Morgiana chalked all the other houses in a similar manner, and defeated the scheme: how? simply by obliterating the difference of appearance between that house and the others. The chalk was still there, but it no longer served the purpose of a distinctive mark.

J.S. Mill, Of Names (1843, p. 37).

In the Arabian Nights, the robber puts a chalk mark on the house containing booty with an intention to identify it later. In his classic discussion of proper names, Mill contends that such a scheme of putting a chalk mark on a house is analogous to the practice of giving a name to an individual—a proper name is ‘an unmeaning mark’ that simply stands for an object without connoting anything about it. Mill’s eponymous view has become the present-day orthodoxy on the semantics of names. In his discussion, however, Mill takes the analogy further—Mill notes that Morgiana successfully defeated the robber’s scheme by putting the sameFootnote 1 chalk mark on all the other houses. If names are unmeaning marks, then giving the same name to two or more individuals should have the same defeating effect on the scheme of naming that Morgiana’s ingenious duplication of chalk marks had on the robber’s scheme.

Yet, names are recklessly multiplied—or atleast so it seems at first glance. David Hume, David Lewis, and David Kaplan (apart from some 10 million other individuals) share the same first name, and it is not uncommon for more than one person to have the same full name (e.g., ‘David Kaplan’ and ‘Lucy O’Brien’.) Why doesn’t the institution of naming come crumbling down due to the existence of namesakes? This question is well acknowledged by defenders of the Millian view, and there is a surprising unanimity in the answer they provide for it: despite their orthographic or phonetic identity, names of distinct individuals count as distinct names. This response relies on a pre-semantic or ‘metaphysical’ account of names—often given the label ‘common currency conception’ or ‘specific’ names—on which the identity and individuation conditions of names conform to the Millian semantics.

This paper argues that the common currency conception (‘CCC’) of names comes with a heavy price that undermines the very view that motivates it. While even defenders of the Millian view admit that the CCC ‘does not agree with the most common usage’ (Kripke, 1980, p. 8), I will argue further that the CCC makes names exceptional amongst the class of linguistic expressions—if the CCC is correct, then names must have a sui-generis metaphysical nature distinct from every other type of linguistic expression. Such metaphysical exceptionalism would be justified if the Millian view had a clear, uncontested theoretical advantage over its rivals. However, in the context of a semantic debate about names in which the closest competitors of the Millian view—i.e., the Predicate view and Indexicalism—do not result in such exceptionalism, it counts as a strike against the Millian view.

Looking ahead: Sect. 1 begins by identifying some semantic commitments of the Millian view. It also points out that these commitments place a constraint on name-individuation—call it the ‘Millian individuation constraint’—that any metaphysical account of names must satisfy to be compatible with the Millian view. Section 2 gives an outline of two broad frameworks for thinking about the metaphysical nature of words and argues that the Millian individuation constraint forces one to accept a particular metaphysical picture of names—the common currency conception. Section 3 provides arguments for why accepting the common currency conception makes the Millian view less attractive than its rivals.

1 Names and tags

Suppose we randomized as many whole numbers as we needed for a one-to-one correspondence, and thereby tagged each thing. This identifying tag is a proper name of the thing. […] This tag, a proper name, has no meaning. It simply tags.

Marcus (1961, pp. 309–310).

1.1 Millian commitments

One central concern of the philosophy of language can be broadly framed as the following question: How do expressions of language relate to objects in the world? The Millian view (or ‘Millianism’) represents one manner of answering this question for one class of linguistic expressions—proper names. The linguistic nature of proper names according to Millianism is neatly captured by the image of a tag employed by Marcus in the passage quoted above and can be summarized in terms of the following three commitmentsFootnote 2,Footnote 3:

  1. 1.

    Proper names are exclusively devices of reference: There is a class of expressions in natural language—i.e., proper names—which function like tags (think valet parking tickets). It is in the very nature of the use of an object as a tag that it stands for another object. Similarly, a proper name ‘simply refers to its bearer and has no other linguistic function.’ (Kripke, 1979, pp. 239–240).

  2. 2.

    Proper names have unique bearers: It would defeat the very purpose of a tag—which is to stand for a particular object—if the same tag is assigned to more than one object. (A parking ticket should allow the valet to identify the unique car that needs to be brought instead of a bunch of cars.) Analogously, a proper name has a unique bearer.

  3. 3.

