Kampanarou & Alexiadou (2026): Genitive alternation in possessives and beyond #
@cite{kampanarou-alexiadou-2026} @cite{alexiadou-2003} @cite{michelioudakis-chatzikyriakidis-spathas-2024} @cite{mertyris-2014} @cite{mertyris-2023} @cite{sims-2006} @cite{alexiadou-stavrou-2020} @cite{kampanarou-2023} @cite{horrocks-stavrou-1987} @cite{holton-mackridge-philippaki-warburton-spyropoulos-2012}
K&A's central claim: in Standard Modern Greek (SMG), the alternation between inflectional genitive and apo-PP is NOT a morphological alternation but a structural one — the two constructions are introduced via distinct syntactic mechanisms. The inflectional genitive is broad-coverage; the apo-PP is restricted to part-whole and source readings, and its felicity is gated by a partitive-coercion semantics that requires the possessor to be construable as a SET.
Structure of this file #
- §1 Empirical taxonomy of apo-PP licensing, factored as (relation × possessor-animacy × set-construability), per K&A §5.
- §2 Paradigm gaps in
-akidiminutives (K&A §3, citing @cite{sims-2006} and @cite{mertyris-2014}; also @cite{alexiadou-2024}). - §4 Scope diagnostic for inalienable vs alienable vs apo-PP (K&A §7,
exx 38–39); reuses
Morphology.DM.PossessionType. - §5 Three syntactic analyses (eq 41 N-selects-PP, eq 43 Pred-SC, eq 47 light-p), each tied to substrate primitives; convergent no-stacking prediction.
- §6 Single Argument Restriction (K&A §8 strengthening of the familiar
Single Genitive Restriction); imports
Wood2023.NominalizationReadingfor the derived-nominal vs result-nominal distinction. - §7 Cross-framework theorems against Myler 2016, AissenPolian 2025, Michelioudakis et al. 2024, plus stubs against Alexiadou & Stavrou 2020, Cardinaletti & Giusti 2006, Barker 1998.
- §8 Diachronic note + negative theorem on Heine 1997 typology coverage.
Out of scope #
- Greek noun-class declension table (Ralli 2000 / Mertyris 2014 Tables 1–2): no second consumer; defer.
- Distinctness substrate (@cite{horrocks-stavrou-1987} / Richards 2010): no current Lean consumer; we use a local predicate with TODO.
- Diachronic substrate covering inflection-to-adposition trajectories (negation theorem in §8 documents the gap).
- Promoting
Wood2023.NominalizationReadingto a substrate file: separate refactor, awaits 3rd consumer.
Per K&A pp. 5–7: the apo-PP's felicity depends on (a) the relation type between possessor and possessee, (b) the animacy of the possessor, and (c) whether the possessee is a body-part. Body parts are formally part-whole but block apo-PPs when the possessor is animate (5c).
- notion : Typology.Possession.PossessiveNotion
- possessorAnimate : Bool
- possesseeIsBodyPart : Bool
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Per K&A §5: the apo-PP is partitive-coerced — felicitous when its complement nominal can be construed as a SET. Modification, plural marking, and common-noun status enable set-construal; proper names and unmodified animate singulars cannot. This factoring is K&A's actual analysis (footnote 8 + ex (28)), not a flat list of features.
- isPlural : Bool
- isModified : Bool
- isProperName : Bool
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Set-construability per K&A §5: enabled by plural OR modification, BLOCKED by proper-name status (which K&A note resists set-denotation; pp. 19).
Equations
Instances For
K&A's apo-PP licensing prediction: felicitous when (a) the relation is partitive-friendly (inanimate part-whole or inanimate source-like) OR (b) the possessor can be construed as a set, coercing partitivity.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Ex (5a) "the door's handle / the handle of the door": part-whole with inanimate possessor, no modification needed. apo licensed.
Ex (6a) "#brother of girls": kinship with animate possessor — not partitive-friendly, and bare common nouns don't auto-coerce. apo blocked.
Ex (5c) "?body part of animate possessor": body-parts are formally part-whole but K&A flag them as degraded with animate possessors.
Ex (28) "father of the quiet kid": kinship with animate possessor BUT modified — coerces a set-of-kids reading. apo licensed. K&A's footnote 8 diagnostic: the modifier creates a contrastive set.
Ex (11b) "of the doors" (plural): plural marking enables set-construal even when the bare-singular variant degrades.
