WITNESS OF A PRACTICE:
While working in the software localization field for a big IT company, I figured out how sometimes translators can’t avoid bad translation renderings. I found out that software strings are translated out of context, mainly visual context in this case. Once the translation was passed onto engineers, ‘words’ were assembled to the software itself and just a quality review used to follow this stage of the process. At this stage translators were not allowed to correct so many things (as the main work was already done), while, instead, a lot of errors were spotted when words were finally associated to menus, windows, etc.
Someone said that the translator is a rational agent aware of and controlling his actions and that he/she surrenders to external coercions only if his/her reason decides that it is the wisest thing to do.
I don’t agree with such an out-of-date and utopical thinking.