Part I
Framing the Project
1 Two Distinctions Within the
Category of Designators
1.1 PRELIMINARY DISCUSSION OF REFERENCE AND STRUCTURE
Humans gather, manipulate, and share information; these phenomena constitute much of our conscious lives and underlie most of our higher cognitive capacities. Philosophy of language is a part of the piecemeal interdisciplinary project of coming up with a comprehensive theoretical description of these phenomena.
One central sub-task within the philosophy of language concerns the individuation of information—e.g., what criteria determine whether two strings of symbols semantically express dif erent information (as do ‘John loves Mary’ and ‘Mary loves John’) or express the same information differently (as do ‘John loves Mary’ and ‘Mary is loved by John’)? A related canonical sub-task within that larger project, which occupies a considerable amount of the attention and energy of twentieth-century philosophy of language, concerns drawing a principled distinction between singular (or object-dependent) vs. general (or object-independent) information.1 This research project aims to clear further ground toward the proper understanding of this ancient and deeply signifi cant categorical division between types of information.
Alternatively, this particular task concerns identifying a centrally important, substantive fault line within the category of designators in Kripke’s (1972) loose and intuitive sense of the term .2 More specifi cally, this essay is focused on two ways of dividing up the category of designators: fi rst, structured vs. unstructured expressions, and second, referring vs. denoting expressions. I will argue that there are many deep and illuminating connections between these two distinctions, so that the notion of semantic structure, properly excavated, underlies and grounds various important points in the theory of reference. I begin by drawing these two distinctions in a preliminary way; the contours of both distinctions will be mapped more precisely as this work proceeds.
The structured/unstructured distinction is relatively straightforward.3 An expression is structured only if it has independently meaningful proper parts. So, sentences are typically structured, 4but individual terms can either be structured (e.g., compound words like ‘snowball’) or unstructured (e.g., atomic words like ‘ball’). ‘Scott’ vs. ‘the author of Waverley’ and ‘nine’ vs. ‘the number of baseball positions’ are two paradigmatic pairs of co-designative terms—the first of which is unstructured, the second of which is structured. Only structured expressions involve molecular compositional complexity; unstructured expressions are, in contrast, atomic and primitive. (They are ‘primitive’ in that it seems reasonable to hold that, typically, structured expressions ultimately factor out into unstructured bits. There may be a limited set of oddball idiomatic exceptions.)
The referring/denoting distinction is more complex and involved, as it is inextricably tied up with (among other things) Russellian exegesis and controversial theses in the philosophy of language. The most important episode in the pre-history of Russell’s (1905) distinction between referring and denoting is Frege’s (1879: Ch.1, §9) argument that the following subject-expressions illustrate dif erent semantic mechanisms:
[1] 20 can be represented as the sum of four squares.
[2] Every positive integer can be represented as the sum of four squares.
Designators like ‘20’ are semantically suited for singling out a specifi c individual and expressing a proposition about it; in contrast, designators like ‘every positive integer’ are semantically suited for expressing general (i.e., quantifi cational) relations among concepts. As Frege’s primary interest here is in logic, his seminal theory of quantifi cation is largely motivated by the different inferential patterns associated with these two types of designators.5
One of the key dif erences between Frege’s and Russell’s view of these matters concerns exactly how to draw the singular/general divide—i.e., notoriously, whereas Frege (1892a) treats defi nite descriptions (e.g., ‘the heaviest animal in the Toronto zoo’) as akin to ‘20’, Russell (1905) argues that they belong on the ‘every positive integer’ side of Frege’s divide. In Russell’s view, whether a proposition should be counted as singular or general is not simply a question of whether its truth-condition concerns exactly one individual; rather, it is a question of what kind of semantic mechanism is at work—i.e., referring or denoting. Down this avenue, Frege’s distinction between singular and general truth-conditions develops into Russell’s more fundamental semantic distinction between these two dif erent types of designations.
While referring is a conventional or stipulative relation between certain kinds of designator and what they are used to designate (paradigm cases include ‘I’, ‘this’, ‘20’, ‘Jones’, and ‘gold ’6), denoting is a distinct sort of connection that holds between certain kinds of designators (such as ‘the person who denounced Catiline’ or ‘a squirrel I saw yesterday’) and which satisfies the condition semantically expressed .7 Utterances of sentences whose subject-expressions are referring expressions semantically express object-dependent propositions (e.g., ‘I am cold’ or ‘20 is even’). Utterances of sentences whose subject-expressions are denoting expressions semantically express object-independent propositions (e.g., ‘A squirrel I saw yesterday was albino’ or ‘Some cats are fond of dogs’).8 Even if these latter cases are about exactly one individual, in the attenuated sense of aboutness that Russell calls ‘denoting’, still only a compositionally determined condition is essential to their content (i.e., the content of the proposition stays constant across situations in which distinct individuals, or no individuals, satisfy the relevant compositionally determined condition).9
Initially, it might be thought that I have given two defi nitions of the term ‘reference’ in the preceding paragraph—i.e., one that takes the criterion of reference to be a conventional, stipulative semantic relation, and a second that ties reference to the notion of object-dependent information. For example, certain indexicals—e.g., ‘she’—might seem to count as referring expressions in only the second sense. By Part IV, I will be in a position to demonstrate that these two defi nitions are equivalent. In the interim, several refi nements are required.
