基于图式理论的中国古典诗歌意象的人工翻译与AI翻译比较

A Comparative Study of Human Translation and AI Translation of Chinese Classical Poetry Imagery Based on Schema Theory

摘要: 中国古典诗歌的意象承载着深厚的民族文化内涵，其跨语言翻译是中华文化“走出去”战略下的重要课题。本文以认知语言学中的图式理论为框架，将翻译过程视为语言、内容、形式三层图式的解码与重构。通过对比分析AI翻译工具(DeepL, Kimi, ChatGPT)产生的译本与以许渊冲为代表的人工译本在分别处理《静夜思》《春望》《望庐山瀑布》《天净沙·秋思》四首诗歌意象时的翻译差异，结合翻译美学、阐释学及具体翻译策略等理论观点，深入探讨了人工翻译和AI翻译在图式传递能力上的优劣。研究发现，AI在语言图式层面表现稳定，但在内容图式，特别是形式图式的审美重构上，与能够进行张建英所提出的“文化调停”和审美再创造的人类译者存在显著差距。

Abstract: The imagery in Chinese classical poetry carries profound national cultural connotations, and its cross-language translation is a significant issue under the strategy of Chinese culture “going global.” This study employs schema theory from cognitive linguistics as its theoretical framework, conceptualizing the translation process as the decoding and reconstruction of three types of schemas: linguistic, content, and formal. By comparing and analyzing the translations generated by AI tools (DeepL, Kimi, ChatGPT) with human translations represented by Xu Yuanchong in their respective treatments of poetic imagery in four classical Chinese poems—Quiet Night Thought, Spring View, The Waterfall in Mount Lu Viewed from Afar, and Autumn Thoughts—and by integrating theoretical perspectives from translation aesthetics, hermeneutics, and specific translation strategies, this paper explores the strengths and limitations of human versus AI translation in terms of schema transmission capacity. The findings reveal that while AI demonstrates stable performance at the linguistic schema level, it exhibits significant gaps compared to human translators at the content schema level, particularly in the aesthetic reconstruction of the formal schema, where human translators are capable of the “cultural mediation” and aesthetic re-creation proposed by Zhang Jianying.