Documentation

Linglib.Studies.Graf2019

Graf (2019): Monotonicity as an effective theory of morphosyntactic variation #

[Gra19]

[Gra19] explains typological gaps — *ABA in adjectival gradation, person-pronoun syncretism, case syncretism, noun stem allomorphy — with two components: a fixed base hierarchy of prominence relations, and the requirement that the form assignment from it be feasibly monotonic (Morphology.Containment.FeasiblyMonotone, his def. (6)). The keystone here is isContiguous_tfae: contiguity, feasible monotonicity, Elsewhere-generability, and Superset-spellability coincide over linear hierarchies — Graf's "effective theory" and the two insertion mechanisms are provably one constraint. Graf establishes the 3-cell case by exhaustion and verifies larger hierarchies instance-by-instance; the general equivalence is Morphology.Containment.isContiguous_iff_feasiblyMonotone, which this file instantiates on his Table 1 (adjectival gradation, via the English fragment) and Table 2 (person-pronoun syncretism, [Har15]'s survey).

Main declarations #

TODO #

Graf's preferred case-layer hierarchy, the PCC person-pair poset, and the GCC gender poset are genuinely partial orders — and the PCC/GCC maps go into the fixed two-element truth-value algebra (upper sets, not kernels) — so they need a partial-order generalization of Pattern and an upper-set substrate. Concrete deferred targets: his ban (14) on multiple cross-level case syncretisms; McFadden's nominative stem-allomorphy generalization (16) via the conflated hierarchy's cycle-forces-identity argument; the exactly-nine monotone PCC maps as a finite check.

The keystone: four characterizations coincide #

Over a linear containment hierarchy the following are equivalent: the pattern is contiguous; it is feasibly monotonic ([Gra19] def. (6)); it is generable by Elsewhere insertion over a terminal antihomophonous vocabulary (DM, [Bob12] ch. 2); it is spellable by Superset competition over a context-free antihomophonous lexicon (nanosyntax, [Cah09]). Graf's "effective theory" and the two insertion mechanisms are one constraint.

Table 1: adjectival gradation #

Every English fragment suppletion pattern is feasibly monotonic — [Gra19] Table 1's attested rows (AAA smart, ABB good), run over the whole fragment.

The unattested *ABA row: no codomain order makes good – better – goodest monotone ("Irrespective of how this set is ordered, there can be no monotonic function f with f(1) = f(3) ≠ f(2)").

Table 2: person-pronoun syncretism #

[Gra19] §3.1 runs the same hierarchy shape over person, 1 > 2 > 3 (the Zwicky hierarchy, [Zwi77]; the repo's feature-level anchor for it is Person.hierarchyRank), against [Har15]'s pronoun-inventory survey (Graf's Table 2): attested partitions include 1|2|3 (Jarawa, Kiowa), 1|23 (Damin), and crucially 12|3 (Winnebago) — an AAB shape — while 13|2 is missing from the survey, as monotonicity predicts: "the account provides a perfect fit for the typology [of pronouns], whereas it only carves out a superset of the attested patterns for adjectival gradation and case syncretism." The encoding runs the hierarchy with index ascending 1 → 3, the dual of Graf's prominence order — equivalent for (feasible) monotonicity, his fn. 1.

Jarawa, Kiowa: all three persons distinct (1|2|3).

Equations
Instances For

    Damin: second and third person share a form (1|23).

    Equations
    Instances For

      Winnebago: first and second person share a form (12|3) — an AAB shape, attested in person though not in gradation.

      Equations
      Instances For

        The unattested partition 13|2: first and third persons syncretic to the exclusion of second.

        Equations
        Instances For

          The missing partition is exactly the non-monotonic one.

          The AAB division of labor #

          [Gra19]: "monotonicity cannot give a unified explanation of the absence of both ABA and AAB patterns. However, this is actually a welcome state of affairs because AAB patterns do show up in other empirical domains" — attested in person (Winnebago), unattested in adjectival gradation, where the absence "has to be stipulated", e.g. in terms of syntactic containment (his suggestion). The pair below states the division precisely: AAB is feasibly monotonic, and its gradation-side exclusion is carried by realization-level conditions ([Bob12]'s markedness condition (202), via Morphology.Containment.csg2) — which the person attestation shows are not category-general, the same point the pronominal case/number data make in Studies/SmithMoskalEtAl2019.lean.

          AAB is feasibly monotonic: Graf's account cannot — and, given Winnebago, should not — exclude it.

          Under antihomophony plus the markedness condition (202), no vocabulary realizes an AAB pattern — the gradation-side exclusion, carried by the realization engine rather than by monotonicity.