The Same Name

“what makes tokens of a word … tokens of the same word … is,

in the first instance, history, not form or function” (Millikan 1984: 72–3)

Abstract

When are two tokens of a name tokens of the same name? According to this paper, the answer is a matter of the historical connections between the tokens. For each name, there is a unique originating event, and subsequent tokens are tokens of that name only if they derive in an appropriate way from that originating event. The conditions for a token being a token of a given name are distinct from the conditions for preservation of the reference of a name. Hence a name may change its reference. Defending the theory requires considerations about the identity of acts, and about empty names and disagreement. The upshot is a causal theory of the identity of names, but not what would normally be counted as a causal theory of the reference of names. Although reference is often transmitted causally, what determines semantic reference is conventionalized speaker-reference.

1 Introduction

Uses of proper names are dated individual events. They may be speech-acts, or silent thought-acts (acts of inner speech). The main question for this paper is: what makes two uses of a name uses of the same name?

The question as it stands is unclear. Are we supposed to count a use of “David” for David Hume as a use of the same name as a use of “David” for David Kaplan? Taken one way, the question answers itself in the affirmative: Hume and Kaplan are both called “David”, and so are called by, and hence have, the same name. Taken another way, we might hesitate: if it’s really the same name, then should not “David was born in the eighteenth century” and “David was born in the twentieth century” be inconsistent? Yet they are not.

There are at least two ways of dealing with this unclarity. One is to liken names to indexicals. This allows us to liken the relation between the two sentences just mentioned to that between utterances of “I am hungry” and “I am not hungry” by different people, or by the same person at different times: of course these utterances are not inconsistent. Another approach, deriving from Kaplan (1990), is to distinguish between generic and specific names. (Kaplan uses “common currency” where I use the more natural “specific”.) A generic name is a template that can be used to create specific names. Hume and Kaplan have names from the same template; they share the generic name “David”. Specific names are more closely related to baptismal acts: Hume’s specific name “David” was bestowed in the eighteenth century, Kaplan’s distinct specific name “David” in the twentieth. Asked to assess a name-containing utterance for truth or falsity, we need to determine which specific name it involves. For it’s only specific names, and not generic names, that have bearers,¹ and the truth or falsehood of a name-containing utterance typically depends upon the bearer of the name.

A generic name can exist when there are no specific names of the genus (e.g. a baby book name is so silly no parent ever uses it for a baby); and a specific name can exist when there is no corresponding generic name, as would happen if no one had seen fit to use the name as a template. There would then be no corresponding generic name, or at least it would be in some kind of limbo (unlike the specific name).

In this paper I shall use the generic/specific way of characterizing the phenomena. In these terms, there are two primary questions: when are two uses of a name uses of the same generic name? And when are two uses of a name uses of the same specific name? Of these two questions, it is the second that is of more direct concern to semantic theory, since different specific names, even if they exemplify the same generic name, will typically be treated differently semantically. For example, the question whether a name can change its bearer is relevant to specific names rather than generic ones, for which the notion of “their bearer” is at best equivocal. In what follows, therefore, I will focus only on the question of what makes two uses of a name uses of the same specific name.

I will argue for two theses:

• 1. The same-name relation among uses is determined by

  • a. unique origination and

  • b. deferential historical chains

• 2. The same-reference relation among uses of the same name is affected by distinct deferential historical chains

The theses taken together allow for the possibility of a name changing its referent, and provide insight into the mechanisms by which this can occur.

The next Sect. 2 shows why one version of the indexical framework should be rejected. Those who are happy with the generic/specific approach could skip directly to Sect. 3.

2 The indexical approach to names

Pelczar and Rainsbury (1998) suggest that the relevant contextual feature to which a name’s indexicality is sensitive is the “dubbing-in-force”, which is the most salient relevant dubbing in the context. A dubbing is the bestowal of a name on some object. They suggest, for example, that

if in the course of a conversation dealing both with President Bush and his son someone says, ‘George Bush occupied the Oval Office for only one term’, Gricean conversational maxims (and in particular, the maxim of Quality) might contribute to raising the dubbing of the President with ‘George Bush’ to prominence, over that of his son. (1998: 295)

Understood literally this is highly mysterious. To the extent that there was a dubbing, it occurred some 50 years before the conversational exchange. One can understand the utterance without knowing when the dubbing occurred, or how, or who was there, or anything about it, or even if there was a dubbing (perhaps there was no deliberate dubbing and the first use of the name was, like many nicknames, an accident that somehow stabilized). It is very hard to see how any such event, now long distant in time, can be raised to prominence, or that it needs to be. In any case, the dubbing involved just “George”. The family name “Bush” was applicable merely in virtue of his parents having that name, not in virtue of a dubbing.² By contrast, in the alternative account I shall offer, a relevant dubbing does indeed play a role in fixing which specific name is being used, but it does so by standing in a suitable causal relation to the utterance, not by being salient. It is not an event that need be accessible to either the speaker or the hearer.

I have further reservations about Pelczar and Rainsbury’s account.

• 1. They describe a dubbing as a dubbing of an object, thus excluding the possibility of empty names. (“We set empty names aside as raising issues beyond the scope of this paper.” 1998: n4.) The methodology is suspect. One should not assume that the right approach is to give an account of names with bearers, hoping that it can somehow be tweaked so as to extend to an account of names without bearers. On the contrary, appreciating the similarity in use between empty and non-empty names dictates a common semantic category at some level in the taxonomy (as in Sainsbury 2005).

