Adamson & Zompì (2025): Polite Pronouns and the PCC

@cite{adamson-zompi-2025}

Polite Pronouns and the PCC. Linguistic Inquiry, Early Access.

Summary

The Person-Case Constraint (PCC) bans certain person combinations in ditransitive clitic clusters. All Italian speakers reject 3>1 and 3>2 IO>DO clitic combinations (Weak PCC); some additionally reject 1>2 and 2>1 (Strong PCC). PCC effects are restricted to clitic clusters — stressed/tonic pronouns are always licit (§2, (5)).

The polite pronoun LEI is formally third person — it triggers 3sg verbal agreement (§3, (8)), patterns with 3sg.f clitics in ordering (§3, (11c)), binds 3rd person reflexives (§3, (10)), and triggers obligatory feminine participle agreement (§3, (14)). Yet LEI triggers PCC effects like a second person pronoun: *Glie La hanno affidata (3.DAT LEI.ACC) is ungrammatical (§4.1, (17)), paralleling *Glie ti hanno affidata (3.DAT 2.ACC), not the licit Glie la hanno affidata (3.DAT 3.F.ACC).

Three independent lines of evidence converge:

1. PCC (§4.1): LEI in DO position triggers PCC effects like 2nd person
2. Fancy Constraint (§4.2): LEI triggers person hierarchy effects in faire infinitif causatives like 2nd person
3. Resolved agreement (§4.3, (30)): LEI in coordination triggers 2PL resolved agreement, unlike imposters which trigger 3PL ((31)-(32))

This falsifies morphosyntactic accounts (@cite{deal-2024}, @cite{coon-keine-2021}, @cite{bejar-rezac-2009}), which predict LEI should behave like 3rd person for PCC purposes. The data supports a syntacticosemantic account such as @cite{pancheva-zubizarreta-2018}, where the PCC reads interpretable person features.

The same pattern obtains cross-linguistically: Spanish USTED (§6.1) and German SIE (§6.2) also trigger PCC effects despite 3rd person agreement. Imposters (e.g., Vostro Onore) do NOT trigger PCC effects (§4.3), confirming the effect depends on syntacticosemantic person features, not addressee reference per se.

structure Phenomena.Agreement.Studies.AdamsonZompi2025.DualPersonFeatures : Type

A nominal's person features split into two layers.

@cite{adamson-zompi-2025} argue that polite pronouns carry two distinct sets of φ-features (following @cite{smith-2017}, @cite{anagnostopoulou-2017a}, among others):

• agreementPerson (uninterpretable): governs verbal agreement, clitic allomorphy, reflexive binding, clitic ordering, participle agreement. For LEI, this is 3rd person.
• interpretablePerson (interpretable): governs the PCC, Fancy Constraint, resolved agreement in coordination. For LEI, this is 2nd person.

For ordinary (non-polite) pronouns, both layers coincide.

• agreementPerson : Core.Prominence.PersonLevel
• interpretablePerson : Core.Prominence.PersonLevel

def Phenomena.Agreement.Studies.AdamsonZompi2025.instDecidableEqDualPersonFeatures.decEq (x✝ x✝¹ : DualPersonFeatures) : Decidable (x✝ = x✝¹)

Equations

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

instance Phenomena.Agreement.Studies.AdamsonZompi2025.instDecidableEqDualPersonFeatures : DecidableEq DualPersonFeatures

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.instDecidableEqDualPersonFeatures = Phenomena.Agreement.Studies.AdamsonZompi2025.instDecidableEqDualPersonFeatures.decEq

def Phenomena.Agreement.Studies.AdamsonZompi2025.instBEqDualPersonFeatures.beq : DualPersonFeatures → DualPersonFeatures → Bool

Equations

• One or more equations did not get rendered due to their size.
• Phenomena.Agreement.Studies.AdamsonZompi2025.instBEqDualPersonFeatures.beq x✝¹ x✝ = false

instance Phenomena.Agreement.Studies.AdamsonZompi2025.instBEqDualPersonFeatures : BEq DualPersonFeatures

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.instBEqDualPersonFeatures = { beq := Phenomena.Agreement.Studies.AdamsonZompi2025.instBEqDualPersonFeatures.beq }

instance Phenomena.Agreement.Studies.AdamsonZompi2025.instReprDualPersonFeatures : Repr DualPersonFeatures

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.instReprDualPersonFeatures = { reprPrec := Phenomena.Agreement.Studies.AdamsonZompi2025.instReprDualPersonFeatures.repr }

def Phenomena.Agreement.Studies.AdamsonZompi2025.instReprDualPersonFeatures.repr : DualPersonFeatures → ℕ → Std.Format

Equations

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

def Phenomena.Agreement.Studies.AdamsonZompi2025.DualPersonFeatures.ordinary (p : Core.Prominence.PersonLevel) : DualPersonFeatures

Ordinary pronoun: both layers are the same person.