Four-way acceptability marker. K&A use * (ungrammatical), ??
(marginal), # (pragmatically anomalous), ? (variable). Flattening
these to a Boolean would erase K&A's empirical fidelity.
- ok : Acceptability
- variable : Acceptability
- marginal : Acceptability
- pragmatic : Acceptability
- ungrammatical : Acceptability
Instances For
Equations
- Phenomena.Possession.Studies.KampanarouAlexiadou2026.instDecidableEqAcceptability 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
A paradigm-gap data point. The genitive form is starred (no inflectional genitive available); the apo-PP variant takes one of K&A's four judgments, conditioned on the relation type.
- noun : String
- glossInflectionalGen : String
- apoVariant : String
- inflectionalAcc : Acceptability
- apoAcc : Acceptability
- apoRelation : Typology.Possession.PossessiveNotion
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
K&A exx (14)–(15): -aki diminutives in part-whole context allow apo-PP despite the genitive gap; in ownership/kinship context, the gap is NOT repaired by apo.
Equations
- One or more equations did not get rendered due to their size.
Instances For
K&A's central paradigm-gap finding: a paradigm gap is NOT sufficient
to license apo-PP across all relation types; only part-whole readings
rescue the genitive function. Stated as a List.all-style structural
sentry over akiGapData.
Quantifier scope reading inside a possessor DP. The diagnostic discriminates by ABSENCE of surface scope under alienable possessors; inalienable possessors and apo-PPs both license both readings.
- surface : ScopeReading
- inverse : ScopeReading
Instances For
Equations
- Phenomena.Possession.Studies.KampanarouAlexiadou2026.instDecidableEqScopeReading 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
The three possessor structures K&A test (pp. 25). Per K&A §7 + Alexiadou 2003, the inalienable structure has the possessor as complement (low), alienable in Spec,PossP (high). apo-PPs pattern with inalienable.
- inalienableGen : PossessorStructure
- alienableGen : PossessorStructure
- apoPP : PossessorStructure
Instances For
Equations
- Phenomena.Possession.Studies.KampanarouAlexiadou2026.instDecidableEqPossessorStructure 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
K&A exx (38)–(39): scope licensing per possessor structure. The KEY discriminator is that alienable genitives BLOCK surface scope; inalienable genitives and apo-PPs both license both readings. The payoff (per K&A pp. 25 ¶3): apo-PPs merging low (as complements, structurally analogous to inalienable possessors) explains their unrestricted scope.
Equations
- One or more equations did not get rendered due to their size.
- Phenomena.Possession.Studies.KampanarouAlexiadou2026.licensesScope Phenomena.Possession.Studies.KampanarouAlexiadou2026.PossessorStructure.inalienableGen x✝ = true
- Phenomena.Possession.Studies.KampanarouAlexiadou2026.licensesScope Phenomena.Possession.Studies.KampanarouAlexiadou2026.PossessorStructure.apoPP x✝ = true
Instances For
K&A's central scope finding: alienable genitives uniquely block surface scope (ex 38b "a toy of each kid" — only inverse).
The diagnostic theorem: apo-PP scope behavior matches inalienable genitive, NOT alienable. K&A pp. 25 ¶3: this supports analyzing apo-PPs as low-merged complements (eq. 41) rather than high specifiers.
Bridge to the substrate's PossessionType: inalienable genitive ↔
PossessionType.inalienable (Spec,nP per Alexiadou 2003 / Myler 2016
refinement; cf. K&A footnote on Alexiadou 2003's complement-of-NP
primary version).
Equations
- One or more equations did not get rendered due to their size.
- Phenomena.Possession.Studies.KampanarouAlexiadou2026.fromPossessionType Morphology.DM.PossessionType.alienable = Phenomena.Possession.Studies.KampanarouAlexiadou2026.PossessorStructure.alienableGen
Instances For
The substrate's PossessionType.canAffectGender (which is true exactly
for inalienable per the GLH) coincides with the structures that license
surface scope under K&A's diagnostic.
The three analyses K&A consider for SMG apo-PPs. Per K&A pp. 30–31, all three converge on the empirical predictions; K&A prefer light-p (eq 47, citing @cite{kampanarou-2023}, inspired by Svenonius 2010 and Kratzer 1996 Voice).
- nSelects : ApoSyntacticAnalysis
eq. 41: possessee N₂ selects the apo-PP as its complement directly. Structural primitive: head-complement selection.
