A Functionalist Perspective on Translating Traditional Chinese Patterns: A Comparative Analysis of Large Language Models vs. Human Translation

Abstract

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.