Descriptivism Distilled

Descriptivism Distilled

Kent Bach

Proper names may be simple on the surface, but they raise some deep questions, about singular thought as well as singular reference. How can a mere name connect us to something in the world? How can using a mere name direct someone else’s attention to something, seemingly without conveying any information about it? The very idea of singular thought suggests that when we attribute a property to a certain object, we are thinking of that object in particular rather than “under a description,” as merely the unique thing of a certain sort. We can do this not just with individuals we are currently perceiving or have previously perceived but even with ones we have learned of and know of only by name. However this is possible, it seems that having a name for something helps us maintain a mental record of it, a record which can be called up and consulted, and added to or corrected in light of new information. Indeed, an individual can come to mind just by virtue of its name occurring to us—think ‘Mongolia’ and you think of Mongolia. As for singular reference, the name of something is generally the linguistic device best suited for calling it to others’ minds, at least if they know of it by name (using a name for something unfamiliar to them serves to introduce them to that individual). Similarly, others can call our attention to something (or introduce us to it) just by using its name.

So calling things to mind seems to be what names are for, in both thought and communication. Mill recognized this when he wrote, “a proper name is but an unmeaning mark which we connect in our minds with the idea of the object, in order that whenever the mark meets our eyes or occurs to our thoughts, we may think of that individual object” (1872, 22). The function of proper names, Mill thought, is not to convey general information but rather “to enable individuals to be made the subject of discourse;” names are “attached to the objects themselves, and are not dependent on … any attribute of the object” (1872, 20). In this way, our use of names can accommodate such pervasive facts as that things can change over time, that one’s conception of something can change over time, and that different people’s conceptions of the same thing can differ. All this is possible if using a name in thinking of or referring to an object is not a matter of representing it as having certain properties but, as Russell said, “merely to indicate what we are speaking about; [the name] is no part of the fact asserted … : it is merely part of the symbolism by which we express our thought” (1919, 175).

No wonder, then, that descriptivism has seemed implausible. Its motivation is understandable enough—to explain how names can play their roles even when they lack bearers—but, in both Frege’s and Russell’s formulations, descriptivism seems unable to do justice to the above observations. Frege held that a proper name (like any other sort of expression) has a reference-determining sense, which imposes a condition that an individual must satisfy in order to qualify as the name’s referent. However, his sense-descriptivism, as we will call it, is subject to the familiar objections made by Donnellan (1970) and Kripke (1980): there no unique such descriptive condition associated with a given name (even taking just one of its bearers into account, if it has more than one), and the name’s reference is not subject to counterfactual variation in the way descriptivism predicts. As for Russell’s view, it seems as implausible as Frege’s, even though it denies that names have senses. It claims instead that names (“ordinary” names, anyway) are “abbreviated” or “disguised” definite descriptions, but this abbreviational descriptivism defy semantic appearances. And, when combined with Russell’s notorious theory of descriptions, according to which singular definite descriptions are not singular terms but complex quantificational phrases, it defies syntactic appearances. But appearances can be deceiving.

That is why Jerrold Katz asks, “Has the description theory of names been refuted?” (the title of Katz 1990). He thinks not. Although his defense of descriptivism is intricate, his basic rationale for it is Russell’s simple point (1918, 242) that if the sole semantic role of names were to denote their bearers, as on Mill's view, then vacuous names would be meaningless. In “Names without Bearers,” Katz points out that vacuous names are perfectly meaningful and that sentences in which they occur can express propositions. This explains why sentences like ‘Santa Claus does not exist,’ which is not only meaningful but true, “receive neither a raised eyebrow from the ordinary speaker nor an asterisk from the linguist” (Katz 1994, 1). Katz acknowledges the familiar difficulties with the usual versions of descriptivism, but his version, which he sees as striking a middle ground between “Fregeanism” and “Millianism,” is metalinguistic: a proper name is semantically equivalent to a definite description that mentions it. This sounds like a version of abbreviational descriptivism à la Russell, but in fact Katz holds that proper names have senses. His view is not fully Fregean, though, in that Katz’s senses do not determine reference—they merely delimit it.

Metalinguistic descriptivism was first suggested by Russell, if only in passing (1919, 174), and it has occasionally been advocated since. Like Katz, I find it far more plausible than its reputation would suggest, at least when it is carefully stated and properly qualified. It can explain or at least accommodate certain phenomena that other theories cannot, and, by adverting not to identifying properties but only to the “purely nominal bearer relation” (Katz 1994, 7), it is not vulnerable to the usual objections to description theories. Also, I agree with Katz that its plausibility depends on enforcing the distinction between “features of language” and “features of language use” (1994, 19). Beyond that, however, Katz’s defense of metalinguistic descriptivism is problematic in a number of ways. For one thing, in billing his account as the alternative to Frege’s and Mill’s views, Katz fails to distinguish Russell's brand of descriptivism from Frege's and modern direct-reference theories from Mill’s. Also, when contrasting his own notion of sense with Frege’s, he falls prey to the common misconception that Frege’s notion belongs to the theory of linguistic meaning rather than to the theory of thought. Separating the two is important, as we will see, because descriptivism at the level of language (Russellian or Fregean) does not commit one to descriptivism at the level of thought. Keeping the two levels separate is crucial to the defense of metalinguistic descriptivism.

As for its precise formulation, I agree with Katz on the importance of invoking only the nominal bearer relation but disagree with him on how to do this, as to both wording and framework. On his view, which he dubs the “pure metalinguistic theory” (PMT), a proper name ‘N’ has the sense of ‘the thing which is a bearer of “N”.’ I prefer what I call the “nominal description theory” (NDT), with its simpler schema, ‘the bearer of “N”.’ It too invokes the nominal bearer relation but not the notion of sense. I will argue that PMT is needlessly complex and that the notion of sense does not do the job for which Katz enlists it. NDT provides a much simpler and cleaner way of explaining how a name with many bearers can be used to refer to one of its bearers in particular. By exploiting the distinction between features of language and features of language use, we can assimilate the case of names with many bearers to the case of incomplete definite descriptions. The resulting pragmatic account picks up the semantic slack left by the meaning of the incomplete definite description with which, according to NDT, a name with many bearers is semantically equivalent. Just as an incomplete description like ‘the table,’ even though it implies uniqueness, can be used to refer to a particular table, so a shared name like ‘Steve Jones,’ (equivalent to ‘the bearer of “Steve Jones”,’ can be used to refer to a particular bearer of the name. We can make sense of this fact by distinguishing the linguistic meaning of a sentence containing the name, the thought in the mind of the speaker, and what is communicated by the utterance of the sentence.

