Semanticism and Ontological Commitment

Abstract

It is widely assumed that if ontological disputes turn out to be verbal they ought to be dismissed. I dissociate the semantic question concerning the verbalness of ontological disputes from the pragmatic question on whether they ought to be dismissed. I argue that in the context of ontological disputes ontologists ought to be taken to communicate views with conflicting ontological commitments even if it turns out that on the correct view of semantics they fail to literally-express their disagreement. I argue, that is, against dismissing ontological disputes on grounds of verbalness. This serves to discharge the ongoing debate on the verbalness of ontological disputes from the metaontological consequences typically associated with it.

1 Introduction

Semanticism in metaontology is the view that (some) ontological disputes are pointless in virtue of being verbal. This paper aims to uncover the fruitfulness of ontological disputes while pointing out the limits of semanticism. Unlike the standard critical response to semanticism¹ it doesn’t aim to prove the ontological disputes in question not to be verbal but instead, to argue that even supposing that they are, they still reflect genuine disagreements and hence, that they are not pointless. Intuitively: a dispute is verbal when the linguistic exchange between interlocutors does not involve a conflict regarding the facts in question but regarding the terminology used to describe the facts. A dispute is merely verbal when it’s pointless in virtue of being verbal. In order to prove ontological disputes not merely verbal, I’ll argue 1. That a coherent formulation of semanticism involves a non-Quinean conception of ontological commitment. I’ll show, that is, that it must dissociate the ontological commitments (of at least some theories) from the values of bound variables. 2. That given such a conception of ontological commitment, securing the verbalness of ontological disputes falls short of establishing semanticism. Roughly and very briefly: reflecting on considerations raised in ontological disputes, it’s the ontological commitments of their theory that ontologists are trying to express. We ought to take ontologists to communicate views with conflicting ontological commitments, even if the different quantified sentences they use fall short of expressing it.

The main point I wish to make is a positive point about successful communication of conflicting ontological commitments. In light of that, the argument is general enough to apply to every form of semanticism. However, in order to emphasize the robustness of the critical aspects, rather than attacking a toy semanticist view, I focus on a specific defence of semanticism due to Eli Hirsch (2010).² Although somewhat out of fashion, this defence is elaborate enough to help me articulate my main point before generalizing.

The paper proceeds as follows: In Sect. 2 I outline Hirsch’s semanticism. In Sects. 3–4 respectively, I distinguish between verbal disputes and merely verbal disputes, and characterize verbal disputes in which a genuine disagreement is communicated. In Sect. 5 I show that if ontological disputes are verbal, ontologists are mistaken about the ontological commitments of what (at least some) theorists express. Based on the insights gained by that point, in Sect. 6 I argue that if even if ontological disputes are verbal, they are not merely verbal: ontologists communicate a genuine disagreement about what our best theory is ontologically committed to. Section 7 concludes.

2 Hirsch’s Semanticism

Hirsch’s semanticism rests on the claim that ontological disputes are verbal. But what is a verbal dispute? Consider the following case (based on Hirsch ([2005], 2010, p.146)):

Ami/Tami: Ami and Tami are having a dispute over whether there are two cups on the table in front of them. Ami says there are two cups. Tami says there aren’t, thinking there’s exactly one cup and one glass. Both agree that there are two drinking vessels on the table (and on all relevant perceptual facts).

Intuitively, this dispute is verbal. The linguistic exchange does not seem to express a conflict regarding the facts in question but regarding the terminology used to describe the facts. Is there a more precise characterization to be had? Rather than relying on Hirsch’s full characterization of verbal disputes, a characterization that is theoretically loaded,³ I will henceforth operate with the following merely sufficient condition (based on Hirsch (1993, p.181)):

Verbal: A dispute over a sentence S is verbal when, on a charitable interpretation, one party takes S to mean S₁, another party takes S to mean S₂, and both parties can (consistently) agree about the truth values of S₁ and S₂.

The Ami/Tami case clearly meets this sufficient condition. On a charitable interpretation, Ami and Tami’s dispute over ‘there are two cups’ rests on the different meanings each disputant associates with ‘there are two cups’. Ami takes ‘there are two cups’ to mean ‘there are two drinking vessels’ (which both parties can affirm). Tami takes it to mean something like ‘there are two drinking vessels with a handle’ (which both parties can deny).

Importantly, as Hirsch emphasizes, verbal disputes can sometimes be settled: if ‘there are two cups’ in fact means something like ‘there are two drinking vessels with a handle’, Ami is simply wrong in saying that ‘there are two cups on the able’ and Tami is simply right in saying that ‘there aren’t’.

