@cite{amato-2025} — Filler-gap case studies via Nested Agree

@cite{amato-2025} (NLLT) §4.2.3 extends Nested Agree from Agree features to Merge features, deriving order-preserving multiple wh-fronting in Bulgarian (and Romanian) from a probe stack on C where the Merge probe [•wh•] is restricted by Nested Agree to the last goal of the prior Agree probe [*wh*]. This file collects the Amato-2025 case studies whose primary phenomenon is filler-gap dependency formation.

Currently only the Bulgarian case (§4.2.3) lands here. Other filler-gap-adjacent Amato-2025 material (none in the published §4) would land in this file as additional sub-namespaces.

Cross-domain validation thesis

NestedAgreeConfig was originally designed against an Agree-feature case (Italian aux selection). Bulgarian uses Merge features — formally identical at the abstract layer (a probe is a probe), but linguistically distinct (movement vs valuation). That the same NestedAgreeConfig shape models Bulgarian without API changes validates the abstraction's neutrality between Agree and Merge.

Out of scope

The Bulgarian formalization captures the structural Nested-Agree restriction on the Merge probe — the central §4.2.3 claim that post-Agree Merge is constrained to the prior Agree's last goal. It does not capture:

• Multiple Agree on the prior probe ([*wh*] actually hits both wh-sbj and wh-obj sequentially; we abstract by setting the goal to wh-obj, the last hit). Multiple Agree as a primitive would require runStack to return List Token rather than Option Token — a substantive API extension deferred until a case study genuinely needs it.
• Probe restart: in Amato's full derivation, after wh-obj raises to inner Spec,C, the Merge probe re-discharges to raise wh-sbj to outer Spec,C. We model only the first movement step.
• Movement itself: SyntacticObject is static. The post-movement tree (with wh-obj in Spec,C and a trace below) is not constructed here. The structural claim about the pre-movement configuration is what the Nested-Agree theorem captures.

Honest scope: we prove that Nested Agree's matryoshka excludes wh-sbj from the Merge probe's truncated domain at the relevant derivational step. The full multi-step Bulgarian fronting derivation is future work.

Part A: Bulgarian multiple wh-fronting (Amato §4.2.3)

Bulgarian permits (and requires) multiple wh-phrases to front to the left periphery, in order-preserving fashion: when a sentence has both a wh-subject and a wh-object, the surface order is wh-sbj > wh-obj matching base-merge order, not the order standard minimality would predict (which would require wh-sbj first regardless of base order).

@cite{amato-2025} §4.2.3 derives order-preservation from Nested Agree applied to Merge features:

1. C bears [*wh*] ≻ [•wh•] — Agree feature first, Merge feature second.
2. [*wh*] undergoes Multiple Agree: hits wh-sbj first (closest), then extends to wh-obj. Last goal: wh-obj.
3. [•wh•] by Nested Agree must target the same goal as the prior probe → wh-obj. wh-obj raises to inner Spec,C.
4. (Out of scope) [•wh•] re-discharges; finds wh-sbj; raises it to outer Spec,C. Surface: outer Spec,C > inner Spec,C → wh-sbj > wh-obj.

This file formalizes step (3): the structural Nested-Agree restriction on the Merge probe. The wh-sbj is in C's c-command (probe 0's domain) but excluded from wh-obj's daughters (probe 1's truncated domain). This is the same apparent_intervener_excluded structure that applied to Italian DPsubj, Icelandic DPdat, and Lak Erg — modulo the Agree-vs-Merge distinction at the consumer level.

def Phenomena.FillerGap.Studies.Amato2025.Bulgarian.bulgarianMultiWh : Minimalist.NestedAgree.NestedAgreeConfig

Bulgarian multi-wh configuration at the relevant derivational step (post-Multiple-Agree, pre-movement). Pre-movement tree: C [wh-sbj [V wh-obj]], goal = wh-obj (the last token Multiple Agree reached, what Nested Agree forces the Merge probe to also target). The Agree-vs-Merge distinction between probes 0 and 1 is implicit in the linguistic interpretation, not in the abstract config.

Equations

• One or more equations did not get rendered due to their size.

theorem Phenomena.FillerGap.Studies.Amato2025.Bulgarian.bulgarianMultiWh_is_nested : Minimalist.NestedAgree.IsNestedAgreeConfig bulgarianMultiWh

theorem Phenomena.FillerGap.Studies.Amato2025.Bulgarian.bulgarianMultiWh_runStack_0 : Minimalist.NestedAgree.runStack bulgarianMultiWh 0 = some (Minimalist.SyntacticObject.leaf Phenomena.FillerGap.Studies.Amato2025.Bulgarian.aWhObj✝)

theorem Phenomena.FillerGap.Studies.Amato2025.Bulgarian.bulgarianMultiWh_runStack_1 : Minimalist.NestedAgree.runStack bulgarianMultiWh 1 = some (Minimalist.SyntacticObject.leaf Phenomena.FillerGap.Studies.Amato2025.Bulgarian.aWhObj✝)

theorem Phenomena.FillerGap.Studies.Amato2025.Bulgarian.bulgarianMultiWh_excludes_wh_subject : Minimalist.SyntacticObject.leaf Phenomena.FillerGap.Studies.Amato2025.Bulgarian.aWhSbj✝ ∈ bulgarianMultiWh.searchDomain 0 ∧ Minimalist.SyntacticObject.leaf Phenomena.FillerGap.Studies.Amato2025.Bulgarian.aWhSbj✝ ∉ bulgarianMultiWh.searchDomain 1

Apparent wh-subject intervention is not actual. The wh-subject is in C's c-command (probe 0's full domain) but is not in wh-obj's daughters (wh-obj doesn't c-command wh-sbj — wh-sbj is structurally above wh-obj). So wh-sbj is excluded from probe 1's truncated domain by Nested Agree.

This is the central §4.2.3 derivational claim: the Merge probe [•wh•], restricted by Nested Agree to the last Agree goal (= wh-obj), cannot target wh-sbj even though wh-sbj is the structurally closer wh-phrase. wh-obj raises first; wh-sbj is raised in the (out-of-scope) restart phase.