Source
Heim (1982), LIT-838, ch. I §1.3 and ch. III, read in NOTE-608; Veltman (1996), LIT-818, §1–2, read in NOTE-617; Krifka (2024), LIT-835, §2 and §6, read in NOTE-602.
What was actually shown
- Anaphoric potential (Heim). “I dropped ten marbles and found all of them, except for one” and “I dropped ten marbles and found only nine of them” are true in the same circumstances, said in the same situation. Only the first can be followed by “It is probably under the sofa”. “John has a spouse” and “John is married” pattern the same way with “She is nice”. A theory that gives sentences only truth conditions (with pragmatic salience on top) must make wording an independent factor. Heim’s file change semantics explains the contrast because the indefinite opens a card. It also shows that two common grounds with the same set of worlds can develop differently under the same utterance. These are judgement data on constructed examples, and they could have come out the other way.
- Tests (Veltman). Updates with epistemic “might” (and “presumably”) pass or fail against the state they meet. They are not persistent, and “Maybe it’s John… It’s Mary… Maybe it’s John” is incoherent where the same sentences in another order are not. Proposition 1.2 states when a dynamic system reduces to a static one: when it is additive. Systems with tests are not additive.
- Performative updates (Krifka). “Citizens can apply for travel abroad” is one proposition. As a declaration it changes every index of the context set, as an assertion it adds a guarantee and proposes that indices be eliminated. Hedges and question–answer pairs show the two readings apart. This is argued from grammatical evidence, not tested.
The common origin is Stalnaker’s account of assertion as narrowing a context set (LIT-811, unread here; known through Heim’s quotations and Krifka’s use). Each of the three extends that account where it falls short.
What this does not say
- Not that truth conditions are dispensable. Each framework recovers them: Heim’s truth of a file, Veltman’s additive fragment, Krifka’s informative update. The claim is that they underdetermine meaning, not that they are wrong.
- Not that the three formalisms are one theory. Heim’s files, Veltman’s information states and Krifka’s index-changing context sets are different objects. Veltman’s updates only ever shrink a state; Krifka’s can add possibilities. Nothing here shows that one framework captures all three effects.
- Not that static rivals have been refuted. E-type and situation accounts of anaphora, and pragmatic accounts of “might”, were developed after or alongside these works. None was read for this document, and the promotion condition asks for exactly that comparison.
- Not a claim about translation. None of the sources compares utterances in different languages. Applying this to whether a rendering preserves an utterance’s context change (CLAIM-061) is a further step these sources do not take.
- Not a measurement. No source gives an observable or an experiment that separates the effects. The evidence is intuitive judgement and formal construction.
Note of 2026-10-28: what holds it back
The status review of this date keeps this reading Proposed (ADR-057). The
three readings are full reads (NOTE-608, NOTE-617, NOTE-602), but
promote_when asks for a developed static rival, read and checked against
the marble and spouse, “might”-order and declaration and assertion
contrasts. None is held. The first candidate is Mandelkern, Bounded
Meaning (2024), its chapters on anaphora and modality, or else Elbourne
(2005). Neither has an open copy, so this waits on a copy from the owner.
CLAIM-061 now cites this reading as corroboration, not as a ground, since applying it to translation “is a further step these sources do not take”.
Note of 2026-10-30: what holds it back
promote_when still asks for a developed static rival, read and checked
against the marble and spouse, “might”-order, and declaration and assertion
contrasts, and none is held. Mandelkern’s Bounded Meaning (2024) and
Elbourne (2005) are unfiled and have no open copy, so this waits on a copy
from the owner. CLAIM-061’s revision of 2026-10-30 removed two unread works
(LIT-785, LIT-780) from its grounds and kept this reading under
corroborated_by; it does not bear on the rival.