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.DualPersonFeatures.ordinary p = { agreementPerson := p, interpretablePerson := p }

def Phenomena.Agreement.Studies.AdamsonZompi2025.pro1 : DualPersonFeatures

io — 1sg: agreement 1st, interpretable 1st.

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.pro1 = Phenomena.Agreement.Studies.AdamsonZompi2025.DualPersonFeatures.ordinary Core.Prominence.PersonLevel.first

def Phenomena.Agreement.Studies.AdamsonZompi2025.pro2 : DualPersonFeatures

tu — 2sg familiar: agreement 2nd, interpretable 2nd.

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.pro2 = Phenomena.Agreement.Studies.AdamsonZompi2025.DualPersonFeatures.ordinary Core.Prominence.PersonLevel.second

def Phenomena.Agreement.Studies.AdamsonZompi2025.pro3 : DualPersonFeatures

lui/lei — 3sg: agreement 3rd, interpretable 3rd.

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.pro3 = Phenomena.Agreement.Studies.AdamsonZompi2025.DualPersonFeatures.ordinary Core.Prominence.PersonLevel.third

def Phenomena.Agreement.Studies.AdamsonZompi2025.lei : DualPersonFeatures

Lei (polite) — agreement 3rd, interpretable 2nd.

Morphological evidence for 3rd person agreement features (§3):

• Morphologically identical to 3sg.f pronominal series (Table 1)
• Triggers 3sg verbal agreement, not 2sg ((7)–(8))
• Patterns with 3sg.f clitics in ordering: follows locative, like 3sg.f la and unlike 2sg ti which precedes locative ((11))
• Binds 3rd person reflexive si, not 2nd person ti ((9)–(10))
• Triggers obligatory feminine participle agreement matching formal (not conceptual) gender, like 3sg.f accusative clitics ((13)–(14))

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.lei = { agreementPerson := Core.Prominence.PersonLevel.third, interpretablePerson := Core.Prominence.PersonLevel.second }

def Phenomena.Agreement.Studies.AdamsonZompi2025.imposter : DualPersonFeatures

Imposters (e.g., Vostro Onore 'Your Honor', il signor Duca): agreement 3rd, interpretable 3rd.

Like LEI, imposters are grammatically 3rd person and are used to refer to addressees. Unlike LEI, imposters do not trigger PCC effects (§4.3, (27)–(29)), Fancy Constraint effects ((28)–(29)), or 2nd person resolved agreement in coordination ((31)–(32)). Their relationship to addressee reference is pragmatic, not encoded in interpretable person features.

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.imposter = Phenomena.Agreement.Studies.AdamsonZompi2025.DualPersonFeatures.ordinary Core.Prominence.PersonLevel.third

def Phenomena.Agreement.Studies.AdamsonZompi2025.weakPCC (ioPerson doPerson : Core.Prominence.PersonLevel) : Bool

The Weak PCC: a 3rd person IO clitic cannot co-occur with a 1st or 2nd person DO clitic.

All Italian speakers have at least the Weak PCC (§2, p. 4–5). The constraint bans 3>1 and 3>2 but allows 1>2, 2>1, and all combinations where IO is 1st/2nd or DO is 3rd.

Note: the person values here are the values the PCC reads — the central question of the paper is whether these are agreement person or interpretable person.

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.weakPCC ioPerson doPerson = !(ioPerson == Core.Prominence.PersonLevel.third && doPerson.isSAP)

def Phenomena.Agreement.Studies.AdamsonZompi2025.strongPCC (_ioPerson doPerson : Core.Prominence.PersonLevel) : Bool

The Strong PCC: in a ditransitive clitic cluster, the DO must be 3rd person. Some Italian speakers have the Strong PCC (§2, p. 5), which additionally bans 1>2 and 2>1.

The Strong PCC entails the Weak PCC: anything banned under the Weak PCC is also banned under the Strong PCC.

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.strongPCC _ioPerson doPerson = (doPerson == Core.Prominence.PersonLevel.third)

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.strong_entails_weak (io do_ : Core.Prominence.PersonLevel) : strongPCC io do_ = true → weakPCC io do_ = true

Strong PCC entails Weak PCC.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.weak_3_3 : weakPCC Core.Prominence.PersonLevel.third Core.Prominence.PersonLevel.third = true

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.weak_2_3 : weakPCC Core.Prominence.PersonLevel.second Core.Prominence.PersonLevel.third = true

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.weak_1_3 : weakPCC Core.Prominence.PersonLevel.first Core.Prominence.PersonLevel.third = true

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.weak_3_2 : weakPCC Core.Prominence.PersonLevel.third Core.Prominence.PersonLevel.second = false

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.weak_3_1 : weakPCC Core.Prominence.PersonLevel.third Core.Prominence.PersonLevel.first = false

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.weak_allows_participant_pairs : weakPCC Core.Prominence.PersonLevel.first Core.Prominence.PersonLevel.second = true ∧ weakPCC Core.Prominence.PersonLevel.second Core.Prominence.PersonLevel.first = true

Weak PCC allows 1>2 and 2>1 (unlike Strong PCC).

