面向AI机译的时政块状话语对应研究  

An AI-Oriented Study of Corresponding Bilingual Lexical Chunks in Political Discourse

作　　者：赵振强 刘绍龙 Zhenqiang Zhao;Shaolong Liu(Zhejiang Yuexiu University,Shaoxing,Zhejiang)

机构地区：[1]浙江越秀外国语学院,浙江,绍兴

出　　处：《教育研究前沿(中英文版)》2024年第4期73-83,共11页Education Research Frontier

基　　金：浙江省教育厅科研项目资助(Y202455035);科研项目“文化翻译视域下新闻话语对外传播效能的优化策略研究”研究成果。

摘　　要：本研究以功能翻译理论为基础,设计四类Prompt指令,探讨ChatGPT翻译中国时政话语中三字格四字格潜隐施事义的表现,并通过多维显译分析框架对比其译文与官方译文的差异。研究表明,ChatGPT在处理文化内涵深厚的语言结构时,即使通过特定的Prompt进行引导,仍然难以准确传递原文的深层含义。此外,不同的Prompt指令会显著影响翻译策略的选择,从而导致译文在连贯性和一致性方面存在不足。尤其是在涉及政治术语和文化特定表达的翻译中,ChatGPT表现出明显的局限性,容易产生误解或不当表达。同时,翻译系统在跨文化传播中,对于敏感政治文化内容的理解和处理能力也显得较为有限。研究揭示了当前主流AI翻译技术在翻译中国时政话语时的局限性,并为改进机器翻译系统提供了方向。This study,grounded in Functional Translation Theory,designs four types of prompt instructions to investigate how ChatGPT translates implicit agentive meanings within three-character and four-character structures in Chinese political discourse.A multidimensional analytical explicitation framework is applied to compare ChatGPT’s translations with official ones.Research indicates that ChatGPT struggles to accurately convey deeper meanings when handling language structures with rich cultural connotations,even when guided by specific prompts.Furthermore,different prompt instructions significantly influence the choice of translation strategies,leading to deficiencies in coherence and consistency in the translated text.ChatGPT demonstrates notable limitations particularly in translating political terminology and culture-specific expressions,often resulting in misinterpretations or inappropriate renderings.Additionally,translation systems show limited capability in understanding and processing sensitive political and cultural content in cross-cultural communication.This study underscores the current limitations of mainstream AI translation technologies in handling Chinese political discourse and offers insights for improving machine translation systems.

关 键 词：时政块状话语 功能翻译理论 认知加工范式 潜隐施事义 显译 

分 类 号：H31[语言文字—英语]