现代汉语歧义句的AI翻译研究——ChatGPT 4.0与“文心一言 3.5”的汉韩翻译对比
A Study on the AI Translation of Ambiguous Sentences in Chinese ― A Comparative Analysis of ChatGPT and ERNIE Bot in Chinese-Korean Translation
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
本文以汉语歧义句的汉韩翻译为研究对象，重点分析ChatGPT4.0与“文心一言3.5”两款AI在处理不同类型歧义句时的表现差异，揭示其优劣势及局限性。研究从词汇、语法、语义三种歧义类型出发，评估两者在翻译中对双重语义的解析能力，旨在探讨AI翻译工具在高复杂度语境中的适配能力和改进方向。研究结果显示，ChatGPT4.0在词汇歧义处理上表现较优，能更灵活地结合上下文生成译文，双重语义解读的覆盖率显著高于“文心一言3.5”。然而，在语法和语义歧义的处理上，两者表现差距不大，且均存在双重语义解析不足、复杂语法结构处理能力有限的共性问题。此外，针对多义负载型句子的翻译，二者的语义解析和语境适配能力仍显不足，常倾向于单一解读，未能充分体现语境的多样性。本文不仅揭示了两款人工智能在歧义句翻译中的表现边界，还指出当前AI翻译在高阶语义处理中的技术瓶颈，为未来AI翻译技术的优化方向提供了参考。
Abstract
This study examines the translation of modern Chinese ambiguous sentences into Korean, comparing ChatGPT 4.0 and Ernie Bot 3.5. It evaluates their ability to interpret dual meanings across lexical, grammatical, and semantic ambiguities, identifying strengths, weaknesses, and areas for improvement in handling complex contexts. The results show that ChatGPT 4.0 excels in lexical ambiguities, leveraging contextual information more effectively and providing broader dual-meaning coverage than Ernie Bot 3.5. However, for grammatical and semantic ambiguities, both tools perform similarly, with common issues such as insufficient dual interpretation and difficulty handling complex structures. Both also struggle with high semantic load sentences, often defaulting to a single interpretation and lacking contextual adaptability. This study highlights the performance limitations of ChatGPT 4.0 and Ernie Bot 3.5 in ambiguous sentence translation and identifies key challenges in advanced semantic processing, offering insights for future AI translation optimization.