We are familiar with the idea that a complete account of semantics (the study of the meanings expressed by different vocabularies) can require attention also to pragmatics (the study of the use of vocabularies). The paradigm is indexical and demonstrative expressions. Determining the referent of tokenings of types such as ‘I’, ‘here’, ‘now’, ‘this’, and ‘that’ depends on features of the use of those very tokenings: who uttered them, where, when, and in what surrounding context. A principal contention of this work is that the class of vocabularies whose meanings cannot be specified without attending to features of use is much larger, and the sorts of dependence of meaning on use much more varied, than is suggested by focusing on this central kind of example. I argue that alethic modal vocabulary, deontic normative vocabulary, and even classical logical vocabulary are also of this general kind. They differ from indexical and demonstrative expressions in that what is essential to understanding their semantics is their relation to the use of other vocabularies (non-modal, non-normative, non-logical ones), not to the pragmatic features of their own tokenings. One of the reasons this kind of semantic relation has been overlooked is an overemphasis on semantic compositionality. In the fifth of these lectures I offer some specific technical results aimed at clarifying some of the issues in the vicinity in a way that may lead to a more balanced assessment.
Modal, normative, and logical vocabularies are all ones that have been of central, perennial interest to philosophers of language over the past century or so. But this distinctive and essential feature of their semantics—as I will put it, the constellation of pragmatically mediated semantic relations they stand in to other vocabularies—has not been studied, or even (with some notable exceptions) much noticed by the tradition. The culmination of my argument in this book is that focusing on relations of this kind makes visible crucial features of what is expressed by the use of semantic, and more broadly, intentional vocabulary. (The title of the sixth and final lecture is “Intentionality as a Pragmatically Mediated Semantic Relation.”) Further, this pragmatically mediated semantic feature of semantic and intentional vocabularies is intimately related to the pragmatically mediated semantic relations in which alethic modal and deontic normative vocabularies stand to each other.
Along the way to this result—as a necessary means of formulating, clarifying, and justifying it—I introduce, and develop through application to other examples, a general method of formally representing relations between meaning and use (in the broad way in which I will use the terms, between the topics of semantics and pragmatics). I call it “meaning-use analysis.” Analyses of this kind are codified and expressed in their most perspicuous form in meaning-use diagrams, which the reader will find peppered throughout the book. Each such diagram presents one or more of the infinite number of pragmatically mediated semantic relations among vocabularies, whose representations are recursively generated by the methods of meaning-use analysis. By way of sharpening and making this conceptual apparatus more definite, I consider along the way a number of further applications of it to a variegated budget of examples: computational linguistics, artificial intelligence, even indexical vocabulary itself. In this same spirit, I also use it to introduce a new kind of formal semantics for alethic modal and classical logical vocabularies that brings to light hitherto invisible relations between them.
The result is an approach to the philosophy of language that in my subtitle I call an “analytic pragmatism.” It is pragmatism pursued in an analytic spirit. By calling it “pragmatism” I mean a view inspired by insights of the later Wittgenstein, which situates concern with the meanings of expressions in the broader context of concern with proprieties governing their use. It counsels us to start our thinking about the meanings expressed by various vocabularies by thinking about discursive practices, about the abilities whose exercise constitutes using the vocabularies to express those meanings. Pursuing those pragmatist ideas in an analytic spirit is rejecting the anti-theoretical, anti-systematic conclusions that are often drawn from them. Instead, we can, the claim is, think about the relations between meaning and use every bit as rigorously and systematically as it has proven possible to think about the sorts of relations between meanings that are codified and explored in classical formal semantics, for instance as developed within the dominant Tarskian model-theoretic and possible worlds frameworks.
This work presents my first big philosophy of language project since the one reported in Making It Explicit.1 As I indicate in more detail in the Afterword, these are different projects. Although they have some topics in common (for instance, both are concerned with the relations between semantics and pragmatics, and with the nature of logic and its role in philosophy), this book is not a further working-out of the theory presented in that earlier one. The two are largely orthogonal enterprises. I think they are broadly compatible, but I have not worried overmuch about reconciling, or even relating, them in a compare-and-contrast sort of way (and devote none of the present volume to doing any of that).
The genesis and development of the two works is also as different as it well could be. Both are ambitious, and each is in its own way systematic. But Making It Explicit took shape very slowly. Working out the details so as to fill in each part of the initial plan, adjusting the remainder as necessary so as to balance the sometimes conflicting demands of both remaining true to the original animating vision and taking suitable account of the lessons learned along the way, took eighteen years to produce a stable product I could call finished. (I’ve been at my Hegel project even longer than that.) I was determined throughout to take as long as it took. Ars, indeed, longa, however brevis the vita. By contrast, this project came from the beginning with a deadline: a date, fixed in advance, on which I would have to stand up and present the material to a demanding audience. In the spring of 2002 I received from the Oxford Philosophy Faculty the kind invitation to deliver the John Locke Lectures there. Their initial suggestion was that I do so in 2004. I felt strongly that that was not enough time to prepare new work worthy of the occasion, and suggested the following year. It turned out that a previous, equally conscientious and foresighted invitee (Ernie Sosa), had had the same sort of resistance to a short lead time, and so was already scheduled for 2005. So we settled on 2006. (Four years is today a not atypical length of time for someone to spend writing a philosophy PhD dissertation—a work of about the same length I would be aiming for.)
