“Well, That Depends on What You Mean by 'Φ'”

“Well, that depends on what you mean by ‘φ’”

On Substantive and Merely Verbal Disputes in Philosophy

Ludvig F. Fuglestvedt

Thesis presented for the degree of Master of Philosophy

Supervised by Professor of Philosophy Herman Cappelen (University of Oslo, University of St. Andrews) and Willis S. and Marion Slusser Professor Emeritus of Philosophy Barry Stroud (University of California, Berkeley)

Department of Philosophy, Classics, History of Art and Ideas, University of Oslo, 2017

©Ludvig F. Fuglestvedt, 2017

“Well, that depends on what you mean by ‘Ф’” On Substantive and Merely Verbal Disputes in Philosophy

Ludvig F. Fuglestvedt http://www.duo.uio.no

Artwork: Microsoft Office Word™ shapes

“Well, that depends on what you mean by ‘Ф’”

On Substantive and Merely Verbal Disputes in Philosophy

Abstract Are philosophers just talking past each other? Are they only arguing about what words mean? Is it all pointless? I often get that feeling, and many people on the street share that idea about what philosophers are up to, and there are also several philosophers who have placed related, deflationary, verdicts upon philosophy, that it is the mere analysis of language, such as members of 1930s’ Vienna Circle including Rudolf Carnap and A.J. Ayer, as well as some philosophers at the University of Oxford in the 1950s inspired by the ideas of Ludwig Wittgenstein. The problems tended to concern whether philosophical questions really are nonsensical, linguistic constructions with the mere illusion of depth. Related to that is the more pressing problem of whether philosophical disputes are merely verbal, pointless disagreements over language. This thesis in philosophy of language, or philosophical methodology, explores the problem of verbal disputes: that there may be many of them—or are there?—and that they are pointless—or are they? It starts by motivating the idea with a focus on the way it is presented in the philosophical paper “Verbal Disputes” by David Chalmers (2011) (chapter1). Then some arguments for and against the idea that philosophy is dominated by verbal disputes are discussed along with the idea that philosophical questions are questions of language and in what way mere words should matter to philosophers (chapter 2). Lastly, a response to the problem is discussed and dismissed. It says that philosophical disputes can be understood as disputes that are verbal but substantive nevertheless, what David Plunkett (2013, 2014, 2016) and others call “metalinguistic negotiations” (chapter 3). I attempt to maximize the magnitude of the arguments in “Verbal Disputes”, and offer a less diplomatic reading by which is a radical paper. I conclude that Plunkett’s theory in “Which Concepts Should we Use? Metalinguistic Negotiations and the Methodology of Philosophy” cannot provide a solution to the problems it poses. The thesis first and foremost renders the two authors’ statements, but aims to place the tension between them within a wider philosophical context. The author’s perspective that plays this part in this is made clear by such as “I think,” and “…it seems reasonable to suppose,” and contributions coming from me are otherwise confined to the more speculative, less analytical ends of each chapter, or opinionated conclusions. Typical analytical tools like the use/mention distinction and stipulative definitions are used, and when literature is surveyed and compared their terminology are merged, and textual references are given to primary sources. Footnotes contain helpful information, credit, and further reading. References are in the text and the literature list contains the full sources as well as works on which rely the provided interpretations.

iii

Acknowledgements The thought of being trapped within one’s own language, philosophizing privately in a system of ideas that makes sense only to oneself, isn’t it the worst? The esteemed fellow students and academic acquaintances that have helped me in this project have made sure that this dissertation is not a verbal dispute, and I am only the mediator between their many interesting perspectives, although they might not agree with the interpretations and arguments provided in this thesis. For reading and discussion, I am thankful to Conrad Bakka, Mark-Oliver Casper, Jørgen Dyrsdad, Patrick J. Winther-Larsen, Erlend Finke Owesen, and Hilde Vinje. And also to Derek Ball, Hannah Ginsborg, David Plunkett, Shamik Dasgupta, and Mons Andreas Nyquist for illuminating discussions, and special thanks for writing help to Herman Cappelen and Barry Stroud. Thanks also for funding from Centre for the Study of Mind in Nature (CSMN) and from Concept Lab at Instutitt for filosofi, ide-, og kunsthistorie og klassiske språk, Universitetet i Oslo. I became interested in this topic seeing a talk by Matti Eklund titled “Philosophy as Conceptual Engineering” at the Nordphil conference at the University of Oslo in 2015, and developed the interest as a result of a very engaging seminar the same year at the University of Uppsala, “Metaontology: Themes from Carnap and Quine.” L.F.F.

Table of Contents

Abstract ...... i Acknowledgements ...... iii Table of Contents ...... iv Introduction ...... vi I. What is a verbal dispute? ...... vi II. Is there a problem of verbal disputes in philosophy? ...... viii III. Substantive verbal disputes ...... xi IV. “Philosophy as Conceptual Engineering” ...... xiv V. Thesis statement: If philosophy is verbal, it is verbal in the bad way ...... xvi VI. Disposition ...... xvii

Chapter one—David Chalmers on verbal disputes

1.1. Verbal disputes defined ...... 2 1.2. Substantivity and pointlessness ...... 9 1.3. Can there be broadly verbal disputes? ...... 11 1.4. A heuristics oriented theory ...... 17 1.5. The method applied to philosophical disputes ...... 24 1.6. The subscript gambit...... 30 1.7. “Answers are answers” ...... 34 1.8. Which disputes are verbal? ...... 37 1.9. Conclusion ...... 43

Chapter two—Against the idea of theoretically conflicting conceptual analyses

2.1. Conceptual Analysis ...... 47 2.2. Real Definition ...... 51 2.3. Questions of language and questions of reality ...... 56 A summary of the argument ...... 58 Externalism and natural kinds—Are questions about language questions about reality? ...... 59 Linguistic evidence, contextualism, and transcendental arguments— Are questions about reality questions about language? ...... 61 Conclusion ...... 64

2.4. Was Humpty Dumpty right? ...... 64 Frank Jackson and the Canberra Plan ...... 65 Against Jackson: Are philosophers disagreeing about what it is they are disagreeing about? ...... 67 Theoretical background: Chalmers’ Conceptual Pluralism ...... 69 Why bother about what something means? ...... 70 2.5 Conclusion ...... 73

Chapter three—Normative philosophy of language to the rescue?

3.1. David Plunkett on metalinguistic negotiations ...... 77 3.2. Metalinguistic Negotiations in Philosophy ...... 82 3.3. Against the idea of substantive verbal disputes ...... 84 Reasons for revising concepts ...... 85 Substantive in virtue of which types of reasons? ...... 87 Epistemic and practical reasons ...... 88 3.4. The Is of reality as the Ought of language ...... 89 First response: Are philosophical questions ethical questions? ...... 90 Second response: Are philosophers negotiating frameworks? ...... 91 Conclusion ...... 96 3.5. Broader context: Chalmers’ theory compared to Plunkett’s theory ...... 96

Conclusion ...... 100 Literature ...... 104 Index ...... 112

Introduction ”Depends on what you mean by ‘F,’” when uttered in response to some question or claim where “F” figures as a central term, is so much a natural response that it’s become a philosophical cliché. But to avoid it is to neglect the longstanding wisdom that clarification of central terms and concepts should be in place before inquiry can begin. Any imprecision, vagueness or conflation inherent in the original concepts will otherwise be transferred to the questions we ask using them, as well as to the consequent philosophical disputes. I want to make a case for this neat but no less useful dialectical tool. A common thread in what follows is that the “depends on what you mean by…” response should be elevated from cliché to routine. In eager pursuit of truth, we all too often leave it idle, and embark instead under careless faith that concepts and terms will work themselves out as we go along. But that’s only the lucky outcome; the likely one is that the pursuit will keep going on end, without getting anywhere, owing to linguistic misunderstandings, congenial and undetected. In such cases the disputants would unknowingly be engaged in what’s called a verbal dispute. This introduction works as an overture of a general line of reasoning that the dissertation aims to comment on. After a short sketch of what a verbal dispute is and why verbal disputes seems to entail pointlessness (I), some reasons are provided for why philosophy can be suspected to suffer from it (II), before the idea that verbal disputes can be substantive is introduced (III), an idea which is placed into the context of the literature on “philosophy as conceptual engineering” whereby philosophy can be both verbal and yet substantive (i.e. the opposite of pointless) (IV), before my reasons to doubt this are sketched (V), and the structure of the ensuing dissertation is presented (VI).

I. What is a verbal dispute?

The paradigm example of a verbal dispute is Karen Bennet’s martini case:

[Metaphysical disputes] are more like the dispute between the purist who says that only cocktails made of gin or vodka, dry vermouth, and perhaps an olive or two count as martinis, and the sorority girl who calls practically anything a martini as long as it served in the classic V- shaped glass. If these two are seated at a table on which such a glass contains some nonsense made of sour green apple liqueur, the latter will say that there is a martini there, and the former will deny it. This is a paradigm case of a verbal dispute. The disputants agree on all the facts, but disagree on how to use the word ‘martini’. (“Composition, Colocation, and Metaontology.” (2007, 13))

Referred to colloquially as “just semantics,” “talking at cross purposes,” and “arguing over mere terminological issues,” verbal disputes occur when, typically unbeknownst to the disputants, the disagreement centres on the mere meanings of words, for the simple reason that they are using their words differently or have different beliefs about what they mean. In so far as a divergence in meanings amounts to real disagreement at all, it’s only over what to call things. The plausible consequence is that, in so far as a dispute is verbal, the disputants

(1) each address different topics, and so (2) disagree on nothing substantial.

vii

And so, as verbal disputes are commonly attributed properties (1) and (2), they have a name for being pointless; if you settle the meanings, you settle the dispute at large. This is not so if the dispute is not verbal. In disputing the origin of life, the effect of trade policy on economic growth, manmade climate change, water on the moon, weapons of mass destruction in Iraq, vaccines causing autism, the earth being flat, the existence of a monster in Loch Ness, or whether the Holocaust happened, settling meanings won’t settle much more as far as the dispute goes. Two bird spotters might agree entirely on the classification scheme of birds, but still disagree over whether the sound they heard was a magpie or a jay. Put otherwise, the meanings are already settled, or sufficiently so for substantive disagreement to be present. (They are also known as factual disputes, but we’ll stick to calling them not-verbal for the reason that it sounds somewhat odd to talk of “philosophical facts.”) Verbal disputes are a problem because they are pointless. The problem of verbal disputes is that disagreeing about meanings and using words differently appears to be a waste, so if we are doing that then we are not doing what we should do; dispute substantively. Are insignificant disagreements being invested in, without any epistemic output? The idea of verbal disputes in various certain types of philosophical debates is often attributed to Rudolf Carnap and his interwar period Vienna Circle, and defended in recent times in Hirsch (2005), Chalmers (2011), and Jenkins (2014), and it is often held to explain the apparent lack of progress or consensus over grand philosophical questions that have been disputed for centuries. As in David Hume at the start of “Of Liberty and Necessity” where he considers whether the old question “Do humans have free will?” involves cross purpose talk (1) because the disputants haven’t properly defined their terms, “affixed them the same ideas,” and how the mere topic of words is not substantive with respect to “the true and real subject,” which is the freedom of the will (2):

It might reasonably be expected in questions which have been canvassed and disputed with great eagerness, since the first origin of science, and philosophy, that the meaning of all the terms, at least, should have been agreed upon among the disputants; and our enquiries, in the course of two thousand years, been able to pass from words to the true and real subject of the controversy. For how easy may it seem to give exact definitions of the terms employed in reasoning, and make these definitions, not the mere sound of words, the object of future scrutiny and examination? But if we consider the matter more narrowly, we shall be apt to draw a quite opposite conclusion. From this circumstance alone, that a controversy has been long kept on foot, and remains still undecided, we may presume that there is some ambiguity in the expression, and that the disputants affix different ideas to the terms employed in the controversy. (David Hume, An Enquiry Concerning Human Understanding: VIII: Of liberty and necessity, 62-63.)

Besides being an old idea, verbal disputes are also occurring all around us, from cultural and political arguments to everyday nothings. But most of the time we have the acumen to avoid them:

Is this dress pretty? Well, that depends on what you mean by pretty—pretty to look at, or fitting to the wearer? Is Torfinn Huvenes tall? Well, it depends on what you mean by tall—tall for an academic, or tall for a basketball player? Is Herman a good supervisor? Well, that depends on what you mean by good—knowledgeable, or compassionate? Is chess a sport? Well, that depends on how you define sport—in terms of physical athleticism and dexterity or just as any competitive rule governed activity? Is Aronian or Carlsen the better chess player? Well, it depends on what you mean by best—most wins or sophistication? Which literary work is greater; 1984 or Brave New World? Well, it depends on what you attribute the most

viii

significance when speaking of greatness—being influential as against, let's say, pioneering? Can you brew espresso in a capsule machine? Well, it depends on whether you use espresso in the sense preferred by the marketers or by the enthusiasts. Is it true that the banana is a berry, the tomato a fruit, and the coffee bean a seed? Well, it depends on how you use the words—by the ordinary or botanical taxonomy? Is the Oslofjord really a fjord? Well, it depends on what you mean by fjord—the geological or the topographical sense? Which is the highest mountain on earth? Well, it depends on what you mean by height—there are different answers depending on whether you measure by sea level (dry) prominence, wet prominence, or prominence from the earth’s centre. Is Mount Elbrus or Mont Blanc the highest peak in Europe? Well, the European continent or Europe? Is British withdrawal from the European Union to leave Europe? Well, the governmental union or the political alliance at large? Does Europe’s asylum wave consist primarily of refugees or migrants? Well, it depends on your definition of refugee—the operative one in a given state, or one defined in some international legal document like the UN Refugee Convention? Is Obama a socialist? Well, socialist by American or European standards? Is same-sex marriage really marriage? Well, it depends on what you mean by marriage—a given state’s legal definition, or some other group’s usage? Do small start-ups or larger companies produce more innovation? Well it depends on whether you mean innovation in the sense of coming up with a new idea or the transforming for that idea into a product with value? Is mental illness relative to societal values? Well, that depends on whether you define mental illness in terms of symptoms or causes. Is humanism a secular movement and critical thinking in conflict with religion? Can posttraumatic stress disorder be a war injury? Am I an alcoholic, Doctor Kevorkian a murderer, Pluto a planet, coffee a drug, philosophy an objective science, and black a colour? Well…

Each arguably demands analyses of different types, and can be classified in various ways, but we can start off from the observation that they all have something in common, namely that some sort of linguistic clarification, here introduced “Well…,” is a relevant response in order to prevent verbal disputes. Verbal disputes typically ensue when preventive strategies of this sort are forgotten, and the parts go on, each having adopted different meanings. If that happens, they are arguably wasting their time: As difference in meanings normally entails difference in topics, they would then have the unhappy property 1 from above of talking past one another. And they would have the equally undesirable property 2 of lacking a substantive disagreement, because the disagreement would be over mere words at best, and non-existent at worst.

II. Is there a problem of verbal disputes in philosophy?

There’s above all one group of disputes that run a special risk of being deemed verbal, and that’s philosophical ones. Now, just as it’s beyond question that verbal disputes occur, it’s beyond question that they occur in philosophy also. That verbal disputes inhibit philosophical progress is a problem well known, but the question concerns the extent of this phenomenon. Though normally put as just a witty remark, it’s frankly quite true that the extent of philosophy comes down to an insider/outsider conflict: While they rarely question the disputes in which they’re themselves engaged, the tendency among philosophers is to have misgivings about the significance of those of their colleagues. In the metaphysics of time, to take an example, you may hold that the dispute between three and four- dimensionalists as well as the dispute between presentists, and eternalists are verbal, while the dispute between perdurantists and endurantists is not. I also think that the questions and subquestions I am interested in are substantive, but that those I’m not interested in are

ix verbal. It is impossible to determine these things from the outside, because there is a “chicken or the egg?” problem with how things appear; are we outsiders to these disputes because they are verbal, or do they appear verbal because we are outsiders? We face an odd situation in which there’s much consensus that verbal disputes are undesirable and far too widespread, but with no consensus on which disputes are verbal. This calls for increased metaphilosophical investigation. That insubstantive talkings past extend wide within philosophy is a concern not only for some philosophers of language, or philosophers in general, but it also seems to be the prevailing, if not default view among the uninitiated. “In one sense, it’s clear that P is true,” you might have heard them say, “but from this other perspective, it’s not true that P!” You understand perfectly well the concern expressed, yet it’s too imprecise for a proper response. “Then what’s the point in disputing P in the first place?” they might naturally reason on. What’s in the back of it, in embryotic form, can be labelled antirealism, deflationism, or maybe relativism. And what’s meant when they talk of different senses can be quite simply word meanings, or something more vague like “ways of being true” or maybe “perspectives.” It can be many things, but if we centre on word meanings, I think such skeptical worries can largely be captured by the more robust notion of a verbal dispute. Not only do I recognise it as a legitimate challenge in its own right, but even if only for reasons of its prevalence among the uninitiated, it’s deserving of a clear treatment. It’s what I’ll attempt in this essay. It’s not a challenge pertaining to philosophical disputes only, but also to the questions themselves, from which they originate. Take, again, the metaphysical question of whether objects endure or perdure through time. Endurance is the property of an object of being wholly present at any moment, while perdurance, as it is used by metaphysicians, is to be only partially present, having temporal parts in the past and the future which are not present (Sider 2001). It might be replied by the skeptic that the answer depends on what’s meant by the word object. If objects are individuated by the totality of their temporal parts, then the answer is a trivial yes: objects perdure through time. If, on the other hand, objects are individuated only by what’s present here and now, then it’s trivial that they are wholly present at any moment. If each answer follows trivially after the distinction in meanings has been drawn, then the question itself was conceived out of a linguistic misunderstanding to begin with. Thus, we not only have the notion of a verbal dispute, which might arise out of a substantive question even, but also the notion of a verbal, or trivial question, disputes over which are always verbal (Bennet, ibid). (On top of my unfairly rendered endurance example, the discipline of metaphysics makes up a rich pile of examples to pick from in diagnosing verbal disputes and questions, as it’s arguably the subfield that most evidently suffers from verbal problems all while laying the strongest claim to language- independence.1) The hypothesis that philosophical questions are trivial should be distinguished from the hypothesis that they are verbally inclined, which means they run a distinctive risk of giving rise to verbal disputes. While they make up the background, neither hypothesis will

1 Not all metaphysicians pursue language-independent objectives. For the contrary of that, see for example the “descriptive metaphysics” in P.F. Strawson’s Individuals (1959) which only seeks to describe how we think and talk. See also critical analysis and discussion of how language can be relevant to metaphysics in Sterjnberg (2009).

be defended in this dissertation: It would have been an unrealistic objective to establish the former, and the latter is too obviously correct to be in need of defence. It is obvious that philosophical questions demand a special attention to meanings, which is evident from the fact that the “Well, it depends on what you mean by that” response has a special place in the discipline. “Depends on what you mean by ‘F’” isn’t only a preventive measure against verbal disputes, but a tool to aide theoretical progress. And where it serves these functions more effectively than anywhere else is in the context of philosophical questions. To see this, compare the earlier set of examples of verbally prone ideological, cultural, socio-political, and everyday questions with this second set of bread and butter philosophy:

Do we perceive the world directly or only mediated by representations? Well, it depends on whether you mean veridical perception or perception in the sense of what appears to you. Can we know anything if we can’t rule out the possibility of error? Well, it depends on whether knowledge presupposes certainty, or whether you set the threshold elsewhere. Can a belief be justified by a falsehood? Well, it depends on whether it’s facts or beliefs that justify. Does this table exist? Well, it depends on whether you mean existence in the ordinary sense or in a metaphysically fundamental sense. Do we have free will given that things could not have been different than how they actually are? Well, it depends on whether you mean compatibilistic or incompatibilistic freedom. Is meaning “in the head”? Well, it depends on whether you mean externalistic or internaltistic meaning. Are sentences or utterances the primary bearers of semantic content? Well, it depends on where you draw the boundary between semantics and pragmatics. Does a difference in meaning entail a difference in topic? Well, it depends on how you individuate topics and meanings relative to one another. Can persons survive teleportation? Well…

The former non-philosophical set comprised verbally inclined questions exclusively, but only for the simple reason they were hand-picked with that property in mind, while the latter, in contrast, was arbitrarily chosen, and can include almost any philosophical question; general or narrow, epistemological or metaphysical; and still share those features with the set that was hand-picked. Equipped, anyhow, with the related notions of trivial and verbally inclined questions, the idea will at least be looming in the background that they both abound in philosophy. Starting from the observation just made that the “Well…” response is especially appropriate in this context, here is a blitz reason to think that many philosophical questions are either trivial or verbally inclined. First, the “depends” in “depends on what you mean by ‘F’” is not just reporting the ordinary dependence of a sentence’s meaning, and so truth, upon the meaning of its constituent terms. Had it been so, it would have been a licenced response to any question or claim, whatsoever. If to the question “Is there magma underneath earth’s crust?” it’s responded “It depends on what you mean by ‘magma,’” the answer is an obvious “no” if “magma” means Arizona Iced Tea. But the fact that it’s licenced only in certain contexts suggests that the dependence relation here reported signals something more. “Depends on what you mean by ‘F’” signals that there are several interpretations of “F” that are more or less equally natural in that context, and so several interpretations of the “F”-involving question itself. Now, it might turn out that under both interpretations, the “F”- question has an answer worth disputing, or that it does under neither. But you never make use of that reply in cases where there’s only one natural reading of “F.” So “it depends on what you mean by ‘F’” is a very revealing dialectical tool, which, wherever licenced, suggests that the question either is verbal to begin with, or has two interpretations both under which it is substantive, which is to say it is verbally inclined. And so, the fact being

xi that this response has a special place in the context of philosophy increases the likelihood that the discipline is dominated by questions of either type. It either has two or more equally salient meanings, or only trivial answers. Not proving much on its own, it definitely increases the stakes, because given that that many of these questions are either trivial or verbally inclined, it is only natural that many of the ensuing disputes will suffer from similar, hereditary problems. That being said, if they are verbally inclined, or if there is evidence to think they are trivial questions, it might just show that that these questions are especially deep in some interesting way, and so demands of the disputants a lot of knowledge about the topic and about the expert usage of these terms, and maybe a special heart and soul sensitivity towards profound subtleties.2 But in any case, we would like to find out. Any finding would be significant if those are the alternatives. 3 I mentioned above the many concerns outsiders may have about the pointlessness of philosophy, followed by a conjecture that these concerns can best be captured by the idea of a verbal dispute. A similar point is this. Due, if only, to the abstract nature of the subject, it’s only reasonable to expect philosophical disputes to suffer from defects. If there is some defect or another, then the most likely diagnosis is that it’s verbal, since there are few other candidate explanations in the running. If there are other diagnoses than that it’s verbal— for example that it’s nonsensical, they will be more severe and so harder to prove. In all, there are many reasons to investigate the idea of philosophical verbal disputes in philosophy. We have entertained now the ideas that philosophy is widely thought to be verbal, that philosophical questions are verbally inclined, and that verbal disputes arise frequently also outside of the field. Besides, there are also a number of theoretical benefits, which will become clear as the discussion proceeds in this text, to be enjoyed by metaphiloophical views wherein verbal disputes play an explanatory central role. For these reasons, it’s an endeavour worthwhile to explore what the consequences would be supposing it’s true that philosophy is largely or entirely verbal. But to be sure, it’s not within the bounds of this dissertation to argue any further than the foregoing for a conclusion to this effect. Neither that philosophy actually is so (if that were possible, then somebody would already have done it), nor which particular disputes therein suffer from verbality (which would require special attendance to the distinctive features of each dispute, but this project aims to study philosophy from the outside). Rather, supposing that some, many, or all philosophical disputes do, what more can be said about them? What would be the consequences? Are all philosophers each other’s strawmen?

III. Substantive verbal disputes

From what so far has been said, “verbal” entails “pointless,” for the reason that it (1) involves talking past, and (2), lacks a substantive disagreement. Recently, however, there

2 See Bridges and Kolodny’s “The quest to understand philosophy” in their The Possibility of Philosophical Understanding -- Essays on the Philosophy of Barry Stroud (2011, eds) for more on the distinctive characteristics of philosophical questions. 3 In addition to verbal disputes, and the stronger notion of verbal questions, there is also the even more contested idea of a verbal or “analytic” statement or proposition. Roughly, it is the answer to a verbal question, and so depends for its truth or falsity solely upon the meanings of its constituent terms, but I will limit my discussions to verbal disputes only. (Verbal disputes have indeed been likened to analytic truth before. See, for example, Graham 2014.)

xii

has been an increased interest in the fact that (1-2) might not always hold; that some disputes might be verbal but all the same free of these unwanted properties, and so worthwhile nonetheless! There are both good and bad verbal disputes, it has been emphasised and researched. Indeed, debates proceeding from many of the social and political questions in the example set above have had centre stage as case studies in substantive verbal disputes, or metalinguistic negotiations. A dispute can be substantive only if the negatives of (1-2) hold, namely that the disputants

(The negation of (1)) discuss the same topic, and (The negation of (2)) that topic is itself substantive.

The idea of a metalinguistic negotiation satisfies the two conditions. The first because even though they use the words with different meanings, in these types of disputes, the same topic is addressed indirectly; the second because the topic of metalinguistic negotiations is “How should words be used?” which is almost certainly substantive. Take a much used case, here from the most influential paper on the topic, Plunkett and Sundell’s “Disagreements and the semantics of Normative and Evaluative Terms” (2013) where they invite us to consider two disputants disagreeing over the sentence “waterboarding is torture.” It’s a verbal dispute because it owes to the two different definitions subscribed to by the disputants, the United Nations and the (previous) U.S. Justice Department, respectively. The former defines torture as any act inflicting severe suffering, physical or mental, in order to obtain information or to punish. The latter defines it as any such act inflicting pain rising to the level of death, organ failure, or the permanent impairment of a significant body function. They argue:

Even if we suppose that the speakers mean different things by the word “torture,” it is clear that we have not exhausted the normative and evaluative work to be done here. After all, in the context of discussions about the moral or legal issues surrounding the treatment of prisoners, there is a substantive question about which definition is better (19).

This is the central interpretative proposal. The dispute centres on meanings, all right, but the key is to understand it normatively, as centring on appropriate language use, how it ought to be used. The discourse theory at work here is that an utterance of “X is Y” need not say anything about X or Y.”. Divergent to its literal meaning, it can also be used to express propositions about how the words “X” and “Y” ought to be used, what they ought to mean:

By employing the word ‘torture’ in a way that excludes waterboarding, the speaker of [“waterboarding is not torture”] communicates (though not via literal expression) the view that such a usage is appropriate to those moral or legal discussions (19).4

4 There is a lot of jargon surrounding these topics. The term “metalinguistic negotiation” might confuse the Alfred Tarski reader who is familiar with “metalanguage” meaning the language whose objects of discourse are words. But, in contrast, a metalinguistic negotiation, because words are used, not mentioned, is carried out precisely in Tarski’s object language. ”Metalinguistic” in Plunkett does not mean “conducted in the metalanguage,” but “implicitly about language,” semantic content in particular. Saying that something is metalinguistic is like saying it’s semantic. “Metalinguistic dispute” is the same as “verbal dispute.”

xiii

Plunkett and Sundell’s theory of substantive verbal disputes will be the subject of closer scrutiny in chapter 3. Though there is plenty to be said on exactly how this is going to work out, the main take away is very roughly that the verbal dispute is substantive, and this for two reasons. First, they are disagreeing over the same topic, namely appropriate language use—hence (1). And secondly, as this topic is normative, it is itself substantive—hence (2). To further drive the message home, take as a second example the verbal dispute over same sex marriage from above. Like the torture case, the disputants are using the word differently, and would have agreed on whether the sentence “same sex couples can marry” is true had they only agreed on the meaning of “marriage.” But they don’t. Let’s say one disputant is using the word in accordance with a liberal state’s current legal definition, whereby same sex couples can marry, while the other disputant is using it in accordance with a religious text, whereby only opposite sex couples can. The fact that the linguistic clarification “Well, it depends on what you mean by ‘marriage.’” is licenced here suggests that the dispute is verbally prone. But the fact that neither disputant sees the need to make use of it suggests that they prefer to remain in the verbal dispute and not agree on meanings. This suggests, in turn, that meanings is the very topic itself. And hence, they are disagreeing on the same topic, namely which meaning ought to be used. So much for topic-sameness, or neg-1. Assuming that this interpretation is right, the mere fact that they are discussing the same topic is unfortunately not sufficient for neg-2: substantivity. The topic must itself be substantive, remember, which is somewhat harder to put in place. To be sure, the fact that they are discussing the same topic strongly supports the view that it’s substantive. Besides, the fact that the topic is normative is a very strong indication too: “You/we/I ought to do this/that/the other” is a sentence form that is almost guaranteed to express substantive propositions, and constitute substantive disagreement. If you make a proposition about what to do or what not to do, how can that be substantive, right? The more complex questions, however, have to do with the explanation of this linguistic normativity, and what it is grounded in. For why does it matter what things are called? It’s relatively clear that it matters, but the more complex questions concern the hows and whys. In general terms, whatever reason the verbal dispute is substantive equals whatever reason the disputants refuse to synchronize. Whatever the reasons, it’s not always arbitrary which meaning to go by. In the case of marriage, it quite to the contrary has consequences as to the relative treatment of gays and certain religious scriptures, because former meaning emphasises gay rights whereas the latter emphasises scriptural authority— or the opponent may have some other reason to see that usage as a corruption of the word. And so, by disagreeing over which meaning ought to be used, they are disagreeing indirectly over the substantive topic of their relative priority, a question that obviously has practical consequences. The primary idea behind metalinguistic negotiations is that these practical consequences explain the normativity, and thence the substantivity, of the linguistic proposal that is under dispute. The same story can explain the waterboarding case. The normative disagreement over the meaning of torture inherits its substantivity from the substantivity of the question of whether to exercise waterboarding:

[The speaker of “waterboarding is not torture”] communicates the proposition that waterboarding itself is, in the relevant sense, unproblematic—a proposition that is, we submit, well worth arguing about (19).

The disputes are substantive in virtue of being normative, and the disputes in the above examples were normative, in turn, in virtue of having practical consequences. The details of

xiv

how mere definitions can have practical consequences will be among the central subjects to be put to closer examination in this essay. We will look at many ways in which the normative question of what to call things matter. By now, the upshot of this section is that if you add normativity to the interpretation, to be verbal does not have to entail being pointless. The oughts of language is a topic that’s both substantive and shared by both parts to many verbal disputes, and so it is necessary to draw a distinction among verbal disputes between substantive, and mere verbal disputes.

IV. “Philosophy as Conceptual Engineering”

In defence against the above sort of skeptical misgivings, the natural strategy of choice is to reject that philosophy is verbal. But if we bar this route, could we instead opt for the alternative defence that concedes that philosophy is verbal, but belonging to the category of substantive verbal disputes? Rather than being based on linguistic misunderstandings, can it be based on substantive linguistic disagreements over normative issues on how language ought to be, just like the disputes over marriage and torture? This idea has gotten some wind in recent literature. First, “Conceptual engineering,” which is also called “conceptual ethics” is the normative study of improving concepts. Typically contrasted to descriptive methodologies, such as analysis, that ascribe importance to actual concepts, conceptual engineering is a revisionary methodology that gives priority to the concepts we should use. Plunkett and others’ idea of substantive verbal disputes from the previous section is an instance of this trend. The conceptual engineering literature will strike the reader by the ease with which it brings together so many distinct fields and topics. The authors are gliding seamlessly between everything from theoretical issues typical of philosophy of language and metaethics to argumennts in moral philosophy and metaphysics at the level of particular cases. Even empirical sciences such as biology and anthropology are thrown into this blend. The current thesis is limited to the metaphilosophical aspects of this conceptual engineering, namely on the idea of philosophy as conceptual engineering. Philosophy as conceptual engineering is the metaphilosophical view that either prescribes conceptual engineering as the right methodology, or interprets philosophy as already engaged in that type of activity, in other words that philosophical disputes are or should be metalinguistic negotiations about conceptual engineering. In addition to the above theory he defends along with Sundell of metalinguistic negotiations over evaluative terms, Plunkett also has a metaphilosophical theory of philosophy as conceptual engineering in his “Which concepts should we use? On metalinguistic negotiations in philosophy” (2016). His views are representative of many other conceptual engineers, and belong to the strand that considers actual philosophy to be engaged with this already, but implicitly. And most conceptual engineers, besides seeing it as the way philosophers have been doing philosophy for a long time, along with Plunkett they also see this as the right way to do it. Many proponents of these interpretative ideas, furthermore, not only seek to justify traditional philosophical methods, but also support revisionism in regard to improving actual methods used in today’s philosophy, saying that conceptual engineering should be elevated from implicit metalinguistic negotiations to a programme where it is carried out explicitly. All these aspects of the family of views I call “philosophy as conceptual engineering” are expressed in the following excerpts from recent literature:

Too much ink has been spilt on philosophy as conceptual analysis. The alternative view, that philosophy is at least as much, if not actually more, engaged with creating, refining, and fitting together our conceptual artefacts [...] has received too little attention. (Floridi 2011, 293)

Semantic claims and conceptual analyses abound throughout contemporary philosophy. In their purest forms, semantics and analysis are exclusively descriptive projects, specifying the meanings or contents of representational vehicles as they are actually used by particular speakers, thinkers, or communities thereof. As such inquiry shades into linguistics and psychology, it is increasingly turned over to empirically informed philosophers of language and mind. But of course, philosophers of language and mind are also quite busy with the theory of such inquiry – studying the nature and methodology of semantics and analysis. In light of this heavy workload, it may come as no surprise that comparatively little has been written on the nature or methodology of semantic and conceptual prescriptions. As we underscore in the present paper, however, claims about how one ought (or would do well) to think and talk are nearly as ubiquitous in philosophy as their descriptive counterparts, not to mention their prevalence in ordinary discourse. (Burgess and Plunkett 2013, 1091)

While philosophers often have been concerned with our actual concepts [...] philosophers should also be asking themselves whether these really are the best tools for understanding the relevant aspects of reality, and in many cases consider what preferable replacements might be. Philosophers should be engaged in conceptual engineering. (Eklund 2014, 295.)

Very large chunks of epistemology are concerned with what exactly the word ‘know’ is used to pick out. Gettier’s 1963 paper ‘Is Justified True Belief Knowledge?’ triggered an enormous amount of literature aimed at figuring out the conditions under which a justified true belief becomes knowledge. This goal is primarily what Robert Pasnau calls ‘lexicography’: to describe our concept of knowledge or the extension of ‘knowledge’. That is, the goal is purely descriptive. This contrasts sharply with the revisionary tradition. Those engaged in conceptual engineering would ask: what should our concept of knowledge be? Can we improve on what we have? Why should we be intellectually complacent and stick with the concept that we have if we could have a better concept of knowledge? (Cappelen, forthcoming)

It has dawned on me that this kind of philosophical methodology (i.e., replacing defective concepts, which are responsible for philosophical troubles) can and should play a much larger role in philosophical theorizing. … Following Simon Blackburn, I’ve called this methodology conceptual engineering... Conceptual engineering is taking a Socratic (critical) and Nietzschean (active) attitude toward one’s own conceptual scheme. Many of us already think that we should take this critical and active attitude toward our beliefs. We should subject them to a battery of objections and see how well we can reply to those objections. If a belief does not fare well in this process, then that is a good indicator that it should be changed. By doing this, one can sculpt and craft a belief system of one’s own rather that just living one’s life with beliefs borrowed from one’s ancestors. The central idea of conceptual engineering is that one ought to take the same critical attitude toward one’s concepts. Likewise, if a concept does not fare well under critical scrutiny, the active attitude kicks in and one crafts new concepts that do the work one wants without giving rise to the problems inherent in the old ones. By doing this, one can sculpt and craft a conceptual repertoire of one’s own rather than just living one’s life with concepts borrowed from one’s ancestors. (Scharp 2015, blog.)

xvi

What concepts are, and exactly what it means to engineer them will not be a major concern, because the current issues primarily will be addressed in terms of meanings instead, and I use the two more or less synonymously, the latter being the thinking equivalents of words, concepts.5 As for a rough idea on what is meant by philosophical concepts, take the following suggestions from Scharp:

On my view, the concepts relevant to philosophy include truth, knowledge, nature, meaning, virtue, explanation, essence, causation, validity, rationality, freedom, necessity, person, beauty, belief, goodness, time, space, justice, etc. (ibid.)

And as for what it means to engineer, revise, or improve concepts, what is important for the current issues is only what it means to intend, propose, or defend a revision, an improvement or an engineering blueprint; not how can be carried out or whether it is even a plausible theory. For a general idea of what engineers here have in mind, take this excerpt from Patrick Greenough:

Conceptual defects are many and varied. Concepts (and terms) can be incomplete (‘open- textured’), confused, unsatisfiable, vague, or inconsistent. They can be too inclusive, too narrow, or simply empty. They can be too complex, too simple, or not fit to feature in any useful explanation; they can be superseded, tired, hackneyed, or systematically misapplied. They can be too parochial, too elitist, or too recondite. They can be loaded with inappropriate connotations, bad ideological baggage, or serve as ongoing devices for deceit, discrimination, or oppression. A concept may be flawed on more than one dimension – broken in many different ways. Conceptual Engineering, as a result, is a multifarious business. (2017, 3, book manuscript).

All these statements make it sound plausible that many verbal disputes are substantive. Philosophers haven’t gone astray in hopeless armchair meditations; instead they are fixing and repairing concepts. How can engineering not be substantive? The important point to take away is that if philosophy is engaged in this type of activity, then philosophical disputes can be understood as concerning the shared topic (neg- 1) of which concept to use, or some other normative linguistic or conceptual business. If so, then metalinguistic disagreements would not be obstacles to progress and defective communication, but substantive disagreements over normative issues (neg-2). And many verbal disputes will not be pointless, but substantive metalinguistic negotiations over conceptual engineering.

