I used to have an accepted analysis which included a particular lexical type that arises through the morphological interface. Suppose the lexical type was called N_CP*_P-C_LE.
Then I decided to change the interface and to instantiate certain generic lexical entries as N_-_MC_LE rather than N_CP. A different type of noun. I make the change, I recompile the grammar, I run it and I get a tree that want with ACE (LUI displays lemmas instead of inflected forms but the tree is fine), you can see the new lexical type in the AVM to the right:
But when I reparse the profile and attempt to treebank, I have a problem: somehow, the profile still thinks the noun is of type N_CP? Where would it be coming from? I am not even updating the treebank, this is just as if I were to treebank from the beginning.
What am I likely to be doing wrong? This looks confusing but the only explanation I have for now is that I am using new parses where the noun is of a different type (namely, one without complement) but somehow in the profile there are remnants of the old analysis?..