Donkey Anaphora: Empirical Data

@cite{geach-1962} @cite{heim-1982} @cite{kadmon-1987} @cite{kanazawa-1994} @cite{schwarz-2009}

Theory-neutral data on donkey anaphora and related binding puzzles.

The Phenomenon

Donkey sentences involve pronouns that are apparently bound by indefinites in syntactically "inaccessible" positions:

"Every farmer who owns a donkey beats it"

The indefinite "a donkey" is inside a relative clause, yet "it" seems to depend on it for its reference.

Properties

1. The indefinite doesn't c-command the pronoun (scope puzzle).
2. Often interpreted as "...beats every donkey they own" (universal force).
3. Sometimes "...beats some donkey they own" (weak readings).
4. "Most farmers who own a donkey beat it" (proportion problem).

structure Phenomena.Anaphora.DonkeyAnaphora.DonkeyDatum : Type

A donkey anaphora datum records a sentence with its readings

• sentence : String

  The sentence

• boundReading : Bool

  Does the pronoun depend on the indefinite?

• strongReading : Bool

  Is a universal ("strong") reading available?

• weakReading : Bool

  Is an existential ("weak") reading available?

• grammatical : Bool

  Is the sentence grammatical?

• notes : String

  Notes on the reading

• source : String

  Citation

@[implicit_reducible]

instance Phenomena.Anaphora.DonkeyAnaphora.instReprDonkeyDatum : Repr DonkeyDatum

Equations

• Phenomena.Anaphora.DonkeyAnaphora.instReprDonkeyDatum = { reprPrec := Phenomena.Anaphora.DonkeyAnaphora.instReprDonkeyDatum.repr }

def Phenomena.Anaphora.DonkeyAnaphora.instReprDonkeyDatum.repr : DonkeyDatum → ℕ → Std.Format

Equations

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

def Phenomena.Anaphora.DonkeyAnaphora.geachDonkey : DonkeyDatum

Classic Geach donkey sentence

Equations

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

def Phenomena.Anaphora.DonkeyAnaphora.conditionalDonkey : DonkeyDatum

Conditional donkey

Equations

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

def Phenomena.Anaphora.DonkeyAnaphora.negatedDonkey : DonkeyDatum

Negated donkey

Equations

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

def Phenomena.Anaphora.DonkeyAnaphora.mostDonkey : DonkeyDatum

"Most" donkey (proportion problem)

Equations

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

structure Phenomena.Anaphora.DonkeyAnaphora.WeakStrongDatum : Type

Data on weak vs strong reading availability

• sentence : String
• strongAvailable : Bool
• weakAvailable : Bool
• preferredReading : String
• context : String
• source : String

@[implicit_reducible]

instance Phenomena.Anaphora.DonkeyAnaphora.instReprWeakStrongDatum : Repr WeakStrongDatum

Equations

• Phenomena.Anaphora.DonkeyAnaphora.instReprWeakStrongDatum = { reprPrec := Phenomena.Anaphora.DonkeyAnaphora.instReprWeakStrongDatum.repr }

def Phenomena.Anaphora.DonkeyAnaphora.instReprWeakStrongDatum.repr : WeakStrongDatum → ℕ → Std.Format

Equations

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

def Phenomena.Anaphora.DonkeyAnaphora.strongDominant : WeakStrongDatum

Strong reading dominant

Equations

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

def Phenomena.Anaphora.DonkeyAnaphora.weakFacilitated : WeakStrongDatum

Weak reading facilitated by context

Equations

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

def Phenomena.Anaphora.DonkeyAnaphora.onlyStrong : WeakStrongDatum

Only strong reading possible

Equations

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

def Phenomena.Anaphora.DonkeyAnaphora.bathroomSentence : DonkeyDatum

Bathroom sentences (Partee)

