Notes on discussion of representation of discourse status and separating it from quantifier rels.
Lars and Dorothee's presentation on the treatment of definiteness in NorSource: We show how the morpho-syntactic encoding of definiteness is treated in NorSource. Semantically NorSource treats definiteness through interlingua relations accounting for deixis, discourse status and genericity.
Everyone seemed convinced that it was worthwhile to unpack definiteness (also demonstratives, other indications of discourse status) from the quantifier rels and into features, most likely features of indices. As part of this, we would expect a fairly flat hierarchy of quantifier relation types.
We spent some time wondering about whether putting the information on indices would in fact cause problems. One potential case was:
- In a language like Greek or Hebrew with definiteness agreement on adjectives (also Norwegian?)
- NP with both an adjectival modifier and a relative clause, where the gap in the relative clause is also modified by an adjective (perhaps as secondary predication).
- Agreement evident on both the upstairs and downstairs adjectives.
If it's possible to have differing agreement in the two places (indefinite in one and definite in the other) *and* we still want to identify the index of the gap with the head noun, then there's a problem.
Berthold suggests that we might want to treat relative clauses as anaphora anyway, i.e., not coindex. (This makes relative clause extraposition more tractable, too.)
- What features are required for representing discourse status, how 'harmonized' can we get across languages?
- What should the inventory of quantifier relations be, and how 'harmonized' can it be across languages?
- How many contrasts among constructionally introduced quantifier relations?
Emily might make a first attempt at adapting the proposal of Borthen and Haugereid 2005 and including it in the Matrix.