COMPARATIVE ANALYSIS OF LITERARY AND MACHINE TRANSLATION (A CASE STUDY OF THE FRAGMENTS OF S. PLATT’S NOVEL “THE BELL JAR”)

Abstract

The article offers a comparative analysis of the translation of literary text fragments carried out by automated online translation systems MateCat and Pragma 6 On-Line. The content of the concepts of adequate and equivalent translation is highlighted, which became the starting point for assessing the quality of machine translation. The choice of machine translation programs reflects two global strategies in terms of machine translation software modeling: rulebased and statistical machine translation systems. The translation transformations applied by the author of literary translation Olga Liubarska are characterized, serving as a standard of quality that meets the criteria of equivalent / adequate translation: accurately reproduces the content and renders the communicative attitudes of the author. Translation errors and flaws of automated online translation programs have been examined in detail. The detected errors were grouped according to linguistic levels, forming lexical, morphological and syntactic types of errors. Lexical errors include the use of words in uncharacteristic meaning, which does not correspond to their contextual environment; untranslated words and phrases, word-for-word translation, the use of Russianisms; inability to correctly interpret phraseological units, incorrect translation due to errors in the source text, etc. Morphological errors consist of incorrect rendering of parts of speech in cases of grammatical homonymy, violation of grammatical categories, in particular, the category of tense of verbs, errors in the formation of case endings of nouns and adjectives. Syntactic errors include violation of semantic relations in the construction of phrases and sentences, namely coordination between sentence members; incorrect use of conjunctions within syntactic constructions, violation of word order in the sentences.