Vago (2017): *VV in Hungarian #
Vago analyses the vowel sequences that suffixation creates in Hungarian with Casali's position-sensitive MAX constraints and a ban on vowel sequences, *VV. Between a root and a suffix the second vowel usually deletes, under the hierarchy MAX LEX ≫ *VV ≫ MAX MI, and a handful of suffixes are indexed to rerankings that delete the first vowel instead or keep both. Between two suffixes the second vowel deletes after a long vowel or a suffix of one vowel, which high-ranked MAX constraints on long vowels and on morphs of one segment protect, and the first vowel deletes after the infinitival suffix -ni. Vago concludes that Casali's constraints make the right predictions for Hungarian.
Main results #
optimal_regular: between a morph and a suffix that are not roots, the regular hierarchy deletes the second vowel when the first is long or alone in its morph, and the first otherwise.tableaux: for every example, the tableau of its four candidates (no deletion, either vowel, both) under the hierarchy of its suffix selects the attested deletion, and its output spells the attested form.starVV_dominates_of_erdesz: a ranking that deletes the first vowel of erdő-ész ranks *VV above MAX on long vowels and MAX LEX, since the deleted root vowel is long.
Implementation notes #
- Each example is a row of
Data/Forms/Vago2017giving its two morphs, their kinds, the morph before them, the attested deletion and the hierarchy its suffix is indexed to. Long vowels are read off the accents of the spelling, length being prosodic in the Hungarian fragment. - The hierarchy that deletes the first vowel is taken as Table 2 prints it, without the
constraints on long vowels and morphs of one segment, which the paper ranks above the regular
hierarchy; with MAX on long vowels on top it would keep the long vowel of erdő
(
starVV_dominates_of_erdesz). - Vago derives the deletion after -ni in Table 4 from the reranking (8b), and the row follows
him; the regular hierarchy derives it too (
optimal_regular), since MAX LEX does not bear on two suffixes.
References #
Equations
- Vago2017.instDecidableEqCon x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- Vago2017.instFintypeCon = { elems := { val := ↑Vago2017.Con.enumList, nodup := Vago2017.Con.enumList_nodup }, complete := Vago2017.instFintypeCon._proof_1 }
Equations
- Vago2017.instReprCon.repr Vago2017.Con.maxLong prec✝ = Repr.addAppParen (Std.Format.nest (if prec✝ ≥ 1024 then 1 else 2) (Std.Format.text "Vago2017.Con.maxLong")).group prec✝
- Vago2017.instReprCon.repr Vago2017.Con.maxMS prec✝ = Repr.addAppParen (Std.Format.nest (if prec✝ ≥ 1024 then 1 else 2) (Std.Format.text "Vago2017.Con.maxMS")).group prec✝
- Vago2017.instReprCon.repr Vago2017.Con.maxLex prec✝ = Repr.addAppParen (Std.Format.nest (if prec✝ ≥ 1024 then 1 else 2) (Std.Format.text "Vago2017.Con.maxLex")).group prec✝
- Vago2017.instReprCon.repr Vago2017.Con.starVV prec✝ = Repr.addAppParen (Std.Format.nest (if prec✝ ≥ 1024 then 1 else 2) (Std.Format.text "Vago2017.Con.starVV")).group prec✝
- Vago2017.instReprCon.repr Vago2017.Con.maxMI prec✝ = Repr.addAppParen (Std.Format.nest (if prec✝ ≥ 1024 then 1 else 2) (Std.Format.text "Vago2017.Con.maxMI")).group prec✝
Instances For
Equations
- Vago2017.instReprCon = { reprPrec := Vago2017.instReprCon.repr }
A candidate is the set of vowels it deletes, and deleted c D the positions they occupy.
Equations
- Vago2017.deleted c D = Finset.image (Casali1997.position c) D
Instances For
The form a candidate gives.
Equations
- Vago2017.resolve c D = ((c.stemBody ++ if Data.Hiatus.ElidedVowel.first ∈ D then [] else [c.v1]) ++ if Data.Hiatus.ElidedVowel.second ∈ D then [] else [c.v2]) ++ c.suffixBody
Instances For
The violations of a candidate, with the long vowels at the positions long.
