OBJECTIVE PARAMETERS FOR SUBJECTIVE EVALUATION OF TRANSLATION QUALITY

Abstract

Тhe article explores modern tools of translation quality evaluation. Various aspects of translation quality became the focus of attention at different periods of time depending on the approach of research, the needs of the translation industry, and the educational tasks. The interest in the study of evaluation criteria has remained unchanged as the researchers strive to identify universal and the most objective evaluation criteria possible. For a long time, manual evaluation by experts based on the error typology remained the only way to evaluate the translation quality. Modern society demands rapid transmission of large volumes of information, thus creating a need to assess the quality of materials produced by machine translation. These changes lead to modifi cation of the expert evaluation and active development and implementation of automatic metrics. The evolution of expert tools – from formal evaluation tools to systems aimed at evaluating texts of different genres – calls for a comprehensive study of practical solutions in close interrelation with theoretical provisions. In order to overcome the disadvantages, which are primarily associated with the degree of evaluation objectivity, researchers make an increasing number of attempts to improve the set of evaluation parameters and to test their possible combinations. The search for the optimal solutions brings about various versions of evaluation systems. The article considers the Multidimensional Quality Metrics typology, which represents the cutting edge approach to error systematisation by categories and severity criteria. Google has recently presented its version of this typology. The study aim is to identify the distinctive features of the traditional and the updated versions of the typology. We compared their key parameters and tested the typology in evaluation of the materials translated by PROMT. The results revealed the impact of the modifi cations on the degree of objectivity in translation quality evaluation. Further study is necessary to determine whether it is reasonable to use several versions of one evaluation system.