Jardine (2016): transformations as correspondence-graph relations #
[Jar16] (Ch. 7) presents a phonological process as a relation between input
and output, given by correspondence graphs carved out of GEN by banned-subgraph
constraints. This file exercises the Autosegmental.Correspondence substrate on the
intervocalic-voicing schema (Jardine's running example): a faithfulness constraint
*[p↔p] — forbidding a p that corresponds to a p — admits the voicing
correspondence apa ↔ aba and rejects the faithful apa ↔ apa.
Scope #
This demonstrates the constraint mechanism at the correspondence-graph level. The full
relation-level result R(CG(φ)) = R_voice additionally requires GEN (CG(Γ), the
correspondence graphs built from primitives), which fixes the correspondence structure;
and Jardine's output-only markedness *VTV needs the arc-labelled-subgraph refinement
(Ch. 7 fn. 7). Both are deferred — see Autosegmental/Correspondence.lean.
Equations
- Jardine2016.instDecidableEqSeg x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- Jardine2016.instReprSeg.repr Jardine2016.Seg.a prec✝ = Repr.addAppParen (Std.Format.nest (if prec✝ ≥ 1024 then 1 else 2) (Std.Format.text "Jardine2016.Seg.a")).group prec✝
- Jardine2016.instReprSeg.repr Jardine2016.Seg.p prec✝ = Repr.addAppParen (Std.Format.nest (if prec✝ ≥ 1024 then 1 else 2) (Std.Format.text "Jardine2016.Seg.p")).group prec✝
- Jardine2016.instReprSeg.repr Jardine2016.Seg.b prec✝ = Repr.addAppParen (Std.Format.nest (if prec✝ ≥ 1024 then 1 else 2) (Std.Format.text "Jardine2016.Seg.b")).group prec✝
Instances For
Equations
- Jardine2016.instReprSeg = { reprPrec := Jardine2016.instReprSeg.repr }
A correspondence graph over the one alphabet (input and output both Seg).
Instances For
The faithfulness banned subgraph *[pⁱ↔p]: a p corresponding to a p. Forbidding
it forces an intervocalic p to change.
Equations
- Jardine2016.noPP = { upper := LabeledTuple.ofList [Jardine2016.Seg.p], lower := LabeledTuple.ofList [Jardine2016.Seg.p], links := {(0, 0)} }
Instances For
gVoice reads input apa, output aba.
The voicing correspondence is admitted by the *[p↔p] grammar (no p↔p).
The faithful correspondence is rejected — it contains a p↔p.
Hence apa ↔ aba lies in the relation presented by the local grammar *[p↔p],
witnessed by gVoice ([Jar16] Def. 25).