Abstract:

Implementation of IT technologies into our lives bring new insights into research, development and education too. Nowadays, machine translation and post-editing are an integral part of translators' work as they facilitate the process of translation. Despite of this fact, not many universities in Slovakia focus on training their students in this field. The aim of the paper is to confirm the need for post-editing as machine translation outputs are still not flawless and they require post-editor's intervention or correction. We identify and analyse the errors of neural machine which was launched for the Slovak language in 2016. We study its quality in the direction English - Slovak and evaluate the errors in terms of Villar's error framework. In comparison with its predecessor – a statistical machine translator – neural machine translation outputs seem to be more fluent but not perfect. According to our research, the most of the raw machine translation output is acceptable, only approximately 15% of the text is unacceptable. The most of the errors concern mainly morphological, syntactic level of the language, then typographical level and punctuation.

Keywords:

Translation, machine translation, translators' training, post-editing.