摘要
本文剖析了大语言模型(LLMs)驱动的民族文化翻译中存在的伦理问题,并提出了系统性的应对策略。研究发现,由于数据稀缺、文化理解不足等技术瓶颈,LLMs在处理民族语言文化时,容易导致文化挪用、文化失真、文化偏见、文化失衡、文化版权侵犯以及文化安全风险。针对上述问题,本文倡导构建“预防为主、技术赋能、协同治理”三位一体的应对框架,旨在推动构建负责任、可持续的民族文化翻译新生态,促进LLMs技术在民族文化翻译领域健康发展。
This paper analyzes the ethical issues in the ethnic cultural translation driven by Large Language Models(LLMs)and proposes systematic strategies to address them.The research finds that when translating ethnic languages and cultures,LLMs are prone to cultural appropriation,distortion,bias,resource imbalance,copyright infringement,and security risks due to technical limitations such as data scarcity and insufficient cultural understanding.To address these issues,this paper advocates for the construction of a three-pronged response framework based on"prevention-first,technological empowerment,and collaborative governance."This study aims to promote the building of a responsible and sustainable new ecosystem for ethnic cultural translation,and promote the healthy development of LLMs technologies in this field.
作者
王华树
陈佳慧
WANG Hua-shu;CHEN Jia-hui(Beijing Foreign Studies University,Beijing 100089,China;Jilin International Studies University,Changchun 130117,China)
出处
《语言与翻译》
2025年第4期62-69,共8页
Language and Translation
基金
2022年国家社科基金项目“数字人文视域下译者数字素养研究”(项目编号22BYY043)。
