Start Submission Become a Reviewer

Reading: Are Reference Rules Inessential to Meaning?


A- A+
Alt. Display


Are Reference Rules Inessential to Meaning?


Kirk Ludwig

Indiana University, US
X close


This article responds to a case-based argument by Mark Richard that rule of reference is not essential to meaning. It objects that the argument requires shifting between understanding the relevant term in the case, ‘marriage,’ as a determinable, in order to support one premise, and a determinate, in order to support another. On no univocal interpretation can both premises be made true.
Keywords: meaning, reference
How to Cite: Ludwig, K., 2020. Are Reference Rules Inessential to Meaning?. Metaphysics, 3(1), pp.92–102. DOI:
  Published on 03 Dec 2020
 Accepted on 10 Nov 2020            Submitted on 07 Nov 2020

1. Introduction

In ‘Is Reference Essential to Meaning?’ Mark Richard argues that reference is not essential to meaning (Richard 2020). This may seem, at least for many words, to be a trivial thesis. The meaning of ‘I’ does not change with different uses but its referent does, and it has no referent except on an occasion of use. Richard explains, though, that he groups ‘the rules which map use to semantic value on the side of reference broadly construed’ (2020: 15). So the thesis is not (really) that reference (as we usually think of it) is not essential to meaning but that the rule that we associate with a word that determines relative to context its reference or extension (reference rule, for short) is not essential to its meaning. That is, sameness of meaning does not entail sameness of reference rule. Of course, the rule that determines the contribution of an expression to the truth conditions of a sentence in which it is used is usually thought to be its meaning. Thus, we might say, tongue in cheek, that the thesis is that meaning is not essential to meaning. More sympathetically, the thesis is that what people usually think of as meaning is not meaning, but can vary while meaning (what it really is) remains the same.

Let’s suppose that the rules for complex expressions are derived from the rules for their components and their mode of combination. Then we can focus on semantically primitive expressions whose meanings are not determined compositionally. Let’s further restrict attention to primitive expressions that in use in a context have a referent or an extension in the sense of a set of things or n-tuples of things that they are true of. Call these extensional primitives. Then we may put the thesis as follows:

[Rule Anti-essentialism (RA)]

It is possible for two extensional primitives to have the same meaning (in the same or different languages) but have associated with them in (in those languages) different rules for determining reference or extension relative to a context.

Richard offers a case-based argument for this, but also gestures toward a background theory developed in his new book Meanings as Species (Richard 2019), which argues that meanings are a kind of historical entity like species. This general thesis doesn’t entail (RA). But it does play a supporting role in the argument for (RA), though even there its role is merely to prepare the ground for a proposal, which might yet be false even if one accepts the idea that meanings are like species in being individuated historically.

In the following, I argue that Richard’s case-based argument for (RA) is unsuccessful.1 Before I explain the argument and objection, I want to look at some assumptions in the preliminaries that are relevant to the discussion.

2. Stage Setting

Richard rejects Quine’s behaviorism. While he agrees that we should look to verbal behavior to understand meaning, he does not think of meaning as reducing to dispositions to verbal behavior neutrally described. Instead, he suggests,

[Tracking Thesis (TT)]

Our dispositions to verbal behavior are grounded in mental representations we form as a result of our interactions with other speakers. Talk of meaning is, I think, talk that tracks the creation, development, and modification of these sorts of representations (Richard 2020: 3).

What sorts of mental representations? It’s a grab bag: intentions to convey something, beliefs about usage patterns, what speakers expect you to infer, presuppositions about the situation they are talking about, etc. (E1–E3 in the paper). It is not clear why we should think that all are equally relevant or relevant in the same way to meaning.

Setting this aside, what is meant by saying talk of meaning tracks ‘the creation, development, and modification of these sorts of representations’? If this is intended to be about the function or concept of meaning, it is hardly obvious that the function of talk of meaning is to say something about the creation, development, and modification about these sorts of representations or that the concept of meaning is about these things. Rather, talk of linguistic meaning is talk of properties of expressions in a public language that enables them to perform the functions they perform for us in communicative interchanges—what we learn when we learn a language. There are such properties and it is our shared knowledge of them that enables us to communicate using those expressions. They perform these functions without our having to have any detailed knowledge about their history. (TT) looks, not like a commonsense characterization of what talk of meaning is about, as it is presented, but the expression of a theory. Let us hold it at arm’s length while looking at the case-based argument. It can’t be presupposed.