My initial response to the invitation was blind panic. Although the Oxford people professed politely to be willing to listen to anything I might have to say, it was clear to me that I had not been invited to talk about Hegel—the topic then occupying the center of my attention. Further, under what they charmingly called the “McDowell rule,” Oxford University Press, who financially support the lectures, now insist that one explicitly agree in advance to let them have first consideration of any book arising out of the material presented in those lectures. (Prior practice had been to leave this expectation implicit. McDowell refused to comply, sending his deep, path-breaking book Mind and World instead to Harvard, in part because of OUP’s having let Gareth Evans’s work go out of print.) I had no objection to such a distinguished venue. But my long, long-gestating Hegel book has long been promised to my long-suffering editor at the Harvard University Press, Lindsay Waters. So the Hegel was out. I felt strongly that this invitation should be understood as a challenge to present the best philosophy of which one is capable. Rehashing, or even just extending, the views I had already presented in excruciating detail in Making It Explicit and subsequent works would be failing to rise to the occasion. I have a good deal of material on Frege, which I had been working up over a period of time, with an eye to an eventual monograph. But even my most optimistic view of what it could aspire to achieve would not put it in the same class as the contribution made by Macbeth’s fabulous book Frege’s Logic,2 whose manuscript I had recently been privileged to read. I would have been glad to present her work, had I written it, but was not going to set myself up to suffer the comparison.
Then it occurred to me that I did have a plan for a three-article set that would develop the ideas I had sketched earlier in a piece called “Modality, Normativity, and Intentionality.”3 There I had stated and explored some of the consequences of accepting what I called the “Kant-Sellars theses” about alethic modal and deontic normative vocabularies. These are the claims, roughly, that anyone who knows how to use ordinary, non-modal, non-normative vocabulary already knows how to do everything she needs to know how to do to deploy also modal and normative vocabulary. In this sense, those kinds of vocabulary have the expressive role of making explicit something that is implicit already in the use of ordinary descriptive vocabulary. But I had not offered arguments that those claims are true, nor thought very much about what kind of claim they are. I proposed to do that in the first of the three elements of the triptych. In the next, I hoped to make good on a long-term project of using the notion of material incompatibility as the semantic primitive of a new sort of formal semantics—one that would represent the content of claims (including those formulated using descriptive, modal, and logical vocabulary) in terms of the set of claims materially incompatible with them. The use of this ultimately modal semantic primitive could in turn be introduced in terms of the normative statuses of commitment and entitlement, according to the principle that to treat claims as materially incompatible is to take commitment to one to preclude entitlement to the other. I thought this idea could be developed into a unified setting that would permit the representation of the contents of alethic modal claims, about what is possible and what is necessary, at exactly the same level and in the same terms as the contents of ordinary logical claims, as well as non-logical ones. But there were substantial technical challenges to be overcome, which at the time I had as yet made little progress on, despite fiddling with them over a period of years. Finally, to complete the triad, I wanted to exploit the relations between what is expressed by deontic normative vocabulary (paradigmatically ‘committed’ and ‘entitled’) and alethic modal vocabulary (‘necessary’ and ‘possible’) that were revealed by putting together the Kant-Sellars theses with the way a semantics for modal vocabulary could be elaborated from what is expressed by normative vocabulary. In those terms I thought I could say something new and interesting about the intentional nexus between knowing and acting subjects, who are obliged to reject or resolve incompatible commitments and objects, which are individuated in part by the impossibility of their exhibiting incompatible properties. (I had already had the idea that something like these two senses of “determinate negation”—his term for “material incompatibility”—lies behind some of Hegel’s darker, but also deeper, claims about the relations between the subjective and the objective.4) It would be worthwhile, I thought, to try to motivate and elaborate this idea wholly in contemporary terms, without having to say anything about Hegel.
