A Comparative Study Of Human Translation And Machine Translation Post-editing Based On Metaphor

With the advancement of globalization,the global demand for translation is growing dramatically with frequent international information exchange.This is where the new translation mode of machine translation and post-editing comes in.To meet the needs of language service market,MT post-editing is increasingly being used as an alternative for traditional human translation.Benefit from its advantage on productivity,this new mode has developed rapidly in the language service industry.Besides,it has also attracted research interest from many translation scholars.Metaphor translation has always been an important theme in the field of translation studies.With the innovation of research tools,many scholars have begun to pay attention to the cognitive process of metaphor translation,but there are few researches on post-editing machine translated metaphors.Therefore,this thesis seeks to explore the applicability of post-editing in metaphor translation,aiming at deepening the understanding of the post-editing process of metaphor translation,and providing implications for metaphor translation practice and post-editing training.The independent variable of this study is translation mode,namely,human translation and post-editing.Based on the triangulation of eye-tracking,keystroke logging,retrospective protocols and questionnaires,a comparative study was carried out to explore translation speed,cognitive effort,translation quality and translation strategies of 30 student translators in post-editing and manually translating metaphors from English into Chinese.The purpose is to verify the following four hypotheses for translating linguistic metaphors from English into Chinese:(1)post-editing is faster than human translation;(2)post-editing requires less cognitive effort than human translation;(3)post-editing produces higher translation quality than human translation;(4)in post-editing,M-M(direct transfer)is used more frequently than in human translation,while M-P(paraphrase)is used less frequently than in human translation.The findings are summarized as follows.Firstly,while dealing with linguistic metaphors,the average duration per word in post-editing was significantly shorter than that in human translation.Hypothesis One was verified,i.e.post-editing was faster than human translation.Secondly,fixation counts,fixation duration in ST AOIs and TT AOIs,pause duration and PWR(Pause to Word Ratio)for post-editing were all lower when compared to human translation,although the differences between fixation counts,fixation duration in TT AOIs and pause duration were not significant.It suggested that post-editing could be cognitively less effortful than human translation when only fixation counts,fixation duration in ST AOIs and PWR were taken into account.Hypothesis Two was partly verified.Thirdly,compared to human translation,both the fluency and accuracy of metaphor translations were higher in post-editing,which verified Hypothesis Three,i.e.metaphor translations produced by post-editing were of significantly higher quality than those produced by human translation.Fourthly,M-M strategy was more frequently used,and M-P strategy was less frequently used in post-editing than in human translation.And there was significant difference,which verified Hypothesis Four.In addition,M1-M2(substitution)strategy and M-?(omission)strategy were both more frequently used in human translation,indicating that metaphor translation strategies were more diversely used in human translation than in post-editing.Although further work will need to be done to investigate thoroughly the differences between human translation and post-editing in metaphor translation,our findings would seem interesting in that they provide a new perspective towards the long-established assumptions about the unlikelihood of post-editing texts rich with metaphorical expressions.