视听文本机器翻译质量测评和译后编辑研究——以《畅游赣鄱——水文化英文之旅》慕课字幕翻译为例

Quality assessment and post-translation editing of audio-visual text machine translation:a case study of subtitle translation for MOOC Dating with Lakes and Rivers in Jiangxi

机器翻译给人们带来了便利,同时促成了译后编辑的诞生。本文以《畅游赣鄱——水文化英文之旅》慕课字幕文本为研究样本,在MQM(Multidimensional Quality Metrics)质量评估框架基础上提出视听文本机器翻译产品质量评估标准,量化其错误值,测试机器翻译质量,并由此提出译后编辑的三级模式:轻度译后编辑(Light Post-editing, LPE)、中度译后编辑(Moderate Post-editing, MPE)和重度译后编辑(Full Post-editing, FPE)。

With the industrialization of machine translation(MT),the language service industry has gained tremendous impetus for rapid development,and post-translation editing(PE)has been consequently turned into an essential part of translation process.Artificial intelligence does bring dividends to translation industry,but,it also has its unavoidable limitations.Taking the subtitle text of MOOC Dating with Lakes and Rivers in Jiangxi as the research sample,this paper proposes the machine translation quality assessment standard of audio-visual text on the basis of MQM quality assessment framework released by the German Research Center for Artificial Intelligence(DFKI),by which the error value is quantified,and its machine translation quality defined.Therefore,three-level model of post-translation editing is proposed:light post-editing(LPE),moderate post-editing(MPE)and full post-editing(FPE).