This was an idea for three articles or lectures, and the Locke Lectures were to be six. But I thought it likely that there was enough material there to make up six meaty lectures, if I just thought about it all hard enough. It was on that basis that the panic subsided (it took about two weeks), and I felt I could responsibly accept the invitation. In the presentation that eventually emerged, the three pieces I had originally conceived correspond pretty well to the last three of the six lectures. The surprising and (as it seems to me) suggestive story about intentionality and its relation to what is expressed by alethic modal and deontic normative vocabularies that unfolds in the last three lectures ended up prefaced by a somewhat more abstract methodological story told in the first three, which develops some of the lessons about intentionality appealed to in the final lecture. The big challenge—since the whole thing would not work without it—was getting the technical results needed to make the incompatibility semantics work. Through great good fortune (though the timing was not just a bizarre coincidence), I was able to spend the academic year 2002–2003 as a fellow at the Center for Advanced Study in the Behavioral Sciences. (This is where Quine wrote Word and Object, and Rawls wrote A Theory of Justice—but no pressure!) Though I was not aware of it in advance, it turns out that my fellowship there was financially supported by the Andrew W. Mellon Foundation—just the first of their contributions to this project. Able to devote myself full-time to philosophical work, I made good progress on my Hegel project: producing drafts of the three chapters I was then in a position actually to write. But I devoted substantial pie-in-the-sky, staring-off-into-space, let’s-try-anything-that-might-work time to the incompatibility semantics. By the end of that golden period, I had made the basic conceptual advances (along with writing some useful computer programs) that I was confident would eventually result in a successful up-and-running product. (It turned out to be a bumpier ride to that destination than I had thought it would be.)
Upon my return from Stanford, I began to think about the overall project again. It was at this point that I realized that the Kant-Sellars theses asserted a distinctive kind of pragmatically mediated semantic relation between modal and normative vocabularies, on the one hand, and ordinary descriptive vocabulary, on the other. Trying to characterize the genus of which this new sort of semantic relation was a species led to the first meaning-use diagrams, and then to the more general idea of meaning-use analysis. Looking for an index example, where the results of a non-trivial meaning-use relation were not philosophically controversial (as all my other proposed examples were bound to be), I realized that I had stumbled on a new kind of metavocabulary for discussing the relations between automata and vocabularies that are studied in the relatively clear-cut syntactic arena by computational linguistics. This provided the example I exploit in the first lecture, to introduce the more general metaconceptual apparatus.
I had for many years taught an undergraduate lecture course at Pitt (originally conceived by John Haugeland, based on his fine textbook Artificial Intelligence: The Very Idea5) that used current debates about the possibility of artificial intelligence to introduce more general themes in the philosophy of mind. So I was familiar with that literature, and had had the opportunity to think through some of the issues. I soon realized that the automaton-theoretic issues I had been considering at the level of syntax would generalize nicely to the semantic level, and that, when they did, the result would be a substantially new way of thinking about the central issues of artificial intelligence—one that shows debates about the essentially symbolic character of thought to be parochial and peripheral. This line of thought led to the formulation and (skeptical) discussion in the third lecture of a pragmatist (as opposed to the classical intellectualist) version of the thesis of artificial intelligence: the claim that the capacity to engage in some autonomous discursive practice (a language-game one could play though one played no other) could be algorithmically decomposed into abilities each of which can, in principle, be exhibited by non-discursive systems.
I taught all this new material in my graduate seminar in the winter of 2005. So that this idiosyncratic stuff would have some context, and the students would learn at least something of value even if the rest of it fell apart, I started off by talking about the history of analytic philosophy from Russell to Quine, as a foil to the pragmatist line I would be taking. In the course of telling that story, I realized that in the systematic form in which I could now tell my pragmatist story, that form need not at all be thought of as an alternative to the analytic tradition. Rather, once the classical project of philosophical analysis is described broadly enough to encompass most of its principal variants—including at least Russell and Moore, the Tractatus, the Cambridge analysts of the 1920s, the Vienna Circle, Ayer, Goodman, and Quine—what I was doing could be seen at least as much as a continuation and extension of that project, bringing substantial new expressive resources to bear on its behalf. That is how I came to introduce the project in the first lecture. At that point, almost exactly a year before I was scheduled to deliver the lectures, I sat down to produce the actual text.
By happy accident, at the end of 2003 I had received a letter telling me I had been awarded a Distinguished Achievement in the Humanities prize from the Andrew W. Mellon Foundation. This astonishingly generous award provides one and a half million dollars to enable its recipient to pursue his or her research in ways not otherwise feasible. Though I have found many other (I trust) worthwhile endeavors to which to apply these funds (for instance, cataloguing and making available to the scholarly public the papers of Wilfrid Sellars), the project reported in these pages has been a principal beneficiary and product of the extraordinary opportunity they afford. I indicated earlier that I knew from the beginning that the technical adequacy of the new modal incompatibility semantics I propose was going to be a critical criterion of adequacy determining whether the story could take the shape I hoped it could. The Mellon money made it possible to assemble a research team to help work out and exploit the basic ideas. My colleague Ken Manders, my student Jukka Keranen, and my research assistants Jeremy Heis and Alp Aker, all made significant contributions to moving the work forward, and to furthering our understanding of the, in many ways, very unusual formal setting of this semantics. (For instance, as detailed in Lecture 5 and its long technical appendix, this is a holistic semantics that is nonetheless fully recursive. At the level of non-logical sentences, it corresponds to a multivalued logic in which the number of multivalues varies depending on how many atomic sentences there are in the language.) The one among these who made the greatest contribution is Alp Aker. He is responsible for almost all of the metatheoretic results reported in the appendix to Lecture 5. Without his ingenuity, insight, and hard work the basic ideas would not have been developed nearly as far nor as well as they are here. Nonetheless, it is early days, and what is offered here is offered in the spirit of a report on work in progress.