    Proper names are directly referential: A tag can be simply assigned to an object and may not contain any information about the object to which it is assigned (valet parking tickets often contain just a number and no information about the parked car.) Analogously, proper names are ‘directly referential’—they are not associated with any referent-determining or reference constraining information.Footnote 4

The Millian view is often contrasted with classical descriptivism— the view of names advocated by Frege (1892) and Russell (1911, p. 114). However, the Millian view and classical descriptivism do not disagree about each of the above three commitments. (For instance, both views agree that when used referentially, a proper name denotes/refers to a unique object.) My aim in this article is to examine the very way the relation between language and the world is construed on the Millian account, and this aim is better served by focusing on the contrast of the Millian view with two of its most prominent contemporary competitors—the Predicate view and Indexicalism. These views reject the Millian equation of a name with a tag and disagree with the Millian view on each of the above three commitments. I will discuss these views and how they fare vis-à-vis the Millian view in Sect. 3; but before that, I will highlight an important consequence of construing the linguistic function of a name on the model of a tag. This is the agenda for the rest of this section and the next section.

1.2 The individuation of tags

The successful use of a tag presupposes some prior understanding of the constitution of the tag and the criterion of its identity and individuation. In Marcus’s example, a tag is a whole number such that distinct numbers count as distinct tags. Valet parking tickets are often paper or plastic objects bearing a number—a physical embodiment of Marcus’s number tags. Two physically distinct parking tickets with different numbers count as distinct tickets, and despite being garbled or damaged by a user, a ticket still counts as the same ticket insofar as it retains the number that it is individuated by. Furthermore, if two copies of the same ticket need to be issued—say, because either of two different people would like to collect a car—it can be done by issuing two distinct physical paper or plastic objects bearing the same number.Footnote 5

Such complex tagging manoeuvres are possible because we have an intuitive grasp of what constitutes a tag and the conditions under which two tags count as the same or distinct. A reader’s acceptance of Morgiana’s success crucially depends on the (natural) assumption that the tag used by the robber is a chalk mark individuated by its shape and colour—only then it makes sense to accept that Morgiana succeeded in putting the same tag on the other houses by drawing a chalk mark with the same shape and colour. Morgiana’s success would not make sense to an eccentric or unsympathetic reader who takes the robber’s tag to be individuated by the exact chemical composition of the robber’s chalk apart from its shape and colour (unless, of course, the proviso that Morgiana used a chalk made of the same material as the robber’s own is made part of the story.)

Tags are physical or abstract objects, and names are wordsFootnote 6—also a kind of object. Therefore, it is not surprising that the same issues concerning constitution, identity, and individuation that arise for tags arise for names as well. Analogous to the unsympathetic reader of the Arabian Nights, one can imagine a critic who objects to commitment (1) above by pointing out that proper names are not exclusively devices of reference (e.g., ‘Alfred’ in ‘No Alfred has ever walked on the Moon’ is a name but does not refer to any individual) or objects to (2) by pointing out that more than one individual can bear the same name. The Millian response to such criticism has been on the same lines as how one would respond to the unsympathetic reader of the Arabian Nights—the critic individuates names in a manner different from how the Millians individuate names. In defence of the Millian view, for instance, it has been argued that ‘Alfred’ in ‘No Alfred lives in Princeton’ is not a name but a homonymous common noun true of a ‘bearer of “Alfred”’ (e.g., Jeshion, 2017, p. 234) and names of distinct individuals are, properly speaking, distinct words (Kaplan, 1989, p. 562, 1990; Kripke, 1980, pp. 7–8; Sainsbury, 2005, p. 121; Soames, 2002; Stojnić, 2021, p. 54).

1.3 From semantics to metaphysics

Note, however, that unlike the reader of Arabian Nights it is less clear (certainly not intuitively clear) that the critic of the Millian view is eccentric or unsympathetic—‘Alfred’ is easily recognized as a name in ‘No Alfred lives in Princeton’, and the phenomenon of namesakes is a linguistic fact. It is therefore reasonable to ask: why should the manner of name-individuation favoured by the Millian view be adopted in the first place?

Before venturing further, however, it will be useful to distinguish two tasks—one semantic and the other non-semantic. The task of semantics is, broadly speaking, to assign meanings to words and expressions. However—as Kaplan notes—‘given an utterance, semantics cannot tell us what expression was uttered, and what language it was uttered in. This is a pre-semantic task’ (Kaplan, 1989, p. 559). The Millian (or any other) semantic view is not under an obligation to provide or defend an account of the constitution of names or their criterion of identity/individuation. This pre-semantic task belongs to a different project that, following Kaplan (1990), I will call a ‘metaphysics’ of words. In principle, a semantic theory can be consistent with different ways of thinking about the metaphysical nature of words.

