Donkeys, bishops, and the distinction between weak and strong definites

Much recent work on definiteness has assumed that definite descriptions come in two types, weak and strong. Weak definites make use of a simple situation-relative uniqueness operator; strong definites add to this a variable available to dynamic binding. Drawing on evidence from Mandarin, Jenks (2018) influentially proposed that strong definites but not weak definites can serve as donkey anaphors. We show that this claim is incorrect for Mandarin as well as for German and English. Certain special cases of donkey anaphora nevertheless have diagnostic value for the distinction between weak and strong definites. We discuss two. First, in structures where there is no dynamic quantifier available to serve as an antecedent, strong definite descriptions cannot serve as donkey anaphors. Second, in so-called bishop sentences, the uniqueness requirement of the weak definite is not met; in these cases weak definite descriptions cannot serve as donkey anaphors. We apply these tests to definites in German, English, and Mandarin. Our results indicate that English the and Mandarin bare nouns are always weak definites, but also that important aspects of crosslinguistic variation with respect to weak definites remain to be explained.