1 Introduction: Re-evaluating the Carnap–Quine Debate

Recent years have seen a resurgence of interest in Carnap and Quine’s pivotal debate over the nature and status of ontological inquiry.¹ One motive for this reexamination is a growing sense of the inadequacy of standard historical narratives. According to the customary interpretation, Quine sought to (and succeeded in) undermining the antimetaphysical position that Carnap defends in “Empiricism, Semantics, and Ontology”, thereby “giving a kind of credibility to metaphysics that Carnap would never have countenanced” (Stroll 2000, 200). Contemporary correctives challenge this orthodoxy, arguing that the traditional interpretive camp neglects deep affinities between Carnap and Quine’s metaontological commitments and misidentifies the nature of their dispute.² However, these revisionary interpretations face a pressing challenge. If Carnap and Quine do not disagree about the status or methodology of inquiry into what there is, why are they at odds? What is really at stake in their debate? By deflating their differences, the revisionary camp risks losing any sense of genuine disagreement between the two and so failing to provide an adequate account of the motivations that drive their dispute.

This paper proposes a novel interpretation that answers these questions and resolves the interpretive challenge facing the revisionary camp. Critics are right to contest the customary narrative that positions Quine as the savior of metaphysics. On the status of existence questions, I will argue, Carnap and Quine stand in near alignment. Both reject the traditional metaphysics that Carnap calls “ontology”. And both endorse the pragmatic, empirically-grounded inquiry that Quine calls “ontology”. The appearance of sharp disagreement arises due to their differing use of “ontology” and its cognate terms. Once this difference in usage is accounted for, their seemingly contradictory claims about the subject can be reconciled.

Crucially, however, this reconciliation does not render their dispute merely verbal. Rather, it emerges as a remarkably consequential case of metalinguistic negotiation (MLN) over the term “ontology” itself. Carnap wants to replace “ontology” for positive philosophical purposes, reserving it solely to demarcate the traditional metaphysical enterprise that he considers meaningless. Quine, on the other hand, advocates that “ontology” be revised or reinterpreted, freed from its first-philosophical and suprascientific implications. On this reading, the two are engaged in a substantive disagreement about which of these ameliorative strategies offers the best response to what they see as the collapse of traditional ontology.³ And they prosecute this negotiation at the metalinguistic level by actively using the term in competing ways.⁴ This reading resolves the interpretive challenge facing the revisionary camp and draws attention to a source of substantive disagreement that interpreters have largely overlooked. It provides a novel extension of MLN from ontological to metaontological disputes and also offers a detailed historical case study for philosophers interested in evaluating different ameliorative strategies.⁵

The paper will proceed as follows. Section 2 introduces the notion of MLN and describes three hallmarks by which cases can be identified. Sections 3 through 5 provide a detailed interpretation of the central texts of the debate. They aim to show that the initial appearance of sharp first-order disagreement between Carnap and Quine is misleading. Sections 6 through 8 explore their views about the use of ontological terminology in positive philosophical theorizing. Ultimately, Carnap’s view on how “ontology” should be used—pejoratively, as a label for an enterprise that he considers fundamentally misguided—is what drives this negotiation over the term. As such, the Carnap–Quine debate over ontology is best understood as a case of metalinguistic negotiation. To close, we will consider a broader metaphilosophical contrast between the two that stimulated this negotiation.

2 Metalinguistic Negotiation

The concept of MLN was developed to capture a common discursive phenomenon—intractable disputes that hinge on different views of the appropriate application of a particular term. In a recent series of papers, David Plunkett and Timothy Sundell analyze a number of proposed examples.⁶ Consider the Secretariat case, originally inspired by an actual debate on sports talk radio.⁷ One of the radio hosts defends Sports Illustrated magazine’s claim that Secretariat is one of the fifty greatest athletes of the twentieth century. The other host disagrees. During their heated exchange, however, neither host disputes Secretariat’s racing prowess or equestrian accomplishments. The relevant descriptive facts form a shared background of agreement between the two. What their disagreement may really be about, according to Plunkett and Sundell, is whether or not it is appropriate to apply the term “athlete” to a racehorse. One of the radio hosts holds that Secretariat is an athlete and so deserves a place in the rankings. The other thinks that Secretariat is not an athlete and thus contests the racehorse’s inclusion on the list.

In metalinguistic disputes, the disputants negotiate between their divergent views by actively using the contentious term in competing ways, rather than merely mentioning it in a debate about its definition or proper use. This feature makes such disagreements hard to diagnose. In the “athlete” case, conflicting views on how the term “athlete” is used may drive the disagreement. If so, it is a descriptive metalinguistic dispute. Alternatively, it may be driven by divergent views on how the term should be used, irrespective of prevailing standards.⁸ In that case, it is a normative metalinguistic dispute or, equivalently, a case of MLN.

Negotiating over “athlete” might seem like a pointless enterprise. But consider a more consequential candidate case—a dispute over whether a given action counts as racist. This dispute could, of course, be a factual disagreement concerning the intention or impact of the action in question. But such disputes also arise due to verbal differences, even when both parties agree on all the relevant first-order facts. One disputant may apply the term “racist” to (roughly) any action that reflects race-based prejudice, while another may require (something like) a systemic position of power in addition. These differences in use suggest that the disputants possess different concepts or express different meanings with the term “racist”. If so, it follows that they “talk past” each other on the semantic level, making their dispute “non-canonical”, in Plunkett and Sundell’s terms.

