Angelopoulos 2026: On clausal complementation, once more #
Greek oti- and pu-clauses present three puzzles (§1): near- complementary distribution after verbs (oti with saying/belief, pu with emotive factives, ex. 1); freedom as internal arguments and derived subjects but a ban from external-argument position (§2.2); and a novel stativity restriction on complement pu-clauses (§2.3).
The analysis reverses selection (§3.1): oti and pu bear an uninterpretable [n]-feature checked by a light noun merged in their specifier (partly adopting Arsenijević 2009; the paper is neutral on the categorial status of oti and pu, fn. 3). The noun must incorporate into a lexical verbal head — possible from complement position, impossible from Spec,vP (nearest head T) or under P — which derives the argument asymmetry and the P-ban (§3.1 ex. 27–32). The oti ~ pu distribution follows from the content/situation dichotomy (§3.2, adopting [Bon22]); the stativity restriction from aspectual-head selection (§4.1: vState selects both otiP and puP, vEvent only otiP). §7.1 extends the adjunct-selection account to Uyghur dep (= de 'say' + converb -ip, per Major 2024) — the structural parallel of Buryat gɘ-žɘ.
§7.3 departs from [Bon22]'s transparent syntax–semantics mapping: bare oti-clauses are merged in COMPLEMENT position (the §2 argumenthood diagnostics: clitic doubling, passivization, derived subjects) while composing via Predicate Modification (the explanans reading, [Ell20a]) — the same syntactic position yields either composition mode.
Main declarations #
bearsN— the §3.1 [n]-feature datum over the fragment entriesNounHost,ClausePosition,licensedIn— the incorporation mechanism and the derived argument-position asymmetryselectsClause,pu_requires_stative— the §4.1 stativity locusclauseSort— oti = content, pu = situation, consumingBondarenko2022.NominalSort(§3.2 ex. 33–34)bareOtiAttested,transparency_conflates_axes— the §7.3 counterclaim againstBondarenko2022.transparentSSMapping
Typed paradigm sentences (ex. 1, 31–34) live in Angelopoulos2026.Examples,
generated from Data/Examples/Angelopoulos2026.json.
Reversed selection: the [n]-feature (§3.1) #
oti and pu bear an uninterpretable [n]-feature checked by a light noun merged in their specifier (§3.1); na does not (its licensing is mood-driven). Paper-specific datum projected over the fragment entries; the paper is neutral on the category of oti and pu (fn. 3).
Equations
Instances For
Equations
- Angelopoulos2026.instDecidablePredComplementizerBearsN c = id inferInstance
Incorporation licensing and the argument asymmetry (§3.1) #
Equations
- Angelopoulos2026.instDecidableEqNounHost x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- One or more equations did not get rendered due to their size.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- Angelopoulos2026.instReprNounHost = { reprPrec := Angelopoulos2026.instReprNounHost.repr }
Whether a host licenses light-noun incorporation.
Equations
Instances For
Positions a bare oti/pu-clause can occupy, each with the nearest potential incorporation host: internal arguments sit under an aspectual v; incorporation precedes movement for derived subjects; the nearest head above Spec,vP is T; P cannot host (§3.1 ex. 27–32).
- internalArgument : ClausePosition
- derivedSubject : ClausePosition
- externalArgument : ClausePosition
- pComplement : ClausePosition
Instances For
Equations
- Angelopoulos2026.instDecidableEqClausePosition x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- One or more equations did not get rendered due to their size.
Equations
Equations
- One or more equations did not get rendered due to their size.
Instances For
The nearest potential incorporation host from each position.
Equations
- Angelopoulos2026.ClausePosition.internalArgument.nearestHost = Angelopoulos2026.NounHost.vLex
- Angelopoulos2026.ClausePosition.derivedSubject.nearestHost = Angelopoulos2026.NounHost.vLex
- Angelopoulos2026.ClausePosition.externalArgument.nearestHost = Angelopoulos2026.NounHost.t
- Angelopoulos2026.ClausePosition.pComplement.nearestHost = Angelopoulos2026.NounHost.p
Instances For
A bare oti/pu-clause is licensed in a position iff the nearest host licenses light-noun incorporation — the paper's derivation of the distribution, not a stipulated table.
