Our hope is that this page will collect info on MRS wellformedness constraints stated independently of any semantic analyses which motivate them.
- Optionality of arguments: need to know how many roles an EP expects, and provide dummy variables (u37, etc) to fill unrealized roles
- Each referential index needs a quantifier
- Each quantifier's RSTR must be linked to something -- typically a nouny EP, and typically via a QEQ
- Have to get the right labels equal to the right other labels; intersective modifiers and relative clauses, for example.
- Each grammar will probably have its own quirks about where QEQs are required
- Each grammar will have other syntax-determined constraints about what is realizable, e.g. see the MRS below about a cat and a mouse that chase each other.
- For NP fragments, need to supply an unknown_rel that can fit into the lowest quantifier's BODY -- otherwise not scopable.
Possibly Also of Interest
A seemingly wellformed MRS that doesn't correspond to any English syntax. Is it realizable in some other exotic language? (Note that with the second "_chase_v_1_rel" deleted, it generates successfully with the ERG). Is there some additional wellformedness constraint that could be articulated to rule this one out?
[ LTOP: h1 INDEX: e2 [ e SF: prop TENSE: pres MOOD: indicative ] RELS: < [ udef_q_rel LBL: h3 ARG0: x4 RSTR: h5 BODY: h6 ] [ udef_q_rel LBL: h7 ARG0: x8 RSTR: h9 BODY: h10 ] [ "_cat_n_1_rel" LBL: h11 ARG0: x4 ] [ "_mouse_n_1_rel" LBL: h12 ARG0: x8 ] [ "_chase_v_1_rel" LBL: h13 ARG0: e2 ARG1: x4 ARG2: x8 ] [ "_chase_v_1_rel" LBL: h13 ARG0: e14 ARG1: x8 ARG2: x4 ] > HCONS: < h1 qeq h13 h5 qeq h11 h9 qeq h12 > ]