While Richard rejects Quine’s behaviorism, there is still something of the spirit of Quine in his taking up a roughly anthropological stance toward meaning. ‘There are these creatures, they have representations of various sorts, intentions to convey something, beliefs about usage patterns, what speakers expect you to infer, presuppositions about the situation they are talking about. What concepts should we take up to keep track of this?’ But we are not outsiders looking in on the game of language trying to develop a conceptual framework to help us make sense of it. We are already part of the game. We understand that language is a social practice. We understand that we are doing things with others when they speak and we listen and vice versa. We have an array of categories for terms to fit into. In practice we approach interpretation holistically, already equipped with the concepts of meanings conceived of as roles expressions have in communicative exchanges in virtue of its being accepted by the parties involved that they have those roles.2 Even more specifically meaning is a special general type of status function; particular meanings are specific instances of that type; we understand this; and that informs how we interpret what others are doing.3

3. The Argument for (RA)

My main focus is the case-based argument for (RA) given in section 4 of Richard’s paper.4 The argument has two stages.

Stage 1

In 1919 people in the United States and Canada believed a lot of things about marriage (or expressed beliefs using ‘marriage’). They would, for example, assent to (in 1919 English) [1–5] (I choose a subset of the examples Richard uses).5 Since [1–5] are in the present tense, what people in the US in 1919 assent to are claims about the situation in 1919.

[1] Marriage is a contract between a man and a woman (in North America).
[2] You can’t (in North America) marry (or be married to) more than one person at a time.
[3] It is permissible that married people have sex with one another (in North America).
[4] It is widely believed (in North America) that it is wrong to have sex with a married person to whom you are not married.
[5] Married people (in North America) generally live with each other and sleep in the same room.

In 2019, people still assented to [2–5] but not to [1]. Again, since the sentences are in the present tense, what they assented to are claims about the situation in 2019.

Richard claims that the meaning of ‘marriage’ hasn’t changed since 1919. He offers as evidence that people in North America in 2019 accepted [N] (slightly modified from Richard’s version) because they and people in North American in 1919 accepted [O].

[N] In both 1919 and 2019, people in North America thought that married people mostly sleep in the same room.
[O] Married people mostly sleep in the same room.

The idea is that unless people in 1919 and in 2019 both meant the same by [O], we cannot infer validly that people in 1919 and in 2019 believed the same thing on the basis of their accepting it.

Stage 2

Imagine a counterfactual world, W, in which, up until 1919 everything is the same as in the actual world, @, but subsequently social attitudes toward marriage do not change and in 2019 in W people in North America (henceforth ‘NA’) assent to [1] just as people did in 1919. Assuming that what a word means at t in a community supervenes on the history of the world up to t, what ‘marriage’ means (in NA) in @ and in W in 1919 is the same. There have been no relevant changes in usage from 1919 to 2019 in W (and everything comes to a cataclysmic end shortly thereafter), so in W what people mean by ‘marriage’ in NA in 2019 is what people in NA meant in 1919. Thus, given the result of the argument in stage 1, what people in NA in 2019 in @ mean by ‘marriage’ is the same as what people in NA in 2019 in W mean by it. However, in W people in NA in 2019 deny that same-sex couples fall in the extension of ‘married.’ Thus (the argument goes), their reference rule in 2019 (in W) is not the same as the reference rule of their counterparts in the actual world in 2019.

Give all of this, we can formulate the following argument.

[The Marriage Argument]

  1. In 1919 in NA, both English speaking people in @ and English speaking people in W meant the same by ‘marry’, ‘marriage’, etc.
  2. In 2019 in W, English speaking people in NA meant the same by ‘marry,’ ‘marriage,’ etc. as English speaking people in NA did in 1919 in W.
  3. The reference rule for ‘married’ in @ in 2019 in English is not the same as the reference rule for ‘married’ in W in 2019 in English.
  4. From stage 1, in 2019 in @ English speaking people in NA meant (in English) the same by ‘marry,’ ‘marriage,’ etc., as English speaking people in NA did in 1919 in @.
  5. From 1, 2, 4, English speaking people in NA in @ in 2019 mean (in English) the same as English speaking people in NA in W in 2019 by ‘marry,’ ‘marriage,’ etc.
  6. From 3, 5, the meaning of two expressions in a language L can be the same though their reference rules in L are different.