• 2. The authors tend to run together the facts that determine which object is the bearer of an utterance of a name with what evidence an interpreter might have for this. The displayed quotation reveals this slippage: there was a fact about what the use of “George Bush” referred to before an interpreter could begin to apply Gricean maxims. The maxims helped provide evidence; they could not be constitutive, since there could be a fact about what the utterer meant even if there was no audience.

• 3. Pelczar and Rainsbury’s theory does not allow us to distinguish a case like the use of “Paris” for Paris, TX, rather than for Paris, France, from a case like the use of “Madagascar” for the African island rather than the mainland. Yet intuitively the cases are very different. No error occurred when Paris (Texas) was so named by George W. Wright in 1844, and no name changed its reference. By contrast, in the usual telling of the story about Madagascar, a name did change its reference. The cases are collapsed by Pelczar and Rainsbury’s theory since all they can say about both is that a new dubbing came into force.

There may be versions of the indexical approach which are not subject to these problems. I suspect that these will permit a straightforward translation into the generic/specific framework, using the rule: a specific name is a name plus some contextual material. In any case, the generic/specific framework provides all the distinctions needed for this paper.

3 Spelling and Pronouncing

Imagine each use of a name, each dated event occurring at any time and place throughout the word’s history, mapped by a dot on a chart. Faced with a sea of dots, your job is to join up those that represent uses of the same specific name. What do you need to know in order to perform this task?

Addressing the question “What is a word?” without reflection, we are inclined to allude to spelling and pronunciation. This inclination is quickly corrected by reflection. Two words may be spelled and pronounced the same, like “bow”. One word with this spelling and pronunciation (to rhyme with “cow”) comes from the Old English bugan (inclination of the body in greeting), another from the Old Norse bogr (front end of a ship). Since the histories are different it is convenient to classify these as distinct words, rather than as a single word with inconsistent histories. In the case of specific names, we know in advance that sameness of spelling and pronunciation will not be sufficient for uses to be uses of the same specific name, for different specific names of the same genus may be spelled and pronounced the same. But might spelling and pronunciation at least constitute a necessary condition for sameness of specific name?

There are endless counterexamples, cases in which the same specific name is pronounced or spelled differently. For pronunciation, consider the contrast between the French pronunciation of “Paris” and the English pronunciation of that same specific name. For spelling, we are familiar with Shakespeare’s variable spelling of his name. For a specific insistence that one and the same name can be differently spelled consider:

Clodius and Claudius are the same name, the former being the less aristocratic version (Michael Grant’s edition of Cicero’s Selected Political Speeches, p. 184, n. 21).

It’s as correct to speak of Publius Clodius Pulcher as of Publius Claudius Pulcher: same person and same (specific) name, different spelling.³

Sameness of bearer is not sufficient for two uses of a name to be uses of the same name, for many people (and at least one famous planet) have more than one name. Whether or not sameness of bearer is necessary is a substantive question, to be explored in what follows.

Rather than consider intrinsic characteristics we should, as Millikan said for words (quoted under the title of this paper), individuate specific names historically. What I will suggest is that we need to start by considering the event in which the name originated, and then the causal links connecting subsequent uses, links that preserve the name’s identity. The general idea is familiar from Kripke (1980) but there are differences. In particular, I shall argue that the links that preserve a name’s identity may not preserve its reference.⁴

4 Origination

Each name has just one originating use: the first use of the name is its origin. We cannot appeal to the name’s identity in specifying its origin. Although it is true that the origination of N is the first use of N, this does not address our individuative question, for it presupposes that we have no problem about individuating N. Reverting to the image of the dots on a chart, the first thing we need to do is pick out those that are originations—maybe we make them glow red.

A use of a name is an originating one iff it does not look back to or defer to earlier uses of any specific name. It is not constrained to follow existing usage. That is why the explicit conferring of a name (as in a baptism) is such a pleasing model of origination. Within limits, one can confer a name as one likes.⁵ No deference to existing usage is involved, since there is no existing usage.

In a simple model of a dubbing or bestowing, there is a single bearer, and a pre-existing generic name. The generic name serves as a kind of template from which a new specific name is forged by a stipulation. Perhaps many specific names originate in this way, but there are also other possibilities. For example, a name might be introduced from scratch without being copied from a generic template.

In some cases, a name is introduced when there is no bearer, perhaps as a result of error or perhaps as part of deliberate fiction. In these cases, no bearer can play a part in individuating the originating event. Take a case of error, Leverrier’s introduction of “Vulcan”. The originating episode perhaps started with a false quantified thought on the lines “There must be another body there affecting the orbit of Mercury”. This can sustain a grammatically singular thought, involving a definite having no bearer: “Let’s call it Vulcan”. The originating act is clear and straightforward, even in the absence of a bearer. The intention to introduce a (new specific) name is normally successful, even if the simultaneous intention to introduce a name with a bearer is not.