Descriptivism is defensible, I believe, if its scope, form, and content are properly delimited. It should be understood as (1) a thesis about a name’s contribution to sentential meanings, not to thought contents, as (2) claiming not that names have senses but that they are semantically equivalent to definite descriptions, and as (3) taking these to be metalinguistic descriptions of the form ‘the bearer of “N”.’ This version of descriptivism avoids the problems that confront the direct-reference view of names, while doing justice to the intuitions that support that view. It also explains the fact, acknowledged even by Mill, that when we refer to persons or things by name, we do not convey “any information about them, except that those are their names” (1872, 22; my italics). It does this, as we will see, in a way that is compatible with the fact that sentences containing proper names are ordinarily used to communicate singular propositions.

1. Abbreviational vs. Sense-Descriptivism

Katz’s failure to distinguish abbreviational from sense-descriptivism makes his version of metalinguistic descriptivism less pure than he thinks it is. He defines descriptivism as the thesis that “proper names have a sense and that their sense is somehow necessary to fixing their reference” (1990, 33). This definition may be neutral between different conceptions of sense, including Frege’s and Katz’s, but sense however construed is a level of semantic significance along with reference. So even though Katz claims to be defining description theories “in the broadest possible way consistent with how they are generally understood in philosophy” (1990, 33), his definition does not even encompass the theory due to that arch-descriptivist Russell, who repudiated two-tiered semantics altogether. Russell held that ordinary proper names are abbreviated definite descriptions, but he denied that definite descriptions (or expressions of any other sort) have two levels of semantic significance. This was the central point of “On Denoting.”[1]

For Russell, what distinguishes both definite descriptions and ordinary proper names from genuine, “logically” proper names, like the individual constants of logic, is not that they do have senses but that they do not have references, at least not as semantic values. That is, semantically they are not referring expressions but quantificational phrases. We should not be misled by Russell’s characterization of descriptions as “denoting complexes”—they generally do have denotations—because for Russell denotation is a semantically inert property. That is, the proposition expressed by a sentence in which a description occurs is the same whether the description has a denotation or not. So its denotation does not enter into that proposition.[2] As Russell explains,

The actual object (if any) which is the denotation is not …a constituent of propositions in which descriptions occur;[3] and this is the reason why, in order to understand such propositions, we need acquaintance with the constituents of the description, but do not need acquaintance with its denotation.[4] (1917, 222)

Thus, for any sentence containing a definite description, grammatical form is misleading as to logical form. According to Russell’s famous theory of descriptions, a simple subject-predicate sentence of the form ‘The F is G’ does not express a singular proposition, of the subject-predicate form ‘a is G,’ but a general proposition, specifically what I will call a ‘uniqueness proposition,’ whose quantificational structure is revealed only after the definite description is “broken up,” to yield (in modern notation) the form ‘($x)(("y)(Fy º y=x) & Gx),’ in which the description, not being a semantic unit, does not even appear.[5] Similarly, the bearer of a name does not enter into the proposition expressed by a sentence in which the name occurs. This is not because (ordinary) proper names have senses but because, for Russell, they are disguised definite descriptions. Russell’s descriptivism is abbreviational.

Frege is a descriptivist of a different kind. His view is not that every proper name is equivalent to some definite description but that expressions of both kinds are of the same semantic genus, which he calls “Eigennamen” (literally translated as ‘proper names’ but better paraphrased as ‘singular terms’). Unlike Russell, Frege does not assimilate definite descriptions to quantificational phrases but treats them, like proper names (properly so-called), as semantic units capable of having individuals as semantic values, as determined by their senses. The sense of such an expression plays the semantic role of imposing a condition that an individual must satisfy in order to be the referent. A proper name, like a definite description, contributes its sense to that of a sentence in which it occurs regardless of which individual it belongs to and even if it has no reference. This is because the condition imposed by sense, the determinant of reference, is independent of that which it determines. So, for example, “the thought remains the same whether ‘Odysseus’ has reference or not”[6] (Frege 1892, 63).

Thus neither Frege’s sense-descriptivism nor Russell’s abbreviational descriptivism is susceptible, as Mill’s view is, to the problem of names without bearers. On both views, a proper name can play its semantic role whether or not it belongs to anything. This is so for different reasons, of course. For Russell, the reason is the semantic inertness of denotation, whereas for Frege it is the independence of sense from reference.

Having a sense does not entail being equivalent to a definite description and being equivalent to a definite description does not entail having a sense—the two kinds of descriptivism are logically independent. Yet it might seem that sense-descriptivism entails the abbreviational kind, especially considering that Frege often illustrates senses of proper names by means of definite descriptions. In fact, however, Frege’s conception of sense does not entail that every proper name has the sense of some definite description, or that the sense of every proper name is an individual concept expressible by some definite description. His conception of sense leaves open the possibility of non-descriptive senses, such as percepts. In this regard, it would seem that thinking of an object by means of a percept, as one does when visually attending to it, is not equivalent to thinking of it under a description of the form ‘the thing that looks thus-and-so.’[7] One might verbally express a thought about an object one is looking at by saying something of the form, ‘the thing that looks thus-and-so is …,’ but, as Frege says in connection with indexical thoughts generally, “the mere wording ... does not suffice for the expression of the thought” (1918, 24). He does not explicitly make the analogous point in regard to proper names, but nowhere does he explicitly assert that each proper name is equivalent to some definite description, and his overall theory of sense and reference does not require this equivalence.

Although abbreviational and sense-descriptivism are logically independent, it is common to equate the two, and Katz is not the only one who does. Kripke, for example, alludes to “the Frege-Russell view” or “the theory of Frege and Russell” (1980, 53 and 60) as if both held the same view, and he seems to equate a name’s having a sense with its being synonymous with a description (1990, 58-59). However, the distinction between the two kinds of descriptivism should be maintained, and one should not assume that valid objections to one apply to the other as well. For the same reason, a distinction should be drawn between descriptivism at the level of language and descriptivism at the level of thought.

2. Semantic vs. Epistemological Descriptivism

When Katz initially characterizes Millianism as the view “that names have no sense (linguistic meaning)” (1994, 1; my italics), it is clear not only that he is contrasting it with sense-descriptivism but that, in equating sense with linguistic meaning, he is concerned with descriptivism at the level of language. Accordingly, we ought to distinguish this semantic level of descriptivism from descriptivism at the level of thought (epistemological descriptivism). Neither entails the other.

