机器翻译错误类型分析与译后编辑策略 ： 以新加坡 YouTube 创作者的视频字幕为例 = Machine translation error type analysis and post-editing strategies : a case study on Singaporean YouTube creators' video subtitles

• Main Author: 王筱玥 Ong, Xiao Yue
• Other Authors: Arista Kuo
• Format: Final Year Project
• Language: Chinese
• Published: Nanyang Technological University 2021
• Subjects: Humanities::Language::Chinese
• Online Access: https://hdl.handle.net/10356/146862
• Institution: Nanyang Technological University

随着计算机技术的发展完善，机器翻译在网络视频中的应用越来越广泛。为了满足YouTube字幕翻译的需要，YouTube自带机器翻译的功能。机器翻译能够提高翻译效率，降低成本，但译文质量仍无法达到像人工译者一样准确。由于机器翻译的局限性，为了呈现准确无误的译文，译后编辑可以在提高翻译效率的同时保证翻译质量，是必不可少的。 作为YouTube的使用者，笔者在观看新加坡视频创作者的视频时，发现YouTube的机器翻译系统对新加坡英语对话的中文翻译的质量仍有进步的空间。因此，本文以知名新加坡视频创作者Ryan Sylvia (Night Owl Cinematics) Food King系列节目中的对话为例，通过实例分析，分析机器翻译存在的典型错误类型，并论证机器翻译在对词汇和句法结构的处理的表现，以展示“机器翻译加上译后编辑”模式的应用价值，同时对比机器翻译译文和译后编辑译文，并探讨相对应的译后编辑技巧。 With the development and improvement of computer technology, the application of machine translation in videos on the internet has become more extensive. In order to meet the demands of subtitle translation, YouTube has developed its machine translation function. Machine translation can improve efficiency and reduce costs, but it is unable to achieve the same accuracy level as a human translator. Due to the limitations of machine translation, to present more accurate translations, post-editing is indispensable in improving translation efficiency while ensuring translation quality. As an avid YouTube user, it is evident that YouTube's machine translation system still has room for improvement in the quality of the Chinese translation of Singapore English conversations in the videos uploaded by Singaporean YouTubers. Therefore, this thesis will focus on the dialogues in the Food King series videos, which are uploaded by the well-known Singaporean YouTuber Ryan Sylvia (Night Owl Cinematics). Through case analysis and comparison between machine-translated text and post-edited text, this thesis analyses the types of typical errors made by machine translation and the performance of machine translation from a lexical and syntactic level, to demonstrate the application value of the “Machine Translation plus Post-editing” model, along with the corresponding post-editing strategies.