I. METALINGUISTIC NEGOTIATIONS

In his paper “Cheap Contextualism,” Peter Ludlow reports hearing a lively debate on sports radio.Footnote ¹⁶ The debate concerned a recent issue of the magazine Sports Illustrated that contained a list of the fifty greatest athletes of the twentieth century. The horse Secretariat was on the list. Some of the people who called into the radio show were really enthusiastic about this choice. Others disagreed strongly with it. Simplifying a bit, we can imagine the following dialogue occurring as part of the sports radio show:

1. (1)

   1. (a) Secretariat was not one of the fifty greatest athletes of the twentieth century. This Sports Illustrated ranking is dead wrong about this.

   2. (b) No way. Secretariat really was one of the fifty greatest athletes of the twentieth century. Sports Illustrated was correct about this.

We can of course imagine a version of this dialogue in which the following is true: at least one of the speakers does not know much about the racehorse Secretariat, as I imagine will be the case with some philosophers reading this paper. (If you do not know much about Secretariat, the main fact to have in mind here is that Secretariat is generally regarded to be one of the best racehorses of all time.) But it is also easy to imagine that most of the people calling into a sports radio show to debate this topic will be sports enthusiasts and will know a lot about Secretariat. So it is easy to imagine that both speakers in (1) know about many (or perhaps all) of the empirical facts that would intuitively be relevant here, including facts how fast Secretariat was, how many races Secretariat won, how dominant Secretariat was with respect to other horses that he raced against, and so on. Still, it is easy to imagine them arguing with each other at length despite both being aware of these empirical facts (as well as mutual awareness of the fact that they are both so aware). Moreover, it will also strike many as intuitive that the speakers disagree with each other and that they are expressing that disagreement in their linguistic exchange.

Now suppose that in the course of their argument, the speakers engage in the following dialogue.

1. (2)

   1. (a) You are missing the point here. Secretariat wasn't even an athlete at all, let alone one of the fifty best ones of the twentieth century.

   2. (b) No. You are wrong. Secretariat was obviously an athlete. In fact, he was one of the fifty best ones of the twentieth century!

Following this, the speakers then go on arguing. In so doing, the following becomes clear: the speaker of (1a) and (2a) never uses the term “athlete” to refer to a nonhuman animal. In contrast, the speaker of (1b) and (2b) calls many nonhuman animals “athletes.” Moreover, this happens even when each becomes aware that the other speaker is using the term in a divergent way.

What is going on in this dialogue about Secretariat? To regiment my discussion of the range of options here, let me pause to introduce some terminology that Sundell and I put forward in other work.Footnote ¹⁷ Let us use the term dispute to refer to a linguistic exchange in which two or more speakers appear to disagree. In turn, we can say that when two people disagree with each other, there is a rational incompatibility between their mental states, such that one person accepts content that is incompatible with content accepted by the other person. The most straightforward way this might happen is this: one person believes a proposition p, the other person believes a proposition q, and q entails not-p. However, for my purposes here, we can be open to there being any number of ways in which people might accept contents (in the relevant senses of “accept” and “content” here) to yield disagreement between them and another person (or even with their past or former self); for example, perhaps there are disagreements that involve a certain kind of clash of noncognitive attitudes such as desires or intentions.

On this way of carving up the terrain, a dispute is a kind of activity that might or might not actually express a disagreement. A disagreement, in contrast, is not a kind of activity at all. Rather, on our definition of it, when two people disagree with each other, that is a state that they are in rather than an activity that they engage in. With that in mind, here are two questions that we can then ask about any given dispute: (1) Does the dispute actually express a disagreement? and (2) If so, how?

We can then draw a distinction between two different classes of dispute. Some disputes are what Sundell and I dub canonical ones. These are disputes that center on information that is expressed by the semantics of the statements involved, that is (roughly), the linguistically encoded content of those statements. If we were to try to model the Secretariat case as a canonical one, we should then say this: the speakers each mean the same things by the words that they use, including, crucially, the word “athlete.” If we think of the meaning of words (in the sense of “meaning” that is explicitly about semantics) in terms of the concepts they express, this amounts to the following idea: there is a common, shared concept (call it athlete) that both of them express by the word “athlete” in this dispute. The reason we should want to say this is as follows. First, consider that, crucially, it seems that speakers in the Secretariat case do not just appear to disagree (as they do on any dispute, given the definition of “dispute” that I am working with). Rather, it is also compelling to think that we as theorists should want to vindicate this appearance, such that we end up saying that the speakers really do disagree. Now consider that if the speakers meant different things by “athlete,” then it seems likely that they would be expressing propositions (or contents of some other sort) that were mutually compatible, in terms of the semantics of what they have each said. But if that is true, then we cannot explain the disagreement in terms of the semantics of what each of them has said. So we should hold that speakers mean the same thing by “athlete.”

The problem with this approach in the Secretariat case is that it seems false to say that speakers mean the same thing by “athlete.” There seems to be compelling evidence that they mean different things by the term. The reason is simple. One important piece of evidence for what a speaker means by a term is how she is disposed to apply that term in nondefective conditions. This is evidence that a wide range of linguists and philosophers of language should grant is important. In the Secretariat case, the speakers diverge significantly in their dispositions to apply the term—moreover, in conditions that certainly seem to be nondefective (in the relevant sense of “nondefective” that matters here for our purposes of linguistic analysis, even if not for the purposes of doing epistemology). So it seems that we have good evidence that the speakers mean different things by their terms. Such evidence, of course, might be outweighed, defeated, or undercut by other evidence. I talk more about this in Section II. For now, what is important is this: at least prima facie, we have strong evidence that the speakers in the Secretariat case mean different things by “athlete.”Footnote ¹⁸

If we remain committed to the view that the Secretariat case is a canonical dispute, then this conclusion would make it hard (if not impossible) to vindicate the idea that the speakers really do disagree with each other. This is for the reason that I gloss above. But, importantly, not all disputes are canonical. Some disputes, rather than being centered on information communicated via semantics, are instead centered on information communicated via pragmatics, that is, information communicated in virtue of features of how we use words or sentences in a given context rather than in virtue of the linguistically encoded content of those words or sentences. Such disputes are what Sundell and I call noncanonical ones. There is a variety of different kinds of noncanonical disputes. Some involve information communicated by implicature, others involve information communicated by presupposition, and others still involve information communicated by connotation.

Roughly, the thought is this. First, take whatever pragmatic mechanisms you think we have for communicating information in general (something that different philosophers of language have different views on). Then, suppose that one or more speakers in the dispute is using one (or more) such mechanism to communicate information in a linguistic exchange that at least appears to be the subject of disagreement. You then have a noncanonical dispute.Footnote ¹⁹

To illustrate, consider a case involving implicature.

