I'm trying to find out if the equivalence problem (or its complement) for context-free and context-sensitive languages is semi-decidable.
In my opinion, the complement is decidable for both classes: since the grammar rules are noncontracting (if I use Chomsky normal form for CFG), I can systematically create words of length 1,2,3,... for both languages. If I find that they differ, I know the languages are not equivalent.
Am I right? Is there some result published somewhere? I couldn't find anything in any of my textbooks.