I made a zip file with some English (original)-Spanish (translation) pairs as the latter is my native language. Sorry for the bias. I guess some other users can use those files for testing language-related features. All of them are in the public domain, so it's safe to use them in any context (pun intended). I'll try to tackle other similar problems which were tricky to solve using LaTeX: the format-agnosting expex does fancy things as glosses and the like, but a better approach based purely on ConTeXt capabilities must be possible for linguistics, translation and language learning, considering the fact ConTeXt already outperforms LaTeX wrt language support.
Regards,