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.strong_3_2 : strongPCC Core.Prominence.PersonLevel.third Core.Prominence.PersonLevel.second = false

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.strong_1_2 : strongPCC Core.Prominence.PersonLevel.first Core.Prominence.PersonLevel.second = false

structure Phenomena.Agreement.Studies.AdamsonZompi2025.CliticJudgment : Type

Acceptability judgments for Italian ditransitive clitic clusters.

Person values are interpretable person — the paper's claim is that PCC effects track this layer, not agreement person. For non-polite pronouns the distinction is vacuous; for LEI it is crucial.

PCC effects are restricted to clitic clusters: when one argument is a stressed (tonic) pronoun, person restrictions vanish (§2, (5)):

• Gli hanno affidato te (3.DAT entrusted 2SG.STRESS) = ✓
• Ti hanno affidato a lui (2.ACC entrusted to 3SG.STRESS) = ✓ Similarly for LEI: Gli hanno affidato Lei (stressed) is licit (§4.1, (21)), and La hanno affidata a lui is licit ((21b)).

• label : String
• ok : Bool
• ioPerson : Core.Prominence.PersonLevel
• doPerson : Core.Prominence.PersonLevel

instance Phenomena.Agreement.Studies.AdamsonZompi2025.instReprCliticJudgment : Repr CliticJudgment

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.instReprCliticJudgment = { reprPrec := Phenomena.Agreement.Studies.AdamsonZompi2025.instReprCliticJudgment.repr }

def Phenomena.Agreement.Studies.AdamsonZompi2025.instReprCliticJudgment.repr : CliticJudgment → ℕ → Std.Format

Equations

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

def Phenomena.Agreement.Studies.AdamsonZompi2025.dat2_acc3 : CliticJudgment

(2a) Te la hanno affidata — 2.DAT > 3.F.ACC = ✓

Equations

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

def Phenomena.Agreement.Studies.AdamsonZompi2025.dat3_acc3 : CliticJudgment

(2b) Glie la hanno affidata — 3.DAT > 3.F.ACC = ✓

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.dat3_acc3 = { label := "3.DAT > 3.F.ACC", ok := true, ioPerson := Core.Prominence.PersonLevel.third, doPerson := Core.Prominence.PersonLevel.third }

def Phenomena.Agreement.Studies.AdamsonZompi2025.dat3_acc2 : CliticJudgment

(3a) *Gli(e)/Le ti/te hanno affidato/affidata — 3.DAT > 2.ACC = ✗

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.dat3_acc2 = { label := "3.DAT > 2.ACC", ok := false, ioPerson := Core.Prominence.PersonLevel.third, doPerson := Core.Prominence.PersonLevel.second }

def Phenomena.Agreement.Studies.AdamsonZompi2025.acc2_dat3 : CliticJudgment

(3b) *Ti/Te gli(e)/le hanno affidato/affidata — same combination, reversed clitic order, still ✗. PCC is about person combination, not linear order.

Equations

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

def Phenomena.Agreement.Studies.AdamsonZompi2025.leiDat_acc3 : CliticJudgment

(16) Glie la hanno affidata — LEI.DAT > 3.F.ACC = ✓ LEI as dative with 3rd person accusative: not a PCC violation because the interpretively-2nd-person argument is the IO, not the DO. (The clitic form glie is ambiguous between regular 3.DAT and LEI.DAT; both readings yield a grammatical sentence.)

Equations

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

def Phenomena.Agreement.Studies.AdamsonZompi2025.dat3_leiAcc : CliticJudgment

(17) *Glie La hanno affidata — 3.DAT > LEI.ACC = ✗ The paper's central data point. LEI as accusative clitic with a 3rd person dative triggers PCC, just like 2nd person accusative ti. Note: this is string-identical to the licit Glie la hanno affidata (3.DAT > 3.F.ACC), differing only in whether the accusative is polite LEI or ordinary 3sg.f ((17) fn. 9).

Equations

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

def Phenomena.Agreement.Studies.AdamsonZompi2025.leiDat_acc3_racc : CliticJudgment

(19a) Glie la hanno raccomandata — LEI.DAT > 3.F.ACC = ✓ Same pattern with raccomandare 'recommend'.

