ChatGPT能否翻出“言外之意”——基于语用推理的AI翻译失误分析
ChatGPT’s Ability to Translate Implicature—An Analysis of Failures in AI Translation Based on Pragmatic Inference

在当前全球化背景下，翻译技术的进步对于跨文化交流具有重要意义，而ChatGPT等翻译工具正被广泛应用。尽管这些工具在处理源文本字面意义方面表现出色，但在理解和传达的非字面意义、隐含意义及语境依赖性方面，仍存在一定的挑战。教师口头禅作为一种承载教育语境和教师意图的语言现象，其翻译准确性对跨文化交流至关重要。首先，本研究旨在分析ChatGPT在翻译教育语境中教师口头禅所蕴含的“言外之意”时的不足，进而探讨基于语用推理的AI翻译语用失误问题。其次，通过对ChatGPT翻译教师口头禅的实例进行深入分析，揭示出AI翻译系统在识别和传达言外之意方面的局限性。
In this globalized era, advancements in translation technology play a paramount role in cross-cultural communication, and ChatGPT is being widely used. While adept at translating literal meanings, these tools often fail to capture non-literal meanings, implicature and context-dependent subtleties. This is particularly relevant for “teacher’s pet phrases,” which are deeply rooted in educational contexts and convey specific teacher intentions. This study investigates ChatGPT’s inability to translate the “implicature” in such phrases accurately, shedding light on the broader problem of pragmatic failures in AI translation systems that utilize pragmatic inference. Through a detailed analysis of ChatGPT’s translations, the author will demonstrate the limitations of AI translation system within recognizing and conveying implicature.