Skip to content

LadUW20161013

GlennSlayden edited this page Oct 13, 2016 · 1 revision

ZhenZhen: Recap of last time: we went back and tried some solutions. It worked for pronouns but not for proper nouns.

[...scribe setup time...]

Emily: The rules that do something special are Head-Modifier.. If you look at non-zero-arg, that's the one that says 'my slash is empty'. But basic-one-arg; I take my slash req que from the sole argument.

Francis: zero-arg-non-slash does not inherit from basic-zero-arg... In fact nothing inherits from basic-zero-arg. Looks like a bug.

Emily: Laurie, could you please enter a ticket

Emily: [examining various 'basic-*-phrase' types.]

Francis: we found a further problem in optional subject. It wasn't closing them off in the subject.

ZhenZhen: we have a rule for optional subject. [shows 'opt-subj-declarative-phrase']

Emily: this rule is referring to ARG-ST, which it shouldn't. Any such solution won't generalize, it shouldn't have worked except by accident if the ARG-ST was not yet not-propagated. They have a 'main-verb-lex'... looks like you've lost the link between valence features and ARG-ST somewhere. Needs re-debugging.

ZhenZhen: Moving on to 4.5 Q5. EmphaticSHI...DE

Emily: In example 88 it seems hard to analyze as shi being a normal copula. So there must be something special going on here and you probably need an analysis for this one where shi is a marker.

ZhenZhen: Like a 'cleft shi' like in English. 88 We can handle with a special 'shi' and even get the semantics we want. 89 is a bigger problem. Ignoring this difficult one for now. 92 is a variation of transitive verb, similar to 89.

[discussion of 91,92]

Emily: I'm starting to get a feeling for this shi-de construction..

ZhenZhen: [reviews proposed solutions from handout, especially #2]

Emily: So then when you have an example that DE is not at the end,... #89... make sure to retain the gap. [Draws tree with ternery branching VP: V XP deP/NP de wants something that is gappy and comps not empty.

Francis: we built something similar but we found that RELS list was coming in but we didn't want a predicate. Question about how we're using XARG in cop-shi-pp-de-lex.

Emily: Ann might have an opinion, but seems legit. Because shi is taking two complements and has to act as a surrogate at the same time.. yeah, weird. The other possibility I can see is, have shi attached low to the thing it's emphasizing. Instead of the three branching thing, you have a non-branching that takes a de/P and makes something that can be picked up.

Francis: That sounds like the adverb analysis of shi that's been explored in literature. We disprefer it for certain reasons.

Francis: teaser for next week, there are many things in Chinese where it looks like we want to combine two things together as a lexical rule. Current machinery doesn't allow this. Generally, a lexical rule can only have one daughter. Chart mapping?