Skip to content

TheAbbey_Chrysalis2014Nominalization

EmilyBender edited this page Feb 18, 2014 · 1 revision

[Picking up a discussion from Saarland Summit: SaarlandSententialArgument]

Woodley: It surprised me that the balloon popped/I was surprised that the balloon popped

Woodley: It surprises me that that the balloon arose surprised me is not grammatical (according to the ERG).

Dan: That surprises me, too. No, I've waffled so many times on this analysis… but it still surprises me

oe: That the chair arrived surprises me. parses fine.

[Surprising variation in ACE/PET over what we believe is the same ERG.]

Woodley: That Mary went away persuaded John to leave the store. ?

Dan: [recuses himself]

Emily: Sounds impeccable to me.

oe: What's your dog in this race?

Dan: I want there to be an asymmetry between subject and object. NP-obj VP-comp S-comp would be required for It persuaded John to leave the store that Mary entered the race. That's a new frame, otherwise unknown.

Woodley: That's derived from…

Emily: Dan does intraposition, not extraposition.

Dan: Don't want to derive a grammatical sentence from an ungrammatical one.

Woodley: Do we have one that we agree is ungrammatical?

Dan: What about the one that everyone disliked in St. Wendel?

Woodley: Sounds okay to me.

Emily: Are you willing to switch to extraposition, not interposition?

Dan: If you can convince me that that asymmetry is true.

Emily: There's also the semantic question, of the asymmetry of treatment in the semantics of subjects v. objects.

oe: Actually ARG1 v. ARG-other. The Kim left was was believed by everyone/Everyone believed that Kim left

Dan; Those are asymmetrical.

Emily/oe: No, symmetrical. That's a sentential complement, and so it doesn't get reified/nominalized regardless of passivization.

oe: Makes these paraphrases, which is good.

Dan: Yes, had to do a special passive rule.

oe: My question at the end of that discussion was couldn't we dispose of the nominalization of ARG1s and be left with more symmetry, and with more of the scopal v non-scopal shebang. That seems at most an issue of degree, not … because we have that shebang already.

[ Reviewing notes from St. Wendel ]

Emily: Can you remind us of the relevance of PPs and hole arguments?

Dan: That depends on what Kim says.

Emily: Is that on really meaningful?

Dan: look into … you might ask the same question.

oe: Might even be a test?

Dan: From two examples?

oe: It's a phenomenal day!

Dan: From what he said, I expect we can think of a better example. From where he sits, …

oe: Not free relatives?

Dan: The test is that why doesn't do it. Are you going to raise the opaque P flag for I'm going to concern myself with…

Emily: Yes

Dan: Circling around

Emily: It might be contentful, but can you circle any other way

Dan: Why we were dancing around why Woodley insisted on having hot chocolate…

Emily: Would it be uncooperative to say that dancing around is

Dan: I'm going to change the question to why Woodley orders hot chocolate.

oe: Our attention is on why Woodley…

Dan: But also His reliance is on fish.

Emily: The PP with put

Dan: Let's put energy into why something happens./Let's put time into why that happens.

Woodley: Star on put time into why this happens.

Emily: okay with put time into research

Woodley: Yes.

Emily: The question is why we have to allow semantically contentful prepositions to take wh clauses.

Dan: Frequent in DeepBank

Emily: Semantically contentful?

Dan: Haven't worried about it because don't have a test for semantically contentful.

Woodley: You nominalize these.

Dan: Yes, but only for wh clause complements. But not that complements.

Woodley: What is nominalized? The verb.

Dan: Always a label (LTOP) because it might be negated. Let's focus on why we didn't study that question more.

Woodley: Not interested in the event of the verb, but more in the why.

Dan: You think the why is the head of the question?

Emily: why will come out as for what reason. In why didn't we look into that more, we have for what reason as modifying look. Is your intuition that that's right.

Dan: There is a potential for an ambiguity: I didn't sing on Sunday.

Emily: I've been here --- with Alex. Negation is a focus sensitive operator. But can we tell that same story for for what reason. (See paper in Sag festschrift volume)

Woodley: I asked what we didn't eat yesterday. Going back to object languages.

