A Quantitative Study Of Mediation Universals Based On CONTEC

After reviewing the conceptual and theoretical background of relevant studies, particularly nonnative (or ELF) writing and translation studies, the present study, based on an English monolingual corpus, CONTEC, is intented to investigate "mediation universals". The corpus consists of three subcorpora:translational (TR) corpus, nonnative writing (NNW) corpus and native speaker’s (NS) corpus, with each in two genres, namely news and fiction.The main purpose of this study is to investigate whether translational and nonnative texts share such "universal" linguistic features as simplification, explicitation, normalization, levelling out, traditionally named "translation universals". A Three-Phase Comparative Analysis has been conducted to verify the hypothesized mediation universals via the data in the three subcorpora of TR, NNW and NS by means of WordSmith tools, L2 Syntactic Complexity Analyzer, and Readability Analyzer 1.0, etc.As regards simplification, the results between the genres of fiction and news suggest that the hypothesis of simplification is not so simplistic, but a rather subtle phenomenon of hybridity both at the lexical and syntactic levels. The similarity between the translational and nonnative texts is also observed, hence supporting the hypothesis of mediation universals.The results of the cross-comparisons of conjunctions between the three subcorpora strongly support the explicitation tendency. It is believed that explicitation is a universal strategy inherent in the process of language mediation, as in nonnative and translational texts.Though sensitive to different genres, the data of passives seem to provide evidence in support of the normalization hypothesis.The conflicting results of standard deviations between translational, nonnative and native texts seem to reject the existence of levelling out and exhibit the difficulty of measurement.