Documentation

Linglib.Phenomena.Complementation.Studies.Bondarenko2022

Bondarenko 2022: Anatomy of an Attitude @cite{bondarenko-2022} #

Tatiana Bondarenko (2022). MIT PhD dissertation. Thesis advisors: P. Elliott, K. von Fintel, D. Fox, S. Iatridou, R. Schwarzschild.

Locator convention #

The §-numbers below DO NOT track a single dissertation chapter; the file is paper-anchored as a whole and the sections were added across multiple phases:

Cross-chapter equivalence claims in the chapter-1 sections (e.g. "Table 1.1 ↔ Table 4.1"; "ex. 1 ↔ §4.2 ex. 1-2") were noted from context but not re-checked across chapters and carry implicit UNVERIFIED: status. References to chapters not yet read carry explicit UNVERIFIED: flags inline.

Headline thesis #

Embedded CPs receive two kinds of denotations corresponding to a syntactic distinction in the left periphery (§1.1.1, paper ex. 1):

In some languages ContP is overtly exposed (Korean -ta declarative; Buryat gɔ 'say'); in others (English, Russian) it is null.

Three substantive claims this file formalizes #

  1. Equality semantics for displacement (§1.1.1.2, paper eq. 7), not subset semantics (eq. 10) or existential (eq. 15). The Cont-CP denotes equality with the embedded proposition, so CONT is a function rather than a downward-closed relation. Three arguments converge: (i) nominal CP interpretation (paper §2.4), (ii) impossibility of CP conjunction (Bassi & Bondarenko 2021), (iii) Russian NPI subjunctive distribution (paper ch. 5).

  2. Sit-CPs are referentially transparent; Cont-CPs are opaque. Paper §1.1.1.1 ex. 5 vs ex. 6: under a Sit-CP, embedded expressions evaluate at the actual world (substitution preserves truth). Under a Cont-CP, they evaluate at content-related worlds (substitution may change truth value).

  3. Transparent Syntax-Semantics mapping thesis (§1.1.2, ch. 4): bare CPs are always modifiers (PM via the situation argument path); nominalized CPs are always arguments (FA via the DP argument path). The syntactic structure of the CP determines the composition path. The (Sit-CP, via DP argument) cell of Table 1.1 is unattested by semantic triviality; the (bare, FA-argument) and (nominalized, PM-modifier) cells are ruled out by the structural correlation.

    This is the thesis @cite{angelopoulos-2026} attacks (autonomy of syntax: same syntactic position can yield either composition mode at LF). The conditional refutation theorem in Phenomena.Complementation.Studies.Angelopoulos2026 cites the transparentSSMapping definition below as its premise.

Out of scope #

Bondarenko's typology of noun sorts that combine with embedded CPs. Studies-local enum per the typological-discipline rule: the substrate (ContentIndividual, SituationIndividual) lives in Theories/Semantics/Attitudes/ClauseDenotation/; the typology- level tagging stays here.

Cross-linguistic robustness varies (English the situation vs the case; Korean sanghwang; Russian slučaj / situacija). This enum follows the paper's primary distinctions (§2.2).

  • content : NominalSort

    Content nouns: claim, belief, rumor, idea, thought, hypothesis, non-eventive fact.

  • situation : NominalSort

    Situation nouns: situation, case, circumstance, event, eventive uses of fact.

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

      Existential semantics for Cont-CP (paper eq. 15). CONT(x) ∩ p ≠ ∅ — content of x is compatible with p.

      Distinct from compC (equality, paper eq. 7) and ContentIndividual.entails (subset, paper eq. 10). The thesis argues against this candidate (§1.1.1.2).

      Equations
      Instances For

        Equality (compC) is strictly stronger than subset (ContentIndividual.entails) per paper §1.1.1.2. Re-export of ContentIndividual.eq_implies_entails under the paper-aligned name.

        Equality (compC) implies existential, provided the content is nonempty. The empty-content witness in ContentIndividual shows this is a real proviso.

        Equality forces functionality: if x has content p1 under equality semantics AND x has content p2 under equality semantics, then p1 = p2.