Dan: In free relatives, the wh is the head of that phrase. I'll have what she's having. I sat where he sat --- locative adverbial not a clause. The claim is that for non-free relatives is that the argument is a clause. I wonder where she that. wonder is a relation between an individual and a proposition.

Woodley: What is it that makes that piece formula an interrogative.

Dan: It's a state of affairs whose polarity is not yet fixed (whether),

Woodley: what/how … also the polarity?

Dan: There's an unfilled argument position/variable under discussion.

Woodley: That interrogative logical formula has a special position in it.

Emily: And that's a separate question from scope.

Woodley: I see that there is the same question at least to some extent for the main clause.

Dan: In the case of the wh guys, the asymmetry is more uniformly distributed in the grammar. When it's a complement, the wh clause does not get nominalized, anywhere else it gets nominalized. Who Kim listens to on the radio always amazes.

Woodley: I wondered whether Kim would come./I wondered about whether Kim would come. quite different in the semantics.

oe: As long as the about is semantically contentful.

oe: I feel we're talking about different types of clausal arguments:

  • to VPs

  • that CPs

  • wh CPs

  • wh to Cps: who to talk to (I wonder who to talk to)

Dan: finite/non-finite and wh-/non-wh

Emily: and can show up as complements or subjects.

oe: What about non-finite non-infitival clauses?

Emily: for-to

Dan: no, -ing depictives.

Woodley: I saw John sing. Non-finite?

Dan: No clause there, because Who did you see sing?

Woodley: I demand that the dog be kept outside

Dan: Subjunctive, still finite.

Emily: So we're using clause to mean subject saturated?

Dan: Yes.

Emily: But in the semantics it may not matter so much.

Emily: What we're missing are non-finite subject saturated wh guys. I wonder why Kim to arrive… it's bad.

Dan: What's the shape of non-finite express subjects in English? for NP

Emily: I wonder why for Kim to arrive…

Dan: I wonder who for Kim to hire…

Dan/Emily: Maybe an idiosyncrasy of English wh syntax?

Woodley: What about: for who to arrive

Emily: Can you make a sentence out of it?

Woodley: Maybe… For who to arrive is it important that the clock strike three?

Emily: star.

Dan: In situ: We're waiting for WHO to arrive?

Woodley: Who are we waiting for Kim to meet?

Emily: Not that one, because the main clause there is finite.

Dan: Enough with this corner of the jungle. Do we have to find the white-striped no tailed tiger?

Emily: If we do, we should surely put the Tiara of Doom on it.

oe/Dan: We do have a good menagerie of creatures in their cages. And now we can see what happens when we put them in various argument positions.

Emily: Why just argument positions?

oe: I don't think any of these can be non-arguments.

Emily: Right, they can be arguments of things that head modifiers, but if they do they would still be arguments.

Woodley: Isn't the problem with trying to create an embedded clause out of these is that they are shaped like modifiers?

Emily: I wonder for who to arrive would be a disaster. But weren't we leaving the jungle?

Emily: So we have the creatures, and now we can ask which argument positions they appear in and what the distribution of nominalization v. non-nominalization over those. And what we're really interested in is semantic argument positions, because the asymmetry that Dan has implemented so far is ARG1 v. everything else.

Dan: Yes, I even go so far as to implement that in the special passive rule.

oe: Taking the paraphrase constraint as a higher goal than structural symmetry.

oe: I was going to ask to be reminded of the need to nominalize. The underlying generalization is in subject position (ignoring passives).

Dan: The VP/CP subject of an active (wh or non-wh) always gets nominalized.

Woodley: interposition example

Emily/oe: That's like passive. Need to refine our description.

Woodley: It is the ARG1.

Emily: The vanilla ones, the house cats, when they show up in subject position, you want them nominalized why?

Dan: What Kim said angered Sandy.

oe: You consider that a house cat?

Dan: It has claws, and likes to go outside, but yes.

Dan: What Kim said chased Bill out of the room. Would you agree that It chased Bill out of the room what Kim said. is strange?

Woodley: No.

Francis: Star.

Woodley: That's because she said it with disdain.

Francis: Okay, can I have it with love. … Invite Tom Wasow.

