Abstract
Generically interpreted bare plural noun phrases license donkey anaphora.
This fact has unexpected consequences both for the analysis of generics and
for the analysis of donkey anaphors. Specifically, if we assume a kindsbased
analysis of bare plurals as in Carlson (1980), we will be forced to
give up the idea that donkey anaphors are variables – presumably in favor
of an E-type analysis. Conversely, if we assume that donkey anaphors are
variables, we will be forced to give up Carlson’s treatment of generically
interpreted bare plurals as denoting kinds – presumably in favor of an
analysis employing a generic operator like that of Wilkinson (1991), which
quantifies over objects directly, without the mediation of kinds.
The two issues are thus dependent on one another, and cannot be decided independently. This paper will not attempt to choose between an E-type and
a variable-based analysis of donkey anaphors, or between a kinds-based and
a kinds-free analysis of generic bare plurals, but only to point out that the
choice one makes with regard to either one of these issues appears to force
a particular choice with respect to the other issue