Equations

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

def Phenomena.Agreement.Studies.AdamsonZompi2025.dat3_leiAcc_racc : CliticJudgment

(19b) *Glie La hanno raccomandata — 3.DAT > LEI.ACC = ✗

Equations

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

def Phenomena.Agreement.Studies.AdamsonZompi2025.italianData : List CliticJudgment

Equations

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

def Phenomena.Agreement.Studies.AdamsonZompi2025.morphosyntacticPrediction (d : DualPersonFeatures) : Bool

Morphosyntactic prediction: the PCC reads agreement person.

Under morphosyntactic accounts (@cite{deal-2024}, @cite{coon-keine-2021}, @cite{bejar-rezac-2009}), LEI's agreement features (3rd person) determine PCC behavior. Since 3>3 is licit, 3.DAT > LEI.ACC should be licit.

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.morphosyntacticPrediction d = Phenomena.Agreement.Studies.AdamsonZompi2025.weakPCC Core.Prominence.PersonLevel.third d.agreementPerson

def Phenomena.Agreement.Studies.AdamsonZompi2025.syntacticosemanticPrediction (d : DualPersonFeatures) : Bool

Syntacticosemantic prediction: the PCC reads interpretable person.

Under a syntacticosemantic account (@cite{pancheva-zubizarreta-2018}), LEI's interpretable features (2nd person) determine PCC behavior. Since 3>2 is illicit, 3.DAT > LEI.ACC should be illicit.

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.syntacticosemanticPrediction d = Phenomena.Agreement.Studies.AdamsonZompi2025.weakPCC Core.Prominence.PersonLevel.third d.interpretablePerson

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.morphosyntactic_wrong_for_lei : morphosyntacticPrediction lei = true

Morphosyntactic account WRONGLY predicts 3.DAT > LEI.ACC is licit.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.syntacticosemantic_correct_for_lei : syntacticosemanticPrediction lei = false

Syntacticosemantic account CORRECTLY predicts 3.DAT > LEI.ACC is illicit.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.lei_prediction_matches_data : syntacticosemanticPrediction lei = dat3_leiAcc.ok

The syntacticosemantic prediction matches the actual judgment.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.lei_morphosyntactic_mismatch : morphosyntacticPrediction lei ≠ dat3_leiAcc.ok

The morphosyntactic prediction does NOT match the actual judgment.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.lei_banned_under_both_variants : weakPCC Core.Prominence.PersonLevel.third lei.interpretablePerson = false ∧ strongPCC Core.Prominence.PersonLevel.third lei.interpretablePerson = false

LEI triggers effects under BOTH PCC variants. The Strong PCC is strictly stronger, so if the Weak PCC bans a combination, the Strong PCC does too. The LEI data doesn't depend on which variant a speaker has.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.ordinary_accounts_agree (p : Core.Prominence.PersonLevel) : morphosyntacticPrediction (DualPersonFeatures.ordinary p) = syntacticosemanticPrediction (DualPersonFeatures.ordinary p)

For ordinary pronouns (no mismatch), both accounts agree.

def Phenomena.Agreement.Studies.AdamsonZompi2025.fancyConstraint (causeePerson : Core.Prominence.PersonLevel) : Bool

The Fancy Constraint (@cite{postal-1989}): in analytic causative constructions (faire infinitif), a person hierarchy effect obtains between the accusative clitic causee and 1st/2nd person arguments.

In Italian, the faire infinitif construction shows: a 3rd person accusative causee is licit ((24) Micol la fa pettinare a Carlo), but a 1st/2nd person causee triggers a hierarchy effect ((24) *Micol ti fa pettinare a Carlo). Crucially, this is NOT a ditransitive clitic cluster — it involves a causative verb — so it constitutes independent evidence for the same person-sensitivity.

LEI patterns with 2nd person, not 3rd:

• (25) *Signor Biagi, Micol La fa pettinare a Carlo = ✗ (LEI as causee)
• (26) Micol fa pettinare {te / Lei} a Carlo = ✓ (stressed: no effect)

We model the Fancy Constraint as reading the same interpretable person features as the PCC.

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.fancyConstraint causeePerson = (causeePerson == Core.Prominence.PersonLevel.third)

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.fancy_3f_ok : fancyConstraint Core.Prominence.PersonLevel.third = true

3sg.f causee is licit: Micol la fa pettinare a Carlo (24).

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.fancy_2_bad : fancyConstraint Core.Prominence.PersonLevel.second = false

2sg causee is illicit: *Micol ti fa pettinare a Carlo (24).

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.fancy_lei_bad : fancyConstraint lei.interpretablePerson = false

LEI causee: interpretable 2nd → illicit, like ti ((25)).