Hirsch believes that some ontological disputes, qua disputes concerning what there is (the truth value of certain quantified sentences), are verbal. One of Hirsch’s favourite examples is the composition debate (COMP), concerning the question ‘what (sorts of) composite material objects are there?’⁴ Here are some of the views philosophers have stood up for in the context of this debate:

Organicism: Organisms are the only sort of composite objects that exist. (Strictly speaking, there are no artifacts, no planets, no noses etc. though there are people, tigers and trees).⁵

Common Sense Ontology: There are ordinary objects “medium sized dry goods” and, perhaps, whatever science as it is currently practiced says there is (molecules, planets, tribes etc.), but no extraordinary objects (like the mereological sum of my nose and the Eiffel-tower).⁶

Universalism: for any plurality of (non-overlapping) objects, no matter how otherwise unrelated, there is an object (their fusion) they jointly compose. (Besides simples and ordinary objects, there are arbitrary mereological sums, like the sum of my nose and the Eiffel tower, and other sorts of extraordinary objects).⁷

Hirsch takes this debate to be verbal. Rather than an objection to Hirsch’s verbalness point, this paper is a study of its metaontological consequences. For this reason, rather than focusing on the details of Hirsch’s argumentation, I’ll begin by presenting four theses central to Hirsch’s semanticism:

• A. Some ontological disputes (COMP, for instance) are verbal disputes: On a charitable interpretation of the disputes in question, for any disputed sentence, S, between any two parties, one party takes S to mean one thing, S₁, another party takes S to mean another thing, S₂, and both parties can agree on the truth values of S₁ and S₂.

• B. Quantifier Variance (QV): There are several equally good ways to use ‘there is’, ‘there exists’, ‘there are’ and interconnected terms (‘something’, ‘object’ etc.); Among a certain variety of quantifiers, none is metaphysically privileged ([2002a], 2010, p.84); all are equally good for expressing ontological theory.

  Combining the two points (A-B): In the case of COMP, given that the parties agree on all perceptual facts, a charitable interpretation will associate a different quantifier with each view such that all parties can (consistently) affirm ‘there are tables’, if ‘there are’ means what some parties take it to mean, and deny that ‘there are tables’ if ‘there are’ means what other parties take it to mean. For instance: by replacing ‘there are’ once with the quantifier charitably associated with mereological-universalism and once with the quantifier associated with organicism (respectively) we reach this result.

• C. QV is compatible with realism: “the world and the things in it exist for the most part in complete independence of our knowledge and language. Our linguistic choices do not determine what exists, but determine what we are to mean by the words “what exists”” (p. 220). This idea should be read so as to shed light on the nature of QV. QV is not some sort of conventionalism, or ontological anti-realism. In Hirsch’s own words: “the basic idea of quantifier variance can be nicely formulated by saying that the same (unstructured) facts can be expressed using different concepts of “the existence of a thing”, that statements involving different kinds of quantifiers can be equally true by virtue of the same (unstructured)⁸ facts in the world” ([2002a], 2010, p.78).

• D. The verbalness of ontological disputes vindicates Common Sense Ontology. In general: as emphasized above, verbal disputes can sometimes be settled. This happens when one of the parties is charitably taken to employ the disputed terminology in accordance with its actual meaning. In the case of COMP: if the debate is carried out in English, the common sense ontologist ought to be taken to win it. ([2002b], 2010, ch.6)

Putting everything together: Hirsch takes ontological disputes like COMP to be verbal. According to Hirsch, the best way to explain the verbalness of disputes like COMP is by noting that each party takes the existential quantifier to mean something else; the way the quantifier is taken to be employed varies across parties, where no quantifier usage can be said to be more metaphysically (non)misleading than another.⁹ On this view, it is natural to think of organicists as if they are using a certain language that operates with a certain existential quantifier, ‘${\exists }_{\mathrm{org}}$’, of common sense ontologists as if they are using a different language with another quantifier, ‘${\exists }_{\mathrm{com}}$’, and of universalists as if they are using another language that operates with yet another quantifier, ‘${\exists }_{\mathrm{u}}$’. etc. (Call the languages ‘Organicese’, ‘Ordinarish’ and ‘Universalese’ respectively). Strictly speaking, all COMP parties speak English of course (just like Ami and Tami). But since they associate different meanings with at least one of the terms, viz. the quantifier in question, it would be convenient to think of their linguistic behaviour in terms of speaking different languages. (For convenience, I will allow myself to use the different languages locution in what follows). With this in hand, we can say that if we were to speak the language of one of the parties, so to speak, i.e. if we were to take the existential quantifier mean what one of the parties takes it to mean, then we could consistently affirm every quantified sentence that this party affirms, no matter which party we originally belong to. This would be similar to thinking of Ami as if he was using a language in which ‘cup’ means one thing, and of Tami as using a different language in which it means something else. And to say that if we were to speak Ami’s language (so to speak), such that ‘cup_Ami’ means ‘drinking vessel’, we could affirm ‘there are two cups_Ami on the table’ in the circumstances described in Ami/Tami.

