Prompt-Output Dataset for Analyzing Linguistic and Cultural Bias in Multilingual LLMs

This dataset contains prompts and large language model (LLM) responses used to analyze cultural and linguistic bias across English and Chinese outputs. It includes four prompt-response catalogs—two for ChatGPT and two for DeepSeek—in both English and Chinese, as well as a bilingual prompt list. These materials support a research study, which includes a critical discourse analysis of how multilingual LLMs reproduce dominant cultural norms, submitted for peer review.