大语言模型对批评隐喻分析中隐喻识别的应用性研究  

A Study on the Applicability of Large Language Model(ChatGPT)to Metaphor Identification in Critical Metaphor Analysis

作　　者：于艳春 Yu Yan-chun(School of Foreign Languages and Literature,Heilongjiang University,Harbin 150080,China;School of Foreign Languages,Harbin University,Harbin 150080,China)

机构地区：[1]黑龙江大学外文学院,哈尔滨150080 [2]哈尔滨学院外国语学院,哈尔滨150080

出　　处：《外语学刊》2025年第1期47-56,共10页Foreign Language Research

基　　金：黑龙江省哲学社会科学研究规划一般项目“数字时代红色文化国际传播研究”(23XWB196)的阶段性成果。

摘　　要：批评隐喻分析(critical metaphor analysis,CMA)作为一种语篇的隐喻分析方法,意在揭示语言使用者的潜在意图,不但可深化我们对语言隐喻的理解,还能揭示隐喻在人类思维和交流中的重要作用。隐喻识别作为CMA分析框架中的一个关键步骤,为隐喻阐释和隐喻说明等后续研究工作提供了坚实基础,是CMA中极其重要且不可或缺的一部分。本研究首先对CMA中隐喻识别的融合性和枢纽性进行深入讨论,探究CMA中隐喻识别的发展特性,熟悉CMA的整个分析框架。同时,针对CMA在隐喻识别中存在的客观性不足问题,提出尝试结合大语言模型(ChatGPT)进行隐喻识别的方法,并基于真实研究案例了解该方法的具体实践操作。通过案例,本研究也对使用大语言模型(ChatGPT)进行隐喻识别过程中遇到的问题进行分析和总结,希望通过合理、恰当地应用技术手段,探索人机协作的最佳方式,以提升隐喻识别的客观性,完善CMA的分析框架,更深入地进行语篇的隐喻研究。Critical Metaphor Analysis(CMA),an approach to metaphor analysis in discourse,aims to reveal the hidden intention of language users,which not only deepens our understanding of linguistic metaphors,but also reveals the important role of metaphor in human thinking and communication.As a key stage in CMA,metaphor identification provides a solid foundation for the subsequent research work such as metaphor interpretation and metaphor explanation,and is an extremely important and indispensable part of CMA.This paper first delves into integration and pivotability of metaphor identification in CMA,to explore the developmental characteristics of metaphor identification in CMA,and to become acquainted with the entire analytical framework of CMA.Then,the lack of objectivity of metaphor identification in CMA is discussed,and a method that attempts to apply Large Language Model(ChatGPT)to metaphor identification in CMA is proposed to address this issue.In the meantime,the specific practical operation of the method is demonstrated based on a real case study.It is found that,in addition to enhancing the objectivity of metaphor identification,ChatGPT mainly plays an enlightening role in the analytical process.At all once,three issues have emerged in the process of metaphor identification applying ChatGPT:First,the results of metaphor identification are not entirely consistent with expectations;Second,the frequency statistics of metaphor keywords are inaccurate;Third,the operation process is repetitive.These issues have prompted us to further ponder.Firstly,LLM(ChatGPT)has not yet possessed fully developed human intelligence,and consequently it may have not been capable of independently completing the identification of complex metaphor phenomena involving multiple dimensions such as language,culture,and cognition.It needs to be combined with human identification to play an auxiliary and enlightening role.Secondly,the application of technology to empower language research should be determined based on the research objectives and r

关 键 词：大语言模型 ChatGPT 批评隐喻分析 隐喻识别 隐喻阐释 隐喻说明 

分 类 号：H08[语言文字—语言学]