It does not follow, however, that this dissolves any genuine disagreement between them. Even if the disputants use the term “racist” in a divergent manner, they may be engaged in a pragmatic disagreement about which of these uses is or should be appropriate in the context at hand. Thus, metalinguistic disputes are not merely verbal, in the way that confusions over semantically distinct but homonymous uses of “bank” might be.⁹ Drawing attention to the “bank” confusion will resolve any disagreement over whether or not there is a canoe on the bank. With a metalinguistic dispute like the one described above, noting the terminological divergence may not help: the debate may persist, since appropriate usage of the term “racist” is precisely what is at issue.¹⁰ This analysis allows one to acknowledge the linguistic or verbal character of the dispute, while still accounting for the sense of genuine disagreement between the disputants. Philosophers have applied this form of analysis to debates in an assortment of subfields, from philosophy of law to ontology itself. Here, I will argue that it aptly captures a landmark metaontological debate drawn from the history of analytic philosophy.

To show that some candidate dispute represents a case of MLN, one must demonstrate that it exhibits three distinct hallmarks. First, the initial appearance of sharp first-order disagreement between the disputants must be misleading and stem from differing applications of the central term. Second, it must be shown that the dispute nevertheless expresses a genuine disagreement or negotiation over which of these uses should prevail. And third, one must demonstrate that this disagreement is prosecuted at the metalinguistic level with “at least one speaker employ[ing] a metalinguistic usage of the term” (Plunkett 2015, 835).¹¹ A metalinguistic usage involves actively using a term, rather than mentioning it, it in order to “communicat[e] something about how to use it appropriately” (Barker 2002, 2).¹² Such a use is instructive. In the descriptive case, it provides guidance as to how the term is used. In the normative case, it conveys one’s views on how the term should be used. The presence of these instructive usages is the third and final hallmark of metalinguistic negotiation. We will begin, in sections 3 through 5, by establishing hallmark one, the illusion of first-order disagreement.

3 Carnap’s Two Projects

Carnap wrote “Empiricism, Semantics, and Ontology” (ESO) in response to the objections of philosophers that he considered fellow “anti-metaphysical empiricists—like Ernest Nagel, W. V. Quine, [and] Nelson Goodman” (PRC, 65).¹³ These philosophers argued that Carnap’s willingness to employ linguistic frameworks that include abstract entities amongst the range of the values of the variables represents “bad metaphysics” and involves “violating the basic principles of empiricism”, in his words (1950, 206, 221). In response to these objections, Carnap contends that admitting such entities “does not imply embracing a Platonic ontology but is perfectly compatible with empiricism and strictly scientific thinking” (1950, 206). His aim is to “overcome…[the] nominalistic scruples” of his objectors by showing that his admission of abstract entities does not involve “bad metaphysics,” or indeed, any metaphysics at all, in his sense of the term (1950, 206; PRC, 65). I will call this Carnap’s negative project. It comprises an elucidation of what it is for a question to be “ontological” and an argument that such questions are fundamentally misconceived. However, Carnap also seeks to clarify how debates over what entities to admit can be reformulated and fruitfully pursued. I will refer to this as his positive project. It consists of a proposal that traditional ontology be replaced by a new enterprise centered on matters of “practical decision concerning the structure of our language” (1950, 207).

To accomplish these projects, Carnap introduces his well-known distinction between framework-internal and framework-external questions. He stresses the importance of this “fundamental distinction” in a letter written to Quine just after his first reading of “On What There Is”:

Much in your discussion is illuminating: but it seems to me there is still a basic underlying ambiguity. To formulate it in the terminology of my current paper, there seems to me to be a lack of distinction between two questions: the question of existence within a framework and the question of the existence of the framework itself. (DCDV, 416)

Carnap proposes his distinction to clarify this “basic underlying ambiguity”. The key to grasping how this proposal helps to realize Carnap’s two projects is to recognize that he actually draws two importantly different distinctions in ESO by identifying three potential senses in which existence questions can be posed.¹⁴ This tripartite division distinguishes meaningful theoretical questions posed within a linguistic framework from two different construals of external questions: (i) ontological pseudo-questions about the existence or reality of the framework itself, posed prior to its adoption; and (ii) pragmatic questions about which framework we ought to adopt for some purpose.

For Carnap, the questions commonly thought of as ontological may be posed in any of the above senses. Consider, “Do numbers exist?” We can distinguish three interpretations:

This triple disambiguation establishes two different internal-external distinctions, each of which aligns with one of Carnap’s stated aims:

Carnap employs the I/E₁ distinction in service of his negative project. By circumscribing those questions that he considers “ontological” or “metaphysical”—the External₁ questions—and arguing that they amount to meaningless pseudo-questions, he aims to clear himself of the accusation of having lapsed into “absolutist metaphysics” (MN, 44). Carnap employs the I/E₂ distinction in service of his positive project. His aim is to clarify the status of the reemergent dispute over abstract entities, arguing that it amounts to a pragmatic question of simplicity and expedience for science. By these lights, he suggests, the objections of “would-be nominalists” to his use of intensions and the like are insufficient.

Carnap’s key thought is that, if interpreted as framework-internal questions, traditional ontological questions are easy. To introduce a new class of entities, like numbers, into a Carnapian linguistic framework, one must specify the rules that govern their use. Once these are in place, one can formulate internal existence questions about entities of this type. One can ask, for example, “Is there a prime number greater than 1,000,000?” One can also pose the traditional philosophical question, “Are there numbers?”, in an internal sense. A positive verdict trivially follows from countless internal claims, such as “There is a number between 3 and 5”. Carnap thinks this very triviality “makes it plausible to assume that those philosophers who treat the question of the existence of numbers as a serious philosophical problem and offer lengthy arguments on either side do not have in mind the internal question” (1950, 209).