Hirsch takes ontological disputes to concern the truth of the disputed quantified sentences (as their surface grammar often suggests). For this reason, he takes the verbalness of ontological disputes to ground their mere-verbalness. Taking himself to have shown that ontological disputes are, indeed, verbal, Hirsch holds that nothing substantial is left to disagree about. Particularly: nothing that involves deep theory-choice considerations we typically find in ontological disputes. If each standpoint turns out true in the language associated with it, the only question left open according to Hirsch is which of the views is true in the language we are in fact speaking; i.e. which of the views can be said to be true in English. This question can be dealt with by shallow conceptual-analysis considerations, leading to Common Sense Ontology and leaving the practice of ontology as we know it utterly pointless.

Suppose that Hirsch is right about the verbalness of ontological disputes like COMP. Will this be enough to ground semanticism, the view that ontological disputes are merely verbal, i.e. pointless in virtue of being verbal? That depends on how we ought to draw the line between verbal and merely verbal disputes. Before coming back to ontological disputes (Sects. 5–6), the following two Sects. (3–4) offer a closer look at the distinction between verbal disputes and merely verbal disputes.

3 Verbal vs. Merely Verbal

As briefly presented in the introduction, a merely verbal dispute is a dispute that is pointless or insubstantial due to being verbal. As opposed to the essentially semantic character of ‘verbal dispute’, ‘merely verbal’ involves normative and pragmatic components. To determine whether a dispute is merely verbal on top of being verbal, we need to add some pragmatic condition like: ‘given the disputants interests and their rationality, the dispute ought to evaporate’ (or something to that effect).¹⁰

Given this extra condition, not all verbal disputes are merely verbal. For instance, when words are the primary domain of concern (e.g. in linguistics) a dispute that arises wholly in virtue of a disagreement about the meaning of a term can be considered worthwhile, i.e. not merely verbal. To use the Ami/Tami case again: If Ami and Tami care mainly about the number of drinking vessels on the table and have a verbal dispute about whether there are two cups on the table, the dispute will most likely evaporate as soon as they both notice that it’s verbal. But if Ami and Tami are linguists that care mainly about the meaning of ‘cup’ in English, they’re not likely to (nor ought they) abandon the dispute over how many cups there are on the table upon noticing the verbalness of their dispute.

Verbal disputes that are not merely verbal are not specific to linguistics and neighbouring fields (philosophy of language, literary criticism etc.) They’re everywhere: think of cases in which a term is used to discuss the prevailing standards (Barker, 2002): e.g. when a dispute about whether Feynman is tall is not about information concerning Feynman’s height but about the applicability of the predicate ‘tall’. E.g., in a context in which Feynman’s height is common knowledge and ‘Feynman is tall’ serves to communicate that ‘in this country, someone as tall as Feynman is considered tall’. For another sort of cases, think of Chalmers’ (2011, p.517) examples for verbal but not merely verbal disputes in ethics. According to Chalmers: “disputes…about what falls under the extension of ‘marriage’ and ‘murder’ [and ‘torture’ and ‘terrorism’] may in some sense be verbal, but the answer to these questions may make a serious difference for people’s lives.” Consider the following example from Sundell (2011, p.12) for instance:

Torture: Two speakers have a dispute about whether waterboarding is torture or not. It’s very clear that each of them applies different standards to ‘torture’ (e.g. American vs. UN law standards), and that this is what gave rise to the dispute in the first place. Both speakers are familiar with the relevant facts about waterboarding (what’s actually involved, the side-effects, etc.)

While Torture meets Verbal (Hirsch’s sufficient condition for verbal disputes discussed earlier), it does not seem merely verbal.

These examples are all fairly intuitive. But how, in practice, can we tell whether a certain verbal dispute (a particular linguistic exchange) is merely verbal or not? Chalmers characterizes merely verbal disputes as verbal disputes in which words don’t matter: “It often happens,” he writes, “that we are concerned with a first-order domain, not with the usage of words, and in such a way that nothing crucial to the domain turns on the usage of words. In this case, a verbal dispute is a mere verbal dispute.”