V. Thesis statement: If philosophy is verbal, it is verbal in a bad way

5 But for a rough characterisation to make sense of the excerpts above, concepts are something like the categories within which we think and talk, and are often referred to as representational devices. It’s a broad term with many precisifications, typically used in clauses like “the concept of…” followed by a noun to denote something like a linguistic entity, but which has got more to do with thought than language. Animals have some concepts, in that sense, and we have concepts for many more things than we have words for. The concept of a concept is therefore closely related to the concept of word meanings, but more abstract, and less connected to particular words and contexts, than meanings are. Some construe concepts to be the same as word meanings, or at least what gives words their meaning, and or some vague idea or principle that unifies several related meanings under the same linguistic expression; but what exact role they can play in a semantic theory I think would amount to a verbal dispute over the word “concept” itself. 5

xvii

My own contribution to the issues now introduced consists in critical analysis and comparison between David Chalmers and David Plunkett’s respective metaphilosophical theories within the wider context of “philosophy as conceptual engineering.” Besides comparative analysis, I also provide some reasons to doubt that Plunkett’s hypothesis of metalinguistic negotiations can constitute a viable response to the problem of verbal disputes in philosophy. If philosophy is verbal, it is merely verbal, I conclude. More precisely, the target of these arguments is that even supposing that disputants are addressing the same topic, the topic itself still cannot be substantive merely in virtue of being normative. The theory is probably correct for the paradigm social and political cases such as “marriage” and “torture,” and related ones that have practical consequences, which is to say, implications about what to do; but not theoretical questions in philosophy. Insofar as verbal disputes in philosophy are substantive, the linguistic normativity in question must be based on truth and how things really are (epistemic normativity), and not in what ought to be done (practical normativity). But nothing supports the odd idea that language is governed by epistemic reasons, or that the way that language ought to be is connected with the way things are in reality. Unless these difficult requirements can be met, verbal disputes in philosophy are merely verbal, and not, like Plunkett and the above others seem to think, substantive metalinguistic negotiations.

VI. Disposition

The first chapter addresses David Chalmers’ (2011) heuristics driven theory of verbal disputes, which prescribes both particular methods as well as more general methodological principles to philosophical practice. First the family resemblance concept of a verbal dispute is fleshed out and given a wide meaning which encompass many similar sorts of activities, which share the feature that there is something pointless about them (1.1.) a feature glossed at in 1.2. Some discussion of how the idea of linguistic deference relates to verbal disputes follows (1.3.), before the method of elimination is introduced (1.4.), and test driven on the philosophical case “Are humans free?” (1.5). What emerges from the discussion is that a certain type of question, of the form “What is freedom?” probably gives rise to many disputes. These socratic questions are dealt with by Chalmers’ special type of method, called the subscript gambit (1.6), before an objection is addressed, which goes to the effect that it doesn’t matter if disputes are verbal, because what matters is to be right about the question irrespectively (1.7). Then the question is asked “Which disputes in philosophy are verbal?” which is concluded conditionally because for the same reason that Chalmers’ “dispute parthood” relation is, the relation of “same topic” is rather unclear. Finally, some conclusions are drawn about the theoretical significance and background that supports these methodological prescriptions (1.8). How many verbal disputes are there in philosophy? The second chapter discusses some ways to respond to the problem of verbal disputes in philosophy in the face of the insider/outsider problem (it is difficult to diagnose disputes because we can’t read the topic of a dispute off from the literal meaning). Many disputes might resist the method of elimination, because they are not disputes over some term’s meaning, like “Knowledge,” but theoretical disagreements over what that property is. First, I argue that disputes over conceptual analysis are verbal (2.1). Then I argue that disputes over real definition are verbal (2.2). Then, in 2.3 I argue that questions of language cannot be questions of reality (2.3.2), and that questions of reality cannot be questions about language (2.3.3). Finally, I discuss a trait shared by all the conflicting methodological outlooks that have been brought out, namely that they have a non-liberal conception of language use, while Chalmers and the character Humpty Dumpty say that “meanings are our servants, not our masters” (2.4).

xviii

Can verbal disputes be substantive? After a rendering of David Plunkett’s theory of substantive verbal disputes, or metalinguistic negotiations (3.1), and his theory of metalinguistic negotiations in philosophy and philosophy as conceptual ethics (3.2), I raise some criticism that centres on the distinction borrowed from moral philosophy between episemic and practical reasons, or normativity (3.3.) and consider some responses (3.4.). Lastly, the discussion of Plunkett is placed into the broader context of David Chalmers’ theory, with the occasional point or two from the even broader context of Rudolf Carnap (3.5.) I make a vague point in the end about the vague notion of a topic that what the continuation of a topic is might be constrained by reason, and that this might provide a marginal type of epistemic constraint on language, otherwise metalinguistic negotiations are pointless (conclusion). The examples in the text are confined to the grand questions of philosophy, typically in the form of the treatment they have received from the 20th century analytic philosophical tradition, however, I think most of this can be extended mutatis mutandis to other areas of philosophy from other times. Like the talk focuses on meanings, and not concepts to the same extent, the text is about disputes, and not about thought and talk; even though much of what is argued can be applied mutatis mutandis to what philosophers are doing outside of paid hours, but alone, in bed, or when they gaze at the stars.

Chapter one—David Chalmers on verbal disputes

The aim of this chapter is to develop a richer notion of what verbal disputes are, and also shed some light on the idea of metalinguistic negotiations, and to fix the framework that will be in use later, drawing from philosopher David Chalmers’ “Verbal Disputes” (Philosophical Review 4, 2011). In this seminal paper, Chalmers aims to define the notion, and then offer practical remedies to avoid verbal disputes in philosophy, for then to draw some first-order conclusions in the philosophy of language. While the former two make an excellent base for easing in on the arguments that will be provided in the second and third chapter, the first-order conclusions of the second half of this long and comprehensive paper do not directly concern this present project; we are interested only in verbal disputes and whether they can be substantive. Take a moment to reflect on the various ways in which disputes can be verbal. First, we have the explicit verbal disputes where issues of language use equal the actual first- order issue, meaning these verbal disputes are never talking pasts or misunderstandings. But they can be pointless nonetheless, albeit less prone to, because being explicit on the topic it is easier for disputants to detect whether and to what degree that topic is worthwhile. It’s a type of verbal dispute can be set somewhat to the side, because they are just like any other dispute, only that the topic happens to be language, and again, this can be either descriptive, as in Chalmers example of disputes over lexicography; or normative, as in his example with legal disputes over institutionalised definitions. Since the line between what is implicit and explicit is a bit unclear and would demand a lot of theory, and because an implicit disagreement might rise to the surface during the course of a dispute, leave it open for “verbal dispute” to refer to both implicit and explicit ones. They have many relevant things in common, most importantly that they concern language. Against this is the fact that to blur the distinction is to leave out important differences between when disputes are talkings past and known to be verbal, and differences also between when terms are used and when they are mentioned, such as some of the differences between the sentences “Martini is a cocktail with gin and dry vermouth” and “’Martini’ means ‘cocktail made of gin and dry vermouth.’” Further, “metalinguistic negotiations” might or might not be synonymous with “substantive verbal disputes,” because there might be other ways than normativity for verbal disputes to be substantive, but it is hard to imagine what exactly this would amount to if not for a metalinguistic negotiation: Assume that some implicit substantive is not normative, but over how language actually is. Now, since the two disputants are using their terms to convey their respective views about how language actually is, how can you

make sense of that other than that they are also conveying the normative standpoint that the words ought to be used in accordance with those descriptive beliefs?6 (The dispute would indeed be non-revisionary, but no less normative for that reason—which is to say, a metalinguistic negotiation, which thus appears to be the only member of the species substantive verbal dispute:

Verbal disputes

Implicit Explicit

Merely verbal Substantive Descriptive Normative

Metalingusitic negotiatons

The taxonomy is approximate, and the various forms that verbal disputes can take will be clearer by the end of this chapter. The chapter begins by going through Chalmers’ various definitions of a verbal dispute, and especially his favoured, “broad” definition which is designed to capture cases that would not amount to different meanings under various semantic theories often associated with semantic externalism (1.1), and then the second section (1.2) raises come concerns about how many verbal disputes can be captured without reference to differences in meaning. The third section (1.3) introduces Chalmers’ method of elimination for identifying and resolving verbal disputes, while the subsequent discusses its application to philosophy (1.5), before, lastly, some conclusions are drawn about the concept of a verbal dispute (1.6).

1.1. Verbal disputes defined

Chalmers settles on the following definition, which in this dry and technical section will be dissected a piece at a time (522):

6 This effect might not be preserved if the words were mentioned, meaning this type of normativity might be exclusive to implicit verbal disputes.

A dispute over S is (broadly) verbal when, for some expression T in S, the parties disagree about the meaning of T, and the dispute over S arises wholly in virtue of this disagreement regarding T.

The very first thing to ask is, whence the modifier “broadly?” Chalmers distinguishes broadly from narrowly verbal disputes, the latter being defined in terms of propositions:

A dispute over S is [narrowly] verbal iff S expresses distinct propositions p and q for the two parties, so that one party asserts p and the other denies q, and the parties agree on the truth of p and q.

The notion of disagreement at work in the paper is, crucially, extra-linguistic in that it centres on disagreements in beliefs rather than over the truth of linguistic utterances— sentences. Disagreement over the latter is reserved the term “dispute” instead. Thus, in a narrow verbal dispute over S, there’s no disagreement over the proposition expressed by S. The reason the broad definition is preferred to the narrow definition is that the latter rules out cases where the disputants, under the supposition of the doctrine of semantic externalism, both mean the same by their terms, and express the same propositions by their sentences, even though some sort of semantic disagreement is clearly in force. It’s therefore found to be insufficiently theory-neutral. According to externalist views on meaning, what speakers believe they mean, and what they actually mean by their words and sentences can come apart in cases of semantic deference. But Chalmers’ crucial point is that it’s only what speakers believe about meanings, and not what they actually mean, that is relevant for a dispute being verbal (521). 7 To semantically defer, or to use a word deferentially, as Chalmers prefers to call it, is very roughly to leave it up to circumstances exterior to one’s mind what the word, and so the sentence, means when you use it. For various views on whom or what we defer to when we use words, consult the externalist literature of the precise details of this “division of linguistic labour,” as it is coined by the early proponent Hilary Putnam, followed by his conclusion that “Meanings just ain’t in the head” (which is to say meanings are individuated “broadly”) (1975). For now let’s just say we defer to something like the language community of which we are a member, or to some linguistic authority therein, whoever they may be. The views also differ along the dimension of how voluntary this is: An internalist, who hold that meanings are in the head, or narrow, might regard deference as part of the term’s meaning, so that to use a word with deference to one’s community is to mean something like “whatever my community means by this term,” while to an externalist, merely being a member of that community might be enough to decide what you mean when you use a term, irrespective of whether that usage is accompanied by an

7 Besides deference, two other reasons for meanings to come apart from beliefs about meaning are also discussed: Cases of inadequate reflection, and cases of hidden facts about meaning, such as in Williamson (1994)

intention to defer, or a deferential attitude or anything of the sort. Under these suppositions, there will be a host of same-proposition, or same-word meaning verbal disputes that the narrow definition, or “dual proposition framework,” is too narrow (as it happens) to be able to include, and the broad definition is therefore preferred (521). (If you’re not an externalist, however, (or a deferentially hard-nosed internalist) narrowly and broadly verbal disputes pick out the same set.) Besides its compatibility with exernalist interpretations of semantic deference, it enjoys some structural and economic benefits. In short, the broad definition says just that a dispute is verbal if it arises in virtue of a disagreement over the meaning of one of the disputed sentence’s terms. Instead of a basis in what speakers say, believe, and believe they say, like the narrow definition, it evades mention of all three in favour of the more covering notion of a disagreement. And instead of relying on a particular configuration between the three, it involves simply the relation “in virtue of.” We will go through both, beginning with “disagreement.” So what is a “metalinguistic disagreement”? A metalinguistic disagreement is spelled out in terms of conflicting metalinguistic beliefs by which he means beliefs about what words and sentences mean. So a verbal dispute is a dispute that owes to conflicting beliefs about the meaning of an expression, even though meanings might actually be the same, for reasons of semantic deference. There’s only a handful of constraints on these metalinguistic beliefs and the meanings they’re about, all easy to swallow. They can be (a) tacit, in that you don’t actually have to entertain them. And (b), they need not be articulable, in that the believer doesn’t have to be able to describe the meaning or come up with a definition or anything of the sort. Two further constraints on meanings is that they be described as more than (c) the set of objects to which the term applies (“extensionalism”), and (d) disquotation schemes of the form “’T’ means T” (“deflationism”) (522-3). But why doesn’t it read

A dispute over S is verbal when, for some expression T in S, the parties use T differently, and the dispute over S arises wholly in virtue of this difference in the meaning of T.8

instead? This definition here is both more natural and closer to the pre-theoretical notion. It captures all the cases of the narrow definition, on only the assumption that word meaning

8 Following Chalmers, usage and meaning are here used interchangeably, although there is another widespread sense of ”usage” whereby the usage of some term, T, can differ between two speakers not because they mean different things by T, but simply because they disagree substantively over T’s extension:

… two parties who disagree substantively over whether O. J. Simpson is a murderer will also disagree about the extension of ‘murderer’, but intuitively this does not disqualify them from agreeing on the term’s meaning (523).

determines propositional content. And with a basis in the former in favour of the latter, it enjoys all the economic structural benefits of the broad definition mentioned above without appeal to its more technical notion of a metalinguistic belief. So why replace the italicized parts above with an unwieldy technical term? The reason for this inconvenience is, again, externalism. For reasons of externalism, and also other potentially complicating theories of meaning, even this definition would be too narrow. When semantic deference is understood externalistically, verbal disputants would mean the same by T. To remedy that, the novel concept of a metalinguistic belief is introduced with the purpose to replace meaning-talk with belief-talk. So instead of having to talk about T meaning different things to different speakers, and the speakers expressing different propositions by S, Chalmers will instead talk about the speakers having disagreeing metalinguistic beliefs about the meaning of T, and by extension, disagreeing metalingusitc beliefs about the meaning of S. But again, if you subscribe to an internalistic understanding of semantic deference, be free to stick with word meanings. These ways of talking are interchangeable for our purposes in a way that they are not for Chalmers, due to the dialectical position he is in. Because he later in the paper will use the idea of a verbal dispute to argue for first-order philosophy of language positions about meaning, like semantic translucency, and an a priori-like relation he later will call scrutability, as well as a variant of analyticity—all at odds with externalism, he has to suspend all talk of meaning and use the broad definition instead. The broad definition requires only the view that whenever speakers use an expression, they do so with metalinguistic beliefs about the expression’s meaning (522), which they will do as long as (a-c) are satisfied. It’s a neat way to get around the present externalism complications. The metalinguistic belief-notion of a verbal dispute allows for disputants to mean the same by all their terms, and to utter the same propositions by the same sentences, but still, due to metalinguistic disagreements, dispute verbally. To say that again, two speakers can mean the same by T in S, and express the same proposition by S, but still disagree verbally over S. In a way, all the work we’d expect meanings to play in defining what a verbal dispute is is loaded over on to this particular, very meaning-like, type of belief. To use an example, return to Bennett’s verbal dispute with the purist who believes that “martini” means only drinks made of gin or vodka with dry vermouth, and the sorority girl who believes that practically anything served in the V-shaped glass qualifies as one. Since their metalinguistic beliefs about the key term “martini” differ, and since that metalinguistic disagreement is what’s responsible for the dispute, it comes out verbal. So even if they both defer to the same linguistic community, and thus mean the same, the dispute is by the broad definition no less verbal. So while the dual proposition and meaning frameworks look like this,

Disputants Meanings Propositions

Purist gin and dry vermouth This drink not a martini Sorority girl anything served in a V-shaped glass This drink is a martini

the same case looks like this in Chalmers’ dual belief framework:

Disputants Metalinguistic beliefs

Purist “martini” means gin and dry vermouth Sorority girl “martini” means anything served in a V-shaped glass

An important difference between the two is that “broadly verbal” encompasses the verbal disputes where the words are not used differently. Only differences in meaning and not differences in beliefs about meaning can distinguish between implicit and explicit verbal disputes, meaning the narrow definitions are more precise, encompassing only those where the meaning differences are at work on an implicit level. But sometimes, when talking about verbal disputes on a more general level, it is beneficial to include the explicit ones too, maybe excluding some narrowly verbal disputes depending on whether differences in usage entails metalinguistic disagreement or not.9 Either way, “broadly verbal disputes” includes many of the large, longstanding debates that shift continuously back and forth between metalinguistic negotiations (which are implicit) and explicit verbal disputes. “Is Pluto a Planet?” is a good example, and of many philosophical disputes too it is a fair description that linguistic issues at times become first-order issues, for example a debate over truth that cycles between talking about the predicate “true,” and concept of truth, and the property truth, the same metalinguistic disagreement being responsible for all three. Since the two are just the same apart from this and that meanings replaced with beliefs about meaning in the name of theory-neutrality, unless specified otherwise, the broad definition will remain in the background as I simply stick to talking about meanings.10 Under any definition there is also the converse possibility of verbally approving, when there’s actually a disagreement, but due to linguistic differences both speakers assent to the same sentence (526).11 Now for the second key constituent. Along with the clauses “owes to,” “gives rise to,” and “is responsible for,” what’s reported by “in virtue of” is some sort of non-causal explanatory relation. Casual would not suffice, because a disagreement over T in S can cause a dispute over S that’s not verbal, for example an explicit verbal dispute about the meaning of T, which is, as we saw, just a normal dispute that happens to be about

9 Which it does if there being differences in how disputants interpret each other entails that they have a metalinguistic disagreement. (See section1.3.) 10 Therefore, problems coming from theories of meaning such as externalism are addressed directly in 1.3., 2.3.2, and 2.4.4. 11 An example can maybe be a church sermon where the participants think they all agree over the sentence “God exists,” but due to widely differing conceptions of the meaning of “God,” each believer is a non-believer to the next. It should be qualified that in this example the participants are only disagreeing literally; there of course will be non-literal content expressed by that sentence the assent to which is an element that unites them.

language, or a personal enmity that then causes first-order disagreement (523). But other than being non-causal, if a more precise characterisation demanded, then Chalmers’ discussion will have to disappoint, because the relation isn’t further explicated, but treated as a primitive. To Chalmers’ defence, leaving a concept up to the intuitions like this, treating it as a primitive, undefined term is a move that philosophers often make use of. I think he is right that we have a good intuitive grasp of what “arises in virtue of” means. Thankfully, “Arises in virtue of” has the counterfactual implication that if there had been a disambiguation of S, say into S1 and S2, there would no longer have been a dispute over either. So should there be a problem with treating it as a primitive, then this implication can provide us with yet another definition of a verbal dispute, namely the counterfactual one:

[A] dispute over S is broadly verbal when for some term T in S, if the parties were to agree over the meaning of T, then they would (if reasonable) agree over the truth of S. (524n.)12

The purist and the sorority girl would both have agreed that the sentence “this glass contains true martini” is false and “this glass contains Appletini” is true, at least insofar as it is a mere talking past, but the metalinguistic disagreement over whether “martini” picks out both true and bogus Martini, or the former only, is what upholds the dispute over that sentence. It follows that all verbal disputes are verbal with respect to a term in the disputed sentence. The idea should be quite clear; the term with respect to which a dispute is verbal is just the term a metalinguistic disagreement over which gives rise to a verbal dispute.13 There’s also the adverb “wholly,” for reasons of the distinction between a wholly and a partial verbal dispute. If a dispute over S arises only partially in virtue of a disagreement regarding T in S, then the dispute is only partially verbal. This means that even if the meaning of T were fixed between the disputants, there would still be disagreement left regarding the truth of S. There’s a case to be made that most verbal disputes are actually only partially so, which starts from the premise that behind every verbal dispute some actual disagreement or another is mirrored. The non-verbal disagreement at the back of the martini-dispute, might, for example, be over the value of authenticity versus commercial misuse of established labels, or whether arrogant purism behaviour should be considered acceptable, or who knows?

12 The definition would be broadly verbal if it “agree over the meaning of T” means “agree on what T means.” If it means “mean the same by T,” it would be a narrow counterfactual definition:

[A] dispute over S is [narrowly] verbal when for some term T in S, if the parties were to use T with the same meaning, then they would (if reasonable) agree over the truth of S.

13 There may be cases where a dispute is verbal with respect to more than one term. For example a dispute over whether Front National is a racist political party will both depend on the meaning of “racist” and also on what defines it as a “party.” (It may have been founded as a racist party, but drifted away later, for example.)

To take an example from real life, I recently had a Sunday supper verbal dispute with my physicist brother Herman over whether Jupiter’s core is solid or gassy, and it contained both a verbal element as well as a substantive disagreement. The verbal element arose from my layman conception of the distinction between solids, liquids, and gasses, which construes it simply as distinction among intrinsic molecular states without taking pressure into consideration. On my usage, solid water is synonymous with frozen water, for example, while by the scientific usage, only given the sufficient amount of pressure, any chemical can be solid in any temperature. On this usage it would be true that water can be solid even at boiling temperatures, not in virtue of a frozen microstructure, but for the reason that that it’s compressed by outside forces, which equals the reason why under the scientific usage of “solid,” it’s true that “Jupiter’s core is solid.” Since the pressure in relatively constant in our ordinary atmospheric circumstances, it is a variable that can safely be ruled of our talk of liquids and gasses, and so ruled out of their corresponding meanings likewise. But for a physicist who might have to deal with cases where superhot water is squeezed into a state of solidity, called Ice IV, the meaning of solidity has to be different in order to encompass that—a difference which gave rise to a verbal dispute. The non-verbal element of the dispute, moreover—the actual substantive disagreement, was due to my mistaken belief that Jupiter be penetrable (maybe only as long as the penetrator is heat resistant enough, and with forward propulsion sufficient to withstand a planetary storm or two). Pointing forward to section 1.4, partially and wholly is tied to the idea of subdisputes and dispute parenthood. My above talk about verbal- and non-verbal elements of disputes is what Chalmers captures in terms of the “part of” relation. The dispute “Is Jupiter’s core is solid?” is partially verbal because the two subdisputes “is Jupiter’s core lay-solid” and “is Jupiter’s core scientific-solid” reveals substantive disagreement over the former only.14 To summarize, aside from the last few paragraphs’ miscellaneous constituents, this section has first and foremost addressed the difference between the dual proposition and dual word meaning frameworks, on the one hand, and the dual belief framework on the other. We saw how the former two were found to be too narrow under the supposition of externalism, so that the latter framework’ broad definition was devised in their place, because even though meanings might not be, metalinguistic beliefs are still “in the head,” and so invariant with regard to how the deferential story goes. Chalmers’ central idea

14 As a final note on this, the conception of solidity that is used in science is an example of a conceptually engineered variant of the original solidity concept, but that does not mean that the verbal element in the dispute here sketched is substantively verbal. It’s still just merely verbal, because we weren’t discussing which one ought to be used, or anything along those lines, but we were using the two different meanings without any such normative purpose in mind. Even if physicist Herman here had been aware of my conceptual shortcomings, and yet with a pedagogical grand plan in mind has kept using that word in that way knowing that I would not understand it at first, it is still doubtful that it would qualify as a metalinguistic negotiation. Because I had no such standpoint, and because I did not pick up on his nonstandard usage, the verbal part of the dispute was no less a pointless talking past—you need two to tango (more on this in 1.3.).

behind the broad definition is that it is these beliefs that are relevant for whether a dispute is verbal or not, which means that the notion of a verbal dispute does not depend on any particular theory about meanings (apart from constraints a-c above, that is). Externalistic objections have at least to my experience often been raised in connection with these issues, in arguments that since disputants mean the same, they are disagreeing subsatntively, but if Chalmers is right, then such objections are missing the point, because it is strictly speaking not meanings that are relevant for whether or not a dispute is verbal, but beliefs about meanings. Yet, we are still free to talk about differences in meanings rather than metalingusutic disagreements as much as we like, the latter only working as a rally point in the background.

1.2. Substantivity and pointlessness

Some may understand the concept of a verbal dispute to entail pointlessness essentially, in which case “substantive verbal dispute” would be a contradiction in terms. But that usage is rare. The question of whether verbal disputes are merely verbal is not verbal, at least not merely. Though “verbal” typically conveys some negative verdict, true, those connotations are commonly due to the context of that particular dispute. And it is normally clear from context, likewise, whether it is used in that excluding sense. Chalmers’ usage is no exception. “Sometimes words matter,” he takes care to stress early as on the second page (516); “Disputes over words are sometimes important disputes, when something important rests on matters of linguistic usage.” In not being defined in terms of pointlessness, the notion of a verbal dispute employed in the paper thus opens up for substantivity, and is therefore ideal for our current purposes. He first goes on to list the trivial cases of substantivity, such as why the meanings of words matter to lexicographers, philologists, and historians as they interpret textual sources, for example. The second class of examples of how such disputes can be worthwhile are the clear cases where word meanings have practical consequences, such as legal definitions and the choice of words in contracts and official documents. The final, and less obvious, group that he mentions are words whose connotations and associations have some fixed normative content, and so have things at stake with them. The examples “torture” and “terrorism” are words we all agree apply to evil actions only, so to dispute what counts as falling within their extension by a dispute over their meaning, is indirectly to make claims about which actions should and should not be condemned, which counts for substantivity. If conducted indirectly, by using rather than mentioning the contested terms, these are precisely the metalinguistic negotiations that are the topic of chapter 3, where the question is whether disputes in philosophy can belong to that category, and so be substantive even though they are verbal. The immediate problem is, obviously, that in contrast to “terrorism,” “torture,” and “marriage,” philosophical terms like “perception,” “mind,” and “personal identity” appear to lack such “fixed normative content.” Fixed normative content should therefore not be the only way for negotiations over mere words to matter. Until further notice, it suffices for substantivity that metalingusutic negotiations involve normative claims about words, whether or not that normativity ultimately be grounded in the content of those words or elsewhere. So though chapter 3’s arguments indeed turn on what these sources can amount to, the mere fact that claims like “’X’ ought to be used thus and so” are

normative is for the time being all that needs to be the case for a verbal dispute to count as substantive. What is substantivity? I take it that substantivity is among the most straightforward concept in Chalmers paper, the surrounding literature, as well as the discussions in this thesis. It is a normative concept which can be taken to hold by the criterion “Should this dispute be discontinued?” The degree to which a dispute over some sentence should carry on is the degree to which the dispute over that sentence is substantive. The concept of substantivity employed here is used without any theoretical backing, as just the negation of pointlessness, covering most things it can mean, but still there is a sense in which the question “Well, that depends on what you mean by ‘substantivity’” can be raised. If substantivity were theoretically defined exclusively in terms of empirical testability, for example, this would be a short affair, as it is contested which disputes satisfy that condition. The the only necessary condition that matters for the present purposes is (1-2), that the disputants address the same substantive topic, and for that topic to be substantive it must be worthwhile or relevant to discuss, or something along those lines. Chalmers argues in terms of “relevance” in several places in the paper, and much hangs on what is relevant to the other thing. A recurring question is “are linguistic facts relevant to the topic we are talking about, our domain of concern?” I take that pointlessness and relevance are attributes we can easily judge on a case-to-case basis, as they derive from beliefs about what should and should not be done, so nothing of importance hinges on how the details are spelled out. All these notions can be thought of as gradeable, so that it is a matter of degree how relevant, pointless, or substantive disputes and topics are, as well as how much a topic be shared between two disputants, and so on, but that poses no problem as long as the cases considered fit well within a reasonable threshold. Likewise for the fact that all these notions are relative to goals and purposes:

For example, a substantive dispute over the exact location of a specific fundamental particle would normally be pointless, but one can imagine a case in which the trajectory of an asteroid heading for the earth depends on this location. (537)

The fact that most of these notions are relative actually makes things much less complicated, as the goals and purposes needed to be addressed here will have to do with philosophy, and only that. So it might well be that, say, lexicography, which is to say the study of how and with what meaning words are used, is substantive relative to the goals and purposes of an historian or a lawyer, but when we consider its relevance to first-order philosophical questions, however, the judgement might land in another place—why should it matter to philosophy what words do or should mean? In very general terms, therefore, the topic whose relevance will be scrutinized is language—as it is actually used, or how it ought to be used, and the goals and purposes relative to which that relevance will be assessed is philosophical truth, or understanding, or something along those lines. And much more need not be said about substantivity. (Terms for this property that are also to be found in the literature include “significant,” “substantial,” “genuine,” “important,” and “serious.”) Intuitively, the reason verbal disputes are the opposite of these things, pointless, is that words are not relevant to the domain of discourse. “Intuitively, a dispute between two

parties is verbal when the two parties agree on the relevant facts about a domain of concern and just disagree about the language used to describe that domain (523, italicized).” Even if the domain of discourse itself concerns words, disagreements over the words used to describe that domain (the object language, i.e. the language used to talk about the objects in that domain, words) would be verbal, because metalinguistic agreements (i.e. agreement about the object language), do not matter for the disagreement about that object-level domain. Think if it as a first and a second level; even if the object of dispute itself was words, the words used to describe those words would not matter. So disputes are verbal in general, because:

Often, however, words do not matter. It often happens 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” (517).

1.3. Can there be broadly verbal disputes?

Before moving on from his definition to his theory of verbal disputes in the next section, it is beneficial for our grasp of the concept to look at a possible problem with the idea of a broadly verbal dispute, with Chalmers’ preferred definition in focus. An objection can be raised, which goes like this:

First: Because metalinguistic beliefs can be about many different things, “metalinguistic disagreement” is found to be ambiguous. Second: Once disambiguated, most verbal disputes are found to lack metalingusitc disagreements altogether. Therefore: It cannot be metalinguistic beliefs, but meanings, that are relevant for whether a dispute is verbal

If this objection is correct, there will be a host of disputes that can be rendered verbal only when or if meanings differ, and will thus be ruled out under suppositions along the lines of deference and externalism. I argue that there are actually only a small number of metalinguistic disagreements in ordinary speech, because “belief about meaning” can be so many different things, and the two have to be the same type of belief to be able to constitute disagreement. Because one man’s modus ponens is the other man’s modus tollens, this can go two ways: Either the broad definition is correct about these cases, and several apparent verbal disputes are found not to be verbal after all; or these disputes are indeed verbal and so constitute counter-examples to the broad definition. More generally, either externalism holds for these cases, and they are not verbal; or they are verbal and meanings differ, which presumably only a non-deferential or internalistic conception of meanings can account for. In any case, Chalmers would be mistaken about the theory neutrality of the notion of a verbal dispute. Very briefly, the exact theories he takes the idea of verbal disputes to be compatible with are any externalisms that does not involve that “there are sources of semantic externalism other than semantic deference and inadequate reflection, resulting in hidden facts about meaning” (561). So which cases do I have in mind here? Which verbal disputes lack metalinguistic disagreements once the notion has been disambiguated? I think they are many, and would

contend that we don’t have to look further than to Bennett’s martini to find an example, in which case the problem would be a quite significant given how representative this case is of how verbal disputes come to be. I will provide some reasons to doubt that there is in fact a metalinguistic disagreement between the purist and the sorority girl, meaning the broad definition fails to capture even that paradigm case. If by the end of the day it’s not a severe problem, it still is worth the attention. Beginning with the premise that Chalmers’ notion of metalinguistic disagreement is ambiguous, a metalinguistic disagreement regarding a term, T, is spelled out in terms of conflicting metalinguistic beliefs about T, but little is said about what it means to have a belief about what an expression means. In particular, what would such a belief be about, and to what would it be answerable? He says in a footnote that “the relevant beliefs for each speaker may concern the community meaning, or the content of the term in the current context” (522n). Now, it can’t be both at the same time, because “T community-means M” and “T does not contextually mean M” can both be true as well as believed at the same time, so let’s pursue only one of them, namely the latter option, supposing all verbal disputes come with a context. Further, beliefs about contextual meaning can itself come down to many things, so an additional disambiguation is demanded. A natural way to make this further specification is to draw the divide between a descriptive and a normative type—interpretative beliefs, and metalinguistic standpoints. The former is a belief about your conversational partner and their communicative dispositions within the context of a conversation, roughly, how to understand them and how they will understand you. To a first approximation, it is a belief about what can be called a speaker’s meaning: the contribution made by a particular word in a particular context to a particular speaker’s utterances. One sort of interpretative belief is a belief about what T means in the speaker’s mouth, in other words what they mean by it when they use it, and another sort is a belief what T means in their ears, in other words what they believe you mean by it when you use it. Metalinguistic standpoints, in contrast, are beliefs about what a term should mean in the context, regardless of what actually it means to one’s conversational partner. The reasons for having such a standpoint can come to many things—maybe a view on how it should be used in any context by anyone whomsoever, or about how it is actually used elsewhere, for example by the language community at large or by some authority such as a dictionary or an expert—who knows? Let’s just say a metalinguistic standpoint is any metalinguistic belief that is not an interpretative belief. As such they are arguably less common, as they go beyond the minimal assumptions one needs to make about a term’s meaning in order to conduct a linguistic exchange. To illustrate the difference between the two, return to the sorority girl and the purist, and their respective mealingusitic beliefs about the meaning of “martini.” To tell from Bennett’s description and the names which the characters are given, the sororist expects the purist both to understand her usage of “martini,” and to use it likewise, which means her metalinguistic belief is a splendid example of the interpretative type. The purist’s belief that “martini” refers only to drinks made of dry vermouth and gin, moreover, is by all accounts a metalinguistic standpoint about the correct or ideal use of that term, since we can suppose that he does not expect the sorority girl to understand that elite usage, at least not from the outset. Telling from Bennett’s description, moreover, the sorority girl

subscribes to no such standpoint; her metalinguistic beliefs are oriented exclusively towards communication. The problem should now become apparent—there is no metalinguistic disagreement here at all! The purist’s metalinguistic standpoint is not in conflict with any of the sorority girl’s beliefs, since none of hers are standpoints, at least in so far as she is not an exercising anti-purist; and her interpretative beliefs are no more in conflict with his purism, as they concern only communication. More generally, because interpretative beliefs are about a conversational partner, they will never constitute disagreement as they are never about the same thing; and since standpoints about what words mean seem to be far rarer than verbal disputes, or so I will argue, it is doubtful that a dispute being verbal entails such a disagreement. To sum up the argument in the form of a reductio to the absurd conclusion that there are no verbal disputes:

Assume: If a dispute is verbal, it entails a disagreement in metalinguistic beliefs

1) A metalinguistic belief is either an interpretative belief or a metalinguistic standpoint

2) If a dispute is verbal, it does not entail a disagreement in interpretative beliefs

3) Neither does it entail a disagreement in metalinguistic standpoints

4) Therefore, no dispute is verbal

The argument is valid, though the detail is left out that the term, T, with respect to which a dispute is verbal equals the same T over which there is a corresponding metalinguistic disagreement, but nothing turns on this stylistic choice. 15 Given that there are verbal

15 That is, the conclusion that ~P follows by classical truth-functional logic from the premises with the form: P  Q Q  (R v W) ~ (P  R) ~ (P  W) where “P” in this argument reads “There is a dispute which is verbal with respect to some T;” “Q” reads “There is a disagreement in metalinguistic beliefs regarding T;” “R” reads “There is a disagreement in interpretative beliefs regarding T;” and “W” reads “There is a disagreement in metalinguistic standpoints regarding T.”

disputes, it follows that either Chalmers’ broad definition (Assumption), or one of my premises (1-3) are false. The three premises will be addressed in turn. Can Chalmers reject premise 1? The distinction just falls out of the fact that we can understand what somebody means by an expression although we have different beliefs about it, so a reply to the effect that “metalinguistic belief” is only homonymical and not ambiguous would not work. While it can be contended that the distinction could be drawn in other ways, this would only hold for the standpoint-part of the distinction, and not for the clearly delineated interpretative belief-part, as it is beyond doubt that beliefs about what a conversational partner means by an expression can differ from other types of beliefs about that expression’s meaning. So while non-interpretative beliefs can come to many things, and might be further divided into several categories, the former notion of an interpretative belief is well defined as the beliefs that are necessary for communicating with a conversational partner in a context, which is all that’s needed. Outside of a context, of course, this disambiguation has no place: When a child asks for an explanation of an expression’s meaning, for example, it would be an unusual response to demand a clarification of whether they were asking for an interpretative key or some sort of standpoint about the meaning of that word; the answer would just be “T means X” where “X” may refer to something very general, such as what you believe to be the community meaning. Within a context, however, a disambiguation is necessary, because a belief that one’s conversational partner means thus-and-so by an expression does not have to be at odds with other beliefs about the meaning of that expression, like what it means in the community. Since we can reasonably expect the purist to know that his Greek society conversational partner is yet to take up martini purism, this might in fact be what is going on in Bennett’s case. It would at least be a realistic scenario that the uncompromising purst very much knows how his nonstandard usage will be interpreted, and yet sticks to it in order to exercise his convictions, setting his interpretative beliefs aside in favour of using the word in accordance with his standpoints about that word instead. If so, this is a case where standpoints and interpretative beliefs come apart, which demands a disambiguation of “the purist believes that martini means a drink made of dry vermouth and gin.” In any case, as the two at least can come apart, that’s all that’s needed in for a disjunction to be in place in the first premise above. If anybody has to accept this premise, it is Chalmers, who as we will see, holds that there are multiple interesting concepts “… in the vicinity of philosophical terms such as ‘semantic’, ‘justified’, ‘free’, and not much of substance depends on which one goes with the term” (539, italicized). Can he reject premise 2? Since an interpretative belief is just a communicative key to understanding one’s conversational partner’s utterances, and to be understood by them, interpretative beliefs seem unable to be in conflict, as they are about different things. The purist’s interpretative beliefs are only about the sorority girl, and vice versa. It can be replied that if both disputants believe they mean the same thing because they defer to the

same community, presumably because on a subconscious level they are semantic externalists, then a belief about the community meaning equals a belief about what the speaker means.16 However, these beliefs would not have counted as interpretative beliefs, because they would not have been about the speaker and their communicative dispositions. It would only have been to use “means” with a different meaning, namely community meaning, and there would still have been two different beliefs, which would only be conflated by someone who were communicatively incompetent. It is of course true that we base ourselves upon beliefs about community meanings inferentially when we interpret others, but we are always prepared to leave those beliefs behind when it comes down to actual linguistic exchanges where each partner is interpreted individually. So this isn’t a viable route either. Furthermore, interpretative beliefs can of course be true and false, and this might indeed be crucial for why verbal disputes are pointless—the martini dispute is after all not a disagreement about “this glass contains gin and dry vermouth.” But if the dispute is verbal partly because the sorority girl’s interpretative beliefs about the purist are wrong, then that is just saying it is verbal because the purist means such-and-such by “martini,” which, as we saw, might not be an option given the current views on meaning under consideration. To say anything to the effect of an interpretative belief’s truth or falsity is just to say something about what a speaker means by an expression, which is the very thing the idea of a broadly verbal dispute was devised to circumnavigate. So under the assumption of deference and externalism, to define verbal disputes based on whether one or more disputant is mistaken in their interpretations would be nothing more than to define them in terms of differences in meanings all over again, in other words to define them narrowly.17 (I might be wrong about this; it might be that disagreements in metalinguistic beliefs in fact can depend on at least one part being mistaken yet without the two meaning different things by the expression. So leave open this hybrid, externalism- friendly alternative.) Can Chalmers reject premise 3? I think very rarely do we have standpoints about what a word ought to mean in a given context, and there is even something odd about this idea. Maybe the purist is one among few, but at least something seems integrally irrelevant about the idea of a word meaning something; it is to people that words have meaning—the conversational partners. Does the purist want the sorority girl to adopt his way of speaking? Even granted that there are such things as beliefs about what a term ought to mean in a given context, then even more rarely than having them, do they come in conflict and give rise to disputes. If metalinguistic beliefs are beliefs about a word’s meaning in a

16 In a different definition of verbal disputes, Eli Hirsch grants that meanings can be decided by the community, but defines them in terms of community meanings nonetheless, but in order to capture the relevant differences in meaning, he instead assumes the disputants to be part of different communities, speaking slightly different languages (2005). 17 It might be replied that an interpretative belief’s truth-value does not entail anything about meanings, for the reason that even speaker meanings are determined by external factors. It’s a radical view which I don’t know how to respond to other than that it would completely sever meanings and beliefs about meanings from communication.

context, moreover, it has to concern interpretation, since Chalmers’ number one assumption about metalinguistic beliefs that “whenever speakers use an expression, they do so with beliefs about that expression’s meaning, where these beliefs may be tacit beliefs rather than explicit beliefs” (522) seems to be a description of interpretative beliefs precisely, and not metalinguistic standpoints, whatever they are. Since none of the three premises are up for the taking, the only available route is to construe metalinguistic beliefs as beliefs about community meanings instead. Against this, there is still the sorority girl and the purist who seem not to be in disagreement about the meaning of “martini.” Even if metalinguistic beliefs are beliefs about community meanings, there is still a sense in which their respective metalinguistic beliefs are about different things, at least if we suppose she is not at the bar to defend anti-purism in a metalinguistic negotiation. So while her belief may be about the community meaning, assuming purism isn’t an empirical view on what the community means, maybe his is about something else; namely which meaning is correct irrespective of how it goes with the deferential story.18 So even if my communication/standpoint distinction among metalinguistic beliefs failed to hit the divide between these different subject matters, the argument can be run again mutatis mutandis, this time with a distinction between descriptive community meaning beliefs and community meaning standpoints, for example. But without further ado, call it a draw. Though this discussion must now cease right when it gets interesting, the intuition remains, anyhow, that the dispute is verbal not because of a disagreement in what they believe about the word, but because they use it with different meanings. If it was responded that the Martini dispute is actually a freak case, I wouldn’t think so. It is actually quite an exemplary circumstance that one disputant uses a word in accordance with one type of metalinguistic belief, while the other uses it in accordance with another type—recall also the neglect of interpretative beliefs by physicist Herman above. I think speakers can have many different types of metalinguistic beliefs simultaneously: one interpretative belief and one standpoint, and another about what it means in the community, and another about how experts use it, and so on. Words, in contrast, can only be used in one way at a time. Maybe the relevant differences in meanings come down to which type of metalinguistic belief the word is used in accordance with? Maybe verbal disputes arise in precisely those cases where deferential usage is resisted? None of the above is especially important. The primary reason for Chalmers to define verbal disputes broadly is that the notion of a verbal dispute will be used later in the paper in arguments against certain views associated with externalism, and so should start out neutral vis-à-vis those views. An important upshot is that there are many things that

18 Alternatively, he might believe that the deferential chains end at the experts which he takes to be the purists, so that the word meaning even in the sorority girl’s mouth is determined by the linguistic labour of the purists. If she too believes that she defers to the same experts, then they might disagree in metalinguistic beliefs.