1. (3)

   1. (a) I missed the party last night. I am curious how many people were there. Can one of you two tell me?

   2. (b) Sure. I can. There were twenty-five people at the party last night.

   3. (c) No there were not. You are wrong. There were forty people there.

According to the familiar Gricean story about number terms, the speaker of (3b) literally says that there were at least twenty-five people at the party, whereas the speaker of (3c) literally says that there were at least forty people at the party. Those are consistent propositions. What is not consistent is what each implicates by making those claims. In such a context, given the expectations that the speaker of (3a) brings to the table—expectations, roughly, that the other speakers will provide information that is maximally informative and relevant, relative to her interests in this subject matter—the speakers of (3b) and (3c) communicate inconsistent propositions.Footnote ²⁰

What Sundell and I propose is that the Secretariat case is a noncanonical dispute. More specifically, we propose that it is a particular kind of noncanonical dispute that we call a metalinguistic negotiation (or, equivalently, a normative metalinguistic dispute).Footnote ²¹ In a metalinguistic negotiation, at least one speaker uses (rather than mentions) a given term in order to advocate a normative view about what the term should mean, at least for the given context at hand.

One way to get a handle on metalinguistic negotiations (or at least an important subset of them) is as follows. Sometimes, when we use words, we hold fixed facts about what our words mean and then use those words to communicate our views about some part of reality. But we sometimes do the reverse. Sometimes we hold our views about some part of reality fixed, and then use our words to communicate not about that part of reality (or at least not directly) but rather about what those words do or should mean. In such a case, then, speakers use (rather than mention) a word to communicate something about the word itself—for example, what it does mean (a descriptive issue) or what it should mean (a normative issue). Such usages are what Sundell and I, drawing on work from Chris Barker, call metalinguistic usages of a term.Footnote ²² Speakers can employ a metalinguistic usage of a term in the context of a dispute, and they can express a disagreement by a metalinguistic usage of a term.Footnote ²³ A descriptive metalinguistic dispute is one in which at least one speaker employs a metalinguistic usage of a term to put forward a view about what a term does mean. And a metalinguistic negotiation (or, equivalently, a normative metalinguistic dispute) is a dispute in which at least one speaker employs a metalinguistic usage of a term to put forward a view about what a term should mean.

Here is what this means for the Secretariat case. In that case, Sundell and I claim, the speakers of (1a) and (1b) both speak truly, in terms of the semantics of what each of them says, given what each of them in fact means by the relevant term “athlete.” (The same goes for when one says (2a) and the other says (2b).) But, in addition to communicating information through the semantics of what they say, the speakers each communicate information through the pragmatics of what they say. In particular, each is advocating for their preferred meaning of the term “athlete” and doing so by metalinguistic usage of the term “athlete.” Each of these speakers, we claim, means something distinct by the term “athlete,” such that each expresses a different concept by that term. So, put in terms of concepts, we can say the following: each speaker is advocating for using one of a range of competing concepts in the context at hand, a concept that each aims to have paired with the term “athlete” in this context.

I am putting things in terms of concepts, so it is worth pausing to say something briefly here about what I take concepts to be and what this means for Sundell's and my proposal about the Secretariat case. The first thing to emphasize is that the idea of a metalinguistic negotiation does not depend on a specific, fully developed view of concepts. Neither do the more general views about thought and talk that I gloss above and which I draw on throughout this paper. Rather, they are compatible with a wide range of leading theories from the philosophy of mind and philosophy of cognitive science of what concepts are.Footnote ²⁴ That being said, it is worth adding a bit more about the basic view of concepts that I favor, at least for the purposes of this paper.

In rough terms, I take concepts to be the equivalent in mental representation to what words are in linguistic representation. In turn, we can then think of different concepts as being marked by the specific cognitive significance that they have for an agent or the specific cognitive role that they play. In many cases (including the concepts that the speakers express by “athlete” in the Secretariat case) the different concepts that each speaker is using play a kind of representational role, such that they are each categorizing things in the world in a different way. These different ways of representing the world partition the world into ways that are in accordance with that representation and those that are not. If one wanted, one might then model that way of representing as something along the following lines: a function from possible worlds to extensions.Footnote ²⁵ This function is a way of determining the extension of the concept in a given possible world. In turn, as mentioned above, I take the meaning of a given word (in the sense having to do with “semantics”) to be given by the concept that it expresses.

Why bother arguing about which concept should be paired with the term “athlete” in the Secretariat case? There are two main things to say here. The first is that not all concepts are equally good for using in a given context, relative to our purposes in that context. For example, when engaged in mathematical inquiry, it will be good to have (and sometimes use) the concepts set and number. Compare that to a concept that picks out the union of numbers, elephants, and tennis racquets. Such a concept is less helpful in such a context, given our aims in engaging in mathematical inquiry.Footnote ²⁶ Second, consider that terms often carry with them significance in a given social-historical setting that goes well beyond what is built into the semantics of a term. For example, quite aside from what is built into the semantics of a term like “athlete,” that term has a rich set of associations tied to it. The term is also closely connected to certain kinds of functional roles in a given context. For example, in the context of a list of the fifty greatest athletes of the twentieth century published in Sports Illustrated (a very widely read magazine), consider some of the things that are likely to go along with calling a human (or nonhuman animal) an “athlete.” These include such things as certain forms of esteem, praise, and social recognition. In light of this, speakers might have very good reason to have normative views about not only which concept should be deployed in a given context but also which word should be used to express that concept in the context at hand.Footnote ²⁷

Of course, speakers might not be—and, indeed, I think often are not—aware that they are engaged in a metalinguistic negotiation, even though they in fact are.Footnote ²⁸ So my point here is not that speakers engaged in this metalinguistic negotiation are consciously aware of the fact that they are engaged in it or that they would quickly affirm that description of their linguistic exchange. Rather, my point is that the basic topic I am claiming is at issue in the Secretariat dispute—a topic about which concept should be paired with a relevant term—can be a substantive one well worth arguing about. And if this topic can be worth arguing about explicitly (via a canonical dispute in conceptual ethics), then so, too, can it be worth arguing about via metalinguistic negotiation. The mode of communication about the subject matter does not change the substantiveness and import of the subject matter itself. In short, there are lots of issues in conceptual ethics that matter, and they matter independently of anything having to do with the way in which we communicate about them.Footnote ²⁹

In making our metalinguistic proposal about the Secretariat case, Sundell and I are following the basic suggestion made by Ludlow himself in his original discussion of the case. Ludlow puts things in different terminology—terminology that is tied to a broader view of language that he advocates for in “Cheap Contextualism,” and which is extended in his recent book Living Words.Footnote ³⁰ Ludlow's core thought is that in the Secretariat case, there is no antecedently settled meaning of the term “athlete” as used by all of the speakers on the sports radio. Rather, the speakers advocate for different ways of fleshing out that meaning. Moreover, Ludlow thinks that many of our words have meanings that are massively underdetermined and that we flesh out their meanings via negotiation in particular contexts relative to the particular purposes at hand. Importantly, one need not accept a radical picture of language in general (e.g., that meaning is massively undetermined in general) to grant Ludlow's basic reading of the Secretariat case: a reading on which the speakers are advocating for different meanings of the term “athlete” for the context at hand. Sundell and I think Ludlow's view here is on the right track, but we develop it in a different way—a way that is more schematic and which, we think, is harder to resist.