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.fancy_lei_morphosyntactic_wrong : fancyConstraint lei.agreementPerson = true

LEI causee: agreement 3rd → morphosyntactic account wrongly predicts licit, like la.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.pcc_fancy_converge : syntacticosemanticPrediction lei = false ∧ fancyConstraint lei.interpretablePerson = false

PCC and Fancy Constraint agree for LEI: both read interpretable person, both ban LEI.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.imposter_no_pcc_either_account : morphosyntacticPrediction imposter = true ∧ syntacticosemanticPrediction imposter = true

Both accounts predict no PCC effect for imposters, matching the data ((27)–(29)).

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.lei_imposter_contrast : syntacticosemanticPrediction lei ≠ syntacticosemanticPrediction imposter

The contrast between LEI and imposters: LEI triggers PCC, imposters don't. Both refer to addressees, but only LEI encodes this in interpretable person features.

inductive Phenomena.Agreement.Studies.AdamsonZompi2025.ResolvedNumber : Type

Resolved agreement in coordination (§4.3, (30)–(32)).

When LEI is coordinated with a 3sg nominal, resolved agreement is obligatorily 2PL ((30) Lei e l'ambasciatore... vi incontrerete), consistent with LEI's interpretable 2nd person features. When an imposter (Vostro Onore, il signor Duca) is coordinated with a 3sg nominal, resolved agreement is 3PL ((31)–(32)), consistent with interpretable 3rd person features.

This is a third line of evidence, independent of both the PCC and the Fancy Constraint, showing that LEI's interpretable person surfaces in the grammar.

• pl2 : ResolvedNumber
• pl3 : ResolvedNumber

instance Phenomena.Agreement.Studies.AdamsonZompi2025.instDecidableEqResolvedNumber : DecidableEq ResolvedNumber

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.instDecidableEqResolvedNumber x✝ y✝ = if h : x✝.ctorIdx = y✝.ctorIdx then isTrue ⋯ else isFalse ⋯

instance Phenomena.Agreement.Studies.AdamsonZompi2025.instBEqResolvedNumber : BEq ResolvedNumber

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.instBEqResolvedNumber = { beq := Phenomena.Agreement.Studies.AdamsonZompi2025.instBEqResolvedNumber.beq }

def Phenomena.Agreement.Studies.AdamsonZompi2025.instBEqResolvedNumber.beq : ResolvedNumber → ResolvedNumber → Bool

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.instBEqResolvedNumber.beq x✝ y✝ = (x✝.ctorIdx == y✝.ctorIdx)

instance Phenomena.Agreement.Studies.AdamsonZompi2025.instReprResolvedNumber : Repr ResolvedNumber

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.instReprResolvedNumber = { reprPrec := Phenomena.Agreement.Studies.AdamsonZompi2025.instReprResolvedNumber.repr }

def Phenomena.Agreement.Studies.AdamsonZompi2025.instReprResolvedNumber.repr : ResolvedNumber → ℕ → Std.Format

Equations

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

def Phenomena.Agreement.Studies.AdamsonZompi2025.resolvedAgreement (d : DualPersonFeatures) : ResolvedNumber

Equations

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

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.lei_resolved_2pl : resolvedAgreement lei = ResolvedNumber.pl2

LEI + 3sg → 2PL resolved agreement ((30)).

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.imposter_resolved_3pl : resolvedAgreement imposter = ResolvedNumber.pl3

Imposter + 3sg → 3PL resolved agreement ((31)–(32)).

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.three_lines_converge : syntacticosemanticPrediction lei = false ∧ syntacticosemanticPrediction imposter = true ∧ fancyConstraint lei.interpretablePerson = false ∧ fancyConstraint imposter.interpretablePerson = true ∧ resolvedAgreement lei = ResolvedNumber.pl2 ∧ resolvedAgreement imposter = ResolvedNumber.pl3

All three lines of evidence (PCC, Fancy Constraint, resolved agreement) converge: LEI patterns with 2nd person, imposters pattern with 3rd.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.lei_interpretable_is_participant : (Minimalism.decomposePerson lei.interpretablePerson).hasParticipant = true

The PCC's sensitivity to interpretable person is equivalent to sensitivity to [+participant] in @cite{preminger-2014}'s decomposition.

The Weak PCC bans 3rd person IO with [+participant] DO. LEI has interpretable [+participant] (2nd person = [+participant, −author]).

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.lei_agreement_not_participant : (Minimalism.decomposePerson lei.agreementPerson).hasParticipant = false

LEI's agreement features lack [participant] (3rd = [−participant]).

def Phenomena.Agreement.Studies.AdamsonZompi2025.pccViaParticipant (ioPerson doPerson : Core.Prominence.PersonLevel) : Bool

The Weak PCC reduces to: if IO is [−participant], then DO must be [−participant]. Equivalently: ¬(IO lacks [participant] ∧ DO has [participant]).

