An Empirical Study Of Translator's Revision Behaviors In Relation To Text Complexity And Translation Experience

In the past several years,the empirical study of translation revision behaviors has been becoming popular in the translation researches.Based on the review of previous studies,it is found that most studies tend to make a surface description of the influence of a single variable,like translation directions and translator expertise,on the allocation of the number of revisions in different translation phases and on different linguistic levels where revisions occur.However,few studies focus on the combined influence of text complexity and translation experience on revision behaviors and the cognitive analysis of revision behaviors are also seldom discussed.Hence,this study addresses three questions as follows:Under the influence of text complexity and translation experience,1)how are revision behaviors allocated in drafting and post-drafting phases? 2)how are revision behaviors allocated on different linguistic levels? 3)how is processing effort allocated to conceptual/procedural encoding revisions? As a part of a National Social Sciences Project(No.15BYY023),research data were collected by the project members together.In this research,32 final year undergraduate students and 30 second year MIT students are asked to translate two texts with different levels of complexity,and their key-logging activities and TAPs data are also collected.Meantime,a 2(translation experience: more;less)×2(text complexity: high;low)mixed research design is adopted.Text complexity and translation experience are the independent variables,and the number of the revisions is the dependent variable.Translation experience is the between-subject variable,and text complexity is the within-subject variable.Moreover,the repeated measures ANOVA testing in the SPSS 19.0.offers the statistical support.The major findings are as follows.1)The number of revisions occurring in drafting phase is higher than that in post-drafting phase,and this distribution is not related to text complexity and translation experience.Meantime,the number of revisions in drafting phase increases with the text complexity.2)The text complexity and translation experience don't influence the allocationof the number of revisions on different linguistic levels.Specifically,the lexical level is found to be the level where most translation revisions occur,and the group of word is the second.The rest two linguistic levels,clausal level and sentence level,are hard to identify which linguistic level is the third and which one is the forth.3)The number of procedural encoding revisions is higher than that of conceptual encoding revisions and thus the procedural encoding revisions consume more processing effort during the translation,which is not influenced by text complexity and translation experience.Meantime,the number of conceptual encoding revisions increases with the text complexity,indicating when facing the more complex text,the translators' effort tends to concentrate on conceptual encoding words,but it is still less than the effort exerted in revising procedural encoding words.What's more,in this research,it is concluded that the number of revisions doesn't become lower as the translation experience increases.Theoretically,this study expands the research scope in terms of revision behaviors in relation to text complexity and translation experience,and further explores the relationship between the processing effort and revision behaviors in translation revision,which offers a new perspective to translation revision research.Methodologically,the present study employs the mixed data collected via key-logging and retrospective TAPs,remedying the weakness of single data-collecting method in previous studies.Pedagogically,since most of translators' effort during the process of revising is spent on lexical level,especially procedural encoding words,the lexical learning should be particularly stressed in translation course.Despite these findings,this research,as with many empirical researches,still suffers from a few limitations in the choice and classification of subjects and ignoring the TT quality,etc.