- predSC : ApoSyntacticAnalysis
eq. 43: small clause with Pred head; apo-PP in complement of Pred. Structural primitive:
Minimalist.SmallClausewithpredCat = .P. - lightP : ApoSyntacticAnalysis
eq. 47: functional light-p relator selects the apo-PP, with possessee as external argument. Structural primitive: parallel to
Minimalist.ApplType.lowSource(low source-of-possession Appl).
Instances For
Equations
- Phenomena.Possession.Studies.KampanarouAlexiadou2026.instDecidableEqApoSyntacticAnalysis 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
The substrate primitive for the Pred-SC analysis (eq. 43). NOT
necessarily den Dikken 1995 — K&A do not cite den Dikken 1995 here;
they use "Pred" generically. SCPredCategory.P is the appropriate
SC predicate-category since apo is a P.
Equations
Instances For
The substrate primitive for the light-p analysis (eq. 47).
@cite{kampanarou-2023} positions the light-p as the nominal counterpart
of Kratzer's Voice, with apo as the spell-out of the relator head. The
closest verbal-domain analog already in linglib is
ApplType.lowSource — the low Appl with possessionFrom denotation.
Equations
Instances For
K&A's preferred analysis (p. 29 ¶4: "The benefit of this analysis is..."; p. 31 ¶1: "Although the latter analysis seems to account better for the facts").
Equations
Instances For
All three analyses converge on the no-stacking prediction (K&A pp. 17–18, exx 25–27). The structural source differs per analysis:
nSelects: complement uniqueness (each head selects ≤ 1 complement)predSC: SC complement-of-Pred uniquenesslightP: relator-p selects ≤ 1 PP complement All three trace back to the principle that selectional complement is unique.
Equations
Instances For
Bridge theorem against Pylkkanen2008.Applicative substrate: K&A's
light-p relator (eq. 47) IS Pylkkänen's lowSource specialized to the
DP-internal level. Both denote possessionFrom; they are sibling
realizations of the same theoretical move (low functional relator with
from-semantics). Stating this prevents silent divergence: a future reader
loading both files would otherwise see two parallel "light-p with
from-semantics" types.
K&A's strengthening of the Single Genitive Restriction (familiar from @cite{horrocks-stavrou-1987} et seq.) into a Single Argument Restriction: Greek DP allows at most ONE genitive-OR-PP argument, since both compete for the same single argument-introducing slot (per K&A's three analyses). English is a counterexample (allows ʼs-genitive + of-genitive co-occurrence).
Equations
- Phenomena.Possession.Studies.KampanarouAlexiadou2026.singleArgRestriction genCount apoCount = decide (genCount + apoCount ≤ 1)
Instances For
English allows multiple genitives in the DP (s-gen + of-gen), so does NOT obey the SAR. K&A p. 32 cites this as a parametric difference.
Per K&A §6 (pp. 21–24), apo-PPs in derived nominals come in two sorts:
- by-phrase counterparts (external-argument introduction; complex event nominals, eq 31 with Voice/v adjunct) — distinct apo, distinct syntax
- apo-PPs alternating with theme genitives — only on result nominals
(per Grimshaw 1990's CEN/RN distinction, formalized as
Wood2023.NominalizationReading).
Equations
- Phenomena.Possession.Studies.KampanarouAlexiadou2026.licensesApoAlternation Morphology.DM.Allosemy.NominalizationReading.complexEvent = false
- Phenomena.Possession.Studies.KampanarouAlexiadou2026.licensesApoAlternation Morphology.DM.Allosemy.NominalizationReading.simpleEvent = true
- Phenomena.Possession.Studies.KampanarouAlexiadou2026.licensesApoAlternation Morphology.DM.Allosemy.NominalizationReading.result = true
- Phenomena.Possession.Studies.KampanarouAlexiadou2026.licensesApoAlternation Morphology.DM.Allosemy.NominalizationReading.simpleState = true
- Phenomena.Possession.Studies.KampanarouAlexiadou2026.licensesApoAlternation Morphology.DM.Allosemy.NominalizationReading.simpleEntity = true
- Phenomena.Possession.Studies.KampanarouAlexiadou2026.licensesApoAlternation Morphology.DM.Allosemy.NominalizationReading.content = true
Instances For
K&A §6 prediction: complex event nominals (CENs) are uniquely
incompatible with theme apo-PPs (the theme must be inflectional
genitive). The diagnostic is aspectual modifiers (for x time, ex 34)
and pluralisation (ex 36).