Although explicit dubbings are the clearest cases of origination, we must also allow for first uses which simply grow up. An unusually tall person is called “Shorty” for the first time, without any intention of founding a name-using practice, and with no intention to defer to an existing use of this nickname. A subsequent use of “Shorty” counts as a use of precisely this specific name only if it traces back causally to that originating use. It’s possible that the first use of “Shorty” has no causal impact: no one else follows suit, and the nickname is very short-lived. Yet later someone else calls the same person “Shorty” and this time it sticks. Because on this later occasion there was no attempt to follow an existing practice, it counts as an originating event. In this person’s history, he was at different times called by distinct specific nicknames of the same genus, “Shorty”.

According to the theory defended here, there should be a correlation between names and originating acts. Yet there seem to be many examples in which a single act originates many different names. I shall first give an example (Twins) that leads to a clarification of the account of origination.

Twins: The parents of twin boys finally agree: “Let’s call the elder Peter and the younger Paul”.

This is a single originating act, but two names are introduced. The example refutes the following principle:

(O1):

Each act of name-origination originates just one specific name

Acts can be counted in different ways. We can’t deny that there is one act in which both twins are named. This does not preclude there also being two acts, one the naming of Peter, the other the naming of Paul. I grind some pepper on my pasta: one act. But to grind the pepper, I have to do two things: hold the grinder firm with one hand, and twist with the other. One act is composed of two acts.

Similarly, the act of naming Peter and Paul is composed of two acts, naming Peter and naming Paul. These acts can properly be regarded as the originating acts of two distinct specific names. Though the example refutes (O1), it leaves room for this more nuanced principle:

(O2):

For every name, N, there is a unique act in which N and N alone is originated

The act of naming Peter originates just the specific name “Peter” and the act of naming Paul originates just the specific name “Paul”. These two events are sub-acts of the act reported in Twins.

Since name originations are acts, counting originations involves counting acts, and so involves taking a stand on questions concerning the structure of acts. In the Twins example, there is little room for disagreement, perhaps because it’s plain that the speaker is likely to have separate intentions, one directed at one child and one at the other, and these separate intentions ground a separation of sub-acts within the total act of naming. But in some of the examples that may be brought against (O2), matters are not so clear. I claim there are cases in which an act’s sub-acts don’t require separate object-directed intentions. Some acts are animated by a plural intention, an intention directed to a group of objects as a group. In these cases, there may be a sub-act relating to a member of the group, even if no specific intention is directed at that object. In other cases, the intention is plural even though there are in fact no objects upon which the intention is directed. Such acts too are composite, the relevant sub-acts involving not objects that are members of the plurality but objects which the subject believes, supposes or imagines are members of a plurality which the subject believes, supposes or imagines she is confronted with. The composition of such acts is fixed internalistically. The aim of the next Sect. 5 is to establish these claims on the basis of examples not involving the metaphysics of names.

5 The Composition of Acts

Given the purposes of this paper, I have no quarrel with theorists who are very liberal about act composition.⁶ For example, a very liberal view is that anything done in the course of an act is a sub-act of that act. On this view, if, on my walk, I tread on an ant, treading on the ant is a sub-act of the walk. The reason such views are unproblematic from the present point of view is as follows: counterexamples to (O2) purport to offer cases in which more than one name is originated in a single act which is not composed of as many sub-acts as names. Very liberal views about the compositional structure of acts are likely to yield as many sub-acts as (O2) needs.

A standard form of illiberalism can be constructed from two claims. One is that every sub-act of an act requires a proprietary intention. The other is that sub-acts are individuated by the objects involved in the act. The views are distinct, though they could be combined in the opinion that sub-acts (for a large class of acts) are to be individuated by an intention directed upon a specific object. In this section, I try to undermine both elements of this opinion.

Some acts are animated by intentions that are plural, in the sense that they are directed upon a group of objects, as a group. A plural intention of this kind does not require the agent to have an intention specific to each member of the group.

Beans: I sweep the beans into the pot, all 100 of them. My intention was directed on the beans, but some beans were not the object of any intention specific to them.

I might not form any intention specific to any one bean. Even if I form some such specific intentions, I will not form one per bean. I am not paying that much attention. Some beans might conceal others, but even a hidden bean, lacking any proprietary intention, falls within the scope of the intention. If a hidden bean somehow escaped my sweep, I did not do what I intended.

I think that there are as many sub-acts of the sweeping as there are beans. In sweeping the beans into the pot, I swept bean A, and I swept bean B, and so on. I did one thing, sweep the beans, but in doing that I did 100 things, one thing per bean. I did not have 100 intentions, yet each bean-sweeping was intentional, because it was a sub-act of the intentional sweeping of the beans. The sub-act can be intentional, as revealed by what would and would not constitute doing what the agent intended, without having an intention of its own. The structure is like that of the act in Twins, except that the sub-acts do not have proprietary intentions.

It may seem that in Beans the sub-acts are individuated by the objects. But there are empty cases for which this does not seem the right explanation.

Nightmare: In your dream you are in a large classroom. One of the students asks “What grade are you fixin’ to give us?” As you are about to explain that grades are based on performance and so may vary from student to student, you notice that every student has placed a large pistol on his or her desk. In a flash of indignation, you say “Everyone hereby gets an F”.