“meaning” can mean, (corresponding to how many things “metalinguistic belief” can mean), and therefore problems that involve “meaning” are verbally inclined.19 Other than that, there likely no definition that can capture all and only the cases we take to be verbal disputes, but “we can instead see the [broad] characterization as pointing us toward a salient and familiar phenomenon, rather than delineating its contours precisely” (525). It’s a view on definitions and conceptual analyses that will be further addressed in chapter 2, that the various definitions work together to elucidate the phenomenon from their own different angles, jointly clarifying the notion. Even paradigm cases does not need to be among a definition’s criteria of success, as “Nothing substantive rests on the term ‘verbal’” and “we could in principle call them ‘schmerbal disputes’”(520n). So if I’m right in insisting that metalinguistic disagreements only rarely give rise to verbal disputes, so that the broad definition only captures a small number of cases, it should not disrupt the broad definition from playing its role in the later sections of “Verbal Disputes.”

1.4. A heuristics oriented theory

In this section the notion will be further elucidated by a discussion of the practical protocols of the body of “Verbal Disputes” that both prescribe ways to detect verbal disputes and ways to solve them. Very generally, a verbal dispute is solved by attending to meanings (527). For example by following Eli Hirsch’ “Method of Charitable Interpretation” (2005), or, simply, by asking “Well, it depends upon what you mean by that.” But it’s a theme in Chalmer’s philosophy to find alternative ways to talk about matters otherwise requiring a treatment in terms of meaning, which, as the previous section just displayed, is a theory laden notion that it’s desirable to steer clear of. Since meanings are difficult, and since it is hard to form a schematic meaning-based heuristic short of an extensive semantic theory, Chalmers proposes an alternative, meaning-free method called “the method of elimination.” It can take the operator to the conclusion that a dispute is due to semantic differences, either word meanings, sentence meanings, metalinguistic beliefs, or differences in what’s said by an utterance relative to the different conversational partners, without mentioning anything like that. It’s relevant to our topic because of the theory of verbal disputes that’s in the background, which we can derive from this part of the paper and the discussion that follows. The heuristics there introduced play an important role in characterising the phenomenon. Consider it a sort of operative definition, like:

19 In 1.5. the line of reasoning prescribed by the method of elimination, and its extension the subscript gambit (1.6.) is opened up, the idea, in particular, that one can avoid verbal disputes by stipulation. The methodological approach with which Chalmers is operating accords well with this attitude, which is called a Humpty Dumpty attitude (2.4.), because if his concept of a broadly verbal dispute does not pick out many verbal disputes, because the two are unrelated concepts, then that might not be the role it was meant to play.

A dispute over S is verbal just in case it can be resolved by the method of elimination.

The method is simple. Let’s say we dispute the sentence S and want to check whether that dispute is verbal. The method asks us to bar the use of the key term T in S and state our disagreement over S in terms of another term T’ in another sentence S’, and if that’s not possible, it is verbal. In more detail, the method gives the following instructions:

Step 1, eliminate. Bar the use of T. Step 2, the practical step: Look for an S’ in the newly restricted vocabulary such that the dispute over S’ is part of the dispute over S.

This will almost certainly yield some progress in almost any case, which means there may be faint elements of verbality present in almost any dispute, but too small to be regarded as a problem (more on this below). The method can thus be applied whenever a dispute is suspected of verbality, if only for the faintest suspicion.

That was the practical step, but now for the more theoretically interesting

step 3, the diagnosive step. If there is such an S’, the dispute over S is not wholly verbal, or at least there is a substantive dispute in the vicinity. If there is no such S’, then the dispute over S is wholly verbal (except in the special case of vocabulary exhaustion, discussed below) (227).

And then, in order to check whether S’ is verbal, reapply the method. The definition makes use of the “part of” relation, which is explicated further in sections 7-9 of “Verbal Disputes,” as well as in later work, where Chalmers asks questions such as “is there a hierarchy of disputes?” and “do all disputes have subdisputes as their parts?” To cut an interesting theory short, we only need to assume a “non-symmertical structure of parthood relations” wherever some dispute is taken to be part of another, but not necessarily one that stretches across the whole language. That parthood structures across disputes are non-symmetrical does not presuppose a hierarchy of disputes, but only that some are more fundamental than others. For example, it is suggested that a dispute over whether mice are phenomenally conscious is part of the dispute over whether mice have sensory experience. Later in the paper this is taken to views about the former dispute explaining the latter two, and about the “concept” of phenomenal consciousness being more fundamental than the concepts of experience and pain. But neither the explanatory relations nor the talk about concepts are presupposed by the idea of parthood relations and certainly not the idea of a verbal dispute. So even granted that the idea of dispute parthood bottoms down to a view on meanings, it would only amount to words having imprecise meanings that can potential a number of more clearly defined submeanings; it alone would still be compatible with semantic holism and most other semantic theories that might be at odds with the conceptual explanatory view that come later. To get a notion of parthood alone, we don’t need to go into details about meanings, as Chalmers again provides the notion with a non-semantic “counterfactual gloss:”

We can say that a dispute over S’ is part of a dispute over S when: (i) if the parties were to agree that S’ is true, they would (if reasonable) agree that S is true, and (ii) if they were to agree that S’ is false, they would (if reasonable) agree that S is false. (528).20 21

The idea of parthood relations across disputes seems like an unproblematic aspect of the theory of verbal disputes, especially as this gloss is made in terms of the rather neutral idea of a reasonable counterfactual agreement. Besides, it enjoys intuitive underpinning from its resemblance to the same relation across topics, meaning we can extend this so as to licence talk about sub-disputes, sub-topics, sub-disagreements and sub-questions more or less interchangeably (henceforward without the hyphen). If a dispute over a sentence contains two other disputes, for example, then the sentence potentials two different topics. And the extent to which a dispute is verbal, moreover, equals the extent to which the dispute fails to touch upon a topic that’s disagreed upon. And to wrap it up, since method of elimination demands the same question to be stated in different words, this is to demand the words to be changed while the topic remains the same. If this can’t be done— verbal! Let’s test drive the method on a couple of cases. Eliminating the use of “martini,” is there an alternative way to state the sentence “This glass contains martini” such that there still is disagreement? No. There are of course alternative ways of stating the sentence— “this class contains gin or dry vermouth” and “this glass is V-shaped,” but none which preserves the disagreement, and therefore, it’s verbal. Let’s also apply it to the apparently substantive question of whether there’s water on the moon’s surface. After barring the key term “water,” several candidates report themselves. If at least one disputant had been ignorant of the harsh conditions on surfaces where there’s no atmosphere, then replacement of “water” with “liquid water” could have amounted to a disagreement, but since, as Chalmers stresses, “we cannot just speak of a ‘verbal question,’ independent of context” (518), let’s also illustrate the method with in more realistic context where the disputants have (and know that the other has) some knowledge of chemistry. If so, could the sentence “Is there H2O on the moon?” have amounted to a disagreement? Assuming mutual knowledge that there are always some molecules of H2O wherever there are gasses and radiation, even if existing only for milliseconds: not likely. So had this been the sole candidate, the dispute over whether there is water on the moon would have been found to be verbal. However, since it’s doubtful that an agreement over the H2O-question entails an agreement over the water-question,

20 I am thankful for the help that Anders Strand provided with figuring out the implications of these counterfactual conditionals. 21 Note the relevant difference between a subdispute that the disputants take to continue the disagreement, and a subdispute that does not preserve the disagreement. On Chalmers’ usage of “dispute A is part of dispute B” only the ones that actually continue the disagreement count as parts; whereas all the merely possible subdisputes are ruled out. Unless anything else is made clear I will use this and the related expressions to cover many more possible subdisputes so that contexts can be left much more unspecified.

then, as we saw, this cannot be the correct subquestion given the present context. So here is the better suggestion: “Are there significant persisting arrangements of H2O on the moon?” If this is where the disagreement lies, then the dispute over that question is part of the dispute over the original water-question, which is thus found to be substantive, and the elimination method helped us get closer to a more precise way of stating the dispute there over. In all, the method predicts that the martini-dispute is verbal while the water on the moon-dispute is not. The operative definition would read:

A dispute over S is verbal with respect to T in S iff, there is no sentence S’ such that S’ does not contain T, and the dispute over S’ is part of the dispute over S.

The definition has the strength that it captures the gradability of “verbal,” because even though there typically is only one subdspute that continues the disagreement, parthood is no less a matter of degrees. If the dispute over S’ is only a minor part of the dispute over S, and S’ was the best replacement sentence we could find, then S is almost completely verbal. But if S’ is a significant part of S, then only certain elements of the dispute over S were verbal. A consequence is that most verbal disputes will likely contain some non-verbal elements. The dispute over whether Pluto is a planet, for example, is verbal because if you bar the use of “planet,” the most prominent subdisputes reveal only agreement, however, there will still be substantive elements there in the periphery, disagreements about various issues related to that dispute. Chalmers suggests “Astronomical terms should be used in the way that is most useful for science”, ‘Traditions should be respected,” and “X is president of the astronomical Society” (if the parties agree that the president is the arbiter of meaning and agree about what X says) (528). Even in the martini case we can find substantive disagreements in the periphery: “the names of classic drinks should not be exploited for commercial use,” or “you should be familiar with the traditional terminology of cocktails before going out drinking with your sorority.” Even “I am a classic type of man who knows a lot of stuff” might be among the sub sentences at stake here. And why stop there? The line can be drawn even further from the original dispute if some inspiration is drawn from the holistic view commonly attributed to W.V.O. Quine that beliefs are assessed not individually, but all at once as a corporate body. From his influential “Two Dogmas of Empiricism” we have the idea that it is not individual statements but only larger collections of them that are the unit of (empirical) significance (1951). “Empirical” set aside, it may be true of every dispute that all or large chunks of background disagreements play a part, so even if the part played by each individual one is miniscule, when the method of elimination reveals a sentence that is agreed upon, it will always be dwarfed by the large conjunction of all the sentences that make up that background disagreement. If this is correct, it might pose a problem; because every dispute that is not just a total misunderstanding would only be a nod in the disputants’ large scale disagreement. The inspiration to draw from this doesn’t have to with semantic holism (which Chalmers already considers not to accord well with the view (549)). It can be a holism about every background disagreement together playing a part in every dispute, for example, so that whenever the conflicting beliefs in conjunction are more responsible for

the dispute than the incentives to address them individually or in the form of their literal topics, then the dispute will not be much verbal by Chalmers’ theory.22 Not only the Pluto case would be found acquitted, but also the Martini case, and also the case with Secretartiat the racehorse considered later. Verbal disputes are sometimes known and upheld on purpose, even when the meaning of the word does not have any intrinsic interest to the disputing parts. They keep using the word with different meanings, knowingly engaging in verbal disputes, the disputed sentence working as a bottle’s neck through which a mix of background disagreements are channelled. The idea with the method of elimination is that removing the word would splinter these background disagreements into independent topics that can be disputed individually. But in cases where disputants do not want to do this, but instead address a series of background disagreements all at once, mediated by a metalinguistic view on how a word ought to be used, then these will not be diagnosed by Chalmers’ method. In other words, its diagnosive capability seems dependent on the disputants’ having a desire to resolve the dispute, which far from often is the case. Chalmers has a smart way to fix this, just by stipulating them away:

In these cases [i.e. metalinguistic negotiations approximately], resolving the residual dispute resolves the original dispute over Pluto precisely in virtue of resolving a metalinguistic dispute over the key term ‘planet’. We might say that in these cases, the relation between S and S’ is metalinguistically mediated: the parties disagree over S in virtue of disagreeing over a metalinguistic sentence M and disagree over M in virtue of disagreeing over S’. We can then stipulate that for a dispute over S’ to be part of the dispute over S, in the sense relevant for our purposes, the relation must not be metalinguistically mediated (528).

What licenses this stipulation? The stipulation seems to enjoy support from the pre- theoretical concept of a verbal dispute. Even though there are substantive disagreements at stake, the only reason they would be relevant for the original dispute has got to do with meanings, which means the dispute centres on the mere meanings of words after all; the desired result. Rule metalinguistic disagreements out, for if they are substantive, then mention the contested words instead of using them. A metalinguistic sentence is just a sentence where the word is mentioned rather than used, as in an explicit verbal dispute. (Implicit) disputes over metalinguistic sentences and disputes mediated by such disagreements now stipulated out of the set of subdisputes, if the method of elimination is applied to a verbal dispute, it will almost always reveal that there is no residual disagreement. To state this important point again, all substantive

22 Note that this is not a problem with counterfactual criteria as such. Both the narrow and the broad counterfactual definitions earlier in the chapter will classify these wilful verbal disputes or meatalingusitic negotiations as verbal, because an agreement on meanings—either how words are used, or beliefs about meanings—would have led to agreement on the original sentence.

disagreements that make up a verbal dispute will in almost23 all cases be metalinguistic disputes or metalinguistically mediated disagreements. The method can therefore distinguish verbal from non-verbal disputes, even if the dispute has substantive parts like metalinguistic negotiations do. However, as far as I can tell, it cannot distinguish a mere verbal dispute from a metalinguistic negotiation, because after elimination any residual disagreements over words are already stipulated away. To understand why, note that the method does not spit out agreed upon sentences like “Pluto is planet-like,” for example, or “This glass does not contain gin and dry vermouth.” It asks only for residual disputes, based solely upon the disputants’ dispositions to continue and discontinue disputing. It is up to them, as it were, which disputes are part of their dispute and which are not. Frankly, I can only speculate, but to distinguish merely from substantively verbal disputes would maybe demand something like a distinction between literal and non-literal topics, or help from some other contested theory in semantics.24 Maybe something like a key is packed into the parenthesised clause “if rational” in the counterfactual gloss, as there is arguably a sense in which metalinguistic negotiations would not come about if the disputants were sufficiently rational, maybe if rationality entails a certain attitude oriented towards solving verbal disputes, or to address each topic individually, or to stay on the literal topic, or what have you. In various places in his work, Chalmers makes heavy use of the related notion of an idealized reasoner, which can shed light on this; and idealization is also found later in the current paper in its verdict on metalinguistic negotiations:

Ideal agents might be unaffected by which terms are used for which concepts, but for nonideal agents such as ourselves, the accepted meaning for a key term will make a difference to which concepts are highlighted, which questions can easily be raised, and which associations and inferences are naturally made. (541.)

Call these differences lexical effects, following Cappelen (2014) Idealisation is also employed in the paper’s last few chapters where the “is part of” relation is upgraded to the more idealised “underlies” relation. A metalinguistic dispute about “Planet,” for example, may be part of a dispute about Pluto, but it would not underlie it, as such lexical effects would be ruled out by idealisation.

23 The conclusion is modified by “almost,” because there are some special circumstances considered in (528n) and (529n), that are verbal but with substantive subdisputes that are not metalinguistically mediated. But, as we saw, neither of the definitions of the phenomenon are intended to delineate its contours precisely. 24 For what it’s worth, it can distinguish a substantive normal dispute from a metalinguistic negotiation if we take the liberty to read off from the counterfactual gloss on parthood something like prominence: The most prominent subdispute is the disputed subsentence S’ an agreement over which would have led to an agreement over the original sentence S (if the disputants were rational). If S’ is metalinguistic or metalinguistically mediated, then the dispute over S is a substantive verbal dispute, but if it is not, then the dispute is substantive and not verbal.

This is only to shed some light on the rationality-clause and its ability to distinguish mere verbal disputes from metalinguistic negotiations, given that disputants can dispute in principle anything they like indirectly. The point is just that if we help ourselves to Chalmers’ extended inventory of theoretical tools the difference might be far clearer. For now, what’s in the rationality constraint can be left intuitive, as I think we have a clear grasp that such as “traditions should be respected” is not the same topic as “Pluto is a planet.” If you buy into his theory of “concepts,” however, and his “anchored inferentialism” whereby “primitive concepts serve as anchors in which other concepts are grounded by inferential relations” and “the method of elimination consists in moving to inferentially related concepts within the web of concepts” (553-4), then all this will be put on sturdier legs, as disputes about traditions would not underlie disputes about stellar objects, because these concepts would not be inferentially related according to those theories. But as it stands, the following italicized part of the method’s step 3 remains a very important qualification: “…if there is such an S’, the dispute over S is not wholly verbal, or at least there is a substantive dispute in the vicinity” (527). These interesting extensions can be ignored in what’s to come, because the main concern here is philosophy, and it can be reasonably expected that potential verbal disputes is something philosophers have more than an attitude, but a sincere interest to resolve. Finally, if the method of elimination reveals a sentence S’ the disagreement over which cannot be stated in other sentences, then we have a case of vocabulary exhaustion. As we saw, the dispute over “this glass contains martini” cannot be stated in a way that preserves the disagreement and yet involves other terms than “martini.” The sentences we found regarded other topics, like purism. Vocabulary exhaustion entails that the dispute is verbal, unless the dispute is a bedrock dispute. The idea of bedrock disputes (but not “bedrock concepts”) falls out of the idea of disputes being part of one another, and as little as possible will be said about them here, note only that “bedrock disputes should not be identified with disputes over epistemologically bedrock theses: theses that one cannot argue for and that constitute brute intuitions or premises. Rather, what is relevant is a sort of conceptual bedrock” (547), and also, “[q]uestions about which concepts are bedrock should be distinguished from questions about which properties are metaphysically fundamental (551)” and also from the idea of “… primitive concepts. I do not think that bedrock concepts are the concepts of which all other concepts are composed” (553). There are only very few candidates to (non-verbal) disputes where the key term cannot be replaced—for example the dispute over whether there’s something it’s like to be a bat. You can, of course, replace it with “do bats have phenomenal consciousness?” but that just means that there is something it’s like to be them—same concept but different modes of expression. Phenomenal consciousness, or something it’s like-ness is therefore a bedrock concept, at least according to Chalmers. Another suggested bedrock concept is the concept of morality, or being morally right, because as we proceed with the method of elimination and bar moral terms, then

the disagreement gets harder and harder to state. It is plausible that once all moral terms are gone, no disagreement can be stated. We might agree on all the nonmoral properties of the relevant actions but still disagree on whether it is right. (543)

The idea is further developed in the last three sections of the paper as well as in later work, but we can leave it with this. Suffice it to say, because bedrock disputes involving bedrock concepts are so rare, vocabulary exhaustion strongly indicates verbality. Here is first a negative conclusion. Because the “part of” relation rests on what disputants would have done counterfactually if they were rational, it is hard to diagnose disputes from the outside, because only the disputants have access to that information. And because the having of substantive subdisputes, say “Astronomical terms should respect tradition” is what makes disputes substantive, these can differ from the literal topic, as Pluto, or whatever it is, of the canonical dispute. This can be an explication of what I have called the insider/outsider problem:

Tt is difficult to diagnose disputes as verbal because we can’t read the topic of a dispute off from the literal meaning of the disputed sentence

From blame to praise, if an opponent rejects the very idea of something being verbal on the grounds that it can be ascribed to near anything, then the method of elimination provides a solid response: The questions that cannot be raised, the claims that cannot be made, and the disputes that cannot be continued after some constituent term, T, has been eliminated, are all due to T.

1.5. The method applied to philosophical disputes

The problem of verbal disputes is everyone’s problem, and not unique to philosophy, maybe even to the contrary; philosophers might be better equipped to deal with them than anyone else. Think, for example, of taxonomical disputes in biology.25 But in philosophy, there is a sense in which their roots are harder to pull out of the ground. In this dense and tangled section, the method will be critically discussed as applied to the go-to case study that is free will, which many readers will be familiar with, which near every philosopher has had a say on, and which maybe more than any other philosophical problem has been accused of verbality; the question being “Is our will free?” A handful of cases where the word appears to be eliminable are discussed, before Chalmers’ special case of the method called the subscript gambit is introduced in the section after. The first step asks for the word “free” to be eliminated, and the second step asks us to look for a dispute that is part of that original dispute. In keeping with Chalmers’ suggestions, let’s say it takes us to the following four candidates:

a. Could our will have been different than it actually was? b. Are we responsible for our actions?

25 See, for example Dupré (1981).

c. Can we do what we want? d. Do we instantiate effects without causes?

The disputants can now go on to discuss one or more of the more specific (a-c), meaning the dispute most certainly has made progress. Whether the disagreement continues over both (a) and (b), or there is agreement on the one but not on the other, or on neither, there is progress all the same. But progress itself does not show that the dispute was verbal to begin with. It might be that both disputants take, say, (a) to be the continuation of the free will dispute, and even if they disagree on all the other alternatives too, they never conflated them with each other, for example. And even if there is disagreement on all the alternatives, it might still be that they themselves are found to be verbal as the method is reapplied. As we saw in the previous section, it is the diagnosive step that is the harder part. Chalmers discusses various outcomes depending on the possible stances the disputants might have on the subquestions, with reference to certain historical debates over the question of free will (530- 31, 531n). The discussion is for the most part oriented towards progress and clarification, while in the following I will be more concerned with the method’s ability to diagnose disputes as verbal. If there is agreement on all the subquestions, the dispute over free will is entirely verbal, and the elimination of the term “freedom” resolved it all. If there is agreement only on one of them, furthermore, the dispute can still be wholly verbal if the remaining three are considered exterior to the topic of the original dispute. And even if a disagreement is preserved, over (b) let’s say, the method can be reapplied eliminating “responsible,” which might for example reveal the further subdisputes b.1. Should punishments for wrongdoings be gentler? and b.2. Do people deserve praise or blame? both over which there might be agreement. If so, (b) was itself found to be verbal. How about (a), where there is no obvious candidate term to eliminate, but only a series of modal functors (“could have been;” “actually”) and the terms “different” and “will”? There is still a sense in which this question can have many meanings, and so give rise to verbal disputes, but without a clear candidate expression responsible for those differences. The whole sentence can of course be barred, and the disagreement there over stated in a different sentence, which would likely yield some progress, but it would not show that the dispute is verbal. Take the question “If you could travel anywhere, which destination would you have visited?” for an analogy. The fact that it can be made more precise and so have different answers depending on the different precisifications (duration, company, budget …) does not show that there are differences in meaning, but there is nonetheless a sense in which it is verbal with respect to something like “dream destination,” even if that word does not appear in the sentence. The term with respect to which (a) is verbal is harder, but, departing somewhat from Chalmers now, a suggestion is that all the modal terms together makes the question verbal with respect to them, or with respect to something like the concept of modality. Can these modal terms be eliminated, for example in favour of the term “determined,” so as to continue the dispute over (a) in form of “Is our will determined by

prior events?” Not likely, as there is probably modality built into it already, as an elimination of “determined” would have taken us back to “necessitated by prior events,” for example, and similarly with “Do we have alternative options?” as there is probably modality involved also here. Modality is as such a good candidate for one of Chalmers’ primitive concepts, 26 but I don’t think this in and of itself is the key notion responsible for the free will dispute. Even if it cannot be eliminated there is a sense in which it is not modality per se, but the type of modality that would give rise to a verbal dispute over question (a). Very roughly, a type of modality can be thought of as the type of reason for why a modal claim is true. The Stanford Encyclopedia of Philosophy entry “Varieties of modality” suggests for example nomological modality (“You cannot travel faster than the speed of light”), conceptual modality (“No bachelor can be married”), legal modality (“You cannot deduct your holidays from your taxes”), technological modality (“One cannot get from London to New York in less than one hour”), etiquette modality (“You cannot start a job application cover letter with ‘hey guys’”), and many more (Kment, 2017). So modality does not have to be eliminated, but only a particular reading of the modal notions in the disputed sentence. We can cling to modality, therefore, and take (a) to ask whether the states or acts of the will are necessitated by laws and prior events, or, as we might just call it, determined, the question being what type of laws and events it is necessitated by. There are still several candidate modalities the question of free will can, and certainly has, turned on:

a.1. Is our will metaphysically determined? a.2. Is our will determined by physical laws? a.3. Is it sociologically determined? a.4. Biologically or genetically? a.5. Determined by Fredudian subconscious goings on?

At least intuitively, if there is no residual disagreement, for example if one disputant had metaphysical modality in mind while the other had physical modality in mind, then the dispute is verbal. Even if there was no single obvious term with respect to which it was so, there is still a clear difference in meanings here of the modal notions. To be determined in a metaphysical sense, for example, is to be necessitated by causes, in other words that the causes could cause no other outcome. Or it can mean physical or nomological determination, whereby the state of the will is necessitated by physical laws and

26 However, see David Lewis (1986) who takes drastic measures in an attempt to eliminate it in favour of logical concepts and the concept of a possible world, which in turn is eliminated in favour of mere causal and spatiotemporal relations.

preconditions.27 Or it can be sociological, and biological, and so on. There is still a clear sense in which a dispute arising in virtue of these differences is verbal. In all, if the modal terms are understood with different modalities, it can lead to a verbal dispute over determinism, which leads to verbal a dispute over whether our will could have been different than it actually was. To briefly put this dispute into context, it is certainly a familiar picture that one disputant takes the relevant sort of modality to be nomological, while the other has something different in mind. The former may depart from the indeterministic Copenhagen interpretation of the laws of quantum mechanics to the conclusion that determinism is false, and say something like

One may legitimately wonder why worries about determinism persist at all in the twenty-first century, when the physical sciences—once the stronghold of determinist thinking—seem to have turned away from determinism (Anscombe, 1971, 24), while the other disputant takes the relevant type of determinism to be metaphysical and so unaffected by what scientists in Copenhagen have said about physical laws (for an example of this view, see Strawson 1994). If so they might agree that physical determinism (a.2.) is false, while agreeing that metaphysical (also called causal) determinism is true (a.1.) and so have a verbal dispute over determinism (question (a), approximately28), which in turn leads to a verbal dispute over the original question about whether we have free will. Now, if these steps from a series of modal functors to the modal notion of determination to, finally, a clarified set of determinisms do not count as elimination, then the method fails to diagnose disputes like this; disputes that are verbal not with respect to a particular term, but with respect rather to something like a concept. If it however does count as elimination, which I think it does, the method would be particularly useful in philosophy, where there is not always one contested expression, but where the dispute is verbal all the same. Returning to Chalmers, the above discussion of determinism leads us to a second possible complication, having to do with the way subdisputes relate to one another, and to the arguments provided therein. Even after elimination of “free will,” the residual disputes might relate to one another in a way that raises the question of whether they really are separate disputes. The residual dispute over whether we have moral responsibility (b), for example, might depend for agreement upon the opinions about the other residual dispute over whether determinism is true (a). This, however, is not a problem for the method, as

27 Physical determinism is an empirical hypothesis, while metaphysical determinism, in contrast, “has often been held to be a priori, a necessity of thought, a category without which science would not be possible." (Russell, 1914, 179) 28 “Approximately,” as there are other less widespread hypotheses than determinism by which (a) could fail, including fatalism (we are powerless to change the world’s courses of events), predeterminism (all information about any future state of the world has been present since the beginning of time, available for example to God or Laplace’s all-knowing demon), and necessitatianism (every truth is necessarily true). See Taylor (1962) and Dasgupta (2016).

the procedures that it prescribes entail no constraints on how the residual disagreement can be framed in the restricted vocabulary. The dependence of the one answer upon the other can simply be framed into a third, separate dispute. This link might for example come to a disagreement over whether a negative answer to (a) is compatible with a positive answer to (b): “One possible outcome is that the parties will disagree over a sentence such as ‘Moral responsibility is incompatible with determinism’ as part of the original dispute. If so, this is a prima facie indication that the dispute is nonverbal—though one may want to reapply the method to ‘moral responsibility’ to be sure”(531).29 To capture these relations and frame them into new disputes might be a non-trivial task, meaning that step 2 will involve some effort, and actually some philosophizing, but it is not in principle a problem with the method of elimination itself, because it was never supposed to be operated without any theorizing, only streamline it. The more complicated questions draw near when the eliminated word is seen needed again in one of the subdisputes’ arguments. I will now consider three ways in which the word “freedom” is not obviously eliminable. First, if there is agreement that we have moral responsibility, for example, but disagreement over whether it is incompatible with determinism, what if a disputant sees the need to draw “freedom” back in and argue:

To deny the free-will thesis is to deny the existence of moral responsibility, which is absurd...Therefore, we should reject determinism. (Van Inwagen, 1986, 223),

an entailment which the other disputant rejects. We have here a first purported entailment from moral responsibility to freedom, and then a second from freedom to indeterminism. Now, because the existence of moral responsibility according to Van Inwagen entails a positive answer to the original question about free will, and the truth of determinism entails a negative answer to that same question, the elimination of “freedom” is equally an elimination of this crucial entailment, it might be objected. However, the method of elimination can be applied on this argument, in order to see whether the disagreement can be captured by other means. Without the middle man “freedom” in it, this would just be an argument from moral responsibility to

29 To reapply the method, «moral responsibility» can be eliminated in favour of terms that designate degrees of moral responsibility, Chalmers adds:

For example, the parties might agree on ‘Determinism is compatible with degree D of moral responsibility’, ‘Determinism is not compatible with a higher degree D’ of moral responsibility’ (for example, a degree involving desert that warrants retributive punishment), and other relevant sentences. This outcome is a prima facie indication that the dispute is verbal, resting on a disagreement about whether the meaning of ‘free will’ requires more than degree D of moral responsibility. (531.)

To allow the substitution of a term “X” for “degree D of X” opens up for the method to diagnose and solve a large number of further disputes, such as “How much certainty is required for knowledge?”

indeterminism directly, like before—or at least so I think Chalmers would respond. And if there is anything more to this disagreement than, says the method, moreover, it should be possible to state it without use of that term, substituted for example with “the ability to do otherwise.” Still, there is something about this argument that seems left out; but hold that thought until we return to this in a minute. Another way in which the key term can appear ineliminable is when it is used to make a point that relates a subdispute, like (a.2.), to the original question, as in

Indeterminism does not confer freedom on us: I would feel that my freedom was impaired if I thought that a quantum mechanical trigger in my brain might cause me to leap into the garden and eat a slug. (Smart, 2003, 63.)

A disagreement over such an argument does not appear any more verbal than the key term “freedom” appears eliminable. Neither moral responsibility and certainly not anything along the lines of indeterminism would be a possible substitute. Maybe instead of “Indeterminism does not confer the ability to do otherwise on us,” it could say “Indeterminism does not confer on us the ability to be the ultimate origin of our choices,” so that the dispute would continue over subquestion (d). But there is still a lot that seems left out. And for a third and final example, if we merge Smart and Van Inwagen’s two arguments, we get what is sometimes called the standard argument against free will. Here in the words of P.F. Strawson, free will can be seen as an inherently self-defeating idea if

…we consider the consequences either of the truth of determinism or of its falsity. The holders of this opinion agree with the pessimists that these notions lack application if determinism is true, and add simply that they also lack it if determinism is false (1962).

In other words, freedom is incompatible both with determinism and indeterminism, and either the one or the other has to be true. The standard argument avoids having to defend either determinism or its opposite, because the same consequence follows from both, namely, that the will is not free. “Freedom” eliminated, any further disagreement over this might, for example turn on whether the same dilemma holds for morality (b), or of whether it holds for free action (c). Or maybe it can be cast in terms of whether there is a third alternative in between randomness and determinism, over whether the will is causa sui—a self-caused cause and the ultimate origin of a causal chain (d). Still, much of the point of the argument seems eliminated away. In all these three cases, important parts of the disputes that were eliminated appeared impossible to capture without the use of “freedom.” How can this residual disagreement be understood? In defence of the elimination method, maybe arguments like these can be understood as involving a standpoint about which of the subdisputes best captures the original disagreement over free will. Smart’s can maybe be understood as an argument that the question of free will is not a question of whether we have the ability to do otherwise, or of whether our will could have been different than it actually was, for example; something like (a), and (a.2.) in particular. And like Smart’s, the standard argument can also be understood as saying that the subdispute (a) is irrelevant, because it makes no difference to the original question whether there is agreement or disagreement here. I suggest they can be seen in analogy to Chalmer’s points in the beginning of this

chapter that a dispute can be verbal even when disputants mean the same by their terms— “externalism does not confer subsatntivity on us” he could have said in the same tone of voice. Chalmers and Smart are making the analogous claims that the phenomenon or the desirable property in which we are interested should be captured in this one way rather than in that other way. Where Chalmers earlier says “We might agree that disputants typically mean the same, but it does not solve the problem of verbal disputes” Smart says “We might agree about indeterminism, but it does not solve the original problem about free will!” The fact that the key term is ineliminable, it can be concluded from this, is only because this residual disagreement would fit in someplace between the method of elimination’s step 1 and 2, at the point when the disputants are looking for a continuation of the original dispute. But don’t take my word for this. The drawback with this response, moreover, is that a lot of non-trivial philosophical argumentation and many important points will be classified into this intermediary phase where the residual disagreement is sought to be identified, which means that the method will be far harder in practice than what is said on the tin. As the discussion of these three cases shows, the elimination of the key term is not always straightforward, but probably possible, even though it might transform the questions by quite a bit. However, it is impossible to give these cases conclusive diagnoses because the disagreement will in most of these cases continue into more than one subdispute, and so would be hard to keep track of. And because it depends on the disputants which subsentences continue the disagreement, and since that is something they might disagree over, it cannot be mechanically determined which disagreements are continued into which subdisputes. Having said that, the method yielded some progress by the fact that the philosophical questions were made more precise. But apart from those elements that fit into the subquestions (a-d) considered initially, there are probably many aspects of the above three points that the above transformations failed to encompass. My best suggestion was that these elements can be captured as disagreement over the continuation of the topic, that they are making points about what it is that is being discussed. In conclusion, it is far from obvious what constitutes a continuation of the disagreement. They are both difficult to find, and not always agreed upon.

1.6. The subscript gambit

Many of the above difficulties can be attributed to the fact that many disagreements turn, at least in large part, on disagreements over what the proper continuation of the topic is. If the key term cannot be eliminated because the disputants disagree over what they are discussing, or what the thing they are discussing is, such disagreements may be addressed in terms of questions such as “What is freedom?” Philosophical questions are often framed in this form, and many philosophical disputes will turn out that way. You may have heard “You are not talking about freedom, but something else.” The phenomenon can also be described as the dispute taking a nominal turn, ceasing to be cast in adjectives, adverbs and verbs so that and term in its predicate form (“free,” “freely” “perceive,” “represent,” true...”), can be described to be eliminated to reveal a dispute where it instead functions as a subject (“freedom,” “perception,” “representation,” “truth…”). How does the method deal with such cases?

