Ginzburg & Cooper (2004): Clarification, Ellipsis, and Contextual Updates #
@cite{ginzburg-cooper-2004}
Formalization of the core running example from @cite{ginzburg-cooper-2004}:
A: "Did Bo leave?" B: "Bo?"
This study applies the KOS framework (DGB, IS, C-PARAMS, coercion operations) to derive both CE readings — clausal and constituent — from the same antecedent sign, and demonstrates the speaker/addressee IS asymmetry.
Key Claims Formalized #
- Proper names introduce C-PARAMS (referent binding) — ex. 28
- The running example has 5 C-PARAMS — ex. 32
- Parameter focussing yields clausal CE reading — ex. 53/54
- Parameter identification yields constituent CE reading — ex. 59/60
- Both coercions take the antecedent sign and target the same SAL-UTT
- Existential generalization removes a parameter without clarification — ex. 77/78
- Speaker resolves all params; addressee may not — ex. 82
- Partial assignment triggers PENDING, not grounding
- Updates require structured representations (Hybrid Content Hypothesis) — ex. 2/16
C-PARAM for "Bo": binds variable b to the referent named "Bo". @cite{ginzburg-cooper-2004} ex. 28.
Equations
- Phenomena.Ellipsis.Studies.GinzburgCooper2004.cpBo = { index := "b", restriction := "named(Bo)(b)" }
Instances For
C-PARAM for temporal precedence.
Equations
- Phenomena.Ellipsis.Studies.GinzburgCooper2004.cpTime = { index := "t", restriction := "precedes(t,k)" }
Instances For
C-PARAM for speaker.
Equations
- Phenomena.Ellipsis.Studies.GinzburgCooper2004.cpSpkr = { index := "i", restriction := "spkr(i)" }
Instances For
C-PARAM for addressee.
Equations
- Phenomena.Ellipsis.Studies.GinzburgCooper2004.cpAddr = { index := "j", restriction := "addr(j)" }
Instances For
C-PARAM for utterance time.
Equations
- Phenomena.Ellipsis.Studies.GinzburgCooper2004.cpUttTime = { index := "k", restriction := "utt-time(k)" }
Instances For
The full C-PARAMS set for "Did Bo leave?" — 5 parameters. @cite{ginzburg-cooper-2004} ex. 32.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- Phenomena.Ellipsis.Studies.GinzburgCooper2004.suDid = { phon := "Did", cat := "AUX", cont := "ask" }
Instances For
Equations
- Phenomena.Ellipsis.Studies.GinzburgCooper2004.suBo = { phon := "Bo", cat := "NP", cont := "b" }
Instances For
Equations
- Phenomena.Ellipsis.Studies.GinzburgCooper2004.suLeave = { phon := "leave", cat := "V", cont := "leave-rel" }
Instances For
Equations
- Phenomena.Ellipsis.Studies.GinzburgCooper2004.suDidBoLeave = { phon := "Did Bo leave", cat := "S", cont := "ask(i,j,?.leave-rel(b,t))" }
Instances For
Full utterance skeleton for "Did Bo leave?" with all 5 C-PARAMS. @cite{ginzburg-cooper-2004} ex. 32.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Speaker (A) resolves all parameters: she knows who Bo is, who she is, who the addressee is, and the temporal parameters. @cite{ginzburg-cooper-2004} ex. 82b.
Equations
Instances For
Addressee (B) resolves all parameters EXCEPT b (Bo's referent). B doesn't know who "Bo" refers to. @cite{ginzburg-cooper-2004} ex. 82c.
Equations
Instances For
A's IS after uttering "Did Bo leave?": fully grounded. Speaker resolves all C-PARAMS, so the utterance goes straight to FACTS.
Equations
- One or more equations did not get rendered due to their size.
Instances For
B's IS after hearing "Did Bo leave?": partial assignment → pending. Addressee cannot resolve b, so the utterance goes to PENDING.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Parameter focussing on "Bo" (parameter b): clausal CE reading. @cite{ginzburg-cooper-2004} ex. 53–54. Output: SAL-UTT = "Bo" constituent, MAX-QUD = ?b.ask(i,j,?.leave-rel(b,t)) Paraphrase: "Are you asking if b left?"
Equations
Instances For
Parameter identification on "Bo" (parameter b): constituent CE reading. @cite{ginzburg-cooper-2004} ex. 59–60. Output: SAL-UTT = "Bo" constituent, MAX-QUD = ?c.spkr-meaning-rel(addr,Bo,c) Paraphrase: "Who do you mean by Bo?"
Equations
Instances For
Existential generalization on "Bo" (parameter b). @cite{ginzburg-cooper-2004} ex. 77–78. Removes b from C-PARAMS, weakens content to ∃b.ask(i,j,?.leave-rel(b,t)).
Equations
Instances For
B applies parameter focussing to set up clarification context.
Equations
- One or more equations did not get rendered due to their size.
Instances For
B applies parameter identification to set up clarification context.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The running example has exactly 5 C-PARAMS.
The running example has 4 constituents (Did, Bo, leave, Did Bo leave).
Speaker resolves all C-PARAMS.
Addressee does NOT resolve all C-PARAMS (missing b).
The unresolved parameter for B is exactly {b}.
Addressee's utterance is NOT grounded (no new facts).
Addressee's utterance goes to PENDING.
Parameter focussing succeeds on "Bo".
Parameter identification succeeds on "Bo".
Both coercions target the same SAL-UTT (the "Bo" constituent).
The SAL-UTT is "Bo".
Focussing and identification produce different operation types.
Existential generalization removes exactly one parameter.
The removed parameter is b.
Existential generalization wraps content with ∃.
Focussing MAX-QUD is a question about the antecedent content.
Identification MAX-QUD is a speaker-meaning question.
The KOS theory's coercion operations correspond to the empirical CE readings: parameterFocussing ↔ clausal, parameterIdentification ↔ constituent. @cite{ginzburg-cooper-2004} §5.
Equations
- One or more equations did not get rendered due to their size.
- Phenomena.Ellipsis.Studies.GinzburgCooper2004.coercionToReading Theories.Pragmatics.Dialogue.KOS.CoercionOp.paramFocussing = Phenomena.Ellipsis.ClarificationEllipsis.CEReading.clausal
- Phenomena.Ellipsis.Studies.GinzburgCooper2004.coercionToReading Theories.Pragmatics.Dialogue.KOS.CoercionOp.paramIdentification = Phenomena.Ellipsis.ClarificationEllipsis.CEReading.constituent
Instances For
The CE data's two readings map to distinct coercion operations.
Hybrid Content Hypothesis (@cite{ginzburg-cooper-2004} ex. 2/16): The content updated in dynamic semantics consists of structure expressing detailed relationships between the content and formal properties (syntax, phonology etc) of the various parts of an utterance.
Evidence: The same propositional content ("Bo left") yields different clarification potentials depending on phonological/syntactic structure. The utterance skeleton encodes this structure via CONSTITS and C-PARAMS.
The CE proper name example matches our running example.