Result nominals license theme apo-PPs (K&A §6, exx 32-33: 'sense of the chocolate', 'cutting of the meat').
Cross-framework theorem 1 (vs @cite{myler-2016}). K&A §5 explicitly
rejects a realizational/VI-style account for SMG (apo-PP is NOT an
alternative spell-out of inflectional genitive). Myler's Icelandic
hafa/eiga IS a VI-style alternation (Myler2016.icelandicHaveVI
bidirectionally conditions on PP-internal possessor). The structural
asymmetry: Icelandic is realizational, SMG is not.
Cross-framework theorem 2 (vs @cite{aissen-polian-2025}). Both
papers commit to NominalSize.nP as a structural locus for inalienable
possessors. A&P predict A-extraction is available only from
non-specific (PossP/nP) possessors — ExtractionAvailable .stranding .nP
is true under their architecture. K&A §7 (with Safir 1987 and
Angelopoulos & Michelioudakis 2023) say SMG apo-PPs (which are DP-internal
complements per all three analyses) NEVER extract or front. The two
architectures make opposite predictions about the same nP slot.
Cross-framework theorem 3 (vs @cite{michelioudakis-chatzikyriakidis-spathas-2024}).
Michelioudakis et al. analyse Grevena Greek (GG) apo-PPs as reduced
relative clauses adjoining within the DP — like Romance de/di. K&A §4
show this analysis CANNOT extend to SMG: SMG apo-PPs cannot stack, cannot
front, cannot sub-extract. The Lean-checkable contrast is the dialect
profiles' adnominalStrategy mismatch and the empirical-distribution
asymmetry encoded in the Fragment files.
Stub theorem (vs @cite{alexiadou-stavrou-2020}). A&S 2020 treat apo as a LEXICAL preposition in partitive contexts. K&A §5 reject this for the possessive uses (apo is FUNCTIONAL/light-p per their preferred analysis). Empirical handle (K&A pp. 20): apo functions as the complement of another preposition/adverb (45a) and does not license clitics (45b) — diagnostics for non-lexical status. TODO: full proof requires lexical-vs-functional substrate; deferred.
Stub theorem (vs Cardinaletti & Giusti 2006, cited in K&A footnote 15 p. 28). C&G 2006: quantifier directly selects the PP — no underlying small clause. K&A entertain this as compatible with their analysis (p. 28 ftn 15) but do not adopt it.
Stub theorem (vs @cite{barker-1995} double-genitives). Barker 1998 (NLLT 16:679–717, cited K&A p. 28 fn 15) treats partitives as a distinct nominal type with anti-uniqueness presupposition. K&A §5: possessive apo-PPs are coerced THROUGH partitivity, suggesting the K&A-Barker direction (possession ← partitive) opposite the more common partitive ← possession direction.
Per @cite{mertyris-2023} + K&A §3: the partitive use of inflectional genitive was lost early in Greek diachrony, with apo already carrying the partitive load by Classical times (ex 1, oligoi apo pollo:n 'few of many'). The K&A claim is that the modern possessive apo-PP traces back via partitive-coercion to this earlier partitive apo.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Negative theorem against @cite{heine-1997}: Heine's PossessionSource
enum (8 schemas) is for grammaticalisation paths to PREDICATIVE
possession. K&A's case-loss-to-adposition trajectory for ADNOMINAL
possession has no slot in Heine's typology — .source is for
"from-possessor" predicative constructions, not for the inflection→PP
reanalysis K&A document. This is a substrate gap; the honest move is
to negate Heine fit rather than misuse .source.
TODO: extend Diachronic substrate when 2nd consumer materializes
(Mertyris 2023 + K&A 2026 = current candidates).
Local Distinctness predicate per @cite{horrocks-stavrou-1987} + Richards
2010 framing in K&A §8. The Single Argument Restriction follows from
Distinctness applied to the [+arg] feature within the DP.
TODO: promote to Theories/Syntax/Minimalist/Distinctness.lean when a
second Lean consumer (likely Wood 2023 Icelandic) materializes. The
pattern matches the deferred-substrate convention used elsewhere
(e.g., SC particles, Indefinite paradigm).
Equations
- Phenomena.Possession.Studies.KampanarouAlexiadou2026.distinctnessLocal sameTypeArgCount = decide (sameTypeArgCount ≤ 1)