Exploration Of The Boundary Between Human Translation And Machine Translation

With the development of artificial intelligence,machine translation has become an increasingly popular research field.For some language service providers(LSP),it is a common practice to do translation in MTPE(machine translation post editing)mode.The author believes that translators do not have to worry about being replaced by machine translation,but should embrace new technologies and improve its quality,so that they could be better assisted to increase translation efficiency and meet the requirements of the new era.The author selects the general report part of Annual Report on Development of New Media in China(2018)as the source text.This thesis is divided into five parts.The first part explains the background of MTPE and the significance of choosing this text.The second part introduces the status of machine translation and standards of post-editing.The third part describes the translation task,including preparation before translation and the translation process.The fourth part is the core chapter,in which the author uses five leading neural machine translation engines of Google,Baidu,Bing,Sogou and Huawei to do translation.On this basis,the author does post-editing to obtain the final translation.The author compares five MT results and conducts qualitative and quantitative analyses respectively.Qualitatively,the author analyzes the typical errors in fluency,adequacy,and style made by machine and sees whether MT helps human translators improve efficiency.Quantitatively,the author compares differences among five engines based on TER(translation error rate),and summarizes the characteristics of Chinese sentences corresponding to three TER levels at 0≤TER≤0.25,0.5<TER<1,and TER=1.The fifth part is a summary of the full text,and the author also points out the shortcomings of this thesis here.By writing this thesis,the author has the following gains.First,she has a good general knowledge of the latest development of China’s new media;Second,she gets familiar with the theory and practice of MTPE process;Third,she tries to measure machine translation quality with TER and explores the boundary between human translation and machine translation,hoping to provide some inspiration for the optimization and development of machine translation.