Lexicogrammatic translationese across two targets and competence levels

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 4102–4112 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 4102 Lexicogrammatic Translationese across Two Targets and Competence Levels Maria Kunilovskaya*, Ekaterina Lapshinova-Koltunski** University of Wolverhampton* and University of Tyumen*, Saarland University** and University of Hildesheim** Wolfruna Street, Wolverhampton, WV1 1LY, UK, Saarbrücken Campus A2.2 Germany maria.kunilovskaya@wlv.ac.uk, e.lapshinova@mx.uni-saarland.de Abstract This research employs genre-comparable data from a number of parallel and comparable corpora to explore the specificity of translations from English into German and Russian produced by students and professional translators. We introduce an elaborate set of human-interpretable lexicogrammatic translationese indicators and calculate the amount of translationese manifested in the data for each target language and translation variety. By placing translations into the same feature space as their sources and the genre-comparable non-translated reference texts in the target language, we observe two separate translationese effects: a shift of translations into the gap between the two languages and a shift away from either language. These trends are linked to the features that contribute to each of the effects. Finally, we compare the translation varieties and find out that the professionalism levels seem to have some correlation with the amount and types of translationese detected, while each language pair demonstrates a specific socio-linguistically determined combination of the translationese effects. Keywords: translation norms, parallel corpora, translationese, multivariate analysis, translation competence, translation varieties, contrastive analysis, machine learning 1. Introduction: Aims and Motivation This paper presents the analysis of the linguistic speci- ficity of translations with English as a source language made by professional and student translators into Russian and German. The peculiarity of our setup lies in the na- ture of the resources we are using. Advantage is taken of the existing resources developed within several inde- pendent projects. To ensure consistency and interoperabil- ity of the research data, we calculate the subsets that are genre-comparable across all the corpora used. We intro- duce an extensive and elaborate list of morpho-syntactic and text-level translationese indicators that are shared by the three languages involved. They are used to measure the overall degree of translationese in both translation pairs and to establish the differences in the professional norms manifested in our data. Furthermore, we automatically de- compose the observed translationese effect into the two types: (1) source-language-induced translationese (‘shin- ing through’) and (2) language-pair-independent transla- tionese (‘non-shining’). The translationese indicators that contribute to each of the trends are used to describe and compare professional and student translations. On the one hand, we show that our hand-engineered and linguistically motivated feature set can reliably differentiate translations and non-translations, regardless of the language pair and the competence level of translators. On the other hand, we try to measure the proficiency level in translation as the amount of translationese assuming that translator edu- cation is about learning to overcome the natural tendencies typical for the translation process (Bernardini and Castag- noli, 2008). Thirdly, we demonstrate that, although the two translation varieties share most …

Content truncated to 500 words