Docs / Phonology / Prosody / Domination

Constraints on prosodic domination

import Linglib.Phonology.Prosody.Domination · namespace Prosody

Selkirk 1996 decomposes the Strict Layer Hypothesis into four constraints on which prosodic category may dominate which. Layeredness forbids a node from dominating a higher level and Headedness requires every node above the syllable to dominate the next level down; Exhaustivity forbids a node of a level from immediately dominating a node more than one level down, and Nonrecursivity forbids it from dominating its own level. Each is a violation count on the prosodic tree. The three stated of a node and its daughters count such pairs; Headedness, like the Parse-into-ℓ of Itô & Mester 2003, is an alignment of the positions of the tree under dominance, each node of one family needing a partner in another below or above it.

Layeredness, counted on mothers and daughters, holds of everything a node dominates, and in a layered tree Nonrecursivity on mothers and daughters is the absence of recursion at any depth. A tree is strictly layered exactly when it violates none of the four, and a tree satisfying Headedness has a node at every level below its root.

The constraints are facts about one map, from the nodes of a structure ordered by dominance to the categories. Layeredness says that the map is antitone, and the Strict Layer Hypothesis that it grades the nodes: a daughter is one category below its mother and a node without daughters is of the lowest. Nothing in that reading asks the nodes to form a tree.

Implementation notes

The constraints are stated for trees over any linearly ordered categories and any marks, the lowest category in the role of the syllable, so they apply to a tree with or without prominence and to hierarchies with other categories than those of Level. Exhaustivity and Nonrecursivity take the level of the dominating node as an argument, as Selkirk 1996 ranks them level by level; noRec is the unindexed No-Recursion of Itô & Mester 2009. Headedness reads "dominates" at any depth, so that a prosodic word whose only foot lies inside a prosodic word it dominates satisfies it.

References

Declarations

L66–77
L79
L81
L83
L85

Counting mothers and daughters

Layeredness and Nonrecursivity

L114–117

Maximal and minimal projections

The projections of a level ℓ (Itô & Mester 2009) are its extremal nodes under dominance. The maximal ℓ-projections are the ℓ-nodes no ℓ-node dominates, and the minimal ones are the ℓ-nodes dominating no ℓ-node. Since the root is the least position, these are the Minimal and Maximal members of the ℓ-positions. That no ℓ-node dominates another at any depth says that the ℓ-positions form an antichain, and it collapses the two. nonrecursivity ℓ counts only a daughter, so ω(f(ω)), which recurses through an intervening foot, satisfies it while its maximal and minimal ω-projections differ; that tree violates Layeredness, and in a layered tree the two agree. The intonational utterance υ is the maximal ι-projection.

L173–175
L194–200
L234–238

Exhaustivity

L246–249

Headedness and the Strict Layer Hypothesis

Headedness fills the hierarchy: a tree satisfying Headedness has a node at every level from its root down, so that an utterance of a single syllable is a syllable, a foot, a word and a phrase at once (Selkirk 1996).

theorem exists_isSubtree_level_eqL311–327
{L : Type u_1} {M : Type u_2} [LinearOrder L] {t : RoseTree (Constituent L M)} {ℓ : L} [DecidableRel fun (x1 x2 : L) => x1 ⋖ x2] [OrderBot L] [WellFoundedLT L] (h : headedness t = 0) (hℓ : ℓ ≤ t.value.level) :
∃ (s : RoseTree (Constituent L M)), s.IsSubtree t ∧ s.value.level = ℓ

Built from

L329–333

The Strict Layer Hypothesis is the four constraints together (Selkirk 1996): a tree is strictly layered exactly when it violates none of Layeredness, Headedness, Exhaustivity and Nonrecursivity.

theorem isStrictlyLayered_iffL343–365
{L : Type u_1} {M : Type u_2} [LinearOrder L] {t : RoseTree (Constituent L M)} [DecidableRel fun (x1 x2 : L) => x1 ⋖ x2] [OrderBot L] :
IsStrictlyLayered t ↔ layeredness t = 0 ∧ headedness t = 0 ∧ ∀ (ℓ : L), exhaustivity ℓ t = 0 ∧ nonrecursivity ℓ t = 0

Built from , , , ,

The hierarchy grades dominance

The positions of a tree are ordered by dominance, the root least, and levelAt sends each to its category. Selkirk's constraints are order-theoretic properties of this map.

The Strict Layer Hypothesis says the hierarchy grades dominance: a tree is strictly layered exactly when each daughter is one category below its mother and each position without daughters is of the lowest category.

theorem isStrictlyLayered_iff_covByL384–400
{L : Type u_1} {M : Type u_2} [LinearOrder L] {t : RoseTree (Constituent L M)} [OrderBot L] :
IsStrictlyLayered t ↔ (∀ ⦃p q : t.Positions⦄, p ⋖ q → levelAt q ⋖ levelAt p) ∧ ∀ ⦃p : t.Positions⦄, IsMax p → levelAt p = ⊥

Built from ,

L402–407
L409–416

Well-formed constituents