Most (all?) languages in LT don't have their own spell checking, they simply use hunspell one way or the other, sometimes with extensions of the dictionary. Thus, integrating this into the POS tagger isn't so easy I think, as we still want to be able to update to more recent hunspell dictionaries in the future. Also, the uncompounder isn't relevant to that many languages, and German already has one (jWordSplitter) that works well enough. It's indeed unfortunate that jWordSplitter has yet another dictionary, but I'm not sure it makes sense to spend time on changing that.