learn.ifs.name

Conversation and its Structure

Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 26

20 cards in “26.1 Properties of human conversation” · whole deck

26.1 Properties of human conversation

  1. What is a turn in dialogue?
    A single contribution from one speaker, as in a game: I take a turn, then you. A turn can be one word (“OK.”), a sentence, or several sentences.
  2. Why does turn structure matter for spoken dialogue systems?
    A system must know when to stop talking (the user may interrupt, perhaps with a correction) and when to start: people begin turns almost immediately after the other finishes, because they predict when the other is about to finish.
  3. What is endpointing, and why is it hard?
    Detecting whether the user is done speaking so the system can respond; it is hard because of noise and because people often pause mid-turn.
  4. What is the key insight behind speech acts?
    Each utterance in a dialogue is an action performed by the speaker (Wittgenstein, worked out by Austin 1962).
  5. What are the four major classes of speech acts (Bach and Harnish 1979)?
    classwhat it does
    constativescommit the speaker to something being the case (answering, claiming, denying)
    directivesget the addressee to do something (asking, ordering, inviting)
    commissivescommit the speaker to a future action (promising, planning, betting)
    acknowledg­mentsexpress attitude toward the hearer about a social action (apologizing, greeting, thanking)
  6. Classify the speech acts: “Turn up the music”; “What day in May did you want to travel?”; “I need to travel in May”; “Thanks!”
    Directive; directive (a question politely commands an answer); constative (stating a constraint); acknowledgment.
  7. What are common ground and grounding?
    Common ground: what the participants both agree on (Stalnaker). Grounding: acknowledging that the hearer has understood the speaker (Clark), or signaling problems in understanding so mutual understanding can be reached.
  8. What non-linguistic example illustrates grounding?
    An elevator button lights up when pressed, acknowledging that the elevator has been called (Norman 1988).
  9. What are the two phases of a contribution (Clark and Schaefer 1989)?
    Presentation: the speaker presents an utterance, performing a speech act. Acceptance: the hearer grounds it, showing whether understanding was achieved.
  10. What methods can a hearer use to ground an utterance, from weakest to strongest?
    strongercontinued attentionkeeps attendingnext contribution“And what city…?”acknowledgmentuh-huh, OK, that’s greatdemonstrationparaphrase, completiondisplayverbatim repetition
  11. Find three kinds of grounding in a travel agent dialogue.
    Explicit acknowledgment (“OK”, “Right”); demonstration by repeating (“what day in May…”); next contribution starting with “And”, implying the answer was understood and this question is in addition.
  12. What is a continuer (backchannel)?
    A short optional utterance like uh-huh that acknowledges the other’s utterance and doesn’t itself require acknowledgment.
  13. What are adjacency pairs?
    Paired dialogue acts from conversation analysis, a first pair part setting up an expectation for a second: QUESTION → ANSWER, PROPOSAL → ACCEPTANCE/REJECTION, COMPLIMENT (“Nice jacket!”) → DOWNPLAYER (“Oh, this old thing?”).
  14. What is a side sequence (subdialogue)? Give an example.
    Material separating the two parts of an adjacency pair. A correction subdialogue: while the agent seeks a May 15 return flight, the client asks “what day of the week is the 15th?” and then decides to stay until Sunday, changing the plan.
  15. What is a clarification subdialogue?
    A side sequence between a REQUEST and its RESPONSE, common after ASR errors: “What do you have going to UNKNOWN on the 5th?” — “Going where on the 5th?” — “Hong Kong.” — “OK, here are some flights…”
  16. What is a presequence?
    A preliminary exchange before a request, e.g. “Can you make train reservations?” — “Yes I can.” — “Great, I’d like to reserve a seat on the 4pm train.”
  17. What is conversational initiative? Contrast system-, user- and mixed initiative.
    Control of the conversation. An interviewer has initiative over the interviewee. Mixed initiative, where control shifts back and forth, is the human norm. Primitive systems use system initiative (the user must answer the system’s question) or user initiative (a search engine passively answers queries). LLM systems come closer to mixed initiative but still don’t switch naturally.
  18. What does “I need to be there for a meeting from the 12th to the 15th” in answer to “What day in May did you want to travel?” illustrate?
    Conversational implicature: the client doesn’t literally answer, but the agent infers travel dates (likely arriving on the 11th).
  19. What is conversational implicature?
    Grice’s term for a class of licensed inferences: the speaker communicates more than the words say, and the hearer infers it using maxims, general heuristics that guide interpretation.
  20. What is Grice’s maxim of relevance, and how does it license inference?
    Speakers try to be relevant, not utter random speech acts. Hearing about a meeting, the agent asks why it was mentioned: the meeting is probably the reason for the trip, so the flight should arrive before it.

Source