The question posed at the end of the first paragraph of this subsection is then a question concerning the metaphysics of names and not their semantics. Categorizing it as metaphysical, however, does not lessen its significance for the Millian view. Corresponding to commitments (1) and (2) above, the Millian view presupposes a metaphysics of words on which names are individuated such thatFootnote 7:

  1. (a)

    a name is a word that stands for an individual, and

  2. (b)

    a name is a word that stands for at most one individual.

Call the conditions specified by (a) and (b) the ‘Millian Individuation Constraint’ or ‘MIC’. If the Millian view is correct, then the only plausible metaphysical views of names must be those that satisfy MIC. If so, then the plausibility of the Millian view cannot be considered independent of the plausibility of the metaphysical accounts that satisfy MIC. It would speak against the Millian view if—as I will argue in this article—the metaphysical accounts that satisfy MIC do not fit within a general metaphysical picture for words.

2 The common currency conception

…for serious semantics, I think that it is my common currency conception that would be important.

David Kaplan, ‘Words’ (1990, p. 111)

2.1 Two frameworks

Philosophical views concerning the metaphysics of words typically develop within one of two broad metaphysical frameworks of thinking about words. The first framework—associated with the work of Peirce (1906), Quine (1987), and Wetzel (2002; 2009)—treats words as abstract types that are tokened in their various occurrences, i.e., inscriptions, utterances, etc. The second framework—associated with the work of Kaplan (1990; 2011) and Sainsbury (2015)—treats a word as a continuant: a four-dimensional object like a person or an artefact. Within this second framework, words (like persons) are thought to be objects created at a point in time that may cease to exist at another point. Furthermore, the relation between a word and its occurrences is thought to be analogous to the relation between a person and the stages of that person’s life—the various inscriptions or utterances of a word are not instances but rather stages in the life of a word. Following Kaplan (1990), call the first framework the ‘type-token’ model and the second the ‘stage-continuant’ model.

Within these two broad frameworks, different commitments concerning the ontology of types and continuants will result in different metaphysical pictures of a word. For instance, within the first framework, if types are individuated orthographically (such that all tokens of a type have an orthographic resemblance), then ‘bow’ would instantiate a single word associated with two distinct meanings, but ‘connection’ (as it is written today) and ‘connexion’ (as written by Hume in the manuscript of the Treatise three hundred years ago) would count as instances of distinct words. Within the second framework, if continuants are individuated by their historical origin, then ‘bow’ could be an inscription of either of two distinct words—‘One word […that…] comes from the Old English bugan (inclination of the body in greeting), another from the Old Norse bogr (front end of a ship).’ (Sainsbury, 2015, p. 198) Furthermore, if it is assumed—as Kaplan (1990, p. 104; 2011, p. 518) does—that the parts of a word continuant are bound together into a single word-continuant not by resemblance between the parts but by the intention of language users to repeat an inscription or utterance, then ‘connection’ and ‘connexion’ can be distinct stages in the life of the same word.Footnote 8

The choice between the type-token and the stage-continuant model is essentially the choice between thinking of words as Universals—instantiated as written signs, spoken sounds, etc.—versus thinking of them as Particulars—with the various inscriptions, utterances as their parts or stages.Footnote 9 In sharp contrast to Universals, a word on the stage-continuant model—to use Kaplan’s words—is an ‘earthly, created thing’ that lives ‘in the world, not in Plato’s heaven’ (Kaplan, 1990; p. 111; 2011, p. 509). In this model, unlike Universals, a word simply does not exist if it does not have any utterances/inscriptions. In the type-token model, however, words are Universals, some of which may be instantiated, but not others. Theorists of this persuasion take it to be an advantage because it provides a neat explanation of how morphemes may combine to form new words, some of which may not have actual instances—e.g., ‘anti-anti-Missile’ (Hawthorne & Lepore, 2011, p. 455).

The question of the relative merits of the two metaphysical frameworks is a substantial question, but I will not take sides on this issue here. My concern here is the question of whether the Millian Individuation Constraint forces one to accept one way of thinking about the metaphysics of words. In what follows, I will argue that it does.

2.2 Incompatibility of Millianism with the type-token model

A metaphysical account of names on the type-token model cannot satisfy the Millian Individuation Constraint. To see this, it will be helpful to begin by drawing a distinction between what may be called ‘Pure Universals’ and ‘Impure Universals’. Pure Universals are Universals that do not make any reference to any Particular. Some examples include the Universals described by the predicates ‘is white’, is a unicorn, ‘is a bag’ etc., which do not require the specification of a Particular (their instance or otherwise) for their full characterization. Impure Universals, on the other hand, make an essential reference to a particular—e.g., the universals described by the predicates ‘is a neighbour of Socrates’, ‘is a neighbour of Plato’, ‘is sitting between Aristotle and Alexander’ etc. As these examples demonstrate, Impure Universals are essentially partially saturated Relations that have one or more Particulars as their relata.