II. THE DISPUTE OVER LEGAL POSITIVISM

Here is a basic thought about the relationship between communication in everyday life and communication in philosophy: if there are pragmatic mechanisms that we use to communicate information in everyday life, the chances are that we also sometimes use those very same pragmatic mechanisms to communicate information in a philosophical context. I think this thought is on the right track. Consider the following. Suppose that we start with a tool for communicating information and that this tool can potentially be used to communicate about a wide range of topics in a wide range of different contexts. Then we start communicating about some specialized topic in some particular context. Suppose that in theory we could use this tool to help us communicate about this topic in this context. If so, would we not then expect that we would at least sometimes use this tool in this context, at least if we were being rational? After all, the tool is already there, at our disposal, and ready to be used for communicating about this topic in this context, just as it can be used for communicating about any number of other topics in a wide range of other contexts.

Based on this, I think we should start with the following idea about the pragmatics of philosophical communication—that is, communication among philosophers about philosophical topics. We should expect that the pragmatic tools that we use in everyday conversation (e.g., implicature, connotation, presupposition, metalinguistic usage, etc.) will also be ones that we should expect that we also sometimes use when talking about philosophy. (The same points holds true for first-order legal discourse as well as any subarea of discourse in general, absent special reason to think otherwise).

If this thought is on the right track, it gives us reason to think that philosophers, when talking about issues in philosophy, might well sometimes use terms in a metalinguistic way, for example, in a way that parallels Jill's use of the term “cold” at the Antarctic research station or the use of the term “athlete” in the Secretariat case. Of course, there might be reasons that the philosophical context turns out to be special, such that this available pragmatic tool just is not used in the philosophical context at all. But we have at least prima facie reason to think otherwise. If so, this should then make us at least open-minded about the possibility that there is an important metalinguistic element to the dispute over legal positivism. Indeed, it gives us reason to take such a proposal seriously and to investigate whether it is true. With that in mind, let us then turn to the dispute over legal positivism; that is, the dispute that pretheoretically we can gloss as one about legal positivism versus legal antipositivism.

In broad terms, here is the question on the table: Should we think of the dispute over legal positivism as a metalinguistic negotiation? In order to zero in on the core issues that matter here, we should make this question a bit more precise. Consider again the dispute about Secretariat on sports radio, as well as the dispute between Jill and Anand at the Antarctic research station. These were disputes that occurred over a relatively short period of time and involved only two people. In contrast, the dispute over legal positivism is something that has occurred (in one form or another) throughout the history of modern legal philosophy and has involved many different philosophers. It is also something that has not just been discussed in conversation but has also been written about at great length by many different people. Such a dispute would be likely to take different forms at different times. Based on this, I think it would be quite surprising if the dispute were always a purely canonical one—or, even more specifically, if it were always a canonical dispute about the exact same topic. For the same reason, I think it would also be quite surprising if it were a purely noncanonical one—or, even more specifically, if it were a purely metalinguistic one.

This, I think, provides good reason to think that the real question that matters to us here is something more nuanced than the question I introduce above (namely, the question of whether we should think of the dispute over legal positivism as a metalinguistic negotiation). I think the relevant question here is more carefully stated as follows: How much of the dispute over legal positivism has been (and is) a metalinguistic negotiation?Footnote ³⁵

The claim that I want to put forward is this: a significant part of the dispute over legal positivism is a metalinguistic negotiation. How much of it? Enough that it can help us illuminate important (and initially confusing) aspects of what is going on in this dispute. Or at least so I suggest in what follows. Put more carefully, what I argue can be stated as follows: the same kind of evidence that supports thinking that the Secretariat case is a metalinguistic negotiation also supports thinking that an important part of the ongoing dispute over legal positivism is a metalinguistic negotiation—a part that might not make up the majority of the dispute (or even anything close to it) but which is nonetheless explanatorily important for our self-understanding of this dispute within legal philosophy.

In order to build this case, what I propose to do is straightforward: to consider the kind of evidence that was relevant in the Secretariat case and then see if the same kind of evidence is present in the dispute over legal positivism. I start by putting the kind of evidence we are looking for in schematic form, and explaining why, in general, this sort of evidence supports thinking that a given linguistic exchange is a metalinguistic negotiation. I then turn to looking at the dispute over legal positivism.

Here is a schematic summary of the kind of evidence that Sundell and I draw on in arguing for reading the Secretariat case as a metalinguistic negotiation.Footnote ³⁶

1. (A) There is good evidence that the linguistic exchange is a dispute; that is, there is good evidence that it is a linguistic exchange that appears to express a disagreement.

2. (B) There is good evidence that the dispute really does express a disagreement.

3. (C) There is good evidence that speakers in the dispute mean different things by (at least) one of the terms in that dispute. There are different things that might provide such evidence for a given term. (What one takes to provide such evidence will, of course, depend on one's more general commitments in the philosophy of language). I claim that one such important piece of evidence is this: the speakers in the dispute have dispositions to apply systematically the same term in divergent ways in the same (nondefective) conditions. And I claim that many different philosophers with a range of different commitments should take that as good evidence, even if there are many other sources of evidence as well.

4. (D) There is good evidence that the disagreement expressed in the dispute, insofar as there is one, is not just about descriptive information about what a word means or how it is used. (This is something that helps provide evidence that it is a normative metalinguistic dispute, i.e., a metalinguistic negotiation, and not a descriptive metalinguistic dispute.) For example, one good piece of evidence here would be that speakers persist in their dispute even when they agree on the facts about a term's current meaning or current use.

It should be emphasized that these conditions are presented in a highly schematic way. What it would take to show that each condition holds would, of course, depend on many other issues. Critically, this includes figuring out what, in general, gives us good evidence for variation between speakers in what they mean by a given term.

