小说《苍老的浮云》的AI英译及人机协同的合与斗
Synergies and Conflicts in Human-AI Collaborative English Translation of Chinese Novel Old Floating Cloud
摘要: 随着生成式人工智能与大语言模型(LLMs)的迭代升级,机器翻译(MT)已全面渗透至各类应用型文本的转写之中。然而,先锋文学翻译——因其充斥着大量梦呓叙事、身体解构、视觉与听觉互换的通感意象以及不可靠的叙述视角,始终被视为人工智能语言表征里抵御机器程式化复刻的文学领地。本报告以中国当代前卫先锋作家残雪的代表作《苍老的浮云》(及《天堂里的对话》等系列隐喻文本)为翻译实践蓝本,深入探讨AI在大先锋文学、超现实主义语境下进行英译的潜能与局限。本研究借助SWOT分析矩阵,对AI翻译输出的语料进行量化统计与质性评估,重点剖析大模型在处理先锋文学中通感隐喻、机械硬化表述、主体性游移等维度的表现。研究发现,AI在词汇扩展、复杂长难句转换及初步格式梳理中展现出强大的协同效应,极大地缩短了初译周期;但在面对残雪式荒诞叙事、多重感官交织的意象以及深层文化语境时,AI往往表现出过度理性化、词汇晦涩堆砌以及文学审美感知的严重缺位,从而引发高频的冲突。基于此,本报告尝试提出了译者在人工智能时代的提升路径,希望为中国文学的海外传播提供实践范式。
Abstract: With the continuous iteration and upgrading of Generative Artificial Intelligence and Large Language Models (LLMs), Machine Translation (MT) has comprehensively penetrated the corpus transcreation of various applied texts. Nevertheless, avant-garde literary translation-replete with somnambulistic narratives, bodily deconstructions, synesthetic transmutations between visual and auditory senses, and inherently unreliable narrative subjectivities-remains recognized as a literary domain in AI language representations that resists machine-programmed replication. This report takes the representative avant-garde works of contemporary Chinese writer Can Xue, specifically Old Floating Cloud (and related metaphysical texts like Dialogues in Paradise), as the source text for translation practice. It delves into the potential and constraints of AI in rendering avant-garde and surrealist Chinese fiction into English. Utilizing the SWOT analysis matrix, this study conducts a quantitative text analysis and qualitative evaluation of AI-generated translations. It focuses on the model’s performance across dimensions such as synesthetic metaphors, lexical hyper-intellectualization, and shifting narrative subjectivities. The findings indicate that AI exhibits a powerful “Synergy” effect in lexical expansion, syntactic restructuring of convoluted sentences, and basic structural arrangement, substantially compressing the initial draft cycle. However, when confronting Can Xue’s absurd narratives, multi-sensory intertwined imagery, and deep cultural undercurrents, AI frequently suffers from over-rationalization, mechanical rigidity, and a deficiency in literary aesthetic perception, leading to intense “Conflicts”. Based on this, this report attempts to propose paths for translators to improve in the era of artificial intelligence, hoping to provide a practical paradigm for the overseas dissemination of Chinese literature.
文章引用:肖倩, 赵冰. 小说《苍老的浮云》的AI英译及人机协同的合与斗[J]. 现代语言学, 2026, 14(8): 302-310. https://doi.org/10.12677/ml.2026.148787

参考文献

[1] 白玉. AI时代机器翻译技术对文学翻译的协助介入作用[J]. 兰州文理学院学报(社会科学版), 2021, 37(3): 84-88.
[2] 孙德喜. 病入膏肓世界的梦呓——残雪小说语言论[J]. 中南大学学报(社会科学版), 2005, 11(1): 97-100.
[3] Toral, A. and Way, A. (2018) What Level of Quality Can Neural Machine Translation Attain on Literary Text? In: Moorkens, J., Castilho, S., Gaspari, F. and Doherty, S. Eds., Machine Translation: Technologies and Applications, Springer International Publishing, 263-287.
https://doi.org/10.1007/978-3-319-91241-7_12
[4] Guerberof-Arenas, A. and Toral, A. (2022) Creativity in Translation: Machine Translation as a Constraint for Literary Texts. Translation Spaces, 11, 184-212.
https://doi.org/10.1075/ts.21025.gue
[5] 邵璐, 李平艳. 英语世界“残雪热”现象与机理考辨[J]. 外国语(上海外国语大学学报), 2024, 47(6): 81-89.
[6] 张茜, 张欣. 国际传播视域下翻译人才的SWOT分析及能力建设[J]. 黑河学院学报, 2023, 14(7): 62-64, 76.
[7] 残雪. 残雪文集·第一卷 苍老的浮云[M]. 长沙: 湖南文艺出版社, 1998.
[8] 陈志兰. 多元文化视域下《格萨尔》史诗传译的SWOT分析[J]. 西藏研究, 2015(5): 103-110.
[9] Li, Y., Zhang, R., Wang, Z., Zhang, H., Cui, L., Yin, Y., et al. (2025). Lost in Literalism: How Supervised Training Shapes Translationese in Llms. In: Che, W.X., Nabende, J., Shutova, E. and Pilehvar, M.T., Eds., Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Association for Computational Linguistics, 12875-12894.
https://doi.org/10.18653/v1/2025.acl-long.630
[10] Venuti, L. (2026) The Human Translator versus the Machine: Algorithms of Interpretation. Critical Inquiry, 52, 459-477.
https://doi.org/10.1086/739756
[11] Wu, M., Xu, J., Yuan, Y., Haffari, G., Wan, L., Luo, W., et al. (2025) (perhaps) Beyond Human Translation: Harnessing Multi-Agent Collaboration for Translating Ultra-Long Literary Texts. Transactions of the Association for Computational Linguistics, 13, 901-922.
https://doi.org/10.1162/tacl.a.25
[12] 张明, 赵丽. 大语言模型翻译文学前卫文本的“具身认知”转向与局限[J]. 外语教学与研究, 2026(1): 88-97.