I find this demarcation unsatisfying. First, the sense in which something crucial can really turn on the usage of words in verbal (but not merely verbal) disputes about norms, for example, is not clear. Disputes in ethics are significantly different from disputes in linguistics or literary criticism in this respect. So, for example, while there is a sense in which what falls under the extension of ‘torture’ does seem to matter (in the verbal debate about waterboarding considered above), it seems misleading to say that the practical importance of this debate turns primarily on the meaning of ‘torture’, or on how this word happens to be used.¹¹ Suppose that some sophisticated linguist equipped with an unfamiliar, yet well-established, metasemantics used empirical evidence about the genealogy of the word ‘torture’ or about competent speakers’ reactions to a certain test, to show that pace what we may have thought, ‘torture’ is not morally loaded, and its extension includes certain practices but fails to include waterboarding. Although findings of this sort may be highly illuminating in the context of a dispute in linguistics, in the context of a dispute about rights and the limits of state authority, they are not. Disputants in this latter dispute do not primarily care about the meaning, or even the extension of the word ‘torture’ (especially in case it turns out to lack an ethically loaded aspect); their dispute about ‘torture’ is a means for expressing a moral disagreement about norms. In light of unanticipated findings about meaning, the disputants in Torture are likely to rephrase their disagreement, not to take it to be settled. Since we would not want to classify the verbal dispute in question as merely verbal, and since it’s unlikely that anything crucial hangs primarily on meaning in this context, Chalmers’s proposal calls for refinement.

Secondly, it would be more satisfying if by way of characterizing what makes some verbal disputes merely verbal, we would thereby be explaining their pointlessness (and articulating the particular way in which they are pointless). Such an account would also explain how verbal disputes can nevertheless make a difference to people’s lives, in case they are not merely verbal. Chalmers’s criterion (while good enough for his own interest) cannot give us that, as it does not go beyond ‘nothing turns on it’ terminology.

To make some progress, consider Sundell’s Analysis of Torture (ibid.):

Even if we suppose that the speakers mean different things by the word torture, it is clear that we have not exhausted the evaluative work to be done. […] By employing the word torture in a way that excludes waterboarding, [one of] the speaker[s] communicates (though not via literal expression) the proposition that such a usage is appropriate to those moral or legal discussions… that waterboarding itself is, in the relevant sense, unproblematic” [my emphasis]

It would be misleading to describe Sundell as taking this dispute to be not merely verbal in virtue of anything hanging on the meaning of the word ‘torture’. Primarily, he takes it to be not merely verbal because it reflects a genuine disagreement; in this case, one that does not concern words but norms.

In line with Sundell, I take disputes which communicate genuine disagreements not to be merely verbal no matter what the domain actually is, or how much practical import it has. To use a Barker-style example again: by using ‘expensive’ differently, two speakers may be saying the truth (relative to the relevant counterfactual language) both with ‘these glasses are expensive’ and with ‘these glasses are not expensive’ respectively. Given that the glasses’ price is common knowledge in this context, the norms of interpretation approved by Hirsch will presumably require us to take the dispute they engage in to be verbal. But given that the speakers are using ‘these glasses are expensive’ to communicate a genuine disagreement over the prevailing standards for ‘expensive’, (if it’s a seller and a buyer, this can serve to negotiate the price of the glasses), it would be wrong to take their dispute to be merely verbal.

The fact that the dispute over ‘expensive’ is not merely verbal can be explained by its being centered on a disagreement about the prevailing standards (in the market). Likewise, the fact that the verbal disputes about the extension of ‘torture’ ‘marriage’ etc. have practical implications, can be explained by their being centered on (/communicating) a disagreement about norms. It is in virtue of communicating a genuine disagreement that some verbal disputes are not merely verbal. Merely verbal dispute are pointless, they (rationally) ought to evaporate, because they fail to communicate a disagreement about the subject matter addressed (be it words, in the case of linguistics, or norms, in the case of ethics).

4 Communicating a Genuine Disagreement

Following Plunkett and Sundell (2013), I think that for a verbal dispute not to meet the necessary conditions for being merely verbal, is for it to communicate a genuine disagreement.¹² To give a better idea of what communicating a genuine disagreement means here, let me say a bit more about ‘disagreement’ and about ‘communicating’. We can use Plunkett’s and Sundell’s definition for ‘disagreement’, in which disagreement is cashed out in terms of rationally incompatible demands:

If two subjects A and B disagree with one another, there are two propositions (/desires /plans etc.) p and q such that A accept p and B accepts q, and p is such that the demands placed on a subject in virtue of accepting it are rationally incompatible with the demands placed on a subject in virtue of accepting q (though not necessarily in virtue of p entailing not-q) (2013, p.18-19).