        Bassi & Bondarenko 2021 derive the impossibility of true CP conjunction from this functionality (paper §1.1.1.2).

        Counterpart for subset semantics (ContentIndividual.entails): subset does NOT force functionality. A non-vacuous content individual can entail multiple distinct propositions (any superset of the content) without being identified with them.

        Witness: a content individual whose CONT holds at exactly {true} (non-empty); it entails both (· = true) and (fun _ => True) — these are distinct propositions yet both satisfy subset semantics.

        Weak transparency at evaluation world s: substitution of predicates that agree AT s preserves clause truth. This is the notion that distinguishes Sit-CPs (transparentAt) from Cont-CPs (NOT transparentAt).

        The everywhere-quantified variant (∀ s', p s' ↔ q s') is funext+propext-trivial for any clause and carries no empirical content; only At is the substantive notion.

        Equations
        Instances For

          A Sit-CP (modeled as fun p s' => p s') IS weak-transparent: its evaluation is at the actual situation s, so coextensional-at-s predicates yield the same value.

          compC is operator-level intensional (NOT weak-transparent): there exist two propositions p, q coextensional at the evaluation world such that compC p xc and compC q xc differ. Constructive witness via W := Bool:

          • xc.cont := (fun b => b = false) (true only at false)
          • p := (fun b => b = false) — agrees with xc.cont
          • q := (fun _ => True) — disagrees at b = true

          At evaluation world false, p false ↔ q false (both true); but compC p xc HOLDS (= xc.cont) while compC q xc FAILS (xc.cont differs from q at true).

          Caveat (per syntax expert): this is the propositional- operator shadow of Bondarenko's claim, not the full DP-level de dicto reading. Bondarenko §2.2.3 ex. 120 is specifically about substitution failure of two DPs (this woman / the queen) coextensional at the evaluation world but diverging across content-related worlds. The DP-level reading requires a binding-theoretic encoding of de dicto vs de re (Moltmann 2020/2021 actuality-condition apparatus, Elliott 2020 CONT-as-content-restrictor) that is NOT yet a linglib substrate primitive. The theorem below establishes the operator-level intensionality that underwrites the DP-level de dicto reading; the DP-level theorem itself is queued.

          The four logically possible (clause shape, composition path) combinations.

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

              Bondarenko's transparent Syntax-Semantics mapping thesis (paper §1.1.2 + ch. 4): a bare CP composes ONLY via the modifier path (PM via the situation argument); a nominalized CP composes ONLY via the argument path (FA via the DP argument).

              The two off-diagonal cells (bareArgument, nominalizedModifier) are predicted impossible by this thesis.

              @cite{angelopoulos-2026} attacks the bareArgument cell with Greek oti/pu: bare clauses pattern as internal arguments (clitic doubling, passivization, extraction transparency) while still composing via PM (the explanans reading).

              Caveat (mathlib audit). This def-as-table encoding is the paper-fidelity stub form — it lossily compresses Bondarenko's distinction between situation-argument and DP-argument composition paths into a single bareArgument cell. A genuinely structural derivation would replace this with a predicate composes : ClauseShape → ArgumentSlot → Prop defined over actual derivations and prove the four-cell pattern from a deeper claim (e.g., that bare CPs lack the nominal projection required to saturate a DP argument slot). The current encoding is honest as a paper-fidelity tag but should not be mistaken for a derivation. TODO: structural refactor when a composes predicate over SyntacticObject derivations is available.

              Equations
              Instances For

                Bondarenko predicts the (bare, argument) cell is empty. This is the EXACT cell @cite{angelopoulos-2026} attacks. The proof is by table-unfolding; the substantive refutation lives in Phenomena.Complementation.Studies.Angelopoulos2026.

                The two composition paths Bondarenko distinguishes (paper §1.1.2, Table 1.1): clauses can either modify a verbal situation argument by Predicate Modification (viaSituation) or saturate a DP argument slot by Functional Application (viaDPArgument).

