Sozdanie parallel’nogo korpusa dlja analiza chudožestvennych perevodov detej-bilingvov

This paper presents an experiment of creating a parallel corpus of Russian-Italian literary texts and its analysis on SE (Sketch Engine) to identify the different levels of translation difficulties occurring while comparing the two linguistic systems. Participating as a jury in the competition “Kul'turnyj most” held in Italy since 2018, we have collected very rich and diverse ma- terial for the analysis. Our experiment has shown that natural translation, which plays a major role in the acquisition of bilingualism by so-called heritage speakers, is not sufficient to develop bilin- gual translation competence. Typical for non-professional translation, the word-for-word approach prevails as a result of the contact between the first and the dominant languages.