The situation as depicted by the premises of the argument is illustrated in Figure 1, where ‘=’ indicates that in the relevant communities ‘married’ means the same or that the reference rule is the same (as appropriate). ‘R-rule’ means the reference rule for ‘married’ in the community. The first row is the situation (across worlds) in 1919. The second is the situation (across worlds) in 2019.

Figure 1 

The Assumptions of the Case-Based Argument against (RA).

The situation in 2019 shows sameness of meaning but difference of reference rule.

There is a question mark next to the representation of the R-rule in @ in 1919 and 2019 being the same in Figure 1. The argument doesn’t explicitly take a stand on this. But as Richard notes (his objection 2), it might seem obvious that in 1919 in @ the reference rule excludes same-sex couples while the reference rule in 2019 in @ does not (on the same ground as the claim that the reference rule in @ in 2019 differs from that in W in 2019). If this is so we could advance a shorter argument with two premises: (1) in 2019 in @ English speaking people in NA meant (in English) the same by ‘marry,’ ‘marriage,’ etc., as English speaking people in NA did in 1919 in @; (2) in 2019 in @ English speaking people in NA attached (in English) a different reference rule to ‘marry,’ ‘marriage,’ etc., than English speaking people in NA did in 1919 in @.6

Richard responds that the rule of reference associated with an expression supervenes not just on past use of the expression but also on future use. So the rule in 1919 in @ is the same as in 2019 in @ but not the same as the rule in 1919 in W, because it is beholden to future use.7

4. Is the Argument Successful?

I start with an objection to the thesis that reference rule (old-fashioned meaning) is determined in part by future usage of an expression. Then I turn to the main objection, which is, roughly, that the argument requires shifting between different things meant by ‘marriage’—more general versus more specific—in support of different premises.

Richard claims is that the rule of reference that attaches to an expression at a time is determined (at least in part) by the term’s future use. However, rules of references are our rules. They don’t magically attach to words. We attach them by conventionally associating them with words.8 Conventions are realized in the intentions and attitudes of members of a community. The conventions realized in a community at a time are determined by the intentions and attitudes of the members of that community at that time. No time in the future (or the past) of that time is relevant to the conventions that obtain in the community at that time. No conventions in continuations of the community in the future determine what the conventions are now. In slogan form: sufficient unto the time are the conventions thereof. This is a condition on our being in a position to grasp the rules and so to know the truth conditions of what people say.9

So the future uses of an expression are not relevant to what the rules of reference are now. Therefore, rules of reference attaching to words at a particular time do not supervene on the entire history of use (including future use) of the words.10 Therefore, the rule for ‘marriage’ in 1919 in @ and in 1919 in W is the same because by hypothesis everything relevant is the same. If the rule in 2019 in @ is different from that in 2019 in W, then the rule in 2019 in @ is not the same as the rule in 1919 in @. While Richard’s argument doesn’t require the assumption that the rule in 2019 in @ is the same as the rule in 1919 in @, the assumption gives the argument a point, since otherwise there is a simpler within-worlds argument for the same conclusion. Do I concede therefore that there is a shorter sound argument? No, the shorter within-worlds argument is unsound for the same reasons that the between-worlds argument is.

Richard offers an argument for the relevance of future use to current rules of reference. The argument is that the following dialogue is fine as it stands.

A (in 1919): It is impossible for men to marry one another.

B (in 2019): When A uttered ‘it is impossible for men to marry one another’ in 1919, he said that it was impossible for men to marry one another. And what he said was false.

The idea is that if what B says is to make sense, she has to be supposing that the reference rule in 2019 is the one that determines the extension of ‘marry’ in 1919, since counterfactually it might have been different in 1919 from what it is (e.g., in W). Considerations developed in the main objection will show what goes wrong here.

The main objection is that the sense in which ‘marriage’ means the same in @ in 1919 and in 2019 and determines the same reference rule is not the sense in which the meaning of ‘married’ in 1919 and 2019 in W is the same and determines a different reference rule.

