生态翻译学视角下基于LLM的博物馆展览说明牌英译研究——以良渚博物院为例

LLM-Based English Translation of Museum Exhibition Labels from the Perspective of Eco-Translatology—A Case Study of Liangzhu Museum

摘要: 本文考察了大语言模型在良渚博物院展览说明牌英译中的表现。本研究以生态翻译学为理论基础，从语言维、文化维和交际维三个层面评估了LLM的译文质量。结果显示，大语言模型在处理低文化负载的一般性术语或常规考古陈述时选择适应性较高，在词汇选择、句法结构、语篇衔接以及语体风格等方面能够符合目标语的语言规范，且能保持语义清晰和连贯，甚至能够弥补人工译文存在的语义遗漏、词汇搭配不适等问题。但在文化负载较高、术语精确性要求较强的翻译情境中，仍存在明显局限，主要表现为文化功能遗漏、文化内涵与象征意义丢失、核心文化概念误译等。研究表明，“三维转换”框架对评估LLM的博物馆说明牌翻译具有解释力，为博物馆在实践中合理地使用LLM提供一定参考。

Abstract: This paper examines the performance of large language models (LLMs) in the English translation of exhibition labels at the Liangzhu Museum. Grounded in eco-translatology, the study evaluates the quality of LLM-generated translations from the linguistic, cultural and communicative dimensions. The results show that LLMs exhibit high adaptive selectivity when dealing with general terms of low cultural load or routine archaeological statements. They are able to conform to the linguistic norms of the target language in terms of lexical choice, syntactic structure, discourse cohesion, and stylistic register, while maintaining semantic clarity and coherence. In some cases, they can even compensate for issues in human translations such as semantic omissions and inappropriate lexical collocations. However, in translation contexts with high cultural load and a strong demand for terminological precision, notable limitations persist, mainly manifesting as omissions of cultural functions, loss of cultural connotations and symbolic meanings, and mistranslations of core cultural concepts. The study suggests that the “three-dimensional transformation” framework has explanatory power for evaluating LLM performance in museum label translation and provides a reference for the judicious use of LLMs in museum practice.