For an account of ‘communicate’ that will serve our purposes, we can use a well-entrenched distinction between two roles of context. On this distinction, as well as the same sentence expressing different propositions in different contexts (e.g. ‘I am Josef’ uttered by different speakers), the same proposition can perform different things (have different effects) in different contexts. (A proposition to the effect that you’re a genius can be used to communicate ‘you’re an idiot’ in some contexts). According to Stanley and Szabo (2000) we can distinguish the semantic role of context: taking part in fixing what is said (/which proposition is expressed) given what is uttered, and the pragmatic role of context: fixing what is communicated given what is said. We can use this terminology for convenience.¹³

In most disputes, disagreements are literally expressed: the propositions interlocutors express in most disputes are strictly incompatible. But sometimes noticing the disagreement involves a more comprehensive reflection on context: accounting for the disputants’ interests in conducting their dispute, the sort of evidence they consider relevant etc. Based on acquaintance with such factors, we sometimes have independent reasons for taking a dispute to generate a genuine disagreement. In those cases, discovering its verbalness would be pragmatically harmless: it would merely shift its center to what is communicated, and make any interpretation that takes it to revolve around its literal content inadequate. Consider the following case:

Healthy: Two military doctors are trying to decide whether Noa is physically qualified for the unit they serve. One of them utters ‘Noa is healthy’ the other utters ‘Noa is not healthy’. As it turns out, they do not disagree on anything regarding Noa’s physical condition; they only have different standards for ‘healthy’. Despite that, they conduct their dispute in the terms described.

Is this dispute merely verbal or not? Well, it depends. If it concerns Noa’s physical data (as its literal content may suggest), then it is. But here’s a case in which it definitely is not: suppose that despite appearances, the main issue for the doctors does not concern Noa’s physical condition, but whether someone in such-and-such conditions (they commonly-know Noa to possess) is physically qualified for the unit. Like the disputants in Torture, the disputants in Healthy can, in that case, rationally go on and use the allegedly problematic terminology. They can use ‘Noa is (not/) healthy’ despite being aware that each side uses language in a way that makes its utterances come out true in their language (so to speak).

To judge whether the actual dispute is merely verbal or not requires some aquaintance with the doctors’ considerations and the sort of evidence they take to be relevant. If the sort of evidence they consider indicates that they’re trying to describe Noa’s condition, unaware of the fact that they merely have different standards for ‘healthy’, the dispute would be merely verbal (given what is being addressed and the context). If the evidence brought up in discussion concerns the capability of someone in such-and-such condition to survive the unit’s trainings, an adequate interpretation would take that to be communicated by their utterances. In that case, acknowledging the verbalness of their dispute, one ought not to take it to be about which of their utterances truly picks out Noa’s actual physical condition (pace literal content), but about whether someone in Noa’s physical conditions is qualified for the unit. The general lesson here is that in case of a genuine disagreement, discovering the verbalness of the linguistic exchange merely advises against taking its surface grammar to tell us the whole story. The rest of the paper goes back to COMP, to see how this general lesson can be applied to ontological disputes.

5 Ontological Commitments

Hirsch takes ontological disputes like COMP to be verbal. On a charitable interpretation, each party takes the existential quantifier to mean something else. With respect to any party, if we were to use the quantifier like they do (speak their language, as it were), we would agree that they speak the truth. We can consistently accept both ‘there are_org tables’ and ‘there are_u no tables’, for instance. The disputants themselves, can, in any given circumstances, agree on all perceptual facts, and on the way matter is distributed in space. When looking at my room, for example, both Organicists and Universalist can consistently affirm: ‘there are_org simples located here-and-there (arranged table-wise), but there is_org no table’ and ‘there are_u simples located here-and-there (arranged table-wise) and there is_u also an object_u, the table, that they jointly compose’. This can be extended to theories purporting to describe reality as a whole.

Organicism could read:

There are_org simples located at [l_t1…l_tn] (arranged table-wise), and there are_org simples located at [l_c1…l_cn] (arranged chairwise), and there are_org simples located at [l_e1…l_en] (arranged elephant-wise)…[‘…’ specifies all simples’, possibly by 4D locations and says they exist_org] and there is_org an elephant, located at [l_e1…l_en] and… [‘…’ specifies all particular organisms and says they exist_org], and that is all there is_org.