The concept of marriage is a concept of a certain kind of social relation between people that has a recognized status in the community. Its reality consists in that status that it is given in the community. The status reconfigures economic, legal, and social relations between those who are married and between them and other institutions in their community. The details, however, vary between different communities, and vary even within communities at different times. Thus, we recognize that there are many different determinate varieties of marriage. In different communities or jurisdictions, there are different entry and exit conditions for being married, as there are during different historical periods in the same societies or countries. There are differences in the form of the social relationship between the married parties, and differences in the rights and duties of those married to each other, as well differences in the rights and duties of others with respect to them. There are differences in eligibility conditions, differences in whether the relation is one-one (monogamy) or one-many (polygyny and polyandry) or many-many (group marriage), and so on. In a country like the United States, what counts as marriage varies from state to state, since laws governing marriage are state laws. Today, in most states, individuals can marry without parental consent or other authorization on reaching 18. But in Nebraska the age is 19, and in Mississippi it is 21 (the age of majority in both states). What marriage implies is different in different jurisdictions, e.g., California, Louisiana, Arizona, Texas, Washington, Idaho, Nevada, New Mexico and Wisconsin are community property states: all property acquired after marriage is owned by both spouses. Other states are common law property states in which the property acquired by one member of a married couple is owned solely by that person. Marriage between first cousins is illegal in most states, and a criminal offense in some (the Dakotas, Nevada, Oklahoma, and Texas), but not in all. There can be differences also in the number of people who can be united by the relation in different jurisdictions and at different times. Polygamy was outlawed in the United States by the Edmunds Act in 1882 but was both legal and practiced (in some jurisdictions) before that. There are still today many countries where polygamous marriages are recognized. To complicate matters there is a further distinction between civil and religious marriages. Civil marriages are legal contracts; their nature and status is tied to the legal system in which they are defined. Religious marriages are recognized instead within a religious community. One may be married in accordance with the rites of a religious community but not legally married, and vice versa. Thus, the generic concept of marriage has to subsume all of these particular varieties.

The generic concept of marriage then is the concept of a determinable.11 When we talk about marriage, we often shift back and forth between using the concept of the determinable and concepts of its determinates. The concepts of the determinates have their application conditions determined by the rules in particular communities or jurisdictions at particular times, or sometimes a range of jurisdictions or communities (e.g., the United States). So we say, e.g., in 2019 an individual can marry in Mississippi without parental consent only upon reaching the age of 21. This signifies that we intend to be talking about the variety of marriage which is defined by the laws of the state of Mississippi in the United States in 2019.

The relevance of this to Richard’s argument is that in asking whether people mean the same by (their uses of) ‘marriage’ in 1919 and 2019 in North America, we have to ask whether we have in mind uses that express the generic concept (the concept of the determinable) or a concept of one of the more determinate forms of marriage. We get different answers depending what we have in mind. In 1919, plausibly, if we have the generic concept in mind, people who were reflective enough would have been able to recognize that one determinate form of marriage could involve same-sex unions. Indeed, Steindorff reported in 1904 on male-male marriage ceremonies in the Siwa Oasis in Egypt (Steindorff 1904: 111). Someone reading the report in 1919 would hardly have denied the truth of the report on conceptual grounds, but have understood it as a straightforward empirical claim about marriage customs among the Berbers at Siwa.

So when Joe Someone asserts (or assents to)

[1] Marriage is a contract between a man and a woman (in North America)

in 1919, he either (i) accepts that marriage, as a determinate legal practice in North America, does not subsume same sex couples or (ii) just that the extension of the generic concept of marriage at that time in North America (or perhaps world-wide, though in that case he would be wrong) did not include same sex couples.12 If it is the generic concept that is expressed, there is certainly no change in what is meant by ‘married’ and no change in reference rule between 1919 and 2019. The same goes even for ‘legal marriage in North American jurisdictions’ because the extension is just whatever (in this case) pairs have satisfied the legal entry requirements defined for marriage and have not satisfied the exit requirements at any given time. The extension changes with time both because of changing relations between individuals and because the conditions specified by law change with changing laws. But the meaning and rule of reference remains the same because ‘legal marriage’ expresses a concept that covers whatever at a time the laws then governing marriage count as marriage. We get a shift in reference rule between 1919 and 2019 only if ‘married’ is used to express not the generic concept of marriage or the concept of legal marriage at a time, but concepts of marriage as defined specifically by the legal rules of 1919 and of 2019 (in one or another jurisdiction). But then it is trivial that the rules of reference for and what is meant by the relevant uses of the expressions ‘marriage,’ ‘married,’ and so on, differ.

What are the implications for The Marriage Argument? Focus on premise 3 of the argument, repeated here.

  • 3. The reference rule for ‘married’ in @ in 2019 in English is not the same as the reference rule for ‘married’ in W in 2019 in English

We have to understand this to be about the reference rule for particular uses of ‘married’ in 2019 in @ and W. What are the options for understanding what concept the relevant uses are intended to express? Here are three:

  1. They express the generic concept of marriage.
  2. They express the concept of civil marriage, i.e., legal conjugal union.
  3. They express the most determinate concept of marriage as the type of legal union specified by laws in a jurisdiction at a particular time (or a disjunction perhaps if we have in mind more than one jurisdiction).

