Donnellan distinguished how definite descriptions can be used either referentially (to talk about a specific individual regardless of their actual properties) or attributively (to talk about whoever has certain properties). This distinction shows that the same sentence can have different truth conditions depending on speaker intent, suggesting that semantic content depends partly on context of utterance in ways classical theories miss.
Use Donnellan's cases: someone points at an innocent person and says "The murderer is tall" intending to refer to the person they see (who isn't the actual murderer). Attributively, the sentence would be false; referentially, it's true. Practice identifying which use is at play in examples, then examine how Russell's theory and intentionalism each handle the distinction.
Russell's theory of definite descriptions gives a powerful analysis of expressions like "the murderer." On Russell's account, "The murderer is insane" means roughly: there is exactly one person who committed the murder, and that person is insane. The description does not *refer* to anyone directly — it makes a quantified claim. You know this analysis well. Keith Donnellan's 1966 paper "Reference and Definite Descriptions" challenged this picture not by refuting Russell's logic but by showing that language use outruns logical form in an important way.
Donnellan distinguished two uses of a definite description. In the attributive use, the speaker intends to talk about *whoever* satisfies the description — the description's role is to pick out the object that fits. If a detective sees a bloody corpse and says "The murderer is insane" without knowing who committed the crime, she is using the description attributively: she means whoever turns out to be the murderer. If it emerges that no one committed a murder (the death was accidental), her statement doesn't apply to anyone. In the referential use, the description is used to refer to a *specific individual* the speaker has in mind, and the description serves merely as a means of identifying that person in the conversational context. If Smith points at Jones at a party and says "The man drinking a martini is a philosopher," Smith refers to Jones — even if Jones is actually drinking tonic water. The success of reference doesn't depend on Jones actually satisfying the description.
The referential/attributive distinction reveals a gap between semantic content (what the sentence means as a matter of linguistic convention) and speaker meaning (what the speaker intends to communicate). In the referential case, the speaker successfully refers to Jones even though the Russellian analysis delivers the "wrong" answer about who the sentence concerns. This is where the puzzle bites. Does this show Russell's theory is incorrect? Or just incomplete as an account of use? Russell himself might respond (and Grice later articulated this in detail) that what the speaker means can diverge from what the sentence *means* — the referential use is a case of pragmatic deviation from semantic content, not a counterexample to the semantics. Kripke's own response follows this line: the referential use exploits a semantic description for pragmatic purposes, but the underlying semantic mechanism is still Russellian.
The deeper lesson is that context of utterance partly determines what a speaker is talking about in ways that semantic content alone cannot capture. This connects forward to broader debates about context-dependence in language: indexicals ("I", "here", "now"), demonstratives ("this", "that"), and referential uses of descriptions all suggest that meaning cannot be read off the sentence in isolation from the situation in which it is uttered. Donnellan's distinction is often cited as one of the key moments that pushed philosophy of language toward a more thoroughly pragmatic and context-sensitive picture of how language works.
Topics in reflective domains aren't scored by quiz answers. Read, reflect, and mark when you've thought it through.