Equations

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

def Phenomena.Anaphora.DonkeyAnaphora.paycheckSentence : DonkeyDatum

Paycheck pronouns

Equations

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

def Phenomena.Anaphora.DonkeyAnaphora.sagePlant : DonkeyDatum

Sage plant sentences (Evans)

Equations

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

structure Phenomena.Anaphora.DonkeyAnaphora.DiscourseDonkeyDatum : Type

Discourse-level donkey anaphora

• sentences : List String
• pronounSentenceIdx : ℕ
• antecedentSentenceIdx : ℕ
• grammatical : Bool
• notes : String
• source : String

def Phenomena.Anaphora.DonkeyAnaphora.instReprDiscourseDonkeyDatum.repr : DiscourseDonkeyDatum → ℕ → Std.Format

Equations

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

@[implicit_reducible]

instance Phenomena.Anaphora.DonkeyAnaphora.instReprDiscourseDonkeyDatum : Repr DiscourseDonkeyDatum

Equations

• Phenomena.Anaphora.DonkeyAnaphora.instReprDiscourseDonkeyDatum = { reprPrec := Phenomena.Anaphora.DonkeyAnaphora.instReprDiscourseDonkeyDatum.repr }

def Phenomena.Anaphora.DonkeyAnaphora.crossSententialDonkey : DiscourseDonkeyDatum

Cross-sentential binding

Equations

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

def Phenomena.Anaphora.DonkeyAnaphora.indefinitePersists : DiscourseDonkeyDatum

Indefinite persists across sentences

Equations

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

structure Phenomena.Anaphora.DonkeyAnaphora.ProportionDatum : Type

The proportion problem: what is being counted?

• sentence : String
• farmerCountingReading : String

  Farmer-counting: proportion of farmers (who own donkeys)

• pairCountingReading : String

  Pair-counting: proportion of farmer-donkey pairs

• preferredReading : String

  Which reading is more natural?

• source : String

def Phenomena.Anaphora.DonkeyAnaphora.instReprProportionDatum.repr : ProportionDatum → ℕ → Std.Format

Equations

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

@[implicit_reducible]

instance Phenomena.Anaphora.DonkeyAnaphora.instReprProportionDatum : Repr ProportionDatum

Equations

• Phenomena.Anaphora.DonkeyAnaphora.instReprProportionDatum = { reprPrec := Phenomena.Anaphora.DonkeyAnaphora.instReprProportionDatum.repr }

def Phenomena.Anaphora.DonkeyAnaphora.proportionProblem : ProportionDatum

Classic proportion problem case

Equations

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

def Phenomena.Anaphora.DonkeyAnaphora.donkeyData : List DonkeyDatum

All basic donkey data

Equations

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

def Phenomena.Anaphora.DonkeyAnaphora.weakStrongData : List WeakStrongDatum

All weak/strong reading data

Equations

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

def Phenomena.Anaphora.DonkeyAnaphora.discourseDonkeyData : List DiscourseDonkeyDatum

All discourse-level data

Equations

• Phenomena.Anaphora.DonkeyAnaphora.discourseDonkeyData = [Phenomena.Anaphora.DonkeyAnaphora.crossSententialDonkey, Phenomena.Anaphora.DonkeyAnaphora.indefinitePersists]

Donkey anaphora as a definiteness use type

@cite{schwarz-2009} §3: donkey pronouns in German pattern with anaphoric definites (strong article von dem), not with uniqueness definites (weak article vom). This means donkey anaphora selects the familiarity presupposition type — the same as regular anaphora.

The cross-linguistic pattern:

• German: strong article (von dem) — two-article system distinguishes
• Thai/Mandarin: demonstrative — demonstratives fill the strong-article role
• Shan: bare noun — no articles, so no morphological signal

This parallels the general pattern for anaphoric definites in @cite{moroney-2021} Table 4.4.

def Phenomena.Anaphora.DonkeyAnaphora.donkeyUseType : Features.Definiteness.DefiniteUseType

Donkey anaphora is classified as its own use type in the definiteness typology, distinct from regular anaphora but sharing the same presupposition type (familiarity).