It cannot be denied that Frege and Russell, in their respective ways, are descriptivists at both levels. For Frege, the sense of an expression is that “wherein the mode of presentation is contained” (1892, 57). The same object can be presented in different modes (ways). Russell disallows this: objects that can be presented at all cannot be presented in different ways. Russell’s restrictive notion of acquaintance is a “direct cognitive relation” and, indeed, is “simply the converse of the relation of object and subject which constitutes presentation” (1917, 202). Notoriously, Russell disqualifies public objects as objects of acquaintance, but this is the price he is willing to pay to get around Frege’s problem of identity statements and to avoid Frege’s solution based on senses. The notion of sense, as the determinant of reference, has no place in Russell’s theory of language or thought. Constituents of propositions are individuals (particulars and universals), and the Principle of Acquaintance requires that “every proposition which we can understand must be composed wholly of constituents with which we are acquainted”[8] (1917, 211). For Frege modes of presentation are the constituents of thoughts, and the objects which modes of presentation present are not. Because the relation between subject to object is mediated by a sense, this relation is indirect, unlike Russellian acquaintance.[9]

So the difference between Frege’s two-tiered and Russell’s one-tiered semantics mirrors their different epistemological views on presentation. In particular, they are, in their respective ways, descriptivists about singular thought as well as about proper names. However, a descriptivist about proper names need not be a descriptivist about singular thought. One could hold that names are abbreviated descriptions without holding, as Russell did, that thinking of an object by name is to think of it under a description. One need not accept Russell’s view that (public) objects do not enter into propositions that we are capable of entertaining and that the propositions that seem to be about them are really general propositions (to the effect that a unique object of a certain sort has a certain property). Equally, one could hold that names have reference-determining senses without holding, as Frege did, that thoughts about objects involve senses (individual concepts), which determine an object by fixing a context-insensitive condition of satisfaction. Neither abbreviational nor sense-descriptivism about proper names has to be saddled with a descriptivist view of singular thought. I will illustrate this in connection the metalinguistic version of abbreviational descriptivism that I will be defending later.

Metalinguistic descriptivism, as a thesis merely about the meaning of names, does not imply that when one refers to an individual by name (in uttering a sentence of the form ‘N is F’), one is communicating a general (uniqueness) proposition, of the form ‘the one and only bearer of “N” is F.’ Ordinarily, using a name is just a convenient way for us “to indicate what we are speaking about” (Russell) by making its bearer “the subject of discourse” (Mill), and metalinguistic descriptivism, not being an account of use, does not imply otherwise. It is compatible with the fact that statements made with sentences containing a proper name generally communicate a singular proposition about the bearer of the name (or, if the name has many bearers, about a particular bearer of the name). For this fact concerns not the semantics of names but their use in communication, and, as I will suggest later, it is this fact to which the intuitions supporting the Millian and other direct-reference views of proper names are sensitive.

So far we have the distinction between the general (uniqueness) proposition determined by the linguistic meaning of a sentence of the form ‘N is F’ and the singular proposition expressed in the utterance of that sentence. What about the thought in the mind of the speaker? Its content cannot be a singular proposition, of the form ‘a is F,’ for then one could not consistently think of the same object that it is not F. But one can do this, e.g., consistently think that London is pretty and that London is not pretty. This is possible, according to Frege, if one thinks of the object under distinct modes of presentation, under one that it is F and under the other that it is not F.[10] Now if these modes of presentation were expressible by definite descriptions, the contents of the two thoughts would be in effect that the object that satisfies one descriptive condition has a certain property and that the object that satisfies the other descriptive condition lacks that property, in which case the contents of “singular” thoughts would really be general propositions.[11] If a thought is truly singular, its object cannot be thought of as that which satisfies a certain description. So we should not assume that singular modes of presentation are expressible by definite descriptions. Let us suppose instead that the object of a singular thought is determined relationally, as the individual that stands in one of a certain class of relations to the token of the thought in question.[12] The (nondescriptive) mode of presentation of the object determines what relation that is—it can involve perception, memory, or communication—but the object is not represented as the thing which stands in that relation. These nondescriptive modes of presentation, which do not determine an object apart from a context of thought, function as mental indexicals, and include the mental counterparts of proper names.[13]

So there is the meaning of the sentence (of the form ‘N is F’), what the speaker communicates in using the sentence, and the thought in the mind of the speaker. The meaning of the sentence expresses a general (uniqueness) proposition, of the form ‘the bearer of “N” is F,’ what is communicated is a singular proposition of the form ‘a is F,’ and the thought in the mind of the speaker is an indexical thought, true or false, relative to the context, with respect to the object in question. The hearer forms a thought of the same sort, though he need not think of the object under the same nondescriptive mode of presentation—what matters is getting the object right.[14]

The idea, then, is that the communication of singular thoughts does not require the communication of singular senses, Fregean or otherwise.[15] What is communicated is merely a singular proposition, which is an abstraction from both the speaker’s and the hearer’s (or hearers’) singular thoughts regarding the object in question (that it is F). The sentence (‘N is F’) used to communicate a singular thought does not encode the full content of either the speaker’s thought or the thought to be formed by the hearer, but it does not encode a singular proposition either. The sentence itself encodes merely a general, uniqueness proposition. So, for example, the sentence ‘David Kaplan is clever’ encodes the uniqueness proposition that the bearer of ‘David Kaplan’ is clever. The bearer of the name, the man, enters only into what is communicated in the use of the sentence.[16] However, this singular proposition, that David Kaplan is clever, is not the content of the singular thought entertained by either the speaker or by the hearer, who must each think of Kaplan in one way or another, but not necessarily in the same way.

Interestingly enough, both Russell and Frege recognized that the speaker’s and the hearer’s ways of thinking of the referent need not be the same. Russell allowed that “the description required to express the thought will vary for different people, or for the same person at different times. … But so long as [the object to which the name applies] remains constant, the particular description involved usually makes no difference to the truth or falsehood of the proposition in which the name appears” (1917, 208). And Frege allowed that for “a proper name such as ‘Aristotle’ opinions as to the sense may differ, … [but] so long as the reference remains the same, such variations of sense may be tolerated” (1892, 58n.). Russell’s and Frege’s theories differ, of course, in how they provide for these possibilities.

In sum, descriptivism about proper names does not require a descriptivist view of singular thought and its communication. In particular, the metalinguistic version of abbreviational descriptivism is compatible with the view that singular thought is indexical and that what is communicated in the expression of a singular thought is a singular proposition.

3. On Sense and Reference: Frege, Katz, and Kaplan

As noted earlier, in defining description theories as claiming that “proper names have a sense and that their sense is somehow necessary to fixing their reference” (1990, 33), Katz erroneously excludes Russell’s abbreviational view. As we will see now, he also inadvertently includes Kaplan’s direct-reference view. In addition, he gives the false impression that even though his view of sense is different from Frege’s, his sense of ‘sense’ is the same as Frege’s.