Equations
- Vago2017.marks c long D Vago2017.Con.maxLong = OptimalityTheory.maxAt (fun (x : ℕ) => x ∈ long) (Vago2017.deleted c D)
- Vago2017.marks c long D Vago2017.Con.maxMS = OptimalityTheory.maxAt (c.Protects Phonology.Hiatus.Prominence.soleSegment) (Vago2017.deleted c D)
- Vago2017.marks c long D Vago2017.Con.maxLex = OptimalityTheory.maxAt (c.Protects Phonology.Hiatus.Prominence.lexical) (Vago2017.deleted c D)
- Vago2017.marks c long D Vago2017.Con.starVV = Phonology.Hiatus.count (Vago2017.resolve c D)
- Vago2017.marks c long D Vago2017.Con.maxMI = OptimalityTheory.maxAt (c.Protects Phonology.Hiatus.Prominence.morphemeInitial) (Vago2017.deleted c D)
Instances For
The tableau of the four candidates under a constraint order, most dominant first.
Equations
- Vago2017.tableau c long order = OptimalityTheory.Tableau.ofOrder order (Vago2017.marks c long)
Instances For
The hierarchy a suffix is indexed to.
- regular : Hierarchy
The regular hierarchy (8a), MAX LEX ≫ *VV ≫ MAX MI, below MAX on long vowels and on morphs of one segment.
- v1Deletion : Hierarchy
The reranking (8b) that deletes the first vowel, *VV ≫ MAX MI ≫ MAX LEX.
- noDeletion : Hierarchy
The reranking (9b) that deletes neither vowel, MAX LEX ≫ MAX MI ≫ *VV.
Instances For
Equations
- Vago2017.instDecidableEqHierarchy x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- One or more equations did not get rendered due to their size.
- Vago2017.instReprHierarchy.repr Vago2017.Hierarchy.regular prec✝ = Repr.addAppParen (Std.Format.nest (if prec✝ ≥ 1024 then 1 else 2) (Std.Format.text "Vago2017.Hierarchy.regular")).group prec✝
Instances For
Equations
- Vago2017.instReprHierarchy = { reprPrec := Vago2017.instReprHierarchy.repr }
The constraint order of a hierarchy, most dominant first.
Equations
- Vago2017.Hierarchy.regular.order = [Vago2017.Con.maxLong, Vago2017.Con.maxMS, Vago2017.Con.maxLex, Vago2017.Con.starVV, Vago2017.Con.maxMI]
- Vago2017.Hierarchy.v1Deletion.order = [Vago2017.Con.starVV, Vago2017.Con.maxMI, Vago2017.Con.maxLex]
- Vago2017.Hierarchy.noDeletion.order = [Vago2017.Con.maxLex, Vago2017.Con.maxMI, Vago2017.Con.starVV]
Instances For
Between two suffixes #
Between a morph and a suffix that is not a root, under the regular hierarchy, the second vowel deletes when the first is long or the sole segment of its morph, and the first vowel deletes otherwise. This holds when the second vowel is short and not alone in its morph, and deleting either vowel or both leaves no vowel sequence.
The examples #
The context of two spelled morphs of the given kinds, if the first ends and the second begins with a vowel.
Equations
- One or more equations did not get rendered due to their size.
- Vago2017.context? before left right l r✝ = none
Instances For
An example records its context, the positions of its long vowels, the segments of the morph before the context, the hierarchy of its suffix and the attested deletion.
- context : Phonology.Hiatus.Context
- long : Finset ℕ
- before : List Phonology.Segment
- hierarchy : Hierarchy
- deletion : Finset Data.Hiatus.ElidedVowel
Instances For
The morph kind of the paper's label.
Equations
Instances For
The hierarchy of the paper's label.
Equations
Instances For
The deleted vowels of the paper's label.
Equations
Instances For
The example the paper's analysis of a form records.
Equations
- One or more equations did not get rendered due to their size.
Instances For
For every example, the tableau of its candidates under the hierarchy of its suffix selects the attested deletion, and the form it gives, after the morph before it, spells the attested form.
The ranking that deletes the first vowel #
Vago's constraints on the candidates of an example.
Equations
- Vago2017.con x i D = Vago2017.marks x.context x.long D i
Instances For
The example erdő-ész of Table 2.
Equations
Instances For
A ranking of Vago's constraints under which erdő-ész deletes its first vowel ranks *VV above MAX on long vowels and above MAX LEX, since that vowel is long and in a root.