It may well be that a disagreement over Van Inwagen’s argument comes down to a disagreement over something like “freedom is the ability to do otherwise,” for example, or that a disagreement over Smart’s argument comes down to a disagreement over how freedom is connected to the self, such as “freedom is the ability to ultimately originate one’s choices” (since leaping into the garden to eat a slug, would doubtfully count as a choice of one’s own when it is caused by a quantum trigger in the brain), or that the standard argument also rests on the conception that freedom is for the will to be neither determined nor undetermined (which is impossible, and therefore there is no freedom). Though it would be an undertaking of its own to put forth such interpretations, the apparent complications that these three examples pose for the method of elimination may—and this is a big “may”—make more sense if they are understood as turning at least in part on disagreements over what freedom is. If that is the case, Chalmers has a ready response to all three. He calls questions of the from “What is X?” Socratic questions, and introduces a special case of the method to deal with them, called the subscript gambit (532). I quote at length:

Suppose that two parties are arguing over the answer to “What is X ?” One says, ‘X is such- and-such’, while the other says, ‘X is so-and-so’. To apply the subscript gambit, we bar the term X and introduce two new terms X1 and X2 that are stipulated to be equivalent to the two right-hand sides. We can then ask: do the parties have nonverbal disagreements involving X1 and X2, of a sort such that resolving these disagreements will at least partly resolve the original dispute? If yes, then the original dispute is nonverbal, and the residual disagreement may serve as the focus of a clarified dispute. If no, then this suggests that the original dispute was verbal … For example, in the dispute over free will, one party might say, ‘Freedom is the ability to do what one wants’, while the other says, ‘Freedom is the ability to ultimately originate one’s choices’. We can then introduce ‘freedom1’ [=our freedom 3] and ‘freedom2’ [=our freedom 4] for the two right-hand sides here and ask: do the parties differ over freedom1 and freedom2? Perhaps they will disagree over “Freedom2 is required for moral responsibility”, or over ‘Freedom1 is what we truly value’. If so, this clarifies the debate. On the other hand, perhaps they will agree that freedom1 conveys a certain watered-down moral responsibility, that freedom2 would be really valuable but that freedom1 is somewhat valuable, and so on. If so, this is a sign that the apparent disagreement over the nature of free will is merely verbal.30

Since many philosophical questions take this socratic form, and the disagreement over a vast number of other questions might turn on something like that, the subscript gambit has some powerful implications. In the above two examples, my best explanation of why “freedom” was eliminable was that the disputes centred on an apparently substantive

30 The idea that watered-down variants of contested terms can be introduced is similar to the idea in the previous footnote of allowing for degrees. The actual verbal disputes over Pluto being a planet, XYZ being water, and whales being fish can all be solved by replacing the disputes terms with such as “planet-like,” “fish-like,” and “waterly,” for example. Again, in philosophy this might be especially useful.

disagreement over which subquestion was the original dispute’s proper continuance, which in turn can be understood as a disagreement over what freedom is. If this is so, the subscript gambit opens for even these types of disputes to be diagnosed verbal. In more detail, let’s say the disputants are disagreeing over which of the subdisputes (a-c) is the proper continuation of the dispute over whether we have free will. Now, we transform them into socratic disagreement by inserting “Freedom is…” in front of the right hand sides of (a-c): “Is freedom the ability to do otherwise?”, “To be free is to have moral responsibility;” “Look I can move my hands. That is all it takes to be free.”, “We are free if and only if we can instantiate uncaused effects”, or just “You are not talking about freedom.” If the disagreement over “freedom,” or over the topic freedom, turns on socratic disputes like these, which I think would be a likely outcome, then the subscript gambit can be applied, in order to check whether the disagreement evaporates. Corresponding to (a-c) respectively, we obtain

the alternative options-freedom, the moral responsibility-freedom, the freedom in the sense of free action—simply, and the ultimate originator (causa sui) freedom.

Alternatively just:

freedom1, freedom2, freedom3, and freedom4.

Reapplied to these further subquestions (a.1, a.2, b.1, b.2…), it might take us to a dozens of further freedoms (freedom 5, 6…), possibly corresponding to any or a large number of turns that the further arguments might take. And if no disagreement is continued into the various questions and claims that can be stated with this brochure of freedoms, then by the subscript gambit’s last step, the original dispute from which they were conceived receives the diagnose. It is a dispute arising out of a disagreement over what freedom is. So what justifies this last step? What is the subscript gambit’s warrant to diagnose socratic disputes? More precisely, why does the loss of a disagreement by a pair of subscripts show that the original dispute was verbal? It’s a question that will be pressed further in chapter 2 and 3 when I discuss objections to the idea of philosophy being verbal, but for now, the following two intuitive reasons can be given. If we start by asking what socratic questions really are asking for, then a natural reading will be something metalinguistic. For one, if the disputants were unwilling to give up the term for a subscripted version thereof, it’s a strong indication that something hangs on the way language is used, as what else do they sacrifice than a manner of speaking? For a second, if we try to state the question “What is X?” in another way, then at least intuitively, the most natural substitute for the copula “is” would be “means;” and the most natural substitute for “X” would a noun, “’X,’” which mentions the word “X;”

and so the best candidate for the continuation of the question at large would be “What does ‘X’ mean?” That these two questions are often used interchangeably in ordinary speech adds plausibility to this, and as we saw in the previous section, disputes over such questions are what Chalmers calls metalinguistic. Should it even turn out that the metalinguistic reading is not the correct one after all, there is an even higher likelihood that whatever turns out to be, it will be metalinguistically mediated at least. Either way, since both metalinguistic as well as metalingustically mediated subdisputes were, as we saw above, taken to entail that the dispute is verbal, this provides at least an intuitive justification for the diagnose, which suffices for the time being, before these issues will be tackled more thoroughly in the next chapter when we look at a defence of irreducibly socratic questions. In all, the subscript gambit diagnoses socratic disputes positively when the disagreement is impossible to state after subscripts are applied. In addition to the likelihood that many philosophical disputes will boil down to subdisputes of the “What is X?” form, there are also a host of philosophical disputes that start out that way, of which Chalmers mentions: What is free will? What is knowledge? What is justification? What is justice? What is law? What is confirmation? What is causation? What is color? What is a concept? What is meaning? What is action? What is life? What is logic? What is self- deception? What is group selection? What is science? What is art? What is consciousness? And indeed: What is a verbal dispute? (532.)

I think that the philosophical literature over almost all of the questions in the last paragraph is beset by verbal disputes, in a fashion that is occasionally but too rarely recognized. Of course each debate involves an admixture of substantive elements too (532n). it is also added in a footnote. The remainder of Chalmers’ discussion consists in applying the method to some disputes he thinks are largely verbal. It should be clear by now that a definitive diagnose of a dispute will require detailed information about the dispositions of the disputants, meaning that he can only provide conditional diagnoses, like my above discussion of free will that was fit to burst with ifs and thens. But with a basis in what he takes possible disputants to likely agree and disagree on, maybe warranted by a long career in the field engaged with these disputes himself, equipped with the method of elimination and the subscript gambit, some general diagnoses are made. Besides free will, and disputes over socratic questions, the ones he considers to be verbally beset if not wholly verbal include the dispute over the semantics/pragmatic distinction, the dispute over the formulation of physicalism about the mental, and the dispute between internalist foundationalists versus externalist reliabilists over the nature of epistemic justification (532-34). We can stop for a second on what he says about physicalism, because there is something odd about the idea of socratically disputing what the –ism physicalism is, as the answer to that can be whatever we want it to be, “physicalism” being just a piece of terminology that names a proposition about the mental. Have philosophers completely lost themselves by debating the meaning of a word that they themselves have made up, believing there is a definite answer? With reference to actual disputes on the issue, Chalmers applies the subscript gambit to reveal issues that are sociological and normative, such as “Physicalism1 is what people in a certain debate are concerned with,”

“Physicalism1 is the more important issue” and “Physicalism1 is what matters for purpose X” (534).31 They qualify as a normal subdisputes that are neither metalinguistic nor metalinguistically mediated, but does this mean that the dispute over what physicalism isn’t verbal? It is clearly a mere terminological issue what physicalism is, and if I understand Chalmers right here, and the reason that it can be deemed a merely verbal dispute—at least merely verbal with respect to philosophy—is that its subdisputes somehow are irrelevant. What is the justification for deeming them irrelevant? To fill this little hole, I think the reason is that these subsentences are irrelevant is that they are about various propositions about how the physical relates to the mental, any yet they are not themselves not saying anything about how the physical relates to the mental, because they aren’t saying anything about these propositions’ truth or falsity. In other words, they might be substantive relative to some issues that are sociological, etcetera, but not to the first-order issues that philosophers of mind are interested in; as such, they are not really philosophical: “if these are the residual disagreements, then one can focus on these issues, putting the debate in the sociological or normative realm where it belongs.” (534). Chalmers advice to set these disagreements aside is a prescription and not a diagnosis, although a dispute over such a question obviously is verbal with respect to first- order issues in philosophy. Whether the method can or should classify them as verbal is another question, it depends on which types of possible subdisputes can be stipulated to disqualify a dispute from constituting a proper continuation of a topic, much like metalinguistically mediated topics were stipulated out in the previous section. What going on in the physicalism dispute is quite a transparent matter , and I’m very certain Chalmers is with me in saying that nothing hangs on the question of whether this is verbal philosophy or just misleadingly framed sociology. As we also saw in the previous section, unless a more utilizable way to rule out irrelevant issues is found in Chalmers’ extended theory, our intuitions about relevance can get us a long way. Similar points about substantive but irrelevant subdisputes will be frequenting the coming discussions, and both the Pluto case as well as the Physicalsim case make good examples for reference to this phenomenon. From this discussion it can be concluded that a very frequent type of subdispute is the socratic type because many disputes, including metalinguistic ones, can be expected to turn on disagreements over what something is. In many of these cases, I doubt that the disputants would agree to give up the term, or agree to apply subscripts to it, in which case it is likely that they are verbal.

1.7. “Answers are answers”

31 It’s a type of subdispute that will examined closer in 1.7. and 2.4.

Arguments against philosophy as verbal disputes can either target the notion itself, or they can make it harder to ascribe it to disputes, limiting their scope as a consequence. Or, a third strategy is to accept the notion and its ascription to some dispute or set of disputes, but limit any unwanted consequences that ascription has for those disputes, as the argument from Plunkett in the last chapter. Disregard the first type; from what has been said the notion is clear.32 Consider instead the following objection which belongs to the third type. This line of resistance, which in a footnote is ascribed to Jonathan Schaffer and Timothy Williamson, targets the consequence of diagnosing a dispute as verbal that it is pointless. I think this objection and Chalmers’ answer to it are key to understanding the theoretical background picture of the two methods we have been looking at. The objection says in short that it does not matter, or it is uninteresting, whether a dispute is verbal or not. We would like to get the questions right and to answer them, and this process should be unaffected by theories about subquestions and differences in meaning (536). If the dispute that’s ascribed verbality is the free will dispute again, the objection would say that

disputes are disputes, and questions are questions. Even though we agree on the answers to all the sub-questions, there is still value attached to getting the original question right—“Is our will free?” Various theories about verbal disputes might show that this is verbal, but so what? Answers are answers. We still want to know them. Therefore, since there is value in getting the answers right, the dispute might be verbal but still not pointless

What can be replied to this? Chalmers’ reply starts from the premise that the significance and value of getting the original question right reduces to the significance and value of getting the subquestions right. So it enjoys its value only derivatively on one or more of the more precise questions that it can be found to express once senses are distinguished. To illustrate, return to the different senses or subquestions (a-d) of “is our will free?” Well, that depends upon what you mean by “free?” a. Could our will have been different than it actually was? b. Are we responsible for our actions? c. Can we do what we want? d. Are we able to instantiate effects without causes?

Suppose that the question is verbal, so that the disputants agree for example that we can do what we want and are responsible for our actions (b,c), and they also agree that the will cannot instantiate effects causa sui or be different than it actually was (negative answers to (a,d). Once we have distinguished these four senses and there is agreement on all points, then there is no value connected to getting the original question right. The

32 See, however, arguments of this type in Graham 2014

question of free will is interesting and valuable only in virtue of those subquestions themselves being interesting or valuable; the value of being right is then transferred over to the two subquestions that preserve the disagreement. It’s a key element in the background picture; as it were very loosely, there is nothing intrinsically valuable about a certain one linguistic mode of presentation over the other. Why should the words matter? It might be replied to this that there still are residual disagreements here that cannot be reduced to any of these subquestions. Recall the difficulties that were encountered when attempting to eliminate “freedom.” One of the parties may say “You are right that we can do what we want and that we are morally responsible, but that doesn’t mean we are free.” And the other party might reply “I don’t care whether determinism is true or whether we can instantiate effects without causes; we are free because we can do what we want, and because we are responsible for it.” It is not an unrealistic scenario that the disputants still disagree on the original question even if they agree on all the subquestions, which suggests that the value of getting the original question right cannot be reduced to the value of getting right the subquestions. “You aren’t talking about “freedom” they might say. However, with the method of elimination and the subscript gambit in hand, I think this response can be refuted. The apparently irreducible value of the original question can most likely be captured by sentences such as:

e. Only (a) is the interesting continuation of the original question, f. Only freedom 1 is what we really value, or g. Only the ability to instantiate effects causa sui deserves the label “freedom.”

Now, even if these conceptual questions are valuable, substantive, and significant in and of themselves, that value cannot be transferred to the original question. Just as with planet Pluto and the Martini, it might be a substantive sub-question who the leader should be of the astronomical society, or whether the purist is a well-informed man, or whether purism about martini is a doctrine with which also the Greek community should associate, but since these are too distant from what the original question asks for, they are not able to help it with any substantivity. The reason is that, as was discussed above with the physicalism case, none of these subdisputes would touch on the first order issue that’s at stake; while the answers to (e-g) will have implications about the answer to “Is the will free?” they do so only in conjunction with answers to (a-d). This means that the value of the original question still depends on the value of all the subdisputes (a-g) together. Unless the question of free will is a question about language to begin with, none of these metalinguistic, normative and psychological issues are relevant to the first-order issues at stake. Assuming it’s not, the answers-are-answers objection does not show that the ascription of verbailty is uninteresting.

If we are concerned with language, for example, to respect tradition or to facilitate communication, or if we are linguists or philosophers of language, then it may be important to get the answer to this question right. But if we are not, then there is no importance in this residual question (536).

Even if one of the disputants are right on one of these residual metalinguistic questions, so that it in some way entails that it is true that we do not have free will, then Chalmers

argues that this true belief would not “correspond to any significant increase in our understanding of the world. And in practice, sensible philosophers and scientists who are not concerned with metalinguistic matters are almost always willing to set aside these residual questions as insignificant” (537-8). He also has this to say about a possible further opposition:

Perhaps there could be an intellectual value system that gave true beliefs about roundness or physicalism or planethood significant nonderivative value, over and above the value of the other true beliefs in question. But such a value system would seem fetishistic. (537).

Why did the objection fail? It tried to appeal to the intrinsic value of being right, but Chalmers reply showed that what has value is not to be right about sentences, but about the topic under discussion. When a question has several interpretations, subquestions follow, and it is being right about those subtopics that matters. So there might well be intrinsic value in being right, but the bearers of that value aren’t sentences, but what is said by those sentences. That is at least one explanation. As noted in the introduction, when a dispute is pointless, either

(1) the disputants are discussing different topics, or (2) they are discussing the same topic, but it is itself pointless.

They might be disagreeing indirectly over such as (e-g), so that (1) holds, but this objection shows that they are mere issues of linguistic modes of presentation, and so pointless with respect to the first order issue at hand To sever the verbal-pointless link, as chapter 3 will also try to, (1-2) will have to be tackled directly. What can be concluded is that to be substantive with respect to (first-order) philosophy, linguistic, psychological, sociological, and normative (however, see chapter 3) cannot help. If subdisputes like (e-g) make up the sole continuation of the dispute, the dispute is not philosophically substantive.

1.8. Which disputes are verbal?

Verbal are those disputes that disappear after elimination, Though the foregoing discussion may have raised more questions than it has answered, the picture that emerges is that the method of elimination can reveal either a (A) normal, first-order type of subdispute, (B) a metalinguistic/metalinguistically mediated type of subdispute, (C) a socratic subdispute, or (D) a normal but irrelevant subdispute. In the same order, “Do we have free will?” can reveal disagreement on “Do we have the ability to do otherwise?” “What does ‘freedom’ mean?”, “What is freedom?” and “Which type of freedom do we value?” There is also the possibility that it reaches (E) a bedrock dispute, which we did not go into. To assess the first type for verballity, the method of elimination can be reapplied, as have been done in this chapter. The second type is as we saw a guarantee that it’s verbal with respect to philosophy. And to assess the third, use the subscript gambit, which either reveals a further non-socratic dispute, a metalinguistic one, or no disagreement at all. Unless it’s the former, the dispute receives the diagnose. This section addresses the possibility of diagnosing verbal disputes and the theoretical background that supports the method’s

perscriptions, especially in relation to the conclusion from 1.4. about the unclear conditions under which something can be said to be the same topic. Which disputes are relevant to philosophy, and why can’t metalinguistic and ineliminable socratic disputes be substantive? Why is metalinguistic philosophy (i.e. the type of philosophy we are doing right now, mentioning words rather than using them) not relevant to the issues of first-order philosophy (i.e. philosophy where the words are mentioned rather than used)? And why do not questions like (e-g) in the previous section count as philosophically substantive? Well, it depends upon what you mean by “philosophy.” Philosophy can be so many things, and consists of many traditions each with their own methodological maxims, not to mention the practices that are idiosyncratic to each follower. In the next chapter I will discuss some positions that might come in conflict with Chalmers, however, time being I think there is sufficiently overlapping consensus on which questions are relevant to which other questions, and maybe also on how metalinguistic issues relate to those questions. So Chalmers’ stance on this can at least be assessed with respect to very general conceptions of philosophy. Here is a set of examples of first-order subjects and first-order philosophy from Harvard Philosophy Writing Centre brochure “A Brief Guide to Writing the Philosophy Paper” by Simon Rippon (2008):

But what is philosophy, and how is it to be done? The answer is complicated. Philosophers are often motivated by one or more of what we might call the “Big Questions,” such as: How should we live? Is there free will? How do we know anything? or, What is truth? 33

Let’s just suppose Rippon’s description is representative. In this sample there are both “What is…” questions and normal first-order questions, but only the first count as substantive in Chalmers’ theory while it is estimated that debates over near all these socratic questions are beset by verbal disputes (532n). As we saw, because they are presumably just asking for meanings. But then again, there might be that the socratic question “What is truth?” actually is just a more general way to raise or introduce a more specific subquestion, as, for example “Is truth a monadic property?” which might not be verbal. If that is the shared topic between the disputants, then this socratic question is not verbal. So Chalmers’ view is so far consistent with this arbitrarily chosen general description of philosophical method. When it comes to Rippon’s second question about how philosophy is to be done, moreover, he seems to affirm that metalinguistic disagreements can be philosophical:

33 Here are some more further examples of first-order topics from the webpages of the University of Oslo’s philosophy department: the nature of existence, the notion of truth and the possibility of knowledge, the nature of science, the nature of the human mind, the standards of logic and argumentation, the standards of beauty,how humans should live and act, what duties we have to one another, the best political organization and the ideal human.

Philosophers insist on the method of first attaining clarity about the exact question being asked, and then providing answers supported by clear, logically structured arguments. (ibid, my italics.)

Does that mean that at least the first of these two phases would count as verbal by Chalmers theory, since attaining clarity about the question just is to conduct linguistic analysis or discuss the value of different questions, like (e-g), and so on? The first phase will indubitably involve disagreements over what words mean, or disagreements about which questions are valuable and so on, so are they verbal? No— it is only with respect to first-order disputes about freedom that questions about “freedom” are irrelevant. Conceptual analyses are technically speaking metalinguistic, but it is only disputes over conceptual analyses that are verbal. Chalmers theory is still compatible with Rippon conception of philosophy, because these things can be substantive if they are done in what he calls the formal mode and not the material mode” which is to say, the words are mentioned rather than used (542). Besides use/mention, the distinction is similar to the distinction between first-order issues and metalinguistic issues, and Plunkett’s distinction between object level and representational level issues (2016). The method of elimination is itself a method that belongs in this stage, and same with the response “Well, it depends on what you mean by “freedom?”34 What’s more, Rippon talks about how the big questions relate to what can be likened to Chalmers’ framework of subdisputes:

While trying to answer Big Questions like those above, philosophers might find themselves discussing questions like (respectively): When would it be morally permissible to push someone into the path of a speeding trolley? What is a cause? Do I know that I have hands? Is there an external world? While arguing about these questions may appear silly or pointless, the satisfactions of philosophy are often derived from, first, discovering and explicating how they are logically connected to the Big Questions, and second, constructing and defending philosophical arguments to answer them in turn. (ibid)

Again, “What is…» questions are among the paradigm examples even of these more precise subdisputes and not only the Big Questions. True, “What is a cause?” appears to ask for something metalinguistic, i.e. the meaning of “cause,” in which case a dispute there over would be verbal with respect to “causation.” But not all socratic questions, we can

34 However, once a question, say, “Are we free?”, is elevated into the material mode it quickly becomes clear that “What does ‘freedom mean’?” is a separate, preliminary issue, and that any material mode dispute that arises in virtue of that disagreement is unmistakably verbal. And again, it may be that this preliminary metalinguistic phase just resolves the dispute, or continues “all the way down” without getting to the real subject-matter, in which case the dispute over “Do we have free will?” is found to be verbal; they agree about all the non-linguistic facts. If everybody agrees about the non-linguistic facts about a question it is trivial. Deflationary outlooks on a philosophical question or on a set of philosophical questions are those whereby those problems are resolved by the time they get out of the womb of this opening stage. Chalmers (2012).ascribes a very general deflationism about philosophy to Rudolf Carnap

suppose, are reducible to metalinguistic disputes or to nonsocratic subdisputes like “are there uncaused events?” as we saw in 1.6. and 1.7. Sometimes the disputants would resist the subscript gambit and instead of disagreeing on “are there uncaused-1 events?” say things like “Only causation-1 is causation.” and “You aren’t talking about causation, you are talking about bogus-causation!” or “schmausation” with the derogatory sch-prefix added instead of the neutral subscript. You might have experienced similar turns of argument. Those disputes centre on what we can call distinctly socratic questions; irreducible disagreements over “What is …?” questions, more precisely disagreements where the disputants refuse to eliminate the term, apply subscripts or enclose it in quotation marks. Since the subscript gambit as a matter of premises demands that there are none, Chalmer’s theory of verbal disputes classifies distinctly socratic questions as verbal. In all, the extent of verbal disputes in philosophy by Chalmers’ theory depends on what the subdisputes are about. Verbal are all those disputes that come down to disagreements about words, disagreements about irrelevant subjects, or to distinctly socratic disagreements. By way of speculation, many philosophical disputes are like this, in which case, a lot of disputes in philosophy are verbal—still, these verdicts can only be given on a case-to-case basis in the context of particular disputes. This is where the method of elimination enters, so how does it do in actual disputes? From approval to blame, the prospects that the method of elimination can return definitive diagnoses seems bleak, as the parthood relation issues few or no directions on what constitutes a continuation of a topic, and the theory at large contributes only by ruling out metalinguistic and socratic questions as well as disputes over topics that belong to sociology, and other obvious derailments, all of which are contestable. Apart from these directions, it is still by and large up to the disputants what constitutes a continuation of the original topic and what doesn’t, and so whether the original dispute should be dissolved or not. Philosophers like to debate each other, so I take it to be an extremely unrealistic outcome that elimination leads the disputants just to lay down their weapons and agree that it was all due to language. And if they don’t agree by the first disambiguation, then it is even less plausible that this will happen at later points in the process—it is very rare to hear “OK, we might have nomological determinism, but biological determinism suffices for degree D of moral responsibility, so it turns out we are not disagreeing after all.” For every new application of the method, there will always be some dispute or another that continues their disagreement, and even if the disputants are rational and with a conscientious attitude, it might wander quite some distance away from what the original topic appeared to be. In the process, a lot of progress will have been made, and interesting nuances in their views and in what they associate with linguistic expressions will have been uncovered, but when the disputants stop and ask: “Are we still discussing the same thing as what we started out with?” the answer, I suspect it, will in most cases be “No,” because a significant part of what the disagreement consisted in, I take it, regards precisely what that topic is. A myriad of related disagreements will have risen to the surface, but the disagreement over the original question will still be standing. Large scale diagnoses are even more difficult. For every (non-verbal, non- metalinguistic, and on the face of things relevant) subdispute, the method can be reapplied, and then reapplied again, and again without a stopping until, maybe, bedrock level. So at least after a while it would get quite complex. The fact that for most verbal disputes the disagreement would split and continue under several equally prominent subdisputes adds

to this difficulty. There is probably also the possibility that an opinion on a socratic question takes the form of a conjunction, such as “to be free is to be X and to be Y.” It can be freedom 2 and 3, for example, so that freedom presumes positive answers to both (b) and (c).35 And why not also suppose that that they consider free action (freedom 3) to be incompatible with some but not all types of determinisms, so that it is ruled out by negative answers to and all the subdisputes of (a) also, except for, say, (a.2.9.1)? At a certain point, following the method will get more complex than regular philosophical disputing, and at this point they will certainly get far more difficult to diagnose. This increasingly complex map of branching disputes with an ever shrinking vocabulary would also differ from the one pair of disputants to the next, because the parthood relation as it stands more or less leaves it up to the pair disputants what is a subdispute of what, making it practically impossible to diagnose the larger scale, longstanding philosophical debates, let alone the philosophical problems themselves. And importantly, while the above treatment of the free will dispute might have clarified many things, the original question, at least speaking for myself, still feels unresolved. With the Pluto case, in contrast, it is easy to tell by our intuitive grasp of topic continuation that the proposed subdisagreements are too remote from the original pre- analysed dispute to lend it much substantivity, if any. And in the Martini case, it was clear that none of the salient subdisputes constituted any disagreement. But in philosophy, it seems, almost any disputed sentence will have a myriad of candidate disambiguations, and almost any disambiguation will reveal a host of new disagreements, and most importantly, due to the interrelatedness of the many philosophical questions and disputes that are out there, almost any dispute can be said to be a part of another—unless, that is, a lot of theory is packed into the rationality constraints and semantic theories behind the parthood relation that were mentioned in 1.4.. With so much packed into the notions of idealized reasoners and what Chalmers later calls “conceptual dynamics,” the relation is not very theoretically utilizable as it stands, and matters of parthood is by and large up to the disputants themselves. At least I suspect, a major part of most disagreement will turn on precisely the issue of which dispute is a part of the other. These features make philosophy hard, but it makes diagnosing the problems even harder. For the same reasons, Chalmers by and large leaves open the exact extent of verbal disputes in philosophy. Over and above the examples that include free will, the semantics/pragmatics distinction, the nature of justification, and, of course, Socratic questions, the clearest statement is that “many philosophical disagreements are at least partly verbal, and almost every philosophical dispute has been diagnosed as verbal at some point” (517). In contrasting his own view to that of the notorious critic of metaphysics Rudolf Carnap, moreover, in the book Constructing the World he says: “Carnap held that many key disputes, in philosophy and elsewhere, are verbal disputes. … I think that more

35 Thanks to Jørgen Dyrstad for this point.

philosophical disputes are substantive than Carnap did, but the dissolution of verbal disputes is nevertheless part of my metaphilosophical religion” (2014, 289). It doesn’t really matter for the primary theses of “Verbal Disputes” which exact disputes actually are verbal. Chalmers’ “metaphilosophical religion” isn’t as doctrinal as it is ritual. The focus is all on the practical step-by-step procedures for the dissolvement of verbal disputes, which can be put to actual practice in seminar rooms. Whether the method resolves the disputes or just continues all the way down, or actually spits out substantive disputes is up to the operators to decide.

The diagnosis of verbal disputes has the potential to serve as a sort of universal acid in philosophical discussion, either dissolving disagreements or boiling them down to the fundamental disagreements on which they turn. If we can move beyond verbal disagreement to either substantive agreement or to clarified substantive disagreement, then we have made philosophical progress. (517)

This solution-before-blame oriented policy relieves Chalmers much of the task of defending the idea of philosophy as verbal disputes, but “Verbal Disputes” will probably leave the reader with reinforced suspicions about the problem’s magnitude. In all, the theory tells us a lot about what a verbal dispute is, and a lot about how philosophical disagreements function, but at least given the difficulty of a utilizable account of the parthood relation and sameness of topic, its ability to tell us exactly which disputes are verbal is found wanting. This presentation has introduced many important ideas that will be of use in what’s to come, but there are several questions about Chalmers’ theory at large that will be left unattended. Let’s imagine where it had gone had the methods here prescribed been followed to the letter. For one, it can be asked if it would have reached bedrock at all, or just continued in circles, or maybe terminated at a dispute that is clearly not anything like bedrock, or maybe revealing verbal disagreements all “the way down.” For a second, if there are such things as bedrock disputes, are they the only substantive ones, the whole elimination process being but a preparation phase? And for a third, wouldn’t this process have been far harder than regular philosophical theorizing as it is usually carried out? The theories of verbal disputes just addressed are only the first green shoots of Chalmers’ extended metaphilosophical theory of what philosophy ultimately consists in, which is very ambitious, and with few or no rivals when it comes to fine details and systematization.36 We can finish the chapter with a glance at where he thinks it might lead in the end:

36 Compare it, for example, to Kevin Scharp’s theory that was mentioned introductorywise and which is forthcoming in his Replacing Philosophy, according to which most philosophical disputes result from concepts that are inconsistent, and that philosophy consists in replacing them with conceptually engineered concepts.

[T]he method of elimination can be seen as a method for resolving philosophical disputes into two basic kinds of disputes: first, disputes involving bedrock concepts (or at least, putative bedrock concepts); second, disputes over which concepts are bedrock. If we could agree on the answers to these questions, then answers to all other important philosophical disputes would be within our grasp. Sadly (or happily), this does not provide a silver bullet for solving philosophical problems: both sorts of residual dispute are as hard as any dispute in philosophy. (552.)

1.9. Conclusion

In the astronomical society, the dispute over whether Pluto is a planet is still strapping, and though its status as verbal is largely acknowledged, this is not because the meaning of a word has any intrinsic interest to the disputants. It is not uncommon that an individual sentence works as a bottle’s neck through which a mix of background disagreements purposefully can be channelled. The idea of the method of elimination is that removing the word would splinter these background disagreements into independent topics that can be disputed individually. Whether a dispute is verbal is therefore a function of (1) the disagreement on the sub-sentences, and (2) how much that disagreement can be said to be relevant to, to be the same topic as the original disagreement. This also enables an explication of the idea of verbal- or trivial questions, I can add: A question is trivial to the extent that disputes over its most prominent sub questions would have been verbal. However, such characterisations say very little, as it is practically impossible to investigate in terms of original disagreements and topics which disputes are verbal. These things are often indeterminable from the outside, especially in philosophy. As we saw, there might be uniform disagreement on many related points, maybe because the individual background disagreements are too small to be worth the dissevering, or there might be hidden agendas that can only be advocated in disguise. With all these things considered, it is impossible to diagnose verbal disputes because even what the topic is might be under dispute, perhaps especially in the case of philosophy. In an interview, planet advocate Alan Cern, whose team of astronomers had a probe underway to what that was at the same time downgraded to a dwarf planet, has an example of such a hidden agenda:

There was an astronomer named Brian Marsden who for decades had a grudge against Tombaugh [who had discovered Pluto]. He had public fights that many people observed. Tombaugh died in 1997 and Marsden went on a jihad to diminish his reputation by removing Pluto from the list of planets. He eventually found a way to do that at a convention of astronomers, a meeting with thousands of people … where the vote was taken.” (“2006: A Space Oddity – The Great Pluto Debate,” Observer, may 1, 2016)

“Tombaugh should not be credited,” we can suppose, is the sole reason for Marsden to engage in this dispute, even though the disputes he engaged in prior to the vote would not even mention Tombaugh, but turn only on more objective points along the lines of “Words should have the meaning that is most useful in science,” as suggested above by Chalmers, or maybe, “’Planet’ should track the natural kind planet” (Ludlow 2012).37 It would be his hidden agenda, which means he would in any dispute reject “Pluto is a planet,” and possibly even agree that Tombaugh should be credited. This is irrational behaviour, but it is only as easy to determine what the topic of a dispute is as it is to determine whether it arises in virtue of differences in meanings, or in beliefs about meanings. Apart from that, the concept of a verbal dispute is only vindicated. This chapter has looked at the framework and the theory about verbal disputes, and, two things can be concluded about them. First, the concept of a verbal dispute is a family resemblance concept consisting of various related notions, which can be used for several different purposes, and some of which are in conflict with Externalism about meanings, and some of which are not. A verbal dispute can be carried out explicitly when philosophers seek to attain clarity about a question, but verbally disputing can also be downright defective communication when disputants are merely talking past each other and mean different things by their words. And they can arise due to disagreements on many different levels, and over various types of metalinguistic disagreements. For example, they can arise in virtue of disagreements about what is the right conceptual analysis, or in virtue of disagreements about what the topic is, in virtue of disagreements in standpoints about words, or about what the community meaning is, which community it is being deferred to, or in virtue of disagreements about which questions are valuable, and so on. These subjects might be substantive in and of themselves, but when they are mediated by disagreements over meanings, there is a shared sense of pointlessness that unites them; Why disagree about mere words? Secondly, the theory side of “Verbal Disputes” which is derived from the heuristic oriented theory of elimination appears promising. Although the idea of shared topics remains vague, several valuable insights about language and philosophical method are at work, and very central among them is the empowering motto that linguistic investigation into philosophical disagreements does not have to be shut down by silentism, or just healthy skepticism towards meta-theorising, as was seen with “Answers are answers!” While aspects of philosophy like evidence, belief formation, content, and truth might remain outside the reach of linguistic explanation, Chalmers theory warrants optimism that philosophical practice can be “opened up” as it were, and given semantic explanations, at least as far as disagreements, questions, and linguistic exchanges in

37 Ludlow analyses this dispute as a metalinguistic negotiation which is substantive (see III, 1.2, ch.3), which might be right about certain parts of this dispute, but there is in any case a sense of irrationality about this dispute at large, about the way it is treated and fact that it is upheld, so it suffices to describe it as verbal, with the merely verbal connotation added to it.

philosophy. And, importantly, it does not appear to be a slippery slope; these issues can be addressed without suddenly standing deep in deflationary, pessimistic assumptions about the legitimacy of the discipline. As we now step back it can seem as if the method of elimination prescribes only trivial things, and that it is just commanding: “Be more precise in your debates—avoid vague and too general questions, like “What is X?” questions!” which is just something that is very trivial and is an obviously reasonable academic standard, quite simply. Of course, not separating topics, being too vague, using too vague concepts like “objective” and “relative” and so on is bad. So a critique of the method’s practical applicability might be that the same effect can be achieved simply by asking “Well, it depends on what you mean by ‘objective’?” However, this critique only is directed at the work it can do out at the barricades, defeating verbal disputes, helping philosophers find truth. In metaphilosophy, in contrast, the method’s practical applicability gives off a promising impression. It provides a rather theory neutral and intuitive vocabulary in which to analyse philosophical disputes, and though it might take a paper’s length if not more, I can clearly picture the framework being put to use in metaphilosophical arguments concerning actual disputes, either that they are verbal or that they are not.38 Maybe it can be useful; I will anyhow make use of it in the next chapter, and it provides a helpful framework. It is helpful the same way the host animal, in biology terms, is to the endosymbiotic organism that lives inside of it. I feed on the terminology and on much of the theory’s flesh. 39

38 See Dahlberg (2016) for a manner of argument that’s very much in line with the method of elimination. He argues against Eli Hirsch’ verbal diagnosis of ontological disputes that there are subdisputes (“ancillary disagreements”) that are substantive and expressible in a restricted vocabulary. 39 A note on the method used in the following two chapters. I use Chalmers’ framework, which has been laid bare in the better half of this dissertation in order to avoid verbal disputes. Where I deviate from the framework is that the theory of bedrock concepts is left out do decide what is a subdispute of what. To fill in the void, I instead rely on the reader’s judgements on particular cases as to topic continuation and relevance, cases I take to be generalizable to larger chunks of philosophical disputes.

Chapter two—Against the idea of theoretically conflicting conceptual analyses

Philosophy is difficult, and yet more difficult to diagnose as verbal, even equipped with the method of elimination. By the time an individual case has been analysed thoroughly enough to licence a verdict, it wouldn’t come as a surprise at least to myself if it consistently turns out to reveal substantive issues at the bottom of each dispute. From a sufficiently insider perspective, most disputes have to my experience at least felt worthwhile. At a certain level of complexity the original dispute and the question of whether it’s verbal would be too far out of sight by the time things are clearer. Still, that conviction doesn’t prove much in the way of defence against accusations of pointlessness coming from the outside. Philosophy is complex, and philosophers can always retreat into the complexity of their subject-matter. The suspicions nonetheless remain to be addressed that philosophy is largely or entirely verbal, and should be met from an equally outsider perspective, without reliance on the difficulty of diagnosing particular disputes. Does semantics leak into philosophical debates and philosophical debates leak into semantics? The way to motivate the magnitude of the problem of verbal disputes is to take a chunk of philosophy and argue that it is verbal, and consider the ideas that come in conflict with this idea and argue that they cannot provide support or vindicate that chunk of disputes. Which chunk do I have in mind? Philosophers often disagree because, vaguely, they subscribe to different analyses of the same concept, or, yet more vaguely, have conflicting beliefs about what the thing they are talking about is. From the above, a critical point is that many disputes might come down to things that are found to be a different topic from the philosophical topic, language above all, as when philosophers disagree about analyses like “What does ‘X’ mean?”, “What is X?”, and “What counts as X?” It has been a recurring theme—separate and dispose of them, they are irrelevant to the philosophical

question and give rise to verbal disputes. Let’s call these disagreements “Chunk,” they are the ones that resist the elimination method (short of eliminating the disagreement), and the subscript gambit. So define it very roughly as

Chunk = the disputes that come down to disagreements where either the use or mention of the key term, and synonyms, cannot be eliminated

It is not obvious, but highly intuitive that these are verbal. Still, there are many disputes where the key term cannot be eliminated that arise in virtue of disagreements about conceptual analysis, and related philosophical endeavours. Many of them are not disagreements in meanings, but disagreement sin theories, or so goes the objection. Disputes that arise in virtue of these types of disagreements (those inside Chunk) have been assumed to be verbal in chapter 1, and I will defend this assumption in the following, against suspicions like Why were they presumed to be mere disagreements in meaning? Why don’t they constitute substantive theoretical disagreement instead? The following chapter explores this gap, and some methodologies that represent this style of opposition are dismissed. This chunk of philosophical disagreements is a verbal chunk, I argue, by considering methodological outlooks that can legitimise Chunk or parts of Chunk. First, the most prominent part of Chunk are those disputes that arise in virtue of disagreements over conceptual analyses, and it is found that they are verbal when language is irrelevant to the philosophical topic of the dispute (2.1.). Then it is considered whether disputes that arise in virtue of disagreements over real definitions are verbal, and it is found that they are, because they are just disagreements over conceptual analysis again (2.2.). Then it is considered whether conceptual analyses can be not only about language but about reality too, and so be substantive nonetheless. The conclusion is that questions of meanings must be separated from questions of reality, or else one runs the risk of verbally disputing (2.3). Lastly it is considered whether meanings, or language in general, should matter to philosophers, or whether disagreements about them are pointless. I argue that they are irrational and facilitate verbal disputes, and that one should keep meanings out of the game by overt stipulative practices, because our words can mean just what we choose, following Chalmers’ “conceptual pluralism,” and Humpty Dumpty (2.4.).

