大语言模型与语言对比研究  

Large language models for Contrastive Linguistics

作　　者：秦洪武[1] 周霞[1] QIN Hongwu;ZHOU Xia(School of Foreign Languages,Qufu Normal University,Qufu 273165,China)

机构地区：[1]曲阜师范大学

出　　处：《外语教学与研究》2024年第2期163-176,318,共15页Foreign Language Teaching and Research

基　　金：国家社科基金重大项目“围绕汉语的大型多语汉外平行语料库集群研制及应用研究”(21&ZD290)资助。

摘　　要：本研究探讨大语言模型时代语言对比研究的发展机遇和发展趋势。研究认为,大语言模型具备语义分析能力,能够为比较第三方找到翻译关系之外的选择,提供较传统翻译关系更为客观的、可计量的共享语义信息,为多层次和多维度语言对比提供优质知识资源。研究指出,大语言模型提供的质性判断结果本身可以转换为量化数据,利于开展更具解释力的混合型研究。研究表明,大语言模型的多语言、多层次和多门类知识调用能力超越了人类个体的智能,语言研究中人工智能与人类智能(研究者)协同工作势在必行,具有广阔的应用前景。This study examines emerging opportunities and trends in the field of Contrastive Linguistics driven by the advent of large language models(LLMs).We argue that the semantic intelligence of LLMs offers novel possibilities for establishing tertium comparationis(TC),enabling objective and measurable comparisons that were previously difficult to attain,surpassing the limitations of traditional methods rooted in translational relations.This advancement has led to the creation of high-quality knowledge resources that facilitate multi-level and multidimensional contrastive analyses.This study highlights how the qualitative analysis results generated by LLMs are transformed into quantitative data,empowering more mixed-method research.The study suggests that LLMs excel in multilingual and multi-level knowledge retrieval,and collaboration between LLMs and researchers is imperative and holds immense potential for various research applications.

关 键 词：大语言模型 语言对比研究 比较第三方 

分 类 号：H043[语言文字—语言学]