Explanation
A donkey sentence typically introduces an indefinite expression inside a conditional or quantified structure and later resumes it with a pronoun even though straightforward grammatical binding is unavailable. In “If someone owns a donkey, they feed it,” it appears to refer to a donkey, yet neither one particular animal nor simple coreference explains the generalising reading. The phenomenon has become a test case for theories of anaphora, quantification, and discourse representation. Dynamic approaches model how a discourse referent introduced locally becomes accessible to subsequent interpretation. Donkey sentences therefore show with unusual clarity that meaning is constructed incrementally in discourse and that pronouns do not always point to already fixed things; their interpretation depends on logical structure, possible cases, and an evolving context.
Movement of thought
1. An indefinite expression introduces a possible referent within a conditional or quantified structure. 2. A later unit resumes it pronominally although its ordinary syntactic binding domain has already been left behind. 3. Understanding keeps the referent accessible across possible cases and associates each relevant owner with an appropriate object. 4. A counter-test varies quantification and clause structure, separating donkey anaphora from simple coreference and ordinary variable binding. 5. The referent returns not as one fixed thing but as a discourse-maintained possibility within every suitable case.
Scientific use case
Formal semantics, pragmatics, and computational linguistics compare dynamic, e-type, and situation-semantic accounts; controlled variants test quantificational force, accessibility, and readings.
Example
“Every researcher who finds a rare attestation documents it” requires an appropriate pronoun reference for each possible finding without presupposing one single attestation.
Counterexample
In “Maria owns a donkey. She feeds it,” the name, indefinite, and pronouns can refer to particular discourse entities. This simple cross-sentential anaphora is not yet a donkey sentence.
Distinction
Distinguish it from simple discourse anaphora, coreference, syntactically bound variable, relative clause, generic statement, and cataphora; anaphoric dependence within quantification is decisive.
Effect on thought and perception
The donkey sentence forces models to treat meaning as evolving context and reveals the limits of assigning pronouns to referents within a purely static representation.
Resonance
An indefinite referent enters discourse only conditionally yet remains reachable; the pronoun follows its trace through a landscape of possible cases.
Cite this entry
Reich, Michael (Hrsg./ed.): „Donkey sentence“. Atlas der Sprache, AT-DEP4-0001. ATLAS 260, data model 1.6.0. https://atlasoflanguage.com/concept/AT-DEP4-0001The address remains stable; editorial content may evolve. Editorial reading state: 2026-10-15. Include your access date for precise references.
Historical background
Modern discussion of the donkey sentence is commonly linked to Geach's 1962 Reference and Generality, but Seuren's historical study shows that related quantificational problems were already formulated in medieval logic, including work by Walter Burley. The canonical sentence type contains an indefinite noun phrase apparently resumed by a pronoun beyond a simple quantifier-scope analysis. Donkey is therefore the name of an example, not a semantic class of animals. German Eselssatz mirrors the English research label, without a verified borrowing chronology here; the medieval antecedents do not license backdating the German term.
How conceptual history changes the analysis
The donkey sentence tests binding, quantification, and discourse reference; the example tradition and the later technical label must be kept historically distinct.
Further explanation
The apparently quaint name addresses the scope of ordinary claims. “Anyone who owns a bicycle must register it” does not identify one previously named bicycle. The pronoun is interpreted through a condition. A simple picture of pronouns referring to fixed antecedent objects needs further work.
Move from familiar understanding to what a formal account must explain. What if someone owns two bicycles: does registering one suffice? Testing possible readings exposes the problem without declaring one intuition universal. The historical antecedents belong to the logical issue, not necessarily to the modern label.
Additional example
Original example: “Anyone who owns a bicycle must register it.” Two bicycles expose a question about quantification. This is invented language data, not an actual legal requirement.
References for the historical section
- P. A. M. Seuren (2006): Donkey Sentences · Historical discussion of Geach 1962 and medieval predecessors