Katz’s definition excludes Millianism but not direct-reference theories in general. According to Mill, “proper names are not connotative: they denote the individuals who are called by them; but they do not indicate or imply any attributes as belonging to those individuals”[17] (1872, 20). However, a direct-reference theory need not be Millian, and certainly Kaplan’s is not. While maintaining that only the referent enters into the proposition expressed by a sentence in which a proper name occurs, Kaplan explicitly states that the referent is fixed by a sense.[18] It is not Kaplan’s view that the reference of a name must be fixed by the content of a definite description, but a direct-reference theorists could consistently hold this—with the proviso that the descriptive content does not enter into the proposition expressed. Such a view does not imply that names have Fregean senses, but it is not Millian either. However, when Katz describes direct-reference theory as holding that “the reference of a name is not mediated by a sense” (1990, 34), he is equating it with Millianism. This eases his task of defending descriptivism, allowing him to award points to descriptivism by scoring points against the Millian thesis that there is nothing more to the meaning of a name than its bearer.

Katz’s conception of sense is quite different from Frege’s. For Katz, senses are properties that expressions have in virtue of which they can be synonymous, ambiguous, redundant, contradictory, hyponymous, antonymous, etc. These are the properties and relations that concern lexical semantics (Cruse 1986), as opposed to philosophical semantics. Katz takes sense to be an intralinguistic property, one that does not concern, at least not directly, the relationship between language and the world. In his view, sense should be defined not as “a determiner of referential properties and relations …but [as] a determiner of sense properties and relations” (1994, 9). Sense merely “mediates” reference. Armed with this concepton, Katz chides Frege for defining sense as the determiner of reference. Such a definition, Katz writes, is like giving “a definition of pronunciation that is not formulated in terms of phonological features or a definition of number that is not formulated in terms of arithmetic features” (1994, 9).

Given his conception of sense, it is no wonder that Katz takes the main question in the debate between Fregeans and Millians to concern whether or not proper names have linguistic meaning. However, it is a mistake to think that Frege equates sense with linguistic meaning, as Tyler Burge pointed out some time ago (1979, 398-407). Burge cites various passages in which Frege says, in regard either to indexicals or to names themselves, that they can have different senses associated with them in different contexts.[19] Frege gives no indication that what is variable is linguistic meaning.

Whereas for Katz sense constitutes linguistic meaning, for Frege sense plays a rather different role. Indeed, as is well-known, Frege casts sense in three roles: as (S1) determinant of reference, as (S2) “mode of presentation” (of reference), and as (S3) “indirect” (as opposed to “customary”) reference of an expression embedded in a context of indirect quotation or propositional attitude ascription.[20] S1 is one of two levels of semantic significance (along with reference), and S2 is often described as cognitive significance (Frege uses the phrase “cognitive value”). Frege agrees with Russell, and with Mill for that matter, that words are ordinarily used (i.e., when S3 does not come into play) to talk about things, not ideas: “If words are used in the ordinary way, what one intends to speak of is their reference” (1892, 58). Even so, we must associate reference-determining properties (S1) with our words. Moreover, since our words also express our thoughts, they must correspond to constituents of those thoughts (S2). Thus, for Frege, the semantic and the cognitive significance of expressions are intimately related. Indeed, because sense is possible without reference—as the determinant of reference it is independent of that which it determines—Frege holds that the constituents of thoughts are senses, not references.

Katz departs from Frege by denying that sense determines reference. He is not repeating John Perry’s (1977) complaint that Frege’s notion of sense cannot handle indexical (or demonstrative) reference.[21] Katz’s point has nothing to do with indexicality, for he is rejecting the very conception of sense that would even suggest that sense determines reference. Like Chomsky (1995), he sees reference not as a semantic relation but as belonging to pragmatics. In Katz’s view, “the study of languages concerns the grammatical structure of sentence types, and the study of language use concerns the pragmatic structure of tokens of sentence types” (1994, 2). Semantic structure is included in grammatical structure, but the properties and relations represented in semantic structure are all intralinguistic. They do not include referential relations. In this respect, Katz’s conception of semantics is clearly different from that of most philosophers of language, who generally take semantics, whatever its specific format, to concern reference and truth conditions. There is no need to assess Katz’s general conception of semantics here, for it will have no bearing on our discussion of proper names in particular. Regardless of the details of that conception, it is clear that when Katz contrasts his (sense-) descriptivism with Frege’s, he and Frege are not disagreeing about the same thing, since Frege does not equate sense with linguistic meaning.

Construing them as linguistic meanings, Katz takes senses to be the properties expressions have in virtue of which they can be synonymous, ambiguous, redundant, etc. He thinks this conception of sense provides an immediate benefit to the semantics of proper names: “Since … sense is required to be just informationally rich enough to explain the sense properties and relations of expressions, descriptivists are no longer forced to stuff senses full of reference-fixing information” (1994, 10). There is no need to define ‘Aristotle,’ for example, in terms of substantive descriptions like ‘the teacher of Alexander’ or ‘Gödel’ as ‘the prover of the incompleteness of arithmetic,’ and thereby fall prey to the familiar counterfactual arguments advanced by Kripke. Rather than invoke identifying properties, Katz’s descriptivism appeals only to the “purely nominal bearer relation that makes a sense metalinguistic” (1994, 7). In my view, metalinguistic descriptivism can be made even purer, by not appealing to sense and simply treating proper names as abbreviations of definite descriptions that mention them.

4. Going Metalinguistic

Names without bearers create an embarrassment for Millian and other direct-reference accounts, because such names cannot contribute references to propositions expressed by sentences in which they occur and therefore cannot play their semantic role. For if a sentence containing a proper name must express a singular proposition, a proposition which is contingent on the existence of an object, then if the name is empty there is no such proposition for it express.[22] Yet names without bearers are perfectly meaningful, and sentences in which they occur do express propositions. Indeed, a sentence like ‘Santa Claus does not exist’ is not only meaningful but true.

Description theories are designed to avoid the problem of empty names, but they face a different problem, one of indeterminacy. This problem is that, in general, there is no one property (or one set of properties) that is uniquely associated with a name, even if the name has only one bearer.[23] As both Donnellan (1970) and Kripke (1980) point out, speakers (even the same speaker) can associate different properties (or sets of properties) with a given individual, hence different descriptions (or senses) with a name of that individual. Yet names are not indeterminate in meaning in this way. For example, even though people associate various descriptions with the name ‘Aristotle’ (considered just as the name of the ancient philosopher), it seems perfectly determinate in meaning.

