| Nowadays, with the increasing communication among international academic field, all key journals require an attachment of English abstracts (EAs) to every research paper. Good abstracts are concise and could quickly arouse the interest of international researchers and introduce them to the main idea of the research content. Since EAs are so important, they gained worldwide attention. The previous research of English abstract translating was mainly conducted in lexical, syntactical and discourse level of various professional fields, which were usually concerned with translation strategies and rules. However, the systematic investigation into the agricultural journal paper abstracts has been seldom studied, especially studies base on the corpus made of international and domestic EAs. The fast development of descriptive translation study made it possible to introduce the methodology of corpus study to translation studies, gradually, which provides unique perspective for translation study.Based on the research background above, this paper aims to investigate the following research questions: In what respects do translations tend to differ from non-translated texts? If there is a difference (for instance in the distribution or frequency of a given feature), is this difference indeed significant? Is the average sentence length of the translations shorter or longer than non-translated texts; the distribution of different sentence lengths about the same in the two groups of texts. The ultimate objective of this paper is to let the translators have an acute sense of the difference between two language systems, and give insight into translation practice.To answer the research questions, the comparative model TT(translated text)≈NTT(non-translated text) which was put forward by Chesterman was used. It is an effective translation model, which provides a key theoretical framework for this study. Two subcorpora were constructed. The first 6 abstracts of the latest copy from 5 agricultural journals which are considered domestic authorities, totally 30 made up subcorpus TT; 30 EAs with the same key words as translated text were chosen from international agricultural journals made up subcorpus NTT. The study was conducted through the analysis of the two subcorpora. Webcorp is used as a supplementary tool to subcorpus NTT to confirm related item. Webcorp is an online corpus made by UK from which one can retrieve English language written by people from any country.Through the study of comparable corpora, the major findings are: 1) The differences between TT and NTT represent in all levels. Under the same topic, the length of NTT is longer than TT. The average length of NTT goes over the maximum requirement of 250 ABSTRACT words by ISO. The average length of TT is close to the minimum requirement of 200 words by GB. TT uses fewer abbreviations than NTT; TT uses more"of,"less"moves"and semantic cohesion. 2) The average sentence length of TT is higher than NTT, but they both used more sentences for move 4-results.The study results indicate the differences between translated abstracts and non-translated abstracts are caused by the following reasons: their different academic environment; from the perspective of translators, personal choice of words, stylistic preference, the inadequate knowledge, the lack of sense of English language and translation purpose of translators. Therefore, this paper recommends translators should have a clear translation purpose, adopt"5-move"structure in abstract translation and use various ways of semantic cohesion. Obviously, comparable corpus is an ideal tool to translators. |
PDF Full Text Request