Instead, Carnap suggests, traditional ontologists mean to be posing an External₁ question. Such questions concern the existence or reality of the entire system of entities introduced by the linguistic framework and must be posed and answered prior to its adoption. Given a linguistic framework with the resources to formulate the question, “Do numbers exist?”, the claim that “there are numbers” is true and it is trivial to prove it, Carnap thinks. The traditional ontologist wants to ask a further theoretical question: do numbers really exist? Are they among the dinge an sich—the ultimate furniture of reality? On this understanding of ontology, the introduction of numbers, propositions, properties, physical objects, or any other class of entities is “legitimate only if it can be justified by an ontological insight supplying an affirmative answer to the question of reality” (1950, 214). Only once ontological inquiry has established that there really are such entities will we be warranted in employing linguistic frameworks that quantify over them. This is ontology as traditionally conceived, as First Philosophy.

Carnap rejects such external questions as devoid of cognitive content. Recognizing an entity as real or existent in the “empirical, scientific, non-metaphysical” sense “means to succeed in incorporating it into the system… so that it fits together with the other things recognized as real” (1950, 207). In other words, Carnap holds that existence is an internal concept. Questions about and ascriptions of existence make sense only within a given linguistic framework. Thus, the attempt to evaluate a language from some imagined external position misfires. “To be real in the scientific sense means to be an element of the system; hence this concept cannot be meaningfully applied to the system itself” (1950, 207). In effect, the traditional ontologist wishes to take up a stance “outside” language and survey reality as it is in itself. We may treat H₂O molecules as real within our best chemical frameworks, but does this framework itself correspond to reality? In doing so, however, Carnap holds that they simply deprive themselves of the ability to formulate a meaningful question at all. “The controversy goes on for centuries without being solved. And it cannot be solved because it is framed in a wrong way” (1950, 207). There is no form of direct “ontological insight” into unconceptualized reality, Carnap thinks, and so no way of posing or pursuing meaningful suprascientific questions about whether language mirrors the world as it is in itself.

In pursuit of his positive project, Carnap proposes that we replace these allegedly theoretical External₁ questions with pragmatic External₂ questions, to be decided based on the “efficiency, fruitfulness, and simplicity” of the linguistic framework (1950, 208). Given the utility of the chemical framework, the External₂ question will receive an affirmative answer. Given the methods of justification that this framework provides, the internal question, “Do H₂0 molecules exist?”, will receive a positive (synthetic) answer. Any alleged further theoretical question is devoid of cognitive content.

For other entities, such as propositions or intensions, the pragmatic question of the explanatory utility of the posited frameworks is more challenging. For Carnap, such framework-external questions are meaningful and, in fact, “belong to the most important problems of philosophy” (PRC, 862). However, he insists that pragmatic External₂ questions are replacements for the “pseudo-theoretical ontological questions” that he rejects, rather than ontological questions in their own right (PRC, 871).¹⁵ Choosing between “conceptual frameworks” is an “engineering problem,” like choosing a motor for an airplane (PRC, 871). He sees no reason “why metaphysics should enter into the first any more than into the second” (MN, 43). “The decisive question is not the alleged ontological question of the existence of abstract entities,” he writes, but the question of whether the introduction of such entities will prove “expedient and fruitful” for science (1950, 220–21). Carnap views this latter question as part of a fundamentally new enterprise of conceptual engineering and insists upon the contrast between this enterprise and traditional ontological inquiry.

To complete his positive project, Carnap applies this pragmatic methodology to the case of abstract entities. Given the widespread utility that abstract entities offer in science and semantics, “very strong reasons must be offered” if Quine or other would-be nominalists wish to reject them (PRC, 66). Carnap does not aim to reach a final verdict on this question in ESO. He hopes to show that, if the dispute is understood as a matter of evaluating the pragmatic utility of quantifying over abstract entities, nominalists have failed to offer sufficient grounds for objection. Thus, he is not trying to dodge his commitment to abstract entities in ESO, as is often assumed.¹⁶ Carnap’s semantic method does bring with it an internal commitment to the existence of intensions. He insists, however, that this should not be considered an ontological commitment, as it does not depend on any claims to “ontological insight” or a prior “metaphysical doctrine concerning the reality of the entities in question” (1950, 214). In this way, he aims to defend this commitment and to overcome the objections of his nominalistic critics. For Carnap, empiricism is not about what entities one posits, but why one posits them. He concludes ESO with a call for tolerance toward any entities that may prove useful to scientific inquiry. “Prohibitions based on prejudices deriving from religious, mythological, metaphysical, or other irrational sources” are “worse than futile,” they are “positively harmful because [they] may obstruct scientific progress” (1950, 221).¹⁷

4 Quine’s Rejection of Traditional Metaphysics

Now that we have a grip on the position that Carnap defends in ESO, it is time to examine Quine’s views. Over the course of the next two sections, I will argue that Quine is amenable to each of Carnap’s two projects. To begin, let’s consider Carnap’s critique of traditional ontology. Though he is often thought to have “saved” metaphysics from Carnap’s attacks, this mischaracterizes his position.