The problem of indeterminacy does not arise for metalinguistic descriptivism, since it does not claim that substantive definite descriptions (or senses) are associated with a proper name. In the case of the name ‘Aristotle,’ for instance, it does not force a choice between descriptions like ‘the author of The Metaphysics’ and ‘the teacher of Alexander.’ The only property it invokes is the property of bearing the name in question, and there is nothing equivocal about that. In this regard, it separates the meaning of the name from the various ways in which the speaker (or the hearer) might think of the name’s bearer.

The metalinguistic view does not require that the primary way in which a user of the name thinks of the referent is as the bearer of the name. Referring to an individual by name is generally a matter of convenience, especially when one’s audience is familiar with that individual (otherwise an introduction is needed) or when the name has already been used in the conversation, but otherwise the property of bearing a certain name is not particularly important or salient and does not enter into what is communicated. In this respect, typical referring uses of proper names are like referential uses of definite descriptions.[24]

On the other hand, there are cases in which the property of bearing a certain name is part of what the speaker is trying to convey. The metalinguistic view straightforwardly explains such a use of a name, which is like the attributive use of a definite description. For instance, suppose one sees the name ‘Aaron Aardvark’ at the beginning of a telephone directory and, contemplating the benefits of being first in alphabetic order, says “Aaron Aardvark (whoever he is) is a lucky guy.” This use is clearly attributive, not referential, in that the property of bearing the name, rather than the bearer of the name, is part of what the speaker means. That property also enters in when someone is identified by name (“That is Madonna”) or is introduced (“This is Steve Jones”), and the metalinguistic view accounts for these cases as well.

A crucial point about the metalinguistic view is that in saying that a name expresses the nominal property of bearing that very name, it is making a generic claim about proper names. Thus this claim takes a schematic form, on my formulation for example, that ‘N’ is semantically equivalent to ‘the bearer of “N”.’ As a result, the proposed schema applies routinely to familiar and unfamilar names alike. So no special semantic knowledge is required for learning a new name, for it is by virtue of one’s general knowledge about the category of proper names that one knows of any specific name that it expresses the property of bearing that name. The fact that the schema applies to proper names as a class explains is why the metalinguistic theorist can hold, for example, that the name ‘Socrates’ means ‘the bearer of “Socrates”,’ without being committed to holding that the word ‘horse’ means ‘thing called “horse”.’ Unlike common nouns, proper names are not lexical items in a language.

Understood in this way, the metalinguistic view is not vulnerable to Kripke’s well-known circularity objection. He rightly insists that a theory of proper names must avoid using any “notion of reference in a way that is ultimately impossible to eliminate” (1980, 68). He then objects that if “we ask to whom does [a speaker] refer by ‘Socrates’, … the answer is given as, well, he refers to the man to whom he refers” (1980, 70). In fact, however, bearing a name is not the same property as being referred to by that name. Kripke regards the metalinguistic theory as circular because he takes the fact that we use a name to refer to its bearer to be trivial. But this is not a trivial fact. Although it is certainly more convenient to refer to people by their own name, we could refer to them instead by their maternal grandmother’s name or even by their Social Security number. It is no more essential to the property of bearing a certain name that one be referred to by that name than it is essential to the property of having a certain Social Security number that one be referred by that number. Bearing a name and being referred to by that name are distinct properties. The metalinguistic approach is not the “theory of reference” that Kripke takes it to be.[25]

The metalinguistic view has a number of interesting and plausible consequences, which go unexplained on other views of proper names:

1. A use of a name to refer to an individual that does not bear the name is not literal.

2. Having more than one bearer does not entail that a name has more than one linguistic meaning.

3. Distinct names of the same individual are not synonymous.

4. Names do not belong to particular languages and are not translatable.

Let us take these up in order.[26]

1. A name intentionally used to refer to an individual that does not bear it is not being used literally. For instance, the name ‘Elvis’ might be used to refer to a local singer of old rock-and-roll songs. Katz gives the example of the name ‘Hitler’ being used to refer to a tyrannical college dean (1994, 11). Since the persons in question are assumed not to bear these names (not even as nicknames), such uses are not literal. A metalinguistic theory, which invokes the property of bearing the name, straightforwardly explains how names can be used nonliterally in this way. The speaker intends to be referring not to the bearer of the name but to someone saliently similar to the (famous) bearer of the name.

2. By citing only the nominal property of bearing the name, the metalinguistic view avoids the implication that names with many bearers are semantically ambiguous. Substantive description theories, as well as direct-reference theories, imply that a name like ‘Salem’ or ‘Sally’ has as many linguistic meanings as it has bearers.[27] They would suggest that being ignorant of all the towns named ‘Salem’ or all the people named ‘Sally’ is a deficiency in linguistic knowledge. And substantive description theories imply that a name is even ambiguous with respect to a single bearer, in as many ways as there are descriptions associated with that individual. ‘Aristotle’ is a familiar example.

3. Because bearing the name ‘M’ is distinct from the property of bearing the name ‘N’, the metalinguistic view entails that distinct names cannot be synonymous. It thereby explains why ‘Dylan bears “Dylan”,’ as Katz says, “smacks of redundancy” (1990, 37), whereas ‘Zimmerman bears “Dylan” ’ does not. Similarly, it explains why statements of the form ‘M is N’ are informative in the sense of not being redundant (such a statement may be uninformative in the irrelevant sense of being old news to someone who knows that M is N, e.g., to someone who knows that Robert Zimmerman is Bob Dylan). Substantive description theories imply that distinct proper names can be synonymous (by being associated with the same definite description), hence that identity statements containing two synonymous names can be uninformative (of course Millians would not find this implication objectionable, but their commitment to the synonymy of co-referring names also commits them to the meaninglessness of empty names).

4. For similar reasons, the metalinguistic view implies that proper names are not translatable. This is not because their meanings are ineffable but because, strictly speaking, they do not belong to particular languages. They do, of course, have pronunciations and spellings characteristic of particular languages, and they do have counterparts with pronunications and spellings characteristic of other languages, but these counterparts are not translations of one another. Consider the name ‘John,’ for example, and its counterparts ‘Juan,’ ‘Johann,’ ‘Jean,’ and ‘Ian.’ Despite their distinctive pronunications and spellings, each of them can can be used without anomaly (or italics) outside its home language.[28] For example, if you wish to speak in English about your Spanish friend ‘Juan,’ you do not switch to ‘John,’ and in writing you do not use italics. Proper names have phonological and orthographic properties characteristic of particular languages, but this does not make them lexical items in those languages. If proper names were lexical items in a language, then not knowing a particular name would constitute a linguistic deficiency. But it does not. Learning a new name is not like acquiring a new word in one’s vocabulary (does learning ‘Dweezil’ and ‘Moon Unit,’ the names of the late Frank Zappa’s children, add to one’s knowledge of English?). As noted above, the metalinguistic view denies that distinct bits of semantic information must be learned for each proper name. It proposes a schema instead, applicable to names generally, including unfamiliar ones.

