Slavic Verbalizer — OT Analysis @cite{stojkovic-2026} #
OT evaluation connecting constraint rankings to the empirical surface forms of the Slavic verbalizer (VBLZ). Stojković (2026) argues the VBLZ has a single abstract underlying representation (a defective diphthong) across all Slavic; surface alternation is derived by OT constraint ranking.
Candidates #
The VBLZ in the pre-vocalic context (before the thematic /-a-/) can surface as one of five candidates, differing in how the unspecified slot of the diphthong is resolved:
| Candidate | Vowel | Mechanism |
|---|---|---|
| [ov] | [o] = [+back, −high] | epenthesise [+back] and [−high] |
| [ev] | [e] = [−back, −high] | share [−back] from palatal, epenthesise [−high] |
| [uv] | [u] = [+back, +high] | epenthesise [+back] and [+high] |
| [iv] | [i] = [−back, +high] | share [−back] from palatal, epenthesise [+high] |
| [u.a] | hiatus | monophthongise, delete unspecified slot |
Constraints #
Six constraints, of which two (NOHIATUS, SPECIFY) are undominated and four (*SHARE[−back], DEP[+back], DEP[−high], DEP[+high]) are variable:
- NOHIATUS (markedness): no adjacent vowels
- SPECIFY (markedness): every base node must be fully specified
- *SHARE[−back] (markedness): don't copy [−back] from adjacent segment
- DEP[+back] (faithfulness): don't epenthesise [+back]
- DEP[−high] (faithfulness): don't epenthesise [−high]
- DEP[+high] (faithfulness): don't epenthesise [+high]
Note: DEP[+high] is implicit in Stojković's analysis. Without it, [uv] harmonically bounds [ov] (strictly fewer violations at every constraint), making it impossible for any ranking to select [ov]. The paper derives the same effect from the markedness of [+high] vs [−high]: [−high] is the cross-linguistically unmarked epenthetic height (p. 14), while [+high] epenthesis incurs an implicit faithfulness cost. Making this explicit as DEP[+high] yields the correct factorial typology.
Factorial Typology #
The 4! = 24 permutations of the four variable constraints produce exactly 4 distinct optima: {[ov]}, {[ev]}, {[uv]}, {[iv]}. Three correspond to attested groups; {[iv]} is unattested.
Candidate surface realisations of the VBLZ before the thematic /-a-/. Each represents a different resolution of the defective diphthong.
- ov : VBLZCandidate
[ov]: diphthong fissions, slot 1 → [o] via epenthetic [+back, −high].
- ev : VBLZCandidate
[ev]: diphthong fissions, slot 1 → [e] via shared [−back], epenthetic [−high].
- uv : VBLZCandidate
[uv]: diphthong fissions, slot 1 → [u] via epenthetic [+back, +high].
- iv : VBLZCandidate
[iv]: diphthong fissions, slot 1 → [i] via shared [−back], epenthetic [+high].
- uHiatus : VBLZCandidate
[u.a]: monophthongisation, unspecified slot deleted, hiatus with /-a-/.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
All candidates.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The four fission candidates (excluding hiatus), for factorial typology.
Equations
- One or more equations did not get rendered due to their size.
Instances For
NOHIATUS: assign * for adjacent vowels.
Equations
- One or more equations did not get rendered due to their size.
Instances For
SPECIFY(•→[F]): assign * for unspecified base node. Only violated by the fully faithful candidate (not in our candidate set), so this constraint is vacuously satisfied. Included for completeness.
Equations
- One or more equations did not get rendered due to their size.
Instances For
DEP[+back]: don't epenthesise [+back] on the unspecified slot. Violated by [ov] and [uv] (both epenthesise [+back] to get a back vowel).
Equations
- One or more equations did not get rendered due to their size.
Instances For
DEP[−high]: don't epenthesise [−high] on the unspecified slot. Violated by [ov] and [ev] (both epenthesise [−high] to get a mid vowel).
Equations
- One or more equations did not get rendered due to their size.
Instances For
DEP[+high]: don't epenthesise [+high] on the unspecified slot. Violated by [uv] and [iv] (both epenthesise [+high] to get a high vowel).