The distinctness between Impure Universals can be grounded in the distinctness between Particulars. The Relation described by ‘x is the teacher of y’ can be saturated by two distinct Particulars to yield the Impure Universals described by ‘x is the teacher of Plato’ and ‘x is the teacher of Aristotle’. Further, the Impure Universal described by ‘x is the teacher of Plato’ is distinct from the one described by ‘x is the teacher of Aristotle’ by virtue of the distinctness of Plato from Aristotle. It is evident that the reason why the distinctness of Impure Universals can be grounded in the distinctness between Particulars is because the full specification of Impure Universals requires a reference to Particulars (as partially saturated Relations, Impure Universals contain Particulars as constituents.)

Pure Universals, however, cannot be individuated by appealing to the distinctness of Particulars.Footnote 10 Pure Universals, by definition, do not make reference to any Particular, so they cannot be individuated by appeal to the distinctness of Particulars. Considerations of asymmetry of the grounding relation provide a further reason for why Pure Universals cannot be individuated by appeal to the distinctness of Particulars. Assuming that the distinctness of Particulars is eventually grounded in the distinctness of Pure Universals, to ground the distinctness of Pure Universals on the distinctness of Particulars amounts to violating the asymmetry of the grounding relation.Footnote 11

But why can’t a Millian (convinced of the type-token model) treat names as Impure Universals and maintain that names can be individuated by appealing to the distinctness of their referents? Doing so would force the Millian to treat a name as a Relation, with tokens such as ‘Tim’, /tim/, etc. as one of its relata and a (Particular) individual as the other. While such two-place relations may be the right way of construing the metaphysics of a relation described by the phrases ‘x is the name of y’, ‘x refers to y’, ‘x is borne by y’ etc., the notion of ‘name’ that is our concern here is complete by itself and does not also require any further specification of an individual. For instance, irrespective of whom ‘Tim’ (in say, an inscription of the sentence, ‘Tim is a philosopher’) refers to, it is readily recognized as a token of a name. A particular inscription of ‘Tim’ may, of course, be used to refer to Tim Maudlin, Tim Crane, or Tim Williamson, but it is the property of being a name and not the relation of being the name of a certain individual that is the object of the metaphysical enquiry pursued here.Footnote 12,Footnote 13

2.3 The common currency conception

With the type-token model ruled out, the stage-continuant model would seem the natural choice for a metaphysics of words compatible with Millianism. However, not all ways of construing the constitution of name-continuants will satisfy the Millian Individuation Constraint. Consider, for instance, what Kaplan calls ‘generic names’—name-continuants that have as their stages inscriptions (or utterances) that resemble orthographically (or phonetically) such that two generic names differ if their stages do not resemble each otherFootnote 14 Kaplan gives the example of the generic name ‘David’ (Kaplan, 1990, p. 111)—a name that was first inscribed/uttered in biblical times, has (roughly) retained its orthographic/phonetic form, and is constituted of the utterances/inscriptions of the first names of David Hume, David Lewis, and David Kaplan, and some 10 million other individuals currently alive. Because generic names can be shared by many individuals, they do not satisfy the MIC. (On the model of a generic name, a generic word can also be defined as follows: a generic word is a word-continuant that has inscriptions or utterances that resemble each other orthographically or phonetically as its stages. As some words are names, generic names are a subset of generic words.)

There may be other ways in which word-continuants may be thought of as constituted, some of which may be compatible with MIC but not others. My criticisms in the next section are directed against all metaphysical accounts of names that are compatible with MIC; and as there can be more than one such account, it will be convenient to call them by a name. Call such account(s) ‘Common Currency Conception(s)’ of names or ‘CCC’. On a CCC, (by definition) a name is individuated such that the following condition is always satisfied: only a word that refers to an individual counts as a name, and a name does not refer to more than one individual. A CCC would also provide a suitable story that explains why individuating a common currency name (i.e., a particular name-continuant on the CCC) in the way that it does should lead to the name standing for exactly one individual and no more.

One sketch of a CCC is provided by Kaplan in his 1990 ‘Words’. Sketch, because although Kaplan does make it clear through examples that names of different individuals (e.g., his name and the name of David Israel) are different common currency names ‘which were created at different points, and which have had distinct life-histories’, he does not provide the details of how common currency names must be individuated on his view. This is understandable because Kaplan’s central focus in ‘Words’ is not to propose a detailed metaphysical account of the CCC but to pitch the stage-continuant model as a viable alternative to the type-token model.

