Exploring English-Chinese machine translation on cross level of s-to-p and p-to-s from the perspective of artificial intelligence
人工智能视域下英汉互译句-段和段-句跨层面机译探讨

摘要

有了人工智能的加持，机器翻译如虎添翼，更受推崇和青睐，变成了译者们必不可少的工具。但这是否意味着机器翻译已经出色到可以独立完成所有的翻译任务呢？中文是意合型语言，英语是形合型语言，意合与形合之间的转换是翻译活动中常见的难点。本文通过ChatGPT-4o、通义千问等人工智能进行试验，发现在英语长句与汉语语段的跨层面互换翻译上，机器翻译虽然中译英的表现优于英译中，但在形合与意合之间的转换表现都不理想，出来的译文并不能很好地做到形合与意合之间的转换，需要人工干预，进行大幅度的结构重组，完成从意合向形合化，或从形合向意合化的转换，才能达到理想的的译文效果。这说明到目前为止，在处理中英句段层面的文本翻译时，机器翻译仍不能解决这两种语言之间表达习惯的差异转化，译文仍旧需要人工译者加以完善。

Abstract

With the support of artificial intelligence, the more powerful and more favored machine translation has become the indispensable tool for translators. But does it mean that the AI-empowered machine translation can independently and perfectly undertake all translation projects? This paper probes into this issue by experimenting with ChatGPT-4o, Tongyi Qianwen, and other well-known software to discover and prove the performance of AI-empowered machine translation. The findings reveal that from the perspective of cross-level translation from English long sentence into Chinese paragraph (S-to-P) and Chinese paragraph into English long sentence (P-to-S), the latter seems better. But since Chinese is paratactic language system and English is hypotactic language system, translation between Chinese and English means transformation between these two features. Our experiment proves the big defect in the transformation, whether from E to C or C to E. So, human translator’s intervention is required for desired effect with macro syntactical reconstruction and reorganization if necessary.