明晰化视阈下《红楼梦》视觉行为动词的人机翻译对比研究

A Human-Machine Translation Comparative Study of Visual Behavior Verbs in A Dream of Red Mansions from the Perspective of Explicitation

摘要: 本研究以《红楼梦》中与林黛玉相关的“看”“见”类视觉行为动词为研究对象，在明晰化视阈下，对比分析杨宪益、戴乃迭译本，霍克斯、闵福德译本，以及ChatGPT和DeepSeek生成的AI译文所采用的英译策略。研究发现，在本文所选例句中，人工译本在处理心理层面视觉行为动词时，更倾向于结合完整语境进行语用与情感明晰化，从而强化林黛玉敏感、多思等人物特征；相比之下，在相同提示词与孤立句输入条件下，两组AI译文更侧重于形式对应与表层语义传递，在人物心理信息的深度显化方面相对有限。研究揭示了在林黛玉相关“看”“见”类视觉行为动词翻译中，人工译本与AI译文在明晰化策略及人物心理信息呈现方式上的差异，可为相关人机翻译研究提供参考。

Abstract: This study examines the visual behavior verbs “kan” and “jian” associated with Lin Daiyu in A Dream of Red Mansions. From the perspective of explicitation, it compares the English translation strategies adopted in the versions by Yang Xianyi and Gladys Yang, David Hawkes and John Minford, as well as the AI-generated translations produced by ChatGPT and DeepSeek. The findings show that, in the selected examples, the human translations are more inclined to draw on the full context to achieve pragmatic and affective explicitation when rendering psychologically oriented visual behavior verbs, thereby reinforcing such character traits of Lin Daiyu as sensitivity and thoughtfulness. By contrast, under the same prompt and isolated-sentence input conditions, the two sets of AI-generated translations place greater emphasis on formal correspondence and the transmission of surface meaning, and are relatively limited in the deeper explicitation of psychological information. The study reveals differences between human and AI translations in their explicitation strategies and in their representation of psychological information when translating the “kan”- and “jian”-type visual behavior verbs associated with Lin Daiyu, and provides a reference for related research on human-AI translation.