A more fully developed CCC is proposed by Mark Sainsbury, who contends that a name-continuant is individuated by an originating baptismal act such that ‘for every name, N, there is a unique act in which N and N alone is originated’ (Sainsbury, 2015, p. 201).Footnote 15 On Sainsbury’s account, not only does a baptismal act single out a unique name-continuant, each baptismal act is also associated with at most one thing that gets named in that act, if anything gets named in the act at all. (Therefore, a single act of baptising 100 objects must resolve into 100 distinct sub-acts of baptising each of the hundred objects.) Therefore, corresponding to every common currency name there is at most one named individual—i.e., the individual baptized in the act that individuates the name. Coupled with the choice to consider only those uses of an expression in which it refers to an individual as a name, Sainsbury’s account provides a clear and fully worked out picture of how a CCC of names is supposed to work within the stage-continuant model.

Taking stock, the Millian Individuation Constraint forces one to not only abandon the type-token model as a suitable framework for a metaphysical account of names but also forces one to accept a specific metaphysics of names even within the stage-continuant model—i.e., the common currency conception of names. In the next section, I evaluate the suitability of the CCC as a metaphysical account of words more generally.

3 The metaphysical burden

I believe that many important theoretical issues about the semantics of names (probably not all) would be largely unaffected had our conventions required that no two things shall be given the same name.

Saul Kripke, Naming and Necessity (1980, p. 7)

3.1 The sui-generis metaphysics of names

In responding to the debate between the two metaphysical frameworks outlined above, Sylvain Bromberger complained that the question ‘what are words?’ is ‘woefully underspecified’ and that the parties in the debate ‘approach the question and its requirements with very different concerns in mind and end up often at cross purposes’ (Bromberger, 2011, pp. 486–487). Bromberger’s point is that how one answers to the question of which metaphysical picture should be adopted for an investigation into the linguistic nature of words depends on the purposes of such an investigation: a phonological study of stress and tone assignment may individuate words as sound types (thereby completely ignoring their orthography), while a study of word etymology will take into account the history and the origin of a word. Given the multiplicity of such purposes, it may even be necessary to ‘break up the notion of word into a number of more precise and theoretically manageable ones’, leaving us with not one but many notions of a word, and consequently many ways of thinking about the metaphysics of words (Bromberger, 2011, p. 503).

If Bromberger’s diagnosis is correct, then the search for the final metaphysical account of names is a misdirected endeavour. Any metaphysics of words must be relativised: for some purposes, a shape- or form- theoretic conception of words within the type-token model may be fruitful (say, when defining notions like ‘ambiguity’, ‘polysemy’, etc.) and for others, conceiving of words within the stage-continuant model may be more useful (e.g., in an etymological study of words.) Nevertheless, a particular sort of investigation into the linguistic nature of words (say, phonological, morphological, or semantic) must adopt a single metaphysical picture of words. This requirement is no more than the requirement that the participants in a particular sort of investigation (say, a semantic investigation) must converge on one answer to the question of what words are. Without this common metaphysical ground, one cannot be sure whether the participants in a semantic debate agree (or disagree) about the semantic properties of the same thing, nor can the possibility of their simply talking past each other be definitively ruled out.

A single, uniform metaphysical account of words is needed across one more dimension—the correct account should not give a different metaphysical picture of words depending on how a word is classified within a scheme of classification. In other words, within a particular sort of linguistic investigation—say, semantic, morphological, or phonological—the correct answer to the question ‘what is a word?’ should not change based on how a word is classified semantically (e.g., as a noun, verb, preposition, etc.), morphologically (derivational morpheme, inflectional morpheme, free morpheme), phonetically (e.g., as monosyllabic, bisyllabic, or polysyllabic), etc. This requirement is supported not only by considerations of theoretical elegance and ontological parsimony but also, more importantly, by the fact that any scheme of classification must presuppose that the entities being classified are unified on some ground. For instance, only because there is a prior sense in which a noun and a preposition share something in common—i.e., by both being words—that it makes sense to classify them on a semantic basis as a noun or a preposition. The notion of a word is a fundamental notion that provides the ground for any further classification based on semantic, phonological, or morphological properties and therefore cannot itself be split up based on such properties.