Universalism could read:

There are_u simples located at [l_t1…l_tn] (arranged table-wise), and there are_u simples located at [l_c1…l_cn] (arranged chairwise)… [‘…’ specifies all simples’ and says they exist_u], and for every (numerically distinct) plurality of simples, there is_u an object_u they jointly compose: There is_u an elephant, located at [l_e1…l_en], There is_u a table located at [l_t1…l_tn], there is_u a chair located at [l_c1…l_cn], there is_u an object composed of a table and a chair located at [l_t1…l_tn, l_c1…l_cn]…[‘…’ exhausting all combinations of simples], and that is all there is_u.

To Hirsch, these two theories sound like equally good descriptions of the very same facts. They are true in the same circumstances. Qua descriptions of reality as a whole, as they are spelled out above, Organicism and Universalism are, in Hirsch’s view, truth conditionally equivalent; their truth depends on what simples exist (and the way they are distributed in space), which is something all parties can agree on. Hirsch takes this to ground not just the verbalness of the dispute, but its mere verbalness.

If Hirsch is right, ontologists in the Quinean tradition are deeply mistaken about the semantics of the theories they stand up for in ontological disputes.¹⁴ While they take their theories to have diverse truth conditions, Hirsch takes them to be “truth-conditionally equivalent” (2010, p.xii).¹⁵ We can be more specific. To use Augustine Rayo’s (2007) characterization of ontological commitments, ontological commitments are an aspect of truth conditions: “To describe a sentence’s ontological commitments is to describe some of the demands that the sentence’s truth imposes on the world—those demands that concern ontology”. With this in mind, we can say that according to Hirsch: while ontologists take their theories to involve conflicting ontological commitments, they actually do not. This is important. Using Rayo’s characterization, we can focus our interest to this aspect of QV, i.e. to its implying that theories involving apparent conflict about quantified sentences (some theories entailing the falsity of quantified sentences structured like the ones affirmed by others) can in fact share ontological commitments.¹⁶

Rayo’s characterization is very general; it tells us nothing about how “the ontological demands posed by the truth of a sentence” are to actually be read off of sentences. Different accounts of ontological commitment may differ substantially on this point. For instance, according to Quine, a first-order sentence carries commitment to Fs iff Fs must be counted amongst the values of the variables in order for the sentence to be true. But a different account (which equally conforms to Rayo’s characterisation) may take first-order sentences to have demands of a completely different type. (It may take the truth of ‘there are Fs’ to impose the existence of F-ness/ merely simples-arranged-F-wise/…).

One thing to note about QV is that once formulated in terms of ontological commitments, the difference in semantics of ontological views must involve differences regarding how the ontological commitments of quantified sentences ought to be read off. Applying Rayo’s general notion, according to QV, different theories in the context of COMP have the same ontological commitments. But clearly these views differ on which quantified sentences they affirm and deny (in their language). If all views are true relative to the language they are associated with, at least some quantified sentences are true without being ontologically committed to the values of bound variables. (Or else: some theories implying correlating negated quantified sentences are true without being uncommitted to the values of relevant bound variables). What this means is that a non-Quinean notion of ontological commitment is in play.

To demonstrate: Take any sentence, ‘there are Ts’, such that its Organicist counterpart (the relevant structurally identical sentence, ‘there are_org Ts’) is false, whereas its universalist counterpart (‘there are_u Ts’), is true. (Ts can be tables, for that matter). Now, if we followed Quine and supposed that Organicism (as a whole) is not committed to Ts (in virtue of Organicism entailing $\sim {\exists }_{\mathrm{org}}x(Tx)$), any competing COMP-theory (as a whole) will also not be committed to Ts (since, according to Hirsch, the theories are truth-conditionally equivalent, and ontological commitment is an aspect of truth conditions). If a positive existential sentence, ‘there are_u Ts’, is integral to that competing theory (in the way that ‘${\exists }_{\mathrm{u}}x(Tx)$’ is indeed integral to Universalism) its truth conditions must not include the demand that “Ts must be counted amongst the values of the variables in order for the sentence to be true.” Otherwise, the theories in question (Universalism and Organicism, for that matter) would not be truth-conditionally-equivalent. So when saying that the theories in question “have the same truth conditions” or more specifically, that they share ontological commitments, a non-Quinean notion of ontological commitment must be involved in the specification of at least one of the theories’ truth conditions. This means that ontologists are deeply mistaken about the semantics of (at least some) COMP-theories.

