Error Types And Post-editing Solutions In C-E Machine Translation

In the last decade,huge translation demands have brought unprecedented challenges to the language service industry.Traditional from-scratch human translation is far from meeting the globally ever-increasing demands,hence enabling machine translation to develop.However,machine translation outputs cannot always meet end users’ expectations for quality,thus making post-editing machine translation an effective way to address the challenges.The source text of the translation task is the fifth chapter of High-speed Rail in Color and the machine translated version is exported from Google Translator Toolkit.With reference to the translation quality evaluation model,eight error types of dead translation,mistranslation,missing translation,awkward translation,broken translation,incomplete translation,inconsistency and punctuation are identified from perspectives of accuracy and fluency.The report findings indicate that machine is of poor performance when dealing with professional expressions,four-character phrases with Chinese characteristics,ambiguous phrases and null-subject clauses and sentences.The current machine translated texts have witnessed a demonstrable improvement in quality since what machine adopts in the process of translation is neural machine translation rather than statistical machine translation used several years ago,but most of its outputs are still processed at the syntactic level without enough consideration of the context and far from serving the cohesive and coherent purposes.Solutions including translation strategies,translation methods and translation techniques to each error type are provided in this report and suggestions are proposed as follows.It is advisable to extract terms and work out a bilingual list prior to post-editing to avoid the timeconsuming task of searching for a term and making it a repetitive pattern of action.If practicable,it is recommended to pre-edit the source text in manners that are in accord with the target language’s ways of expression including subject addition and long sentence division.During the process of post-editing,it is sensible to draw support from term extraction tools and quality assurance tools to raise efficiency and quality.