If names are words and if the Millian view is the right semantic account of names, then given the above considerations of uniformity, one would expect that the metaphysical picture of a word that comes along with the Millian view (i.e., CCC) should also be the right metaphysics to adopt for a semantic investigation into words more generally. It is difficult, however, to see how this can be so. Words belonging to many—or rather most—semantic categories (e.g., prepositions like ‘of’, ‘below’, conjunctions like ‘or’, verbs like ‘reading’, etc.) do not refer and therefore it is not clear how the manner of name individuation proposed by the CCC should apply to these words.Footnote 16 (Unlike a name, there is no clear baptismal act that associates the preposition ‘of’ with an object.) Even if attention is restricted to singular terms that do refer—e.g., pronouns, indexicals, some definite descriptions, etc.—it is clear that these words are not individuated in the manner that names are individuated according to the CCC. It will not only be odd to say that the pronoun ‘she’ must be a different pronoun when it is used to refer to different individuals; such a proposal would also make nonsense of the idea that pronouns have characters (or φ-features) that constrain their reference. (Which of the many pronouns is the character of ‘she’ the character of? Is it associated with the word type ‘she’? If so, then what remains of the CCC of pronouns?)

It may be objected that the above construal of words on the CCC unsympathetically assumes that such construal must require words with different referents to be considered different words. It may be said that a more sympathetic construal of words (in accordance with the CCC) would only require words with different meanings to be different words. The objector could say, for example, that the verb ‘reading’ and the noun ‘reading’ (e.g., as it appears in the ‘University of Reading’) are different words because they have different meanings and that the pronoun ‘she’ can be the very same pronoun across its various uses because it retains the same character (which is understood as the meaning of pronouns.)

However, in the context of proposing a metaphysics of words for a semantic investigation into words, such a proposal would have a serious methodological flaw. One objective behind a metaphysics of words is to provide a common ground for semantic disagreements concerning words. A minimal condition for a genuine disagreement concerning the properties of a thing is that the dissenting parties disagree about the properties of the very same thing. However, if the answer to the questions ‘what are words?’ and ‘how are words individuated?’ itself depends on the meaning (or some other semantic property) that one takes the word to have, then two people who think that they disagree about the meaning of a certain word cannot really be disagreeing—given that the meanings are different, they must have distinct words in mind. For genuine disagreement, however, the two parties must differ with respect to the meanings that they assign to the same word.Footnote 17

The CCC, therefore, is not a suitable metaphysical picture for words other than proper names. Kaplan concedes this point when he writes that common nouns must be construed as generic names and not as common currency names. Kaplan also suggests that the CCC is not the right metaphysical account for indexicals and demonstratives (Kaplan, 2011, p. 524). If so, then some account other than the CCC must be the appropriate metaphysical account for a semantic investigation into words more generally—i.e., for all words other than names. For instance, it may so turn out that for semantic purposes a metaphysics of words within the type-token model is the best way to answer the general question of what words (other than names) are.

The category of names is a semantic category; thus, the classification of words into words that are names and words that are not names is also a semantic classification. If the Millian semantics of names is correct and if for semantic purposes the CCC cannot be the right metaphysical account for words that are not names, then proper names must be construed as common currency names but not words other than proper names. This goes against the considerations of metaphysical uniformity outlined earlier—i.e., the answer to the question ‘what are words’ cannot be sensitive to how a word gets classified within a scheme of classification. More importantly, the Millian view sets names apart as words with a sui-generis metaphysical nature, disconnected from the metaphysics of other categories of words. Such exceptionalism for names would be justified if the very semantic nature of names necessitated it. The metaphysical uniqueness of names, however, is not a condition for a plausible semantic account of proper names—or so I will argue in the next section.

At this point, the Millian may raise the following objection: the semantic merits of the Millian View (which, for the committed Millian, there are many) justify the metaphysical exceptionalism of names, but the article puts the cart before the horse by making the exceptionalism a ground for attacking the Millian view of names. This might have been a valid objection against the argument of this article had it been true that the Millian view is the correct (or at least the most plausible) semantic account of proper names. However, in the context of a semantic debate concerning proper names, to assume the truth or plausibility of the Millian view is to beg the question against its competitors and is therefore methodologically problematic.

3.2 Alternative semantic views

The most prominent contemporary competitors of the Millian account are semantic views labelled in contemporary debates as the ‘Predicate View’ and ‘Indexicalism’.Footnote 18 According to the predicate view of names, proper names are (syntactically) common nouns and (semantically) expressions with a predicate-type semantic value (i.e., type <e,t>). According to Indexicalism, a proper name is essentially a pronoun or an indexical—i.e., an expression whose reference is constrained by a condition (i.e., its ‘character’ or ‘φ-feature’) but whose referent can be different in different contexts. As I noted earlier, the disagreement of these two views with the Millian view is deeper than the disagreement between the Millian view and classical descriptivism—unlike classical descriptivism, the Predicate View and Indexicalism disagree with the Millian View on all three of its commitments (1)-(3) discussed above. To get a better grip on what lies at the root of the disagreement, I will consider each commitment of the Millian view in turn and highlight the contrast of the Millian view with these two views.

