Machine translation of Russian connectives into French: errors and quality failures

The paper shows what machine errors and quality failures may occur when translating connectives. To that end, a statistical machine translation (SMT) system has been used in order to generate translation samples. The opening part presents a brief retrospective on how machine translation has been developing for over 60 years; it sets out the necessary background and provides the context. Also, section 1 explains how an SMT system works. Then, the paper takes a closer look at the problem of evaluation of machine translation quality. Several approaches to classifying machine translation errors are considered to finally attempt a taxonomy that covers specifically the errors central to translation of connectives (from Russian into French). The closing section provides examples of these machine translation errors.