The lectures that appear here are essentially as I delivered them at Oxford in Trinity Term of 2006. I have resolutely resisted the temptation to write the long book that is struggling to get out of this short one. Brevity (in art, if not in life) has its virtues. The exception is the two appendices, which were already available when the lectures were given. They provide detail necessary to back up claims made in the lectures proper. Only the first half of the argument concerning indexical vocabulary is presented in the first lecture. The argument would be seriously incomplete without the material in the appendix to Lecture 2, which simply could not be shoehorned into the lectures themselves without pushing out something even more important to the whole project. (Material developed only in the second lecture is needed to fill in the argument about indexical vocabulary gestured to in the first.) The long technical appendix to Lecture 5 provides the definitions and proofs that articulate incompatibility semantics for modal and classical logic. They are the cash for the promissory notes offered by my descriptions of those results in the lecture. As I remark there, in this context, proof is the word made flesh. Those who are willing to take my word for it can skip the dreary details. But anyone who wants to understand how the system really works has to have them all available.
The John Locke Lectures are delivered once a week over the course of a term at Oxford, to an audience whose composition inevitably varies with the exigencies of individual schedules. I was both fortunate in and flattered by the relatively large numbers who continued to turn out. (I’m sure that the exhortations of my gracious principal host, Tim Williamson, helped secure that happy result.) This is a point about which I had been somewhat apprehensive. My friend and colleague John McDowell got great audiences for his 1991 Locke Lectures (as well he should have, given the material he presented), but when Sellars had done his in 1965, the precipitous drop-off in attendance, ending with an audience for the last three of five or six souls scattered through a cavernous auditorium, is something he was bitter about literally until his dying day. (When I visited him in the hospital just before his death, he cited the disparity between the reception of his Locke Lectures and that accorded his University of London lectures on “Empiricism and the Philosophy of Mind” just nine years earlier, as one of the great disappointments of his professional life.) The question periods, and other less formal discussions, including weekly meetings with graduate students, were often helpful to me.
The most intense and thought-provoking conversations about this material, though, took place in Prague, in April of 2007. I had been keen to make these ideas vulnerable to the criticism of a wider philosophical audience. The largesse of the Mellon Foundation and the good offices of my friends Jaroslav Peregrin of the Charles XII University and the Czech Academy of Sciences, and Pirmin Stekeler of the University of Leipzig, made it possible. The result was a week-long meeting, attended by more than two hundred philosophers from twenty-seven countries, including a host of specially invited people whose opinions I particularly value. The lecture portion of the meeting was preceded (at Jarda Peregrin’s suggestion) by a three-day “School of Inferentialism,” in which people intimately familiar with it presented papers and presided over informational sessions devoted to various aspects of my prior work, as background for the new material everyone would hear later. (Michael Williams, James Conant, Björn Ramberg, Mark Lance, Michael Kremer, and Paul Horwich were the presenters.) Each of my six lectures was followed by a paper discussing it. In order of presentation, these were: John McDowell, John MacFarlane, Pirmin Stekeler, Huw Price, Jaroslav Peregrin, and Sebastian Rödl. I was particularly touched that my Doktorvater, Richard Rorty, having known for some time that he was mortally ill, and with the end near (he died on 8 June 2007), made the extraordinary gesture, appreciated by all, to come and to participate. These lectures are dedicated to him—my generous teacher, friend, best critic, and the first reader of these lectures, as he has been of most of what I have written over the past thirty-five years.
In July 2007, I gave all six lectures once more, this time to an impressively sophisticated, varied, and engaged audience at the University of Buenos Aires. Although I have made some relatively minor emendations in response to the comments elicited at all these events, I decided in the end that the material was best served by being presented in something very close to its original form. I have confined myself to appending to the lectures an afterword, in which I respond to some of the larger-scale worries—particularly those expressed by John McDowell and Sebastian Rödl (though similar notes have been sounded in the comments of others). The Afterword offers such arguments as occur to me that the analytic-algebraic form in which I have presented my thoughts should be understood not as fetters binding the radical, reconstructive spirit of Rorty and the experimental, exploratory spirit of David Lewis, but rather as a vocabulary in which to express, apply, and develop those thoughts, consonant with both of those spirits.
Bob Brandom
August 2007