2.1. Conceptual Analysis

Conceptual analysis we can define broadly as the investigation of a word’s meaning, so that a dispute is verbal if it arises in virtue of a disagreement over conceptual analyses. Which methodologies are conceptual analyses? Well, that depends on what I mean by ‘meaning.’ Chalmers ascribes his opposition to “1950s Oxford philosophy, Canberra-plan analyses of folk concepts, contextualism and related theses in epistemology, contemporary linguistics-based philosophy, and some parts of experimental philosophy” (540). But the target of these arguments may be able to be extended even more, so that conceptual analysis not only means metalinguistic analysis, but encompasses also psychological things like “concepts,” and also what is called folk theories, and also experimental survey-style, demographically conscious, approaches to analysis, informally called “Ex Phi,” and maybe also introspective endeavours, like using thought experiments and counterfactual

reasoning and assessing intuitions, and so on. Parts and strands of every one of these metaphilosophical approaches, I think, can be considered conflict with the idea of verbal disputes, but consider the target of the following arguments as a cluster with unclear contours surrounding “conceptual analysis.” The extent of verbal disputes in philosophy depends on which types of disputes arise in virtue of disagreements about conceptual analyses. So why are disputes arising in virtue of conceptual analysis disagreements verbal? If I was right in my argument in 1.3. that metalinguistic disagreements are scarce, then conceptual analysis disagreements are scarce too, which entails that those that dispute them are not really disagreeing. Just like in the “Answers are answers” objection, the argument is that we are interested not in language but in philosophical truth, which means socratic questions understood this way—“What does ‘freedom’ mean?”—would be asking for something irrelevant. While competing analyses of “freedom” may well be theories, both true and substantive ones, they wouldn’t be theories about free will or about whether we have it, but rather the way we think about free will, or talk about it, or the role that concept plays in our society, how we value it, or whatever it may be; nothing that’s philosophically substantive. Competing analyses of “freedom” aren’t competing philosophical theories. This is supported by Chalmers:

Again, once we have agreed on the first-order properties (including the value) of freedom1 and freedom2, it is hard to see that anything else in the first-order domain will rest on this conceptual issue. … And even before agreeing on the sociological questions and the like, there is no important philosophical dispute left to resolve, at least if we are not directly concerned with language. (535)

There might be substantive conceptual, linguistic, and psychological issues reflected in the original question, in other words, but nothing that concerns the first-order domain which is philosophy, meaning these issues can be eliminated. Even for philosophers of language and those working in the intersection with linguistics who are investigating how people are actually using language as well as subjects like meanings, context, and speech, the way in which these terms (“meaning,” “context…”) are used in the larger community is equally irrelevant. It may of course be replied that, yes, language is actually the thing that philosophy is directly concerned with when it investigates apparently non-linguistic concepts like freedom and perception and intentionality and consciousness and etcetera, contra what philosophers tend to regard themselves as doing. This would be a very controversial minority position, and will not be further attended to. Chalmers himself does not even consider this stance, presumably for just that reason. But if somebody prefers to interpret

many or some questions of philosophy as questions of psychology and about what words mean, then what use is there to argue against that?40 Next, following Chalmers, four ways are considered, in which the question “What does X mean,” or “How do we think about X” actually can be relevant to philosophy, with the conclusion that it can be only marginally so.

First, it can lend some evidence as regards which concept we really value, and so which questions are worth investigating (541).

Second, linguistic claims can sometimes take part in inferences to extra-linguistic conclusions. For example, if we know the psychological fact that people value freedom, and the linguistic fact that by “freedom” they tend to mean freedom 1, then assuming that our values provide evidence for what is valuable, some evidence is provided for the nonlinguistic claim that it is valuable for the will to have alternative options (542). (See discussion 2.4.2, below, “Are philosophers disagreeing about what it is they are disagreeing about?”)

Third, he mentions the “genius of our tongue” idea, by which there are deep insights inherent in language, maybe because of the long history of our use, which has built our wisdom into the distinctions and nuances therein (542).

Fourth is the normativity that pertains to language use, which closely relates to the topic of this thesis. How people use language can be relevant if it means “how people should use language” (542). (See chapter 3.)

Along with those arguments, we will go through each in turn and conclude that they are only marginal and cannot constitute the theoretical disagreement of philosophical disputes. To start with the first, while it is true that philosophers can use words the way they like to in order to investigate questions they consider important, what the public finds interesting can be relevant. For example, if nobody cares about compatibilistic freedom, but only the conceptions of freedom that we don’t have if the will is pre-determined, then arguably more time and energy should be spent on philosophising about determinism, and not compatibilism. However, this is would be a side issue at best, and certainly not a theory of what freedom is. Likewise for the ways in which linguistic or psychological premises can partake in arguments: The inferences will be of the above sort where some evidence is provided that freedom 1 is valuable, but it stops there. They will not lead to

40 Parts of experimental philosophy, certain parts of post-war Oxford philosophy, and parts of the Canberra plan indubitably involve some elements of views like this. Philosophical questions understood this way are just questions of language.

conclusions about what freedom is. Moving on, a lot can be said about the genius of the tongue view and how philosophy is connected intimately with language. But Chalmer’s responds by invoking the distinction between context of discovery and context of justification. While knowledge of language might be invaluable for philosophers in coming to new ideas and realisations, seeing faults, coming up with reasons for their views, and so on, one still needs to justify these claims. Unless language is the sole subject matter of the discipline, this type of activity can be useful in the introduction phase of a philosophical question, for example, when the problem is defined, and so on, but not when the topic is set and substantive claims need justification. And in principle, I would like to add, these nuances and deep distinctions inherent in actual language could in principle be reached from the armchair without the need to actually study language users. Fourth, the way in which the normativity of language use can be relevant for philosophy will be postponed until the last chapter. For now it should be mentioned that his treatment of this also leaves it with a marginal role, which is only contingent on our shortcomings as finite nonideal reasoners. (ibid.) Unless the “philosophical” question is just asking for lexicography or psychology, to begin with these different meanings cannot be theories about what freedom is, what perception is, meaning is, and etcetera. Whether it is freedom 1 or freedom 2 that corresponds best to the ordinary usage or way of thinking about it, simply does not matter. So in sum, while these residual issues are not misguided, they are largely irrelevant in the philosophy room, and so can not constitute theories of what freedom is. It is plausible that philosophy consists in large parts of conceptual clarification—“cleaning the cushions.” But all the activities that can be described thus should be done explicitly, or in what Chalmers calls the formal mode, as opposed to the material mode, and not through disagreements over questions like “What is X?” or “Are we free?”

So ordinary language philosophy [=conceptual analysis41] is not unimportant. But one needs to be very clear about its use. If ordinary language philosophy is practiced in the material mode, as it often is, it is easy to move too quickly from linguistic data to substantive philosophical claims in a way that masks potential verbal disputes. Things work best when ordinary language philosophy is practiced in the formal mode, making claims about ‘freedom’ and not about freedom, for example, and being clear when these claims are descriptive or normative claims about actual usage. If one wants to draw nonlinguistic conclusions while avoiding verbal disputes, one needs to be explicit about the bridge, and ideally one should be prepared to cast the conclusion without using the key expression. So one might argue as above that linguistic data suggesting that ‘free’ means X provides evidence about the value of X, for example. Doing things this way minimizes potential verbal disputes and maximizes clarity. (542-3.)

Only explicitly, for attaining clarity about the philosophical questions at hand, can these types of activities that fall under “conceptual analysis” be considered philosophically substantive. Disputes arising out of such disagreements are merely verbal, and testimony that clarity was not attained in advance of commencing of the dispute. If the dispute can’t be stated without using—and now also mentioning—the key term, it is verbal.

2.2. Real Definition

“We are disagreeing about the real property in the world, and not the word’s meaning” is an intuitive response which you might have heard when there is evidence present that a dispute is verbal. The next sort of problem cases are questions of the form “What is…?” If they are not just asking for conceptual analyses or stipulative definitions, can a whole lot of philosophy that appears to be verbal be vindicated on grounds that they are not disagreements in meaning as in “What does ‘freedom’ mean?” but substantive theoretical disagreements over questions like “What is freedom in reality?” In fairness, they appear to ask for something very similar to conceptual analyses, and yet, are seemingly not about semantics. Many or most verbal disputes can therefore be defended on the ground that they don’t arise from differences in meaning but differences in theories or conceptions of such as what it takes to be free, what it is to be free, or what the property or phenomenon freedom is. Data that supports this argument is when the subscript gambit fails and the disputants say things like “You are not talking about freedom, only scheedom” ,“Freedom- 1 isn’t really freedom”, and “Freedom-2 isn’t a meaning, but it’s a theory of what freedom actually is in the world.” §1. What can be replied? Well, one can of course point to the fact that it is an incredibly vague type of question, and ask what precisely it is about. The same question has also been asked by Richard Robinson in his Definition (1950) where he criticizes the question form “What is X?” and the corresponding notion of “real definition.” Here from a review of the book, by John A. Irving for the Philosophy and Phenomenological Research:

Dr. Robinson concludes that in the notion of a real definition, there is a confusion of at least twelve activities: searching for an identical meaning in all the applications of an ambiguous word; searching for essences; describing a form and giving it a name; definition a word, while mistakenly thinking that one is not talking about words; apprehending a tautology determined by a nominal definition; searching for a cause; searching for a key that will explain a mass of facts; adopting and recommending ideals; abstracting, i.e. coming to realize a form; analysing, i.e. coming to realize that a certain form is a certain complex for forms; synthesizing, i.e. coming to realize that a certain form is a certain part of a certain complex form; improving one’s concepts. The basic source of the confusion that has attended the notion of real definition is the occurrence in Indo-European languages, Robinson maintains, of the question form “What is X?”, “the vaguest of all forms of question, except an inarticulate grunt” (p. 190). (1956, 418).

However, that only goes to show that “What is …?” questions are verbally inclined, not that they are verbal. The search for an answer to what something is being such an ambiguous activity, creating favourable conditions for verbal disputes, does not rule out that the questions shares a (sufficiently) definitive meaning for the disputants, so that searching for an answer is a substantive activity nonetheless. Whether or not they are verbally inclined,

if there are such questions, i.e. questions that resist the subscript gambit and are not about meanings, then they can considerably limit the problem of verbal disputes in philosophy. Call these questions distinctly socratic. §2. To reply to this objection, we can say that since they are not asking about language, then what are they asking about? To answer “What is X?” is just to give an analysis or stipulation of what X is. But then it can be immediately responded they are asking about how things really are, independently of how we think and talk, about “the worldly property itself.” Call it real definition, along with Robinson. Chalmers rejects real definition, i.e. “the claims of some opponents of conceptual analysis. Such opponents often say: “I don’t care about the concept of X. I care about what X really is. Even though X1 doesn’t mirror our concept of X, X really is X1.” (538.) And the position is ascribed to Ruth Millikan with her theory of the “natural nature of meaning” (1983), as well as to Hilary Kornblith with his account not of the concept of knowledge, but of “knowledge itself” (2002). Real definitions are therefore not conceptual analyses, and constitute substantive theoretical disagreements, says the argument. §3. Chalmers rejects the view in only one sentence: “I think these proposals about what X “really is” are often implausible, as the concept of X places constraints on what it picks out (539). But why is the constraint a problem? The intuitive reason for this is that the topic of X and the topic of “X” are distinct; X is a property while “X” is a linguistic expression, or more vaguely, a psychological entity or concept. To open up this argument a bit more, what “X” means, it is natural to assume, determines an extension E of the things that “X” applies to (The hymn from analytic philosophy that says “meaning determines reference”.) But it is also a natural assumption that a theory of what the property X is determines which things are X, namely set of things E’ that have that property (“X-ness determines X-hood”). Now, if a real definition of X is not constrained by the conceptual analysis of “X,” then E and E’ should be able to be completely different, which obviously they cannot. Too much discontinuity with the conceptual analysis of “X” tends to disqualify a theory of what X is for being a theory about X. It will be accused of having changed the topic, and that is what is meant by being constrained. In Millikan and Kornblith’s defence, what if the only reason that they are using “meaning” and “knowledge” to describe the properties they are interested in is that these are just the words that happen to come closest to picking it out? It does not mean that they are constrained by the meanings of those terms; had the “natural nature” of meaning differed too significantly from the ordinary concept of meaning, and if the concept of knowledge had been different from “knowledge itself,” then they might as well have used a different term to describe that property. If they are prepared to let go of the linguistic expression, then it cannot be said their theories are constrained by it. §4. However, if it is unconstrained by analysis, Chalmers can reply, then they should be prepared to let go of it or apply subscripts (see 1.6), which it appears they can’t, because much of the theory rests on “knowledge” or “meaning” having one particular meaning. From this, the argument can be made more precise. The problem, we can say, is not that real definition is constrained by analysis, but that there are no obvious criteria of correctness for those theories apart from those given by the analysis. If these theories are not constrained by the meaning of “knowledge,” and “meaning,” then what else guides and constraints their search, what are real definitions answerable to? For example, what can be

said to defend a theory about what some X is, against another theory about what X is, apart from a reference to a conceptual analysis of “X”? §5. Kornblith actually considers the same type of objection coming from Alvin Goldman: “How would one know what to criticize, or what needs to be transcended, in the absence of such a description [of the way we think and talk about knowledge]?” (Goldman 1992). And he makes the same type of reply, with the premise that knowledge is a natural kind:

[I]f knowledge truly is a natural kind, then this sort of response is inadequate. We would hardly think that the chemist's first job is an elucidation of folk chemical notions (especially if this required extraordinary effort by the entire community of chemists over a period of millennia) so that we would know what chemical views need to be transcended. In the case of chemistry; we can simply skip straight to the project of understanding the real chemical kinds as they exist in nature. My suggestion here is that we should take seriously the possibility that a similar strategy might be equally fruitful in epistemology. (2002, 19)

In the book and elsewhere, Korblith argues extensively for why knowledge is a natural kind, by appealing to features like being inductively projectable and belonging to a “homeostatic property cluster” and, so on, and most importantly, he assumes an anti- metaphysical approach that takes the questions of epistemology to be heavily informed by the empirical sciences, and cognitive ethology especially. The question of real definition has turned on the question of natural kinds. Are disputes arising in virtue of disagreements about real definition verbal, or are they about natural kinds? Are they substantive because they are epistemically governed by constrains coming at least in part from reality itself? Roughly, a natural kind is a property or category that exists in the real world independently of how we think and talk about it, and independently of our interests. A simple analogy is the natural kind water, where the answer to “What is water?” is H2O. It can be argued that even if it had turned out that the concept of water picks out another extension, the real definition of water would still be H2O. In other words, the conceptual analysis might reveal that there are things that are not H2O to which we would still apply the term “water” (for example the clear, drinkable, imaginary liquid, XYZ) and things that are H2O but to which we would not apply this term (for example an orange, flammable, imaginary solid made of H2O) but that does nothing to constrain investigation into what water really is, or so goes the reply. According to this view, the fact that the extension of X and the extension of “X” can come apart is because X is a natural kind and so it is the real world, and not the meaning of “X” that determines what X is. Are philosophers disagreeing over real definitions given by natural kinds rather than of analyses given by meanings? Here is the natural kind reply on behalf of real definition stated, the natural kind being freedom mutatis mutandis, and not knowledge:

“Freedom is the ability to do otherwise” is not an analysis because there might be things that instantiate the natural kind freedom, but to which the ordinary term “free” does not apply, and likewise, individuals to which the term does apply, but who really aren’t free. Instead of subscripts, the difference between the word and the property can be signalled with the modifying adverb “really.” Disagreements over the real definition of freedom are constrained not the concept of freedom, i.e. the way we think and talk about it, but by the fact that there exists a clearly delineated phenomenon in the world, for example the ability

to do otherwise, that it is natural to track by a question such as “What is freedom in reality?” Suppose, for example, that there is agreement that humans lack the ability to do otherwise, but disagreement over whether they have free will; many such disagreements come down to socratic disagreements: “What is freedom?” According to the natural kind/real definition defence of such disagreements, if the ability to do otherwise is a natural kind, then humans aren’t really free. And because it’s only facts about the real world, not meanings, that can settle that dispute, it is a substantive theoretical disagreement.

This may be able to legitimise distinctly socratic disagreements where the disputants say things like “Is freedom-1 really freedom?”, “We are disagreeing about the property itself, not the meaning”, and “You are only talking about shmeedom.” 42 §6. I will now give some reasons to reject this view. First and foremost, the fact that water is a natural kind arguably depends on “water” being a natural kind term. But a disagreement over whether something is a natural kind term is a metalinguistic disagreement, meaning disputes arising out of such disagreements are verbal. Isn’t it ultimately due to meanings, or at least a semantic theory, that water being H2O is not just a mere stipulation? It might be that the reason H2O qualifies as water and just not a stipulation owes to something like the fact that “water” is used deferentially, or that externalism holds at least for this term—its status as a natural kind term. If analysis revealed that “water” is used in a non-deferential way, for example, or that whatever experts discover about molecular structures is not related to that concept at all, then in such a language it would be false to say “water is H2O.” But that is not to reveal anything about the substance water. The similar point is made by Chalmers in a related discussion (2012, 176):

Once we know enough underlying facts about substances in non-“water”-language, we know how those substances are, and further debate about whether or not a given one of those substances counts as water is in a certain sense verbal. The opposing view [that a theory of how terms refer could defeat a theory about how the world is] here elevates shallow verbal issues into sources of deep and essential ignorance about how things are in the world.

The question about language can be separated from the one about reality. Which terms are natural kind terms is obviously a verbal issue that substantive philosophical disagreements should not depend on. We are back again to conceptual analysis! 43 44

42 See Ludlow (2012, 42) for an application of the natural kind theory to vindicate the apparently verbal question of whether Pluto is a planet. 43 An analogous argument is exchanged between Chalmers (2009) and Ted Sider (2009) both in Metametaphysics: New Essays on the Foundations of Ontology. The question is whether what exists depends on the meaning of the existence quantifier “there exists.” In normal English Eli Hirsh’ commonsensical descriptive ontology (2005) is true, so that ordinary objects like tables and chairs exist, but in the variant of English that is spoken in ontology seminars, Ontologese, Ted Sider’s revisionary ontology is right, or so goes Sider’s story (before he later abandoned this approach (2005)). In this specialised language the existence quantifier ranges over fundamental objects only, and not ordinary

§7. Apart from the many troubles that exist about the mind-independence and language-independence of natural kinds themselves, here are two possible problems with extending this solution into to philosophy. First, let’s assume that the former objection was mistaken (see 2.3.), so that it is relevant with respect to first-order issues like knowledge, freedom, truth, and so on, which terms are natural kind terms. Why think that philosophical terms are natural kind terms at all, instead of, for example, terms that are used non-deferentially, or deferentially only up to the point at which two candidate real definitions come into conflict. Even supposing that the ability to do otherwise is a natural kind, it does not follow that freedom is the ability to do otherwise; it still in part depends on the meaning of “freedom,” namely that it a natural kind term. So even if one real definition gets closer to a philosophical natural kind, say the ability to do otherwise, than the other, then why think that philosophical theories are governed by a type of semantics where naturalness holds as a criterion of correctness? 45 §8. The third potential problem is that it is not clear what a natural kind could be in philosophy. Even if his naturalistic proposal is correct about knowledge, it is an open question whether his theory about the natural kind knowledge can apply mutatis mutandis to other philosophical concepts like intentionality, justice, and freedom. Another natural kind take on philosophy is Theodore Sider’s in his Writing the Book of the World (2011), where the idea of a natural property is extended beyond the paradigm cases of chemical and biological kinds to into the philosophical discipline of metaphysics, following the comprehensive system of metaphysician David Lewis (1941-2001). However, natural properties in Sider and Lewis’ metaphysics are confined to the properties and relations instantiated by the things that are postulated to be the most fundamental entities of the world, such as spacetime points. For that reason theory can’t lend much help to the current issues; even though it postulates natural kinds that are accessible by philosophical reflection, they are confined to a very small number of metaphysical concepts, and could certainly not shed much light on concepts in philosophy more generally, like freedom, perception, value, truth, meaning, intentionality, and so on, and in turn questions like “What is perception?”, “What is truth?”, and “What is meaning?” Are there alternatives to natural kinds, or something that suits philosophy better? A look at historical conceptions of real definition might give a clue on how they can be right or wrong short of Korblith’s commitment to naturalism and Sider’s confinement to fundamental entities in the physical world. Real definition has a natural place in metaphysical systems that postulate abstract objects and the ability of human reason to connect with these entities. On such views, which we can label Platonism, real definitions

objects like tables and chairs, meaning Hirsh’ descriptive ontology wrong given that philosophers speak Ontologese. Now, the point coming from Chalmers is that any dispute arising in virtue of a disagreement over we speak English or Ontologese is verbal, just like any dispute arising in virtue of a disagreement over which terms are natural kind terms. 44 A reply to this objection continues in section 2.3. 45 See Cohnitz & Haukioja (2013) for an approach where an externalist theory of reference can hold for some philosophical terms while an internalist theory of reference can hold for others.

are answerable neither to physical facts in the empirical world, nor to mere language use, but rather to something like “forms,” as was discussed by Robinson above, and we acquire philosophical knowledge through rational illumination that puts us in intellectual contact with these forms, and it’s from them that the real definitions derive their criteria of correctness. In this platonic spirit that “language ought to carve the world at its joints” can ideas like “natural kinds,” “essences,” and “reference magnetism” be extended into philosophical methodology and vindicate disputes we thought were due to mere disagreements in meaning? Against this, it is unclear what could be the theoretical backing for those who do not hold such metaphysical views; it seems to me that it would at least require an inflated theory of properties that takes facts about which properties there are to be independent of mind and language, but that is only the beginning. To metaphysically speculate for a second, there is also something like a location problem: The mind independent joints would have had to be not in the actual physical world of animal taxa and chemical microstructures, as we are used to in this literature (and not in the fundamental structure of the concrete world like in Sider’s theory), but it is not so clear to me where exactly these mind independent correctness conditions could derive from , if not for a platonic heaven or a “space of concepts,” as in McDowell (1996). And rather than tigers, water, gold, and spacetime points having essential properties, the same ideas need to be extended to the likes of freedom, perception, and intensionality, which might be very difficult. Also in Kornblith, the phenomenon knowledge exists in nature, and is connected with causes and effects, and so on (2002), criteria that would be unrealistic to assume for most philosophical terms. Be that as it may, a theory of “philosophical joints” that are not just “joints in language” or “joints in thought” is beyond the scope of what can be done here. And why not just “abstract objects” in the first place? In sum, it is implausible that philosophers are disputing real definitions, at least insofar as something like natural kinds are the only option. It is clear neither what could play the role of natural kinds for philosophical terms, nor whether philosophical terms are natural kind terms to begin with; and most importantly, it doesn’t matter whether there are natural kinds, for a dispute that arises in virtue of such disagreements would be verbal because the disagreement depends on how the term refers, settling the question by settling meanings. But if this can be made to work, then there could be substantive theoretical disagreements over “What is…?” questions. Though the focus has been on natural kinds, the idea of real definition in general seems inseparable from language and thought. Real definitions are trying to do two things at the same time; they are conceptual analyses disguised as propositions about reality.

2.3. Questions of language and questions of reality

Can we distinguish questions of language and questions of reality? Theoretical disagreements are often intertwined with metalinguistic disagreements, but the spirit behind the method of elimination theory is that those linguistic elements must to be separated out, set to the side. The two subjects, reality and language, can be given independent treatment, and when a question apparently about the former turns on a disagreement about the latter—verbal. Conceptual analyses were found to be irrelevant to first order issues about reality, and real definitions were found to be constrained by

linguistic facts; therefore, disputes that arise out of either are verbal, or so it was concluded. The two hypotheses about socratic questions that have been considered were described so that conceptual analysis in its many forms was supposed to be in tension with the idea of real definition. But if we set aside Millikan and Kornblith’s analysis unfriendly views on this, was Chalmers right in distinguishing the two questions to begin with? An important reply will be attended to in this section, which begins by rejecting the distinction between questions of reality and questions of language; they share the same topic. Because they do, the analysis of language can be philosophically substantive, and likewise, questions about reality being constrained by language does not mean that they are verbal! A hybrid conception of philosophical questions like “Are we free?” and “What is freedom?” may sound odd—shouldn’t questions of language be distinguished from questions of reality? On the other hand, what is so undesirable about a theory that is constrained by language, but which is also answerable to how the world is? What if they are one and the same, such that to find out about what “free” means just is to find out what the property freedom is, and in turn, who instantiates that property? Plenty of philosphers in this literature have said comparable things, along the lines of “let us forget once and for all the very idea of some knowledge of language or meaning that is not knowledge of the world itself” (David Wiggins 200) and that the question of the meaning of “F” is not distinct from the question of what is the nature of the phenomenon Fness (Hacker on Wittgenstein, 2001). Even philosophers the likes of Timothy Williamson, otherwise in opposition to linguistic approaches like these, have said that “perhaps one cannot reflect upon thought or talk about reality without reflecting on reality itself” (2004). 46 Taken out of their respective settings, these general claims can come to many things, but if we reject, like them, that language and reality make up distinct topics, three replies report themselves: The first is to argue that conceptual analyses involve facts about reality for the reason that words depend for their meaning on facts about reality; the second is that conceptual analyses constitute evidence for claims about reality; and the third—from my reading of Frank Jackson—is that conceptual analyses define which parts of reality a theory about reality is a theory of. They are all hybrid views, supporting the idea of “reality oriented conceptual analyses.” If either of them is right, neither semantic constraints nor irrelevancy to questions about reality would be a problem for “What is X?” questions. The three will be addressed in turn.

46 “… What there is determines what there is for us to mean” (111), as with natural kinds; metaphysical theories about which properties there are determines what words can mean. Williamson is a good representative of what can be called a “reality first” methaphilosophy. At least telling from his thesis that expressions like “tall,” “bald” and “grown-up” having vague meanings (or since Williamson is an externalist about meanings: our metalinguistic beliefs about them are vague) does not imply that the property of tallness and baldness are vague properties likewise. In other words, conceptual analysis has nothing to say about the real, language and mind-independent cut-off points of these properties (See Vagueness 1994).

2.3.1. A summary of the argument To recapitulate, disputes over conceptual analyses are philosophically insubstantive because they are merely about language, which is irrelevant, (2.1.); and same with real definitions; because there is evidence that they arise in virtue of disagreements about language too, namely, disagreements over which terms are natural kind terms, (2.2., paragraph §6) or something along the lines of natural kind terms (§7). Solve that linguistic issue, and then the dispute at large is solved. The present problem can be summed up in an inconsistent set of points under consideration: First is the thing that was established in §6.

Disagreements or disputes over real definitions arise in virtue of disagreements about conceptual analysis

Together with some definitions:

Real definitions are about reality Conceptual analyses are about language If a dispute arises out of disagreements about language it is verbal

A benefit with the argument is that none of the premises rely on particular conceptions on real definitions or conceptual analyses, but only disagreements that arise in virtue of them, which makes it very general. We get the strong conclusion that disputes owing to real definition or conceptual analysis disagreements are verbal, which may be many. Neither of the two things that “What is …” questions can mean is vindicated, and these types of disputes are merely verbal. However, if this reasonable premise is added:

If a dispute is about reality it is not verbal

the set turns out inconsistent (real definition disputes are verbal and not verbal). The premise seems reasonable; “about reality” understood broadly to encompass any of the things that are not the words used in the dispute, verbal disputes are precisely those that are not about reality. To underscore the width of the problem, we can just understand “reality” as anything that is not the words used in that dispute, so that any dispute about those words or mediated by one is not about reality in this relevant safe sense. Other words used in metaphilsophy for this include “extra-lingusitic reality,” “worldly fact,” and ”object level issues.” Does this mean that the assumption from §6 must go so that the argument fails? To defend the argument the following fix can be given, an assumption that has been looming in the background for some time:

No disagreement about language is a disagreement about reality

This means that the argument stands and falls on this assumption, which it might be denied by an opponent. It will be discussed in what follows.

2.3.2. Externalism and natural kinds—Are questions about language questions about reality? The first objection assumes where the discussion of natural kinds ended, this time around with the premise that disputes over reality can be about language at the same time. It can be a defence of any dispute that arises in virtue of disagreements over conceptual analysis or real definition, but let’s limit the scope to “What is…” questions in philosophy. It can be argued that since we all speak the same language and so mean the same things by all our terms, and since the facts about what we mean are grounded in facts about reality, finding out about what we mean is equally to find out about reality. It starts off from the idea that there are facts about what expressions mean that stretch across the whole language community and are grounded in facts about reality, which is to say, it is in virtue of reality in part that expressions mean what they do. Call this first premise externalism, and the second premise we can call the socratic premise: “What is X?” questions are asking for such semantic facts, and socratic disputes turn on disagreement about what these facts are. Now, this reply’s third premise is that since these facts depend in part upon how reality is—for example, the meaning of “water” depends in part upon the chemical structure of the real world substance it denotes—then questions about semantic facts, or meanings, are also questions about reality. The conclusion is that socratic questions and disputes are both about meanings and about reality. On this hybrid view of socratic questions, a dispute over what water is would be a dispute about the meaning of that term and just as much a dispute about the chemical composition of water. At least in the case of water, supposing “water” is a natural kind term, the answer to “What is water (in reality)?” and the answer to “What does ‘water’ mean?” are, according to this view, one and the same. There are some reasons that the two first premises can be rejected, for example, even if externalism is correct and there are facts about what words mean, they will not be so precise as to be able to distinguish between different answers to philosophical questions, including socratic questions (2.2., §7).47 Are there philosophical experts that partake in deciding the meanings of expressions like “freedom”?48 If there is reason to doubt such authority about expressions like “martini,” why believe philosophers? But if it is granted, that philosophical terms indeed are governed by such a semantics, and that there are something equivalent to natural kinds on which they depend for their meaning, then another reason is that, where biology has expert usage and experts deferring to chemical facts, what would a “philosophical fact” be, or what would be its equivalent? (2.2., §8) Even if the problems in §7 and 8 are off the hook, this response misses the mark. It is the third premise that I want to take issue with, so let’s assume the first two, i.e. that externalism is true (also for terms in philosophy), and that socratic questions are

47 I am thankful to Tomas Midttun Tobiassen for helpful discussion 48See for example Williamson (2014) discussion of philosophical expertise.

questions of meanings as meanings are construed according to externalism (and that socratic questions in philosophy are too). The third premise’s inference from dependence to a shared topic is questionable. Why think the fact that meanings depend on facts about reality entail that questions about meanings are also questions of reality, or that the two can constitute the same topic? My hunch is that there are still two different things that the socratic disagreement over “What is water?” can turn on—one is about language and the other is about chemistry. To see this, consider what the method of elimination does to “What is water?” when the key term is eliminated. Whatever the disagreement would consist in, it would probably turn on either of the two subdisputes “What does ‘water’ mean?” and “What is the clear, liquid waterly suff’s chemical composition?” And since these are about different things—a word and a chemical—they are different topics. In conclusion, the premise that being dependent on reality means that conceptual analyses are also about reality must be rejected. In more detail, the reason to doubt these two subdisputes are the same topic is approximately the same as the quote from Chalmers in §6. The fact that agreement on the former might depend on agreement on the latter does not go to show that they are one and the same topic. The dependence relation would thereby have had to be symmetrical, or go both ways, but a disagreement over the former topic, i.e. the word “water” does not have any implications as to disagreement over the latter topic, i.e. chemistry, which would have been absurd: There are no ways in which findings in chemistry would be accountable to semantic facts. The discovery that water is H2O is all on the part of the chemists and not on someone who studied the meaning of that word. We can’t discover that water isn’t H20 from studying language. The separate issue of whether the waterly stuff is H2O would only have the implication that water is H2O given a background of semantic facts about how speakers and communities defer. One can agree about the analysis while disagreeing on the first-order issue and one can agree on the first-order issues while disagreeing on the analysis, and so on. Supposing externalism, we have an asymmetrical dependence relation, but not an equivalence between the semantic question and the one about chemistry. It can be responded that things are different with philosophical terms: Maybe the dependence not only holds in the direction reality-to-language as may be granted about the term “water” and maybe some philosophical terms too. But maybe, in philosophy, in contrast, it also holds in the inverse, language-to-reality, direction, so that the dependence is mutual. Mutual dependence seems to be the only way for a question to be both about language and reality, and not only depending for it in part for reasons of externalism. If so, the answer to such as “What is freedom?” could have been about both reality and language, because “freedom” meaning what it does not only depends on external philosophical factors, but these external factors, whatever they would come to, in some sense also depend on meanings. However, I think this reply can be dismissed. From the dependence relation being mutual follows the undesirable consequence that features of reality in some way are dependent on features of language. I think this would be equally absurd for philosophy as it is with the case of chemistry just considered, so consider this reply out of the question. In summary, the topics of language and reality are always separable, for example by the method of elimination or by asking “Well, that depends on what you mean by that expression!” Therefore, they are different topics, and while they might reflect some of

them, disagreements about meanings cannot equal theoretical disagreements about how things are. That would require a mysterious mutual dependence relation between the two subject matters. (In connection to following hybrid view about reality oriented conceptual analysis “transcendental arguments,” I discuss in more detail the absurdity of this type of consequences.)49

2.3.3. Linguistic evidence, contextualism, and transcendental arguments—Are questions about reality questions about language? The thought and talk that analyses are concerned with appears to be very different from the topic of how things are in reality; the four ways that Chalmers mentioned in which the former can be relevant for philosophy were marginal, and certainly none of them had got anything to do with how things are in the world, that part of it which is independent of human linguistic practice. They could at best just inform the philosophical disputes by supplying information about which topics it is that ordinary language users find interesting, or inspire distinctions and nuances that might become relevant at some point later in the dispute, and it stops there. (2.1.) We can never say “’Freedom’ means such-and- such, therefore we aren’t free.” Or can we? The second hybrid response to be considered rejects this relegating verdict on conceptual analysis and says that premises obtained from conceptual analyses, presumably about how words are used, but maybe also about the concepts more generally with which we think, can take us to conclusions about how things are in reality. If that is true, then disputes coming down to disagreement over analyses aren’t verbal after all. Like with the former response, it is difficult to find representatives of such views (it should be researched more), but my theory is that many similar assumptions are at work behind many verbal disputes, and that it is easy to think in this way. The questions “What is knowledge?” and “What is perception?” have been at the centre of many disputes in philosophy where “What does ‘knowledge’ mean?” and “What does it mean to perceive?” have been in the centre to equal extents. In some parts of this literature linguistic data is often appealed to for evidence about who has knowledge and who has not. One example can be found in the various sorts of contextualism that circulate in this literature, about properties ranging from perception and knowledge to morally right action and aesthetic value. Take the epistemic variant, here introduced by epistemologist John Greco:

We may distinguish two questions one might try to answer giving an account of knowledge. The first is the “What is knowledge?” question. This question asks what conditions a person must satisfy to count as knowing. The second is the “What are we doing?” question. This question asks what illocutionary act is being performed when we say that someone knows. (2003, 116).

49 See Cappelen (forthcoming 2018) for more discussion about how externalism relates to change of topic.

His second, attribution question is clearly about language, but the first one is by many philosophers taken to concern reality; a distinction which is mirrored in the divide between the two doctrines: attribution contextualism and property contextualism. The more moderate attribution contextualism concerns the context-sensitivity of the correct application of a term, whilst the latter is about the property itself. Property contextualism is thus characterised by its commitment to facts about reality (about who knows what) being dependent on facts about language (what “knowledge” means). To illustrate: First, linguistic analysis reveals that the meaning of “to know,” is context dependent, as well as the dynamics of this dependence with respect to different contexts. Second—the language- to-reality move, this means that whether or not you have the property of knowing, let’s say that you have hands, depends on which context you are in. In an everyday context where the threshold is low you do have hands. But in the context of a philosophy seminar about skepticism you don’t, because considerations about defeating scenarios like one in which a Cartesian daemon which fools you about everything raises the threshold to the point that you cannot know that you have hands (See, for example Ludlow 2005 and replies by Stanley 2005). Skepticism is a hypothesis about the real world, and not about language, but the current arguments against Skepticism, which comes from many contextualists, tend to jump straight to that conclusion from considerations about language and knowledge attribution only. Ludlow (2012, 131), for example, opens by saying “Contextualism offers the following answer to the skeptic: The person who denies we know we have hands is sometimes right, but only in contained artificial circumstances, and our knowledge claims are otherwise preserved.” before he turns to defend contextualism exclusively by appeal to arguments from linguistics. As a matter of interpretation, maybe the question about knowledge is a question about language in the eyes of the contextualist. If so, they are in harmony with the problem of verbal disputes as it is raised in this dissertation and in Chalmers’ paper. They can be considered to follow the instructions of the method of elimination and eliminate the word “knowledge” in favour of “knowledge with a high threshold” and “knowledge with a low threshold,” and solving the dispute. Those approaches to philosophy that do this are pluralists about the meanings that can be given to philosophical terms, and so would not support the legitimacy of “What is X?” questions all the same. Regarding contextualists and others making the same type of argument, Chalmers’ theory comes with a very general warning. Great caution is advised when moving from conceptual analysis to real definition, because

if one is not careful, one will end up making points that reflect the vicissitudes of one’s language rather than deeper philosophical truths. … To see the point, note that the mere fact that existing words like ‘know’ or ‘intentional’ or ‘see’ behave in a certain way does not suffice to settle substantive disputes about epistemology, action, or perception. (540)

The two are distinct topics, and the topic of “know,” for example, should be treated as different from the topic of whether we have knowledge (in reality). I will take a moment to rather haphazardly run through the many possible consequences of mixing the two, just in order to underline why “making points that reflect the vicissitudes of one’s language”

would be a problem. For one, if what philosophical theories we accept owes to the way our language figures, why should we trust those language based judgements to be right, short of reasons to believe in some sort of preestablished harmony between language and reality? And for a second, under the reasonable assumption that language can change, either from context to context, and if not at our will, at least over time, then what happens to the supposed independence and absoluteness of reality? And no matter what’s your theory of linguistic evidence, it is beyond doubt that our actual language figures like it does only contingently so. For a third, if it’s language that’s responsible for our philosophical beliefs and theories, how can they be about something other than what they are based on? And lastly, if the key to their justification is found in our linguistic competence, or if our evidence essentially is linguistic, how can we on that basis so to speak transcend language to reach conclusions pertaining to extra-linguistic reality? A leap like that would at the very least require support from some interesting assumptions, and in particular, as compellingly argued by Barry Stroud in connection with similar issues, these assumptions would amount to either some sort of idealism (that facts about reality depend on facts about the mind), or verificationsim (that only in accordance with empirical criteria of application can terms be applied meaningfully) (both of which are characteristically anti- realist). For more details on this argument, s “Transcendental Arguments” (1968) and “Kantian Arguments, Conceptual Capacities, and Invunerability” (2000)). The epistemic contextualists should be pluralists in order to avoid these consequences. It would be an odd consequence if changing contexts changes reality, or if something in the world changes when language does. It can be argued that this is just a difference in standards, and not a difference in meanings (Stanley 2005), so that the contextualistst does not have do agree that the dispute over knowledge is verbal. However, who has knowledge also varies along the dimension of ascriber and ascription: When I’m in the seminar on skepticism, does nobody know that they have hands either (i.e. those who are not in the room), or is it only me and my classmates who don’t know that? Do I suddenly know again that I have hands once I enter another context where knowledge has a lower threshold? Unless we are giving into crazy metaphysics, these are variations in language, not in the world. These are paradigm specimens of prima facie verbal questions. Given that the case for contextualism in general is well supported, these reasons should be limited to talking about words and ascriptions; not reality. Given that these are by and large linguistic issues, Chalmers points out that also with respect to those two think that we always or never have knowledge, skepticists and optimists, the contextualist should agree (541-41):

After all, views based on these data may differ only verbally from views on which ‘know’ or ‘intentional’ or ‘see’ pick out something else and that endorse apparently different first-order claims that use these words. Consider a contextualist view that says ‘know’ has two distinct referents in different contexts: knowledgelow, which we often have, and knowledgehigh, which we never have. A proponent of this view may differ only verbally from a proponent of an apparently skeptical noncontextualist view on which ‘know’ always means knowhigh, or an apparently nonskeptical view on which ‘know’ always means knowlow. Any substantive differences between the proponents will turn on what they say about further properties of the referents: for example, about the epistemic value and normative role of knowledgelow and knowledgehigh. Assessing these further properties requires going beyond the linguistic data.