Equations
- Angelopoulos2026.licensedIn pos = pos.nearestHost.licenses
Instances For
Internal arguments and derived subjects are licensed (§2.1–2.2).
The external-argument ban (§2.2): T cannot host incorporation.
Bare clauses are excluded after P (ex. 31c, 32c).
The stativity locus (§4.1) #
Aspectual heads introducing internal arguments (§4.1, following Borer and Merchant as cited there).
- vState : AspectualHead
- vEvent : AspectualHead
Instances For
Equations
- Angelopoulos2026.instDecidableEqAspectualHead x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- One or more equations did not get rendered due to their size.
Equations
Equations
- One or more equations did not get rendered due to their size.
Instances For
§4.1: vState selects both otiP and puP as its complement; vEvent selects only otiP.
Equations
Instances For
The stativity restriction (§2.3), derived: a pu-complement forces the stative aspectual head.
The verb-level reflex over the fragment sample: the pu-only matrix verbs (ex. 1b, 13–14, 20) are all stative.
Content vs situation (§3.2) #
The sort of clause each complementizer introduces — oti content,
pu situation — which must match the incorporating noun's sort
(§3.2). The sorts and their diagnostics ('true'/'mistaken' vs
'happen', ex. 33–34) are [Bon22]'s (Bondarenko2022.NominalSort,
§2.2.3); na is outside the dichotomy.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The assigned sorts pass the §3.2 diagnostics: oti's sort is truth-evaluable, pu's occurrence-compatible (ex. 33–34, matching [Bon22] §2.2.3).
Against the transparent syntax–semantics mapping (§7.3) #
Syntactic position of an embedded clause — one of the two axes
[Bon22]'s ClauseStructurePath conflates.
- complement : SynPosition
- adjunct : SynPosition
Instances For
Equations
- Angelopoulos2026.instDecidableEqSynPosition x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
Composition mode — the other axis.
Instances For
Equations
- Angelopoulos2026.instDecidableEqCompMode x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯
Equations
- Angelopoulos2026.instReprCompMode = { reprPrec := Angelopoulos2026.instReprCompMode.repr }
Equations
- One or more equations did not get rendered due to their size.
Instances For
The paper's claim for bare oti-clauses (§2 diagnostics + §7.3): complement position composing via PM is attested (the explanans reading); FA requires the nominalizing D (§7.3 ex. 57), so bare clauses never compose via FA from either position.
Equations
- Angelopoulos2026.bareOtiAttested Angelopoulos2026.SynPosition.complement Angelopoulos2026.CompMode.pm = True
- Angelopoulos2026.bareOtiAttested Angelopoulos2026.SynPosition.complement Angelopoulos2026.CompMode.fa = False
- Angelopoulos2026.bareOtiAttested Angelopoulos2026.SynPosition.adjunct Angelopoulos2026.CompMode.pm = True
- Angelopoulos2026.bareOtiAttested Angelopoulos2026.SynPosition.adjunct Angelopoulos2026.CompMode.fa = False
Instances For
[Bon22]'s transparent mapping restated on the split axes: a bare clause composing via PM must be a syntactic adjunct (the composition path is reflected in syntax).
Equations
Instances For
The §7.3 divergence: bare oti-clauses occupy complement position while composing via PM — attested here, excluded by the transparent mapping. Conditional on the paper's premises: the clauses are BARE (no covert nominal shell — the analysis rejects Arsenijević's DP layer and Faure's case-less-DP treatment, §3.1) and are internal ARGUMENTS (clitic doubling, passivization, derived subjects, §2.1–2.2). [Bon22] can deny either premise.
Against Bondarenko2022.transparentSSMapping directly: Bondarenko
predicts the (bare, argument) cell empty
(Bondarenko2022.bare_argument_predicted_impossible), identifying
argument position with the FA path; Greek realizes syntactic
argumenthood for bare clauses WITHOUT FA — the identification of the
two axes is what fails.