MACHINE TRANSLATION: COMPARISON OF WORKS AND ANALYSIS OF ERRORS MADE BY DEEPL AND GOOGLE TRANSLATE.

Abstract

The article is focused on the study of machine translation errors on the example of the results of DeepL and Google Translate. The aim of the study is to compare the results of these services on the basis of literary and journalistic texts. The following methods were used to achieve this goal: theoretical analysis, descriptive, comparative, contextual, deductive, and quantitative methods. The results of this study make an important contribution to further detailed research on machine translation services and provide a basis for improving the algorithms of these services. The article will also be useful for researchers who want to deepen their knowledge in the field of translation. The discussion of the results shows that there is currently no firm opinion in favor of one of the above-mentioned competitor services, as the quality of translations by machine translation services varies from year to year. The conclusions of the study present the results of the analysis of the services, namely: DeepL made fewer errors in general than Google Translate. Therefore,translations from DeepL are considered to be of higher quality than translations from Google Translate on the basis that post-editors need more time to process and edit translations from Google Translate. The study is of great novelty, as the constant updating and improvement of machine translation systems makes previous studies obsolete today. It is also one of the first studies for the German-Ukrainian language pair. The results are of great practical importance for practical, lecture and seminar courses in translation-related disciplines. The results can also be used as a basis for a more detailed study of the process of each individual stage of translation or translation programs.