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.pccViaParticipant ioPerson doPerson = !(!(Minimalism.decomposePerson ioPerson).hasParticipant && (Minimalism.decomposePerson doPerson).hasParticipant)

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.pcc_participant_equivalence (io do_ : Core.Prominence.PersonLevel) : weakPCC io do_ = pccViaParticipant io do_

pccViaParticipant is extensionally equivalent to weakPCC.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.pconstraint_matches_weakPCC (io do_ : Core.Prominence.PersonLevel) : Minimalism.PConstraint.pccLicit Minimalism.PConstraint.weakGrammar io do_ = weakPCC io do_

The P-Constraint (@cite{pancheva-zubizarreta-2018}) derives PCC effects from point-of-view encoding. pccLicit from PConstraint.lean evaluates the constraint over interpretable person values.

The Weak PCC (P-Uniqueness inactive) produces the same judgments as our weakPCC function.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.pconstraint_matches_strongPCC (io do_ : Core.Prominence.PersonLevel) : Minimalism.PConstraint.pccLicit Minimalism.PConstraint.strongGrammar io do_ = strongPCC io do_

The Strong PCC (P-Uniqueness active) produces the same judgments as our strongPCC function.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.lei_is_proximate : Minimalism.PConstraint.isInherentlyProximate lei.interpretablePerson = true

LEI is inherently [+PROXIMATE] by virtue of interpretable 2nd person. This is what triggers PCC effects under P&Z's account.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.pconstraint_bans_lei : Minimalism.PConstraint.pccLicit Minimalism.PConstraint.weakGrammar Core.Prominence.PersonLevel.third lei.interpretablePerson = false

The P-Constraint correctly predicts LEI triggers PCC (Weak grammar).

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.pconstraint_allows_imposter : Minimalism.PConstraint.pccLicit Minimalism.PConstraint.weakGrammar Core.Prominence.PersonLevel.third imposter.interpretablePerson = true

The P-Constraint correctly predicts imposters do NOT trigger PCC.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.probe_finds_person_on_both : Minimalism.hasValuedFeature Phenomena.Agreement.Studies.AdamsonZompi2025.lei_agreement_bundle✝ (Minimalism.FeatureVal.phi (Minimalism.PhiFeature.person Core.Prominence.PersonLevel.third)) = true ∧ Minimalism.hasValuedFeature Phenomena.Agreement.Studies.AdamsonZompi2025.lei_interpretable_bundle✝ (Minimalism.FeatureVal.phi (Minimalism.PhiFeature.person Core.Prominence.PersonLevel.second)) = true

A probe seeking person features finds them on both bundles — the probe can Agree with either. The issue is WHICH person value it sees.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.morphosyntactic_probe_misses_participant : (Minimalism.decomposePerson Core.Prominence.PersonLevel.third).hasParticipant = false

Under morphosyntactic accounts, the probe sees LEI's agreement bundle and finds [Person:3rd] — no [participant] feature. The probe treats LEI as a non-participant, predicting no PCC effect.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.syntacticosemantic_finds_participant : (Minimalism.decomposePerson Core.Prominence.PersonLevel.second).hasParticipant = true

Under the syntacticosemantic account, the PCC reads LEI's interpretable bundle and finds [Person:2nd] — [+participant]. PCC effect predicted.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.deal_weak_wrong_for_lei : Deal2024.isLicit Deal2024.weak Core.Prominence.PersonLevel.third lei.agreementPerson = true ∧ Deal2024.isLicit Deal2024.weak Core.Prominence.PersonLevel.third lei.interpretablePerson = false

@cite{deal-2024}'s isLicit directly demonstrates the morphosyntactic failure. Under the Weak PCC (Italian), the probe reads agreement person:

• isLicit weak .third lei.agreementPerson = isLicit weak 3 3 = true (wrongly predicts ⟨3.DAT, LEI.ACC⟩ is licit)
• isLicit weak .third lei.interpretablePerson = isLicit weak 3 2 = false (correctly predicts illicit)

This is a formal, end-to-end test: @cite{deal-2024}'s own licitness function, applied to LEI's agreement features, gives the wrong answer.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.deal_strong_wrong_for_lei : Deal2024.isLicit Deal2024.strong Core.Prominence.PersonLevel.third lei.agreementPerson = true ∧ Deal2024.isLicit Deal2024.strong Core.Prominence.PersonLevel.third lei.interpretablePerson = false

The same failure obtains under the Strong PCC variant: Deal's model reads agreement-3P as ⟨3,3⟩ (licit) rather than interpretable-2P as ⟨3,2⟩ (illicit).

