Docs / Morphology / DistributedMorphology / LateInsertion

Late insertion

import Linglib.Morphology.DistributedMorphology.LateInsertion · namespace DistributedMorphology

The syntax hands PF a spell-out domain: the pronounced terminals of a linearized syntactic object, left to right, each read as a feature bundle (SpelloutDomain.ofLinearized). The postsyntactic pipeline then runs on it and Vocabulary Insertion realizes each surviving position in its neighborhood: late insertion, on which no phonological content travels through the syntax. Early insertion is the degenerate case: over the vocabulary whose items each spell out one terminal's own token as its lexically listed form, the empty-module pipeline returns exactly the carrier's pfPhon, so terminals carrying their forms compute what one context-free item per token computes, and contextual allomorphy is precisely what such a vocabulary cannot state.

Main definitions

Main results

References

Declarations

Early insertion

pfPhon is the early-insertion pipeline: handing the pronounced terminals over as atomic features and inserting each token's lexically listed form returns the carrier's own pronunciation, silent tokens dropped.

theorem pfPhon_eq_earlyInsertionL97–111
(t : Minimalist.LinearizedSyntacticObject) :
Minimalist.pfPhon t = List.filter (fun (x : String) => decide (x ≠ "")) ((earlySpellout (Minimalist.pfYield t)).pf (SpelloutDomain.ofLinearized (fun (x : Minimalist.LexicalItem.Token) => {x}) t)).flatten

Built from ,