An apparent difficulty for metalinguistic descriptivism is that a name often has more than one bearer. ‘Salem’ is such a name—there is no such town as the bearer of that name. Accordingly, one might consider going metalinguistic without going descriptivist. Alternative formulations would also advert to the property of bearing a certain name but would not imply uniqueness. One possible version, not that it has ever been suggested, is that a proper name is equivalent to a metalinguistic indefinite description, of the form ‘a bearer of “N”.’ This suggestion certainly allows for multiplicity of bearers, but it is too weak. It falsely predicts that a sentence like ‘John is tall’ could be used to make merely the existential assertion that at least one bearer of ‘John’ is tall. The use of proper names is never as nonspecific as that.[29]

More plausible is the suggestion that a proper name is a kind of indexical or demonstrative, with fixed meaning and variable reference. The idea is that a name ‘N’ is equivalent to the metalinguistic demonstrative description, ‘that bearer of “N”.’ However, names belong to individuals quite apart from context of use. In contrast, indexicals and demonstratives do not belong to individuals at all—they are simply devices for referring. This difference makes trouble François Recanati’s formulation of the indexical theory: “the meaning of a proper name NN refers the hearer to a relation which holds in context between the name and its reference, namely the name-bearer relation”[30] (1993, 140-1). The trouble is that the name-bearer relation is not context-sensitive at all—a name bears this relation to all of its bearers, not just to the one the speaker is using it to refer to in the context. What is context-sensitive is the speaker-referent relation, but this is not the relation invoked by Recanati’s rule. The mere fact that proper names are used to refer to contextually identifiable individuals does not mean that they are like indexicals and demonstratives. However, it does raise the question, which an adequate version of metalinguistic descriptivism must answer, of how this context-sensitivity is to be accounted for.

5. PMT and NDT Compared

Exactly how should metalinguistic descriptivism be formulated? Should it assign senses to names? And how does it propose to accommodate the fact that a name with many bearers can be used to refer to one of the name’s bearers in particular? Katz and I agree that the only property it should advert to is the nominal property of bearing a certain name but that it is beyond the scope of semantics to give an account of the “sociological” property of bearing a certain name. An explanation of that would have to reckon with the fact that how a name is acquired can vary, depending on the type of name: given name, nickname, stage name, geographical name, brand name, etc. As Katz emphasizes, it is important to keeps the metalinguistic theory “pure” (the ‘P’ in ‘PMT’):

[by] adulterating the bearer condition in the senses of names with real properties and relations [e.g., being called ‘N’], impure metalinguistic description theories forfeit the advantage of having gone metalinguistic in the first place. [This] makes the theory vulnerable to the very counterfactual arguments that refute straightforward Fregean versions of the description theory. (1994, 7)

Katz and I both try to keep our accounts untainted.

As Katz formulates PMT, a proper name ‘N’ has the sense of

DDPMT: ‘the thing which is a bearer of “N” ’

My “nominal description theory” also invokes the nominal bearer relation, but with the simpler schema

DDNDT: ‘the bearer of “N” ’[31]

Despite the fact that DDPMT is more complicated, it seems to be merely a notational variant of DDNDT. The two seem to differ only syntactically, like ‘the thing which is a bottle’ and ‘the bottle.’ So why does Katz opt for the more elaborate DDPMT, with its relative clause containing an indefinite description? He does this to accommodate “the linguistic fact that a name is not limited to a single bearer”[32] (1994, 17). The elaborate “sense structure” of DDPMT is supposed to do justice to the “division of labor” involved in the use of names. As Katz describes this division, “the objects in the domain of the language are first filtered by the bearer condition, then by contextual knowledge of the name-bearer correlations, and finally by descriptive information introduced to make the referent contextually definite” (1994, 21).

Now if this division of labor were linguistically marked, then by parity of reasoning an ordinary (nonmetalinguistic) incomplete definite description like ‘the bottle’ would be marked similarly: its semantic structure would take the form, ‘the thing which is a bottle.’ Yet there is no syntactic motivation for supposing that. Worse, this supposition does not really help explain how ‘the bottle’ can be used to refer to a specific bottle. If one utters “The bottle is empty,” using the description ‘the bottle’ to refer to a certain bottle, say the bottle one is examining, one is using ‘bottle’ to mean bottle and ‘the’ to imply uniqueness. However, one obviously does not mean that there exists only one bottle; what one is saying, strictly and literally, does not make fully explicit what one means. But neither does the more elaborate, ‘The thing which is a bottle is empty.’ With either utterance, one means something regarding the contextually relevant bottle, but its being contextually relevant is not part of what one says. Adding the relative clause with its indefinite description does not help explain the specificity of the reference. Similarly, the use of a proper name with many bearers does not make explicit which of its bearers one is referring to, and Katz’s DDPMT with its relative clause and indefinite description does not pick up the slack.[33] So, for example, one can use the name ‘Gareth Evans’ to refer to the late English philosopher rather than to the Australian politician, but this is not explained by supposing that the sense of the name is given by ‘the individual who is a bearer of “Gareth Evans”.’ DDPMT makes a futile attempt to incorporate into sense elements that are merely pragmatic and relevant to use, not meaning.

Katz goes further and attributes additional sense structure to particular tokens of a name, in order to explain the token’s reference to a particular bearer of the name type. Each token of a name with many bearers has a sense in its own right, according to Katz, and “the sense of literal tokens derives from the sense of their type” (1994, 23). But he does not explain how. He needs to say how the sense of the type gets enriched into the sense of the token, so that its additional properties can effectively constrain the reference. Nor does he explain what these properties could be. For if sense properties (and relations) are characteristics like ambiguity, synonymy, redundancy, and antonymy, they could belong only to expression types, not tokens. Moreover, it does not seem that proper names, considered just as types, even have such characteristics. What, for example, could count as a synonym of ‘Cincinnati’ or as an antonym of ‘Mississippi’? All in all, it does not seem that the notion of sense can play the role for which Katz casts it. Indeed, it seems that what makes Katz’s PMT metalinguistic is not that it assigns senses to names but that, in the fashion of Russell, it treats them as short for definite descriptions.

