INVESTIGATING MACHINE TRANSLATION PROBLEMS BY MEANS OF MULTILINGUAL PARALLEL CORPORA

Abstract

The article deals with the identification and analysis of the main problems that are caused by machine translation. The object of the research is the eTranslation machine translation system, which is used on the official website of the European Union. The main tool of the conducted research is the multilingual parallel corpus EU History, which we have created to compare and analyze texts in different languages. It contains original articles about the history of the creation and development of the European Union in English and their translations into German and Ukrainian. This corpus allowed us to identify and analyze errors and inaccuracies made by the eTranslation system. As a result of the study, several major machine translation problems were identified, including inadequate consideration of polysemy and contextual markers, as the machine translation system often fails to recognize the ambiguity of concepts and does not consider the context in which they are used, leading to translation errors. The machine translation system may also have difficulties translating complex grammatical constructions, abbreviations, and proper names, making the translation inaccurate or unclear. The research revealed the absence of universal methods for transcribing from Latin to Cyrillic, which makes it impossible for the machine translation system to adequately render proper names. This system does not always adhere to the rules of agreement in the Ukrainian language, which also negatively affects the quality of translation. The results obtained indicate that machine translation at the current stage cannot fully replace human translation, especially when it comes to complex texts that require a deep understanding of context and knowledge of language nuances. The article also examines the potential of machine translation in combination with human translation to achieve better results in interlingual communication and access to information.