@cite{pesetsky-2013} — Russian Case Morphology and the Syntactic Categories #

Pesetsky's central thesis: morphological case is the realization of a syntactic part-of-speech category. The four major cases line up with the four traditional lexical categories, formalized here as a restriction of Minimalism's Cat:

Case	POS category	Assigner
Genitive	N	every noun (Primeval Genitive, p. 9)
Nominative	D	the D head of a DP
Accusative	V	a transitive V (under VACC, eq. 77 p. 66)
Obliques	P	a P head; DAT canonical exemplar (p. 7)

Pesetsky writes (book p. 7): "only the distinctions among these oblique cases in Russian will fail to correspond to a traditional part-of-speech distinction" — P_OBL is canonical, P_DAT is one exemplar with additional features. POSCat.cases returns the Russian oblique series for .P (DAT/INS/LOC, matching Fragments.Slavic.Russian.Case.caseInventory); POSCat.canonicalExemplar picks DAT.

Architecture: layered grounding on Minimalism #

This is a Minimalist study. Its tree is SyntacticObject (@cite{marcolli-chomsky-berwick-2025}, formalized in Theories/Syntax/Minimalist/Basic.lean); its structure-building operation is merge; its categories are members of Cat. Per-leaf case stacks are a derived structural property of the SyntacticObject — caseStackAt walks the tree, treating left-Merge as "head + complement" per Pesetsky's Minimalist convention, and prepends the head's POSCat to the complement's leaves. No external state, no side-map: the case stack at a leaf is a function of the tree structure plus the leaf's identity.

The genuinely paper-specific apparatus is small:

POSCat — the {N,D,V,P} restriction of Minimalism's Cat to the four case-bearing categories. The injection POSCat.toCat makes the restriction structural rather than parallel.
caseStackAt / surfaceCase — Pesetsky's FA realized as structural recursion over the SO. Each .node head body step in the tree corresponds to a Merge; if head.outerCat is a POSCat and the queried leaf is in body, the head's category is prepended to the leaf's stack. This is FA "version 3 of 6" (@cite{pesetsky-2013} eq. (76), book p. 65) made implicit: complementation-requirements gating is not modeled (treated as always satisfied), and "designated as feature assigner for β" is exactly the POSCat.ofCat? head.outerCat test (only N/D/V/P heads FA).
One-Suffix Rule (Ch 3) — Russian's spell-out: the outermost stratum (head of the stack) is realized; inner strata are deleted. Encoded by surfaceCase = (caseStackAt ...).head?.map ....
Primeval Genitive Conjecture (eq. (6), p. 9) — NGEN categorizes a Russian root as a noun in the lexicon, in the spirit of @cite{marantz-1997}'s root + categorizer. Automatic from the recursion: a leaf with outerCat = .N carries [POSCat.N] in its base case, so any noun-categorized LIToken has primeval N without explicit initialization. The structural bridge to DM Categorizer.n is primevalN_via_dm_categorizer.

Two corollaries reshape standard Case theory #

Every noun is "born" with NGEN in the lexicon. The base case of caseStackAt for an outerCat = .N leaf returns [POSCat.N] without input.
Nominative as Schützean default. D-Merge assigns DNOM (eq. (87) p. 73 — D does assign nominative morphology); but DNOM is "default" in the sense of Schütze 1997, 2001 (cited p. 73): D's assignment survives in subject position because Russian finite T does not assign features under FA, so nothing later overwrites it. When V intervenes (transitive object under eq. 77 conditions), V overwrites with VACC; otherwise DNOM stays. This inverts the standard "T assigns NOM" view — Pesetsky writes (p. 72) that he has "never needed to assume that nominative morphology is assigned to the elements of DP by anything other than D itself."

What is and isn't modeled #

Modeled: FA version 3 of 6 (eq. (76) p. 65) with α-cat / β-target entanglement implicit in the structural recursion; Russian One-Suffix Rule; Primeval Genitive (with DM Categorizer.n bridge); the four POS categories as restriction of Cat; the basic NOM/ACC/GEN/OBL bridge; noDeletionSpellOut as the limiting "no deletion" case (Lardil itself has partial deletion per Ch 3 §3.2 — modeled here as the limit, with the actual Lardil pattern deferred); cross-framework agreement bridge with @cite{caha-2009}.

Out of scope (deferred to follow-on study files or commits):

The complete VACC restriction (eq. 77 p. 66 — [+FEM,+SG] ∨ [+ANIM] ∨ [+PRONOMINAL]) plus the realization rules eq. (80) p. 67 (assignment-vs-realization is a two-layer story); to be added in a follow-on commit with a paper-local phi-feature side-map.
FA versions 4-6: Spell-Out locality (Ch 8 §8.3 eq. (110)/(111) p. 87-88) and the Prototype mechanism (Ch 9). The current caseStackAt realizes onto every leaf of body; this is wrong for multi-leaf body under Spell-Out — for the single-leaf cases §6 models, the difference doesn't arise.
The Number Mismatch puzzle for paucals (Ch 4) — dva stola showing SG on the noun but PL on demonstratives/adjectives — and the Lebanese Arabic # parallel (Ch 5.3). The structural derivation (Quant-to-D with NBR pied-piping, eq. (61) p. 54) is left to a follow-on study.
The full D-headed PP analysis of POBL↔VACC alternations (Ch 7).
The realistic [DP D [NP A N]] Russian DP geometry (eq. (62) p. 54) — §6 theorems use the schematic D + N direct-merge.
The complementation-requirements gate (eq. (76a)). The current recursion treats every Merge as FA-firing; modeling the gate requires either a Bool-tagged Step extension or filtering the derivation history.

Connections #

Theories/Syntax/Minimalist/Basic.lean — the substrate. Cat, SyntacticObject, LIToken, merge all consumed here. The derivation-step infrastructure in Theories/Syntax/Minimalist/Derivation.lean is the natural next layer once FA gating is modeled.
Theories/Morphology/DM/Categorizer.lean (@cite{harley-2014}) — the structural bridge primevalN_via_dm_categorizer connects POSCat.N to DM Categorizer.n via Cat.N.
Core.Case (Blake) gives the cross-linguistic case inventory; we bridge POSCat → Finset Core.Case (POSCat.cases) so Pesetsky's primitives plug into the existing typology.
Core.Case.Order provides Caha's containmentRank; the pesetsky_caha_compatible_prefix theorem records the cross- framework agreement (Pesetsky's on-Caha-hierarchy image equals Caha's full hierarchy).
Fragments/Slavic/Russian/Case.lean lists the Russian 6-case inventory.
vrač and Pesetsky's feminizing Ж analysis (Ch 5) are noted; the morphological-declension claim itself is owned by Phenomena/Gender/Studies/Kramer2020.lean::vrač_morph_masculine.