NDT is unequivocal about this—it makes no reference to sense and claims only that a name ‘N’ is semantically equivalent to ‘the bearer of “N”.’[34] It does not assign to this schema the job for which Katz employs the more elaborate ‘the thing which is a bearer of “N”,’ a job which cannot be performed at the semantic level alone. NDT relies on the structurally simpler DDNDT for limited semantic duties; a separate pragmatic account is required, should ‘N’ have many bearers, to explain how ‘N’ can be used to refer to one of its bearers in particular. Reference to a contextually definite bearer of ‘N’ is assimilated to the more general phenomenon of using an incomplete definite description to refer to a specific individual, a phenomenon that does not have to be represented at any linguistic level. NDT does not confuse the definiteness of a proper name with the specificity of the reference made in using it. No one would suggest that just because there are many bottles and the description ‘the bottle’ is ordinarily used to refer to a certain one of them, that this description has the underlying form, ‘the thing which is a bottle.’

In this regard, it is no objection to Russell’s theory of descriptions that an incomplete definite description can be used successfully to refer even when it is obviously not satisfied by anything uniquely.[35] Using such a description does not commit the speaker to the existence of one and only individual of the relevant sort. Consider the use of the description ‘the bottle,’ as it occurs in the sentence ‘The bottle is empty.’ There may be some completion of the description (e.g., ‘the whiskey bottle in my hip pocket’) that the speaker intends the hearer to read into the utterance. Or, there may be some contextually identifiable basis (e.g., visual salience) for singling out the relevant individual. Russell’s theory does not require that when one utters ‘The bottle is empty,’ one means precisely what the sentence means. One’s utterance need not be taken strictly and literally. In this regard, a definite description is like any other quantified noun phrase, in that it can be used with an implicit restriction on its domain. But this should not lead us to suppose that the syntactic or semantic structure of the description contains a slot that must be filled in by a specification of the means by which the intended referent is contextually determined, or by a contextual restriction on the domain of quantification or “universe of discourse.” The suggestion would be that the description has underlying form ‘the ___ bottle’ or ‘the bottle which is ___,’ where the blank must be filled by a contextually undertood completer of the description. The trouble with this suggestion is its implicit assumption that the specificity of the reference must be marked at some linguistic level. Inferring such a completion of (or restriction on) the description may help the hearer identify the referent, but this completion is not part of the strict and literal content of the utterance.

By exploiting the distinction between what is literally and strictly expressed by a sentence containing a proper name and what is conveyed in its typical use, metalinguistic descriptivism can answer the complaint that it implies that when we use sentences containing a name with many bearers, “we only rarely mean what we say” (Taschek 1990, 40), hence that it postulates a “massive amount of semantic idleness” (Kobes 1991, 471). For what Kobes calls semantic idleness is really pragmatic efficiency. He is wrong to suggest that we should have expected “such standardized non-literality to drift into the semantics of the language, just as misuse and metaphor become semantically correct and literal with standardization.” We should not expect such inexplicitness to be encoded, for it is not attributable to particular constructions and occurs with a great variety of different ones. Inexplicitness is a general phenomenon.[36]

Here is one way to think of what happens when we use an incomplete definite description or a shared proper name: we do not imply uniqueness but merely pretend (for the sake of discussion) that there is a unique satisfier of the description or a unique bearer of the name. This pretense of uniqueness is analogous to other pretenses involved in certain uses of language. For example, as Mark Crimmins has suggested, when we attribute to Lois Lane the belief that Superman can fly and that Clark Kent cannot, “we pretend that there are different individuals to talk about” (1995, 13). Then there is the pretense of existence we engage in when talking about fictional characters. As Gregory Currie (1990) argues, such talk does not require explanation in terms of specifically fictional language. Rather, it is a special use of ordinary language. The same goes for the use of definite singular terms that do not apply uniquely.

The pretense model will perhaps mislead. It is just a way of putting the idea that the uses of names with many bearers may be assimilated to referential uses of incomplete definite descriptions. This idea was first advanced by Brian Loar (1976, 370-73), who proposed that standard uses of names like ‘Jones’ and descriptions like ‘the table’ include individuating concepts that are not expressed by their conventional meanings (their conventional meanings convey only non-individuating properties, such as being called ‘Jones’ and being a table). How this occurs has a Gricean explanation. A speaker who says ‘Jones is a psychoanalyst’ or ‘The table is bare’ exploits the mutual belief, shared by him and his audience, that there are many bearers of ‘Jones’ or many tables. He can intend and reasonably expect them to rely on this mutual belief to figure out that he must be referring to a certain bearer of ‘Jones’ or to a certain table, identifiable under the circumstances of utterances on the supposition that it is intended to be identified.

Once it is understood that the standard use of proper names—to refer to particular bearers—is not to be explained by their conventional meaning alone, we can see why the usual objections to substantive description theories do not apply to the metalinguistic version. As Donnellan (1970) and Kripke (1980) have shown, there are serious problems with the claim that the name ‘Gödel,’ for example, is synonymous with, say, ‘the discoverer of the incompleteness of arithmetic,’ especially if it asserts that using the name to refer to Gödel requires thinking of him under that description. Also, this claim misclassifies as analytic the obviously synthetic sentence, ‘Gödel was the discoverer of the incompleteness of arithmetic,’ and it rules out real metaphysical possibilities, such as that someone other than Gödel was the discoverer of the incompleteness of arithmetic. However, analogous objections do not apply to metalinguistic descriptivism. The claim that the name ‘Gödel’ is semantically equivalent to ‘the bearer of “Gödel” ’ does entail that the use of this name conveys the property of bearing the name, but it does not entail that a user of that name think of Gödel under the description, ‘the bearer of “Gödel”.’ It does entail that the sentence, ‘Gödel bore the name “Gödel”,’ is analytic, but it does not entail that bearing the name was a necessary property of the man Gödel. Metalinguistic descriptivism avoids the usual objections by exploiting the distinction between the meaning of the name and its use to refer. In uttering, ‘Gödel discovered the incompleteness of arithmetic,’ a speaker is not likely to mean that the bearer of ‘Gödel’ discovered incompleteness. He is likely to be communicating a singular proposition about Gödel.[37]

6. Summing Up

When descriptivism is distilled down to its essence, it is nothing more than the thesis that a name ‘N’ is semantically equivalent (not syntactically or pragmatic equivalent) to a metalinguistic definite description of the form ‘the bearer of “N”.’ To explain the use of a name with many bearers to refer to a particular one of its bearers, it must be augmented by a pragmatic account that parallels the explanation of referential uses of incomplete definite descriptions. Such an account exploits the distinction between the linguistic meaning of a sentence, the thought in the mind of the speaker, and what is communicated in the utterance. The meaning of a sentence containing a proper name or a definite description is a general, uniqueness proposition, but if the use is referential, the speaker’s thought is indexical in character, involving a contextual relation between the speaker and the referent. But what is communicated in such a use is a singular proposition, involving the referent rather than any particular way of thinking of it. This last fact explains the intuition on which direct-reference theories rely, that the referent is what enters into the proposition expressed, but the intuition concerns what is communicated, not literal meaning. Metalinguistic descriptivism does not imply that the property of bearing the name is part of what is communicated, and, unlike direct-reference theories, it does not imply that sentences containing names without bearers are semantically defective. Fully purified, it is a modest thesis about the linguistic meaning of names and, without attributing senses to them, claims simply that they abbreviate definite descriptions that mention them.

