Zusammenfassung
In this paper discuss, from a computational point of view, some misconceptions about machine translation and translation in general. 1 start by making the following three c\'ıa ims: (i) one should concentrate on the differences between the languages in question (not on their similarities) and base translation on contrastive studies; (u) one must pay due attention to the language systems as a whole, and not presuppose equivalence at any level, and (iii) there is no essential difference between machine and human translation: they share several problems and the same misconceptions apply to both. 1 present then briefly the architecture of a particular machine translation system, a set of simple contrastive pairs illustrating several issues, and a descriptive model of translation (the translation network). l also provide some considerations on false friends and translationese, and point to the new trend in computer-assisted trans\'ıation of making use of previous human translations, concluding by criticizing the low status of translation, both human and automatic.
Nutzer