Neither option (a) nor option (b) provides any reason to think the reference rules differ between @ in 2019 and W in 2019 (for the relevant uses).13 In this case, premise 3 is false. (c) does sanction saying that the reference rules (for relevant uses) are different, and so validates premise 3. However, (c) falsifies premise 4, repeated here,

  • 4. In 2019 in @ English speaking people in NA meant (in English) the same by ‘marry’, ‘marriage’, etc., as English speaking people in NA did in 1919 in @

because relevant uses of ‘married’ in 1919 are to be understood to be expressing a determinate concept of marriage given by the content of the laws governing marriage then and similarly for relevant uses of ‘married’ in 2019. And the concept of marriage determined by the laws of 1919 in (say) the United States, and the concept of marriage determined by the laws of 2019 in the United States, are not the same. The argument faces a dilemma. The reading of premise 3 required for it to be true makes premise 4 false; whereas the reading of premise 4 that makes it true makes premise 3 false.

Figure 2 illustrates the problem. To secure the claim that in the actual world @ the meaning and reference rule for ‘married’ is the same in 1919 and 2019, we must suppose that the relevant uses of ‘married’ express the generic concept of marriage or a system relative notion (‘legally married in jurisdiction x at t’). This is shown on the left. To secure the claim that the reference rule is different between 2019 in @ and 2019 in W, we must suppose that the relevant uses of ‘married’ express determinate concepts of ‘married.’ This is shown on the right. No single concept secures the assumptions needed for the conclusion.

Figure 2 

The Dilemma for the Argument for (RA).

In light of this, we can return to the argument that Richard offered to show that we think that the rules of reference that govern the use of words at a time depend in part on future uses of those words. That argument appealed to a dialogue, repeated here.

A (in 1919): It is impossible for men to marry one another.

B (in 2019): When A uttered ‘it is impossible for men to marry one another’ in 1919, he said that it was impossible for men to marry one another. And what he said was false.

When we read this, we may regard B or A charitably. Since B is the second interlocutor, we might feel a slight tendency to favor B. If we interpret B charitably, we suppose B is making a true claim and look for a reading on which it is true. This requires interpreting B and A as asserting something about marriage in general, that is, invoking the determinable concept. On that reading, what B says is true and what A says is false. Of course, if we were considering an actual utterance by someone in 1919, there would be a question about whether that person, call him A again, had intended to be making a general claim about marriage at any time and any place (which would have been false even at the time) or a claim about marriage in North America in 1919. In the latter case, what B says in 2019 is false, because B means both that A was making a general claim and that what he said was false because the general claim is false, when A was not making a general claim.

5. Objection: ‘Marriage’ Is Univocal

To save the argument, one might argue that ‘marriage’ in 1919 English in North America and in 2019 English in North America is univocal and that no one in 1919 or in 2019 works with more than one concept of marriage or ever means in using ‘marriage’ in any context anything different than in any other. One might say:

In 1919, people in the U.S. thought that what they meant by ‘marriage’ made the sentence ‘it’s impossible for men to marry one another’ true. One thinks that the population then—as does the current population—took it for granted that ‘we all mean the same thing by ‘marriage,’’ though of course they might have then (and certainly would now) … add ‘but we disagree pretty strongly with others about whether same sex marriage is possible.’14

However, it is not plausible that people either in 1919 or in 2019 did not recognize that there is a determinable concept of marriage and various more determinate notions and didn’t routinely deploy different notions as appropriate. If your eighteen-year old daughter asks you whether she needs your permission to marry and you live in Mississippi, you should say ‘yes.’ But if you live in California, you should say ‘no.’ Here the relevant concept in each case is the one that that is determined by the actual laws or the one that is relativized to the legal system in a jurisdiction, such as legal marriage in accordance with the laws of Mississippi. For she is concerned with what the entry conditions are for getting legally married in the jurisdiction in which she lives. There is no answer to the question if it is the generic concept of marriage that one uses, and clearly different determinate notions are required for different jurisdictions. Moreover, no one has any trouble today, or would have had any trouble in 1919, in understanding: ‘Marriage takes different forms in different communities. In the United States, in 1919, marriage relates a pair of a man and a woman. But there are many countries in which polygamy is legal and it was legal in the United States not so long ago and practiced by the Mormons in Utah.’ Here we have in mind the generic notion. If someone in North America asserts, ‘You can’t be married to more than one person at a time,’ but would also accept that polygamy is legal in many countries in Africa, and was only outlawed in the US in 1882, he cannot intend, in uttering ‘You can’t be married to more than one person at a time,’ the concept that subsumes both legal and religious marriages in the United States and those in other countries. He must rather mean marriage as a social relation and practice in the United States in 1919. Even in 1919 people in the United States would have understood Steindorff’s report that in Siwa there were same-sex marriage ceremonies. Many in 1919 might have thought it inappropriate, perhaps disgusting, or wrong, and have been glad that marriage in the United States was restricted to pairings of men and women, but they would not have thought it conceptually incoherent.15 It is just a different determinate of the generic concept than the one that they are used to thinking in terms of.

