电气工程文本机器翻译的译后编辑例析

Case Study on Machine Translation and Post-Editing of Electrical Engineering Texts

摘要: 近年来以百度神经网络为代表的机器翻译发展迅猛，对于某些正式文本的翻译质量高达70%，但机器译文仍需人工审校与修改，因此以机器翻译为基础的译后编辑尤为重要。本文以电气工程学术论文的英汉机器翻译译文为研究对象，从词汇、句法和语篇三个方面对该类文本机器译文和人工译文进行对比分析。研究发现，机器译文在词汇层面的错误主要是术语错译、缩略词漏译；句法层面的错误主要是长难句错译、短语错译、分词错译；语篇层面的错误主要是句内句际连贯不当。上述问题在译后编辑时通过逆序法、主述分切法等策略运用均得到解决，改进了电气工程文本英汉机器翻译译文的质量。

Abstract: In recent years, machine translation, represented by Baidu Neural Network, has made rapid ad-vancements, achieving translation quality of up to 70% for certain formal texts. However, ma-chine-translated texts still require manual proofreading and editing, making post-editing based on machine translation particularly important. This study focuses on comparing and analyzing ma-chine-translated and human-translated texts of academic papers in the field of electrical engineer-ing from three aspects: vocabulary, syntax, and discourse. The research findings indicate that vo-cabulary errors in machine-translated texts primarily involve terminology misinterpretation and omission of abbreviations. Syntax errors mainly include mistranslation of long and complex sen-tences, phrases, and word segmentation. Discourse errors mainly manifest as improper coherence within and between sentences. These issues are effectively addressed during post-editing through strategies such as order reversing and subject-predicate separation, thereby improving the quality of English-Chinese machine-translated texts in the field of electrical engineering.