This last point is a crucial one to emphasize. For an important piece of the argument for reading a case as a metalinguistic negotiation, at least one along the lines of the Secretariat case, is that there is compelling evidence that speakers in a dispute mean different things by their terms. But of course, different philosophers will have different views here about what counts as good evidence of speakers meaning different things by their terms. This is tied to different foundational views about the grounds of meaning.Footnote ³⁷ For example, perhaps external facts about a speaker's environment or the patterns of usage by experts matter here as well, independently of anything having to do with facts about the speaker's own psychology (as is true according to certain kinds of well-known externalist theories of meaning).Footnote ³⁸ Or perhaps, as proponents of Lewis-inspired reference magnetism urge us, facts about genuine objective similarity, or facts about what is genuinely explanatory important, matter for settling the meaning of our terms, and do so independently of anything about the speaker's own psychology.Footnote ³⁹ Or perhaps, as Dworkin urges, facts about the content of at least certain concepts (and thus the meaning of terms that express those concepts) are settled by normative facts about what best justifies a set of practices, such that moral facts matter for settling the meaning of terms, and do so independently of anything about the speaker's own psychology.Footnote ⁴⁰

In light of this, one might be tempted to think the following: if one of those views is right, then that changes what we should be looking for in condition (C). It might be harder to find good evidence that speakers in the dispute mean different things by (at least) one of the terms in that dispute. Perhaps the evidence that Sundell and I cite in the Secretariat case—facts about the dispositions of speakers to apply a term one way or another—is strong evidence only on certain theories of meaning, for example, a kind of internalism. On many other theories of meaning, perhaps the evidence we cite is only a very weak kind of evidence.

This charge is onto something important. But the dialectic here is more subtle than it might appear at first blush. There are two things that are crucial to keep in mind. First, as Sundell and I argue elsewhere, everyone needs to hold that there are some metalinguistic disputes.Footnote ⁴¹ The question is how far the phenomenon extends and what sort of explanatory work it can do. But it is a tool that everyone needs in their toolkit anyway, to explain at least some cases. Second, as Sundell and I emphasize, all plausible theories of meaning—including the kinds of views I sketch above—grant that facts about patterns of usage or dispositions matter to some degree for settling facts of meaning. We then need to ask how much they matter and what other facts matter as well. The answers to those questions obviously interact with the question of which foundational theory of meaning is correct. However, the answer to that question also needs to be informed by the question of how plausible it is to suppose that speakers differ in what they mean, including, for example, whether someone who posits significant variation of the meaning of terms between speakers has a good story to tell about disagreement (and, relatedly, about agreement). The question of how prevalent metalinguistic disputes are (as well as, more broadly, how prevalent noncanonical disputes in general are) matters for thinking about that issue—and thus for thinking about which kind of foundational semantic theory is correct.

With that in mind, we might put the point like this: thinking about the viability of reading a given dispute as a metalinguistic one is part of the process of figuring out our best general theory of meaning. To appreciate this, consider that one prevalent kind of argument in favor of rival semantic views is a disagreement-based one, on which a given view (e.g., Dworkinian interpretivism, expressivism, speaker-relativism, etc.) is supported by its (purported) ability smoothly to capture (purported) data about disagreement—or at least do so more smoothly than its rivals. Such a disagreement-based argument is also sometimes given not for a particular view of concepts or semantics but for what might best be thought of as a broad schema of a view; for example, that speakers share a given concept but have different “conceptions.” Without further detail, on such a view the idea of the concept/conception distinction is really more of a schema of a kind of view than a developed view of the foundation of meaning: it is a schema that says speakers share a concept but differ in something else (i.e., whatever it turns out a “conception” ultimately is).

The broad thesis at the core of this “concept/conception” schema is one that a lot of disagreement-based arguments aim at establishing: the thesis that speakers mean the same things by their words in the disputes that express the disagreements in question, such that at bottom they really do share the same concepts (that they express with the relevant words in these disputes). One crucial component of assessing such disagreement-based arguments for sameness of meaning is to figure out whether the disputes are ultimately best understood in metalinguistic terms. In short, one's views about the grounds of meaning both inform and are informed by one's views about the prevalence of metalinguistic disputes.

Disagreement-based arguments often assume, often without argument, that the best way to capture how disagreements are expressed in the relevant class of cases is to see such disputes as canonical ones (where there is then pressure to maintain sameness of meaning between speakers in order to uphold the idea that there really is a disagreement that is expressed). The boldest versions of such disagreement-based arguments for sameness of meaning are ones that appeal to something like the following principle: the only possible way to avoid concluding that the speakers are “merely talking past each other” (and thus not disagreeing) is to posit that speakers mean the same things by their words in a given dispute (and thus that they express the same concepts by those words). The metalinguistic disputes that Sundell and I discuss, in which speakers mean different things by their words but still genuinely disagree, are counterexamples to this principle. Indeed, to provide a counterexample to the above principle, Sundell and I do not even need to be right in our diagnoses about the particular cases; we just need to be right that such disputes are metaphysically possible.

More careful disagreement-based-argument disputes work by appeal to a kind of inference to best explanation. Such arguments claim that positing that the relevant disputes are canonical ones, where speakers mean the same things by their words, is the best explanation of how disagreement is expressed in these disputes. To run this second kind of disagreement-based argument for a given class of disputes, one needs to show that such a view is better than our metalinguistic reading of those disputes. We argue that for many disputes where disagreement-based arguments are run, our kind of metalinguistic reading is in fact the better one. In short, Sundell and I claim that seeing disputes such as the Secretariat one as metalinguistic (rather than canonical) both better illuminates what is going on with them and fits more smoothly with a good general theory of meaning. And, crucially, our theory can do this without positing that speakers involved in these disputes mean the same things by their words (which, for example, is what the concept/conception model claims is so).Footnote ⁴²

Moreover, we can note the following. Inference-to-best-explanation versions of disagreement-based arguments are more philosophically promising than versions that appeal to the (demonstrably false) principle that in order for to two speakers to express a disagreement in a given dispute, sameness of meaning is required. But even these arguments turn out to be in bad shape once metalinguistic disputes are squarely in view. We can put this point as follows. If a philosopher wants to show that speakers mean the same things by their words in a given dispute, then she must provide an argument that this is so. We have just seen that merely citing the presence of genuine disagreement does not show this. So she must provide an independent argument for this conclusion, which, typically, will involve arguing that the dispute is really a canonical one. However, by showing this, she will have done the core work she needs to do in order to show that speakers mean the same things by their words. For in order to argue that a dispute is a canonical one that really does express a disagreement (which, remember, is what the theorist running the disagreement-based argument wants to show), the theorist needs to maintain (at least typically) that speakers mean the same things by their words.