Even if this is too crude to prove that conceptual analyses cannot take us to truths about reality, the contextualist as well as those who make similar language-to-reality moves at least have a number of things to explain if they are to represent legitimate instances of this idea. So the conclusion that the constraint from the former upon the latter is a problem is still standing until a theory whereby they are one and the same is established, which the argument in 2.3.3. provides reasons to be pessimistic about. Unless that can be done, we should distinguish sharply between questions of language and questions of reality.

2.3.4. Conclusion We have seen that questions of language are not questions of reality (2.3.2) and that questions of reality are not questions of language (2.3.3.). I therefore conclude that the premise in (2.3.1.) can be held on to, and the problem about “What is…” questions remains. This suggests that the problem from (2.1.) still stands, that disagreements over conceptual analysis are philosophically irrelevant, and also the problem from (2.2.) that real definitions are constrained by language, and are therefore just conceptual analyses to begin with. Given that many philosophical disputes arise in virtue of such disagreements, the problem of verbal disputes remains a significant problem.

2.4. Was Humpty Dumpty right?

'When I use a word,' Humpty Dumpty said, in rather a scornful tone, 'it means just what I choose it to mean — neither more nor less.' 'The question is,' said Alice, 'whether you can make words mean so many different things.' 'The question is,' said Humpty Dumpty, 'which is to be master — that's all.'50

This section aims to put the above discussion into more context, and question why analysis should matter in general, and why we should not accept Chalmers’ “conceptual pluralism” instead, roughly that we can talk about any part of reality that we like, words not having to mean any particular thing. Can there be right and wrong analyses, right and wrong meanings, right and wrong topics? Humpty Dumpty from Lewis Carroll’s novel seems not to think so. We can mean whatever we like by our words. Given only that our conversational partners follow, disputing meanings is pointless all the same. Humpty Dumpty agrees with the idea in “Verbal Disputes” that we are not in the service of language, and can make it mean whatever we want. In particular, we can stipulate as much as we like—in fact, the subscript gambit itself is a device for stipulation—feed it a

50 Lewis Carroll, Through the Looking Glass

socratic disagreement and it spits out two stipulated terms instead, two new topics to disagree about. However, coming from the other side on these issues are philosophers like Frank Jackson and P.F. Strawson who worry that if philosophers use expressions with their own, idiolectic, or specialized meanings, then they are just talking about different things. Don’t we want to talk about the real phenomena knowledge, intentionality, freedom and truth, rather than schmowledge, and schmeedom, etcetera? Considerations like these make us think that meanings matter, although important to the idea of a merely verbal dispute is that meanings do not matter. Over and above for addressing the same topic, do meanings matter? In this subsection I address some more moderate reasons why analysis could be the subject of substantive theoretical disagreement, namely that analysis is constrained by reality, and disagreements over analysis too, because there are right and wrong analyses which constrains which theories are true and false. It is coming from those that wish to reserve themselves against linguistic mistakes and misuse of stipulation. The earlier verdict on analysis was that it is irrelevant and that disputes are verbal insofar as they come down to disagreements there over, but are analyses relevant because they ultimately decide who is right and wrong? This picture has it fits in well with the actual practice in philosophical disputes, where it’s rare to choose to look for either a mere conceptual analysis or a mere real definition. Typically, both of the objectives are in the air: The conceptual part seems to play a role when a theory about what some X is has to respect the prior usage of “X,” if not to every detail, then resemble it at the very least. And the reality part seems to play a role when philosophers go beyond that to make it a theory of the real property X rather than just the expression “X.” There is something paradoxical about this; if it accords with “X’”s meaning, it is just an analysis, and if it breaks with it too much, it is no longer a theory of X. But given that this is a sufficiently accurate description of actual practice, a theory about what philosophers are disagreeing over should assume its legitimacy and try to make sense of it. If this practice is legitimate, then disputes about conceptual analyses can be disputes over substantive issues, namely of which part of reality a theory that employs that analysis is a theory of. Not about reality itself, like in the externalism objection above, this more moderate proposal at least secures that they are about something relevant to questions about reality, meaning they centre on substantive theoretical disagreements. To find out, “Aha, my theory of freedom is a theory about the ability to do otherwise,” might be based solely on linguistic evidence, but it is still a substantive finding about the world, because given that the theory is true, substantive consequences follows about what it is true of. If so, disputes aren’t verbal even if they arise in virtue of disagreements about conceptual analyses, as these analyses constrain what the disagreement can be about. Is this conception of conceptual analysis compatible with Chalmers’ critique?

2.4.1. Frank Jackson and the Canberra Plan In his Form Metaphysics to Ethics—A Defence of Conceptual Analysis (2012), Frank Jackson advances a view on conceptual analysis like this whereby it can partake in the discovery of truths about the real world. He defends hybrid view whereby the two are distinct topics, yet interact together in mutual dependence:

Although metaphysics is about what the world is like, the questions we ask when we do metaphysics are framed in language, and thus we need to attend to what the users of the language mean by the words the employ to ask their questions. When bounty hunters go searching, they are searching for a person and not a handbill. But they will not get very far if they fail to attend to the representational properties of the handbill on the wanted person. These properties give them their target, or, if you like, define the subject of their search. (30)

To find out about what it is one is talking about is an obligatory activity in philosophical theorizing, on Jackson’s view, and is answered by the attendance to conceptual analysis, his preferred Canberra version of which is roughly something like surveys of “the folk’s” intuitions, or their dispositions to apply the given word to possible cases. If it reveals “freedom” means “the ability to do otherwise,” then the ability to do otherwise is the topic of the free will question. Once the topic is thus defined, the philosopher can proceed to step two which is the reality part. Handbill in hand, they will look at whether that exists, or how it relates to other types of properties by way of relations such as supervenience, reduction, entailment, and elimination, without us having to go into the details. As such, conceptual definition is assigned only a modest, but crucial role. To help ourselves to an example, take the non-philosophical term “water.” If we were to discover that what we thought to be water was actually another chemical substance, XYZ, then conceptual analysis of “water” would decide whether that means a discovery that water isn’t H2O or whether it means elimitivism about water, which is to say, there is none of it in our rivers and oceans. He says further that if you left out the conceptual analysis part, or failed to observe the constraints that it places upon the topic, then the reality oriented theories you propose would not “have much of an audience.” To find this thing out about water, for sticking with the example, would not have had much of an audience given the result that we didn’t use the word in such a way that water were HO2 in the first place. And Edmund Gettier’s counterexample to Chilsholm and company’s justified true belief-analysis of knowledge would have been short on audience likewise had we all used that word to mean simply justified true belief in the first place, and without an audience entirely had Chisholm and company not intended that analysis to align with the ordinary usage. So solutions to philosophical puzzles cannot be obtained on the cheap simply by stipulating what we mean—never mind what others mean—by our philosophical terms. It would have been to turn “interesting philosophical debates into easy exercises in deductions from stipulative definitions together with accepted facts” (31). (Jackson’s position is inspired by Lewis (1970) and Ramsey (1990).) In more detail, conceptual analysis, as construed by Canberra plan to collect what they sometimes call intuitions or platitudes about the property X by analysing how people are disposed towards applying the term “X” including imagined counterfactual circumstances. The platitudes are thereby unified and put into order, so that if there are conflicts (often revealed by thought experiments), some of these platitudes must go—for example, the platitude that knowledge requires justification must go if we still think the term “knowledge” can be applied in an imagined case where the knower lacks justification. For example, one such imagined scenario is the chicken sexer, who can sort chickens into male and female by a mere glance, without knowing on which grounds their classification is based on and which features suggest that they are either male or female. The upshot is a list of platitudes that must be

satisfied for something in the world to have the property X. A theory involving “X” can thereby be assessed with respect to whether the property it denotes by “X” is X. If it does not, then it has just changed the topic, and is no longer about X. This theory has several benefits, primarily owing to its will to compromise. Analysis is philosophically substantive not because it tells us anything about reality itself, but because it tells us which part of reality we are talking about. Language and reality are still separate topics. Analysis informs us about which property or phenomenon in the world it is that is being talked about, without appeal to transcendental argumentation or by postulating natural kinds. It also yields what appears to be a reading of “What is X” questions that makes them relevant, precisely because they are about reality, and about X, because of the constraint owing to the meaning of the expression “X.” In a sense, they are not saying what X is, but which things are. The “What is X?” part is covered by conceptual analysis, while the “in reality” part is covered by the theories of supervenience, existence, entailment, and all that.

2.4.2. Against Jackson: Are philosophers disagreeing about what it is they are disagreeing about?

As philosophers we play language games. … And when we play language games, we do so rather in order to find out what game it is that we are playing.51

Over and above informing us about which topics would have had an audience, as if philosophy does that to begin with, the above moderate defence of conceptual analysis does not show that language can be a relevant topic. In spite of its appeal, the arguments from Chalmers still stand. We can see this if the subscript gambit is applied again, let’s say to “Are we free?” According to Jackson, we can only settle this question by conceptual analysis, by finding out what it is we are talking about. Are we talking about freedom-1 or freedom-2? Let’s say there is agreement that we have freedom-1, but not freedom-1; if analysis reveals that we are talking about freedom-1, then we are free. According to Chalmers, in contrast, there is no one thing we are always talking about when we talk about freedom; if there are several analyses, then, quite simply, there are several questions “Are free free-1?”, “Are we free-2?, and so on, and we were talking about different things. Debating analyses are still to verbally dispute. Analysis about what we mean by our terms cannot inform us about the thing we are talking about; it can at best inform us about which thing it is we are talking about. Either way, it would only be information about ourselves, as for example, what we would be interested in talking about:

51 Jonathan Miller and John Cleese, “Oxbridge Philosophy” from “An evening without Sir Bernard Miles,” at Her Majesty’s Theatre, 1977 from Youtube.

if it turns out that freedom-1 and not freedom-1 is picked out by our ordinary concept of freedom, so that compatibilism is true of the ordinary concept, then that gives some extra interest to freedom-1 and perhaps is evidence that this is the sort of freedom that we really value. (541).

It can supply some interest to the question of freedom-1, in other words, but no answers. Conceptual analysis is limited to this marginal role. If we apply the subscript gambit, we might see that our current theory of knowledge or freedom is about something different than the nonlingusitic property we find valuable, but that does not mean we don’t have knowledge or freedom, let alone that that the theory is false. If analysis would be able to tell us about something other than ourselves, then we will have to suppose that facts about ourselves provide evidence not just about the question or the topic under discussion, but the nonlingusitic property itself. Chalmers actually opens for such a use of conceptual analysis, but assigns it a marginal role. Return to the objections from 2.1, and objection number two in particular:

Second, linguistic claims may play an evidential role with respect to nonlinguistic claims. One such role is the trivial disquotational role: for example, the fact that ‘knowledge’ (in my mouth) refers to knowledge1 allows me to conclude that knowledge is knowledge1. For reasons discussed earlier, however, these disquotational inferences do not settle much: my view may still differ just verbally from that of someone else who says, ‘Knowledge is knowledge-2’. Some roles are less trivial, though. I know that I value knowledge, so if it turns out that ‘knowledge’ refers to knowledge1, it follows that I value knowledge-1. Assuming that our values provide evidence for what is valuable, then this provides evidence for the nonlingusitic claim that knowledge-1 is valuable. (541-2.)

But again, this is just an inference from facts about a linguistic agent to other facts about that linguistic agent; nothing new is inferred about knowledge other than that it can be referred to with another term too. To see this, consider what happens if the analysis had turned out incorrect, “knowledge” not referring to knowledge-1; Chalmers would still have valued knowledge-1. (Let’s say “knowledge-1” refers to justified true belief, or whatever property is that we do not have in Gettier cases, but in any other circumstance in which we have the property of knowing.) So the difference between Chalmers and Jackson’s approaches, while they agree on many points, is that in Chalmers’ different analyses amounts to different topics, while in Jackson our values and our language use places constraints on which topic is that we are actually discussing, and the linguistic data can tell us more about that topic than whether it is interesting or valuable; it can inform us about the conditions under which a theory about freedom-1, freedom-2, and so on is a theory of freedom. But unless, this comes down to something like real definition, this is still just information about ourselves, our values and our language, which means disputes that arise in virtue of conflicting conceptual analyses are still verbal. “What is…?” questions and other disputes that turn on disagreements over conceptual analysis are still in need of justification if the extent of the problem of verbal disputes in philosophy is to be limited.

2.4.3. Theoretical background: Chalmers’ Conceptual Pluralism One topic or many topics; one right analysis or many fine grained ones? This all comes down to the indeterminate meaning of the same-topic relation that was discussed in 1.8. We saw how it is often up to disputants what constitutes a continuation of the original question, and that this itself might be among the central topics of disagreement underlying philosophical disputes. The two conceptions of analysis that we have looked at are very similar, and might by the end of the day be found do differ only verbally, or with respect to what is emphasised: Jackson emphasises the importance of staying on topic and the pointlessness of just stipulating answers on the cheap. Chalmers emphasises that these answers can also be interesting on their own, and that nothing hangs on whether the original label is put on them or not. But given that there is actual disagreement between the two, it can be better displayed by raising these differences in emphasis into differences in ideology, rather than fiddling with vague concepts of topic-sameness and lingusitic constraints. If Jackson’s view represents opposition at all, it is in the form of opposition to Chalmers’ conceptual pluralism.

There are multiple interesting concepts (corresponding to multiple interesting roles) in the vicinity of philosophical terms such as ‘semantic’, ‘justified’, ‘free’, and not much of substance depends on which one goes with the term. The model also leads to a sort of pluralism about the properties that these concepts pick out. For example, it naturally leads to semantic pluralism: there are many interesting sorts of quasi-semantic properties of expressions, playing different roles. It leads to epistemic pluralism: there are many different epistemic relations, playing different roles. It leads to gene pluralism: there are many different things that deserve to be called “genes,” playing different roles. The same goes for confirmation pluralism, color pluralism, and so on. (539.)

Chalmers says “Yes!” to pluralism. It is the spirit you need to have to avoid verbal disputes. Above all, the value of the method of elimination, and the subscript gambit in particular, as I see it, consists not in its applicability but in its theoretical interest. It entails that inasmuch as a dispute is substantive, the disputants should always be prepared to give up a specific term. Historically, this pluralism about what terms can mean in philosophy is related to Rudolf Carnap’s pluralism about which linguistic/conceptual framework to use in science, and his Principle of Tolerance, which says that we are free to cast our conclusions in any language we please, instead of being prohibited by what words actually mean. Here the framework in question being logical frameworks:

In logic, there are no morals. Everyone is at liberty to build his … own form of language, as he wishes. All that is required of him is that, if he wishes to discuss it, he must state his methods clearly, and give syntactical rules instead of philosophical arguments. (The Logical Syntax of Language, 1959, 10)

To give syntactical rules is to stipulate, and the syntactical rules of a theory are not something that can be subject to argument. It is the content of the theory, as opposed to the framework (in our case the language, and not a logical system); only the former can be assessed for truth or falsity, and constitute disagreement. Chalmers pluralism about which concepts and properties can be associated with the same expressions is similar; nothing of

substance hangs on which property goes with which term (539), and they both have the same consequence that disagreeing about the language in which a theory is stated is not a theoretical disagreement. Cast the philosophical proposition or disagreement in any language suitable, it doesn’t matter which, or perhaps in slogan form: “Let go of language.” I think both of these principles are correct; nothing of substance, only practicalities, hangs on the meaning of terms. If Chalmers is right, the role of conceptual analysis is limited considerably by conceptual (or property) pluralism. Analysis would still only inform us about people’s intuitions and dispositions to apply a term; it would be a theory of “X” and not X. Even if one analysis of “X” captures the platitudes about X better than the rest, it does not mean that the analysis can claim to be a theory of X in competition with other theories of X. For example, if “freedom” means the ability to do otherwise, then it does not follow that theories about freedom wherein it has a different meaning is not about freedom. It would just be about one of the many topics corresponding to that word—moral responsibility for example. Each theory capture different aspects of the same property, or two related properties; and we should not expect there to be only one property that plays them all (or most of them). We might have freedom 1 and 2, but not freedom 3 and 4, for example. Which one happens to be a theory of freedom would be contingent on how people uses words, and so irrelevant. Which topic (freedom 1-4) that people tend to discuss when they use the word should not place constraints upon which topics philosophers can seek answers about:

For example, instead of asking “What is semantic content?” and expecting a determinate answer, one can instead focus on various explanatory roles one wants semantic content to play. One can then say: here are some interesting properties (of sentences or utterances): S1 can play this role, S2 can play these roles, S3 can play these roles. Not much hangs on the residual verbal question of which is really semantic content. Likewise, instead of asking “What is a belief? What is it to believe?” and expecting a determinate answer, one can instead focus on the various roles one wants belief to play and say: here are some interesting states: B1 can play these roles, B2 can play these roles, B3 can play these roles. Not much hangs on the residual verbal question of which is really belief. (538).

The reason the ideas of Chalmers entails pluralism is the same reason why verbal disputes are pointless, namely that, as might be said, “Words do not matter.” Or as it is put in a later work,

language is merely our tool, and we can use it do divide the world up as we please, without thereby creating deep and essential ignorance of how things are in the world. (2012).

2.4.4. Why bother about what something means? Now, against this very general idea, something like the following objection can be raised:

Jackson’s worry: If pluralism is right, then philosophy can’t be about reality; analysis is what makes sure a question is about reality and not just about made up properties.

What Jackson seeks to avoid a construal of philosophy where terms are just stipulated out of thin air so that the theories involving these terms are true in their own stipulated language and thence true merely because their words mean whatever they want them to mean. So where is the line between pluralism and an anything goes view? In this imagined dialogue, what Chalmers responds next is that language poses no constraints upon philosophy. In principle anything goes, as long as the others go along. In later work where the position is criticized that facts about language could defeat a theory about reality, Chalmers brings up what he calls Humpty Dumpty’s principle:

At this point, I think we should recall Humpty Dumpty’s dictum: Words are our servants, not our masters. (2012, 176)

Humpty Dumpty’s principle seems reasonable. As long as no confusion is caused, one can use words however one likes to, and the linguistic framework in a philosophical theory should not be an object of substantive disagreement. Avoiding verbal disputes seems to be the only constraint on language, at least it so with typical philosophical questions like “Do we have free will,” “Is the mind physical?”, “How can beliefs represent reality?”, “Does perception represent the perceived world” and “Can we know anything?” Myself I would even grant that one word can have two meanings in the same sentence as long as the difference is clear. For example, one can say “The mind is the brain, but the mind is not the brain” to signal that these are different usages of the word “mind” and so yield different but compatible answers to “Is the mind the brain?” signalling that it amounts to two different questions (let’s say if in the first “mind” is understood in observational third- person terms, and in the second it is understood as pure phenomenal consciousness). It does not have to be both at the same time as long as one is precise, for example by adding subscripts or modify the noun by such as phenomenal, or just make a new word for it altogether, like “qualia,” as it is called by many philosophers. However, as for who is right, the monist or the pluralist, it is difficult to tell on such a general level, especially with pluralism being defined more as an attitude to bring to disputes, rather than a theory. But then, does the pluralism attitude come with an anything goes attitude? The discussion will be ended by a look into these two different attitudes and their consequences. First, those types of disputants Jackson are worried about are often likened to Lewis Carroll character Humpty Dumpty who is known exploiting and playing with language, changing and stipulating meanings as he likes, confusing his conversational partner Alice in chapter six of Through the Looking Glass. However, in the context of Chalmers’ theory of verbal disputes, Humpty Dumpty is also a good fellow who represents the idea that nothing substantive hangs on which terms goes with which meaningswhile the argument from Jackson seems to presuppose precisely that it does; if terms are used differently they are not about the same thing; Jackson emphasises that change of meaning entails change of topic, an idea often ascribed to Strawson, (See Haslanger 2010; Cappelen, forthcoming):

..to offer formal explanations of key terms of scientific theories to one who seeks philosophical illumination of essential concepts of non-scientific discourse, is to do something utterly irrelevant—is a sheer misunderstanding, like offering a text-book on physiology to someone who says (with a sigh) that he wished he understood the workings

of the human heart. ... typical philosophical problems about the concepts used in non- scientific discourse cannot be solved by laying down the rules of exact and fruitful concepts in science. To do this last is not to solve the typical philosophical problem, but to change the subject. (Strawson 1963: 505)

It is a challenge to Rudolf Carnap’s idea of explicative or revisionary analysis—coming up with analyses that change the meaning of the analysandum for the benefit of its scientific application, a key example being the explication of the comparative concepts of “warmer than” and of “colder than” into the quantitative concept of temperature. Strawson’s objection to this proposed methodology is that it changes the topic; although the explication might mean something similar to the explicandum, it is no less a stipulation. While the exchange that ensued between Carnap and Strawson is highly relevant, let’s just grant that (synchronically) differences in meaning entails difference in topics, and that (diachronically) changes in meaning entails changes in topic; one topic for each different meaning that can be given to the term.52 Here Chalmers’ pluralism enters; we can agree that there are different topics for each meaning, but not one of those topics is the right one; just like words are not our masters according to Humpty Dumpty, topics are not our masters either. For example, the argument in 1.3., “Can there be broadly verbal disputes?” attempted to show that there are actually not many metalinguistic disagreements in ordinary speech, because “belief about meaning” can be so many different things, and the two have to be the same type of belief to be able to disagree. If the argument is right, there are not many broadly verbal disputes, hence, that definition does not align with the ordinary, broad notion of a verbal dispute. But so what? The interesting property might be broadly verbal disputes, or as he suggests, “Shmerbal dispute” (520n). Chalmers is a pluralist about “verbal dispute” and many other expressions too, for example “mind.” And by distinguishing the different senses of that word, following Strawson, we distinguish different philosophical questions. For example, the above disambiguation of “mind” led to a disambiguation of “How does the mind relate to the brain?” Two different topics, two different questions, which in his earlier work, Chalmers calls The Easy Problem and The Hard Problem respectively. In all, pluralism seems compatible with Strawson’s principle. Almost all of us are pluralists about the terms we don’t care about, but verbal disputes are occasioned when a philosopher prefers to use a term with one specific meaning, being monist for a second, or to defend such a usage, denouncing all the others. It is still just an attitude, but it is an attitude that any disputants need to assume if they are to operate the method of elimination, and not refuse to eliminate the key term, saying

52 It appears that whether Strawson’s principle is true is just a matter of how topics and meanings are individuated with respect to one another. In line with the general case made in this thesis, the acceptance of this principle is just to provide a stipulation about how things are individuated without the further question of whether they really are individuated differently. However, against this, see Cappellen (forthcoming).

things like “Your theory isn’t about knowledge!” Chalmers’ theory splits this into two different disagreements: Whether it is to know-1 or to know-2 that really is to know is a mere verbal disagreement, and there is no such thing as what knowledge really is. There are many properties, and none of them has a special claim to that word. When the subscript gambit reveals a handful of new properties, the monist says in the tone of the child who turns vegetables down: “I don’t want these related properties, I want knowledge!” The method of elimination is revisionary because among the instincts that guide philosophical theorizing, a major one is this socratic urge to unify the many different things associated with an expression under the same theory. The spirit that comes with the method of elimination is to give up on the idea of unary concepts and of a theory’s unifying abilities as a measure for its worth. Insofar as there is an interesting relation in between several possible subdisputes and among the various things an expression can be taken to mean, then the imperative goes: Address it as its own topic! The response we have been looking at seems to require the opposite attitude, call it monism about the properties and concepts that words denote, and by extension, the topics that they determine. It assumes that there is one correct analysis, and so one topic for each term. Chalmers’ pluralism about linguistic frameworks in philosophy is more radical than what it might look like, because, as I have tried to show in this Chapter, there is not only Jackson, but a number of first-order philosophical theories as well as metaphilosophical views, that rely on the opposite assumption, monism. Monists are those that take it to be definitive singular answers to question of the form “What is…?” The ravine between conceptual analysis and questions about the real world contains some conceptions of philosophical method that might be in tension with the idea of verbal disputes in philosophy, and a clear divide can now be discerned between those that are and those that are not—the divide between monism and pluralism. A conceptual analysist can be monistic about their analysis and claim that all other analyses are false; a natural kind real definer can be monistic about their real definition and claim that theirs is the only natural kind associated with that expression; and Frank Jackson can be a monist about topics, and argue that his topic is the topic they are disagreeing about, and so on. If disputes over these topics aren’t merely verbal, and they actually constitute substantive disagreements about which thing in the world the key term picks out, then are philosophers who ask such questions just disagreeing about what it is they are disagreeing about?

2.5 Conclusion

In a parenthesis immediately following the Martini story, Bennett adds: “The purist is of course right about the use of the word. Remember that there can be verbal disputes in which one side is straightforwardly mistaken!” (ibid, 13) The notion that one can be right and wrong about language, in any relevant sense, must be rejected. Two lexicographers may be able to disagree about language, but if philosophers do the same, they are wasting their time, just like the two bar goers in our thought scenario. There is something inherently wrong about disputing language, a sad error which rises from the monistic assumption about language that it can be wrong or right. Over and above the purpose of avoiding verbal disputes, perhaps also for communicative purposes in general, there is no philosophically relevant sense in which one can be wrong or right about language. Right

about the community meaning? Why care? With analysis, why is it important how we actually think and talk about things? Right about a philosophical fact, a socratic truth? There are shallow reasons to believe in these things? A real definition grounded in natural kinds? Language all over again. None of these ideas seem to offer any help. In the line of argumentation that has been entertained, some likelihood has been raised that verbal disputes abound, on various levels, because of assumptions that found in the wanting of validation about language being important. If anyone is in the wrong, it is the purist, according to this current Humpty Dumpty friendly conclusion. Meanings are not the types of things that can be substantively disagreed. A little summary is that since philosophical disputes are difficult to diagnose, the discussion of verbal disputes must limit itself to a domain, or a set of them, of disputes with certain features, and discuss them on a general level. First, conceptual analyses of many types were assessed for relevance and confined to what was called the “analytic stage,” and beyond that, disagreements of that origin would not be substantive. Second, real definition was rejected as an implausible idea, primarily because it is inseparable from language, even with further semantic assumptions in place coming from Externalism, and deference theories of meaning. Then, the idea was rejected that language and reality somehow come together, so that analysis can be substantive, before the idea was rejected that analysis can tell us about which part of reality it is we are talking about so that philosophers are not stipulating truths, in favour if the idea that stipulation stipulates topics instead, and that nothing changes by a change of words. By this chapter’s end the idea should have been enrichened about which disputes are verbal. Chunk=those disputes that are grounded monistic disagreements, especially over socratic questions like “What is freedom?”, “what is justice?”, “you are not talking about knowledge,” and “you are only talking about schmeaning, not meaning,” and maybe also mutatis mutandis disputes that centre on how to individuate things, like “Are propositions fine grained or coarse grained?” or “Are meanings in the head or in the world?” and “Is a belief evidence only if it is true?” and “Are smartphones parts of our minds?” To be sure, the case is not aimed at old Socrates himself. This type of question is an excellent conversation starter in the agora of ancient Athens, if there is any interest to finding out about which properties people value, or which topics they care about; and it is also useful in other contexts—I even use the same question form in this writing. In particular, it is suitable in the stages of a philosophical dispute where the problem is being zeroed in on, as in chapter 1, and meanings are being specified, what can be called the “analytic phase.” But in this phase conceptual analyses are no more right than stipulations, and we can be like Humpty Dumpty and stipulate anything we like, as long as it is creates an interesting topic to explore, or picks out an interesting property, or however one likes to put it. When used in the stipulative sense, “What is…” is clear from context it can be used without the presupposition of a distinct, monistic answer that reflects the true nature of reality. If a Canberra planer objects: “Your theory is not a theory about X,” unless we assume monism it is still a proper response to ask “Well, that depends on what you mean by ‘X?’” A pattern is that many things in philosophy seem to come down to meanings, or are explainable in those terms, which means that semantic theory becomes more important for methodological investigation, like the current one. Call those methodological outlooks on philosophy that emphasise the importance of language, and maybe something like

concepts, in explanations of philosophical practice, linguacentric metaphilosophies. The chapter 1 theory of verbal disputes can be seen as a way to explain philosophical practice by appeal to language, and so too can certain parts of conceptual engineering from part IV of the introduction, although the former comes with a negative verdict and the latter sounds more positive and substantive. Now, from what has been said in this chapter, it appears to be likely that the linguacentric approaches, which, according to Williamson (2004:107) belong to a linguistic turn in philosophy with its best years behind it, and that “fewer and fewer of those who accepted the label ‘analytic philosophy’ for their work would also claim to take the linguistic turn.” What is the alternative metaphilosophical outlook on things, what type of explanation that’s not linguistic do they accept, of features like belief-formation, content, discourse, and maybe justification and truth?53 At least, limiting ourselves to disagreements and analysis, what is the alternative position on these things? We can distinguish linguistic or conceptual analysis outlooks on the subject-matter of philosophy from what we might call realism. Realism very generally, highlights the discipline’s claim to relate to “reality.” The following four vague points about what it means to be about reality, italicized, are tenets associated with this metaphilosophical realism:

First, philosophy is about reality, or the real world. And second, this philosophical reality is absolute and objective as opposed to relative or subjective, it’s out there, and independent of and external to us and what we do and believe. Thirdly, our philosophical beliefs are answerable to the worldy facts, or to how reality really is in itself, we discover how it really is rather than creating or making it up, meaning our beliefs about it can be true and false. And fourth, our beliefs are rich in content and are substantial, and attaining philosophical knowledge requires epistemic effort.

The broader conclusion here is that both with the idea of verbal disputes and with the idea of conceptual analysis there appears to be conflicts with this optimistic outlook on philosophy, realism. Here enters the conceptual engineer (see IV), “maybe philosophers aren’t just analysing language, but engineering it, and so realism can be preserved on that ground. Albeit largely verbal, philosophy is substantive.”

53 Justification and truth are the two most controversial explanatory objects, because linguistic explanations appear to undermine the substance of the propositions that are true or justified. See Boghossian (1996) and Williamson (2007), and also Boghossian (2010) and Williamson (2011).

Chapter three—Normative philosophy of language to the rescue?

Can philosophy be verbal in a good way? Can philosophers be engaged in verbal disputes that are substantive nevertheless? Maybe a large part of what philosophers are disagreeing about comes down to words, but since the disagreements aren’t descriptive, but normative, they aren’t talking past each other when they use the words with different meanings. Instead, they are advocating substantive propositions in conceptual engineering, or so goes the reply. And when they are asking “What is X?” neither are they seeking a conceptual analyses nor real definition, but an answer to “What should ‘X’ mean?” This chapter addresses a line of argument to the effect that philosophers actually are engaged in substantive “normative metalinguistic disputes,” or metalinguistic negotiations over conceptual engineering. The idea of a metalinguistic negotiation and the general theory of that phenomenon was introduced in the paper “Disagreement and the Semantics of Normative and Evaluative Terms” by David Plunkett & Tim Sundell (2013) which we looked at in (III) above. While that paper confines itself to normative and evaluative terms, typically drawn from an American political context, such as “marriage,” and “torture,” in a later paper in Inquiry, “Which Concepts Should We Use? Metalinguistic Negotiations and the Methodology of Philosophy” (2016) Plunkett develops the theory further and he also extends it to metaphilosophy, claiming that the same holds for philosophical terms too, which are not obviously evaluative and normative. His paper provides a response to the idea that philosophers are wasting their time verbally disputing mere questions of language. The idea instead, is that they are metalinguistically negotiating conceptual engineering. I introduce this idea (3.1, 3.2.), and then criticize that theory in light of the ideas from chapter 2 as well as in light of considerations about which types of reason that govern philosophical disputes. I argue that disputes in philosophy, insofar as they are

philosophically substantive, are about how the world is and not how worlds ought to be used. How to use words is a merely practical question, but the question of how things are is supported by epistemic reasons and is very different. Finally, I locate the foregoing discussion of conceptual ethics within the context of Chalmers’ theories (3.4.)

3.1. David Plunkett on metalinguistic negotiations

…or harmless verbal disputes. Many of the problem cases discussed in connection with the method of elimination were due to disagreements over what something is, turning on questions of the form “What is X?” After ruling out “What does ‘X’ mean?” and “What is X in reality?” as plausible interpretations in the next chapter, maybe they can be understood disagreements over “What should ‘X’ mean?” instead. In “Which concepts should we use?” Plunkett proposes this interpretation, as both a hypothesis about discourse in philosophy but also as a justification of actual methodological practices. The paper opens by a distinction between questions about reality and questions about language; or what he calls object-level issues and representational level issues (analogously what I called questions of reality and questions of language), together with the assurance that in philosophy it is primarily the former that concerns us. After listing some typical kinds of philosophical issues including real definitions and essences as were discussed above, he says that

when philosophers want to know about these kinds of issues, their inquiry isn’t about our representations of reality. Rather, it is about reality itself. This fact about the aim of much philosophical inquiry—namely, that it often ultimately aims at better understanding some part of reality, rather than our representations of it—is a crucial one to keep in mind when thinking about the nature of philosophy and its methodology.(829) before the main thesis of the paper is stated:

In this paper, I am going to present a framework on which this kind of understanding is sometimes (perhaps often) mistaken. According to this framework, in many philosophical disputes that are seemingly about [object-level issues] the disagreement that is expressed is not actually about these things. Rather, it is about our ways of representing reality, and, more specifically, about which ways of representing reality we should adopt. In slightly more general terms, the normative question is often this: which concepts we should employ for the purposes at hand. (830)

Plunkett is, in other words, defending an interpretation of philosophy as conceptual ethics. Philosophers are verbally disputing substantive topics like “Should freedom be the ability to do otherwise?” “Is freedom 1 a defective concept, so we should instead talk about freedom2?”, “No, freedom 1 is more valuable” or simply, “What should ‘freedom’ mean?” instead of what philosophers appear to do on the surface, disagreeing over whether we are free. “It is tempting to think that such disputes straightforwardly express disagreements about these topics. And, indeed, this is exactly what many philosophers think. In contrast to this, I suggest that, in many such cases, the disagreement that is expressed in the dispute is actually one about which concepts should be employed” (832).

Before this position gets to be defended, the paper’s first part is devoted to a thorough explication of the theory of metalinguistic negotiations with a focus on its role in ordinary conversation (830-52). This part of the paper presents a framework of technical vocabulary which is slightly more fine-grained and linguistics-oriented than Chalmers’, but at bottom they come down to the same important distinctions and emphasise the same important points. For example, the distinction Chalmers draws between first-order and metalinguistic issues is just the same as the one drawn by Plunkett between object-level issues and representational level issues, the only difference being that representational level issues includes not only issues of language, but also issues that regard our ways of thinking and talking more generally, how we conceptualize things, and so on. As it were, Plunkett’s framework is just a development of the basic notions used by Chalmers’, with some added details and more precisely defined terms. A word more on the two theoretical frameworks. Where Chalmers talks in terms of subdisputes—the disputes that would have continued the disagreement, in Plunkett many of the same points are expressed in terms of the distinction between literal and pragmatically conveyed content instead, and in terms of word meanings, as well as the idea which he draws from Chris Barker (2002) of a metalinguistic usage. When a term is mentioned (rather than used) to express a view about the meaning of that term, or, relatedly, how to correctly use that term, then it is used metalinguistically (833):

An important part of what goes on in much communication is this: speakers hold fixed their views about what their words mean (and what their interlocutors words mean), and then use those words to communicate about parts of the world that one speaker (or more) doesn’t hold fixed views about. The process, however, can also work in the reverse. That is: we can hold fixed our views about the world around us, and then, in turn, use (rather than mention) a word to communicate views about the meaning of that word. These views can either be ones about the descriptive issue of what a word does mean or about the normative issue of what it should mean.

A good example of a metalinguistic usage is when young children are taught language; most of the words we know we have likely learnt by this communicative tool. We sometimes say things that are trivial, literally, but convey metalinguistic propositions. Language learning is by means of the descriptive type; and ostensive definitions by example too, as in “that cabinet is Art Deco.” In Chalmers, moreover, a dispute occurs (presumably within a context) when one speaker assents, and the other doesn’t to the truth of a sentence (whether or not it actually expresses a disagreement), and in Plunkett it is essentially the same: a linguistic exchange that appears to express a disagreement. It is also added that when the topic of a dispute equals its literal topic, which is to say, when a dispute centres on the content that is literally expressed by the disputants, then it is a canonical dispute. And when the actual disagreement differs from what it appears to be, it is non-canonical. In Chalmers’ terms, a metalinguistic dispute is either a dispute that is explicitly about words, or a subdispute that is metalinguistic or metalinguistically mediated (meaning its parent dispute is verbal), whereas in Plunkett’s usage, “metalinguistic dispute” picks out the implicit, or non- canonical ones only, while it is canonical metalinguistic disputes that have words as their literal content:

There are two parts to what makes something a metalinguistic dispute. There is the part marked by the term ‘metalinguistic’: at least one speaker employs a metalinguistic usage of a term. And then there is the part marked by the term ‘dispute’. For our purposes here, we can define dispute as follows: a linguistic exchange that appears to express a disagreement. (835).

