Pearson (2015): The interpretation of the logophoric pronoun in Ewe

[Pea15] [Sel87] [PS03] [CS14] [Chi90] [Kra09b] [Lew79a] [Alo01]

[Pea15] (Nat Lang Semantics 23(2)) gives the definitive modern semantics of the Ewe logophoric pronoun yè — the carrier ye : LogophoricPronoun (Syntax/Pronoun/Logophoric.lean, requiredRole = .self). The traditional Heim & von Stechow view (after [Chi90]) is that yè bears an [log] feature that must be bound by the individual abstractor an attitude verb introduces in the embedded left periphery — so yè obligatorily occurs under an attitude predicate and takes the attitude holder as antecedent (this is the .self requirement: Sells' antecedent-must-be-a-self). That view predicts yè is obligatorily de se.

[Pea15]'s fieldwork finding is that this prediction is wrong: yè is de se / de re ambiguous. Her account keeps yè bound by the attitude abstractor (preserving the distribution) but lets it additionally sit inside a resP — a covert constituent housing a concept generator variable ([PS03], [CS14]) — yielding the de re reading. A concept generator maps a res to an individual concept; it is suitable when reliable (returns the res in the actual world, with a variable over the holder's epistemic alternatives overwritten with the holder) and acquaintance-based (the concept is one of the holder's available ways of identifying — a member of her conceptual cover, [Alo01]).

This file

• Concept / ConceptGenerator — over Intensional.Intension; a Concept is an element of an Acquaintance.Cover, so acquaintance is membership in the holder's cover.
• sayDeSe (eq. 76), sayDeRe/claimDeRe (eq. 77/79b) — the two denotations of say, with Reliable (eq. 82.i, reliability-with-epistemic-overwrite) and Suitable (eq. 82 = reliable ∧ acquaintance-based-via-cover).
• Scenario 1 (eq. 75, §5.3): John says "whoever wrote this is clever", unaware he is the author. ye_de_se_de_re_ambiguous proves the de se LF is false while the de re LF is true — via an explicit witnessing "author" concept generator. [Pea15]'s central finding, made true by construction.
• Scenario 2 (Napoleon, §6, eq. 80–85): John believes he is Napoleon and, not recognising himself on TV, says the patient is delusional. napoleon_contrast proves "John claims he is delusional" true but "John claims Napoleon is delusional" false — the bound pronoun (overwritten to John, whom John is acquainted with) succeeds where the name (the actual-world Napoleon, to whom John bears no acquaintance, so no cover concept reliably picks him) fails.
• ye_antecedent_is_attitude_holder grounds the carrier: yè's antecedent is the attitude holder in both readings, i.e. the carrier's requiredRole = .self.

Deferred to prose: the yè-vs-PRO φ-feature asymmetry (§7.1: PRO is a φ-less [Kra09b] minimal pronoun, so — unlike yè — it takes no long-distance antecedent).

Concept generators ([PS03], [CS14])

@[reducible, inline]

abbrev Pearson2015.Centered (W : Type u_1) (E : Type u_2) : Type (max u_1 u_2)

A (Lewis) centered attitude alternative: a world paired with the individual the attitude holder identifies as herself there ([Lew79a]).

Equations

• Pearson2015.Centered W E = (W × E)

@[reducible, inline]

abbrev Pearson2015.Concept (W : Type u_1) (E : Type u_2) : Type (max (max u_2 u_1) u_2)

An individual concept à la [PS03]: a function from centered worlds to individuals (Intensional.Intension over Centered). This is exactly an element of an Acquaintance.Cover (Centered W E) E.

Equations

• Pearson2015.Concept W E = Intensional.Intension (Pearson2015.Centered W E) E

@[reducible, inline]

abbrev Pearson2015.ConceptGenerator (W : Type u_1) (E : Type u_2) : Type (max u_2 u_1)

A concept generator: from a res to an individual concept.

Equations

• Pearson2015.ConceptGenerator W E = (E → Pearson2015.Concept W E)

@[reducible, inline]

abbrev Pearson2015.CProp (W : Type u_1) (E : Type u_2) : Type (max u_1 u_2)

A centered property (type ⟨e,⟨s,t⟩⟩): holds of an individual at a world.