Before we proceed, however, one crucial issue must be directly addressed. It may seem puzzling that we have made it through a discussion of Carnap’s views on ontology without identifying some crucial role for the analytic-synthetic distinction. Analyticity was arguably the central issue in the Carnap-Quine debate and is often taken to be intimately tied to their disagreement about ontology. Quine himself did much to popularize this interpretation. He begins his 1951 essay, “On Carnap’s Views on Ontology”, with the following passage:

Though no one has influenced my philosophical thought more than Carnap, an issue has persisted between us for years over questions of ontology and analyticity. The questions prove to be interrelated; their interrelations come out especially clearly in Carnap’s paper “Empiricism, semantics, and ontology”. (WOP, 203)

Over the course of the paper, which he presented at a University of Chicago ontology symposium with Carnap in 1951, Quine argues that Carnap’s internal-external distinction ultimately depends upon the analytic-synthetic distinction. Because he rejects the latter, he claims that he must dismiss the former as well.

However, Quine’s interpretation of their disagreement misses the mark.¹⁸ The I/E₁ distinction, and so Carnap’s criticism of traditional metaphysics, does not depend upon the analytic–synthetic distinction. It does not depend on either of the first two dogmas of empiricism. Rather, it reflects Carnap’s adherence to what Donald Davidson called the “third dogma of empiricism’: the dualism of conceptual scheme and empirical content (Davidson 1973).

Scheme–content dualism is a commitment that Carnap and Quine share. Quine’s Duhemian holism about empirical confirmation leads him to reject the idea that empirical content can be parceled out sentence-by-sentence. For this reason, among others, he rejects the claim that individual sentences can be true solely in virtue of their meaning. Analyticity goes by the board, along with Carnap’s notion of a formal linguistic framework with explicit semantical rules. In its place, Quine talks of “conceptual schemes”—sets of intertranslatable languages with truth-values distributed across them. However, Quine preserves the notion of empirical content. The difference is simply that the scheme as a whole must face up to “the tribunal of sense experience”, rather than individual sentences (FLPV, 41). The core conception of a system of linguistic categories—a conceptual scheme—imposing form upon the chaos of experience remains in place. Indeed, this is precisely the role that ontological “posits” play on Quine’s view—they serve to “work a manageable structure into the flux of experience” (FLPV, 44).

This distinction between a conceptual scheme and the empirical content it serves to systematize is all that one needs to formulate the I/E₁ distinction, the distinction between questions posed within a scheme and external questions about whether the scheme corresponds to reality. In fact, this is just what Quine himself does. In his essay, “Identity, Ostension, and Hypostasis”, published in 1950 at the height of their debate, Quine writes:

We can improve our conceptual scheme, our philosophy, bit by bit while continuing to depend on it for support; but we cannot detach ourselves from it and compare it objectively with an unconceptualized reality. Hence it is meaningless, I suggest, to inquire into the absolute correctness of a conceptual scheme as a mirror of reality. Our standard for appraising basic changes of conceptual scheme must be, not a realistic standard of correspondence to reality, but a pragmatic standard. (FLPV, 79)

Working within a given conceptual scheme, we can query the existence of one set of entities or another. However, the attempt to ask whether the scheme as a whole corresponds to the world as it is in itself “is meaningless”.¹⁹ Quine lodges this charge of meaningless on the same empiricist basis as Carnap—there is no way to step outside of one’s scheme to “compare it objectively with an unconceptualized reality”. There is simply “no such cosmic exile”, as Quine will later put it (WO, 275). He echoes Carnap’s conclusions as well. “Reality” and “existence” are, for Quine immanent notions; ontological commitments are internal to a conceptual scheme.²⁰ And our standard of evaluation is “not a realistic standard of correspondence to reality, but a pragmatic standard”.

Thus, Carnap’s negative project offers no point of contrast between the two. Quine agrees wholeheartedly with Carnap’s critique of traditional External₁ questions. Before we move to Carnap’s positive project, consider one final passage from Quine’s later writing that articulates this bedrock of agreement explicitly:

To ask what reality is really like, however, apart from human categories, is self-stultifying. It is like asking how long the Nile really is, apart from parochial matters of miles or meters. Positivists were right in branding such metaphysics as meaningless. (POT, 9)

5 Quinean Ontology and the Point of Carnap’s Positive Project

We have seen that Quine agrees that the External₁ questions Carnap calls “ontological” are meaningless. Carnap likewise agrees that the questions Quine calls “ontological” are both meaningful and important. He writes:

I agree, of course, with Quine that the problem of “Nominalism” as he interprets it is a meaningful problem; it is the question of whether all natural science can be expressed in a “nominalistic” language, that is one containing only individual variables whose values are concrete objects, not classes, properties, and the like. However, I am doubtful whether it is advisable to transfer to this new problem in logic or semantics the label “nominalism” which stems from an old metaphysical problem. (MN, 43)

Carnap does not challenge the significance of Quine’s interpretation of “nominalism.” He seems to think it obvious that he has no reason to challenge it. He objects only to Quine’s decision to apply the word “nominalism” to what he considers a “new problem[s] in logic or semantics” (MN, 43). Had Quine described this position as “a thesis in the philosophy of science”, rather than an ontological thesis, Carnap would not have objected at all (Quine 1939, 708).