And finally, in Chalmers’ system, a metalinguistic negotiation is a verbal dispute with a normative metalinguistic dispute as its part. If it has a descriptive metalinguistic dispute as its part I suggested in (1.1.) that they also be understood as metalinguistic negotiations. The reason for this is that the descriptive issues at stake are either addressed explicitly by mention of “X” (as in “Expression ’X’ means T,” “No, ‘X’ does not mean T”), or it is addressed implicitly and “X” is only used and not mentioned. If it be the first, then it would not be a metalinguistic negotiation, but just a normal (canonical) metalinguistic dispute. If it is the latter, then it seems to me that it also has to involve some normativity, they are after all using the words differently, so if that’s what the disagreement is about, then to use words in accordance with one descriptive metalinguistic belief rather than another involves an assent to the normative proposition that that word should be used with that meaning at least in the current context. In other words, since metalinguistic negotiations aren’t explicit, there is no such thing as a descriptive equivalent of a metalinguistic negotiation; descriptive metalinguistic disagreements are always merely verbal. So metalinguistic negotiations fit neatly into Chalmers’ system. The way Plunkett defines a metalinguistic negotiation is as a normative metalinguistic dispute (where metalinguistic disputes are always non-canonical): “As I will define things in what follows, a metalinguistic negotiation (or, equivalently, a normative metalinguistic dispute) is a specific type of metalinguistic dispute. In basic terms, it is one that centers on normative issues about what a word should mean, or, similarly, about how it should be used, rather than the descriptive issue about what it does mean or about how it is used” (838). In all, despite certain terminological nuances, the two frameworks merge seamlessly, and there appears to be few complications with either. Chalmers’ framework emphasises more what’s at stake (the subdisputes) while in Plunkett the focus is more on disagreed content (pragmatically conveyed), but the content can be the same whether it’s a subdispute or an implicit disagreement, and they are theories of verbal disputes, and answers to the question “What is really going on with disputes in philosophy?” Now, the paper goes on to look for the symptoms by which to diagnose a dispute as a metalinguistic negotiation. Here are the four symptoms that support such a diagnosis:

(A) There is good evidence that the linguistic exchange is a dispute. That is: there is good evidence that it is a linguistic exchange that appears to express a disagreement. (B) There is good evidence that the dispute really does express a disagreement. (C) There is good evidence that speakers in the dispute mean different things by (at least) one of the terms in that dispute. […] (D) There is good evidence that the disagreement expressed in the dispute, insofar as there is one, isn’t just about descriptive information about what a word does mean, or how it is used. (This is something that helps provide evidence that it is a normative metalinguistic dispute, i.e. a metalinguistic negotiation, and not a descriptive metalinguistic dispute). For example: one good piece of evidence here would be that speakers persist in their dispute even when they agree on the facts about a term’s current meaning or current use. (847).

The case study from which these clues are extracted is the often used dispute about Secretariat the racehorse. It’s an oft used example of a metalinguistic negotiation, originally from Ludlow (2008) after he overheard it on a sports radio show. The dispute centres on something like the sentence “Secretariat is an athlete” after the horse was featured on a list of the fifty greatest athletes of the twentieth century in an issue of Sports Illustrated. “Ludlow proposes that what is happening in the Secretariat case is this: each speaker is advocating for their preferred way of using the term ‘athlete’, in the context at hand. Sundell and I think this basic idea is correct. Put in our terminology: Ludlow’s Secretariat case is a metalinguistic negotiation (841).” (Ludlow refers to the same phenomenon as “lexical warfare.”) It is without doubt a dispute, and it is also clear that they are using “athlete” with different meanings—(A) and (C). In terms of verbal disputes, (A) and (C) are evidence that a dispute is verbal. The dispute over whether Secretariat is an athlete is obviously verbal, owing to a disagreement over the meaning of “athlete,” and in particular, whether horses qualify as athletes. By the method of elimination it would turn into subdisputes like “Does Secretariat deserve praise” and metalinguistic subdisputes like “don’t use ’athlete’ about animals” and metalinguistically mediated subdisputes like “Horses can’t be athletes.” But it also has features (B)—that they are disagreeing on the same topic, and (D)—that this topic is normative, which is what makes it a metalinguistic negotiation and not just a mere verbal dispute. So what establishes the latter two? After analysis and some further suppositions about the details of the case, it is found that

there is pressure to think the speakers in this dispute are engaged in a dispute that really does express a disagreement, and is not just a ‘mere talking past’. Part of what supports this idea is this: the speakers are engaged in a heated debate, take themselves to be at odds, and continue in their dispute even as their divergent patterns of applying the term ‘athlete’ become clear.

And the reason to think it is normative is that

it makes little sense to think that they are engaged in a dispute about what the term does mean. To see this, suppose that a linguist who is an expert about word usage came on the air of the radio show and said ‘I am certain that “athlete” just means something that excludes non- human animals’. This might settle the issue for someone who thinks that that way the term should be used is one that conforms to current usage. (There is no reason, after all, that one can’t advocate for a conservative view in conceptual ethics, rather than a reforming or revolutionary one, when engaged in a metalinguistic negotiation). (842)

We can now ask: Why is the dispute substantive? Bring back the two conditions for substantivity that have been in use throughout:

(1) The disputants address the same topic (2) That topic is substantive

Since to address the same topic, as that expression is used here, just means that the dispute expresses a disagreement, Plunkett’s criterion (B) satisfies condition (1), and the two are

essentially the same: The shared topic is how the term “athlete” should be used. So even though they use the term with different meanings, they are disagreeing about the same topic; the synchronic variant of Strawson’s principle (2.4.) is false: One can use different meanings, but still address the same topic, without having to let go of the very compelling diachronic variant, that changing meanings changes topic. But now, why is it a substantive topic? Why does it matter what the term “athlete” means? The secretariat case is a good case-study because the answer to this is not as obvious as in the paradigm cases considered introductory, such as “marriage” and “torture,” which have obvious substantive issues at stake, such as the relative treatment of gays and (interpretations of) the scriptures, and about the ethical constraints on how intelligence officials can collect information. Plunkett’s theory in this less transparent case is that the disputants are engaged in conceptual ethics, the same thing that has been referred to in the above as conceptual engineering. His two primary examples of conceptual ethics are drawn from philosophy. The first is Sally Haslanger’s argument in her “Gender and Race: (What) Are they? (What) Do we want them to be?” (2000) that we should engineer, which is to say edit or replace, the concepts of gender and race so as to align better with our beliefs about social reality; how it is and how it ought to be. Her argument is that insofar as these words should be used at all in talk about social identity, racial and gender identity in particular, they should be used with different meanings than how they are commonly used. And so Haslanger proposes a series of definitions for these terms that are supposed to be better at accomplishing the social and political aims that she thinks we should be pursuing. The second example of conceptual ethics is from theoretical philosophy, and it is the position defended in Writing the Book of the World by Theodore Sider (2011) that metaphysicians ought to use the terms that match up with the “objective joints in reality,” an hypothesis very similar to the appeal to natural kinds that was considered in connection with the above defence of real definitions. In Sider’s metaphysics there are right and wrong concepts to use, and the reason some are preferred to others are purely theoretical, namely, that they better mirror the fundamental structure of the real world. Following David Lewis’ theory of natural properties from his “New Work for a Theory of Universals” (1983), this fundamental structure is given by which properties are more natural than others. In Lewis’ ontology, i.e. his theory of what fundamentally exists, natural properties are the properties instantiated by spacetime points, roughly, the spatial and temporal relations that hold between them and whether they are occupied or not. These properties being natural, Sider argues, means they impose linguistic normativity upon metaphysicians to cast their theories in terms that pick out, or are constructed out of terms that pick out, natural properties. From fundamental metaphysics to socio-political moral philosophy,

These are arguments in what we can dub conceptual ethics. We can use this term to cover normative and evaluative issues about thought and talk, including, centrally, normative issues about which concepts one should use in a given context. Haslanger’s and Sider’s arguments concern this central topic in conceptual ethics. And so does the dispute in the Secretariat case. They all concern questions of what concepts we should use, in a given context. (843)

And so, because metalinguistic negotiations concern the topic of conceptual ethics, they concern a topic which is substantive—condition (2):

This basic normative topic is a substantive one. It is substantive because not all concepts are equally good for using in a given context, and so we should be using certain concepts rather than others. This basic thought can be accepted by people with a wide range of views about what makes a concept better to use in a given context than others. (843-4).

And since metalinguistic negotiations are verbal disputes, they are substantive verbal disputes, and the condition is added that a dispute is substantively verbal if it arises in virtue of a disagreement over conceptual ethics.

3.2. Metalinguistic Negotiations in Philosophy

The two examples that were presented of conceptual ethics in philosophy were examples of explicit—applied—conceptual ethics. Haslanger does not convey her revisionary proposals through metalinguistic usage of “race” and “gender,” and Sider’s proposal is purely theoretical; it would have been practically impossible to actually talk in terms of occupied spacetime points and the relations between them. But that conceptual ethics occurs explicitly is not a significant finding. “Philosophers often overtly and explicitly debate which concepts we should employ, or, more generally, normative and evaluative issues about the tools we are using in our thought and talk. Indeed, these kinds of debate occur throughout the field, in subfields ranging from meta-metaphysics to formal epistemology to applied ethics (830). Plunkett’s idea of philosophy as conceptual ethics is first and foremost a thesis about apparently object level disputes where normative linguistic proposals are not made explicitly, but through metalinguistic usage. It is on one hand a linguistic, theoretical interpretation of disputes in philosophy, which on the other hand, it can also be described as a normative standpoint about the activity we call “conceptual ethics,” that it is legitimate and substantive, and certainly not pointless. The argument for this thesis is that “if we have good reason to think that some (perhaps many) prosaic disputes are metalinguistic negotiations, then we also have good reason to think that some (perhaps many) philosophical disputes are metalinguistic negotiations.” (853). More precisely, argument is that many of the diagnosive criteria (A-D) can apply to philosophical disputes, the example being, again, the free will case, “Is our will free?”, or, supposing agreement that determinism is true, “is freedom compatible with determinism?” Can it be extended to philosophy?

Now consider the following. Can we flesh out the details of such a dialogue such that it has the features (A)–(D) that I introduced above (the ones that give us evidence for thinking that the dispute is a metalinguistic negotiation)? And, moreover, is such a way of fleshing out the details one that is representative of how disputes of this kind sometimes happen in philosophy? The answer, I think, is a clear ‘yes’. (855)

Since the idea of metalinguistic negotiations in this last chapter is supposed to constitute a possible solution to the problem of verbal disputes in philosophy, we can just suppose for the sake of argument that the free will dispute is verbal, so that both the clues (A) and (C) hold. That is, the dispute we are considering over whether free will is compatible with determinism is a dispute where the term “freedom” is being used with different meanings; one by which it is compatible with determinism, and one in which it is

not. How about the features (B) and (C)—that they disagree on the same topic, and that the topic is normative—that support thinking that the dispute is a metalinguistic negotiation and not just a merely verbal talkng-past dispute? In very short words, Plunkett supports (B) and (D) because (B) is supported by what philosophers consider themselves as doing, and while (D) also adds some support to the idea that philosophy is not pointless, the reason to think that (D) is that given the disputants are using the term with different meanings, then even if they were aware of that, their divergent usages would persist, they would not synchronize. These two points in more detail. Here we go again, following a similar line of reasoning as with Secretariat the racehorse above. The former, i.e. (B), that the dispute expresses a disagreement, would fail to hold if the dispute were merely verbal. The central idea behind metalinguistic negotiations is that the dispute can express genuine disagreement over the same topic even when the terms are being used in different ways, which is suggested if the (C) clue holds. It is natural to think that if one is using “freedom” with a compatibilist meaning while the other one is using it with an incompatibilist meaning, then they are discussing different topics—“Do we have compatibilistic freedom?” and “Do we have incompatibilistic freedom?” respectively. And supposing they agree on the truth of determinism, then that would dissolve the dispute by an agreed “Yes.” to the former and a “No.” on the latter—they were just talking about different topics, failing condition (1). At this point Plunkett comes along with the idea of a metalinguistic negotiation and says: “But they are not talking past each other, even if that is so. Instead, they are each advocating, via competing metalinguistic usages of the term ‘free will’, for a view about what the word ‘free will’ should mean in this context. They disagree in their normative views here about how this term should be used, and that disagreement is being expressed in the dispute” (587). More precisely, the topic of disagreement is which meaning “freedom” should have in the context, or, in Plunkett’s terms, which concept should be connected to that expression:

As with the Secretariat case, the issue here is about concept-word pairing: it is about which of a range of competing concepts should be paired with the term ‘free will’ for the context at hand” (857).

So (B) holds, but how about (D)? What is a reason to think this metalinguistic disagreement is normative and not descriptive? Because the disputants are using the term differently rather than addressing the metalinguistic issue directly, it is crucial that their disagreement is normative rather than descriptive, as noted in (1.1.) and in the previous section. For a descriptive disagreement about what a term actually means in the current context, is tantamount to saying the dispute is a misunderstanding. Descriptive verbal disputes can only be substantive if they are explicitly concerned with words, but we have already supposed that the current one is implicit, i.e. that the terms are being used differently. Plunkett makes a very similar point to the ones that were used to refute conceptual analysis approaches to socratic questions in the previous chapter, that it is largely irrelevant to the first-order issues that philosophers are interested in. He invites us to consider what would happen if the disputants were presented with descriptive facts about

what the key term means, and he argues that the dispute would likely persist unaffected by agreement over such facts.

Consider research into what everyday people mean by ‘free will’, or, similarly, what a group of philosophers mean by it. For many philosophers involved in disputes like (8) [i.e. the free will dispute] that is certainly not going to be taken as dispositive for the issue in question. Indeed, many will think that such information is not only not dispositive, but, indeed, largely irrelevant. (856).

This is analogous to the marginalization of conceptual analysis of chapter two. Descriptive facts about what words mean are insignificant to the conceptual engineers, as they are interested, rather, in what they should mean. The analysis of this current dispute over “Is our will free?” is taken to be representative of many disputes about the same question, and it is also held to be generalizable beyond that particular question to several other disputes in philosophy (857). Even when philosophers aren’t engaged in explicit conceptual ethics like Sider and Haslanger, disputes that appear to be verbal because the terms are being used with different meaning are actually philosophers engaged in conceptual ethics on an implicit level. Disputants are Plunkett’s theory about philosophy as implicit conceptual ethics “matters for the contribution it makes to develop a reflective self-understanding of what we are doing when we are doing philosophy.” And to underscore that it is a proposal about philosophy on a deep level, “…notice that part of what makes this proposal viable has nothing at all to do with what goes on in disputes, nor even with what is going on in linguistic communication. Rather, it has to do with a view about what is going on when someone is engaged in inquiry (862).

3.3. Against the idea of substantive verbal disputes

In this section I provide two related criticisms towards Plunkett’s theory of metalinguistic negotiations in philosophy. It centres primarily on the distinction between practical reasons (what ought to be done), and epistemic reasons (reasons for something being true). but since Plunkett’s is a multifaceted theory that’s both about philosophical inquiry in general and also particular disputes and pragmatic mechanisms to convey information, there is such a wide selection of points of controversy that one might easily become stuck for choice, so these many points will be drawn into conflict with the epistemic/practical distinction. I begin by discussing reasons, concluding vaguely that the reasons for a given modulation of language must somehow equal the reasons provided in the object level philosophical argumentation. This is strange, because they appear to be very different, and then I nod at the distinction between epistemic and practical reasons (3.3.1.) Next, I ease in on the problem by discussing why metalinguistic negotiations are substantive and state the line of argument that follows. (3.3.2.) Next, I claim, like the methodological discussion of section 2.4. suggested that philosophers can, that epistemic reasons are the ones we are interested in (3.3.3), and then argue that practical reasons are marginal in philosophy.

3.3.1 Reasons for revising concepts The immediate problem with Plunkett’s hypothesis is that the normative metalinguistic proposition P at stake in a metalinguistic negotiation over some sentence S needs to be epistemically supported by the premises and reasons that are given for S. Otherwise, it is dubitable that P can be the topic of the dispute over S. To do that, the reasons, or premises that are made use of to support those metalinguistic bids about B must be interpreted so as to favour P or disfavour P depending on which side to the dispute over S they come from. Reasons given for P needs to equal the reasons given for S. (For a similar argument, see “The argument argument” against metalinguistic negotiations in (Ball, forthcoming 2018)). Take, for example, the argument for skepticism that

I can’t know whether an evil daemon is fooling me so that I’m always wrong Unless I can rule that out, I can’t know that I have hands If I can’t know that I have hands, then I can hardly know a thing

So which one expresses the metalinguistic proposition P? Let’s say the metalinguistic proposition expressed is that the proper threshold for knowledge is one whereby you can know that you have hands, premise three, but it is also one that sets the threshold at the point where that hypothesis can be ruled out, premise two. In other words, the threshold for “knowledge,” interpreted metalinguistically, is sat by those two metalinguistic usages at the point at which, first, one can know that one has hands, which in turn is precisified even more to the point at which one is able to rule out the defeating daemon hypothesis. Let’s say the premises area advocating this high threshold position on the term “knowledge.” Let’s call the property the want to bring into play, knowledge-1 (ruling out- 1, knows-1, and so on). Now, given that one can’t rule out-1 the defeating hypothesis, it may surely follow that we know-1. But if “know” just meant know-1, in that premise would have been false: If the threshold of knowledge is already set to the point at which that hypothesis can be ruled out. The only way the argument would work is if “knowledge” referred to knowledge-1 to begin with. Following Strawson’s principle (2.4.), that is just a different topic. It might be that the interpretation of the premises is mistaken, but I think that however they are interpreted, the same point will hold: At least one of them would have been false if the metalinguistic propositions advocated attained acceptance. In Chalmers’ system this is not a problem, for implicit disagreements are understood just as subdisputes—the disputes that would have continued the disagreements, and not as actual pragmatically conveyed content that is actually the topic of the dispute. It is therefore up to the disputants what is a relevant reason for what. Plunkett makes an important point:

Crucially, the fact that normative issues about concept choice matter does not depend on the way in which speakers argue about those issues. Rather, it depends on the content of those issues. Thus, this kind of issue in conceptual ethics remains a substantive one regardless of whether it is argued about via the assertion of literal content (i.e. via the semantics) vs. other aspects of the communication (i.e. via the pragmatics).

which can be granted, but it is important to remember that it is the disputes that need to be legitimised and shown to be substantive, and not the disagreements of disputes. There are two different questions, as we saw, whether a dispute is substantive and whether the disagreement that it arises in virtue of is. Language can sometimes be substantive; but that does not mean verbal disputes are. Something must relate the topic of the object level controversy to the linguistic one to the literal disputed one, otherwise its verbal for the same reason irrelevant and metalinguistically mediated disputes are—they are about a different topic, such as knowledge-1 and not knowledge, or “knowledge” and not knowledge (whichever one of them it means). About this we can be liberal, and set only a vague condition, because “topics” is such a vague concept already. Let’s say, very generally, that the metalinguistic proposition must have something to do with the object level proposition, for example that we do not know anything. For example they must have consequences as to each other’s truth, or be epistemically grounded in similar evidence, or share similar consequences, or something like this. To draw some inspiration from another advocate of the metalinguistic negotiations hypothesis, Ludlow (2012):

[T]he mere fact that there is variation in the meaning of these expressions does not mean that anything goes and certainly does not mean that one is entitled to stubbornly dig in on the meaning of a term. To the contrary, the process by which we become entrained often involves argumentation and argumentation is a normative activity. (144.)

It’s an idea that is central in conceptual ethics; language is rationally constrained, and therefore substantive. But those reasons must connect somehow with the actual arguments in the dispute, or else they are different topics. Recall the minimal qualification of Humpty Dumpty’ principle, about why not “anything goes;” very simply, we need to stay on the same topic, avoid verbal disputes and understand each other, in general. But, these minimal requirements are not enough to unite the substantive linguistic topic with the reasons provided for and against the thesis. Something more is needed, namely, as Ludlow says, that there is normativity, like the normativity governing argumentation, that one can be wrong or right. The sought wrongness and rightness is supposed to be closely connected somehow to the truth and falsity of corresponding beliefs. For example, if it is right to assign the G- implying meaning rather than the wrong H-implying meaning to some predicate “F,” then it should be true that Fness entails Gness rather than Hness, and that all Fs are Gs, all F-ers are G-ers, and so forth. The way we ought to believe these latter truths has to relate in a certain way to the way these former meanings ought epistemically to be assigned. But how this can be done is far from straight forward. The reasons for these beliefs have, in some crucial sense, to equal the reasons for the meanings we assign to terms, the former of which are under Realist assumptions supposed to be truth-directed, answerable to reality, apt for truth and falsity, and all that we now in short call epistemic. It might be true and false, and necessarily so, that all Fs are Gs, and we might have good reasons to believe it, very well, regardless of that having any direct link to which meaning is the right meaning to assign to “F.” Or so it appears. These questions seem on the face of things unrelated, with a different set of reasons pertaining to each. So what explains this? We have that being reason governed provides substantivity, but aren’t questions of language choice governed by a different type of reason?

3.3.2. Substantive in virtue of which types of reasons? The previous discussion concluded speculatively, about reasons, and what it is for language to right and wrong, and Plunkett urge was that we focus on the metalinguistic content and set the reasons provided for it aside. But the same problem with different topics; what ought to be done and how things are; I think reappears also if we focus on the mere content itself. To ease in on the general problem I have in mind, consider the question again of why language choice is substantive, besides what was suggested in the previous subsection that it’s rational. The line of reasoning that leads to doubt Plunkett’s theory starts from the discussion with the Secretariat case over why linguistic normativity is a substantive topic. The reason provided was that it was an instance of conceptual ethics— but, what does that mean?—a better explanation can be demanded. Something more is needed, for why is not this itself a pointless exercise? I think the answer to this is something like:

How can propositions about what ought to be done not be substantive? It seems that sentences of the form “We/you/they ought to do this/that/the other” simply cannot fail to have that feature—they have direct practical consequences about what should be done in the world, and the same with disputes over these types of topics.

Few other types of disputes have equally strong claims to substantivity, so this line of reasoning makes sense. Maybe claims of the form “if we do this, then that will happen” has a similarly strong claim, and maybe do also claims that are empirically verifiable and falsifiable like “this jug contains twelve marbles,” but normative statements and disputes there over are certainly among those few types of claims that are near guaranteed to be substantive. However, even granted all this, there is still the further question of what explains or, maybe, grounds the normativity of these questions. For example, the source of normativity for “torture is evil” is that it condemns a certain sort of action; it is connected to action, and assessed with respect to whether that action should be done. Another sort is with the racehorse Secretariat; it is substantive in virtue of being a normative bid about what should be done. These paradigm metalinguistic negotiations are grounded in disagreements about who should be tortured and who should feature in magazines and who should get to marry, and so on. But still, these disagreement centres on ideological outlooks on morals, quite broadly. Recall Carnap’s “In logic there are no morals. Everyone is at liberty to build his … own form of language, as he wishes” (ibid, 10). There might be substantive moral disagreements about who should feature in magazines and who should be tortured, and who should get to marry, but why believe there are morals in language itself? If a philosophical dispute is a verbal dispute, why believe a term should be used with one meaning rather than the other? Plunkett’s reason, in connection with the Secretariat case, sounds quite commonsensical:

But it is also difficult, surrounded by confusion, as I see it. That might hold for a case like Secretariat the racehorse, fair and well, and the same type of explanation can be given for other paradigmatic metalinguistic negotiations too, such as “marriage,” or “torture.” But the idea with which I want to take issue is that the same type of explanation also can apply to philosophical disputes, which is just what Plunkett does with the free will case:

As with the term ‘athlete’, there is a lot of important resonance that the term ‘free will’ has. This is resonance that obtains not in virtue of the specific meaning of the term (that is: because of the particular concept it is paired with, in a given context), but rather because of the background social and cultural facts about how that term has been used over time, and what kind of functional role it has played in our philosophical discussions.

Are philosophers really disagreeing about how language ought to be used by disputing what appears to be object level topics? My argument is that either this is right, in which case philosophers are merely verbally disputing, or it is wrong. I will argue like the following:

(i) A dispute is verbal if the parts had agreed if they had agreed about meanings, i.e. accomodated. (ii) Therefore, the only reason to engage in metalinguistic negotiations is if there are reasons not to accomodate. (iii) There are no reasons not to accommodate in philosophical disputes. (iv) Therefore, assuming philosophers are reasonable, no disputes in philosophy are metalinguistic negotiations

The previous chapter has made a case that analyses can’t be in substantive disagreement, and so that there are no reasons for philosophers not to agree about them, i.e. accommodate, premise (iii). However, this premise might be too strong, because there are as many reasons language can be right and wrong that there are reasons not to accommodate, which recent literature on conceptual ethics has helped to bring out. Premise (iii) will have to be modified, and a candidate for replacement is clear—there are no epistemic reasons for using a certain word rather than the other. Language is governed by practical rationality, or “human gain and loss,” as it is put in Carnap (1968, 264).

3.3.3 Epistemic and practical reasons Begin by asking Why are philosophers refusing to accommodate? and What are the reasons not to? There can be many explanations, but why think there are reasons that can justify it. So in order that Plunkett’s hypothesis is not understood as just a hypothesis about philosophers engaging in unfortunate behaviour, it can be supposed that philosophers are actually doing this on purpose and that there actually are reasons to metalinguistically negotiate the ethics of what words ought to mean. If se suppose that there are reasons and not just explanations, we can follow moral philosophers in drawing the most general distinction among reasons there is, reasons that are evidence about how things are, and reasons pertaining to what to do, epistemic and practical reasons. I will try to argue that to rest the substantivity of questions about how the world is upon consequences pertatining to what to do involves an understanding of those questions and indeed the discipline as a

whole that is fundamentally misguided. Normativity is no way to legitimise philosophical disputes being verbal,, because philosophy is not concerned with mere descriptive facts Pascal’s Wager is perhaps the most well-known example of a practical reason placed in an epistemic reason’s stead. This 17th century theistic argument from Blaise Pascal employs game-theoretical considerations in favour of believing in God’s existence, that the choice of believing in vain incurs only insignificant losses in comparison to the potential infinite gains and losses associated with eternity in heaven or in hell, “If you gain, you gain all; if you lose, you lose nothing.” 54 The argument might rationally justify the act of believing in God, but it provides no epistemic reasons to so, or put another way, it is an argument for believing in something and not an argument for why it is true. Arguments for believing why things are true, i.e. reasons about how things are, are what philosophers are dealing in, or at least that’s a charitable supposition to make about their actual practice, and it is a bad supposition that they are dealing in practical reasons, i.e. reasons to do things. It is either the one or the other, as the two are to radically different. All though they are both reasons, we can even go back to the idea of “topics” that they are about different things, and must be distinguished. And now seems like a good point to invoke the idea that some property is what we value more than some other property (1.7., 2.4.), and therefore that property should make up the continuation of the topic. Epistemic normativity is what philosophers value, and not the practical type. What ought to be done and how are things, philosophers are most likely dealing in the former, given that they are rational and value truth and so on. Would not the other alternative reduce philosophical questions to mere practical resolutions?

3.4. The Is of reality as the Ought of language

In this section I consider two responses to the argument provided, and reject both. The first is that philosophical questions have practical implications,or practical import, about what we should do in the non-linguistic world, so it matters what philosophical terms mean, so they are governed by practical reasons. The second is that there are a lot of theoretically valuable aspects of how the linguistic frameworks in philosophy are figured, or engineered. I argue that the first is only marginal, and as for the second, that that there are no reasons not to agree on frameworks. One can’t argue over anything other than the practical features of frameworks, so it’s something we rarely do. Instead, we simply accept the other’s terminology to try to get to the real substantive issue. Neither can constitute subject of substantive dispute. More generally, they have in common that, and are rejected for the reason that- they conflict with the reality-oriented conception of substantivity whereby philosophical questions are about how reality is and governed by epistemic

54 Pensées

reasons. The idea that they are the same—the is of reality as the ought of language—must be rejected. Conflating of the two is the reason these disputes are merely verbal.

First response: Are philosophical questions ethical questions? Admittedly, it is difficult to find the exact point of conflict in this case, but I think that something pointing towards the heart of the controversy is the distinction between being about how the world is and how the wold should be. A way to case the problem in terms of this distinction is like this:

Philosophy is not governed by practical reasons; metalinguistic negotiations are governed by practical reasons only. Philosophy is governed by epistemic reasons, but metalinguistic negotiations are never governed by epistemic reasons

It can be argued that philosophy actually is governed by practical reasons because philosophical questions have practical consequences, in that we will live differently depending on which terms philosophers manage to negotiate into acceptance. If we accept that we do not have free will in many of the senses from section 1.5.,, for example, we might be disposed, or even better: reasonably committed, to treat people differently, for example with respect to punishment.

We can expand on this basic idea as follows. As with the term ‘athlete’, there is a lot of important resonance that the term ‘free will’ has. …This is because, given the importance of the term ‘free will’ in our philosophical discourse, and the relatively fixed functional role we grant those things called ‘free will’ in our practices, it matters that the concept gets paired with this specific word ‘free will’ as opposed to something else. Again, a similar point holds for lots of other philosophical terms that are central to philosophical discussion, e.g. ‘knowledge’, ‘self’, ‘freedom’, ‘morality’, etc. (857.)

Again, this type of explanation strikes me as confused, because reduces theoretical disagreement to war over labels. It is true that “freedom” has some positive connotations, but what can that contribute with in a philosophical context? If the question of free will is substantive, whatever this “cultural resonance” is, it should be irrelevant. Disputants to the question of whether we have free will are not disputing because of the connection that term has to our cultural practices with freedom (punishing free wrongdoings, pitying unfree wrongdoings, and so on). Ethics has got nothing to do with serious theoretical disagreement. Questions about how things are should not be interpreted as questions about how things ought to be. A look in the earlier paper “Conceptual Ethics I” (Plunkett&Burgess, 2014) presents the same type of explanation; the philosophical question of personal identity through change of constituent parts is reduced, as it were, to questions about what to do. They consider the question of whether a fancy teleportation device of the future, which disassembles the user and then reassembles them someplace else, is a quick means of travel or just a fancy way to die:

These are conceptual or semantic choices, to be sure, but they are not ‘merely semantic’, in the popular, pejorative sense. On the contrary, we would expect the natural progression of a society opting for the ‘travel’ paradigm to look quite different from that of a society opting for

‘death’ – different with respect to how widely the technology is used, the market for component parts, property law in cases of teleportation and so on. (1092.)

Argued here is that the verbal question is substantive on the grounds of the practical consequences pertaining to the usage of the device. It might of course be that some philosophical questions do have such implications, but that is not what they rest on for their substantivity. However, in moral philosophy, disputes might have such practical consequences, let’s say that two disputants are disputing whether it is right or wrong to do action A. If it is also true that what is right should be done, then it might follow that a disagreement over A is a disagreement over what should be done. But then again, this does not provide much help, let alone justification, for those parts of philosophy that are not about how we ought to live and what we ought to do.

Second response: Are philosophers negotiating frameworks? In engineering there are substantive normative disagreements over which material to use, or which blueprint is best to implement, and in general, how to build the thing that is going to be built. The hypothesis under consideration is whether philosophers are normatively verbally disputing, over how language ought to be used, and there are actually some ways philosophers are familiar with where reason, or at least normativity, does concern language. There is linguistic normativity associated, for example, with principles of least effort in communication, or with the virtues elegance and simplicity in axiomatic systems, including programming languages. And also in empirical science, let’s say, there are benefits associated with having a theoretical framework that is useful and well- predicting, irrespective of whether and how it represents what, following Carnap’s distinction between a theory’s linguistic framework and its content (2.4). Can it be said that philosophers are negotiating conceptual engineering and that conceptual engineering concerns these types of things? Let’s consider philosophical equivalents for these metaphors on how language can be governed by normativity. What would “least effort” and the others mean for philosophy?

Least effort governs not only communication, but understood more broadly it governs the linguistic frameworks that philosophers cast their disagreements in. Elegance and simplicity go hand in hand, and just like we ought to believe the simplest explanation in ordinary cognition, philosophers should also use the simplest framework Predictive power seems also to be applicable, if not to philosophical terms, at least to frameworks. For example, maybe in moral philosophy what the theory virtue ethics predicts is the right way to live is a way that matches well with our judgements about what the right way to live is, or a theory about knowledge in the face of skeptical doubts might predict that we do have knowledge in those situations where we tend to think we do, and so be well-predicting.

Since these ideas can all be said to govern philosophical disputes, it can be argued, metalinguistic negotiations aren’t pointless nonetheless.

Adding to this, one might also argue that they are very common, because they are necessary preconditions of stating something that can be disagreed about. Linguistic frameworks (like the current one “substantivity”, “verbal”, “about reality” etcetera), are important aspects of a theory, and it might be difficult to distinguish the thesis I am making from the linguistic framework, and so too with many other philosophical theories. Philosophers are arguing over linguistic, or conceptual frameworks; systems of representations, indispensable to philosophical theorizing, and the meanings of our words.55 Against this response. Are philosophers negotiating frameworks? No. Very much like the arguments against conceptual analysis in 2.1., we can point out that these are just marginal considerations. Logicians, philosophers of mathematics, and formalistic philosophers in other sectors, operate with frameworks like these are front and centre, but still these considerations are marginal concerns, and are relevant only with respect to a specific one of the discipline’s wings. While these three points about practical reason, and many more potential ones too that were not mentioned, might apply to philosophy at times; and with the necessary adjustments made extended to explain many things about philosophical disputes; they would be unable to justify them. Metalinguistic negotiations are dubitably about any of these things insofar as philosophers are not all about least effort, elegance and simplicity and predictive power of frameworks. Besides being marginal, there are two more reasons to reject ideas of this sort. Just like with the conceptual analysis objection above, there is a dilemma: Either they are marginal, or philosophers are wasting their time. Like it was proper to ask why analyses should give rise to disputes, why should frameworks be negotiated? Bring back Rippon (2008), the arbitrarily chosen representative of the neutral ground on philosophical methodology. There there were two phases alluded to, of philosophical investigation: The phase where one gets clear on the question, which I called “analytic” and secondly, the phase which we call the object-level phase. These are not only insider perspectives on philosophy, but common sense, and by common sense, to figure out which framework to use is something that should be done before enquiry starts. Verbally disputing to promote one’s framework is counterproductive and not something to dispute about at all. Those, like logicians and philosophers of mathematics, for which considerations of framework are more central, this type of behaviour would have been even more destructive, because what is at stake is not treated directly. Conceptual engineering in philosophy cannot be the topic of a substantive disagreement. Now, given that my objection to this holds, then the point that was considered in support of how important it would be to negotiate frameworks would one amplify the problem of verbal disputes in philosophy. It said that these frameworks are difficult, or maybe even impossible to distinguish from our theories and so on, and that frameworks

55 For more on how the concept of a concept framework is used in philosophy, see for example Davidson (1973)

are central parts of philosophical controversies and disagreement. Without the assumption that such disagreements can be substantive, this point now supports the consequence that merely verbal disputes abound. About how things ought to be or about how things are, the types of considerations just addressed favoured a practical outlook on theoretical philosophy, if that makes any sense at all. It was rejected on the ground that these practical tasks are not something that should be disputed, just agreed upon. Plunkett agrees to some extent, saying things like

Once one realizes that one is engaged in a metalinguistic negotiation, it can often be helpful to move to having a canonical dispute about the issues in conceptual ethics that the negotiation is about. (868.)

However, in philosophy, it is not just “often helpful,” to dissolve verbal disputes, but not doing so is to maintain pointless disagreements, and when elevated to the canonical mode one will see that there is nothing to disagree about. Plunkett, I submit, does not display enough sensitivity towards the fact that philosophical discourse is different from other types of discourse where metalinguistic negotiations might be useful. Discourse in philosophy is a joint effort seeking truth, and should not be reduced to barbarian lexical warfare. Generalizing the argument. The current line of argument can be placed in a wider context and reinforced in terms of the distinction (or many related distinctions) borrowed from moral philosophy, and decision theory, and so on, between practical and epistemic reasons, discussed above.56 To display the manifest difference between the paradigm metalinguistic disputes over words like “marriage,“ “torture,” and “athlete,” on the one hand, and disputes in philosophy over such as “perception,” “freedom,” and “justification” on the other, it is enough just to consider this distinction. I think the presence of practical reasons is what explains why we sometimes metalinguistically negotiate, but that this only occurs outside of philosophy insofar as philosophers are not merely talking past each other. There are reasons deriving from the real-word consequences of using a term in a certain way—for example, “marriage” should not be used with a meaning that excludes same sex couples, because there are practical reason not to which have to do with the treatment of same-sex couples and so on. The practical question of how to live becomes

56 What, more precisely is epistemic normativity? It is the same familiar type of normativity that guides rational belief formation; we want our beliefs to be true, and there is little we can do to convince ourselves of the truth what we take to be a falsehood. In other words, just as there are true and false beliefs, there needs to be right and wrong concepts, on this solution. But here I mean “right” and “wrong” not in the sense in which gruesome and disjunctive concepts, for example, are wrong, nor in the way that discriminating or oppressive concepts are, (both of which are types of practical normativity), but in the way beliefs ought to be true, to aim for knowledge. In this, we can alternatively call it truth-directed or representational, normativity, It is reasoning about how things are rather than on what ought to be done. “Epistemic normativity” is just a word to describe the topic that disputes in philosophy are concerned with—how things really are, and not what ought to be done.

a question about which meaning should be used, and the reasons to use the word with certain meanings over others is a question about how to treat people, and so on. While, as we saw with the Wager, there might be practical reasons the apply to and govern beliefs, it is dubitable whether epistemic reasons about the truth of something in a domain can apply to a linguistic framework that describes that domain, and to the linguistic choices to make about that framework. Consider a line of reasoning like this:

How things are in some domain governs what we should believe about that domain. Beliefs about some domain are not beliefs about how language describing that domain ought to be. How things are in a domain does not govern how a language describing that domain should be like. Only how things should be in a domain can govern how language that describes that domain should be.

Now, let’s imagine what would be required for the three suggested norms above if they are governed or if they were governed by what we should believe about the domain in question, i.e. epistemic and not practical reasons:

Least effort governing our linguistic practices about a certain domain would be indicative of how things are in that domain. More precisely, if a linguistic framework is convenient, the dispute that comes down to a disagreement over this framework, is substantive because it is over how things are in that domain. For example, if the knowledge-1 framework is easier to conduct disputes in, for example, it suggests that we have knowledge because we do in that framework. Elegance and simplicity: the elegance and simplicity of a linguistic framework about some domain governs what we should believe about that domain. Same as with the previous point, if a framework (where “knowledge” means knowledge-1, for example), is more elegant (Let’s say, the definition of knowledge has fewer clauses), that supports the theory that we have knowledge.

I think neither of these methodological outlooks make any sense. See also the same idea applied to number three, (linguistic) predictive power:

Predictive power we can understand, again, as providing the right result in something like thought experiments. If knowledge-1 gives the right results in most thought experiments, for example, it aligns with the intuitive data about when we are disposed to apply it (see 2.4. for a similar line of reasoning from Jackson). Understood as indicative or at least relevant as to the truth of of whether we have knowledge, this makes even less sense.

