A Report On EC Post-editing Practice Of Mt Under The Guidance Of Reiss’s Text Type Theory

With the rapid development of neural network machine translation in recent years,machine translation has been widely used in the translation industry and has been accepted by customers.The advantages of machine translation are self-evident,such as translation speed.However,the shortcomings of machine translation are undeniable.The intervention of post-editing can,to some extent,improve the accuracy of translation output while retaining the advantage of machine translation speed.Therefore,the MT+PE model is favored by the translation industry.This report also adopts this translation model to explore the common types of errors in machine translation in translating economic policy report and propose corresponding post-editing strategies based on Reiss’ s Text Type Theory.The first chapter of the APEC Economic Policy Report 2020 on Structural Reform and Women’s Empowerment is about twelve thousand words,and the author selected it as the translation practice material for this report.As an economic policy report,it is characterized by long sentences,rich terminologies and abbreviations,and complex sentence structures as well.The author selected one part of the source material and applied three major online translation software,namely Baidu,Youdao and Google,to conduct a trial translation.Eventually,Baidu online translation software was favored,which produced the least errors,as the machine translation software for this report.The author found that errors produced by Baidu MT mainly focus on three levels,namely,lexical,syntactic and discourse level.Lexical errors include mistranslation,omission,under-translation and over-translation;syntactic errors mainly include mistranslation,omission and inappropriate order of words;discoursal errors mainly include inconsistent format,redundancy and lack of coherence.By analyzing the data of error types generated by Baidu,the author found that the errors at the lexical level far exceeded those at the syntactic and discourse levels and common types of errors included mistranslation,inappropriate order of words and under-translation.The author holds that in the process of post-translation editing,some common errors caused by MT can be improved by carefully comparing the MT version with the true meaning of the source text,giving full play to translation strategies such as“semantic revision,syntactic reorganization,addition and deletion”.