When one examines the theses that Quine calls “ontological” but advances without recourse to traditional terminology, they epitomize Carnap’s “pragmatic standard.” Consider the conclusion of Quine’s 1953 paper, “On Mental Entities”:

The issue is merely whether… it is efficacious so to frame our conceptual scheme as to mark out a range of entities or units of a so-called mental kind in addition to the physical ones. My hypothesis, put forward in the spirit of a hypothesis of natural science, is that it is not efficacious. (WOP, 227)

Carnap would thoroughly approve. Again, the explanation lies in their shared commitment to scheme-content dualism. Choosing what entities to admit is, for both, a pragmatic matter of framing one’s scheme. “Pragmatic” because, beyond adequately capturing its empirical content, all that matters in evaluating a scheme is convenience, simplicity, and the like. There is no further “realistic standard of correspondence to reality” to which the scheme must adhere (FLPV, 79). Huw Price writes, “The traditional metaphysician wants to be able to say, ‘I agree that it is useful to say this, but is it true?’ Carnap rules out this question, and Quine does not rule it back in” (2009, 326). In fact, as we have seen, Quine does not even attempt to rule it back in. His ontological methodology is every bit as pragmatic as Carnap recommends. Insofar as Quine hews to this methodology, he and Carnap stand in alignment.

However, Carnap’s positive project was no idle undertaking. It was prompted by his reaction to Quine and Nelson Goodman’s 1947 paper, “Steps Toward a Constructive Nominalism”. In this essay, Quine lapses from, or has yet to fully embrace, this pragmatic metaontology (FLPV, 79). He and Goodman open the essay by writing:

We do not believe in abstract entities. No one supposes that abstract entities—classes, relations, properties, etc.—exist in space-time; but we mean more than this. We renounce them altogether... Why do we refuse to admit the abstract entities that mathematics needs? Fundamentally this refusal is based on a philosophical intuition that cannot be justified by appeal to anything more ultimate. (Goodman and Quine 1947, 105)

This passage may have played a role in prompting Carnap to write ESO. Goodman and Quine seem to concede that mathematics requires abstract entities. Yet they refuse to admit them based on “a philosophical intuition that cannot be justified by appeal to anything more ultimate” (1947, 105). They do not describe this as an “ontological insight”, but Carnap seems to have perceived it as such and was duly dismayed.²¹

Quine’s lapse was brief. He disavows these remarks in the bibliography of From a Logical Point of View and explains their “demotion” again in Word and Object (WO, 243). As he himself would later explain:

Whether to [admit abstract entities] may reasonably be decided by considerations of systematic efficacy, utility for theory. But if nominalism and realism are to be adjudicated on such grounds, nominalism’s claims dwindle. The reason for admitting numbers as objects is precisely their efficacy in organizing and expediting the sciences. (WO, 236–37)

This was precisely the point that Carnap aimed to clarify by introducing his I/E₂ distinction in ESO—that the admission of various entities could only be evaluated on pragmatic grounds. Evidently, his discussions with Quine had their intended effect. Quine overcame his “nominalistic scruples” and distanced himself from the dogmatic declarations of “Steps Toward a Constructive Nominalism” (1950, 206). With this, Carnap and Quine’s metaontological positions are once more brought into alignment.

Carnap acknowledges as much toward the end of ESO and reiterates the sentiment in his “Intellectual Autobiography”. At the end of the 1951 ontology symposium, he concluded that he and Quine did not differ “concerning any matter of fact, nor any question with cognitive content” (Stein 1992, 279). According to an attendee of the symposium, “Quine happily assented to Carnap’s diagnosis” (1992, 279).²² Echoing the disputants themselves, then, we may conclude that this debate exhibits the first hallmark of metalinguistic negotiation: it does not hinge upon a disagreement over any relevant matters of fact.²³ In Carnap’s sense, Quine is not an ontologist, he is a fellow “anti-metaphysical empiricist” (PRC, 65). (Or rather, Quine may only have been an ontologist for a matter of some months in 1947). And Carnap is an ontologist in Quine’s sense, one who seeks to defend his provisional commitment to the scientific utility of abstract entities. Thus, contrary to customary interpretations that frame Carnap as a vehement opponent of metaphysics and Quine as its savior, the two in fact endorse closely related and compatible metaontological views.

6 Hallmark Two: Carnap’s Genuine Disagreement

Recognizing that Carnap and Quine “talk past” each other in their proclamations about ontology, Marc Alspector-Kelly suggests that their debate may be “merely verbal” (2001, 115). Arguing along similarly deflationary lines, Matti Eklund concludes that “there is then no real conflict between them” (2013, 246). To the contrary, I will argue, the basic point of contention has just emerged: whose preferred usage of “ontology” should prevail? Should an enterprise that Quine freely admits is “at bottom just as arbitrary or pragmatic a matter as one’s adoption of a new brand of set theory or even a new system of bookkeeping” be considered ontology (WOP, 132)? On the other hand, why should the term be reserved for an enterprise that Carnap considers meaningless? Over the final three sections, we will see that Carnap and Quine engage in a substantive metalinguistic dispute on precisely this issue. Given the meaninglessness of traditional ontology, Quine feels free to adopt the term as a label for his own enterprise. Carnap strongly opposes this on both descriptive and normative grounds. He negotiates in favor of his own view—that “ontology” be reserved as a label for the defunct traditional project—by instructively using the term in his preferred manner (DCDV, 386). With hallmarks two and three thus established, we may conclude that the debate is best understood as a case of MLN. This diagnosis provides an account of the source of disagreement driving Carnap and Quine’s dispute, answering the challenge that faces the revisionary interpretive camp.

