The distinction languages make between singular and plural terms has been correlated in recent work in semantics with a distinction in the world between singular and plural entities. If I say that John and Mary met, then I am saying neither that John met nor that Mary met, but rather I am speaking about the plurality consisting of the two of them. Pluralities correspond then to sets of (singular) individuals. At this point one can ask whether we want also to posit entities corresponding to sets of sets of individuals and even higher order entities? I maintain that we do not, and in this article I focus on the following kind of argument in favor of 'higher order entities'.
Even assuming that we have young and old, male and female Senators, the noun phrase a.the male Senators and the female Senators cannot always replace the related noun phrase b. the Senators without affecting the meaning of the sentence in which the replacement is made. By allowing in higher order entities, this is explained by taking (a) to denote a plurality consisting of just two parts, one female, the other male, while (b) would denote a different plurality composed simply of individual Senators. Similarly, if the noun phrase c. the young Senators and the old Senators is taken to denote yet another plurality with two parts, one young, the other old, it can be distinguished from (a) and (b). As can be seen above, the argument for the higher order approach is based on non- coreference claims and so I begin with a general discussion of such claims.
Next I consider the relatedness of noun phrases like a.-c. above examining the degree to which they can replace one another without affecting meaning. As a result of this discussion it becomes apparent that the examples underpinning the higher order approach provide very weak evidence for it, evidence that can only be maintained at the cost of positing a systematic ambiguity in the meanings of plural verb phrases. The argument is especially weak in light of the fact that the ambiguity itself could not even be expressed without adopting the higher order approach in the first place.
While groups of noun phrases like those in a.-c. are, I maintain, equivalent in reference, they nonetheless differ in other ways. In particular, they differ in the way they affect discourses in which they are used. In the third part of the paper, I discuss these differences, especially in relation to the meaning and use of reciprocal predicates such as were separated from each other. I argue that reciprocals have discourse sensitive meanings. The interaction of this discourse sensitivity with different members of groups of related noun phrases, was, I claim, earlier misdiagnosed as the result of differences in the reference of these noun phrases.