Translating the Zhuang Folktale Ye Xian with DeepSeek: An Analysis of Error Types and Post-editing Strategies

Machine translation in literary translation remains a great challenge due to the high cultural and artistic elements inherent in literary texts. In the context of artificial intelligence, this study used a large language model-based machine translation system DeepSeek-V3 to investigate its performance in translating the renowned Zhuang folk tale Ye Xian. Based on the Multidimensional Quality Metrics framework, this study conducted a qualitative research and identified four common error types involving accuracy, terminology, fluency, and locale convention errors. In response, four post-editing strategies are proposed, namely semantic correction, fluency optimization, terminology standardization, and cultural annotating. These strategies are expected to improve the quality of machine translation in rendering folk literary works, hence contributing to the advancement of machine translation in the field of literature.