Adequacy of Economic Terminology Translation. The Advantage of Human Translation over Machine Translation?

Abstract

The article examines the specific features of translating economic terminology and the problem of ensuring its adequacy. The translation of economic texts functions as an important means of interinstitutional communication, directly influencing market behavior and serving as a tool for ensuring compliance with international norms and standards. The importance of specialized translation of economic texts is confirmed by the practice of multilingual publication of documents by international organizations. Institutions such as the World Bank, the European Central Bank (ECB), the Organisation for Economic Co-operation and Development (OECD), and others provide access to their documents in multiple languages, making translation accuracy critically important in such texts.

Accordingly, consistency and precision in rendering economic terminology are essential, since ambiguous interpretation of terms or phrases in documentation may lead to misinterpretation of provisions, create legal gaps, or complicate the fulfillment of contractual obligations by the parties involved. In this context, the problem of evaluating the quality of economic terminology translation – particularly through a comparison of human translation and neural machine translation systems – gains special scientific and practical relevance.

The active implementation of machine translation technologies contributes to the optimization of translation processes and the reduction of translation costs. The economic efficiency of automated solutions facilitates their increasing integration into modern workflows. Therefore, a comprehensive study of the operational parameters and capabilities of machine translation systems has attracted considerable attention from both the academic community and business sectors.

The study analyzes the terminological adequacy of translated economic vocabulary and provides quantitative calculations of translation errors. The results demonstrate that the translation produced by a neural machine translation system contained approximately 43 errors (about 8.15% of the total), whereas the human translation contained only about 3 errors (0.56%). The findings indicate that although neural machine translation can achieve a relatively high level of adequacy (approximately 91%), this level still does not fully meet the requirements for translating specialized economic texts.