Equations

• Pearson2015.CProp W E = (E → W → Prop)

def Pearson2015.epiAlt {W : Type u_1} {E : Type u_2} (alts : List (Centered W E)) : List E

The epistemic alternatives of the attitude holder = the de se centers of her attitude alternatives.

Equations

• Pearson2015.epiAlt alts = List.map Prod.snd alts

say de se and de re ([Pea15] eq. 76, 77, 82)

def Pearson2015.sayDeSe {W : Type u_1} {E : Type u_2} (alts : List (Centered W E)) (P : CProp W E) : Prop

⟦say^de se⟧ (eq. 76): the embedded property holds of the de se center at each attitude alternative. yè bound directly by the attitude abstractor.

Equations

• Pearson2015.sayDeSe alts P = ∀ p ∈ alts, P p.2 p.1

def Pearson2015.Reliable {W : Type u_1} {E : Type u_2} [DecidableEq E] (alts : List (Centered W E)) (G : ConceptGenerator W E) (x : E) (w : W) (dom : List E) : Prop

A concept generator is reliable for holder x in w ([Pea15] eq. 82.i, the "Reliability" clause; over a finite res domain dom): for each res u the concept returns u in the actual world, or u is an epistemic alternative of x and the concept returns x (the epistemic-alternative overwrite).

Equations

• Pearson2015.Reliable alts G x w dom = ∀ u ∈ dom, G u (w, x) = u ∨ u ∈ Pearson2015.epiAlt alts ∧ G u (w, x) = x

def Pearson2015.Suitable {W : Type u_1} {E : Type u_2} [DecidableEq E] (cover : Semantics.Reference.Acquaintance.Cover (Centered W E) E) (alts : List (Centered W E)) (G : ConceptGenerator W E) (x : E) (w : W) (dom : List E) : Prop

A concept generator is suitable for x in w ([Pea15] eq. 82): it is Reliable and acquaintance-based — each concept it produces (for a res in dom) is one of the holder's available ways of identifying, i.e. a member of her conceptual cover ([Alo01]; Acquaintance.Cover). The cover is what makes suitability non-trivial: a generator whose concepts lie outside the holder's cover is unsuitable, even if reliable.

Equations

• Pearson2015.Suitable cover alts G x w dom = (Pearson2015.Reliable alts G x w dom ∧ ∀ u ∈ dom, G u ∈ cover)

def Pearson2015.sayDeRe {W : Type u_1} {E : Type u_2} [DecidableEq E] (cover : Semantics.Reference.Acquaintance.Cover (Centered W E) E) (alts : List (Centered W E)) (P : CProp W E) (x : E) (w : W) (dom : List E) : Prop

⟦say^de re⟧ for a pronoun (yè) res (eq. 77/79b): there is a Suitable concept generator G such that at each attitude alternative ⟨w',y⟩, the individual G's concept (fed the de se center y as res) picks out at ⟨w',y⟩ has the embedded property at w'. yè sits in a resP, bound by the attitude abstractor.

Equations

• Pearson2015.sayDeRe cover alts P x w dom = ∃ (G : Pearson2015.ConceptGenerator W E), Pearson2015.Suitable cover alts G x w dom ∧ ∀ p ∈ alts, P (G p.2 p) p.1

def Pearson2015.claimDeRe {W : Type u_1} {E : Type u_2} [DecidableEq E] (cover : Semantics.Reference.Acquaintance.Cover (Centered W E) E) (alts : List (Centered W E)) (P : CProp W E) (res x : E) (w : W) (dom : List E) : Prop

⟦say^de re⟧ for a name res (eq. 79): like sayDeRe, but the res fed to G is the fixed individual res (a rigid, actual-world-bound name), not the de se center. The difference from sayDeRe is exactly the pronoun/name asymmetry that delivers §6.

Equations

• Pearson2015.claimDeRe cover alts P res x w dom = ∃ (G : Pearson2015.ConceptGenerator W E), Pearson2015.Suitable cover alts G x w dom ∧ ∀ p ∈ alts, P (G res p) p.1

Scenario 1: the de se / de re ambiguity ([Pea15] eq. 75, §5.3)

John has found an old paper he wrote but does not recognise as his own; impressed, he says "Whoever wrote this is clever." John be yè le cleva ('John said that yè was clever') is judged true — yet John never self-ascribes cleverness.

@[reducible, inline]

abbrev Pearson2015.Wld : Type

Worlds: actual (0) and bel (1), John's say-alternative world.