Equations

• Phenomena.Anaphora.DonkeyAnaphora.donkeyUseType = Features.Definiteness.DefiniteUseType.donkey

theorem Phenomena.Anaphora.DonkeyAnaphora.donkey_presup_is_familiarity : Features.Definiteness.useTypeToPresupType donkeyUseType = Features.Definiteness.DefPresupType.familiarity

Donkey pronouns select the familiarity (strong article) presupposition type. @cite{schwarz-2009} §3: in German, donkey pronouns require the strong article (von dem), patterning with anaphoric uses rather than uniqueness uses.

theorem Phenomena.Anaphora.DonkeyAnaphora.donkey_patterns_with_anaphoric : Features.Definiteness.useTypeToPresupType Features.Definiteness.DefiniteUseType.donkey = Features.Definiteness.useTypeToPresupType Features.Definiteness.DefiniteUseType.anaphoric

Donkey anaphora and regular anaphora share the same presupposition type (familiarity). This is why they select the same article form cross-linguistically: both require discourse familiarity, not situational uniqueness.

structure Phenomena.Anaphora.DonkeyAnaphora.DonkeyArticleDatum : Type

Cross-linguistic data on how donkey anaphora is expressed morphologically. This connects the abstract DefiniteUseType.donkey to concrete article forms.

The article system (articleSystem) is derived from the language's fragment-level articleInventory, not stipulated independently — ArticleInventory is the single source of truth.

• language : String
• isoCode : String
• form : String

  Morphological form used for donkey pronouns

• articleInventory : Core.Nominal.ArticleInventory

  Morphological article inventory (single source of truth from which articleSystem is derived).

def Phenomena.Anaphora.DonkeyAnaphora.DonkeyArticleDatum.articleSystem (d : DonkeyArticleDatum) : Features.Definiteness.ArticleType

Schwarz ArticleType classification, derived from articleInventory.

Equations

• d.articleSystem = d.articleInventory.toArticleType

def Phenomena.Anaphora.DonkeyAnaphora.germanDonkey : DonkeyArticleDatum

Equations

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

def Phenomena.Anaphora.DonkeyAnaphora.thaiDonkey : DonkeyArticleDatum

Equations

• Phenomena.Anaphora.DonkeyAnaphora.thaiDonkey = { language := "Thai", isoCode := "tha", form := "demonstrative", articleInventory := Fragments.Thai.Definiteness.articleInventory }

def Phenomena.Anaphora.DonkeyAnaphora.mandarinDonkey : DonkeyArticleDatum

Equations

• Phenomena.Anaphora.DonkeyAnaphora.mandarinDonkey = { language := "Mandarin", isoCode := "cmn", form := "demonstrative", articleInventory := Fragments.Mandarin.Definiteness.articleInventory }

def Phenomena.Anaphora.DonkeyAnaphora.shanDonkey : DonkeyArticleDatum

Equations

• Phenomena.Anaphora.DonkeyAnaphora.shanDonkey = { language := "Shan", isoCode := "shn", form := "bare noun", articleInventory := Fragments.Shan.Definiteness.articleInventory }

def Phenomena.Anaphora.DonkeyAnaphora.donkeyArticleData : List DonkeyArticleDatum

All cross-linguistic donkey article data.

Equations

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

theorem Phenomena.Anaphora.DonkeyAnaphora.weakAndStrong_uses_strong : ((List.filter (fun (x : DonkeyArticleDatum) => x.articleSystem == Features.Definiteness.ArticleType.weakAndStrong) donkeyArticleData).all fun (x : DonkeyArticleDatum) => x.form == "strong article (von dem)") = true

In languages with a weak/strong article distinction, donkey anaphora uses the strong form — never the weak form. This is predicted by donkey_presup_is_familiarity: strong articles encode familiarity, and donkey anaphora requires familiarity.