Leveling Out in Chinese Translated Fiction
|School||Southwest Jiaotong University|
|Course||Foreign Linguistics and Applied Linguistics|
|Keywords||Corpus-based Translation Studies Comparable Corpora DIY Corpus Levelling out|
This corpus-based study is to testify the hypothesis whether levelling out in terms of lexical, stylistic and syntactic features exists in Chinese translated fiction by corpus approaches. Levelling out refers to the tendency of translational texts gravitate towards the center of a continuum, that is, in this paper compared with Chinese non-translated fiction Chinese translated fiction displays higher level of convergence on three aforementioned levels. In particular, Chinese translated fiction differs significantly from Chinese non-translated fiction with respect to individual features and statistically it indicates that these two groups come from two different populations. In addition, standardize deviations of linguistic features within sub-corpora in corpus of translated fiction are expected to be less than their counterparts in corpus of non-translated fiction.The present paper externalizes lexical, stylistic and syntactic features into seven quantifiable features:lexical density, lexical variety, average sentence length, text readability, the proportion of discourse markers, Part of Speech-Ngrams, and the counts of passive constructions. The scores of these specific features are retrieved from self-built Corpus of Chinese Translated Fiction (CCTF) and the comparable corpus fiction sub-corpora in The Lancaster Corpus of Mandarin Chinese (LCMC) for the comparison of standardize deviation within sub-corpora per se and statistic tests run to test if significant differences exist between two corpora in order to confirm or refute the hypothesis of Universal of levelling out.The researcher chooses the fiction part in LCMC as the comparable corpus but resamples its subcategory of adventure fiction and retags the all five categories with the same software applied to tag CCTF corpus to make it comparable to the DIY corpus, which takes the same sampling frame as LCMC in order to increase their comparability. CCTF contains five genres totaling about 460 thousand characters within one hundred and forty-five 3,200- character text extracts.With respect to the Universal of Levelling out, the researcher postulates that if levelling out was tenable the results would reveal a significant difference between two comparable corpora as far as aforementioned individual corpus features are concerned, and at the same time smaller standardize deviation should be assumed within sub-corpora in CCTF. The research results turned out that CCTF the corpus in question only significantly differs from LCMC the comparable corpus in terms of lexical density and the proportions of discourse markers to running words with smaller standardize deviation within sub corpora. Statistically speaking. CCTF does not deviate from LCMC at a significance level with regard to other individual corpus features.Results show that what Baker hold in esteem the universal of levelling out does not hold with CCTF in terms of aforementioned corpus features. The researcher, however, maintains that this does not necessarily refute the validity of levelling out. but on the contrary, it necessitates that more quantifiable features should be explored to compute the variance between two corpora. The researcher also proposes that universals of translation tend to interact with each other and they are by no means an either-or choice, and holds that the validation of one universal does not presuppose the falsification of another, and even a given universal dominates in the polysystem of features, another may strive to displays itself in some specific aspects.