How is this related to the meaning of ‘marriage’? When I developed the objection to the Marriage Argument, I asked for the most part what concept uses of ‘married’ were meant or intended to express rather than what meaning of ‘married’ we had in mind. This it leaves open whether, on the one hand, when Joe Someone says, ‘Marriage is a relation between a man and a woman,’ he uses ‘marriage’ in a more determinate rather than less determinate sense or whether, on the other, the sentence used fails to express the determinate thought that he has in mind, a thought which could have been expressed by a sentence with additional modifiers: ‘Marriage as it is practiced in North American in the present day (the first few decades or the 20th century) is a relation between a man and a woman.’ For very often the sentences we use underdetermine the thoughts we intend to express and what we mean to communicate. We are laconic because it is efficient. We opt for shorter sentences that don’t fully express our thoughts in the interests of efficiency when it will be clear from the context what we have in mind. For the purposes of the objection to the Marriage Argument, we don’t have to decide between the multiple senses and laconic speech views. The point is that what people are communicating, what thoughts they are expressing depends on what concepts they are using. In practice we look for an interpretation that makes what they say seem a reasonable contribution to the communicative exchange given what we know about their perspective, history, beliefs, etc. Once we see that there are various concepts, more or less determinate, that we operate with, we can see that the Marriage Argument founders because we have to say which ones are relevant for which judgments of sameness and difference of what is meant by uses of ‘marriage’ and what rules of reference are intended. And the intuitive judgments that Richard seeks to use to support the premises of the argument rely on choosing concepts at different levels at different points.16

6. Concluding Remarks

We are used to contrasting meaning and reference. The referent or denotation of an expression is one thing, its meaning, which determines its referent, is another.17 The extension of a predicate is one thing, its meaning, which determines its extension, relative to the way the world is, is another. The truth of a sentence is one thing, its meaning, which determines its truth relative to the way the world is, is another. But we are not used to contrasting the rule we grasp associated with an expression that determines its reference, extension, or truth value with meaning. Rather, we think that the rule we grasp associated with an expression that determines its reference as its meaning. That’s what we teach when we teach people what words mean. We don’t teach them the history of the word (or the trajectory of its future uses). What they come to understand is the rule, not the history. If we want to think of meaning as what we learn and teach, then it is the rule.

However, in an ecumenical spirit, allow that there can be different concepts of meaning, and, perhaps, different concepts are useful for different purposes. Call rule of reference meaning ‘meaningR’ and the species concept ‘meaningS’. Consider the argument with these two notions distinguished. It is immediately apparent that the conclusion cannot be established for meaningR because

Necessarily, if e1 meansR the same as e2, the rule of reference associated with e1 is the same as the rule of reference associated with e2.

That is because,

MeaningR is essential to meaningR, and meaningR is rule of reference.

However, it is an open question, at least, whether

MeaningR is essential to meaningS.

This is the same as the question whether

Rule of reference is essential to meaningS.

This question now seems, however, less freighted with significance, and the answer less likely to be controversial, however it comes out. It is a potentially interesting open question about the relation between two concepts ‘meaning’ might be used to express. It is a question that should be answered not by appeal to intuitions about cases but by further articulation of what meaningS comes to and how it is related to the rules for the use of words that we learn when we learn how to use them.