Equations

• Pearson2015.Wld = Fin 2

@[reducible, inline]

abbrev Pearson2015.Ind : Type

Individuals: john (0) and auth (1), the author John takes to be someone else.

Equations

• Pearson2015.Ind = Fin 2

def Pearson2015.actual : Wld

Equations

• Pearson2015.actual = 0

def Pearson2015.bel : Wld

Equations

• Pearson2015.bel = 1

def Pearson2015.john : Ind

Equations

• Pearson2015.john = 0

def Pearson2015.auth : Ind

Equations

• Pearson2015.auth = 1

def Pearson2015.sayAlts : List (Centered Wld Ind)

John's single say-alternative: world bel, de se center john (himself qua speaker).

Equations

• Pearson2015.sayAlts = [(Pearson2015.bel, Pearson2015.john)]

def Pearson2015.dom : List Ind

The relevant res domain: just the attitude holder.

Equations

• Pearson2015.dom = [Pearson2015.john]

def Pearson2015.cleverB : Wld → Ind → Bool

clever in the belief world bel: the author auth is clever, John is not — John ascribes cleverness to the author, failing to recognise the author as himself.

Equations

• Pearson2015.cleverB 1 1 = true
• Pearson2015.cleverB x✝¹ x✝ = false

@[reducible, inline]

abbrev Pearson2015.cleverP : CProp Wld Ind

Reducible so decide sees the underlying Bool test through the wrapper.

Equations

• Pearson2015.cleverP y w = (Pearson2015.cleverB w y = true)

def Pearson2015.authorConcept : Concept Wld Ind

The "author of the paper" concept: returns the res john in the actual world (John really is the author) but the believed author auth in John's belief world.

Equations

• Pearson2015.authorConcept p = if p.1 = Pearson2015.actual then Pearson2015.john else Pearson2015.auth

def Pearson2015.authorGen : ConceptGenerator Wld Ind

The generator carrying the "author" concept for any res.

Equations

• Pearson2015.authorGen _u = Pearson2015.authorConcept

def Pearson2015.s1Cover : Semantics.Reference.Acquaintance.Cover (Centered Wld Ind) Ind

John's conceptual cover: the "author of the paper" concept he is acquainted via.

Equations

• Pearson2015.s1Cover = {Pearson2015.authorConcept}

The ambiguity, derived

theorem Pearson2015.deSe_false : ¬sayDeSe sayAlts cleverP

The de se reading is false: at John's say-alternative ⟨bel, john⟩ the de se centre (john) is not clever — John did not self-ascribe cleverness.

theorem Pearson2015.authorGen_suitable : Suitable s1Cover sayAlts authorGen john actual dom

The "author" generator is suitable for John: reliable (returns the res john in the actual world) and acquaintance-based (its concept is in John's cover).

theorem Pearson2015.deRe_true : sayDeRe s1Cover sayAlts cleverP john actual dom

The de re reading is true: under the suitable "author" generator, at each say-alternative the individual it picks (the author) is clever — the de re reading rescues truth where de se fails.

theorem Pearson2015.ye_de_se_de_re_ambiguous : ¬sayDeSe sayAlts cleverP ∧ sayDeRe s1Cover sayAlts cleverP john actual dom

yè is de se / de re ambiguous ([Pea15]'s central finding): the same sentence is false on the de se LF (78) but true on the de re LF (79). The Heim & von Stechow prediction of obligatory de se is thereby refuted, by construction.

Scenario 2: the Napoleon contrast ([Pea15] §6, eq. 80–85)

John is delusional and believes he is Napoleon; watching a TV report he does not recognise himself, he says the patient he saw is delusional. "John claims he is delusional" (the pronoun yè, de re) is true; "John claims Napoleon is delusional" is false — though John believes he is Napoleon. A variable bound by the attitude verb (yè) ranges over John's epistemic alternatives and is overwritten with John, whom John identifies via the in-cover "patient on TV" concept; the name Napoleon denotes the actual-world individual, to whom John bears no acquaintance — no concept in his cover reliably picks Napoleon.

@[reducible, inline]

abbrev Pearson2015.Napoleon.Wld : Type

Worlds: actualN (0) and belN (1), John's claim-alternative world.