Dan: What Kim said put me into a lousy frame of mind.

Woodley: It put me into a lousy frame of mind, what Kim said.

Francis: star.

Woodley: It intrigued me whether it would rain.

Francis: Is there anything in the wild that has inspired this?

Emily: By why nominalize? What Kim said chased Bill out of the room.

Dan: I need to put that handle somewhere.

Emily: qeq to the ARG1 of chase.

Dan: But I was pretty sure that chase wanted as its ARG1 an index.

oe: Why?

Dan: Didn't seem like a big leap. We have discussed underspecification. Hiring Kim was easy. Probably know that unexpressed ARG1 of hire.

Emily: Could the type unexpressed not constrain the INDEX of the dropped argument to i.

oe: Do we know that it wasn't that we received a tax reimbursement that hired Kim.? That we received a new grant hired two students.

Woodley: Syntax is there, semantics not yet.

Dan: The fact that we received a new grant hired Kim. Equally bad?

Woodley: Yes.

Emily: Lexical difference between hire and chase? What we promised to work on hired Kim … still doesn't sound good.

oe: That the contract finally closed hired three new consultants.

Emily: Star.

Dan: Can't even pretend to be dispassionate on this one. Ask Woodley.

Woodley: Having the CEO sign the new contract hired three new people.

Emily: Getting closer.

Francis: Having the CEO sign the contract validated the deal.

Dan: How about another verb? What he said persuaded me to change my mind. Is that English?

Emily/Francis/Woodley: Yes.

oe: For Amazon to enter the Chinese market will hire a whole new town of employees.

Emily/Francis: Star.

oe: Can I have a last crack at hire?

Dan: You'll be the last one at that pond.

oe: What we earned 2013 will hire two new programmers this year.

Emily: Ding ding ding.

Francis: Unstar.

Dan: To rain on your parade, free relatives are catch and release.

Woodley: How we doubled our profit last year will earn us two new patents.

Dan: How also has a free relative.

Woodley: Why we doubled our profit last year will earn us three new inquiries from the Justice Dept.

Dan: Seems okay.

Emily: Back to why do you want to nominalize.

Dan: What does the verb say in the lex entry about its SUBJ.CONT.HOOK.

Emily: "grab me"

Dan: "grab me" and do what? Then put in a qeq.

oe: The argument when it's preparing to be grabbed could do it.

Dan: Where?

Emily: The nominalization rule.

Dan: Weren't you trying to get rid of that rule?

Emily/oe: No, just its effect on the semantics.

Emily: The alternative is much more lexical ambiguity than you want.

Woodley: Every verb would have to…

Emily: Not every verb. Hire doesn't.

Dan: If you're not picked up unexpressed can do it, but then I have to be careful to discharge every unexpressed argument.

Emily: Or olist can do it?

Dan: Yes, maybe olist can do it?

Woodley: Are battleships sunk if ps are left resolved at the end of the day.

oe: No one is sensitive to p. They bear no relationship to hs.

Dan: Emerging candidate proposal is that we have no motivation for the nominalization in its own right, and that we can get something more symmetric overall if we just say that ARG1s of verbs could as well be qeq unifications rather than having an additional EP there to mediate between the ref-ind and the handle and we would seek to engineer the rest of the machinery so that in case you didn't fill those it was properly taken care of. Would probably be able to do it with the existing machinery. Not doing lexical ambiguity but rather building the VP first and the duplicating that.

oe: And not doing it for some verbs?

Dan: I believe I'm doing it for some verbs.

Woodley: Why qeq?

Dan: There's going to be a negation and a quantifier: That every dog didn't chase some cat resulted in a disaster. That constituent must be allowed an interpretation with negation in widest scope or one or more quantifiers go higher.

Woodley: Above that embedded verb?

Dan: The assertion is that I can give a reading where it's only inside that argument position.

oe: If that verb were an operator like doubt.

Woodley: In situations where it's always expected to be a scopal argument, it's clear you can come up with those examples, but in situations where it's coerced from subject position, less obvious.

Dan: The burden of proof is on one who wants to say that the variation doesn't happen.

Clone this wiki locally