An experimental study of the impact of machine translation on translation quality

Abstract

The article is devoted to the study of the impact of machine translation (MT) engines (Onlinedoctranslator and DeepLTranslate) on the quality of target text in the field of psychiatry. The translation quality was determined by the number of errors and quality of rendering psychiatric terminology. The modern translation market offers many online engines designed to help with and accelerate the translation of texts in the field of psychiatry, which is in a state of constant developing, evolution and changes. It is important to determine whether there is a difference in the efficiency of different MT engines. In conducting our research, we formulated a hypothesis, selected a psychiatric text rich in psychiatric terminology, compiled a bilingual dictionary based on the selected text, selected MT engines and the procedure of evaluating errors, analyzed target texts in terms of general quality and psychiatric terminology, processed the experimental data and represented the results in tables providing an expert assessment of the empirical data and formulated conclusions. In conducting the experiment, the hypothesis was confirmed: different MT engines provide different quality of translation. According to the results of the study, DeepL Translate was found to be more efficient. However, errors of all types still occur, so a thorough post-editing of the target texts is required. It was also found that there is an important area in the field of psychiatry, which the researched MT systems cannot cope with. This area is a matter of tact, which requires an understanding of the nuances of the source text and appropriate subject knowledge.