Reentrancy in feature structures

[Luis Casillas]

Am I missing something, or is it possible to do away with reentrancies
completely in models for HPSG, and have trees instead of dags as the
modeling structures?

This would work as follows, very roughly:

1. The modeling structures are trees, with the points labeled with types
   and the edges with feature names.

2. A constraint is interpreted as a set of trees that satisfy it.

3. A tree satisfies an atomic value constraint [PATH val] iff there is a
   path with following the sequence of labels in PATH that leads to a
   node labeled with type val.

4. A tree satisfies an atomic path equivalence PATH1 = PATH2 iff the
   the subtree rooted at the node reached by following PATH1, and the
   subtree rooted at the node reached by following PATH2, are identical.

As far as I can see, reentrancy is just a way of guaranteeing that 4
will hold.  Am I missing something?

--
Luis Casillas
Department of Linguistics
Stanford University
http://www.stanford.edu/~casillas/