The upshot is this. Some disagreement-based arguments are unsound (these are the ones that rely on the overly strong principle about sameness of meaning underwriting the possibility of genuine disagreement being expressed in a dispute). And the ones that revolve around an inference-to-best-explanation claim end up being either unsupported or unnecessary.Footnote ⁴³ Thus facts about the presence of disagreement here just will not do much dialectical work in this context. The relevant question we then need to ask is simply this: Should we regard the relevant disputes as canonical ones or as noncanonical ones (e.g., as metalinguistic negotiations)? Sundell and I argue that (1) holding that condition (C) holds in a wide range of cases (e.g., wider than certain externalist rivals hold), based on the kind of evidence about speakers’ dispositions that we emphasize in the Secretariat case, and then (2) viewing disputes where (A) to (D) hold as metalinguistic negotiations is the best overall package deal here. To rebut that claim, one needs to propose an alternative package deal and show that it does better. Simply citing the (purported) presence of genuine disagreement in these disputes is going to get you nowhere.

It is beyond the scope of this paper to go into great detail to defend Sundell's and my approach in relation to all of the salient alternatives here. But it is worth saying a few things here to frame the discussion and the assessment of the view that I am putting forward about the dispute over legal positivism.

One important advantage of our view is that it allows us to vindicate intuitions of disagreement in a straightforwardly descriptivist framework—on which the semantics of the relevant class of statements is explained at the most explanatorily basic level in terms of beliefs that speakers express about how things are. Our approach thus fits smoothly with the default view in much of philosophy of language and linguistics (thus avoiding the range of difficulties faced by noncognitivist approaches, e.g., the Frege-Geach problem about nonassertoric uses of the relevant terminology).Footnote ⁴⁴ At the same time, our approach does not end up positing “talking past” each other in cases where we reflectively think there really is a disagreement being expressed. It is hard to get both of those things at once for many cases (including those such as the Secretariat case). And so the fact that our proposal does so is a mark in its favor.

Another important advantage of our view is that it allows us to explain how speakers express disagreement in these cases without invoking a (purportedly) special type of concept, as on the Dworkinian interpretivist approach. In rough terms, Dworkin argues that we need to introduce what he calls “interpretative” concepts in order to explain an important class of disputes: disputes where (1) speakers agree on all of the relevant nonlinguistic facts, (2) they apply the same term in divergent ways, (3) they are aware of this divergence in use, and (4) we as theorists (and often the participants themselves) take those speakers to disagree with each other. Dworkin introduces the idea of interpretative concepts in order to explain how genuine disagreements are possible in such cases—which, he claims, cannot be done if one posits that speakers simply employ the other kinds of concepts Dworkin posits in his overall theory (namely, what he calls “criterial” concepts and also what he calls “natural-kind” concepts).Footnote ⁴⁵ Dworkin argues that an interpretative concept is one whose correct application does not depend on fixed criteria or an instance-identifying decision procedure but rather on the moral facts that best justify the total set of practices in which that concept is used.Footnote ⁴⁶

Sundell's and my metalinguistic approach to the disputes that Dworkin is concerned with—and which, it should be noted, he claims include the dispute over positivism that I am discussing here—does not require positing any such concepts. Instead, we can stick with our best general, independently motivated view of the nature and content of concepts. Our view thus has the advantage of theoretical parsimony here. We make use of tools that everyone (including Dworkin) needs in their toolkit anyway, including, centrally, noninterpretative concepts (which Dworkin thinks we also need to posit) and the device of metalinguistic usage of a term (which everyone, including Dworkin, needs to posit in order to explain some cases). In contrast, Dworkin introduces a new special type of concept—and one that moreover remains underdescribed and mysterious in many key respects, especially relative to the concerns about concepts that motivate work on them in the philosophy of mind.Footnote ⁴⁷

A third advantage of our view—relative to, say, assessor-relative approaches to the grounds of meaning, such as Andy Egan's and John MacFarlane's—is that we can stick with a more straightforward approach to content and a more straightforward approach to truth.Footnote ⁴⁸

Finally, a fourth important advantage of our view is that it allows one to hew closely to patterns of usage and dispositions when assigning meaning to speakers—an approach that fits well with many of the leading methodological approaches in linguistics (certainly much more so than Dworkinian views do). In general, fit with an explanatorily powerful empirical research program is an important piece of evidence in favor of a view.

With all of this in mind, let us now return to the kinds of evidence outlined in (A) to (D) which I sketch above. If what I say above is on the right track, then, in cases where (A) to (D) hold, there is good reason to think that dispute might well be a metalinguistic negotiation. Perhaps such a reading will ultimately turn out to be wrong. Perhaps, all things considered, a rival reading of the case (e.g., an expressivist reading or a Dworkinian interpretivist one) will turn out to be the best one. But it is at least a proposal worth taking seriously for such cases, and indeed, there is much to be said in its favor.

Before moving on, it is worth emphasizing here that a dispute might be a metalinguistic negotiation even if (A) to (D) do not all hold. Consider the following. In a series of recent papers, Amie Thomasson draws on Sundell's and my work on metalinguistic negotiation in the context of defending her broadly deflationary views in meta-metaphysics.Footnote ⁴⁹ In the course of doing so, Thomasson defends the idea that there can be metalinguistic negotiations in which speakers engaged in a dispute mean the same thing by their words. She focuses on cases where one of the speakers says something false in order to push the usage of the relevant term toward a meaning on which the speaker's statement would be true.

Thomasson's views here will be particularly attractive to those drawn to more externalist ways of thinking about semantics. Nothing Sundell and I say rules out the possibility of such cases. Sundell and I focus on cases where we claim that speakers mean different things by their words in what we take to be a crucial sense of “meaning” relevant to judgments about the truth and falsity of what speakers say (at least in many theoretical contexts where we are interested in assessing such judgments). Our claim about meaning variation in such cases is tied to our more general views in the philosophy of language, which not everyone interested in the idea of metalinguistic negotiation will share. In this paper, I develop my proposal about the dispute over legal positivism based on the idea that this dispute bears important similarities to Ludlow's Secretariat case. Because Sundell and I hold that speakers mean different things by the word “athlete” in that case, and because my work here draws extensively on the overall framework about thought and talk that Sundell and I have been developing, I treat the presence of variation of meaning (satisfying condition (C)) as an important piece of evidence that a linguistic exchange is a metalinguistic negotiation.