Equations

• Pearson2015.Napoleon.Wld = Fin 2

@[reducible, inline]

abbrev Pearson2015.Napoleon.Ind : Type

Individuals: john (0), the patient John sees on TV (1), and napoleon (2).

Equations

• Pearson2015.Napoleon.Ind = Fin 3

def Pearson2015.Napoleon.actualN : Wld

Equations

• Pearson2015.Napoleon.actualN = 0

def Pearson2015.Napoleon.belN : Wld

Equations

• Pearson2015.Napoleon.belN = 1

def Pearson2015.Napoleon.john : Ind

Equations

• Pearson2015.Napoleon.john = 0

def Pearson2015.Napoleon.patient : Ind

Equations

• Pearson2015.Napoleon.patient = 1

def Pearson2015.Napoleon.napoleon : Ind

Equations

• Pearson2015.Napoleon.napoleon = 2

def Pearson2015.Napoleon.claimAlts : List (Centered Wld Ind)

John's claim-alternatives: world belN, de se center john.

Equations

• Pearson2015.Napoleon.claimAlts = [(Pearson2015.Napoleon.belN, Pearson2015.Napoleon.john)]

def Pearson2015.Napoleon.delusionalB : Wld → Ind → Bool

delusional in belN: the patient John saw on TV is delusional.

Equations

• Pearson2015.Napoleon.delusionalB 1 1 = true
• Pearson2015.Napoleon.delusionalB x✝¹ x✝ = false

@[reducible, inline]

abbrev Pearson2015.Napoleon.delusionalP : CProp Wld Ind

Equations

• Pearson2015.Napoleon.delusionalP y w = (Pearson2015.Napoleon.delusionalB w y = true)

def Pearson2015.Napoleon.patientConcept : Concept Wld Ind

The "patient I saw on TV" concept: the actual world → john (reliably himself), the belief world → patient (whom John takes the man on TV to be).

Equations

• Pearson2015.Napoleon.patientConcept p = if p.1 = Pearson2015.Napoleon.actualN then Pearson2015.Napoleon.john else Pearson2015.Napoleon.patient

def Pearson2015.Napoleon.patientGen : ConceptGenerator Wld Ind

The generator carrying the "patient" concept for any res.

Equations

• Pearson2015.Napoleon.patientGen _u = Pearson2015.Napoleon.patientConcept

def Pearson2015.Napoleon.johnCover : Semantics.Reference.Acquaintance.Cover (Centered Wld Ind) Ind

John's conceptual cover: the single "patient on TV" concept. He has no concept that picks out the actual-world napoleon.

Equations

• Pearson2015.Napoleon.johnCover = {Pearson2015.Napoleon.patientConcept}

theorem Pearson2015.Napoleon.claimsHe_true : sayDeRe johnCover claimAlts delusionalP john actualN [john]

"John claims he is delusional" is true on the de re LF: the de se center (= John) is fed to the suitable "patient" generator, which at the claim-alternative picks the delusional patient.

theorem Pearson2015.Napoleon.claimsNapoleon_false : ¬claimDeRe johnCover claimAlts delusionalP napoleon john actualN [napoleon]

"John claims Napoleon is delusional" is false: no concept in John's cover is reliable for the res napoleon (the only cover concept returns John, not Napoleon, in the actual world, and Napoleon is not an epistemic alternative of John). So no suitable generator exists.

theorem Pearson2015.Napoleon.napoleon_contrast : sayDeRe johnCover claimAlts delusionalP john actualN [john] ∧ ¬claimDeRe johnCover claimAlts delusionalP napoleon john actualN [napoleon]

The Napoleon contrast ([Pea15] §6): the bound-pronoun (yè) reading is true while the name reading is false — the discrimination Pearson's suitability (reliability + acquaintance-based cover) delivers, and which the earlier vacuous ∃R could not.

Grounding the LogophoricPronoun carrier

theorem Pearson2015.ye_antecedent_is_attitude_holder : Features.Logophoricity.Logophoric.requiredRole ye = Features.Logophoricity.LogophoricRole.self

yè's antecedent is the attitude holder in both readings ([Pea15]: bound by the attitude verb's individual abstractor) — exactly the carrier's requiredRole = .self (Sells: the antecedent must be at least a self = an attitude holder). The de se/de re ambiguity above is orthogonal: both readings keep the attitude holder as antecedent.