მთარგმნელობითი ტექნოლოგიები, სამანქანო თარგმანი და ლინგვისტური ტაგირების პრინციპები

Abstract

The European Masters in Translation network first published its framework for translator and translation competence, including the well-known EMT “Wheel of Competence”, in January 2009. Since that time, the Wheel of Competence has changed several times. In accordance with the changes implemented in 2017, the central part of the EMT framework is application of computational technologies to the translation studies. The needs of translation industry as a whole oblige young translator-interpreters to provide high-quality translation by means of computer technologies. The mentioned competence encompasses six components including Machine Translation (MT) competence. The translator-interpreter should use MT systems as a part of his/her working process. This paper presents an overview of Machine Translation (MT) with regards to three main directions. The first one could be considered as a Rule-Based Machine Translation (RBMT) used in older types of technologies, the second one – as a Statistical Machine Translation (SMT) used in comparatively new types of technologies till 2016 and the third – as a Neural Machine Translation (NMT) used in modern type translation technologies since 2016. All these systems are somehow interconnected between each other and modern machine translation systems reveal a shift from statistical to deep learning practices. As a result of the above-mentioned shift, Google, Pangeanic, KantanMT, Omniscien Technologies and SDL Trados translation technologies use neural machine translation. But all these systems reveal difficulties in processing languages like Georgian. The contents of the paper are as follows: 1. Introduction; 2. Theoretical and technological issues; 3. Components of machine translation with regards to Georgian language and 4. Conclusions.