Form without function: a reception-theoretic evaluation of LLM translations of classical Chinese poetry and its ethical implications

Background Classical Chinese poetry creates artistic conception through formal gaps, which invite reader imagination. However, these gaps are difficult to preserve in translation, as they may be difficult to interpret without additional context. Large Language Models (LLMs) are capable of preserving the formal structures of these gaps in their translation of classical Chinese poetry, yet they often lose their essential function—inviting imagination. Methods From a reception-theoretic perspective, this study evaluates three LLMs (GPT, GLM-5.1, DeepSeek) by comparing their translations of gaps in two canonical Tang poems against human benchmark translations. Drawing on reception theory's concept of “gaps” and “indeterminacy”, we develop a four-point scale to assess the five categories of gaps (semantic, logical, syntactic, cultural, and atmospheric gaps) in the LLMs' translations. Results Our findings reveal a preservation paradox: LLMs can preserve the formal structure of gaps, but this formal preservation does not guarantee their aesthetic function. Conclusion This suggests that pattern-matching alone fails to convey the literariness of classical Chinese poetry, which requires consciousness of aesthetic function. The paradox also raises ethical questions about the use of LLMs in culturally significant translation tasks, where aesthetic and cultural values are at stake.