  1. 1.

    Names are not exclusively devices of reference: According to both Predicate view and Indexicalism, while proper names are most frequently used to refer to individuals, they can have other, nonreferential uses as well. On the Predicate view, proper names are literally common nouns, and common nouns can be used both to refer to an individual as well as to predicate something of an individual. For instance, as part of a larger noun phrase, the common noun ‘cat’ in ‘the cat wants out’ is used to refer to a cat, but the same common noun has a nonreferential, predicative use in ‘Sir Harrington is a cat’ or ‘I have two cats—Sir Harrington and Commodore Norrington’. Similarly, apart from referential uses, a proper name can also have nonreferential, predicative uses. For instance, ‘Alfred’ in ‘No Alfred has ever walked on the Moon’ or ‘All Alfreds study in Princeton’ appears as a common noun applying to all bearers of “Alfred”, not to refer to an individual. Similarly, according to Indexicalism, proper names can not only be used to refer to individuals, but (like pronouns and indexicals more generally) proper names can also have anaphoric readings—e.g., the name ‘Bambi’ in the consequent of ‘If a child is christened “Bambi”, Disney will sue Bambi’s parents’ is anaphoric to the indefinite in the antecedent (Geurts, 1997; Schoubye, 2020, pp. 62–64).

  2. 2.

    Multiple bearer-hood of names: Both the Predicate view and Indexicalism account for the phenomenon that more than one individual can bear the same name; on these views (when used to refer) the same proper name can refer to one or another individual depending on the features of context. This is rather straightforward on the Indexicalist view, which treats proper names as pronouns. Just as a pronoun ‘she’ can be used to refer to different individuals who satisfy the φ-feature of the pronoun ‘she’ (depending on the context of its use), a proper name such as ‘David’ can also be used to refer to different individuals who satisfy the φ-feature of the name ‘David’.

    On the Predicate view, (when used to refer) proper names are thought to be part of either an incomplete definite description or a complex demonstrative with a covert determiner element (such that the logical form of ‘David is a philosopher’ is either ‘(the) David is a philosopher’ or ‘(that) David is a philosopher’ where the determiner is not overt in the syntax of some languages, like English). Despite their internal differences concerning how to account for the semantics of the referential uses of names, defenders of the predicate view unanimously maintain that just like ordinary common nouns, as part of indefinite descriptions or complex demonstratives, proper names can also be used to refer to one or another individual at different occasions of their use.

  1. 3.

    Direct reference: The predicate view borrows the standard semantics of incomplete descriptions or complex demonstratives to explain the reference of proper names, and the standard semantics of both incomplete descriptions and complex demonstratives involve the commitment that their referent is an object that falls within the extension of their respective nominal components (for instance, the referent of ‘this maroon table’ must be a member of the extension of the nominal ‘maroon table’.) Therefore, names are not directly referential on the predicative view. The same is true for the semantics of names on Indexicalism: the referent of pronouns is constrained by their φ-feature (e.g., the pronoun ‘she’ can only be used to refer to an individual who identifies themselves as belonging to the female gender); and because names are pronouns on the Indexicalist account, the referent of a name is constrained by the φ-feature associated with the name.

At the root of these differences between the Millian view on the one hand and the Predicate view and Indexicalism on the other is the fact that the former view takes seriously the idea that a name is a tag, but the latter views take seriously the idea that a name is a kind of word. This is, of course, not to say that the Millians deny that names are words—on the contrary, I have been assuming an understanding of the Millian view on which names are thought of as words. However, taking names to be words while keeping the Millian account forces a special, limited understanding of the notion of a ‘word’. In treating names as tags, the Millian view ignores the role of words as an interesting linguistic item with wide-ranging semantic, phonetic, and morphological properties. Furthermore, unlike the notion of a word, the notion of a tag is a strictly functional notion—an object is a tag insofar as it performs the function of standing for something else. Thus, the Millian construal of names on the model of tags forces the treatment of literal, nonreferential use of a name as somehow deviant uses of a name, or not a use of a name at all.Footnote 19 The Millian view begins with the assumption that names are tags and then accommodates the assumption that names are words by proposing a conception of words that is most conducive to their treatment as a tag.

