Grounded in the functionalist Skopos Theory, this study systematically compares the English translations of traditional Chinese patterns produced by professional museum translators and Large Language Models (LLMs). It constructs a four dimensional evaluation system covering cultural information accuracy, visual imagery fidelity, cultural adaptability, and linguistic expression quality.

The findings indicate that human translation significantly outperforms LLMs in in-depth cultural transmission, systematic symbolic interpretation, and reader reception. By flexibly employing strategies such as annotation, cultural adaptation, and free translation, human translators effectively achieve the cross-linguistic transfer of cultural functions. Conversely, while LLMs demonstrate high linguistic fluency and efficiency, they struggle with highly culture-loaded symbols, often leading to metaphor loss, contextual dislocation, and cultural simplification, revealing a lack of genuine interpretive capabilities.

Based on these insights, this paper proposes an optimized “human-led, LLM-assisted” collaboration model to enhance both the quality and efficiency of pattern translation. This study provides a theoretical basis and practical strategies for the international dissemination of traditional cultural symbols.