But, for the reasons Thomasson's work brings out, it would be a mistake to think of (C) as a condition that must be satisfied in order for something to be a metalinguistic negotiation.Footnote ⁵⁰ More generally, it would be a mistake to think that (A) to (D) give us anything like necessary and sufficient conditions for an exchange being a metalinguistic negotiation. Rather, my idea is that when conditions (A) to (D) are satisfied, this gives us important evidence that an exchange is a metalinguistic negotiation.

With these qualifications about conditions (A) to (D) in mind, let us now turn to the dispute over legal positivism. Let us start with (A). In this case, there seems to be ample evidence that the linguistic exchange is a dispute. (And thus that (A) is met.) After all, many participants in the relevant linguistic exchanges here—as well as many observers of them—take it that positivists and antipositivists express a disagreement when they argue with each other. And that is all that is needed for something to be a dispute, on the stipulative definition of “dispute” I am working with here. Does the dispute really express a disagreement? Perhaps it does not. Some disputes are ones where speakers think they are expressing a disagreement, but it turns out that they are not doing so at all. For example, that is so in some cases where one speaker uses the term “bank” to refer to the side of a river, and another does so to refer to a financial institution, but each speaker thinks the other is using it in the same way she is. In that case, discovering the difference in usage might dissolve the initial appearance of disagreement.

I think that parts of the dispute over positivism are like that kind of case of “merely talking past each other.” But it is also hard to accept the conclusion that this is all that is going on in the dispute over positivism and that there is no disagreement that is actually being expressed here. It is hard to imagine that there is nothing that positivists and antipositivists disagree about—and that they manage to communicate about—in their dispute. I could say more on this front to motivate that thought. However, for now, let us leave it at that. For the core of what makes the metalinguistic proposal an interesting one is that it shows us a way to avoid the conclusion of “merely talking past each other,” while still granting a variance in what speakers mean by the words they use in the dispute.

What about condition (C)? Is there is good evidence that speakers in the dispute mean different things by (at least) one of the terms in that dispute? For our purposes at hand, the relevant term to focus on here is “law” (or, relatedly, “the law”). Here is a reason for thinking that speakers mean different things by this term, at least some of the time. Suppose one is a convinced positivist. Then one is going to make claims about what kinds of facts are among the grounds of law and also what kinds of facts are not. Those are claims that an antipositivist denies. We know there will be some claims using the term “law” that the positivist is disposed to make which the antipositivist is not. So that is some evidence that they mean different things by their terms. But it also might be pretty weak evidence, for, other things being equal, it seems implausible to hold that every time one philosopher says something of the form “X is grounded in Y” and the other says “X is not grounded in Y” then they mean different things by the term X. After all, it is surely possible for two speakers to mean the same thing by “X” but have different theories about X (including different theories about the grounds of the X facts).

But we can also appeal to more than this evidence in the case at hand. First, in the case of the dispute over legal positivism—in parallel to the Secretariat case—speakers persist in using the term “law” in different ways after being presented with all sorts of arguments for and against positivism. There is a lot of evidence and arguments being offered by both sides for what should count as “law”—just as in the Secretariat case with “athlete.” And just as in the Secretariat case, where both sides of the dispute know the relevant empirical facts about how many races Secretariat won, how fast he was, and so on, many of the speakers in the positivist dispute are well aware of a range of the relevant empirical facts. Moreover, many of them—despite being on opposite sides of this debate in philosophy of law—will agree on a range of moral and political issues as well. Despite all of this, some philosophers seem totally convinced that legal positivism is true and others seem totally convinced that legal antipositivism is true. (Just as one speaker is totally convinced that Secretariat is an athlete and the other is totally convinced that he is not).

Second, it is not just that speakers on one side of the dispute make claims about the grounds of law that the other side does not. Positivists also present purported platitudes about law that (at least some) antipositivists deny, as well as vice versa. Moreover, they often do this knowing full well that others will deny those purported platitudes. This is important data to have on the table when theorizing about what speakers mean by their words.Footnote ⁵¹ Consider, for example, Scott Shapiro's claim (which he rightly notes not all antipositivists will agree with) that “it seems to me an obvious truth about the nature of legality that regimes that are morally illegitimate may still have law. I say this because it is easy for me to imagine legal systems that are evil.”Footnote ⁵² The same thing happens in reverse. For example, antipositivists present ways in which law figures into normative topics such as political legitimacy or moral worth in a way that many positivists would deny is so.

Consider, for example, the following statement from Lon Fuller: “Law, as something deserving loyalty, must represent a human achievement; it cannot be a simple fiat of power or a repetitive pattern discernible in the behavior of state officials.”Footnote ⁵³ When there is significant disagreement on platitudes about those things that a given term (e.g., “law”) refers to—or, in a connected vein, when speakers (in nondefective conditions) insist on basic truths about a subject matter that they take to be nonnegotiable in terms of fixing the topic of the subject matter at hand—that is some evidence that speakers actually mean different things by the term in question. So there is some evidence here that speakers do mean different things by the term “law” in the dispute over legal positivism. In particular, for my purposes here, what I want to focus on is this idea: some positivists mean one thing by “law” that is different from what some antipositivists mean by “law.”

Given my aims in this paper, the relevant question about my proposal is not actually whether it is correct, all things considered. For recall that my ambition is to argue for the following claim: the same kind of evidence that supports thinking that the Secretariat case is a metalinguistic negotiation also supports thinking that the ongoing dispute over legal positivism is partly a metalinguistic negotiation. Now ask about condition (C) in light of this. What evidence supports thinking that speakers meant different things by “athlete” in the Secretariat case? Does the same kind of evidence hold here? I argue above that the answer is yes. If one wants to deny this, then we need a reason for thinking that the cases are different in a respect relevant to the semantic facts about what each speaker means by the relevant terms in each dispute.