There is a question about what ordinary uses of ‘meaning’ express when we have in mind linguistic meaning. My view is that in ordinary use ‘meaning’ expresses a type of status function. The determinate status function of that type that is the meaning of a particular word determines its rule of reference. If this is right, ‘meaning’, as ordinarily understood, does not express not an historical notion. However, the philosophically important question is not what ordinary uses of ‘meaning’ express but what tools help us to understand how language works. On that point, I think that we are better served by the status function conception of meaning. This is well-suited to explain the mechanism and content of communication and does not outrun our capacities to learn what is needed for that in the present, when it is needed. But elaboration of that thought must wait for another occasion.


1The objection developed here was originally presented at a workshop on Essentialism and Anti-Essentialism, September 26–29, 2019, at Banff, Canada. A revision of the paper Richard presented there is the one to which this is a response. 

2This allows for negotiation on the fly about what the role is, given a repertory of co-understood possibilities. This accommodates the sorts of observations that Davidson makes in (Davidson 2005). 

3The notion of a status function is introduced in (Searle 1995: 41). See (Ludwig 2017: chs. 6–10) for elaboration. Searle suggests meanings are status functions but he doesn’t explain what this comes to. See (Jankovic 2014a, 2014b) for a development of the idea. 

4There is shorter argument offered for the same conclusion: if people start applying ‘bachelor’ to unmarried women as well as unmarried men (eliding subtleties), then, since ‘they would not treat this as a change of subject, this looks to suggest both that ‘bachelors are male’ is not analytic—its rejection is not a change of meaning—and that therefore … a word with the meaning of ‘bachelor’ can have a different reference from that which the meaning of ‘bachelor’ currently determines’ (Richard 2020: 2). It is unclear, however, what the argument is. What is a change of subject? We were using W to apply to Ys and now we use it to apply to Xs and Ys. Is the fact that Ys are still included in the scope of what we are now talking about show it is not a change of subject? If we use ‘France’ to talk about ‘France’ and then come to use it to talk about Europe, is that also not a change of subject? Is the study of France and the study of Europe a study of the same subject? And why would this have anything to do with sameness or difference of meaning? Why would that suggest ‘Bachelors are male’ is not analytic? Of course we can use words differently than we do. The question of whether a sentence is analytic makes sense only if we fix on a determine set of uses of its component expressions. And if the meaning of ‘bachelor’ determines that it is true only of men, if we preserve the meaning, do we not preserve what it determines? 

5I relativize to a geographical region to make explicit what is implicit in the examples that Richard gives. Otherwise some of the claims made will turn out to be false. For example, polygamous marriages were then and are even today practiced and legal in many countries. Similarly, same-sex marriages did not first appear on the scene in the 21st century. Many communities before the modern period, in the West, and in Native American, African and Asian cultures, recognized same-sex marriages (Eskridge 1993). There are records of same-sex marriages in antiquity in Mesopotamia, and in Ancient Greece; same-sex marriages in Egypt are referred to in the Sifra in the 3rd century BCE; there is evidence of acceptance of same-sex marriages in Roman society. Reports by Spanish explorers and missionaries of pre-Columbian Native American cultures report same-sex marriages. There were same-sex marriages among the Navaho and Zuni before the influence of Christianity changed tribal mores. The Siwa Oasis in Egypt historically accepted male homosexuality and rituals of same-sex marriage as late as 1928, since repressed. Though his thesis is controversial, John Boswell (1995) argued that same-sex marriages were recognized by the Catholic Church up to the Middle Ages. In pre-colonial Africa same-sex marriages between women were recognized as an option across many African populations, especially West Africa, South Africa, East Africa and the Sudan. Similarly same-sex marriages were recognized in many Asian cultures. See Eskridge for further discussion. 

6And why not also a shorter argument based on geographical variation in 1919 (or 2019) in @? People in North America in 1919 believed that a single man could be married to no more than one woman at the same time in North America. People in North Africa in 1919 believed that a single man could be married to more than one woman at the same time in North Africa. The same word ‘married’ is used to characterize the beliefs of people in North America and in North Africa in 1919, and means the same, though rule of reference (apparently) differs; therefore, meaning does not determine rule of reference. The mistake here has the same source the mistake in Richard’s argument. 

7Given this, the argument secures the possibility that across possible worlds there can be a difference in reference rules without a difference of meaning but not within a world. So we might still have a thesis of some interest connecting meaning and rules of reference: Necessarily, for any world w, if x and y are the same in meaning in w, then the reference rule for x and y is the same in w. 

8It would be possible in principle of index rules to some future community, but since we are interested in knowing what it is that we are talking about, we are also interested in attaching rules to expressions whose content is determined by us. This is connected with the remark in note 9. 