An analogy can perhaps give a better grasp of where we stand. Suppose that in Healthy, doctor-1 thinks that anyone with a score of 80 (or more) is qualified for the unit, and doctor-2 thinks only those with a score of 90 (or more) are qualified, where Noa’s score is commonly known to be 85. Suppose that a third party explains to the doctors that their dispute meets Verbal, by saying: “both parties can equally affirm that ‘Noa is healthy₁’ and ‘Noa is not healthy₂’, (where ‘healthy₁’ is the meaning Doctor-1 associates with ‘healthy’ and ‘healthy₂’ is the meaning Doctor-2 associates with ‘healthy’ in this context). Moreover, a more comprehensive articulations of the situation implies truth-conditional equivalence: S₁: ‘Noa’s score is 5 points above the bar for healthy₁’, and S₂: ‘Noa’s score is 5 points below the bar for healthy₂’, have the same truth conditions.” The doctors originally thought that being above the bar for healthy comes down to being physically fit for the unit. Hence, if S₁ and S₂ have the same truth conditions (and thereby impose the same practical requirements), at least one of those sentences fails to imply what the doctors commonly took it to imply (that Noa is fit/unfit for the unit, respectively).

As stated in the previous section, everything said so far suggests that both with respect to COMP and with respect to this extended version of Healthy, uncovering the disagreement requires us to go beyond literal content. If what I said so far is on the right track, there’s something awkward about Hirsch’s analysis of COMP. On the view advanced here, the right way to go is “bottom-up”: noticing the verbalness of a dispute, we start from a careful reflection on the evidence considered by its participants and see if in the context it’s raised it generates a genuine disagreement reasonably associated with their interests. If it does, the literal content disguises the real issue rather than expressing it, and the semantic character of the dispute (viz. the verbalness of the dispute) is a reason to focus on what is communicated. Instead of that, Hirsch fixes his interpretation of the dispute to concern literal content (whether certain quantified sentences are true), and then goes on to argue that, given its verbalness, “there can’t be anything deep or theoretical [about it]” ([2002a], 2010, p.90).

On a reasonable interpretation: judging by the sort of theory-choice considerations we typically find in ontology, it is the ontological commitments of our best theory that ontologists take themselves to disagree about. Their use of quantified sentences is subordinate to expressing that disagreement. As we have seen, if Hirsch is right about the verbalness point, what is literally expressed by the disputants will not be able to capture this genuine disagreement. But rather than a sign that there is no real disagreement, it is evidence against a literal-content-centered interpretation of the disputes. The next section (last before concluding) further elaborates on this point and examines ramifications for semanticism.

6 Genuine Ontological Disagreements

Suppose that different composition theories are, as Hirsch holds, truth-conditionally equivalent. In that case, as we have seen, the truth of some of these theories is going to have nontrivial ontological demands. In the sense discussed earlier: the ontological commitments of some of them will not be read off directly from their logical structure in Quinean fashion.

There’s more than one framework in which formulating non-Quinean accounts of ontological commitments is possible. Accordingly, there’s more than one way to complete the formulation of the idea that the ontological theories (addressed by Hirsch) express the same ontological commitments, while varying in logical structure.¹⁷ In any event, if we are to accommodate QV, we must be able to make the difference between two sorts of (first order) quantified sentences: quantified sentences that are ontologically committed to what ontologists commonly take them to (i.e. the values of their bound variables) and quantified sentences that are not.

With this distinction in mind, so long as we remain in the realm of semantics (in the sense of ‘fixing what is said’), we can dialectically give Hirsch everything he wanted. We can go along with the assumption that COMP meets Verbal, the sufficient condition for being verbal: we can take all standpoints in ontological disputes to be literally true (in their language) and we can take COMP-theories concerning reality as a whole to be truth-conditionally equivalent; We can require of an adequate interpretation to take Common Sense Ontology to be true in English without that being in any tension with the assertion that Organicism and Universalism are equally true relative to their languages (Organicese and Universalese, respectively); We can hold revisionary ontology, the idea that “many common sense judgments about the existence of highly visible physical objects are a priori necessarily false” (Hirsch, [2002b], 2010, p.101), to be wrongheaded and absurd.¹⁸

However, associating different semantics with different theories will prevent us from taking all theories to be ontologically committed to what ontologists commonly take them to (nothing guarantees that Common Sense Ontology is ontologically committed to values of its bound variables, for instance). As we have seen, conflicting rational demands, in this case: demands generated by conflicting ontological commitments, are enough to establish a genuine disagreement. If a disagreement fails to be literally expressed, judging whether the dispute is pointless requires the disclosure of what is communicated.

