Timothy Smiley

1. ExpositionRichard Sharvy's ‘A more general theory of definite descriptions’ was published in 1980. Its aim was to replace Russell's paradigm by " a general theory of definite descriptions, of which definite mass descriptions, definite plural descriptions, and Russellian definite singular count descriptions are species. … We have an account of the generic ‘the’ along these same lines. " By now his theory has attained the status of a new paradigm. Even a casual trawl of the literature throws up over a score of citations and endorsements, and no rebuttals of its central contention that the primary use of ‘the’ is to indicate totality . We shall argue that it is wrong for mass description, wrong for singular description, wrong for plural description and wrong for generic ‘the’. Sharvy would have approved: his obituary says that his car displayed the slogan ‘Subvert the Dominant Paradigm’.He follows Russell in treating descriptions as incomplete symbols, but nothing in his paper turns on this aspect of it, and we can bypass the issue by considering them in the context of complete sentences. Where Russell translates F into the predicate calculus as Formula Sharvy translates it as Formula Here ≤ is supposed to stand for some or other species of part of relation. His other governing idea is of a predicate's being cumulative. G is cumulative if any number of things that satisfy G have a sum, i.e. a least upper bound with respect to ≤ which also satisfies G. The meanings of ≤, sum and cumulativity vary according to the different kinds of description, which we take in the order in which Sharvy presents them. Meanwhile, we note that he originally defined ‘sum’ to mean simply the l.u.b. , but subsequently redefined …

We present a plural logic that is as expressively strong as it can be without sacrificing axiomatisability, axiomatise it, and use it to chart the expressive limits set by axiomatisability. To the standard apparatus of quantification using singular variables our object-language adds plural variables, a predicate expressing inclusion (is/are/is one of/are among), and a plural definite description operator. Axiomatisability demands that plural variables only occur free, but they have a surprisingly important role. Plural description is not eliminable in favour of quantification; on the contrary, quantification is definable in terms of it. Predicates and functors (function signs) can take plural as well as singular terms as arguments, and both many-valued and single-valued functions are expressible. The system accommodates collective as well as distributive predicates, and the condition for a predicate to be distributive is definable within it; similarly for functors. An essential part of the project is to demonstrate the soundness and completeness of the calculus with respect to a semantics that does without set-theoretic domains and in which the use of settheoretic extensions of predicates and functors is replaced by the sui generis relations and functions for which the extensions were at best artificial surrogates. Our metalanguage is designed to solve the difficulties involved in talking plurally about individuals and about the semantic values of plural items

The history of the idea of predicate is the history of its emancipation. The lesson of this paper is that there are two more steps to take. The first is to recognize that predicates need not have a fixed degree, the second that they can combine with plural terms. We begin by articulating the notion of a multigrade predicate: one that takes variably many arguments. We counter objections to the very idea posed by Peirce, Dummett's Frege, and Strawson. We show that the arguments of a multigrade predicate must be grouped into places, with perhaps several arguments occupying positions at a place. Variability may relate to places or positions. Russell's multiple judgement predicate turns out to be just one example of a family-'is necessarily true of', 'is said of', 'is instantiated by' and so on-of predicates with variably many places. Our main concern, however, is lists. Any adequate account of lists must include plural as well as singular terms. On one account, lists are mere strings of separate arguments, which occupy variably many positions within a place of a multigrade predicate. A quite different account takes the list itself to be a compound plural term. We compare these rival conceptions, and reach some surprising conclusions. As a coda, we deploy the conceptual apparatus developed in the paper to assess Morton's pioneer system of multigrade logic.

We all learn about the mistake of treating ‘nothing’ as if it were a term standing for something; but is it a mistake to treat it as an empty term, denoting nothing? We argue not, and we introduce ‘zilch’, defined as ‘the non-self-identical thing’, as a term which is empty as a matter of logical necessity. We contrast its behaviour with that of the quantifier ‘nothing’, and illustrate its uses. We use the same idea to vindicate Locke’s, Descartes’ and Hume’s handling of ‘nothing’, and we show how Heidegger’s ‘das Nichts nichtet’ comes out as logically true

Russell had two theories of definite descriptions: one for singular descriptions, another for plural descriptions. We chart its development, in which ‘On Denoting’ plays a part but not the part one might expect, before explaining why it eventually fails. We go on to consider many-valued functions, since they too bring in plural terms—terms such as ‘4’ or the descriptive ‘the inhabitants of London’ which, like plain plural descriptions, stand for more than one thing. Logicians need to take plural reference seriously if only because mathematicians take many-valued functions seriously. We assess the objection (by Russell, Frege and others) that many-valued functions are illegitimate because the corresponding terms are ambiguous. We also assess the various methods proposed for getting rid of them. Finding the objection ill-founded and the methods ineffective, we introduce a logical framework that admits plural reference, and use it to answer some earlier questions and to raise some more.