Documentation

Linglib.Phenomena.Modality.Studies.Huijsmans2025

Huijsmans (2025) @cite{huijsmans-2025} #

Timing of evidence and epistemic modal claims. Natural Language Semantics 33: 207–254.

The future morphemes səm (ʔayʔaǰuθəm; Central Salish) and will (English) pattern with strong epistemic modals like ćε (ʔayʔaǰuθəm inferential) and must, but their distributions diverge in inferences about non-future eventualities. Huijsmans argues the difference is temporal, not modal strength: səm/will presuppose the modal base is established prior to the earliest moment the prejacent is claimed to hold, while ćε requires at least one modal-base proposition to hold at or following that moment.

Structural claim #

The temporal-presupposition partition is structurally identical to the EP-condition partition in Theories/Semantics/Tense/Evidential.lean, but indexed on MBT (modal-base time) rather than EAT (evidence acquisition time). EPCondition is reused under the slot reinterpretation MBT ↦ acquisitionTime, PrejT ↦ eventTime. The bridge is MBTProfile.licenses_iff_EP.

Empirical wedge #

Most of Huijsmans's stimuli (cooking, Daniel-home, poison) satisfy MBT < PrejT ∧ EAT < ET — both analyses agree. The Felipe-rain stimulus (ex. 46-49) and clam-digging stimulus (ex. 48-49) have MBT < PrejT but ET < EAT — Huijsmans's MBT analysis correctly licenses will/must; an EAT analysis (with the same partition reinterpreted) does not. MBT_vs_EAT_diverge_on_felipe is the load-bearing theorem.

At-issue content #

The at-issue content is Kratzer strong necessity over the modal base, shared across all four lexical entries (modulo modal-base type). It is not formalized in this file; consumers wanting full denotations should combine Modal.flavor with Theories.Semantics.Modality.Kratzer.necessity.

A stimulus context, supplying the four temporal anchors needed to state both Huijsmans's MBT predictions and the EAT predictions she is contrasted against. The MBT/EAT split lives in which pair of anchors a profile reads.

  • earliestMBT : Time

    ⌜MBT⌝ — the maximum over q ∈ MB of EARLIEST(SHIFT q), i.e. the latest "earliest moment" required by the modal base.

  • earliestPrejT : Time

    ⌜PrejT⌝ — EARLIEST(p) for the prejacent p.

  • eat : Time

    The (Cumming/Hirayama-Matthewson) evidence-acquisition time.

  • et : Time

    The event time of the prejacent.

Instances For
    def Phenomena.Modality.Studies.Huijsmans2025.instReprStimulus.repr {Time✝ : Type} [Repr Time✝] :
    Stimulus Time✝Std.Format
    Equations
    • One or more equations did not get rendered due to their size.
    Instances For

      The three MBT-presupposition profiles Huijsmans's four lexical items instantiate. A domain-flavored selector over the abstract Core.Time.Relation partition (cf. EPCondition in Tense/Evidential.lean, which selects from the same partition for the EAT/ET slot pair). Only three of the five abstract relations name a Huijsmans modal; the others are unused here.

      • strictPrior : MBTProfile

        MBT < PrejT. Modal base established strictly before the prejacent's earliest moment. səm, will. Eq. (35).

      • nonProspective : MBTProfile

        PrejT ≤ MBT. At least one MB proposition becomes true at or after the prejacent's earliest moment. ćε. Eq. (36).

      • unrestricted : MBTProfile

        No temporal restriction. must.

      Instances For
        @[implicit_reducible]
        Equations
        Equations
        • One or more equations did not get rendered due to their size.
        Instances For

          The MBT-licensing predicate, derived from the abstract partition.

          Equations
          Instances For
            @[implicit_reducible]
            instance Phenomena.Modality.Studies.Huijsmans2025.instDecidableLicensesOfDecidableEq {Time : Type} [LinearOrder Time] [DecidableEq Time] (c : MBTProfile) (s : Stimulus Time) :
            Decidable (c.licenses s)
            Equations

            A modal in the Huijsmans typology: a label, a (possibly underspecified) modal-flavor, and an MBT-presupposition profile. All four entries in this paper are necessity modals; force is omitted.

            Instances For
              Equations
              • One or more equations did not get rendered due to their size.
              Instances For

                ʔayʔaǰuθəm future clitic səm. Underspecified for modal base (epistemic and circumstantial uses; Huijsmans §2.3).

                Equations
                Instances For

                  English future auxiliary will. Underspecified for modal base (epistemic in non-future-eventuality uses, circumstantial in future-eventuality uses; Huijsmans §2.4).

                  Equations
                  Instances For

                    ʔayʔaǰuθəm inferential clitic ćε. Epistemic only.

                    Equations
                    • One or more equations did not get rendered due to their size.
                    Instances For

                      English necessity modal must. Underspecified for modal base; this paper only treats epistemic uses.

                      Equations
                      Instances For

                        (40) Cooking. The fish-cooking stimulus: MB props (fish in oven, typical cook time) hold prior to PrejT (it is cooked). Both MBT and EAT analyses predict səm/will/must ✓, ćε ✗.

                        Equations
                        Instances For

                          (46) Felipe-rain. The wedge case. MB props (it is raining, Felipe has no rain jacket) become true earlier than PrejT (Felipe is wet), but the speaker only acquires the evidence later (sees the rain at 6pm, after Felipe has already gotten wet at 5pm). So MBT < PrejT but ET < EAT. Huijsmans correctly licenses will/must; an EAT analysis incorrectly rejects will.

                          Equations
                          Instances For

                            (50) Smell-of-fish. The smell coincides with PrejT: at least one MB proposition (the fish smells cooked) holds at the moment the prejacent holds. ćε/must ✓, səm/will ✗.

                            Equations
                            Instances For

                              The EAT-indexed analogue of MBTProfile.licenses — applies the same partition shape to (eat, et) instead of (earliestMBT, earliestPrejT). Defined inline rather than imported from a (not-yet-existent) Hirayama-Matthewson study file.

                              Equations
                              Instances For
                                @[implicit_reducible]
                                instance Phenomena.Modality.Studies.Huijsmans2025.instDecidableEatLicensesOfDecidableEq {Time : Type} [LinearOrder Time] [DecidableEq Time] (c : MBTProfile) (s : Stimulus Time) :
                                Decidable (c.eatLicenses s)
                                Equations
                                • One or more equations did not get rendered due to their size.

                                The empirical wedge (Huijsmans (46)–(49)).

                                On the Felipe-rain stimulus the MBT and EAT analyses disagree on will: the MBT-analysis licenses it (matching the judgment), the EAT-analysis rejects it (mispredicting). This is the entire empirical content of Huijsmans's argument compressed to a single decide.