Implicit in Stojković's analysis. Without this constraint, [uv] harmonically bounds [ov], making it impossible for any ranking to select [ov] as optimal. Stojković derives the equivalent effect from the markedness of [+high] vs [−high] (p. 14): "The feature [−high] is cross-linguistically more likely to be unmarked compared to [+high]."
Equations
- One or more equations did not get rendered due to their size.
Instances For
The four variable constraints (excluding the undominated NOHIATUS and vacuous SPECIFY). These are permuted in the factorial typology.
Equations
- One or more equations did not get rendered due to their size.
Instances For
All six constraints in one list.
Equations
- One or more equations did not get rendered due to their size.
Instances For
[ov] group ranking (Stojković 2026, (17)): NOHIATUS, SPECIFY ≫ *SHARE[−back] ≫ DEP[+high] ≫ DEP[+back] ≫ DEP[−high]
*SHARE[−back] high: sharing [−back] from palatals is banned. DEP[+high] above DEP[−high]: [+high] epenthesis is costlier than [−high], making [−high] the default → mid vowel [o] surfaces.
Equations
- One or more equations did not get rendered due to their size.
Instances For
[ov]/[ev] group ranking (Stojković 2026, (21)): NOHIATUS, SPECIFY ≫ DEP[+back] ≫ DEP[+high] ≫ *SHARE[−back] ≫ DEP[−high]
DEP[+back] high: epenthesising [+back] is banned. After palatals, sharing [−back] is the only option → [ev]. After non-palatals, [−high] epenthesis yields [ov] (in a separate non-palatal evaluation).
Equations
- One or more equations did not get rendered due to their size.
Instances For
[uv] group ranking (Stojković 2026, (29)): NOHIATUS, SPECIFY ≫ DEP[−high] ≫ *SHARE[−back] ≫ DEP[+back] ≫ DEP[+high]
DEP[−high] high: epenthesising [−high] is banned → [+high] surfaces. *SHARE[−back] above DEP[+back]: epenthesising [+back] is cheaper than sharing [−back] → back vowel [u] surfaces.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The [ov] group ranking selects [ov] as the unique optimal candidate.
The [ov]/[ev] ranking selects [ev] as optimal in the palatal context. (In the non-palatal context, [ev] and [iv] are unavailable because there is no palatal to share [−back]; [ov] wins trivially.)
The [uv] group ranking selects [uv] as the unique optimal candidate.
The factorial typology over the four variable constraints and four fission candidates produces exactly 4 distinct optimal sets.
Three correspond to attested Slavic groups:
- {[ov]}: *SHARE, DEP[+high] ranked high
- {[ev]}: DEP[+back], DEP[+high] ranked high
- {[uv]}: DEP[−high], *SHARE ranked high
The fourth ({[iv]}) is unattested in Slavic.
The four distinct optima are exactly the four singleton candidate sets.
The [iv] pattern (shared [−back] + [+high], giving a front high vowel) is the only unattested pattern among the four predicted by the factorial typology (Stojković 2026, (31f)).
Equations
- One or more equations did not get rendered due to their size.
Instances For
The [ov] group's OT-predicted form matches the empirical data.
The [uv] group's OT-predicted form matches the empirical data.
The [ov]/[ev] group's OT-predicted forms match the empirical data:
[ev] in the palatal context (proved by ev_optimal), [ov] in the
non-palatal context (where [ev] is unavailable).
The surface form of each candidate.
Equations
- Phenomena.Allomorphy.SlavicVerbalizer.Studies.Stojkovic2026.VBLZCandidate.ov.surfaceForm = "ov"
- Phenomena.Allomorphy.SlavicVerbalizer.Studies.Stojkovic2026.VBLZCandidate.ev.surfaceForm = "ev"
- Phenomena.Allomorphy.SlavicVerbalizer.Studies.Stojkovic2026.VBLZCandidate.uv.surfaceForm = "uv"
- Phenomena.Allomorphy.SlavicVerbalizer.Studies.Stojkovic2026.VBLZCandidate.iv.surfaceForm = "iv"
- Phenomena.Allomorphy.SlavicVerbalizer.Studies.Stojkovic2026.VBLZCandidate.uHiatus.surfaceForm = "u"
Instances For
The OT-optimal candidate for the [ov] group produces the attested form.
The OT-optimal candidate for the [uv] group produces the attested form.
The OT-optimal candidate for the [ov]/[ev] palatal context produces the attested palatal-conditioned form.