Docs / Phonology / OptimalityTheory / Correspondence / Transderivational

Transderivational correspondence

import Linglib.Phonology.OptimalityTheory.Correspondence.Transderivational · namespace OptimalityTheory.Transderivational

This file defines Benua's transderivational correspondence theory, in which a derived word and its base stand in an output–output correspondence whose identity constraints are ranked with markedness and with the faithfulness of each word to its own input. A candidate is a subparadigm, the two words with their inputs and the three relations. Every constraint of the hierarchy is read at both words, and recursive evaluation ranks the copy that evaluates the base above the copy that evaluates the derived word.

Main definitions

Main statements

Implementation notes

  • OO-Identity is vacuous at the base, a simplex word that bears no OO relation of its own.
  • An OO-Identity constraint is proper to one OO relation, and is vacuous on subparadigms bearing another, so that affixes subcategorizing for different relations share one hierarchy.

References

Declarations

L66–72

The subparadigm whose OO relation runs through the derived word's input: the base's input stands in the derived word's at offset, so that the OO relation links the realisations of each segment of the base's input, and the epenthetic segments at a common site.

def paradigmL74–89
{α : Type u_1} (bIn b dIn d : List α) (ioB ioD : List (ℕ × ℕ)) (offset : ℕ := 0) :

The definition is too large to print here; read it in the source.

Built from

A constraint of the hierarchy, read at each word of the subparadigm.

structure ParadigmConstraintL91–96
(α : Type u_4) :
Type u_4

Built from

Used by markedness, ioFaith, ooIdentity, baseRecursion, derivedRecursion, recursive, nonrecursive, evalOn_recursive, evalOn_recursive_of_forall_eq, eq_zero_of_mem_evalOn_nonrecursive

Used in Alderete2001, Benua1997

Base priority is ranking priority: the winners of recursive evaluation are the derived word's winners among the subparadigms whose base wins its own copy of the hierarchy.

theorem evalOn_recursiveL137–143
{α : Type u_1} {ρ : Type u_2} {L : Type u_3} {cand : L → Correspondence.Coindexation Role α} {labels : Finset L} {ranking : List (ρ → ParadigmConstraint α)} {r : ρ} :
Constraint.evalOn cand labels (recursive ranking r) = Constraint.evalOn cand (Constraint.evalOn cand labels (baseRecursion ranking r)) (derivedRecursion ranking r)

Built from , , , ,

L145–156

Without recursion underapplication is never optimal: under summed evaluation, if some subparadigm satisfies the top constraints in both words, as overapplication satisfies OO-Identity and the markedness constraint driving the process, so does every winner.

theorem eq_zero_of_mem_evalOn_nonrecursiveL158–171
{α : Type u_1} {ρ : Type u_2} {L : Type u_3} {cand : L → Correspondence.Coindexation Role α} {labels : Finset L} {r : ρ} {top rest : List (ρ → ParadigmConstraint α)} {x y : L} (hy : y ∈ Constraint.evalOn cand labels (nonrecursive (top ++ rest) r)) (hx : x ∈ labels) (h0 : ∀ k ∈ top, (k r).atBase (cand x) = 0 ∧ (k r).atDerived (cand x) = 0) (k : ρ → ParadigmConstraint α) :
k ∈ top → (k r).atBase (cand y) = 0 ∧ (k r).atDerived (cand y) = 0

Built from , ,