It is hard to point out the locus of the absurdity about this way of seeing philosophical methodology, but what they have in common that linguistic facts are taken as evidence for how things are in some domain, which was dismissed as an implausible practice in section 2.3., but without conclusive argument, and only a set of warnings inspired by Barry Stroud about why one should tread carefully over this language and reality ravine. But I still suspect that the ravine is so wide there is no bridging in sight, whether it is normative or descriptive.57 My interpretations, furthermore, of this methodological outlook on things are of course very crude and might not do justice to those who have developed theories about how it can be made to work. 58 But the same sense of irrelevancy with respect to questions about reality remains even if we consider another set of reasons, now by the other metalinguistic negotiations proponent, Peter Ludlow, for why some language should be used rather than some other language, his “norms for meaning modulation:”

(1) Take undisputed cases and argue analogically for new cases (or against familiar cases). (2) Modulations should track (not cross-cut) important properties. (3) Modulations should not be too taxonomically disruptive. (4) Modulations should allow ease of empirical testing. (2013, chapter 2).

He is clear that these are approximations, but whatever they are, I cannot see how any of them can come close guiding the modulation of philosophical frameworks. It is far from obvious how even to begin to assess their relevance to perscriptive philosophical methodology, for the reason that they are provided, it seems, only to deal with non- philosophical litigations. As I see it, these criteria are providing nothing more than the cans of language in that they seem only to report working strategies as to how to get other participants on board with one’s semantic proposals. But philosophers should always go along with each other’s ways of speaking, and dispute the content only (2.4). And inasmuch as they provide oughts, these are but practical, pertaining to the usefulness of the modulation relative to sets of needs, in this case with an emphasis on the need to align with something like natural kinds.59 None that can be brought to bear on theoretical issues in philosophy (2.2.), may even amount to something like transcendental argumentation (2.3.). Philosophical questions, in stark contrast, are concerned with how things are; and not with how things should be with the domain. Humpty Dumpty’s insights (2.4.) apply not only to descriptive metalinguistic disagreements, but also to the normative ones: There

57 However, see the conclusion at the end for some speculation on the alternatives. 58 However, see Williamson’s Modal Logic as Metaphysics (2013), introduction in particular, for an example that’s not made up of straws of an inference to the best explanation methodology which applies precisely to the practical merits of linguistic frameworks with conclusions about how things are in reality. 59 See his case study on Pluto with which these norms are associated, and his natural kind defence of why it is not a planet (2012, ch.2).

are no substantive disagreements of the form “freedom should be the ability to do otherwise,” or “knowledge should be justified true belief.”

Conclusion In conclusion, when proclaiming insights about what philosophers are “really” doing, to help keep their record clean, more can be expected as to why this delicate activity should be thought to be substantive. These types of source of normativity may be able to explain certain disputes, but it does nothing to justify them, I submit; because they are purely practical. There might be various practical reasons for why one concept would be better than the other, but that won’t suffice. Quite to the contrary, it would reduce an alleged representational question of philosophical truth to a mere practical resolution about what to do. Not much of a rescue to the problem of verbal disputes. The central point is that not just any normativity will do. Philosophers can’t simply be discussing which concept is the best one, for example in the sense of usefulness, or some other practical end, and so in virtue of that ought to be used. That might render substantial even the Martini question. Expediency won’t do. Insofar as metalinguistic negotiations are substantive, what the disputants must be disagreeing over is not which concepts are good, but rather, in an epistemic sense, is which one is the right concept.60 But that is a strange idea that I have no idea what to make of—call it the is of reality as the ought of language. With this possibility left unattended, it is still reasonable for philosophers to come together under Rudolf Carnap’s motto that “In logic there are no morals.”

3.5. Broader context: Chalmers’ theory compared to Plunkett’s theory

Starting with an even wider context, the positivist critique of metaphysics (or “ontology”) in the first part of the twentieth century was based on a similar view about language as that of Chalmers. Rudolf Carnap describes in his intellectual autobiography how he in his early years in the profession was accused of inconsistency on the ground that he switched between ways of speaking depending on conversational partners, and adopted to their languages. Assuming, let’s say, a platonist language was perceived as inconsistent with later shifting to a nominalist language, they would accuse him of switching between theories about what exists in the world—properties. But, against that, languages are tools

60 Michael Dummett’s Intuitionism, as extended from mathematics to philosophy, I believe has relevance to this question. To Dummett (The Logical Basis of Metaphysics, 1991, 246-251), using concepts may involve genuine beliefs, and hence epistemic reasons, given that the concept’s circumstances of application (analogous to introduction rules in logic) are inharmonious with its consequences of application (analogous to elimination rules in logic). If concepts are construed to be like this, i.e. that the consequences from applying the concept are different from the conditions of applying it, then a dispute over such a concept, or a dispute owing to disagreements there over, would not be verbal. Pejoratives are excellent examples, because they have “unacceptable content,” as put by Robert Brandom in his Making it Explicit (1998). See critique of this “substantial conception of meaning” in Boghossian (2003).

for saying things that are true or false; they cannot themselves be true or false, or so says his Principle of Tolerance (2.4.). Though not sufficient, it was necessary for his deflationary view on ontology. Carnap saw ontological disputes as over which language to adopt. But by the Tolerance principle, languages are not things that can be assessed for truth and falsity. As a remedy to these confusions, the method of framework explication was proposed. Carnap had faith that if one could specify the vocabulary, grammar and a set of analytic truths for a language, i.e. the set of sentences such that that by holding them true, the expressions in the whole framework gets their meaning. If this was done, he had faith, in a formal and rigorous manner, ontological disagreements would vanish. If the disputants could only agree on the linguistic framework, what would be left to disagree on would be of substance—and it would not be ontology. (“Empiricism, Semantics and Ontology,” 1959) Chalmers is more moderate; there can be substantive disputes even supposing complete framework harmony, over bedrock concepts. So while Carnap thought that framework explication was sufficient for dissolving ontological disputes, in Chalmers they might turn on bedrock disputes, which have not received treatment in the current thesis. However, what is similar between the interwar period philosopher and the contemporary Australian is that some disputes are over which framework to use, and so merely lingusitic, and unable to be true or false. It does not matter to the first order disagreement over such as freedom which linguistic mode the disagreement is presented. Chalmers therefore shares Carnap’s view that language can be separated from content, and so agreed upon by all parties. They both seem to agree with the central argument coming from myself also that that frameworks can only be assessed with respect to their practical features (pragmatism), while epistemic reasons are reserved to the content of the theory, which may be framed within several frameworks:

We also have a Carnapian pragmatism about conceptual frameworks. On this view, instead of focusing on existing words and the concepts they express, we should focus on the role one needs them to play. Carnap’s (1934) principle of tolerance stresses that everyone is at liberty to build their own form of language. In Carnap’s later work (1950a, 1950b), he rejects traditional conceptual analysis for a project of explication that may be revisionary and that gives the fruitfulness of a concept a central role, and he holds that the choice between different conceptual frameworks depends on our purposes rather than on matters of fact. (566.)61

To Carnap this framework pragmatism goes all the way down, and any substantive content can be expressed in any framework, because the substance is given by empirical

61 Continued: “My pragmatism does not run as deep as Carnap’s: Carnap would almost certainly have rejected the claim that there is a privileged basis of bedrock concepts, holding instead that one can equally start from a phenomenal basis or a physical basis (among others), depending on one’s purposes. Still, on his project and on this one, the philosophical interest of most concepts lies in the work that they can do for us.” (566.)

observable consequences. In Chalmers’ theory, in contrast, there are nonempirical philosophical truths that can only be captured by a framework involving bedrock concepts, but as long as it does, there are no non-pragmatic (non-practical) reasons for who one framework is preferred to the other. If metalinguistic disagreements are disagreements over which framework to use, they can’t be relevant to the literally expressed first order issues of philosophy, and negotiating philosophers would, in both systems, be making unwarranted, irrelevant claims. 62 The complex theory from Chalmers that we have analysed consists of many connected key concepts and key theses that are in conflict with many of the ideas that Plunkett represents. But first, the two frameworks do have several things in common. One is that they both provoke investigation into what the real content of philosophical disagreements comes to; which conflicts with the “answers are answers” attitude in 2.2 which promotes silence on these meta-issues. By Plunkett and Chalmers’ “answers are not answers” attitude, in contrast, part of legitimising a dispute is the specification of the content of disagreement, either in the form of the counterfactual notion of a subdispute, or in the form of pragmatically encoded information. What they also agree about is that the content is often metalinguistic, or metalinguistically mediated. Now, metalinguistic negotiations fail the method of elimination, meaning they are verbal. But with the added key constituent that the metalinguistic disagreement is normative, the idea coming from Plunkett is that they can be substantively so, depending on the magnitude of the conceptual ethics disagreement. This excludes all those metalinguistic negotiations where the normative disagreement is peripheral. Unless the disputants were exercising a strange game of using instead of mentioning the word to begin with, the metalinguistic disagreement can in most cases be set to the side as a side topic. The main difference is that Plunkett’s framework is more apologetic towards actual practice by building these metalinguistic contents into the pragmatics of what philosophers are actually arguing about, while in Chalmers system, a dispute is deemed verbal unless it addresses these issues directly; in the formal rather than the material mode. While Chalmers agrees that metalinguistic (and psychological and sociological) issues can be substantive, they are irrelevant to the first order issues of philosophy and so cannot lend them much substantivity—approximately the argument I have been making here. The reason metalinguistic disagreements are pointless is that it is only in non-ideal reasoning that there are substantive issues about language use. The method of elimination and the subscript gambit entails/presupposes pluralism about which terms can be used in

62 That choice of language is independent of choice of theory was challenged by Quine (1951) when he challenged the distinction between analytic and synthetic truth. If one can separate the language from the content of a theory, one can separate two classes of sentences in that theory: those whose acceptance constitute choice of language, and those that constitute the content.

philosophy, and pluralism can be legitimized as an ideal way of reasoning: Nothing hangs on language use; so the metalinguistic disputes hypothesis can explain bad philosophical practise only, and not justify it.

So there are ways in which language can matter for philosophers, but it does nothing to legitimise verbal disputes, so Chalmers’ system seems to be in conflict with Plunkett’s proposal. No certain linguistic mode of presentation is in principle preferred to the other, and linguistic choice does not matter for the first order issue that’s at stake; it’s a mere practical decision. Plunkett in fact agrees with this claim of Chalmers (857). But does that mean agreement that conceptual engineering is a non-idea method, a method that is only rational to be engaged with because of our shortcomings as finite reasoners, creatures subject to linguistic rhetorical and lexical effects encoded in language use? Then it would only be a descriptive theory of how philosophers are psychologically affected by language use, and not reveal any deep insights about reason in philosophy and our truth-seeking endeavours. Plunkett can retreat to a mere descriptive interpretation of philosophical practice, but then metalinguistic negotiations would not be any more vindicated, only rational with respect to rhetorical ends, and not in a way suitable for the ideals of philosophy and its claim to be concerned with substantive activities. That it does not matter which words to use puts great limits upon the idea of conceptual ethics, or engineering, in general. The humpty-dumpty type of pluralism about which terms to use is incompatible with large parts of the idea that choice of concepts is important. Chalmers’ pluralism says, like Carnap’s Principle of Tolerance at we be free to use which ever term is suitable; language does not matter, and it supports the use of stipulation and to carry out meaning-engineering business in the formal, not material mode. The subscript gambit is a device to specify concepts, by openly stipulative definitions signalled by the subscripts. Described from the other point of view, however, the principle of tolerance is not as liberal as it sounds in that it places constraints upon what can meaningfully be disputed. Only certain minor parts of philosophy can be described as practically working on concepts, i.e. those that consist in building theoretical frameworks and so on, and yet fewer can be understood as doing this implicitly through pragmatically encoded information.

100

Conclusion

It’s not for the social well-being of philosophers that these issues about philosophical communication are important. Rather, there are solipsistic motivations for why we should care about them, having to do with how we philosophise in general and the disputes that take place in our thinking. Alone at night gazing at the starry sky and philosophising about life and death—are we merely verbally disputing with ourselves? As Plunkett said, above, “that part of what makes this proposal viable has nothing at all to do with what goes on in disputes, nor even with what is going on in linguistic communication. Rather, it has to do with a view about what is going on when someone is engaged in inquiry” (862). However, as has been argued, something deeper than the fixing and the working with frameworks can be demanded if this is a theory of what is really going on at a profound philosophical level. The conclusion is that far more substantial reasons are in demand for why verbal disputes in philosophy aren’t pointless, but substantive. After reflection on the “Well, it depends on what you mean by that?” response, the line of reasoning went down the steps from, in the first chapter, “Verbal disputes arise in virtue of disagreements in meaning” to, in the second chapter, “Which disputes arise in virtue of disagreements in meaning?” and “Why are disagreements in meaning pointless?” to, in the final chapter, “Why are normative disagreements in meaning pointless?” It is still open how many of the actual disputes in philosophy suffer from these defects, but I at least hope to have given some valuable contributions to the more prescriptive question of whether these disagreements in meaning really are defects in the first place—they are. Verbal disputes in philosophy are mere verbal disputes. Neither of the very broad types of activities conceptual analysis and conceptual engineering can legitimize philosophical disputes insofar as they are substantive. In general, there only one way in which the attendance to language has relevance for philosophical questions, namely to avoid verbal disputes.63 In conceptual engineering

63 An example of this type of activity is this very thesis’ first chapter in its entirety, where ambiguities and conceptual nuances were unpacked and laid bare.

101

terms, this is to divide concepts and precisify topics, laboring towards the substantive non-linguistic disagreement, if there is one, while setting the others to the side, by doing such things as applying subscripts, eliminating contested terms, or asking "Well, that depends on what you mean by ‘φ’" To leave some end remarks about which way further philosophical speculation can take. First, is all the foregoing merely a verbal dispute with respect to “meaning?” I don’t think so, because I have used “meaning” in the broadest sense available, but not so broad that the topic is changed, and same with the other central terms like “substantive.” The central question enjoys practical consequences, because what it means to be substantive and pointless has grounding in what to do—which disputes to continue and discontinue. For what it’s worth, the all of the above survives the elimination test because it is ultimately a perscriptive undertaking, which might be done like in the above, by lining up a type of disputes that are likely to be verbal for the display of their shared sense of pointlessness. They centred on many of the things that have been considered in the above (analysis, real definition, frameworks, topics, labels with normative resonance, and so on), irrespective of whether these things qualify for the label metalinguistic disagreement or not. The identical sense of pointlessness remains whether it is a disagreement about what a property really is or what the topic is independently of what we mean, and so on, or just, broadly, differences in meaning. Meaning, or not, maybe we should stop a little more often and ask ourselves, “Is this really a significant thing to be disagreeing about?” When philosophers are engaged in disputes with one another, secondly, maybe the disagreement in meaning comes down to many of these different senses simultaneously, as was considered with the holism from 1.4. above, inspired by W.V.O. Quine, about all kinds of different beliefs playing a role in any kind of dispute. Alternatively, it can be indeterminate which one it is, so that it ultimately makes less sense to talk about a distinction between metalinguistic and theoretical disagreement. Maybe they are a bit over what terms mean, a bit over what they should mean, at the same time also about which topic should be discussed, and in addition to that about which properties we are interested in talking about, and so on. It may indeed be true that the distinction between talking about words and talking about things is truly indeterminate, the metalinguistic disagreement being inseparable from the rest of the grand background of disagreements. There are many appealing reasons to opt for a response of this sort, especially considering the difficulties encountered with the parthood relation between disputes and the relation of same topic; since they are so vague, maybe it’s indeterminate which disputes are verbal and not, meaning the concept of a verbal dispute is unoperatable and must be abandoned, for example. It being difficult to determine what counts as the same topic and what counts as relevant is for sure a good point of departure for arguments against the case that’s been made here. Can this be right? Whether this poses a problem, however, for the currently considered theses, comes down to which level of generality it is which is sought for. At the level of particular disputes, to begin with it still makes sense to ascribe disagreements over sentences to either disagreements about the subject matter and disagreements about something irrelevant, as meanings. And more importantly, the fact can’t be taken away about which types of disagreements give rise to pointless disputes, even though the operative definition of the elimination method failing to separate, and these disputes all being inseparable from the substantive ones. Even though, in a Quinean spirit, the method of elimination fails because

102

disagreements over words can’t be separated from disagreement over content, that only means the base conditions of application is taken away from the concept of a verbal dispute, but not the pointlessness involved in this sort of activity. In short, that metalinguistic disagreements or meaning being a vague concept with difficult conditions of application does nothing to our strong intuitive grasp of what is relevant for what, which topics are different, and so what substantive and what is pointless. Whether or not these irrelevant disagreements can be separated out from the large holistic dough does not matter as long as we know that they are the ingredients. (For further reading on these continued issues, see Chalmers’ “Irrevisability and Conceptual Change” (2011), “Epistemic Two-Dimensional Semantics” (2004), and “Intentions and Indeterminacy” (2014). For more on this type of indeterminacy argument, see Eklund, “The Ontological Insignificance of Inscrutability” 2007.) The controversy, zooming out one notch, which has been at the heart of the tradition analytic philosophy, or Anglophone philosophy of language at large, is indeed the question about analytic judgements. A cousin of verbal disputes, they too are not about reality, but only about language. The same sort of tension is present between, on the one hand, those who emphasise linguistic explanations of philosophical truth and justification, or disputing—justified beliefs in virtue of linguistic competence, truth in virtue of meaning, or disputes that arise out of differences in meaning. Call them linguistic turn philosophers, just to use Williamson’s (2004) expression again. On the other hand are those who emphasise realism, sketched in 2.5. What to prioritize, the explanatory potential offered by language, or the realistic outlook that philosophy is at heart concerned with how things are independently of that? The tension is due to the many problems encountered in 2.3. of mixing talk about language, either how it is or ought to be, with talk about how things are with the part of reality that is not about language. The problem of verbal disputes is a perfect instance of this greater problem. What was found in this thesis that is relevant to this larger problem is that rationality might constrain judgements about topics, and that disagreements over topics might be governed by epistemic norms (3.3.3.). In other words, judgements about what is a relevant subdispute support many of the (2.5.) intuitive criteria, for something philosophical to be about reality, or something similar to reality. Though it is not directly about worldly things, there might be substantive disagreements over which questions are relevant to the truth of which other questions—for example, from the discussion in 2.3., there was the issue of whether ascription contextualism is relevant to the question of whether humans can know anything. Chalmers’ “if rational” constraint on which disputes are parts of the other (1.4.) is not very informative, in other words, because several moves in philosophical reasoning turn crucially on what exactly the subdispute ought to be, derived from how things are in reality, namely what is a reason for what, and so normative disagreements over topics may have some claim to be substantive. But it is a vague conclusion, and maybe only marginal. Bring back the dramatic consequences of conflating language and reality discussed earlier in connection with Barry Stroud’s critique of transcendental arguments. To relate to reality is a desirable but immensely vague feature for a dispute to have, filled with difficulties. Solving some parts of the greater language/reality problem might not solve others, and so on, and does it even make sense to talk of a distinction between language and reality in the first place? Philosophical methodology is bewildering. Instead of fiddling with details, and particular instances, like in this dissertation, an alternative to trying to

103

solve the problem of semantics leaking into philosophy is to take a more drastic measure and ditch the idea that philosophy is about reality, and in its stead just accept the sweeping conclusion that “philosophy is all about language”. Whether letting go of the grammatically licenced but confusing “in reality” and “extra-linguistic reality” manners of speaking opens up for saying that philosophy is all about language, might be protested, but what if we just accept this sweeping claim, for subsequently to try to make sense of it? The first metaphor to adopt in working out the consequences is that the concept of reality can be diminished in favour of the concept of reasoning, and the second metaphor is that reasoning can be understood in terms of language. The very general idea, when restricted to the domain of philosophy, is that philosophizing is just the practice of reasoning, and even more generally the exercise of language. The crucial controversy that follows, again, surrounds the mediating relations that are treated as primitives in this and in much other literature, such as “in virtue of” and “depend.” “Does philosophical truth depend on only reasons?” it can be replied to this view, and also “Is philosophical reasoning merely a linguistic activity, which derives its criteria of soundness solely from linguistic rules?” Thus, if this is what must be accepted, it can plausibly be called linguistic idealism. It is an option which has not been explored, that philosophers can just let go of the incredibly vague, but as we have seen, decisive concept of an independent philosophical reality. More precisely, it is an option to reject the idea that the real world contains philosophical truths, from which philosophers derive their evidence, and from which philosophical theories derive their truth, independently of reasoning and language.

104

Literature

The list contains the works referred to, as well as those I have relied on for interpretations of the cited sources, including those of the authors themselves.

Anscombe, G. E. M. 1971. "Causality and determinism." In 1981 Metaphysics and the Philosophy of Mind: Collected Philosophical Papers Volume II. Oxford: Basil Blackwell,

Ayer, A. J. 1978, Language, Truth and Logic, London: Penguin Books.

------1956. The Problem of Knowledge London: Macmillan.

Ball, D. forthcoming “Could Women be Analytically Oppressed?” in Burgess, A., Plunkett, D., and Cappelen, H. (eds), Conceptual Ethics and Conceptual Engineering, Oxford University Press

Bennett, K. 2009, “Composition, Colocation and Metaontology,” in D. Chalmers, D. Manley, and R. Wasserman (eds.), Metametaphysics, Oxford: Oxford University Press, pp. 38- 76.

Boghossian, 1994. "The transparency of mental content." Philosophical perspectives 8: 33-50.

------, 1996 “Analyticity reconsidered,” Nous 30, 160-191 .

------, 2003. "Epistemic analyticity: A defense." Grazer Philosophische Studien 661: 15-35.

------, 2010. "Our grasp of the concept of truth: Reflections on Künne." Dialectica 644: 553-563.

Bourget, D. and D. J. Chalmers 2014. "What do philosophers believe?" Philosophical Studies 1703: 465-500.

Bridges, J. 2011. The possibility of philosophical understanding: reflections on the thought of Barry Stroud, Oxford University Press.

Braddon-Mitchell, D. and R. Nola 2009. "Introducing the Canberra plan." Conceptual Analysis and Philosophical Naturalism, (eds.), The MIT Press

Brandom, R. 1998 Making it Explicit: Reasoning, Representing, and Discursive Commitment, Harvard University Press

------2009. Articulating reasons, Harvard University Press.

Burgess, A. and Plunkett, D. 2013, “Conceptual Ethics I&II” Philosophy Compass 8 (12):1091-1101.

105

Cappelen, H. 2017. Fixing Language: An Essay on Conceptual Engineering, Oxford: Oxford University Press. Google Scholar.

Cappelen, H. and Dever, J. 2017 Context and Comuncation: Contemporary Introductions to Philosophy of Language, Oxford University Press.

Cappelen, H. and E. Lepore 1997. "Varieties of quotation." Mind 106423: 429-450.

------2008. Insensitive semantics: A defense of semantic minimalism and speech act pluralism, John Wiley & Sons.

Carnap, R. 1937. The Logical Syntax of Language. Smeaton, A. (trans.), London: Routledge and Kegan Paul,

------1956. “Empiricism, Semantics, and Ontology.” Meaning and Necessity. 2nd. Ed. Chicago: University of Chicago Press.

------1959. "The old and the new logic." Logical positivism: 133-146.

------1963. “W. V. Quine on Logical Truth.” In Schilpp, P. A. (ed.), The Philosophy of Rudolf Carnap. La Salle, Ill.: Open Court. 915–22.

------1968 “Inductive Logic and Inductive Intuition” in I. Lakatos, ed. The Problem of Inductive Logic: Proceedings of the International Colloquium in the Philosophy of Science, London, 257-67

Chalmers 2004. "Epistemic two-dimensional semantics." Philosophical Studies 1181: 153- 226.

------2009. "Ontological anti-realism." Metametaphysics: New essays on the foundations of ontology: 77-129.

------2011 Verbal disputes." Philosophical Review 1204: 515-566.

------2012 Constructing the World, Oxford University Press

------2015, “Why isn’t there more Progress in Philosophy?” in T. Hondrich (ed.), Philosophers of our Times.

Chalmers, D., and Jackson F,. ‘Conceptual Analysis and Reductive Explanation’. Philosophical Review 110 (2001): 315–60.

Chalmers, D., Manley, M. and Wasserman, R. 2009, Metametaphysics; new essays on the foundations of ontology, Oxford: Oxford University Press. 103

Chan, T. 2013, The Aim of Belief, Oxford: Oxford University Press.

Chisholm, R. M. 1966. Theory of Knowledge (Englewood Cliffs, NJ: Prentice Hall).

Cohnitz, D 2015. "Intuitions in philosophical semantics." Erkenntnis 803: 617-641.

Cohnitz, Daniel & Jussi Haukioja 2013 "Meta-Externalism vs. Meta-Internalism in the Study of Reference", Australasian Journal of Philosophy 91, 475-500.

106

Dahlberg, N. 2016 “Diagnosing Verbal Dispute: The Case of Ontology.” Thesis, Georgia State University, 2016 scholarworks.

Das, K. L. 2006. Philosophical Relevance of Language, Northern Book Centre.

Dasgupta, S. 2016. "Metaphysical rationalism." Noûs 502: 379-418.

David, M. 1996. ”Analyticity, Carnap, Quine, and Truth”, Philosophical Perspectives 10, 281-96

Davidson, D., 1974 “On the Very Idea of a Conceptual Scheme” Proceedings and Addresses of the American Philosophical Association, Vol. 47, 5-20

Dummett, M. 1991, The Logical Basis of Metaphysics, The William James Lectures, 1976, (ed.) Cambridge Massachusetts, Harvard University Press, 1991.

Dupré, J. 1981. "Natural kinds and biological taxa." The Philosophical Review 901: 66-90.

Eklund 2007, “The Ontological Significance of Inscrutability” Philoosophical Topics, 115-134

------2007, “The Picture of Reality as an Amorphous Lump,” in T. Sider, J. Hawthorne, and D. Zimmermann (eds.), Contemporary Debates in Metaphysics, London: WileyBlackwell.

------2007. "The ontological significance of inscrutability.” Philosophical Topics 351/2: 115-134.

------2009, “Carnap and Ontological Pluralism,” in D. Chalmers, D. Manley, and R. Wasserman (eds.), Metametaphysics: New Essays on the Foundations of Ontology, Oxford: Clarendon Press, pp. 231–259.

------2013, “Carnap’s Metaontology”, Noûs, 47: 229-49.

------2013, “Evaluative Language and Evaluative Reality”, in Simon Kirchin (ed.), Concepts, Oxfotrd University Press, Oxford 161-81

------2014, “On Quantification and Ontology”, Oxford Handbooks Online, (2014), www.oxfordhandbooks.com.

------2015, “Intuitions, Conceptual Engineering, and Conceptual Fixed Points”, in Christopher Daly (ed.), The Palgrave Handbook of Philosophical Methods, Palgrave Macmillan

------2016, “Carnap’s Legacy for the Contemporary Metaontological Debate”, Stephan Blatti and Sandra Lapointe (eds.), Ontology after Carnap.

Floridi, L. 2011. "A defence of constructionism: Philosophy as conceptual engineering." Metaphilosophy 423: 282-304.

Fodor, J & Lepore, E. 1998. Holism: A shopper’s guide. Oxford: Clarendon Press

107

Gettier, E. L. 1963 “Is Justified True Belief Knowledge?” Analysis 23: 121-3.

George, A. 2000. “On Washing the Fur Without Wetting It: Quine, Carnap, and Analyticity.” Mind 109: 1–24.

Greco, J. 2003. "`Knowledge as Credit for True Belief." In Michael DePaul & Linda Zagzebski (eds.), Intellectual Virtue: Perspectives From Ethics and Epistemology. Oxford: Oxford University Press. pp. 111-134

Greenough, P. 2017. Against Conceptual Engineering. Book Manuscript.

Hacker, P. M. 2007. "Analytic philosophy: beyond the linguistic turn and back again." The Analytic Turn: Analysis in Early Analytic Philosophy and Phenomenology: 125-141.

Hacker, P. M. S., et al. 2009. Wittgenstein and Analytic Philosophy: Essays for P. M. S. Hacker, OUP Oxford.

Hanfling, O. 2013. Philosophy and Ordinary Language: The Bent and Genius of our Tongue, Taylor & Francis.

Haslanger, Sally 2000 ‘Gender and Race: (What) Are They? (What) Do We Want Them to Be?’ Nous 34, 31–55.

Haukioja, J. 2015. "On deriving essentialism from the theory of reference." Philosophical Studies 1728: 2141-2151.

Hirsch, E. 2002a, “Quantifier Variance and Realism,” Philosophical Issues, 12, pp. 51–73. in Hirsch 2011, pp. 68-95.

------2002b, “Against Revisionary Ontology,” Philosophical Topics, 30 (1), pp. 103–127. in Hirsch 2011, pp. 96-123.

------2005, “Physical-Object Ontology, Verbal disputes, and Common Sense” Philosophy and Phenomenological research, 70, 76-97.

------2013, “The Metaphysically Best Language,” Philosophy and Phenomenological Research, 87(3), pp. 709–716.

Hume, D. 1993, An Enquiry Concerning Human Understanding, London: Hacket Publishing. Originally published in 1748.

Irving, J. A. 1956 “Review: Definition by Richard Robinson” Philosophy and Phenomenological Research, Vol. 16, 416-418

Jackson, F. 1998. From metaphysics to ethics: A defence of conceptual analysis, Oxford University Press.

Jenkins, C. 2014. "Serious Verbal Disputes: Ontology, Metaontology, and Analyticity." The journal of philosophy 1119/10: 454-469.

------2014. "Merely verbal disputes." Erkenntnis 791: 11-30.

108

Kment, B. 2012. "Varieties of modality," The Stanford Encyclopedia of Philosophy (Spring 2017 Edition), Edward N. Zalta (ed.), URL = .

Kornblith, H. 1993. "Epistemic normativity." Synthese 943: 357-376.

------2005. "Précis of knowledge and its place in nature." Philosophy and Phenomenological Research 712: 399-402.

Lewis, D. 1970. "How to define theoretical terms." The journal of philosophy 6713: 427-446.

------. 1979 ‘Scorekeeping in a Language Game’. Journal of Philosophical Logic 8, 339–59.

------1983. "New work for a theory of universals." Australasian journal of Philosophy 614: 343-377.

------1986, On the Plurality of Worlds, Oxford: Blackwell.

Ludlow, P. 2005. "Contextualism and the new linguistic turn in epistemology." Contextualism in philosophy: 11-50.

------2008. "Cheap contextualism." Philosophical Issues 181: 104-129.

------2012 Living Words: Meaning Underdetermination and the Dynamic Lexicon, Oxford University Press.

McDowell, J. 1996. Mind and world, Harvard University Press.

Millikan, R. G. 1984. Language, thought, and other biological categories: New foundations for realism, MIT press.

Neta, R. 2008. "How cheap can you get?" Philosophical Issues 181: 130-142.

Plunkett, D. 2015. “Which Concepts Should We Use? Metalinguistic negotiations and the methodology of philosophy” Inquiry : An Interdisciplinary Journal of Philosophy 58 (7-8):828-874

Plunkett, D. and Sundell, T., 2013, “Disagreement and the Semantics of Normative and Evaluative Terms.” Philosophers' Imprint 13, 23.

------2013. "Dworkin's Interpretivism and the Pragmatics of Legal Disputes." Legal Theory 193: 242-281.

------2014 ‘Antipositivist Arguments from Legal Thought and Talk: The Metalinguistic Response’. In Pragmatism, Law, and Language, ed. Graham Hubb and Douglas Lind, 56–75. New York: Routledge,

Putnam, H. 1973. "Meaning and reference." The journal of philosophy 7019: 699-711.

109

------1975 “The Meaning of ‘Meaning’,” Minnesota Studies in the Philosophy of Science 7, pp. 131–193.

------1981, Reason, Truth, and History, Cambridge: Cambridge University Press.

Quine, W. V. 1980. “Two Dogmas of Empiricism.” in From a Logical Point of View. Cambridge, Mass.: Harvard University Press. 20–46. Originally 1951. Philosophical Review 60 (1).

Ramsey, W., et al. 1990. Connectionism, eliminativism, and the future of folk psychology. Philosophy, Mind, and Cognitive Inquiry, Springer: 117-144.

Rippon, S. 2008, “Brief guide to writing the philosophy paper,” Harvard College Writing Centre. Available at: https://writingcenter.fas.harvard.edu. Accessed: 11.12.17

Robinson, R. 1950, Definition, Oxford: Clarendon Press.

Russell, B. 1914. Our knowledge of the external world as a field for scientific method in philosophy, Open Court.

Scharp, K. 2007, “Replacing Truth,” Inquiry: An Interdisciplinary Journal of Philosophy 50 (6):606-621.

------2014, “Replacing Truth?” in Alexis Burgess and Brett Sherman (eds.), Metasemantics, Oxford University Press.

------2015, “Weekly Discussion; Week 16: Conceptual Engineering,” [blog] Weekly Discussion at /r/Philosophy. Available at:: https://www.reddit.com/r/philosophy/comments/3pen67/week_16_co nceptual_engineering/ [Accessed: 26.11.2017].

Scharp, K., and Shapiro, S., (forthcoming) “Revising Inconsistent Concepts” in B. Armour-Garb (ed.) The Relevance of the Liar, Oxford University Press

Schwartz, S. P. 2012. A Brief History of Analytic Philosophy: From Russell to Rawls, Wiley.

Sider, T. 2001. Four-dimensionalism: An ontology of persistence and time, Oxford University Press on Demand.

------2009. "Ontological realism." Metametaphysics: 384-423.

------2011 Writing the Book of the World, Oxford University Press

Simon, M. 2017, “The ‘should’ in conceptual engineering,” Inquiry: An interdisciplinary Journal of Philosophy, 1-15

Smart, J. J. C. and J. J. Haldane 2008. Atheism and Theism, Wiley.

Stanley, J. 2005. Knowledge and practical interests, Clarendon Press.

110

Stjernberg, F. 2009. "Strawsons Descriptive Metaphysics-Its Scope and Limits." Organon 529-541.

Strawson, 1959, Individuals, London: Methuen.

------1966, The Bounds of Sense, London: Methuen.

------1982, “Freedom and Resentment,” in Watson (1982), ed., 59–80.

------1992, Analysis and Metaphysics, Oxford: Oxford University Press.

------2004, Ethics without Ontology, Cambridge, MA: Harvard University Press.

Strawson, G. 1986, Freedom and Belief. Oxford: Clarendon Press.

------1994. "The impossibility of moral responsibility." Philosophical Studies 751: 5-24.

Sosa, E. 2007. Experimental philosophy and philosophical intuition. Philosophical Studies 132:99-107

Stroud, B. 1968. "Transcendental arguments." The journal of philosophy: 241-256.

------1994. "Kantian Argument, Conceptual Capacities, and Invulnerability." Kant and Contemporary Epistemology: 231-251.

------2000. "Practical reasoning." in E. Ullmann-Margalit (ed.), Reasoning Practically, New York 1999 pp. 27-38. Reprinted in: B. Stroud, Philosophers Past and Present: Selected Essays, Oxford

------2000. Understanding Human Knowledge: Philosophical Essays, Oxford University Press

Tappolet, C. 2014, “The Normativity of Evaluative Concepts,” in Anne Rebout (ed.), Mind, Values, and Metaphysics. Philosophical Essays in Honor of Kevin Mulligan, Volume 2, pp. 39-54.

Tarski, A 1933, “The Concept of Truth in Formalized Languages,” in Logic, Semantics, Meta-Mathematics. Tr. J. H. Woodger. Ed. J.Corcoran. Indianapolis: Hackett Publishing Company, 1983.

Taylor, R. 1962. "Fatalism." The Philosophical Review 711: 56-66.

Van Inwagen, P. 1986. "The incompatibility of responsibility and determinism." John Martin Fischer Ithaca, NY: Cornell University Press, 1986.

Waismann, F., et al. 1983. Wittgenstein and the Vienna Circle: Conversations, B. Blackwell.

Wiggins, D. 2001. Sameness and substance (renewed), Cambridge University Press.

Williamson, T. 1994, Vagueness, London: Routledge.

------2002. Knowledge and Its Limits, Oxford University Press.

111

------2002. Vagueness, Taylor & Francis.

------2004. "Past the linguistic turn?" in: Brian Leiter (ed.) The Future for Philosophy, Oxford: Oxford University Press, 106-28.

------. 2004. "Philosphical ‘intuitions’ and scepticism about judgement." dialectica 581: 109-153.

------2008. The Philosophy of Philosophy, Wiley.

------2011. "Philosophical expertise and the burden of proof." Metaphilosophy 423: 215-229.

------2013. Modal logic as metaphysics, Oxford University Press.

112

Index

analytic philosophy, 52, 75, 102 metalinguistic analyticity, xi, 102 beliefs, 5, 11 Carnap's conception of, 97 disagreement, vi, 101 negoatiation, xii Barry Stroud, 63, 95, 102 normativity, 12 bedrock concepts/disputes, 18, 23, usage, 78 37, 42, 97 natural kinds, 53, 59 conceptual analysis, xiv, 47, 64 normativity and substantivity, 87 dispute parthood, 8, 18, 19, 20 practical and epistemic, 88 sources of, 87

ethics/moral philosophy, 91 P.F. Strawson, ix, 65, 69 Pascal’s Wager, 89 explication, 72 philosophy see Rudolf Carnap, 69 "as conceptual analysis", 46 externalism, 3, 11, 30, 59 "as conceptual engineering", xiv about philosophical concepts, 54 "as metalinguistic negotiation", 76 "as verbal disputes", 35 framework, 69, 71, 91, 92 paradigm questions of, x, 38 Frank Jackson, 65 pointlessness see substantivity, vii genius of our tongue, 49 principle of tolerance see Rudolf Carnap, 67 H2O. See water Humpty Dumpty, v, xvii, 17, 47, 64, real definition, 52 71, 72, 74, 86, 95 reasons first-order, 50 idealism, 63, 103 metalinguistic. See metalinguistic in virtue of, 6 normativity , 39, 77, individuation, ix representational level issues disagreements about, 74 78 of meanings, 3, 74 language. See metalinguistic of topics, 72 thought, 47, 100 , 51 insider/outsider, viii, xvii, 24, 37, 43 Richard Robinson linguacentric metaphilosophy, 75 Rudolf Carnap, vii, 41, 69, 72, 87, 96 linguistic turn, 75, 102 Secretariat the racehorse, 80

113

semantic deference see externalism, 3 use/mention distinction, xii Socratic question, 30, 37 about reality. See real definition verbal disputes distinctly socratic, 39 definition of, 2 reducible. See conceptual analysis diagnosing. See insider/outsider stipulation, 17, 21, 31, 64, 73, 99 merely verbal dispute, xiv, 11 subdispute, 19 paradigm example of, vi substantivity pointless. See merely verbal dispute criteria for, vii substantive. See metalinguistic definition of, 9 negotiations the problem of, vii varieties of, 1 topics changing them, 19, 52, 67, 72, 81 verbal question. See trivial question in relation to conceptual pluralism, verbally inclined question, ix, x, 51 72 in relation to meaning, 69, 85 W.V.O. Quine, 20, 101 irrelevant ones, 29, 34, 37, 48, 83, water, 19, 31, 53, 59, 66 90 -boarding, xii trivial question, ix, x

114

115