We begin with the second hallmark, the presence of genuine disagreement. Quine was aware of Carnap’s objections to his use of ontological terminology. Carnap raised them in his letters as early as 1945. At the outset of “On Carnap’s Views on Ontology,” Quine notes, “Carnap does not much like my terminology here... he disapproves of my giving meaning to a word which belongs to traditional metaphysics” (WOP, 203). However, it seems that he struggled to grasp the grounds for Carnap’s resistance. On the back of a letter in which Carnap pressed the point, Quine wrote:

Why reserve “platonism”, “ontology”, etc. when you deny you are reserving them for any meaning? Why not say “platonistic” (or “realistic”) math. & physics? (DCDV, 417)

Quine’s perplexity is understandable. Why would Carnap hold that the term “ontological” should be reserved to designate External₁ questions, when he explicitly argues that such questions are meaningless?

The first reason for Carnap’s resistance is descriptive. He thinks that Quinean ontology diverges too drastically from the traditional enterprise to warrant his adoption of the same term. I will call this the “traditionalist” objection. This objection is, in part, a product of Carnap’s philosophical background. His education in Europe put him in contact with many of the prominent traditional metaphysicians of the era. He wrote his dissertation under the neo-Kantian Bruno Bauch and knew what “ontology” signified for philosophers in this tradition, within which the term first came into use.²⁴ Such philosophers, he felt, would not consider Quine’s project part of the same enterprise. Carnap considers it something like definitional of “ontology” that it purports to involve some avenue of insight that is neither empirical nor conceptual, that it claims “to find and express knowledge that is inaccessible to empirical science” (DCDV, 161). Given Quine’s abandonment of this traditional conception, Carnap thinks his adoption of “ontology” represents an unreasonable divergence from traditional use.²⁵

These points emerge in one of Carnap’s most forceful letters to Quine:

Here is an important question which you must answer in order to make your conception clearly understandable: what is the nature of questions like: “Are there classes (properties, propositions, real numbers, etc.)?” and of the true answers to them? You call them ontological & even frankly metaphysical. I suppose this means that you regard them neither as analytic (purely logical) nor as empirical. Are they then synthetic a priori so that you abandon empiricism? Or what else? More specifically, what is the method of establishing their truth?... Perhaps Kant’s transcendental analysis or Husserl’s “Wesensschau”? (DCDV, 406)

Calling such questions “ontological & even frankly metaphysical” implies to Carnap that they are neither empirical nor analytic and that “the method of establishing their truth” lies outside the ambit of natural science. Quine has no intention of reviving this kind of first-philosophical inquiry.²⁶ For precisely this reason, however, Carnap believes that “the Quinean private sense of the word”, as he puts it, differs too dramatically from “the generally accepted sense” to warrant its adoption (DCDV, 378). Quine, he thinks, is changing the subject and should update his lexicon accordingly.²⁷

If Carnap was merely concerned with the customary use of the term “ontology”, this case would represent a descriptive metalinguistic dispute. But Carnap was no ordinary language philosopher. Given his commitment to engineering language, fidelity to past usage is not an ironclad obligation. Carnap’s primary concern lies not in how the term “ontology” is or has been used, but how it should be used by contemporary philosophers. Given its traditional associations, Carnap thinks that Quine’s reinterpretation of “ontology” will be misleading. He voices this concern explicitly in Meaning and Necessity:

I should prefer not to use the word “ontology” for the recognition of entities by the admission of variables. This use seems to me to be at least misleading; it might be understood as implying that the decision to use certain kinds of variables must be based on ontological, metaphysical convictions. (MN, 43)

This passage lays bare the real core of Carnap and Quine’s dispute.²⁸ Carnap worries that Quine’s adoption of this traditional terminology will be seen as licensing the revival of traditional metaphysical arguments based on a priori intuitions and brute metaphysical convictions. As their correspondence attests, Carnap was initially led astray himself, unsure whether Quine’s talk of ontology meant that he intended to revive traditional metaphysics. Shortly, I will suggest that some later interpreters of Quine may have been similarly misled. For his own part, Carnap always “tried to avoid the terms of traditional philosophy and to use instead those of logic, mathematics, and empirical science” (PRC, 21). As Thomas Ricketts writes, it is hard to overstate “the centrality of Carnap’s interest in distinguishing genuine problems from pseudo-problems” (1994, 193). He advocates reserving the term “ontology” for precisely this function, as a way of demarcating the External₁ pseudo-questions that he and Quine reject. For positive philosophical purposes, he believes that the term should be replaced entirely. As we will now see, Carnap employs instructive metalinguistic usages to advocate for this position, the final hallmark of MLN.

7 Hallmark Three: Carnap’s Metalinguistic Usages of the Term “Ontology”

Given that Carnap explicitly objects to Quine’s use of the term “ontology” in their correspondence, it may be surprising that he only references this objection briefly in ESO.²⁹ This scarcity of explicit commentary is explained by the fact that, throughout the paper, Carnap contests Quine’s reinterpretation of “ontology” and advocates for his own usage not by mentioning the term but by instructively using it in the requisite manner. In this section, I will argue that Carnap employs both descriptive and normative metalinguistic usages of the term “ontology” and its cognates in ESO. In so doing, I aim to demonstrate the explanatory utility of these notions and, more broadly, illustrate how contemporary insights into the complexity of communication can help inform interpretive work in the history of philosophy. With our final hallmark established, we may conclude that Carnap and Quine’s dispute is best analyzed as a case of MLN.

