Potential of Machine Translation in Museum Media Discourse: Analysis of Modern Browser Systems

The article is devoted to the actual problems of neural machine translation optimization, which are caused by the inclusion of multilingualism in the functionality of online resources for providing equal access to information for all users. The results of analyzing the machine translation of English-, German- and French-language content of museums’ media discourse into Russian are based on the usage of built-in browser systems from Yandex, Google and Microsoft. The selection of fragments for comparative analysis is carried out taking into account the random choice of the source language by users. The functional capabilities of machine translation systems, optimized by neural network technologies, are viewed as denotative equivalence provision, correct transference of the majority of proper names and adequate actualization of the lexical units meanings with their reference to the context. Several cases stated that browser-based translators used correspondences without taking into account the context, linguacultural tradition, pragmatic meaning of language units, which demonstrates the limited capabilities of these systems to create an adequate, stylistically and discursively designed translation. It is established that the greatest difficulty lies in the translation of isolated nominations (section names, language selection buttons, etc.), which indicates a high level of artificial intelligence involved in machine translation systems, that imitate humans in their mandatory desire to obtain a context that actualizes the meaning of a word.