In the dream, each student is awarded an F, student #1, student #2, etc. There are as many sub-acts of the act of awarding everyone an F as there are students in the dream. Here the sub-acts are not individuated by objects, for there are none, nor by proprietary intention, for there are more sub-acts than there are these. It is no doubt not determinate how many sub-acts there are, since it is not determinate how many students were involved in the dream, but it is determinate that there are many.

This shows that sub-acts can be individuated in terms of what a subject believes (or supposes, or imagines) rather than in terms of how things actually are.

There are supposed to be two upshots of the discussion. (1) Plural intentional acts may have sub-acts which are intentional despite having no proprietary intention. (2) Plural intentional acts may have sub-acts which are distinct not because they really involve distinct objects, but because the agent believes (or supposes or imagines) that she is acting on several things. If you accept these claims, you will find that (O2) is not damaged by the putative counterexamples to be discussed in the next section.

6 Purported Counterexamples to (O2)

For ease of reference, I reiterate (O2):

(O2):

For every name, N, there is a unique act in which N and N alone is originated

This section presents five purported counterexamples, and shows that each fails.

Vladimir:⁷ The tyrant decrees: all male children born today are to be called “Vladimir”. 100 male children are born on that day.

A single act is responsible for the introduction of 100 distinct specific names.

One possibility is to treat this case like Twins, claiming that the tyrant’s act was in reality composed of 100 sub-acts. I think a better response is to deny that the tyrant’s act was one of naming at all, and that is why it is not a counterexample to (O2). Rather than name anything, what the tyrant did was issue an instruction about how anticipated future namings are to be conducted. If, for example, no males had been born that day, no specific names “Vladimir” would have been introduced, but the decree would not have been violated. Hence the decree was not itself a naming, for a naming must introduce a name. In the case envisaged, the parents of the 100 children, if they obeyed the decree, would have produced the 100 separate acts of naming. There is no conflict with (O2).

Group baptism: A variant of Vladimir. The decree is that no male child born that day is to be named by its parents, but is to be brought to the palace where it will be named by the tyrant. When, at evening, the 100 babies born that day are assembled in the palace, the tyrant says “I hereby name each one of you Vladimir”.

It seems that a single act has originated 100 distinct specific names “Vladimir”, contrary to (O2).

This differs from Twins in the following respect: one would expect the parents of the twins to have distinct sub-intentions for each twin, and this helps sustain the view that the single act of naming both twins was composed of distinct sub-acts of naming each. By contrast, in Group baptism one cannot expect the tyrant to have, for each child, a distinct intention: there are too many of them, and it might be that he can’t even see them all. If his single act consists of 100 distinct sub-acts only if he has 100 distinct sub-intentions, then the single act does not consist of 100 sub-acts.

But, as we saw in Beans (Sect. 4), it’s not the case that every sub-act needs a distinct intention. It is as correct to say that there are 100 sub-acts of naming in the case of Group baptism as that there are 100 sub-acts in the case of Beans. Hence there is an act for each name, and a counterexample to (O2) has not been produced.

Textor’s gods: Suppose that there are three thousand people each … worshipping a different god. Every god is nameless. … a mass baptism is arranged in which every god receives the name ‘Peter’. (Textor 2010: 112)

We have seen that emptiness as such does not provide a difficulty for the account (“Vulcan”), and nor does group baptism. But the combination appears to raise a new problem, since the baptizer cannot have a plural intention with respect to the various gods: there are no gods.

Name introduction can take place without direct contact with the bearer, and this is what must happen in empty cases. Even so, some kind of prop is often required. For Leverrier, this was the unexplained rate of precession in the orbit of Mercury, something for which he postulated Vulcan as a cause. In Textor’s case, the props are the 3,000 worshippers. These can be the objects of a plural intention, just as in Group baptism, except that this time the immediate objects of the intention are just props, not bearers. The supposed bearers are, for each prop, the god it worships. Just as Leverrier’s introduction of “Vulcan” can be seen as based on the supposition that there was a unique cause of the rate of precession, so Textor’s baptizer’s action is based on the supposition that for each worshipper there is a unique god worshipped. The baptizer does manage to do 3,000 things, as well as one thing, just as the tyrant in Group baptism managed to do 100 things, as well as one thing.

For the final examples, we’ll get rid of the props.

Multi-object hallucination (1): A subject hallucinates many gremlins. She thinks that she can discriminate the gremlins visually.

There is no special problem in introducing names in this case, either individually (“You [singular] are to be called ‘Peter’”) or collectively (“I name each of you ‘Peter’”). The names will, of course, all be empty, for in reality there are no gremlins. The subject supposes that she is confronted with many gremlins, just as in Nightmare the professor dreams she is confronted with many students with guns. The plural intention relating to what she supposes to be objects ensures that there are many sub-acts.

Multi-object hallucination (2:) A hallucination like the first case except that the gremlins do not appear to the subject to be discriminable. They are flitting around and are never stable enough for the subject to have any idea how many there are. Even so, the subject says to herself: “I hereby name them all ‘Peter’, however many of them there are”.

We seem to have many names, and a single act of naming that is not composed of discrete sub-acts.

We might doubt whether this is genuinely an act of naming, or something more like a recipe for how to perform such acts (as in Vladimir). Perhaps she needs some more determinate supposed contact with the gremlins to be able to name them, and her remark is a kind of aide memoire for which name to bestow if the opportunity arises.