In the central section of ESO, where Carnap first introduces linguistic frameworks and his internal-external distinctions, he includes a fictional conversation with a traditional metaphysician. If asked whether they intended to pose an internal question, Carnap says, a traditional ontologist “would probably reply: ‘Not at all; we mean a question prior to the acceptance of the new framework’” (1950, 209). “They might try to explain what they mean”, Carnap continues, “by saying that it is a question of the ontological status of numbers; the question whether or not numbers have a certain metaphysical characteristic called reality (but a kind of ideal reality, different from the material reality of the thing world)” (1950, 209, my italics). The decision to include fictional dialogue is an anomaly in the text, and in Carnap’s writings more broadly.³⁰ The explanation, I suggest, is that this conceit allows him to put what I have called the traditionalist objection in the traditionalists own voice, exhibiting how the term “ontological” is used in its traditional sense: as a way of designating External₁ questions. Thus, the dialogue serves an instructive function. Its purpose is to convey what Carnap takes to be the “generally accepted sense” of the term “ontology” (DCDV, 378).³¹ As such, it is best understood as a descriptive metalinguistic usage of the term.

Ultimately, Carnap and Quine’s disagreement does not center on how the term “ontology” is or has been used, but on how it should be used in philosophy. Carnap conveys his views on this subject through additional normative metalinguistic usages. The two clearest cases appear in Carnap’s introduction and conclusion. At the end of the introduction, Carnap writes, “It will be shown that using such a language [i.e., one that quantifies over abstract entities] does not imply embracing a Platonic ontology but is perfectly compatible with empiricism and strictly scientific thinking” (1950, 206). Commentators admit that this passage is “perplexing” (Soames 2009, 428). In one breath, Carnap may seem to both commit himself to an ontology of abstract entities and to disavow this commitment. To rectify this apparent conflict, some have tried to find a way to interpret Carnap as a nominalist, despite the evidence to the contrary.³² By identifying this as a normative metalinguistic usage of “ontology”, we can avoid these interpretive contortions. In this passage, Carnap is not trying to deny that his approach to semantics commits him to quantification over intensions. Rather, he is advocating that this commitment ought not be described as “embracing a Platonic ontology” since it is motivated solely by “empiricism and strictly scientific thinking”. On Carnap’s preferred usage, ontology and scientific thinking are mutually incompatible. We can make sense of the “perplexing” statement above, I suggest, by reading Carnap as employing an instructive metalinguistic usage to get this commitment across.

The normative metalinguistic usage in Carnap’s conclusion is even more striking:

For those who want to develop or use semantical methods, the decisive question is not the alleged ontological question of the existence of abstract entities but rather the question whether the use of abstract linguistic forms or, in technical terms, the use of variables beyond those for things (or phenomenal data), is expedient and fruitful for the purposes for which semantical analyses are made (Carnap 1950, 220–21).

In this passage, Carnap contrasts the “alleged ontological question” with the question whether the admission of abstract entities “is expedient and fruitful.” For Quine, the latter question just is the “ontological question”. Carnap knows as much and seems to be addressing Quine directly, yet still insists upon the contrast. Why? Wouldn’t that render his remark idle? Not if we recognize this as a normative metalinguistic usage, the point of which is to resist Quine’s assimilation of these questions and demonstrate how he thinks the term “ontology” should be used—to mark a sharp distinction between the pragmatic, naturalistic inquiry that Quine is engaged in and the External₁ pseudo-questions that they both reject. Taken together, these interpretations build a cumulative case that Carnap employs normative metalinguistic usages in ESO, establishing the third and final hallmark of metalinguistic negotiation.

Identifying these metalinguistic usages helps to explain some of the more puzzling passages in Carnap’s work. Consider the passage quoted at the outset of Section 5.³³ In his paper, “The neo-Carnapians,” Peter van Inwagen explicitly questions the purpose of Carnap’s use of ontological terminology. He asks:

What explains Carnap’s decision to use the phrase “the ontological reality of universals” in the passage I have quoted? Why did he say “questions... concerning the ontological reality of universals” and not simply “questions… concerning the existence of universals’?’—or, even more simply,”the question whether there are any universals”? (van Inwagen 2020, 24).

These last two sections provide us with the answer: because Carnap thinks that traditional metaphysical terminology should be reserved to demarcate External₁ questions. In the last two sentences of the passage, Carnap mentions the term “nominalism” to make this point explicitly. In the first, he employs a metalinguistic usage of “ontological reality” to convey the same message.

Carnap’s chief worry is that Quine’s reappropriation of these terms will prove misleading, seeming to imply that the choice of conceptual scheme “must be based on ontological, metaphysical convictions” (MN, 43). Van Inwagen’s essay offers a suggestive case study. Van Inwagen, the standard bearer of contemporary “neo-Quinean” ontology, argues that Quine’s formulation of nominalism is a metaphysical thesis, rather than a “new problem in logic or semantics”, as Carnap suggests. To defend this claim, he appeals to Goodman and Quine’s motivations in writing “Steps Toward a Constructive Nominalism.” “Why did they engage in this exercise in ‘logic or semantics’?”, he asks (2020, 29). Precisely because of their prior conviction that abstract entities do not exist. This conviction, in turn, stems from a commitment to certain metaphysical principles.³⁴ Thus, he concludes, it is a metaphysical thesis and Goodman and Quine are metaphysicians.³⁵ As we have seen, Carnap would actually agree with this diagnosis—arguments about what exists based on brute a priori intuitions are precisely what Carnap calls “ontological”. On his view, Quine and Goodman are engaged in “bad metaphysics” here. But that was also Quine’s own considered view. It is striking that, to claim him as a metaphysician, van Inwagen must appeal to the only passage in Quine’s oeuvre that he ever explicitly disavowed.

