A Research On Post-editing Methods For Military-Political Texts

With the growing international military exchanges,there is an ever-increasing demand for Chinese translations of foreign military-political texts.Meanwhile,with the development of science and technology,the capability of machine translation has been continuously improved,and its performance on certain types of texts is close to that of human translators.Military-political texts are mainly informative and are relatively suitable for machine translation,and the translation model of “machine translation +post-editing”(MTPE)is increasingly embraced by translators.Therefore,it is necessary to study this new model of translation practice.This dissertation is a translation report of MTPE.The author first uses Google Translate to translate the full text of the article entitled Turnaround: The Untold Story of the Human Terrain System from the US military journal Joint Forces Quarterly,issue78,2015,and then post-edited the translation.This report attempts to summarize the basic procedure of post-editing of military-political texts and errors tend to occur in machine translation,and explore the causes of errors from the perspective of principles of machine translation.Based on Carl James’ classification of errors in Error Analysis theory,the author classifies the errors in English-Chinese machine translation of military-political texts into three first-level errors: substance errors,text errors and discourse errors;and subdivides each of the first-level errors into several second-level errors,and they are totally eight types: format errors,messy code,mistranslation of proper nouns,mistranslation of non-proper nouns,morphology errors,syntax errors,confusion,incoherence and style deviation.In practical,the author divides the process of post-editing into six steps:identification of parallel texts,the establishment of glossary,correction of errors,polishing,the establishment of titles,proofreading,to initially establish a standardized procedure for post-editing.After post-editing,based on the statistical results and related analysis,the author presents three findings: firstly,lexical mistranslation occurs most frequently,secondly;incoherence also occurs frequently;thirdly,machine translation is insensitive to the differences between Chinese and English.At last,the author tries to explain the causes of errors in machine translation from three perspectives: the limitation of the encoder-decoder model,the lack of prior knowledge and the limitation of vocabulary,to provide a reference for future development of machine translation.After this practice,the author thinks that the MTPE model can significantly improve the translation efficiency,but post-editors need to sufficiently correct the errors and polish the machine translation,so that the final translation can be free from the hints of machine translation and of high quality.