In the context of (extended) Healthy, reflecting on the doctors’ considerations, we were compelled to take their dispute over ‘Noa is healthy’ not to be about whether that sentence truly picks out Noa’s physical conditions (despite what is suggested by its literal content); but rather about whether the practical requirements that the doctors collectively associate with it are the ones that ought to be embraced (i.e. whether someone with the physical conditions they commonly attribute to Noa is physically fit for being recruited to their unit). In analogy, in case a dispute over ‘$\exists xTx$’ turns out verbal, given typical considerations raised by ontologists, we ought not to take it to be about whether ‘$\exists xTx$’ truly describes the world; but rather about whether the ontological commitments ontologists collectively associate with it are the demands (having to do with ontology) posed by the best theory (and truth-conditional equivalents of it). Ontologists collectively associate different ontological commitments with different COMP-views; we ought to take them to communicate a genuine disagreement by expressing those views. If their dispute is verbal, ontologists indeed fail to literally express the disagreement they care about. But that’s no reason to dismiss it.

As long as considerations raised by COMP theorists are not decisive, what the best theory (and truth-conditional equivalents of it) is ontologically committed to remains a question. As Williams (2010, p.106–7) correctly notes, that’s precisely the sort of question ontologists typically care about. In his words: “for the purposes of metaphysics, we are not just interested in what the true sentences or propositions are: we are interested in the way reality is, in the objects and properties and their arrangements that support the truth of propositions”. If we can’t tell what the theories are ontologically committed to, knowing which composition theories are true (in English, e.g.) is not going to be enough to exhaust the ontological inquiry into that matter. Suppose that the parties discover that COMP theories are true (each relative to some language), and that they are truth conditionally equivalent, so there can be no more than one theory that actually has the truth conditions ontologists commonly associate with it. Suppose that at some later stage it turns out that the only theory that has truth conditions ontologists associate with is Organicism. In such a case, the other parties will take themselves to have lost the disagreement. Yet prior to finding out that Organicism is not only true (relative to a language), but ontologically committed exactly to the values of its bound variables, i.e. that as opposed to other theories, what ontologists took it to say is in fact the case, the disagreement is not settled. Hence, proving all COMP theories to be true leaves the real disagreement unsettled.

Recapitulating: Knowing that ontologists (in the Quinean tradition) associate ontological commitments with values of variables, an adequate interpretation would take their views to disagree over that which “supports the truth of propositions”; What Hirsch, qua realist, calls the “…basic notion of “reality”, “the world”, “the way it is”” that “remain[s] invariant through any changes in the concept of “the things that exist”” ([2002a], 2010, p.79). Given the disagreement they take themselves to be expressing (and the conflicting rational demands, in this respect), “all theories in question are true” falls short of semanticism. Shallow considerations, like charity, may be taken to guide us towards the truth, but such considerations fall short of telling us what the truth ontologically requires. On an adequate interpretation, the considerations ontologists actually raise (overall simplicity, non-arbitrariness, etc.) ought to be taken to generate a genuine disagreement precisely about that.¹⁹ If QV is, indeed, correct, reflecting on the sort of considerations taken by ontologists, rather than taking each side to imply that their theory is uniquely true (i.e. the falsity of the other theories), we ought to take it to communicate, in this context, a commitment to a particular ontology, and the disapproval of ontological commitments commonly associated with each of the other theories, whether or not they are part of the truth conditions they actually have.²⁰

7 Conclusion

Hirsch’s main argument for semanticism was:

• (1) The sort of questions ontological disputes address concern the truth of certain quantified sentences.

• (2) According to the correct norms of linguistic interpretation, in some disputes each side ought to agree that the quantified sentences (/negated quantified sentences) used by the other parties are true. Hence:

• (3) In those disputes, there’s nothing substantial left for the parties to disagree about. The dispute is pointless.

For some reason, standard critical accounts address (2), and plea: “not verbal”.²¹ For some reason, semanticists believe that proving the disputes not to be verbal (arguing against (2)) is the single hope for non-semanticists. Hirsch, for example, takes it to be the only real threat to his view ([2005], 2010, p.170). At bottom, I think this has to do with the preconception that in ontology, any gap between verbal and merely verbal is idle. In ontology, it is widely assumed (for good reasons), words don’t matter. But only a bad demarcation between verbal and merely verbal can make that count for anything.

This paper focused mainly on assumption (1), or more precisely on the relationship between (1) and (2). As I argued, given the verbalness of the disputes, alternative descriptions of the practice of ontology become strikingly more plausible. In light of that, if Hirsch is right about (2), rather than agreeing with him on (3), one ought to discard (1).

Given the main argument, this reaction to semanticism seems stable and general. Instead of quibbling with semanticists over whether ontological disputes are verbal, it takes its cue from a close attention to what ontologists are up to. Given that ontologists are trying to express views with conflicting commitments, either ontological disputes are non-verbal and they concern what the views express, or else the disputes are verbal, but they concern what the views communicate and not what they express. On both horns, a genuine disagreement is conveyed. The open question is a semantic question about how this actually gets done.