Emily: Might be able to suspend my allergy…

Dan: Do we want to use the same representation for coreference chains as we do for Binding Theory?

Emily: No, binding theory is potential constraints on coreference chains, not actual coreference statements.

Dan: What about “he craned his neck”? That’s surely the language telling you those have to be coreferent.

oe: Do we really want pron_rel there?

Emily: Maybe like reflexives in French/Spanish type languages where it’s marked morphologically (se-) as opposed to a true independent word. (soi-même is really the emphatic use, not the true reflexive).

oe: Scandinavian language also have inherently reflexive verbs.

Emily: But I think that the sag selv forms are also reflexives that might need a pron_rel, though they are subject to different binding constraints than English -self forms.

Francis: I’ll be marking the reflexive pronouns in reflexive idioms so you don’t have to worry about them in coreference resolution.

oe: He shaved himself — we currently say there’s an x and a y. He bit his tongue, x, y, z possessor of tongue. Binding theory or account of possessive idioms might tell us that x and y are constrained to corefer, or another way of looking at that is that there should be no y.

Emily: For English, I’d want to keep separate pron_rel for the *self forms, but would be happy to see no extra pron_rel in “he bit his tongue”. Does that give us MRSs of problematic shapes?

oe: Don’t think so

Dan: Currently not creating any potential danger for quantifiers into places where they can’t bind their variables — cf odd cases where the tool doesn’t cope right.

Emily: How wild and woolly do they get?

oe: pron(x), bit(x,y) poss(x,y), tongue(y), udef_q(x), udef_q(y)

Francis: They look as though butter wouldn’t melt in their mouth.

Woodley: And can you shave your poor self?

Emily: relevance?

Woodley: Looking for an example where there’s something we wouldn’t want to lose in the restriction of the quantifier.

oe: Not sure that long example will be problematic, but can see why you might worry about it.

oe: Don’t think we’ll actually lose those pron_rels, but can remind ourselves of why they are there.

Woodley: Can you ever get anything interesting on the restriction of the quantifier for the possessive:

You, in the front, close your books. *Your in the front books. Shoot yourself in the front in the foot.

Dan: What about own: He craned his own neck

Francis: Some of those idioms allow it.

Dan/Woodley: Depending on your theory of own, it could be a marker of reflexive, or a modifier. If it’s just a morphosyntactic variant, doesn’t have to make it to the MRS.

Francis: Mind your own business, Paddle your own canoe, Be your own man, give someone a taste of their own medicine.

Dan: So own isn’t productive in the right way for worrying about that restrictor.

Dan: The only thing I can think of that might be worrisome about crossing that embedded clause is if that as though can be in a scopal modifier phrase. smiled as though butter couldn’t melt in his mouth (but that’s not the idiom). Quantifier would be forced to scope outside in order to bind both xs and that would possibly over constrain the possible quantifier scope.

Woodley: But do we care about quantifiers for pronouns?

Dan/Emily: Lower one isn’t a pronoun (tongue, neck) etc in the semantics on this proposal, and the other one didn’t have to be: the children all craned their necks.

Emily: I don’t understand why a clausal complement verb is different from a scopal modifier.

Dan: Because the limited tests I’m using are neg and every scope ambiguities.

Francis: Not every child looked as though butter wouldn’t melt in his mouth.

Emily: Still don’t understand.

Dan: Because if the overt quantifier is in the subject of the upstairs verb, no matter what happens with the rest of the clause, it can scope over the other arguments (scopal or otherwise) of that verb.

Dan: Even though every child chased no dog, butter looked like it wouldn’t melt in its mouth. Negation is only part of the one clause, not both.

[ Bracketing this for now. ]

Last update: 2014-02-17 by StephanOepen [edit]