Dependent Case ↔ Inventory Bridge #
Connects the dependent case algorithm (assignCases) to language-specific
case inventories from Fragment files. For each language, we prove that the
structural cases the algorithm can assign are members of that language's
validated case inventory.
Structure #
- § 1: Language type assignments (accusative, ergative, or split)
- § 2: Inventory coverage — for each language, the structural cases
produced by
structuralCasesForare all in the language'scaseInventory - § 3: Ergative language coverage (Basque, Mam, Kaqchikel)
- § 4: Split-ergative coverage (Hindi, Georgian) — partial, because the algorithm's ABS is realized as NOM in these languages
- § 5: Concrete derivation examples for representative languages
The ABS/NOM Mismatch in Split-Ergative Languages #
The dependent case algorithm assigns ABS (CaseVal.abs) as the unmarked
case in ergative alignment. However, Hindi and Georgian realize this
function morphologically as NOM (no overt marker), not as a distinct ABS
form. Their inventories contain ERG (the dependent case) but not ABS.
This is a well-known typological fact: many split-ergative languages have
a syncretic unmarked case that serves both the nominative (accusative
frames) and absolutive (ergative frames) functions.
The bridge documents this: we prove full coverage for accusative alignment and ERG-specific coverage for ergative alignment, noting that ABS → NOM is a morphological identity, not a gap in the theory.
German is an accusative language (NOM for S/A, ACC for P).
Equations
Instances For
Turkish is an accusative language.
Equations
Instances For
Russian is an accusative language.
Equations
Instances For
Czech is an accusative language.
Equations
Instances For
Polish is an accusative language.
Equations
Instances For
Ukrainian is an accusative language.
Equations
Instances For
Serbian is an accusative language.
Equations
Instances For
Slovenian is an accusative language.
Equations
Instances For
Greek is an accusative language.
Equations
Instances For
Latin is an accusative language.
Equations
Instances For
Finnish is an accusative language.
Equations
Instances For
Hungarian is an accusative language.
Equations
Instances For
Tamil is an accusative language.
Equations
Instances For
Japanese is an accusative language.
Equations
Instances For
Korean is an accusative language.
Equations
Instances For
Basque is an ergative language (ERG for A, ABS for S/P).
Equations
Instances For
Mam (Mayan) is a tripartite language (ERG, ACC, ABS all distinct).
Equations
Instances For
Kaqchikel (Mayan) is an ergative language.
Equations
Instances For
For each accusative language, the structural cases [NOM, ACC] are both members of that language's case inventory.
For ergative languages, the structural cases [ABS, ERG] are in the inventory. Basque and Kaqchikel are fully ergative; Mam is tripartite (ERG, ACC, ABS all distinct).
Hindi and Georgian are split-ergative: accusative alignment in some tense/aspect contexts, ergative alignment in others. We prove: (1) Full accusative structural coverage (NOM, ACC ∈ inventory) (2) ERG ∈ inventory (the dependent case in ergative frames) (3) ABS ∉ inventory — these languages realize the absolutive function morphologically as NOM, so the algorithm's ABS output maps to a case that is in the inventory under a different label.
ABS is not in Hindi's inventory: the absolutive function (unmarked S/P in perfective) is morphologically NOM.
But NOM IS in the inventory, documenting the ABS → NOM syncretism.
ABS is not in Georgian's inventory: the absolutive function is morphologically NOM in both aorist (ergative) and present (accusative) frames.
NOM covers the absolutive function in Georgian.
German Derivations #
"Der Mann sieht die Frau" (The man sees the woman)
- Transitive: subject (higher) + object (lower), both without lexical case
- Subject → NOM (unmarked), Object → ACC (dependent)
"Der Mann schläft" (The man sleeps)
- Intransitive: single NP without lexical case
- Subject → NOM (unmarked, no case competitor)
Equations
Instances For
All cases in the German transitive derivation are in German's inventory.
All cases in the German intransitive derivation are in German's inventory.
Turkish Derivations #
"Adam kadını gördü" (The man saw the woman)
- Transitive: subject NOM (∅), object ACC (-I)
"Adam uyudu" (The man slept)
- Intransitive: subject NOM (∅)
Basque Derivations (Ergative) #
"Gizonak mutila ikusi du" (The man-ERG boy-ABS see AUX)
- Transitive: agent (higher) → ERG (dependent), patient (lower) → ABS (unmarked)
"Mutila etorri da" (The boy-ABS come AUX)
- Intransitive: sole NP → ABS (unmarked, no case competitor)
Equations
Instances For
All cases in the Basque transitive derivation are in Basque's inventory.
All cases in the Basque intransitive derivation are in Basque's inventory.
Hindi Split-Ergative Derivations #
Hindi perfective transitive: "Raam-ne roTii khaayii" (Ram-ERG bread-NOM ate) — ergative alignment
Hindi imperfective transitive: "Raam roTii khaataa hai" (Ram-NOM bread-ACC eats AUX) — accusative alignment
The same structural configuration (agent + patient) yields different case frames depending on the tense/aspect conditioning of the split.
Equations
Instances For
In the perfective (ergative alignment), ERG is in the inventory but ABS is not — it is realized as NOM. The agent case (ERG) is correctly predicted; the patient case (ABS → NOM) requires the morphological identity documented in § 4.
Georgian Split-Ergative Derivations #
Georgian aorist transitive: "K'ac-ma bavšv-i naxa" (Man-ERG child-NOM saw) — ergative alignment
Georgian present transitive: "K'ac-i bavšv-s xedavs" (Man-NOM child-DAT sees) — accusative-like, with lexical DAT on object
In the present series, the patient receives lexical DAT from the verb, not structural ACC from dependent case.
Equations
Instances For
Equations
Instances For
The agent ERG in the aorist is in Georgian's inventory.
In the present series, lexical DAT on the patient bleeds dependent ACC: the agent gets NOM (no case competitor) and the patient gets DAT (lexical from V). Both are in Georgian's inventory.