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

                    Paper Table 1.1: combinations of CP meanings × composition paths. Three of four cells are attested; the (Sit-CP, viaSituation) cell is unattested by semantic triviality (a Sit-CP composing intersectively with the situation argument always yields a trivially false statement, as a situation cannot be both a minimal exemplar of φ and the verb's situation argument).

                    Equations
                    Instances For

                      Bondarenko & Elliott 2026 inherit the equality semantics from this dissertation. The substantive bridge requires referencing a specific equality-using definition from Phenomena/Attitudes/Studies/BondarenkoElliott2026.lean (their MSI/MSO/TECM apparatus presupposes CONT(x) = ⟦S⟧ rather than CONT(x) ⊆ ⟦S⟧). The full bridge is queued: this file does not yet import BE2026, and the cross-file Iff target needs to be a specific BE2026 lemma rather than the compC substrate primitive. TODO: add bridge once BE2026 exposes a stable equalityForm lemma.

                      Semantic types relevant for Bondarenko's type-mismatch derivation. The two-element enum reflects only the cells the chapter exploits: individual (saturated entity, type ⟨e⟩) and predicate (unsaturated, type ⟨e,t⟩). Finer type-theoretic distinctions (intensional types ⟨s,e⟩, generalized quantifiers ⟨⟨e,t⟩,t⟩, etc.) are out of scope for this Studies file.

                      Instances For
                        @[implicit_reducible]
                        Equations
                        Equations
                        • One or more equations did not get rendered due to their size.
                        Instances For

                          Embedded clause types per Bondarenko §4.2 (the bare-vs-nominalized cut). Names emphasize the semantic content (predicate of individuals vs predicate of situations) rather than the morphological exponence (bare vs nominalized) — this is the BPS- compatible framing.

                          • predicateOfIndividuals : ClauseType

                            Nominalized CP: type ⟨e⟩, refers to an individual (typically with propositional content via CONT). Buryat -marked clause with a participial nominaliser; Korean -nun + kes clause.

                          • predicateOfSituations : ClauseType

                            Bare CP: type ⟨e,t⟩, predicate over (minimal) situations. Buryat clause without ; Korean -ta declarative; Russian bare čto-clause.

                          Instances For
                            @[implicit_reducible]
                            Equations
                            Equations
                            • One or more equations did not get rendered due to their size.
                            Instances For

                              Θ-heads relevant to Bondarenko §4.4. Per syntax expert S3, these are NOT three independent stipulations — they share the selectional restriction "requires individual argument." Per-Θ presupposition flavor (Θ_About has pre-existence; §4.4.3) is deferred to a future scope.

                              • causer : ThetaHead

                                Θ_Causer (Pylkkänen / Cuervo). §4.4.1.

                              • theme : ThetaHead

                                Θ_Theme (broad, internal-argument introducer). §4.4.2.

                              • aboutAttitude : ThetaHead

                                Θ_About (Bondarenko's own innovation, Pylkkänen-style refinement; carries pre-existence presupposition not formalised here). §4.4.3.

                              Instances For
                                @[implicit_reducible]
                                Equations
                                Equations
                                • One or more equations did not get rendered due to their size.
                                Instances For

                                  Every Θ-head requires an individual argument (type ⟨e⟩). This is the central claim that drives the bare-vs-nominalized derivation; it does NOT vary by Θ-head, despite per-Θ presupposition variation (§4.4.3 Θ_About pre-existence).

                                  Equations
                                  Instances For

                                    A clause type saturates a Θ-head iff its semantic type matches the head's required type. Per Bondarenko §4.2, this is the central type-mismatch derivation. NOT a stipulation: the Prop is just type-equality on the projection through semanticType / requiredType.

                                    Equations
                                    Instances For

                                      Bondarenko §4.4.1: Bare CPs cannot be Causers. Derived from type-mismatch: bare CP delivers .predicate; Θ_Causer requires .individual.

                                      Bondarenko §4.4.2: Bare CPs cannot be Theme-arguments. Same mechanism as Causer. The explain-class data (@cite{halpert-schueler-2013}, @cite{elliott-2016}, @cite{roelofsen-uegaki-2021}, @cite{pietroski-2000}) is what makes this derivation tight: when explain takes a bare clause vs a nominalized clause, only the latter receives the Theme reading.