Carnap, I suspect, would not have been shocked at strained readings and the widespread sense that Quine saved metaphysics. In his view, Quine courted these misunderstandings by reappropriating ontological terminology. Better to preempt them, Carnap urged, by demarcating as clearly as possible the shift away from traditional metaphysics and towards a new, thoroughly pragmatic approach to philosophy.

8 New Wine in Old Bottles: Quine’s “Ethics of Terminology”

We have now seen that the Carnap-Quine debate exhibits all three hallmarks of metalinguistic negotiation: they talk past each other on the semantic level while engaging in a substantive normative dispute with at least one disputant, Carnap, contesting the appropriate use of the central term via metalinguistic usages. With this analysis in place, we may conclude that Carnap and Quine were confederates, not opponents, when it came to empiricism and ontology (if not semantics). To close, we will briefly consider Quine’s position in this negotiation and the broader metaphilosophical clash that underlies he and Carnap’s disagreement.

Burton Dreben, Quine’s longtime colleague and collaborator, discusses Quine’s taste for appropriating traditional philosophical terminology. “Quine never—at least, hardly ever—resolves a ‘traditional problem’; he dismisses it”, Dreben (1992, 308) writes, but he “constantly reconstrues the labels, puts new wine in old bottles” (1992, 311). In the case of “ontology,” Quine says that since “meaningless words... are precisely the words which I feel freest to specify meanings for”, he has no issue “giving meaning to a word which belongs to traditional metaphysics” (WOP, 203). Dreben argues that this distinctive “ethics of terminology” leads readers to continually underestimate the radical nature of Quine’s views and attributes it to Quine’s deep-seated insouciance towards “philosophical labels” (WOP, 203; Dreben, 1992, 296). Whereas “sharp boundaries were Carnap’s style early and late,” Quine exhibits “supreme indifference” toward the classification of philosophical problems and pursuits (QID, 144, 295, 992).³⁶ Thus, as a general rule, Quine claims to have “no quarrel with traditionalists who protest my retention” of traditional terms (POT, 19). Get the facts right, he insists, and “let the classifications fall where they may” (OQ, 500).

With “ontology”, however, Quine does quarrel with the traditionalist objection and explicitly advocates on behalf of his “adoption” of the term. “My adoption of the word ‘ontology’… is not as arbitrary as I make it sound”, he writes. “Though no champion of traditional metaphysics, I suspect that the sense in which I use this crusty old word has been nuclear to its usage all along” (WOP, 203). In considering Carnap’s position, we have focused on the differences between Quinean ontology and traditional framework-external inquiry. But Quine is still interested in what there is and this has, of course, been central all along. Is this continuity sufficient to maintain accord with the “generally accepted sense” of “ontology”, despite the radical shift in methodology and application of the term (DCDV, 378)? The issue has divided philosophers. David Armstrong, for one, considers Quine an anti-metaphysician; David Lewis disagrees.³⁷

In the end, however, it is not the descriptive disagreement but the normative one that drives this debate: should Quine have called the pragmatic attempt to parse and refine the commitments of our best scientific theories “ontology”?³⁸ Here we must remember that instrumental questions of this sort can be answered only relative to a fixed purpose or aim. In their desire to forge a new, scientifically respectable form of philosophy and jettison components of the tradition that diverge from this vision, Carnap and Quine’s aims align. However, they differ in how much weight they attach to each aspect of this enterprise. For Carnap, the key to progress in philosophy lies in clearly identifying the intractable debates and unjustifiable methodological assumptions embedded in the tradition so that they can be cast aside for good. Carnap’s “negative project” in ESO exemplifies this vision of a worthwhile philosophical contribution. For Quine, however, this critical or “therapeutic” task is distinctly secondary; his primary goal is to develop new positive projects that can be carried out in a naturalized manner.³⁹ If these can be framed as refinements of traditional pursuits, all the better. If the old bottles are empty, why not fill them up?

Thus, their normative metalinguistic disagreement ultimately points to a more general clash of metaphilosophical approaches between the two. Carnap and Quine’s lasting influence is a testimony to the compelling nature of their visions for the future of philosophy. Though they agree on the nature and status of existence questions, these broader visions do not always align. Amie Thomasson has emphasized that MLNs often emerge as a way of working through deeper differences. She writes, “How we use words matters, given their relations to other aspects of our conceptual scheme, and to our non-verbal behavior” (2017a, 11). This can make their use well-worth contesting. In the case of “ontology”, Carnap and Quine are each intent on framing the field in a manner that fits their metaphilosophical inclinations. Carnap, the revolutionary, emphasizes the rejection of ossified tradition; Quine prefers reform and emphasizes whatever continuities he believes will be of use of in the advance of post-positivist philosophy.⁴⁰ Their negotiation over the term “ontology” was one way in which they addressed this discrepancy.

At the close of their 1951 ontology symposium, Carnap concluded that, if he and Quine could live another two hundred years, they might be able to evaluate which vision had proved most fruitful (Stein 1992, 279). By this standard, it is still too soon to tell. Final assessment of this metalinguistic negotiation will remain a task for future historians. Our goal here has been to get the facts right. Let the classifications fall where they may.

Acknowledgements

I would like to thank two anonymous reviewers for their helpful comments on this paper. I am deeply grateful to James Levine, James T. M. Miller, and Amie Thomasson for their guidance on early drafts of the manuscript and to Richard Kimberley Heck, Elizabeth Miller, and Joshua Schechter for suggestions that helped me to complete it. I would also like to thank Frank O’Boyle, Haigen Hessou Messerian, Jacob Russin, and Stavros Orfeas Zormpalas for countless conversations about Carnap and Quine.