There have been reports already (some slavic language, I don't remember exactly). I'm aware of the issue, and will eventually be fixed. I think it will be in the next major release, most likely 2.3.0. Maybe (well, if there will be any) 2.2.3 will have it too, but that I just can't tell now.
It needs a lot of research, which word is used in two(or more) different contexts, which has to have different translations, etc. If anyone would do the research, providing a list of ambiguous words (for example in the form of a .po file), that would definiately speed up context-sensitive translations entering aMule.