Alternatively, one might allow that she can name the gremlins. She can think of them (or take herself to think of them) collectively, without knowing how many there are, just as the professor in Nightmare can think about the students collectively, without knowing how many there are. On this treatment, her act will have many sub-acts, just as the dreaming professor’s did, and so it turns out not to be a counterexample to (O2).

The upshot of this section is that (O2) stands: for each specific name, there is a unique originating event in which it and it alone was introduced.

7 Propagation

What makes two uses of a name uses of the same name? The answer to be defended here is that there is an event of name origination to which both uses trace back. Many would agree that causally isolated uses of names cannot be uses of the same name, just as causally isolated myths do not count as the same myth, even if they are qualitatively similar. But not just any kind of causal contact will sustain the propagation of a name. No doubt when George Wright chose the name “Paris” for Paris, TX, he was influenced by the practice of using the name for Paris, France. But this causal connection is not the kind that makes his use of “Paris” a use of the same specific name as the one for the French capital.⁸

At a minimum, non-originating uses need to be animated by some kind of conformist or deferential intention: to use the name as it is used in the relevant community, by the existing users. George Wright did not have these intentions. He knew he was starting something new.

A name is propagated in two ways:

• A. new users acquire the name.

• B. an existing user uses the name again.

For case A, a new user who does not herself originate the name must acquire it by contact with an on-going practice of its use. This gives rise to our first question:

• 1. Under what conditions does one successfully acquire a name?

Once one has acquired it, one can use it later. This is case (B), and gives rise to our second question:

• 2. Having acquired a specific name, N, what makes a subsequent use of a name a use of N (rather than a use of some other name, possibly of the same genus as N)?

To acquire a name, that is, to become initiated into an existing name-using practice, one must have encountered the name: one must have heard or seen or otherwise perceived a use of it. Acquisition involves a de re intention concerning that token. A self-conscious and sophisticated subject might inwardly think: that’s something I will add to my repertoire, using it as those people do. This intention can be regarded as having two components, one syntactic and the other semantic. The syntactic component relates to reproducing the same word. The semantic component relates to “using the word as the others do”: using the name to refer to what the others use it to refer to.

I suggest that the syntactic component of the intention must be fulfilled, if name acquisition is to occur; and the semantic component must be present, even if it is not fulfilled. Fulfilling the syntactic intention will result in acquiring a mechanism of name-reproduction, a mechanism that produces copies of the source name. It is the fact that they are copies that makes uses controlled by this mechanism uses of one rather than another name. It is the fact that any copying process may be imperfect that ensures that uses of the same name may differ phonetically or orthographically. A blurred or imperfect copy is still a copy.

Typical acquisitions probably don’t involve explicit intentions of the kind described; and talk of copying mechanisms is, in the present state of knowledge, metaphorical. So more needs to be said about both these issues.

We can attribute implicit intentions to creatures we may not consider have formed any corresponding explicit intention. The chimp is trying to get the banana from the Perspex box, but is having trouble as the stick he is using as a tool (for the edification of the psychologist spectators) is too short. We could rightly be entirely confident of the truth of this claim without committing ourselves to a chimp’s capacity to form an explicit intention involving the concepts banana, Perspex or box. In the case of name usings, one sign of the relevant implicit intentions is the acceptance of corrections. If a child calls her toy an ephalant, it’s quite likely she will respond positively to the suggestion that she should call it an elephant. She is trying to conform. She may go on to issue imperious corrections herself to other children, showing that she expects them to conform too. When her acquisition has successfully stabilized, she can produce the word “elephant” without difficulty, as a causal product of previous interactions. That is what can be metaphorically described as the operation of the “elephant” copying mechanism.

When acquiring a name, a new user must aim to use it to refer to what the others refer to with it, but in this she may fail. Allowing for failure does justice to the obvious fact that someone, especially soon after acquisition, may use a name to refer to something to which it does not refer. A child might think it is the new puppy, not the new kitten, that its parents have started to call “Whiskers”. She comes to use the name “Whiskers”, but wrongly applies it to the puppy rather than the kitten. The name “Whiskers” used by the child needs to be the same name as the name “Whiskers” used by the parents, if the child is to count as misapplying it. Otherwise she would be correctly applying a distinct but same-sounding name for a puppy, a name of which she would be the only user. In reality, she is using the very name her parents have introduced, and her mistake is using this name for something that is not so called. The possibility of successfully using a certain name, even if one’s semantic intentions concerning it fail, gives rise to the possibility of a name changing its reference, as will be discussed later.

Acquisition is deferential in that an acquiring subject must aim to use the name as his community does. This is deference to the meaning the name has, not deference to judgments in which the name is used. There need be no experts. Deference is simply intending to join in the practice, and one may be a violent critic of the judgments made by the majority of the other practitioners, for example, an atheist among believers, or a believer among atheists.

In this respect, the kind of deference needed in the present account of name acquisition differs from the kind of deference envisaged by Burge (1979) and Putnam (1975). Everyone agrees that someone may use the word “arthritis” to express the false belief that he has arthritis in his thigh. Burge suggests that one sign that the subject genuinely uses the word “arthritis”, as the word is used in his (and our) community, is that he defers to his doctor’s judgment, relinquishing his belief when the doctor tells him that he can’t have arthritis in the thigh. Burge’s aim is to show that the extensions of some expressions are individuated externalistically, a project not on the radar of the present paper. He compares the actual situation, in which the subject makes a mistake, to a counterfactual one, in which the conventional use of the word in our subject’s community comprises some inflammations in the thigh, the upshot being that in the counterfactual situation the subject does not make a mistake.