Here is one proposal. Speakers in the dispute over legal positivism use more specialized philosophical terminology, such as “ground,” “supervenience,” or “essence,” terminology not present in the dispute on sports radio. Perhaps that is a relevant disanalogy? It is not. To see that, we can consider a modified version of the Secretariat case where speakers assert claims using that terminology on sports radio, for example, claims such as “it is part of the essence of being an athlete to be a human agent,” or “facts about being an athlete are partly grounded in facts about performance relative to physical ability.” The addition of such claims does not change our main reasons for thinking that speakers mean different things by “athlete” in the Secretariat case. If anything, it might give us more reason to think they mean different things—for we are now adding new ways in which the speakers diverge in their application of the term “athlete” in nondefective conditions.Footnote ⁵⁴

Perhaps there are features of the case that support a disanalogy here having to do with the context of a philosophical conversation rather than a sports radio debate. That might be so. But it is also hard to see what those would be. One such feature might be this—speakers in a philosophical context often want to converge on an agreed-upon meaning for relevant terms (e.g., “knowledge,” “free will,” or “law”), in a way that perhaps is not the case in debates on sports radio. But even if so, that does not mean that speakers meant the same thing throughout the dispute. It very well might be that speakers meant different things at the start of the dispute and then, perhaps in part by engaging in metalinguistic negotiation, come to agree upon a meaning of the term for the purposes of conversation. Indeed, I think that is a crucial part of many disputes involving metalinguistic negotiation. So this (purported) difference of the philosophical context from the context of a sports radio debate does not seem to support a disanalogy here in terms of thinking about how much speakers vary in what they mean by the relevant terms (e.g., “law” or “athlete”).Footnote ⁵⁵

In light of all this, I conclude the following: if there is good reason to think that the speakers in the Secretariat case mean different things by “athlete,” then so, too, is there good reason to think that speakers in the positivist dispute mean different things by “law.” The same kind of evidence is present in both cases.

Now consider the final condition, condition (D). The question here is this: Is there good evidence that the disagreement expressed in the dispute, insofar as there is one, is not just about descriptive information about what a word means or how it is used? In thinking about this issue, is important to proceed with care. Sometimes, of course, positivists and antipositivists put forward their views as part of canonical disputes about what a group of people in fact mean by the term “law.” Such canonical disputes—disputes that we can think of as a kind of interpretative, metalegal dispute—make up an important part of legal philosophy and, moreover, illuminate an important part of what is going in the dispute over legal positivism.Footnote ⁵⁶ Philosophers (and linguists and others) often engage in canonical disputes about what people in fact mean by their terms—as exemplified, for example, by what I myself am doing in much of this paper.

In many cases, proposals about what speakers mean by their words—especially when they take a descriptivist form rather than an expressivist one—involve facts about the metaphysics of what that word refers to (such as facts about what kind of things these things are, or about their grounds), insofar as that word ends up successfully referring to anything at all. This is certainly not always the case. But it is a common feature of many semantic proposals that follow a descriptivist pattern of explanation (which, for our purposes here, we can also call a “cognitivist” pattern of explanation). Thus, in the metalegal case, we should expect that for some proposals about legal discourse, those proposals will carry with them a proposal about whether positivism or antipositivism is true.

We can in fact push this point further. Suppose one is drawn to a descriptivist understanding of the discourse in question. And suppose one accepts a view of concepts of the kind I am sympathetic to in this paper—one on which concepts are individuated in terms of their precise cognitive significance, for example, in terms of one particular way of representing reality as opposed to another, and where we want to make fairly fine-grained distinctions between different concepts that individuals have (such that, e.g., the speakers in the Secretariat case employ distinct concepts). When we seek to specify a particular way of representing reality that someone employs when she uses a particular concept C1 (as opposed to nearby concepts C2, C3, etc.), that might very well tell us the following: the only things that fall under this concept are things that have a certain kind of nature or certain kinds of grounds. Because of that, it might very well be that conceptual analysis of a given concept—for example, the concept law employed by a group of people—yields a verdict as to whether positivism or antipositivism is true of the things picked out by that concept.Footnote ⁵⁷

This means that metalegal theorists who are interpreting legal thought and talk might well express, within a canonical dispute in metalegal theory, literal claims that yield a claim about the truth or falsity of legal positivism.Footnote ⁵⁸ Conceptual analysis—which, for purposes here, we can understand as a purely descriptive project aimed at understanding the content of concepts (e.g., those that we ourselves use)—can thus play a potentially important role in making progress in the dispute over legal positivism.

However, the model just sketched does not capture everything going on in the dispute over positivism. In some parts of the dispute, there is reason to think that speakers are not only making rival substantive semantic proposals about what a group of speakers (perhaps including themselves) mean by the term “law” (and in turn, metaphysical proposals about what follows from that proposal) as part of a canonical dispute in metalegal theory. Rather, for the reasons I give above, there is pressure to think that speakers sometimes (1) themselves mean different things by the term “law” (and hence express different concepts by that term), and (2) then use (rather than mention) the term to put forward a view about the term. What kind of view are they putting forward when they do so? As I discuss in Section I, some metalinguistic disputes are about the descriptive issue of what a word does mean. But this seems like a bad model for much of what is going on in the dispute over positivism, at least for much of what is going on when speakers are using the word in a metalinguistic way. Instead, I think, they are putting forward views about what the word should mean. If so, that means they are engaged in a metalinguistic negotiation.

Why think it is a metalinguistic negotiation as opposed to the contrasting kind of metalinguistic dispute that concerns just the descriptive facts about what a word does mean? Consider a committed Dworkinian who makes claims about what the law is and what follows from that. And suppose, for now, that as part of this, she uses the term in a metalinguistic way. Will information from a linguist about what the term currently means in the broader community convince the Dworkinian that she is mistaken (or that she is correct)? Perhaps it will in some cases, if she actually accepts the linguist's proposal. But it is also easy to imagine that she will not. For it might very well be that she thinks she has normative reason—for example, perhaps deriving from moral or political considerations—to use the term one way or another. That would hardly be surprising in the case of a Dworkinian, given that such moral and political considerations are standardly taken by Dworkinians to pervade how we should think about theory-choice in legal philosophy across a wide range of parts of the subfield. If that is right, and if it is correct that the Dworkinian and a positivist both use the term “law” in a metalinguistic dispute, then it looks like their dispute must be a metalinguistic negotiation in particular (i.e., a normative metalinguistic dispute).

To add to this, consider that a number of important proposals in the philosophy of law explicitly involve putting forward reforming definitions of the term “law.” For example, this is so in the case of Jeremy Bentham, who proposes that we use the term “law” in such a way that positivism is true of the thing picked out by it.Footnote ⁵⁹ Bentham proposes this reforming definition based partly on the perceived moral and political benefits of adopting his preferred usage. But not all reforming definitions need appeal to such values at all. For example, one might instead propose a reforming definition of a given term based on the idea that the new way of using the term will help us deliver better social-scientific explanations. One might couple that proposal with the idea that doing so will lead to some kind of moral or political progress (as in the case of Bentham). But one also need not do so—one might propose the reforming definition based on the purported theoretical virtues (rather than practical virtues) of the new way of using the term.