References

Bach, Kent. 1981. “What’s in a Name.” Australasian Journal of Philosophy 59:371-86.

Bach, Kent. 1982. “De Re Belief and Methodological Solipsism.” In A. Woodfield, ed., Thought and Object. Oxford:Oxford University Press.

Bach, Kent. 1987. Thought and Reference. Oxford: Oxford University Press.

Bach, Kent. 1994. “Conversational Impliciture.” Mind & Language 8:124-62.

Bell, David. 1990. “How ‘Russellian’ was Frege?” Mind 99:267-77.

Braun, David. 1993. “Empty Names.” Noûs 27: 449-69.

Braun, David. 1994. “Structured Character and Complex Demonstratives.” Philosophical Studies 74:193-219.

Burge, Tyler. 1977. “De Re Belief.” Journal of Philosophy 74:338-62.

Burge, Tyler. 1979. “Sinning against Frege.” Philosophical Review 88:398-432.

Burge, Tyler. 1991. “Vision and Intentional Content.” In E. LePore and R. Van Gulick, eds., John Searle and his Critics. Oxford: Blackwell.

Chomsky, Noam. 1995. “Language and Nature.” Mind 104:1-61.

Crimmins, Mark. 1995, :Quasi-singular Propositions: The Semantics of Belief Reports,” Proceedings of the Aristotelian Society, supp. vol. 69.

Cruse, D. A. 1986. Lexical Semantics. Cambridge, Eng.: Cambridge University Press.

Currie, Gregory. 1990. The Nature of Fiction. Cambridge, Eng.: Cambridge University Press.

Donnellan, Keith. 1966. “Reference and Definite Descriptions.” Philosophical Review 75:281-304.

Donnellan, Keith. 1970. “Proper Names and Identifying Descriptions.” Synthese 21:335-58.

Evans, Gareth. 1982. Varieties of Reference. Oxford: Oxford University Press.

Evans, Gareth. 1985. “Understanding Demonstratives.” In Collected Papers. Oxford: Oxford University Press.

Forbes, Graeme. 1989. “Cognitive Architecture and the Semantics of Belief.” Midwest Studies in Philosophy 14:84-100.

Forbes, Graeme. 1990. “The Indispensability of Sinn.” Philosophical Review 99:535-63.

Frege, Gottlob. 1892. “On Sense and Reference.” Reprinted in P. Geach and M. Black, eds., Translations from the Philosophical Writings of Gottlob Frege. Oxford: Blackwell, 1960.

Frege, Gottlob. 1918. “The Thought: A Logical Inquiry.” Reprinted in P. Strawson, ed., Philosophical Logic. Oxford: Oxford University Press, 1967.

Heck, Richard G. 1995. “The Sense of Communication,” Mind 104: 79-106.

Kaplan, David. 1977. “Demonstratives.” Published in J. Perry, J. Almog, and H. Wettstein, eds., Themes from Kaplan. Oxford: Oxford University Press, 1989.

Kaplan, David. 1989. “Afterthoughts.” In J. Perry, J. Almog, and H. Wettstein, eds., Themes from Kaplan. Oxford: Oxford University Press.

Kaplan, David. 1990. “Words.” Proceedings of the Aristotelian Society 64:93-119.

Katz, Jerrold J. 1990. “Has the Description Theory of Names been Refuted?” In G. Boolos, ed., Meaning and Method: Essays in Honour of Hilary Putnam. Cambridge, Eng.: Cambridge University Press.

Katz, Jerrold J. 1994. “Names without Bearers.” Philosophical Review 103:1-39.

Kobes, Bernard W. 1991. Review of Bach 1982. Philosophical Review 100:469-71.

Kremer, Michael. 1994. “The Argument of ‘On Denoting’.” Philosophical Review 103:249-97.

Kripke, Saul. 1980. Naming and Necessity. Cambridge, Mass.: Harvard University Press.

Loar, Brian. 1976. “The Semantics of Singular Terms.” Philosophical Studies 30:353-77.

Longobardi, Giuseppe. 1994. “Reference and Proper Names: A Theory of N-Movement in Syntax and Logical Form.” Linguistic Inquiry 25:609-65.

Ludlow, Peter and Stephen Neale. 1991. “Indefinite Descriptions: In Defense of Russell.” Linguistics and Philosophy 14:171-202.

Mill, J. S. 1872. A System of Logic, definitive 8th edition. 1949 reprint, London: Longmans, Green and Company.

Perry, John. 1977. “Frege on Demonstratives.” Philosophical Review 86:474-97.

Neale, Stephen. 1990. Descriptions. Cambridge, Mass.: MIT Press.

Neale, Stephen. 1993. “Term Limits.” In J. Tomberlin, ed., Philosophical Perspectives 7, Atascadero, CA: Ridgeview Press.

Recanati, François. 1993. Direct Reference: From Language to Thought. Oxford: Blackwell.

Russell, Bertrand. 1905. “On Denoting.” Reprinted in R. C. Marsh, ed., Logic and Knowledge. London: George Allen and Unwin, 1956.

Russell, Bertrand. 1918. “Descriptions and Incomplete Symbols.” Lecture VI of “The Philosophy of Logical Atomism.” Reprinted in R. C. Marsh, ed., Logic and Knowledge. London: George Allen and Unwin, 1956.

Russell, Bertrand. 1917. “Knowledge by Acquaintance and Knowledge by Description.” In Mysticism and Logic, paperback edition. Garden City, NY: Doubleday, 1957.

Russell, Bertrand. 1919. Introduction to Mathematical Philosophy. London: George Allen and Unwin.

Searle, John. 1991. “Response: Reference and Intentionality.” In E. LePore and R. Van Gulick, eds., John Searle and his Critics. Oxford: Blackwell.

Taschek, William W. 1987. “Content, Character, and Cognitive Significance.” Philosophical Studies 52:161-89.

Taschek, William W. 1990. Review of Bach 1987. Journal of Philosophy 87:38-45.

Footnotes