The only aspect of the example that bears on present themes is the test for whether or not, in the actual situation, the subject is genuinely using the word “arthritis” with the meaning it actually has in the community. An old-fashioned view might be that it is an analytic truth that arthritis is a disease of the joints, something true in virtue of the meaning of “arthritis”, so that someone who did not know this would not understand “arthritis”, or not understand it fully, and so would not be able to use the word, or not be able to use it competently. In tension with this opinion is the view that our subject’s deference to the doctor shows that he does understand the word “arthritis”, despite his false belief. However, it should not be thought that the doctor’s expertise plays a special role in the story, or that if the subject did not defer to the doctor, that would show he did not understand “arthritis”. A genuine “arthritis”-user might respond: “I don’t accept your verdict: I need a second opinion about whether I have arthritis in my thigh.” This would be equally good evidence that he is a genuine user of “arthritis”. The kind of behavior that would be evidence for the other conclusion is the Humpty-Dumpty style “Well, I have what I call arthritis in my thigh”.

The other species of propagation, case (B), involves an existing user using the name again. One might have a name in one’s repertoire without recalling any specific other occasion of its use, by oneself or another. So, in case (B), other uses are not always the de re target of an intention. That’s consistent with other uses always being the target of de dicto intentions: one aims to use the name as before, however that was. But this intention, though no doubt often present, is unexplanatory, since it involves thinking, albeit unspecifically, of other uses of the same name, and that is the notion we are trying to clarify.

A subsequent use of a name counts as a use of the name N just if it is produced by the copying mechanism that was installed when N was acquired. This does not require explicit intentions, and carries no semantic burdens. Suppose that the specific name for Paris, TX, and also the specific name for Paris, France, are both in Scott’s repertoire and he utters the words “I’m going back to Paris”. It’s easy to spin a tale about the sorts of further behavior that would be evidence that he has used one specific name rather than the other. But our concern is what constitutes this fact. Evidence might be misleading, especially if the subject has crazy beliefs. The claim about copying mechanisms is intended to point in the direction of the constituting fact, which is the kind of causal connection that would obtain between acquisition and current use were acquisition properly described as the installation of a copying mechanism. The relevant causal chain is one that allows for degradation, just as in literal copying machines. Indeed, it is doubtful that there are any clear limits to the qualitative changes that can be induced by copying. That’s why the same name may be spelled and pronounced differently by different speakers or at different times. The crucial point is that the present use must causally trace back to the operation of a functional mental state induced by the subject’s acquisition of the name, and the causal process must be similar to the causal processes involved in copying.

Using a name does not require a capacity to distinguish its bearer from other bearers of the same generic name. If two very similar twins are both called David, you might be lucky and encounter only one of them and learn his name. So long as you never encounter the other, there’s no question but that you have acquired the specific name of the twin you met, even if you would be unable to tell him apart from his twin. If you encounter both, all sorts of different possibilities open up.

Even if later uses do not involve explicit deferential intentions, they are deferential in both a backward-looking and a forward-looking way: they inherit deference from their acquisition, and they are subject to correction. A later use of N is causally derived from an event of acquisition in which the acquirer aims to use the word as it is used in her community. That aim is quite general: to use it now and in future as it is used in her community. A later departure from community use would be inconsistent with the earlier intention. That is a kind of backward-looking deference. Looking forward, a user of a name must be ready to deal with accusations of confusion, accusations, that is, that she has not managed to conform her use of the name to that of the community. A Humpty-Dumpty response is not acceptable.

8 Meaning and Judgment

In name acquisition, syntactic conformity involves using a word you have identified in the mouth of another. Semantic conformity involves using it with the same meaning. This does not involve deferring to the judgments of the other users. Suppose you are firmly atheistical, but are studying the behavior and rituals of a people you regard as steeped in foolish superstitions: they believe in a whole panoply of invisible spirits. You need to learn the names of these spirits, to understand their rituals better. This does not involve sharing any salient judgments with the people you are studying. You might wish to express your disagreement by an assertive utterance of a sentence like “Bahlam does not exist”. For your assertion to have the impact it should, your use of “Bahlam” must in some sense coincide with theirs: the very thing they are so keen to affirm, that Bahlam exists, you wish to deny. In this case as in every other, disagreement in judgment requires agreement in meaning.

If one thinks that names play an essential part in analytic truths, one will not accept this separation of meaning and judgment. For example, on some views it is a matter of meaning alone that the sentence “Hesperus is visible in the evening” is true. Given that understanding is knowledge of meaning, it follows that someone who has understood the sentence is committed to its truth. Contrary judgment would entail misunderstanding. Most people currently reject the view that names play an essential part in analytic truths, and I will align with that majority without further argument.

Here is an objection: If one rejects the name-involving judgments of a community, it seems that one cannot be “using the name as they do”. The atheist supposedly wishes to use names like “Bahlam” in the way the superstitious do. But the superstitious use the name to refer to an entity they believe exists; the atheist does not. So the atheist cannot use the name as the superstitious do.

