On the destructive effects of AI translation in Japanese-Chinese literary translation practice——An analysis based on the Japanese-to-Chinese translation of the burning green tree
论人工智能翻译在日中文学翻译实践中的破坏性效果——基于《燃烧的绿树》日译中翻译文本的分析

摘要

人工智能翻译在文学文本应用中可能会带来破坏性后果，本文以大江健三郎《燃烧的绿树（2）：动摇》日译中实践为实证资料，指出人工智能翻译在三个方面存在破坏性后果，即词藻华丽、过度修饰损伤源文本行文风格、擅自删减原文信息破坏了源文本的内在逻辑，理解错误带来的误译破坏了译文的忠实性。在此基础上指出，可以说，当前人工智能翻译工具在文学文本中的应用，存在以信息传输效率牺牲“文学特异性”的风险，因此文学翻译中译者的干预是不可替代的，译者必须承担文本特异性守护者角色，抵制修饰词添加对作者风格的侵蚀，保护承担隐喻功能的“冗余”信息，干预AI对文化特异性表达的简化处理。

Abstract

Artificial intelligence (AI) translation may incur destructive consequences in literary text applications. Based on empirical evidence from the Japanese-to-Chinese translation practice of Kenzaburō Ōe’s The Burning Green Tree (Part 2: Vacillation), this paper identifies three primary destructive consequences of AI translation: Overly ornate rhetoric and excessive embellishment that compromise the source text’s stylistic integrity; Unauthorized omission of source content that disrupts intrinsic textual logic; Misinterpretation-induced mistranslation that undermines translational fidelity. These findings demonstrate that current AI translation tools, when applied to literary texts, prioritize informational efficiency at the expense of literary specificity. Consequently, human translator intervention remains irreplaceable. Translators must act as guardians of textual specificity by: Mitigating stylistic erosion caused by AI-added modifiers; Preserving functionally "redundant" elements essential to metaphorical systems; Intervening against AI-driven oversimplification of culturally specific expressions.