[KC15]: Trivalent truth-value judgments for embedded plurals

Empirical anchor for the homogeneity-projection literature. Križ & Chemla (2015), "Two methods to find truth-value gaps and their application to the projection problem of homogeneity," Natural Language Semantics 23(3): 205-248, introduce a binary-then-ternary truth-value judgment paradigm and deploy it on plural definites embedded under quantifiers (every, no, exactly N) plus sentential negation, across three experiment batches (A1-A3 binary, B1-B3 ternary, C2-C4 ternary with grammatical boys/presents stimuli).

Main declarations

• examples — per-paper accessor (option B): K&C's contribution to the cross-paper projection pool, i.e. KrizChemla2015.Examples.all.

The encoded stimulus rows themselves live in the generated module Data.Examples.KrizChemla2015 (from Linglib/Data/Examples/KrizChemla2015.json via scripts/gen_examples.py); the pool at [[Generalizations.HomogeneityProjection]] imports them, and is where cross-account testing of rival theories ([Mag14], [Kri16], [KS21b], [BL21]) lands.

Implementation notes

The K&C examples live in the generated Data.Examples.KrizChemla2015 module; the projection hub imports them into its cross-paper test pool. This file remains the paper-anchored entry point and narrative hub.

Three projector-synthesized fields are gone from the prior incarnation, verified against the paper PDF:

• Monotonicity enum + monotonicity function — re-stipulated NaturalLogic.ContextPolarity and contained a return-value / comment contradiction for notEvery.
• ProjectionDatum/EmbeddedTruthConditions structures with Bool predicate-shape fields and String informal-prose fields — violated CLAUDE.md's no-Bool-for-predicates and no-String-in-proof-relevant-code conventions; superseded by the typed LinguisticExample rows.
• Enum cases exactlyOne and atLeastOne — never tested in the paper.

A fourth correction was the no projection finding: the prior encoding's gapDetectable := false reflected the A2/B2 null, but [KC15] §5.2.3 (Exp. C2) explicitly overturns it ("E-no does show a gap, which, albeit small, is robust" — β=1.3 χ²=8.2 p=.004 on Diag. 1, Table 9). The C2 finding lands as Examples.no_C2_gap with gap_detected = "true".

Todo

• The presupposition account named by [KC15] (Schwarzschild 1994 + [Gaj05]) has no linglib study file yet; adding one would close the three-rival theoretical landscape the paper discusses.

def KrizChemla2015.examples : List Data.Examples.LinguisticExample

Per-paper accessor: K&C 2015's stimulus rows (the generated Examples.all), which the projection hub pools for cross-account testing.

Equations

• KrizChemla2015.examples = KrizChemla2015.Examples.all