I will employ terms from Bach (1987) and call the semantic mechanism involved in referring ‘relational’ and the semantic mechanism involved in denoting ‘satisfactional’. In the case of a use of a referring expression, its designatum is that which stands in a certain relation to the utterance—i.e., perhaps a causal-historical relation is appropriate, for the case of a proper name or natural kind term; a more character-driven relation, for the case of demonstratives or indexicals; and so on. In contrast, the designatum of a denoting expression is whatever, in the relevant context of evaluation, happens to satisfy the compositionally determined condition that it semantically expresses.10
To sum up this preliminary discussion of the distinction between referring expressions and denoting expressions, consider the dif erence between my saying the following:
[3] I am wearing glasses.
[4] The tallest person in this room is wearing glasses.
The designators in [3] and [4] happen to designate the same individual, but they instance distinct semantic mechanisms. ‘I’ (as used by me) refers to me, whereas ‘the tallest person in this room’ (in this context) denotes me (i.e., it semantically expresses a compositional condition which I happen to satisfy in this context, but might not satisfy in other contexts). Which individual is relevant to the truth-conditions expressed by [3] stays constant across possible contexts of evaluation, whereas which individual is relevant to the truth-conditions expressed by [4] varies according to circumstances. Crucially, if a seven-foot-tall woman were to have entered this room, in uttering [4] I would designate her while semantically expressing exactly the same proposition.11 Hence, as the following table illustrates, the distinction between referring and denoting lines up with the distinction between object-dependent and object-independent propositions:
Table 1.1 Types of Designator (preliminary version) Type of designator | Referring expression | Denoting expression |
Semantic mechanism | Relational | Satisfactional |
Proposition semantically expressed | Object-dependent | Object-independent |
The central aim of this research is to excavate the many deep connections between these two distinctions (i.e., unstructured vs. structured designators, and referring vs. denoting designators), and to show that a proper understanding of these connections helps to illuminate various important points within, and beyond, the theory of reference.
1.2 THE CENTRAL THESES
We have out on the table two ways of dividing up the category of designators—the fi rst (i.e., structured vs. unstructured) has to do with the absence or presence of molecular complexity; the second (i.e., referring vs. denoting) has to do with the way in which the designatum is semantically singled out. There is some reason to think that both distinctions are both exclusive and exhaustive—i.e., every designator is either but not both structured or unstructured, and either but not both referring or denoting.12 It is therefore natural that questions arise as to how these two distinctions relate.
It seems initially evident that there is some fairly close connection between the notion of semantic structure and the distinction between referring and denoting. Clearly, there is something relatively atomic about, say, Russell’s (1905, 1911, 1918) paradigm cases of referring expressions, or the expressions upon which Donnellan (1970), Kripke (1972), Putnam (1975), and Kaplan (1977) build the causal-historical theory of reference. These paradigmatic referring expressions—e.g., ‘I’, ‘this’, ‘Jones’, ‘gold’—lack independently meaningful proper parts. In contrast, all denoting expressions involve molecular compositional complexity, as is attested by the following representative list of denoting expressions from Russell’s (1905: 479) seminal work:
By ‘a denoting phrase’ I mean a phrase such as any one of the following: a man, some man, any man, every man, all men, the present King of England, the present King of France, the center of mass in the solar system at the fi rst instant of the twentieth century, the revolution of the sun round the earth. Thus a phrase is denoting solely in virtue of its form.
So, at fi rst glance, it seems that referring expressions are typically unstructured, while denoting expressions are typically structured.
A question that arises, then, is: Are the following theses tenable?
[R if U] All and only referring expressions are unstructured designators.
[D if S] All and only denoting expressions are structured designators.
(I will sometimes refer to these as ‘the central theses’, and to views which espouse them as comprising ‘the Russellian orthodoxy in the theory of reference’.) Prima facie, both seem to be open to counterexample. First, there seem to be structured referring expressions, such as complex demonstratives (e.g., ‘that big green-headed duck’) and referentially used descriptions (e.g., ‘the world heavyweight champ’). One might also think that ordinary proper names pose a similar complication, as names are the very paradigm of a referring expression, and yet people tend to be called by complex, molecular expressions like ‘Willard van Orman Quine’ or ‘Louis XIV’. If these appearances hold up to scrutiny, then such structured referring expressions are counterexamples to the [if R then U] and the [if S then D] clauses of the above theses.13
In the other direction, even if unstructured denoting expressions might be harder to fi nd (as it seems that a denoting expression requires a determiner plus a nominal expression), the notion is not entirely without precedent. For example, Russell (1911, 1918) viewed ordinary proper names as disguised or truncated descriptions—e.g., ‘Bismarck’ is just shorthand for, say, ‘the fi rst Chancellor of the German empire’—and so, in a sense, as unstructured denoting expressions .14 Relatedly, perhaps there are grounds for thinking that the relation between utterances of a...