The Predicate view and Indexicalism, however, begin with the assumption that a name is a kind of word; therefore, the semantic, phonetic, and morphological properties of names must be understood in accordance with the ways in which words are thought to possess those properties more generally. No class of words in natural language has an exclusive semantic function, and on these views, the same is true of names. Furthermore, like words generally, names can be part of various word formation processes such as inflection (e.g., names get inflected for number, case, etc.) and allow for morphological derivation or conversion into other semantic/syntactic categories (e.g., the verbs ‘to google’ and ‘to chisholm’).Footnote 20 On the Predicate view and Indexicalism, names are expected to exhibit the full range of linguistic behaviour that one would expect from a word more generally.

The Predicate view and Indexicalism therefore explain the semantic properties of proper names by appealing to the standard semantics of the word categories that they take names to belong—i.e., indexicals and common terms. Irrespective of the semantic dividends of such a move, it certainly has a metaphysical dividend: because on these views, names are common nouns (or pronouns), whatever metaphysical account turns out to be the best account for common nouns (or pronouns) will also suffice as the best metaphysical account of proper names. Unlike the Millian view, therefore, these accounts do not require a sui-generis metaphysics for names.

3.3 Names, words, and tags

Therefore, in a balance of considerations vis-à-vis its competitors, the sui-generis metaphysics of proper names that the Millian view requires counts as a strike against it. However, the argument of this article has thus far relied on the assumption that names are words. After all, if names are not words, then their exceptionalism among the class of words would not be so unpalatable. Although most philosophers concerned with the metaphysics of names have assumed that names are units of language, in wider debates concerning names, it is sometimes held that names are not words and that they do not belong to a language (e.g., Ziff,1960, p. 86). Although this move would save the Millian view from the objections raised in this article, the proposal that names are not words raises many serious problems of its own. I now review some such problems.

Proper names exhibit much of the same linguistic behaviour that words belonging to a language generally exhibit. For instance, proper names follow the same rules for inflection which apply to other words of a language, e.g., in English nouns are inflected into their plural forms by adding the suffix ‘-s’ (e.g., ‘languages’ from ‘language’) and so are proper names (e.g., ‘Alfreds’ from ‘Alfred’). Furthermore, proper names are governed by the same rules for word formation (e.g., morphological derivation) that govern other words of a language. In American English, for example, new verbs can be formed from nouns or adjectives using the suffixes ‘-ize’ or ‘-ify’. Which words form verbs by taking the suffix ‘-ify’ (e.g., classify, mummify, etc.) and which take ‘-ize’ (e.g., hospitalize, tranquilise) is governed by a morphological rule. Consider, for example, the following rule for forming a new word using the suffix ‘-ize’:

-ize attaches to adjectives or nouns of two or more syllables where the final syllable does not bear primary stress. (Lieber, 2009).

Thus, ‘Stalin’—which is a proper name with more than two syllables in which the final syllable does not bear primary stress—takes the suffix ‘-ize’ (and not ‘-ify’) to form the verb ‘Stalinize’. Additionally, it is not uncommon for linguists and philosophers to simply extend to proper names a general account developed for words belonging to a language. For instance, Clark & Clark and Carston do not distinguish between the English word ‘porch’ and the proper name ‘Houdini’ when discussing the phenomenon of lexical innovation and polysemy (Carston, 2019; Clark & Clark, 1979). Finally, proper names, like words generally, often get translated across languages. One example is that of Vienna, which is called ‘Wien’ in German and ‘Bécs’ in Hungarian; another example is that of Jesus, whose name is spelled ‘Jesus’ in German but ‘Jézus’ in Hungarian.

If proper names are not words, then it would be surprising why there should be such similarities in their behaviour. A bigger problem with taking proper names as not words is that it makes it puzzling why an investigation into the nature of proper names must be a part of a semantic investigation. Semantics is concerned with the properties of words and expressions; if proper names are not words but are, say, tags, then an investigation of proper names must properly be a part of a more general project concerning the investigation into the use of objects as tags. The Millian view would then not be a semantic view but rather a theory of a special kind of tag. This is not a petty argument concerning disciplinary boundaries but rather a broad-minded invitation to try and understand the linguistic nature of proper names as they appear as part of language. The use of tags is an undeniably important part of human behaviour, and while most tags are concrete objects, it is also possible that some tags are constructed using the orthographic and phonetic resources of a language (i.e., its letters and phonemes). However, names are an equally important component of human linguistic behaviour, and insofar as human linguistic competence involves a competent use of names, there is a scope (and need) for an investigation into the nature of names while treating them as words belonging to a language.

A semantic study of names must therefore treat proper names as words. If so, then I have argued that the Millian view of names requires a metaphysical account that makes names exceptional within the class of words. Given the considerations of uniformity and parsimony, such exceptionalism speaks against the Millian view but provides support to competing semantic accounts that do not require such exceptionalism.