A reforming proposal about a given term x will often be put forward by mentioning rather than using that term x. But once the reforming proposal has been made, a speaker will then often go on to use the term in a way that aligns with the reforming definition. Such is the case with Bentham and “law.” Such usage will sometimes be metalinguistic—especially when the term is being used in line with a reforming definition in a context where others do not use the term in that way. In such a case, information about the descriptive issue about what a term does mean will not settle the issue of what it should mean. And that is what a proponent of a reforming definition ultimately cares about. This means that when (1) there is a proponent of a reforming definition of a given term, (2) she uses that term in a metalinguistic way that aligns with her proposed reforming definition, and (3) she enters into disputes with others who do not use the term in the same way, then such disputes will be metalinguistic negotiations. This supports thinking that part of what is going on in the philosophical dispute over legal positivism is a metalinguistic negotiation, given the way in which proponents of reforming definitions enter into the dispute.

Let us now take stock of where we are. I argue above that the same kind of evidence that supports thinking that the Secretariat case is a metalinguistic negotiation also supports thinking that a subset of the dispute over legal positivism is as well.

Three things should be underscored about this proposal.

First, I am not saying that all of the dispute over positivism is a metalinguistic negotiation. Indeed, as I indicate above, I do not think that is so. Rather, my proposal is that part of this dispute is a metalinguistic negotiation—and, moreover, an important enough part to be explanatorily important for the self-understanding of this dispute within legal philosophy.

Second, I am not aiming to identify which specific parts of the dispute over legal positivism are metalinguistic negotiations. Rather, I am giving a general, schematic argument for why we should think that at least some parts of it are, based on the same kind of evidence that supports thinking that the Secretariat case is a metalinguistic negotiation. The argument for this general claim lays important groundwork for further, more focused investigation into which particular parts of the dispute over legal positivism are metalinguistic negotiations. It does so by providing reasons to think that at least some parts of it involve metalinguistic negotiation, as well as providing some of the key tools to use when investigating whether specific parts do or not. That does not mean that it will be easy to deploy those tools. Indeed, it would take significant further work to make a convincing case that a given particular linguistic exchange within the dispute over legal positivism really is a metalinguistic negotiation. (Recall, for example, the philosophical complexities involved in condition (C) above, at the start of this section).

But that does not mean that the general claim is not itself already a significant one. After all, the default views of communication that dominate much of the contemporary discussion in legal philosophy—and, indeed, dominate contemporary philosophy more generally—are ones that I am rejecting in arguing for the partly metalinguistic reading of the dispute over legal positivism. If my argument is successful, it will lay the philosophical foundations for future work that would otherwise seem implausible if one were to rely on those default views within the field. At the same time, this general claim, if correct, is an important contribution to our self-understanding in legal philosophy and already has some important methodological upshots for work in the philosophy of law. Or at least so I argue below.

Third, my claim that part of the dispute over legal positivism is a metalinguistic negotiation ultimately rests on a claim about how one case compares to another: the Secretariat case and the case of the dispute over legal positivism. By itself, this comparative claim is compatible with it ultimately being a mistake to think of the Secretariat case as a metalinguistic negotiation. I think that the Secretariat case really is a metalinguistic negotiation. But, for the reasons I give above, I think it is beyond the scope of this paper to provide anything like a conclusive argument on behalf of that proposal—for doing so, as explained, is bound up with theory-choice at the foundations of semantics, of a kind that I do not have room to go into here. But even the conclusion of the more modest argument is still dialectically powerful. For, if the conclusion is correct, it shows the following: if you want to resist the idea that the dispute over legal positivism is partially a metalinguistic negotiation, then you need to either (1) deny that the Secretariat case is a metalinguistic negotiation, or (2) locate a relevant disanalogy between the Secretariat case and the case of the dispute over legal positivism. Neither is easy to do, and especially not when one is aiming to integrate one's views into a plausible overall account in the philosophy of mind and language.

What exactly does my metalinguistic proposal amount to in the case of the dispute over legal positivism, if it is correct? The core picture here is this. In part of the dispute over legal positivism, some philosophers use the term “law” to mean something different from what other philosophers mean by the term. For one meaning of the term (meaning 1), positivism is true of the thing picked out by the term “law.” For another meaning of the term (meaning 2), antipositivism is true of the thing picked out by the term “law.” (We could continue the list for any number of further meanings, e.g., meanings 3, 4, 5, etc.) A philosopher who uses the term with meaning 1 who then says “legal positivism is true” says something that is literally true. Another philosopher who uses the term with meaning 2 and who then says “legal antipositivism is true” also says something that is literally true. But that does not mean that there is nothing at issue between these two philosophers when they enter into a dispute and say such things. For, in addition to making rationally compatible (and true) claims in terms of the literal content of what they say, they also sometimes metalinguistically use terms like “law” (or relatedly, “the law,” “legal facts,” etc.) in a way that involves them advocating for a normative view about how those terms should be used. Put in terms of concepts, each of them is advocating for using the term “law” to express rival concepts. So there is a disagreement between them that is being expressed here, namely, one about which concept should be employed in the context at hand. And one (or both) of them could be getting it wrong about this normative topic.

Like “athlete,” the term “law” has rich associations in many contexts, largely independent of what is built into the semantics of the term. Some of those associations have to do with moral and political ideals, and some have to do with explanatorily important cuts within our best sociological or anthropological theories. In such a case, it can matter a great deal in a given context which concept is expressed by the term “law.” Different philosophers—often drawn to capturing different associations of the term “law”—metalinguistically advocate for attaching their preferred concept to the term and, in so doing, put forward (what are often) competing visions of which thing should play a functional role in a given part of our thought and talk and in a given part of our social practices.

At this point, many philosophers reading this paper will no doubt have started to get worried (or may have been worried for quite some time now) about the following question: Does the form of argument not extend to many other disputes in philosophy? Given the form of argument I give here, should I not also think that many philosophical disputes are at least partially metalinguistic negotiations? I think this is entirely correct. But I do not think that is a problem for my view. In other work, I argue that the same kind of argument that I give here supports thinking that a wide range of philosophical disputes across a range of subareas of philosophy are best thought of as involving metalinguistic negotiation.Footnote ⁶⁰ Or, more carefully, what I argue for is the conditional claim: if the Secretariat case is best thought of as a metalinguistic negotiation, then so, too, are many disputes in philosophy. This includes disputes that we would describe intuitively as “about” object-level issues, such as disputes involving issues of supervenience, ground, and real definition. The basic idea is this: when philosophers make claims of the form “the essence of X is Y” or “the X facts are grounded in the Y facts,” they are sometimes engaged in a metalinguistic negotiation. The dispute over positivism is, on my view, much the same as many other philosophical disputes, including ones, for example, about central philosophical topics such as free will, knowledge, epistemic justification, morality, meaning, objectivity, and justice.Footnote ⁶¹