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.lei_matches_la_cl_person : lei.agreementPerson = Core.Prominence.PersonLevel.third ∧ Fragments.Italian.Pronouns.la_cl.person = UD.Person.third

LEI's agreement features match the 3sg.f clitic la in person.

The fragment entry la_cl has .third person, matching LEI's agreement layer — confirming the paper's morphological evidence that LEI is formally 3sg.f (§3, Table 1).

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.lei_differs_from_ti : lei.agreementPerson = Core.Prominence.PersonLevel.third ∧ Fragments.Italian.Pronouns.ti_acc.person = UD.Person.second

LEI's agreement person (3rd) differs from 2sg clitic ti (2nd).

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.lei_reflexive_is_3p : Fragments.Italian.Pronouns.si_refl.person = UD.Person.third ∧ Fragments.Italian.Pronouns.ti_refl.person = UD.Person.second

LEI binds 3rd person reflexive si, not 2nd person ti (§3, (9)–(10)).

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.fragment_lei_dual_person : Fragments.Italian.Pronouns.lei_formal.person = some UD.Person.third ∧ Fragments.Italian.Pronouns.lei_formal.referentialPerson = some Core.Prominence.PersonLevel.second

The fragment's lei_formal encodes dual person features: person = 3rd (agreement), referentialPerson = 2nd (interpretable). This directly mirrors our DualPersonFeatures structure.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.all_data_match_syntacticosemantic : (italianData.all fun (j : CliticJudgment) => weakPCC j.ioPerson j.doPerson == j.ok) = true

Every Italian judgment matches the Weak PCC evaluated over interpretable person. Since weakPCC checks both IO and DO person, this verifies the full constraint, not just DO.

def Phenomena.Agreement.Studies.AdamsonZompi2025.usted : DualPersonFeatures

Spanish USTED: agreement 3rd, interpretable 2nd.

Like Italian LEI, USTED triggers 3sg verbal agreement but refers to an addressee. @cite{rezac-2011} observes PCC effects with USTED: the accusative clitic la is grammatical in a 3>3 configuration if its referent is 3rd person, but ungrammatical if interpreted as polite USTED (§6.1, (43)). The paper's own consultants confirm the same contrast with encomendar 'entrust' ((44)).

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.usted = { agreementPerson := Core.Prominence.PersonLevel.third, interpretablePerson := Core.Prominence.PersonLevel.second }

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.usted_pcc : syntacticosemanticPrediction usted = false

USTED triggers PCC effects under the syntacticosemantic account.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.usted_morphosyntactic_wrong : morphosyntacticPrediction usted = true

Morphosyntactic account wrongly predicts no PCC effect.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.lei_usted_isomorphic : lei = usted

Italian LEI and Spanish USTED have the same dual-feature structure: both are agreement-3rd, interpretable-2nd.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.spanish_fragment_usted_dual : Fragments.Spanish.Pronouns.usted.person = some UD.Person.third ∧ Fragments.Spanish.Pronouns.usted.referentialPerson = some Core.Prominence.PersonLevel.second

The Spanish fragment's usted entry encodes the same dual-person structure, grounding the study's USTED in fragment data.

def Phenomena.Agreement.Studies.AdamsonZompi2025.sie : DualPersonFeatures

German polite SIE: agreement 3rd (plural), interpretable 2nd.

SIE triggers 3pl verbal agreement ((45)) and binds 3rd person reflexive sich (not 2sg dich or 2pl euch). In the limited PCC environments available in German (Wackernagel clusters), SIE patterns with 2nd person in triggering person hierarchy effects ((47)–(48)).

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.sie = { agreementPerson := Core.Prominence.PersonLevel.third, interpretablePerson := Core.Prominence.PersonLevel.second }

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.sie_pcc : syntacticosemanticPrediction sie = false

SIE triggers PCC effects under the syntacticosemantic account.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.german_fragment_sie_dual : Fragments.German.Pronouns.sie_polite.person = some UD.Person.third ∧ Fragments.German.Pronouns.sie_polite.referentialPerson = some Core.Prominence.PersonLevel.second

The German fragment's sie_polite entry encodes the same dual-person structure, grounding the study's SIE in fragment data.

def Phenomena.Agreement.Studies.AdamsonZompi2025.assumedIdentityEffect (formalPerson : Core.Prominence.PersonLevel) (copulaForm3rd : Bool) : Bool

German assumed-identity copular constructions (§6.2, (49)–(53)) exhibit a DIFFERENT person hierarchy effect — one ameliorated by syncretism of verbal forms and therefore attributed to exponence/morphology, not to the syntax-semantics interface.

The paper's key modularity argument: PCC effects (syntacticosemantic source) are sensitive to interpretable person, so polite pronouns trigger them. Assumed-identity effects (exponence-based source) are sensitive to formal features, so polite pronouns do NOT trigger them.

