Experimental analysis of economic text translatability indicators

This article analyzes the influence of the specificity of English-language economic texts on the quality of their machine translation into Russian. The study is conducted on the basis of a fairly representative amount of experimental data obtained as a result of applying popular PROMT and Google Translate online systems to the texts on macro- and megaeconomics, one of the most important economic sub-domains, information sources of which require operative presentation of their content in various national languages. In the course of the experiment, the types and quantitative characteristics of translation errors, typical for each of the above machine translation systems, domain in question and the English-to-Russian translation direction were examined. Translatability indicators characteristic for each of the systems that cause the identified types of errors were detected and compared. Based on the analysis of the obtained data, rules for pre-editing the source text were developed in order to eliminate translatability indicators. It is shown that reformulating of original English texts following the pre-editing rules prior to machine translation allows the user significantly improving translation quality without contacting the developers.