                                      Bondarenko §4.4.3: Bare CPs cannot be About-arguments. Same type-mismatch; per Bondarenko this Θ-head additionally introduces a pre-existence presupposition (deferred).

                                      The unified Bondarenko §4.4 result: bare CPs satisfy NO Θ-head. This is the single mechanism behind §4.4.1, §4.4.2, §4.4.3 — not three independent stipulations (per syntax expert S3).

                                      Conversely, nominalized CPs satisfy every Θ-head (modulo per-Θ presupposition flavor not formalized here).

                                      A clause can compose via Predicate Modification with a verbal situation argument iff it has predicate type (⟨s,t⟩ after lifting the individual variable in Bondarenko's lift; substantively, both arguments are situation-predicates and intersect by PM). The predicateModification substrate at Theories/Semantics/Composition/Modification.lean is the typed realisation; this Studies-level predicate is its qualitative projection.

                                      Equations
                                      Instances For
                                        @[implicit_reducible]
                                        Equations
                                        • One or more equations did not get rendered due to their size.

                                        Bondarenko §4.5: bare CPs are verbal modifiers. Stated as PM-compatibility, derived from the type.

                                        Nominalized CPs are NOT modifiers via PM (Bondarenko §4.5 contrapositive — they compose via FA through the DP path).

                                        The cell value of the original transparentSSMapping table at a given path, derived from the type-theoretic predicates.

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

                                          The original transparentSSMapping (§5, def-as-table) agrees with the type-theoretic derivation cell-by-cell. This is the bridge mathlib audit demanded: the four-cell pattern is now derived rather than stipulated.

                                          Cross-linguistic overt exponents of ContP per @cite{bondarenko-2022}'s analysis (§4.3.1 Buryat, §4.3.2 Korean) extended with Tigrinya kemzi (@cite{cacchioli-2025}).

                                          Equations
                                          Instances For

                                            A nominal sort is occurrence-compatible iff it can combine with 'occur' / 'happen' (Russian proizojti, slučitsja; Korean ilena) — paper §2.2.3 ex. 108-112. The footnote-28 consultant comment ("alachay just does not sound good with 'claim'") is a weaker rejection than False; the binary encoding here aggregates with the main paradigm.

                                            Equations
                                            Instances For

                                              @cite{bondarenko-2022} §2.3 ex. 150 Cont head denotation IS the existing compC substrate. The denotational identity is one-line — the substantive Layered-Grounding fact is that Bondarenko's Chapter-2 proposal does NOT introduce new machinery; it instantiates the Kratzer/Moulton CONT-equality apparatus already in linglib.

                                              def Phenomena.Complementation.Studies.Bondarenko2022.compHead {S : Type u_1} [Preorder S] (p : SProp) (x evalSit : S) :

                                              @cite{bondarenko-2022} §2.3 ex. 151 Comp head denotation, consuming the Truthmaker substrate (Theories/Semantics/Truthmaker/Basic.lean attHolds-style parthood s ≤ σ x, Theories/Semantics/Truthmaker/Inexact.lean inexactVer for the part-existential variant) and situation parthood (Core/Logic/Intensional/Situations.lean ).

                                              compHead p x evalSit holds iff x is a situation that is part of the evaluation situation AND x exactly verifies p (verifier-membership; distinguished from inexact exemplification where existential closure over parts is allowed).

                                              Type signature follows Bondarenko's ⟨e,t⟩ shape with S as the situation-typed-individual sort: p : S → Prop is the propositional argument, x : S is the candidate verifier, evalSit : S is the evaluation situation.

                                              Equations
                                              Instances For
                                                theorem Phenomena.Complementation.Studies.Bondarenko2022.compHead_implies_inexactVerifier {S : Type u_1} [Preorder S] (p : SProp) (x evalSit : S) (h : compHead p x evalSit) :
                                                (u : S), u evalSit p u

                                                Exact exemplification implies inexact exemplification (a verifier is a part of itself). Substrate connection to Theories/Semantics/Truthmaker/Inexact.lean's inexactVer_of_exact.