Skip to content

CambridgeWhEvents

MichaelGoodman edited this page Jul 16, 2019 · 1 revision

Linking wh-words to a specific event in the MRS

Participants

  • Moderator: OlgaZamaraeva (OZ)

  • Scribe: MichaelGoodman (MWG) (the scribe apologizes for not always keeping up with who said what when.)

  • Other Participants (in order of appearance): Dan Flickinger (DPF), Guy Emerson (GE), Woodley Packard (WP), Ann Copestake (AAC), David Inman (DI), Francis Bond (FCB)

Minutes

  • OZ: Problem is linking a wh word to a specific event when there's more than one in a sentence
    • Do you know who Kim saw?
  • OZ: Goal is to figure out how real the problem is, how it relates to the Matrix
  • OZ: In Oslo we discussed these, but mostly about linguistic context; i want to know technical difficulties
  • DPF: wh gives ARG0, in typical case it is deterministic
  • DPF: but we may have more than one event where it's applicable: Who tried to chase Kim
  • Examples from slides
    • (1b) Sandy said who saw *who*?
    • (2b) Sandy said who *who* saw?
  • DPF: 2b is a silly analysis, with the second *who* being a relative clause
  • GE: With echo questions... from earlier discussion:
  • GE: It doesn't need to be a question: "Sandy said who saw who."
  • WP: What's an echo question?
  • DPF: When A says: "John saw a glarump..." and B says: "John saw what?"
  • DPF: It's interesting because the ERG doesn't do well with polarity questions: "I wonder whether ..."
  • DPF: And these interact with echo questions
  • WP: In the MRS for "John wondered whether Kim left." the only difference between "John was amazed that Kim left" (besides wonder/amaze) is that _leave_v_1 has sf: ques
  • GE: In Oslo it was suggested to use ICONS to link the wh word to the clauses
  • DPF: Can we create an information-structure backstory for that? Currently ICONS is used for information-structure in the ERG.
  • OZ: There is literature on focus with wh-questions. Many believe they are always focused (though there is disagreement)
  • AAC: Back to the echo questions... The "who" of the echo question is not the same as the who in the actual question, but it's a discourse thing; asking the speaker to clarify what they said.
  • AAC: So we need to decide if the MRS should be underspecified here or if we can capture them somehow.
  • DI: "John saw Mary and *mumble*"
    • "John saw Mary and who?" Felicitous response is "John saw Mary and who?" not "John saw who?"
  • DPF: And not "John saw who and Mary?"
  • AAC/DPF: "John saw Mary and whom?" -- do constraints on case persist in echo questions? general agreement that this sounds pedantic
  • GE: Echo in yes/no questions: "Is Kim here?" "Is who here?"
  • AAC: Coming back to an underspecified representation... If it is ambiguous between a polarity question and a wh-question, can this be underspecified?
  • DPF: "Is who here?" is both a wh-question and also the original yes-know question
  • WP: If you have a sentence with many wh-words and one becomes the focus of an echo question, the rest are invalidated... You're really asking to clarify that one, the rest are background.
  • AAC: Yes, let's set aside echo questions, they're a distraction.
  • OZ: "Who do you know whether Kim saw"... In Russian it's better with negation: "Who do you not know whether Kim saw"
  • .... (couldn't keep up here)
  • OZ: If I want to model this in the Matrix, do I have a problem here?
  • OZ: "Do you know who Kim saw" can be polar: literal answer "yes", pragmatic answer may be "Sandy"
  • WP: Can't these have truth conditional effects? If the answer to the polar question depends on the answer to the wh, then we have a problem because ICONS should not change truth-conditions.
  • GE: "I know whether you know who Kim saw."
  • GE: vs "I know who you know whether Kim saw."
  • ... again, the scribe can't keep up with these examples apologies abound.
  • MWG: wait I thought ICONS *could* effect truth conditions
  • DPF: In the current uses, passive, topic, I don't think it does
  • AAC: But if we use them for pronouns it could
  • AAC: But in general they should not affect truth values... Although they could be used to disambiguate truth conditions
  • DPF: Before we had ICONS between an event and an individual that did not affect truth, and between two individuals where it does, but this presents a third case between an event and an individual that does affect truth
  • DI: Use focus to show what is the expected answer... "Who did Kim see?" "Do you know who Kim saw?"
  • DPF: ..
  • WP: What if both wh are individuals?
  • DI: Who saw what?
  • AAC: If we want lists of pairs for responses, then for "Do you know who Kim saw?" then "Yes, Sandy" is a valid answer.
  • DI: But not "No, Sandy"
  • GE: "What do you know who Kim gave?"
  • FCB: In Japanese it's fine, just answer all whs.
Clone this wiki locally