Machine Translation – The Mechanized Translator
It is true that machine translation will replace some translators. Why? Simply because for some text types, the translation process requires the translator to be translate like a machine. A good example would be EU texts. When we examine the EU Legislation Translation Guide prepared by the Ministry of EU Affairs for the English-Turkish language pair in 2017, we see that standardized forms of expression are transferred to Turkish as standardized expressions as well. For example, according to the guide: the equivalent expression for “Having regard to the consent of the European Parliament,” is determined as “Avrupa Parlamentosunun Muvafakatını göz önünde tutarak,” in Turkish. The important factor here that these two expressions, which are regarded as equivalent, can not subject to change in any type of context whatsoever. In short, it is intolerable and unacceptable to translate the expression “Having regard to the consent of the European Parliament,” except its determined equivalent expression, “Avrupa parlementosunun Muvafakatını göz önünde tutarak,” Then, what is the point of humane nature in translation anyway? Where is the translator’s fingerprint? In that case, for EU text, the translator is already a machine.
Machine Translation – Time Machine
As all my colleagues would be aware, CAT tools divide the texts into segments and the translation process works via these segments. In other words, they take us back to thirty to sixty years prior. Bring us back to the rank-bound and unbounded translation approach which was described by Catford. In terms of CAT tools this rank appears as sentence level. In short, CAT tools lead translators to approach a text from a sentence level through their interfaces. Aside from the text types in which the translators are required to translate like machines; I feel obligated to state that I find this sentece level approach quite primitive and I consider this approach as an insult to linguists who work on discourse analysis. I believe CAT tool developers should be reminded that, aside from the text types I previously excluded, approaching a text through a single rank, in this sense the sentence, can not be considered as an adequate approach.
Machine Translation – Domination of the Abstract
Let us examine the CAT tool translation process a little closer. As I mentioned before, CAT tools reduce the translation to a single rank. The source text is divided into segments and each segment is placed inside of its own little box; and next to these little boxes, there are empty boxes to be filled with sentences via translation, which, in turn, would constitute a target text. When we injected machine translation into the CAT tool, we see that these previously empty target text boxes are filled with machine translation output, assuming an advisory role. These sequence of events may seems useful, however, it may have some consequences. The creativity of the translator peaks when s/he encounters the linguistic elements and processing them in his/her cognition. That is to say, when it comes to texts where creativity is the key, these advisory machine translation outputs may interfere with semiotic and pragmatic processes of a translator in a negative manner. This results in an inevitably synthetic discourses and target texts.