Equations

• Phenomena.Agreement.Studies.AdamsonZompi2025.assumedIdentityEffect formalPerson copulaForm3rd = (formalPerson != Core.Prominence.PersonLevel.third && !copulaForm3rd)

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.sie_no_assumed_identity : assumedIdentityEffect sie.agreementPerson false = false

SIE does NOT trigger assumed-identity effects (agreement 3rd).

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.sie_triggers_pcc : syntacticosemanticPrediction sie = false

SIE DOES trigger PCC effects (interpretable 2nd).

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.modularity_contrast : syntacticosemanticPrediction sie = false ∧ assumedIdentityEffect sie.agreementPerson false = false

The modularity contrast: PCC reads interpretable person (SIE triggers), exponence reads agreement person (SIE doesn't trigger).

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.pcc_politeness_prediction (pol : DualPersonFeatures) (h_int : pol.interpretablePerson = Core.Prominence.PersonLevel.second) : syntacticosemanticPrediction pol = false

The paper's cross-linguistic prediction (40):

"If a language displays PCC effects in ditransitives for second person arguments and has a third person addressee-referring polite pronoun, this pronoun should also give rise to PCC effects."

Formalized: for any pronoun with interpretable 2nd person, the syntacticosemantic account predicts a PCC effect in DO position.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.morphosyntactic_misses_polite (pol : DualPersonFeatures) (h_agr : pol.agreementPerson = Core.Prominence.PersonLevel.third) : morphosyntacticPrediction pol = true

The morphosyntactic account wrongly predicts NO PCC effect for any polite pronoun with 3rd person agreement features.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.all_polite_pronouns_predicted : syntacticosemanticPrediction lei = false ∧ syntacticosemanticPrediction usted = false ∧ syntacticosemanticPrediction sie = false

All three cross-linguistic polite pronouns (LEI, USTED, SIE) are correctly predicted by the syntacticosemantic account.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.all_fragments_grounded : Fragments.Italian.Pronouns.lei_formal.referentialPerson = some Core.Prominence.PersonLevel.second ∧ Fragments.Spanish.Pronouns.usted.referentialPerson = some Core.Prominence.PersonLevel.second ∧ Fragments.German.Pronouns.sie_polite.referentialPerson = some Core.Prominence.PersonLevel.second

All three fragment entries encode dual person features, and the study's DualPersonFeatures values match them.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.person_geometry_matches_core_features : Core.Person.Category.s1.toFeatures.hasParticipant = (Minimalism.decomposePerson Core.Prominence.PersonLevel.first).hasParticipant ∧ Core.Person.Category.s2.toFeatures.hasParticipant = (Minimalism.decomposePerson Core.Prominence.PersonLevel.second).hasParticipant ∧ Core.Person.Category.s3.toFeatures.hasParticipant = (Minimalism.decomposePerson Core.Prominence.PersonLevel.third).hasParticipant

The [±participant] decomposition in Core/Person/Category.lean (operating on Category) is the same decomposition as PersonGeometry.decomposePerson (operating on PersonLevel).

This theorem bridges the two: for all singular Categories, toFeatures.hasParticipant equals decomposePerson.hasParticipant.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.cross_linguistic_number : Fragments.Italian.Pronouns.lei_formal.number = some Number.sg ∧ Fragments.Spanish.Pronouns.usted.number = some Number.sg ∧ Fragments.German.Pronouns.sie_polite.number = some Number.pl

Cross-linguistic number validation: the fragment entries record the agreement number of each polite pronoun, matching §3 (Italian), §6.1 (Spanish), §6.2 (German).

Italian LEI and Spanish USTED trigger 3sg agreement. German SIE triggers 3pl agreement — the key typological outlier (§6.2, (45)). SIE's formal plurality is why it doubles as a 2PL form and why it can address multiple addressees, unlike LEI/USTED.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.number_irrelevant_to_pcc : Fragments.Italian.Pronouns.lei_formal.number ≠ Fragments.German.Pronouns.sie_polite.number ∧ syntacticosemanticPrediction lei = syntacticosemanticPrediction sie

Despite different agreement numbers, all three polite pronouns trigger PCC effects identically — confirming that the PCC reads person features (specifically interpretable person), not number.

theorem Phenomena.Agreement.Studies.AdamsonZompi2025.mismatch_is_crucial (d : DualPersonFeatures) (h : d.agreementPerson = d.interpretablePerson) : morphosyntacticPrediction d = syntacticosemanticPrediction d

Only mismatch pronouns distinguish the two accounts. For any pronoun where agreement and interpretable person coincide, both accounts make the same prediction. Polite pronouns are the crucial test case.