The objection calls for a renewal of the distinction between agreement in meaning and agreement in judgment. It is not part of the meaning of “Bahlam” that it refers to something that certain people believe exists. It is not a semantic convention that one should use “Bahlam” only for something one believes exists. One can accept the semantic conventions governing the use of “Bahlam” without prejudice to atheism.

But what are these conventions? One answer is that, in the relevant group (to include the investigator), there’s a convention to use “Bahlam” to refer to Bahlam. The investigator’s intentions are fulfilled if he conforms to the prevailing convention, using “Bahlam” to refer to Bahlam.

There is a usage of “refer” according to which this must be false. If “refer” is thought of as an ordinary extensional relational expression, “x refers y” entails “there exists something, y, such that x refers to y”. But there does not exist anything, y, such that “Bahlam” refers to y. Those who use “refer” in this extensional way will prefer to replace “refer”, in the previous paragraph, by “purport to refer”. In my view, this preserves the truth of the claims. I say more about empty names in Sect. 10.

9 Shifting Reference

On the account offered here, the conditions on being the same name diverge from the conditions on having the same bearer. The present theory is a causal theory of the identity of names, but the causal facts that ensure identity of names may fail to ensure identity of reference.

The “semantic reference” of a name, as used in a community, is its conventionalized, stabilized or normalized speaker-reference in the community. “London” refers to London among many speakers who live in England (and elsewhere) because it’s a conventional or stabilized or normal fact about these speakers that they use the specific name “London” (I hope you know which specific name of the genus “London” I have in mind) only if they intend thereby to refer to London. The notion of semantic reference is a theoretical one, and one that needs to be constructed to suit theoretical purposes. In the present case, we need a conception of semantic reference that will supervene on use and help explain features of usage (for example agreement, disagreement, correction). Basing semantic reference on speaker-reference is the most straightforward, and perhaps the only, way to achieve this. Speaker-reference can be theoretically described without any theoretical commitment to semantic reference, so the supervenience has a reductive character. Much work has been done, and much remains to be done, to sort out what the supervenience relation should be based on.⁹ Here I give a trio of possibilities (convention, stability, normalization); a determinate thesis would need to choose from among them, and also clarify the preferred option. All that is well beyond this paper’s scope. I will use the word “convention” to express the relevant relation; this should be treated as having merely the status of a place-holder. In these terms, shifting semantic reference is a matter of a shifting convention.

For a name to change its reference it has to remain the same name through the change, having one referent at one time, another at another. We have seen that acquiring a name does not guarantee using it to refer to what the source community uses it to refer to (“Whiskers”). This is the key possibility for reference shift.

Evans uses apparent cases of reference shift as problems for “the causal theory of names”, by which he understands a theory which, unlike the one offered in the present paper, makes no distinction between the causal facts that induce name-identity and those that induce reference-identity. The present theory is a causal theory of the identity of names, but not of the reference of names. When Evans targeted “the causal theory of names”, he had in mind a causal theory of the reference of names. His examples tell against such theories of reference, but they help rather than harm the causal theory presented here, which relates to the identity of names.

The first example I shall discuss is Jack:

two babies are born, and their mothers bestow names upon them. A nurse inadvertently switches them and the error is never discovered. It will henceforth undeniably be the case that the man universally known as ‘Jack’ is so called because a woman dubbed some other baby with the name. (1973: 11)

The present account accepts the story at face value: the very same specific name “Jack” was first used for the real baby, and then for the changeling. The shift did not occur in the course of acquisition but in the course of case (B) propagation. In using the name “Jack”, the mother was using the name she had first used for her baby. Even when used for the changeling, it was produced by the very mechanism that was installed when she first called her own baby by that name. Though aiming to use the name as she had before, after the swap she made what at first was a systematic mistake about the reference. At all times, the mother’s uses were governed by the syntactic intention to use the same name, and the semantic one to use it with the same reference as before. Soon after the swap, the syntactic intention succeeded, but the semantic intention, whose presence was necessary for this to be a propagated use of the name, failed. Before the swap, it was conventional (among a small group dominated by the mother) to use “Jack” for the mother’s real child. Some time after the swap, it became conventional to use the name for the changeling. The shift in convention was facilitated by dominance of the mother’s uses in the early stages.

Evans’s other example is Madagascar. He quotes from Taylor’s Names and Their History:

In the case of ‘Madagascar’ a hearsay report of Malay or Arab sailors misunderstood by Marco Polo… has had the effect of transferring a corrupt form of the name of a portion of the African mainland to the great African Island. (Evans 1973: 11)

The story needs to say that, despite the corruption, Marco Polo acquired the name “Madagascar” (otherwise he did not use that name for the island) but mistook its reference. His uses of the name for the island became dominant in the wider European community: it became a convention, among many Europeans, to speaker-refer to the island when using “Madagascar”. Hence the island came to be the semantic referent of the name in some European languages. The shift occurred in the course of acquisition not in the course of case (B) propagation, and so differs from Jack.

Madagascar also differs from Jack in another respect. We don’t generally hear much about the linguistic developments among the “Malays or Arabs”. It would be natural to imagine that their use of the specific name “Madagascar” was unaffected by that of the visiting Europeans. In that case, the same specific name has one reference among the original community of users and another among Europeans. Pre-theoretically we are neutral about whether this is the right thing to say; we can accept it as a consequence of a good theory.

