Multiple coindexes per node

[diertani]

which afterwards was made the ground of war

I need the subject of the matrix here to have two indexes: one on the trace, co-indexed with the wh-operator, and another on the syntactic tag, co-indexed with the null subject of the small clause. There is currently no way in annotald to do this.

• If I make the trace subject in the usual way, and then coindex it to the small clause subject, it just duplicates the index on the trace on the *, so that all three end up with the same index.
• If I try to cheat and do it backwards, by making the small clause subject first, then making a matrix subject coindexed with it, and then coindexing the matrix to the wh-operator, it again uses the same index for all three nodes, plus now the one on the trace is on the empty category rather than on the phrasal node.
• If I try making the small clause subject via co-indexation and before-insertion, the index is still the same as the first index, plus it will insert the default ICH empty category rather than *, (I can change the latter, but it doesn't really matter since the indexes are wrong.)

[aecay]

@beatrice57 and I talked about this, and she agreed to think about whether we need two indices in this case, or whether it is possible to treat all three as links on the same chain. (The latter is my not-very-considered opinion.) Since this might require an update to the annotation manual, maybe she can comment.

[diertani]

If we did the latter, we'd have to revise all the extant corpora, correct?

[aecay]

Eventually we might want to, but it's not clear it would be immediately necessary. Another way of putting it is, it's crucial that, (in CS terminology) *T* sameindex which and *T* sameindex *. Is it ever important whether or not which sameindex *? The old way does not have this relationship; the new way does. But if it's really not important, then it can be encoded in either way, without affecting anything.

[aecay]

See also #37.