摘要
本文以许渊冲、宇文所安和生成式人工智能ChatGPT所译杜甫诗歌英译本为研究对象,从词汇、句子和语篇三个层面进行定量与定性分析。结果显示:与人类译者相比,ChatGPT在翻译中使用的词汇更加丰富多样,且偏好名词,叙事倾向于采用第一人称视角;在词汇使用和翻译风格上更加接近宇文译本,在句式结构上更接近许译本。不过,ChatGPT有时会因过分追求韵律或因人称代词指代不清而导致歧义或影响译文自然度,仍需专业人员进行干预调整,以达到更好的效果。研究指出,大力推动中国文化的国际传播不仅要“善用”,还要“用善”人工智能。
Based on quantitative and qualitative analyses of Du Fu’s poems translated respectively by Xu Yuanchong,Stephen Owen,and generative artificial intelligence ChatGPT,this paper explores the differences and similarities at lexical,syntactic,and textual levels,assessing AI’s potential in poetry translation.The results reveal that compared with human translators,ChatGPT demonstrates a more diverse vocabulary and a preference for nouns and first-person perspective in unfolding the plot.While sharing more similarities in syntactic structures and rhyme patterns with Xu Yuanchong,ChatGPT aligns more closely with Stephen Owen in style and textual effects.Despite AI’s strengths in handling cultural nuances,it occasionally produces ambiguous or unnatural expressions due to pronoun usage or rhythmic emphasis,necessitating human refinement.Findings suggest the importance of using AI both“well”and“wisely”to enhance global cultural exchange.
作者
张拥政
阎睿
于翠红
ZHANG Yongzheng;YAN Rui;YU Cuihong(College of Foreign Studies,Shandong Technology and Business University,Yantai 264005,China;School of Foreign Languages,Beijing Institute of Technology,Beijing 100081,China;School of Foreign Languages,Yantai University,Yantai 264005,China)
出处
《浙江外国语学院学报》
2024年第5期73-82,共10页
Journal of Zhejiang International Studies University
基金
教育部人文社科研究规划基金项目“汉语框式结构二语认知加工机制研究”(22YJA740038)。
