagogy, Gemini is more suitable for developing academic writing and practicing theoretical expression, while Claude provides clearer, more accessible feedback for interactive activities. Used together, they can balance formal and practical aspects of language learning. This study helps fill a gap in research that often emphasizes technical performance and instead proposes an analytical framework linking AI language use to applied linguistics. As this study is based on illustrative examples, its findings cannot be generalized; however, they provide a foundation for future corpus-based comparative research.