Only a theory which, like the one offered here, distinguishes syntactic success (successfully using the same name) from semantic success (using it with the same reference) can accept Jack, Madagascar and similar cases at face value.

10 Empty Names

Ever since I mentioned Textor’s gods, bearerless (empty) names have been a possible source of unease in the present discussion, and this will now be addressed directly. Empty names prompt at least two questions. (1) Semantic deference consists in aiming to refer with a name to that to which it is used to refer in the community from which one acquired it. But if a name is empty, it’s hard to see what this intention could amount to, and in particular how the reference of “Pegasus” can be distinguished from the reference of “Vulcan”. (2) Can we make sense of an empty name changing referent, while remaining empty? Can an empty name come to have a referent? Or a non-empty one come to be empty?

For those who accept an intensional notion of reference, that is, one that does not entail the existence of what is referred to, the background theory is fairly straightforward. A new user of a specific name, empty or otherwise, must attempt to use it with the reference is actually has. The last phrase is to be understood in such a way that, since “Pegasus” refers to Pegasus in our community, conformity with that practice requires using the word to refer to Pegasus. Equally, we can speak without embarrassment of the anthropologist aiming to refer to Bahlam when she uses “Bahlam”, and can treat this intention as distinct from the intention to use “Bahlam” to refer to Pegasus even though, from the extensional perspective, “Bahlam” and “Pegasus” do not differ in reference. But the intensional perspective is not widely accepted, and is likely to provoke accusations of Meinongianism and other supposed sins. I cannot attempt a full treatment here, but will make some remarks designed to appease extensionalist scruples. When these framework observations are in place, I will turn to the specific questions posed in the first paragraph of this section.

Earlier (Sect. 8) I suggested that someone who is firmly extensionalist about reference could think of an intensionalist’s use of “reference” as equivalent to “purported reference”. It’s not that “Pegasus” refers to Pegasus (such theorists might say) but that it purports to refer to Pegasus. Although this does not address the logical issues, and although it is inconsistent with a prejudice in favor of purely extensional semantic theories, it is enough for present purposes that these claims be regarded as correct.¹⁰ For emphasis, I’ll often use “purports” idioms in the remainder of this section, though this is the same in meaning as my normal use of the unadorned idioms of reference.

The semantic intention required to acquire a name is to intend to use it to purport to refer to what one’s source purports to refer to when using it. In this, one does not have to succeed. But in the absence of such an intention, explicit or implicit, there’s nothing to make it the case that the subject is affected by the fact that a name, a vehicle of communication, is at stake. Merely reproducing the word, without semantic intention, would be like reproducing one’s singing teacher’s note, a non-sematic attempt which, if successful, constitutes a non-semantic achievement. By contrast, the child who thought Whiskers was the puppy had achieved something semantic: she regards the word as a name, a vehicle of communication, indeed as a name of a pet. The semantic intention she fails to achieve is more specific.

Although there is, in a sense, nothing to distinguish the reference of “Pegasus” from that of “Vulcan”, purporting to refer to Pegasus is different from purporting to refer to Vulcan. “Pegasus” and its normal users do the first but do not thereby do the second; “Vulcan” and its normal users do the second but do not thereby do the first. This is all the distinction we need.

We can now turn to the second question, whether any reference-shifting phenomena affect empty names. In his Addenda to Naming and Necessity, Kripke comes down firmly in favor of the possibility of names changing bearers, even names which many would regard as empty:

So real reference can shift to another real reference, fictional reference can shift to real, and real to fictional.” (Kripke 1980: 163)

For Kripke, fictional reference is not the same as emptiness, since he believes there are fictional characters and that these are the bearers of fictional names. However, his claim can be supported even if we think of fictional names as a species of empty name.

There are three cases:

• a. an empty name acquires a bearer

• b. a non-empty name becomes empty

• c. semantic change among empty names.

For case (a), Kripke says (attributing the point to David Lewis) that in a variant of Madagascar in which the original native use was for a mythical place, we would still have a change of reference (1980: 163). For case (b), he offers “Santa Claus” (which no doubt once referred to the historical Saint Nicholas, a fourth century bishop) (1980: 93).

He does not explicitly envisage a (c)-type case, but it does not seem too difficult to construct one.

Joke: A small group, as a kind of “in” joke, decides to use “Holmes” for Watson and “Watson” for Holmes. Then all the texts are destroyed in some cataclysm. People learning from the small group don’t realize there has been a jocular inversion. These benighted souls become the only users of the specific names.

Although the purported referent of “Watson” used to be Watson (the side-kick), it has become Holmes (the detective), and the purported referent of “Holmes” is now Watson. We again apply the principle that semantic reference is conventionalized speaker’s reference. In the example, the purported reference of empty names has changed.

11 Conclusion

That which makes for sameness of name, viz. being anchored in some originating event, does not necessarily make for sameness of reference. The question “Which name?” has a historical answer. The question “What referent?” has a non-historical answer, one requiring relativization to a time and a group of name users. The purported referent of a name in a group at a time is what a convention among those people at that time says they should purport to refer to with the name.¹¹