Documentation

Studies.Mizuno2024

Mizuno (2024): Strategies for Anderson Conditionals #

This file formalizes the account in [Miz24] of Anderson conditionals, conditionals whose consequent is an observed fact and which therefore argue for their antecedent ([And51]). English must X-mark them, since over the live domain the consequent is trivially true and O-marking says nothing, while Japanese and Mandarin must O-mark them: their X-marking, the fake past -ta and the perfective le, forces a counterfactual reading. The typological record and its attested minimal pairs are transferred from the paper's examples (andersonStrategy, anderson_judgments_match_strategy), and the correlation with future-less-vivid conditionals holds row by row (flv_anderson_correlation). Both strategies enlarge the modal domain to one containing a consequent-failing world, after the expansion analysis of [Mac15] and [Mac19]: X-marking by enlarging the base (xMarking_expands), Japanese O-marking by a historical-present shift of the evaluation index backward under branching time ([Sch04b]; hp_expands_jones_domain), and on either route the consequent is left open and a true Anderson conditional excludes a live world (expanded_anderson_informative).

Implementation notes #

The Jones scenario is a strict conditional over a historical base whose worlds record whether the symptoms are shown; the marking typology and exponent inventory are those of Semantics/Modality/Exclusion, and the Mandarin exponent le is study-local, resting on the paper's single consultant.

References #

The per-language strategy record (§2–§4.2) #

The felicitous Anderson marking of each language is X for English (§2, ex. 1a), O for Japanese (§3, ex. 4a -ru) and O for Mandarin (§4.2, ex. 13a without le), and none outside the sample.

Equations
Instances For

    FLV X-marking availability (§4.2): English yes (ex. 8); Japanese -ta (ex. 10) and Mandarin le (ex. 12) induce strong counterfactuality instead.

    Equations
    Instances For

      English and Japanese pick opposite Anderson strategies (§3).

      The §3.1 puzzle: available X-marking that must not be deployed #

      Japanese has X-marking (Fake Past -ta, ex. 3; Japanese.Conditionals.xMarking), and §2 predicts it in Anderson conditionals — "strikingly, this is not borne out". Uniformity (§4.1) survives as a meaning/use split: uniform semantics, blocked deployment.

      X-marking exponents, routed from the Fragment inventory. Mandarin le is study-local: fn 15's single consultant is below the Fragment consensus bar.

      Equations
      Instances For

        The inventory-driven prediction X-marks wherever the inventory provides X-marking (§2; a fortiori under the blocking of fn 14, after [Chi98]).

        Equations
        Instances For

          English validates the inventory prediction.

          The §3.1 puzzle: Japanese has the X resource yet must not deploy it.

          Mandarin patterns with Japanese (§4.2).

          The attested minimal pairs #

          Pairs live inside one numbered example: the felicitous primaryText (strategy m) plus the infelicitous alternatives entry (realizing m.other).

          Felicitous when fully acceptable.

          Equations
          Instances For

            An Anderson row matches the record when its primary judgment matches strategy m and each alternative matches the other strategy, O-marking being the absence of X-marking.

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

              Every recorded Anderson judgment matches the strategy record.

              The Future-Less-Vivid correlation (§4.2) #

              def Mizuno2024.flvAvailableTag (e : Datum) :
              Option Bool

              The flv_xmarking tag.

              Equations
              Instances For

                §4.2: Anderson X-marking and FLV X-marking stand or fall together, for every sampled language. An empirical correlation of two independent records, not a definition; cf. Iatridou2000.CounterfactualType.flv (one ExclF layer).

                The triviality puzzle and its resolution by domain expansion #

                §2: an Anderson consequent is an observed fact, true throughout the live domain D, so over D the conditional is trivial; both strategies expand D to a D⁺ containing a consequent-failing world. The Jones scenario as strictImp over a historical base: worlds are Bool (true = symptoms shown), the domain at an index is its live alternatives. D = the domain at utteranceIdx; D⁺ via X-marking (larger base, same index: xMarkedBase) or via HP (same base, earlier index: hpIdx) — same operator, same two conclusions on each route: §4.1's uniformity of meaning under varying use.

                The Anderson consequent, that Jones shows the symptoms.

                Equations
                Instances For

                  A backward-closed history, on which a world is live at an index iff it is the index's own world or the index's time is at most 0.

                  Equations
                  Instances For
                    @[simp]
                    theorem Mizuno2024.mem_historyJones {w : Bool} {s : Reference.Index Bool ℤ} :
                    w ∈ historyJones s ↔ w = s.world ∨ s.time ≤ 0

                    The utterance index; its domain is the paper's D.

                    Equations
                    Instances For

                      The HP-shifted index (§3.3); its domain is D⁺ on the O-marking route.

                      Equations
                      Instances For

                        The X-marked base makes every world live at every index (§2, fn 6), the D⁺ of the X-marking route.

                        Equations
                        Instances For

                          Over D the consequent is trivial, since only the symptom world is live (§2).

                          The symptom-absent world is not live at the utterance index.

                          O-marking without expansion is trivially true for any antecedent, the infelicity of ex. (2).

                          X-marking's expansion is strict — the paper's D ⊂ D⁺ (§2).

                          Over the X-marked D⁺ the consequent is left open (§2).

                          Japanese O-marking: HP expansion under branching time #

                          §3.3: the HP shift moves the evaluation index backward; live possibilities shrink monotonically over time (historyJones_antitone, the backward closure of [Con02]), so the earlier index has a larger domain.

                          The HP shift strictly enlarges the live domain (§3.3): subset by antitonicity, strictness by the newly live symptom-absent world.

                          Over the HP-shifted domain the consequent is left open — the conclusion consequent_open_over_xMarkedBase reaches by enlarging the base (§3.3, §4.1).

                          theorem Mizuno2024.expanded_anderson_informative (antecedent : Set Bool) (h : hpIdx ∈ Conditional.strictImp historyJones antecedent showsSymptoms) :
                          ¬historyJones hpIdx ⊆ antecedent

                          A true Anderson conditional at the HP index excludes a live world (Set.not_subset for the witness form), so "one can make a meaningful, contingent claim" (§2).

                          Fragment marker connection #