This file defines the Match constraints of the syntax–prosody interface. A constituent structure
over a string is the family of spans its constituents cover, so a syntactic input enters as the
spans of its constituents of a category, counted in the syllables of the output, and a prosodic
candidate as the spans of its nodes at a level. Match(α, π) requires each α-span to be the span
of a π-node and Match(π, α) each π-span to be an α-span: two-sided alignment, both edges of a
constituent coinciding with those of one constituent of the other category. One-sided alignment is
Constraint.alignLeft and Constraint.alignRight on the same families. MAX(α, π) asks only that
each α-constituent have a corresponding π-node, one containing its head and no head from outside
it, no π-node answering to two α-constituents.
Main definitions
Main results
Implementation notes
The syntactic family is a list of spans rather than a tree, so the constraints read any syntactic
representation, a phrase-structure tree or a morphological word, once the spans of its
constituents are given in the syllables of the output. The list keeps multiplicity, so a φ
recursively over a φ of the same span is counted twice.
References