9If rule of reference is hostage to future use, even use in the distant future, then no one is ever in a position to say what it is that he or she means or is saying because it depends on facts about the future which no one is in a position to predict. Meaning (and truth conditions) becomes opaque to speakers. We don’t know what we say. We don’t know whether it is true or false. If what we say expresses what we believe and want, we don’t know what we believe or want either; and otherwise we don’t know how to express what we believe or want. 

10Richard considers the objection that his position on reference rules undermines his supervenience principle (S). 

(S) Meaning at a time supervenes on physical and social facts up to that time: If the histories–that is, the physical and social facts–of two worlds are identical up to and including time t, then at t an expression used in those worlds has the same meaning therein (2020: 19).

He responds with the thesis that meanings are individuated historically like species, and species may lose features they once had but be the same. However, surely it is intuitively as plausible that rule of reference supervenes on history as that meaning does, which is the support that (S) is provided in the paper. I take up the argument Richard offers in favor of the future use being relevant to reference rule below.

11Or perhaps it is a genus/species relation. I won’t distinguish these here since it won’t matter for the argument. 

12Of course, it could fall below the level of definiteness of his intentions, but often what people have in mind is brought out by asking them further questions. Since questions about sameness and difference of what people mean in using ‘married’ presupposes they mean something determinate, I leave aside this further possible complication. 

13Either (a) or (b) will underwrite moving from people in NA in 1919 and in 2019 accepting [O] to our accepting [N]. 

14This is a response that Richard included in an earlier version of his paper, and it is worth discussing still because it helps bring out the implausibility of the view that ‘marriage’ is only ever used to express a single concept. 

15Of course, some might have claimed it was conceptually incoherent. Some would do so out of a prescriptive motive, recognizing even if dimly, that it was conceptually possible, but wishing to rule it out and so claiming more than could be properly extracted from the concept. This is a familiar enough maneuver in philosophy as in ordinary life. Others might really think it conceptually impossible. But would that be because they were confused about it or because it was beyond the limits of their ability to easily imagine alternatives? 

16Suppose we insisted on one meaning for ‘marriage.’ What would it be? It would be the determinable or a system relative notion. For everyone in 1919 and in 2019 understands what is meant by saying that in some countries polygamy (a marriage union among more than two people) is legally sanctioned. In this case, premise 3 is false. 

17Modulo directly referring terms and natural kind terms, at least as the story is usually told. Even here a slight shift of perspective brings natural kind terms back into line. For if we take meaning to be determined by the concept a term expresses, and the concept to determine the property it attributes (as the Fregean tradition assumed), when we index a term like ‘gold’ to a property of the yellowish malleable incorruptible metal around here that explains its observable features, we index the concept it expresses to what we discover. What is a posteriori then is both what property is attributed and what concept is expressed (hence, what it means), and there is no need to say the meaning does not determine the property, and, hence, extension of the term (fixing meaning) across possible worlds. A corollary is that ‘gold is an element with atomic number 79’ expresses an analytic proposition, though we determine empirically that it is expressed by the sentence. The stereotype associated with the word on this way of thinking is our mode of presentation of the meaning, and competence in its use in ordinary life consists in grasp of the stereotype and its role in determining the concept and property. 

Competing Interests

The author has no competing interests to declare.


  1. Boswell, J. 1995. Same-sex unions in pre-modern Europe. New York: Vintage Books. DOI: 

  2. Davidson, D. 2005. A Nice Derangement of Epitaphs. In: Davidson, D (ed.), 89–108. Truth, Language, and History. Oxford: Oxford University Press. DOI: 

  3. Eskridge, WN. 1993. A history of same-sex marriage. Virginia Law Review, 79(7): 1419–1513. DOI: 

  4. Jankovic, M. 2014a. Communication and shared information. Philosophical Studies, 169(3): 489–508. DOI: 

  5. Jankovic, M. 2014b. Conventional meaning. Unpublished thesis (PhD), Indiana University. 

  6. Ludwig, K. 2017. From plural to institutional agency: Collective action 2. Oxford: Oxford University Press. DOI: 

  7. Richard, M. 2019. Meanings as species. Oxford: Oxford University Press. DOI: 

  8. Richard, M. 2020. Is reference essential to meaning? Metaphysics, 3(1): 68–80. 

  9. Searle, J. 1995. The construction of social reality. New York: Free Press. 

  10. Steindorff, G. 1904. Durch die libysche wuste zur amonoase. Liepzig: Velohgen and Klasing. DOI: