‘Yo!’ and ‘Lo!’: The Pragmatic Topography of the Space of Reasons 9780674274396

Citation preview

‘Yo!’ and ‘Lo!’

‘Yo!’ and ‘Lo!’ The Pragmatic Topography of the Space of Reasons

Rebecca Kukla Mark Lance

HARVARD UNIVERSITY PRESS Cambridge, Massachusetts London, England 2009

Copyright © 2009 by the President and Fellows of Harvard College All rights reserved Printed in the United States of America Library of Congress Cataloging-in-Publication Data Kukla, Rebecca, 1969– Yo! and Lo! : the pragmatic topography of the space of reasons / Rebecca Kukla, Mark Lance. p. cm. Includes bibliographical references and index. ISBN-13: 978-0-674-03147-0 (alk. paper) 1. Pragmatics. 2. Speech acts (Linguistics) 3. Language and languages— Philosophy. I. Lance, Mark Norris. II. Title. P99.4.P72K85 2008 410—dc22 2008011161

To Wilfrid Sellars ‘Lo, a rabbit!’ —W. V. O. Quine Yo! Word up! —Dead Prez You talkin’ to me? —Travis Bickle

Contents

1

Acknowledgments

ix

Pragmatism, Pragmatics, and Discourse: Mapping the Terrain

1

1.1 Varieties of Pragmatism 3 1.2 Two Distinctions among Normative Statuses 12 1.3 A Typology of Speech Acts 18 1.4 More about Agent-Relativity and Agent-Neutrality 1.5 Several Caveats 29 1.6 Entitlement and Epistemic Responsibility 34 1.7 Where We Go from Here 38

2

Observatives and the Pragmatics of Perception 2.1 Observatives 45 2.2 Observatives and Occasion Sentences 51 2.3 Observing-That and the Declarative Fallacy 2.4 The Ineliminability of the First-Person Voice

3

23

42

53 59

The Pragmatic Structure of Objectivity 3.1 Observatives, Observation, and Answerability to the World 3.2 Intersubjectivity 78 3.3 Objectivity 81

4

5

66 66

Anticlimactic Interlude: Why Performatives Are Not That Important to Us

87

Prescriptives and the Metaphysics of Ought-Claims

95

5.1 The Pragmatics of Prescriptives

97

viii

Contents

5.2 Four Ways of Telling Someone What to Do 5.3 Two Alternative Accounts 113 5.4 Reasons, Claims, and Addresses 122 5.5 Coda: Categorical Imperatives 128

6

Vocatives, Acknowledgments, and the Pragmatics of Recognition 6.1 Two Kinds of Recognitives 6.2 Vocatives 138 6.3 Acknowledgments 145

7

105

134

137

The Essential Second Person

153

7.1 Concrete Habitation of the Space of Reasons 155 7.2 Second-Person Speech 160 7.3 Tellings, Holdings, and Transcendental Vocatives 163 7.4 Speech as Communication and as Calling 171

8

Sharing a World

179

8.1 Interpellation and Induction into Normative Space 180 8.2 Membership in the Discursive Community 190 8.3 How Many Discursive Communities Are There? 195 8.4 Sharing a World and Learning to See 205 8.5 On the Equiprimordiality and Entanglement of ‘Yo!’ and ‘Lo!’ 210 8.6 Fugue 212

Appendix: Toward a Formal Pragmatics of Normative Statuses (with Greg Restall)

217

Index

235

Acknowledgments

This book is the direct result of almost exactly five years of intensive joint philosophical work. Prior to that, each of us had thought hard about certain themes in this book for many years. Our shared discovery of the possibilities for synthesizing the ideas that we had been pursuing separately—occasioned by a graduate seminar at Georgetown University—dramatically transformed each of our thinking and created something wholly new. It would be hard to overstate the intellectual excitement of those early conversations during which this book was born. Finding someone who not only understands what you are up to, but whose work immediately opens up new possibilities for the formulation and development of your own, and with whom you can explore, challenge, deepen, and make that work more precise, all in a context that is intellectually smooth, is a rare and treasured moment in a life. Since those initial meetings, the work on this book has been utterly collaborative. We talked through each major idea and argumentative move in advance of any writing. Though initial drafts of chapters were often undertaken by one of us, subsequent drafts always went to the other, and later drafts were written and rewritten line by line as we sat together in front of the monitor. There is no chance that any part of this book could have existed in anything like its current form without that collaboration. Not only could neither of us have found our way down this road alone, but we are certain that neither of us could have done so with any other companion. But of course if we had talked only to each other along the way we would have descended into madness. We have been supported and joined by a magnificent intellectual community. Two people deserve special mention for their essential, engaged, and generous help: Richard Manning organized and participated in a day-long “jam session” on the book at Georgetown University when it was very much a work in progix

x

Acknowledgments

ress, and the conversations we had that day altered and enriched the book. At least as important, he gave us detailed, line-by-line comments on an early draft, and as always proved himself both a penetrating reader and a maddeningly reliable bullshit detector. We overhauled much of Chapters 1 and 2, in particular, in response to his comments. Margaret Little has been a constant sounding-board for ideas, testing our intuitions, challenging underlying assumptions, directing us to relevant literature in moral philosophy, pushing us to formulate points more clearly, and suggesting everything from clarifying examples to more perspicuous formulations of views. Indeed, much of Maggie’s own work on deontic pluralism and intimacy has tendrils that have penetrated our thought. It is hard to imagine more supportive and stimulating colleagues and friends than Maggie and Richard. Sincere thanks go to our coauthor on the Appendix, Greg Restall, who was kind enough to arrange a grant for Mark to visit Melbourne for two months. During that time Mark and Greg worked out the basics of the formal Appendix and discussed in detail how a formal perspective could illuminate and refine the philosophical meat of the book. The three of us developed later versions of the Appendix together, and we fully expect the tripartite collaboration to continue. Many people have helped us with their suggestions, objections, skepticism, and sympathy. An undoubtedly partial list includes William Blattner, Taylor Carman, Alan Gibbard, Mitch Green, John Haugeland, Elisa Hurley, Paul Hurley, André Kukla, Coleen Macnamara, Chauncey Maher, James Mattingly, John McDowell, Niklas Möller, Mark Okrent, Terry Pinkard, Alex Pruss, Joseph Rouse, Charles Taylor, Michael Williams, and audiences at Queens University, Georgetown University, the University of Virginia, the International Association for Phenomenological Studies, the Johns Hopkins University, the University of Cape Town, the University of North Carolina at Chapel Hill, the University of Melbourne, and the University of Cincinnati. Special thanks go to Colleen Fulton, the world’s greatest R.A., who gave us invaluable comments on the entire manuscript, and to Philip Kremer and Juliet Floyd, who prepared wonderful critical responses to our work for the workshop that Richard organized at Georgetown. We have been exceptionally well supported by various institutions. It is only because the Georgetown University philosophy department, through the efforts of its superlative chair, Wayne Davis, welcomed

Acknowledgments

xi

Rebecca as a visitor for three years that the opportunity for this collaboration came to be. Carleton University awarded Rebecca a Marston LaFrance Research Award, which gave her an entire paid year of release from teaching to finish this book. Several trips between Ottawa, Washington, and Tampa for the purpose of writing together were funded by Rebecca’s grant from the Social Sciences and Humanities Research Council of Canada. Mark was able to work with Greg Restall in Melbourne thanks to a grant from the University of Melbourne visiting scholars program. Camille Smith did a superb job of editing the entire manuscript, and Phoebe Kosman and Lindsay Waters of Harvard University Press helped us throughout the editing and publishing process. Finally, as is standard but no less genuine for that, we thank our wise spouses, Richard Manning (same person, different guise) and Amy Hubbard. They put up with long trips, extra parenting duties, late nights, early mornings, grouchiness when the issues were particularly recalcitrant, and excessive giddiness when the solutions came quickly. They rolled their eyes only internally when we lapsed periodically into a cryptic dialect comprehensible only to the two of us. Our children, Eli Kukla-Manning and Emma Lance, inspired and forbore. Throughout, much slack was taken up.

‘Yo!’ and ‘Lo!’

1

Pragmatism, Pragmatics, and Discourse: Mapping the Terrain “In the beginning was the Word!” Here I’m stuck already! Who helps me go further? The spirit helps me! All at once I see the answer. And confidently write: “In the beginning was the Act!” —Goethe

In this book we examine how speech acts alter and are enabled by the normative structure of our concretely incarnated social world. In other words, we examine language through the lens of pragmatism, in the metaphysical sense that takes the phenomenon of language to be, in the first instance, a concrete, embodied social practice whose purpose is meaningful communication. We argue that, by beginning with our analysis of the normative functioning of speech acts, we can clarify the structure (and sometimes make progress toward a solution) of some key issues in metaphysics and epistemology, including the role of perception in grounding empirical knowledge, how we manage to engage in intersubjective inquiry with objective import, the nature of moral reasons, and the capacity of subjects to be responsive and responsible to norms. Using an image that would grip the imaginations of at least three generations of philosophers, Wilfrid Sellars placed us—that is, us beings capable of language, thought, intention, meaning, and normative accountability—within a ‘space of reasons’, set over and against a space of mere causes. For some close followers of Sellars, most emblematically Robert Brandom, this space is first and foremost a space of inferential relations between declarative propositions.1 John McDowell’s post1. Robert Brandom, Making It Explicit: Reasoning, Representing, and Discursive Commitment (Cambridge, Mass.: Harvard University Press, 1994).

1

2

‘Yo!’ and ‘Lo!’

Sellarsian space of reasons is a provocatively richer and perhaps more ambiguous one. The relationship between the deliverances of experience and the space of reasons is complex rather than merely oppositional for McDowell.2 But none of the authors who have developed and philosophically mined the metaphor of the space of reasons have taken particularly seriously what its overall pragmatic structure may be, nor have they given detailed attention to how different normative pragmatic relations might importantly inflect and constitute this space. We aim to rectify this absence through an exploration of what we call the “pragmatic topography” of the space of reasons. We develop a framework for thinking about the normative pragmatic structure of discursive speech acts, guided by the presumption that the pragmatic structure of the space of reasons can be no less rich than that of discourse. Like any beginning, this beginning in the concrete normative structure of discourse embodies two commitments: that the starting point exists, and that it is a good place to start. Existentially, we are committed to the claim that language has systematic normative effects and functions, that these essentially depend on the concrete ways in which speakers are enmeshed in social communities and environments, and that discursive performances systematically transform the normative statuses of speakers and of those spoken to. Prescriptively, we are committed to the principle that this dimension of language and discursive practice forms an explanatorily useful entering point for thinking about larger questions concerning our contact with the empirical world, with normative force, and with one another. We can afford to be quite liberal about what counts as a speech act; for our purposes, a speech act is a communicative act that functions normatively within a structured system of communication. We don’t much care about nailing down the boundaries of the notion, but it is clear that we can count many gestures, written signs, facial expressions, and more as speech acts. Such acts may or may not have a determinate syntactic or semantic structure, but it is an integral consequence of our account that they have a rich and determinate pragmatic, communicative structure— one that is of the right sort to let them participate in a discursive system that lends itself to semantic and syntactic analysis, and of the right sort 2. See in particular John McDowell, Mind and World (Cambridge, Mass.: Harvard University Press, 1994), and McDowell, “Having the World in View: Kant, Sellars, and Intentionality,” Journal of Philosophy 65 (1998): 431–450. Sellars’s own view of the relation between the spaces is hard to pin down.

Pragmatism, Pragmatics, and Discourse

3

to use for making claims about, and upon, the world and one another. A crucial upshot of our analysis will be that meaningful speech acts are fundamentally indexed to particular agents with particular stances, substantial relationships to other particular agents, and locations within concrete social normative space that are ineliminably first- and secondpersonally owned by this or that living, embodied subject who has a particular point of view and is capable of making and being bound by claims. Our central conceptual tool, introduced later in this chapter, is a typology of speech acts—or, more precisely, of normative dimensions of speech acts—that is orthogonal to the usual systems of pragmatic categorization (by performative force, etc.). We believe this typology has surprisingly large philosophical payoffs. There is nothing uniquely privileged or architectonic about our typology; there are plenty of legitimate ways of dividing up speech acts along pragmatic lines, and surely different ways have different benefits and clarify different philosophical issues. What we claim on behalf of our typology is, first, that the mere fact that it is substantially different from the categorization systems used by linguists and other philosophers of language serves to de-naturalize the more traditional systems, and to broaden our philosophical imagination and vision, and, second, that its use can make some seemingly impenetrable philosophical questions appear quite straightforward.

1.1 Varieties of Pragmatism There are two large camps of philosophers who fly the banner of pragmatism, plus an additional camp of those who do not necessarily identify as ‘pragmatists’ but who study the pragmatics of language. Although we think that we are true to (what ought to be) the spirit of pragmatism, and although we are centrally concerned with the pragmatics of language, we depart substantially from all three camps. First, there are philosophers who find their roots in the classic American Pragmatists such as Dewey, James, and Pierce, and often also in the early work of Heidegger and his French successors such as Pierre Bourdieu and Maurice Merleau-Ponty.3 This group has productively focused on embodied practice as the ineliminable site of human meaning, and has worked to shift 3. Typical recent examples include Shannon Sullivan, Living Across and Through Skins (Bloomington: Indiana University Press, 2001), and Samuel Todes, Body and World (Cambridge, Mass.: MIT Press, 2001).

4

‘Yo!’ and ‘Lo!’

epistemological attention to local and contextual epistemic practices, and away from the quest for transcendent truths, universal principles, and absolute certainty. Second, there is what we might call “Pittsburgh School Pragmatism,” represented paradigmatically by Wilfrid Sellars, John McDowell, Richard Rorty, Robert Brandom, Donald Davidson, and John Haugeland,4 and characterized by anti-reductionism and anti-representationalism in the philosophy of mind and epistemology with roots in Wittgenstein. These philosophers are committed to the principle that the best place from which to begin thinking about intentional phenomena such as meaningful speech acts and contentful mental states is with our practical interactions with the world and with others, and their normative structure. For example, in the preface to Making It Explicit, Brandom writes: The explanatory strategy pursued here is to begin with an account of social practices, identify the particular structure they must exhibit in order to qualify as specifically linguistic practices, and then consider what different sorts of semantic contents those practices can confer on states, performances, and expressions caught up in them in suitable ways.5

Finally, there are philosophers of language such as William Alston and John Searle, who work in close collaboration with linguists and focus on speech act theory, looking backwards to Austin and Grice.6 These philosophers seek to develop a formal pragmatics that can sit alongside formal theories of semantics and of syntax. In the imperfect tripartite division of language into syntax, semantics, and pragmatics, there is rough agreement that syntax is the study of well-formedness, or grammaticality, semantics is the study of meaning, and pragmatics is the study of the way bits of language are used in the performance of speech acts. While 4. For example see Wilfrid Sellars, Empiricism and the Philosophy of Mind (Cambridge, Mass.: Harvard University Press, 1997); Richard Rorty, Philosophy and the Mirror of Nature (Princeton: Princeton University Press, 1981); Donald Davidson, Subjective, Intersubjective, Objective (Oxford: Oxford University Press, 2001); McDowell, Mind and World; and Brandom, Making It Explicit. 5. Brandom, Making It Explicit, xiii. 6. See for example William Alston, Illocutionary Acts and Sentence Meaning (Ithaca: Cornell University Press, 2000); John Searle, Speech Acts: An Essay in the Philosophy of Language (Cambridge: Cambridge University Press, 1969); and Wayne Davis, Meaning, Expression, and Thought (Cambridge: Cambridge University Press 2003).

Pragmatism, Pragmatics, and Discourse

5

many—though not all—philosophers who work on formal pragmatics share with both camps of pragmatism an explanatory privileging of the pragmatic dimensions of language, this last group departs from the first two in generally treating syntax, semantics, and pragmatics as independently analyzable, and taking mental meanings and representations as given independently and in advance of performative utterances.7 We share some methodological commitments with each of the three camps we have just described. The priority of the pragmatic in the order of explanation is important to us; we shall return to this point in detail below. We believe that meaning and normativity are phenomena that are ineliminably grounded in socially located human bodies, that reductionism and classical representationalism are bankrupt projects in philosophy of mind and epistemology, and that there is an important place for formal theories in attempts to understand the pragmatic structure of language. On the other hand, we see each of these three orientations as having serious limitations. The first camp has tended to privilege embodied practice over conceptual discourse and thought, seeing the former as more fundamental and more interesting than the latter.8 To do so is to assume that discourse and thought are not themselves embodied practices,9 and it is also, we think, to undervalue the philosophical centrality of language and discursive judgment in making possible our status as epistemic and moral subjects and our receptivity to the claims and character of the empirical world. Our points of convergence with and divergence from the second and third camps—the Pittsburgh School Pragmatists and the theorists of formal pragmatics—deserve some detailed discussion right up front. Sellars, Brandom, Davidson, and other anti-representationalists are methodologically committed to a particular explanatory starting point 7. For instance, Kent Bach writes: “Different types of speech acts (statements, requests, apologies, etc.) may be distinguished by the type of propositional attitude (belief, desire, regret, etc.) being expressed by the speaker . . . Many philosophers would at least concede that mental content is a more fundamental notion than linguistic meaning, and perhaps even that semantics reduces to propositional attitude psychology” (online.sfsu.edu/ kbach/grice.htm, accessed 10/10/07). 8. Classic examples include Hubert Dreyfus, What Computers Still Can’t Do: A Critique of Artificial Reason (Cambridge, Mass.: MIT Press, 1992), and Todes, Body and World. 9. Joseph Rouse, in How Scientific Practices Matter: Explaining Philosophical Naturalism (Chicago: University of Chicago Press, 2002), does an excellent job of systematically defending a picture of discourse as continuous with, rather than derivative upon, embodied practice.

6

‘Yo!’ and ‘Lo!’

in philosophy of mind and language, namely an account of the role that discursively formed encounters with the world and with one another play in constituting the normative statuses of participants in a discursive community—an account of the acts that form what Brandom (attributing the thought, if not the phrase, to Sellars) calls the “game of giving and asking for reasons.” It is attention to the social practices of discourse, according to this approach, that is our best way into thinking about how language manages to be suitably responsive to the world, and hence how this responsiveness is codified in a semantics and a syntax. Furthermore, Sellars tenaciously argued—following Hegel and Wittgenstein—that intentional mental states are best understood as derivative and dependent upon meaningful discursive practice, and his philosophical descendents have championed this commitment. So on this picture, philosophical explanation moves from discursive use, to content and grammar, to mind. We share a commitment to this order of explanation. In this book we will not argue separately for the rectitude of this order, but we hope to demonstrate its fecundity. We think that only by beginning with discursive practices can we understand, on the one hand, how discourse comes to be responsive to the world and capable of expressing and communicating content, and on the other hand, how any practices manage to be practices of reason-giving, truth-telling, and responsibility-imputing, rather than just elaborate conventional dances. In this sense, we are certainly continuing a project with its lineage in the work of (in particular) Brandom, Sellars, and Hegel. However, authors like Brandom think not only that pragmatics is explanatorily prior to semantics and syntax, but also that the latter are reducible to the former, that meaning just is a pragmatic feature of a speech act, properly understood. The major project of Brandom’s Making It Explicit is to spell out how semantics and syntax can be derived fully from pragmatics. In contrast, we remain steadfastly agnostic on issues of semantic-pragmatic reduction. It is consistent with all we say that semantics retain significant forms of autonomy from pragmatics. Tempting as it will surely be to some readers, we ask that our use of key Sellarsian and Brandomian terms and ideas such as the ‘space of reasons’ and ‘commitments and entitlements’ not be read as our implicit acceptance of this reductive move. We assume that both mental states and speech acts are meaningful

Pragmatism, Pragmatics, and Discourse

7

only insofar as they are properly situated within a body of discursive practices that is their constitutive precondition. Despite our agnosticism about the reducibility of semantics to pragmatics, our acceptance of the pragmatists’ order of explanation puts us at odds with most philosophers working on speech act theory and formal pragmatics. Indeed, for typical theorists of pragmatics, things go almost exactly the other way around. Mental states, particularly intentions, are typically taken for granted for the purposes of linguistic theory. Of course, philosophers such as Searle have accounts of mind, but their theories of mental representation cast it as independent of, and in important senses prior to, language. The job of the theorist, on this view, is to characterize the range of ways a person can then intentionally put a sentence—usually seen as having an unproblematic syntax—to use. Accordingly, such philosophers follow the linguists’ odd practice of treating categories of speech acts that mark pragmatic function, such as declaratives, imperatives, and interrogatives, as definitionally grammatical categories (namely moods), and only secondarily as pragmatic categories. Thus, such categories apply to sentences in virtue of their grammar, and one asks questions such as “what can a person do with a declarative?”10 But in keeping with our commitment to the explanatory priority of pragmatics, we define such use-indicating categories in terms of their use (which ought to seem quite a sensible approach, we think). Hence, for us, the answer to the above question is that what one can do with a declarative is—by definition—declare. This isn’t to deny that we can identify syntactic types as, for example, those that are typically or defeasibly used to produce declaratival acts. But for us, this will be a secondary notion. We always privilege pragmatic categories over grammatical categories when identifying the functional structure of a particular utterance. Thus, rather than “What can one do with declaratives?”, a question for us (though not a particularly interesting one) will be “Which syntactic forms can function as declaratives in English?” While our commitment to the pragmatist order of explanation puts 10. See H. P. Grice’s seminal paper, “Logic and Conversation,” in Donald Davidson and Gilbert Harman, eds., The Logic of Grammar (Encino, Calif.: Dickenson, 1975), 64–75. The assumption that a “pragmatics first” approach to language should follow the lines of Grice is common. See for example Peter Grundy, Doing Pragmatics (New York: Oxford University Press, 2000), a fairly standard introductory linguistics text that adopts the Gricean framework without discussion or argument.

8

‘Yo!’ and ‘Lo!’

us on the side of Sellars and Brandom over against most theorists of pragmatics, our focus on the multiplicity of discursive functions puts us at odds with Sellars and Brandom and the rest of the Pittsburgh School Pragmatists. Given the Sellarsian/Brandomian order of explanation, from pragmatics, through semantics and syntax, to mental states, one would expect members of that group to begin their philosophical accounts with comprehensive analyses of the entire terrain of discursive practice—of the pragmatic topography of discourse, in all its richness and complexity, including the whole variety of meaningful and communicative practices that make it up. That is, apart from the details of theory and argument, we would expect such pragmatists to display a particular interest in pragmatic phenomena. Yet, in fact, among members of this tradition, there is an odd disconnect between their commitment to a pragmatist order of explanation and their interest in the pragmatic texture of discourse. First, when authors such as Sellars and Brandom discuss practices, the lived, acting body planted in a concrete environment does not remain in view. These authors give pragmatic accounts of meaning and interpretation, but they are vastly more interested in language and theoretical reason than in the rest of human bodily activity, and they care little about how these two domains fit together. For Brandom, inferentially articulated discourse forms an autonomous domain of normativity, while perception and action serve as the ways in and out of this domain—that is, as language-entry and language-exit conditions. Indeed, he makes the remarkable claim that it is merely a contingent matter that discourse is bounded by perception and action, and that it could in principle exist without them.11 Although Brandom understands language as a system of shifting commitments and entitlements, he has next to nothing to say about what concrete events such as taking on a commitment or granting an entitlement actually are like. He gives us no story about how to materially identify such events, and he often writes as though different speakers’ respective commitments and entitlements may as well be abstract scores that shift around in Platonic space.12 Both schools of pragma11. Brandom, Making It Explicit, 234. 12. In Making It Explicit he gives many pedagogically inspirational stories, such as the story of the hut that one brings a sacred leaf to enter. But these are explicitly speculative stories about the causal origins of normativity, and they are not intended to give us insight into the real character of our current practices.

Pragmatism, Pragmatics, and Discourse

9

tism, ironically, at least implicitly agree that embodied and discursive practices are separate domains making only peripheral contact—which would seem to be a surprisingly unpragmatic conclusion. In this work we aim to plant discursive practices firmly within the embodied material terrain. For us, concrete action, centrally including the act of perceiving, will form the substance of language and not just a means of entering or exiting it. Second, consider Brandom’s pragmatic account of language. Far from starting with an articulated view of the whole terrain of discursive practices, he offers instead an account that focuses almost exclusively on asserting.13 On the basis of this account, he builds a semantically significant notion of inference, and then proceeds to work out the semantic content of semantically significant sub-sentential bits of syntax. He tells us just enough, that is, about the pragmatics of one type of speech act to define a notion of inference, and then he is off and running with his semantic story, giving hardly a glance and certainly no systematic attention to the rest of pragmatic space. We believe that this narrow focus on assertion is a serious error, and a particularly surprising one for a self-declared pragmatist. We think that it leads not only to missing out on philosophically important dimensions of language, but to hopelessly distorting our understanding of language as a normative phenomenon, including our understanding of assertion and how it works.14 Brandom’s narrow use of assertion as the sole 13. In the introduction to Making It Explicit, Brandom says: “The first step in the project is accordingly the elaboration of a pragmatics (a theory of the use of language) that is couched in terms of practical scorekeeping . . . The pragmatic significance of performances—eventually speech acts such as assertions—is then understood to consist in the difference those performances make to the commitments and entitlements attributed by various scorekeepers . . . The defining characteristic of discursive practice is the production and consumption of specifically propositional contents” (xiv, first emphasis added). The final sentence of this passage—not to mention the semantic inferentialism—is a clue that assertions are more than an example of a speech act Brandom will analyze. Indeed, when one turns to the first two chapters of Making It Explicit, in which the “elaboration of a pragmatics” is carried out and the bridge between it and semantics developed, one searches in vain for any discussion of any speech act other than assertions. Indeed, a search of the index under “imperatives” yields “See commands,” which takes us to historical discussions of Pufendorf, Wittgenstein, and Kant on normativity. “Interrogative” “performative,” etc., do not appear in the index at all. 14. We develop these objections to Brandom, and argue in particular that his focus on declaratives precludes an adequate account of observation—material that will connect with the discussion of Chapter 2—in Mark Lance and Rebecca Kukla, “Perception, Language, and the

10

‘Yo!’ and ‘Lo!’

building block for his entire theory of the pragmatic and semantic structure of language and the connection between language and the world recreates and instantiates a failure of vision and methodology that is nearly universal in analytic philosophy. To be specific, analytic philosophers, of any stripe, act as though the most fundamental, important, and common thing we do with language is use it to make propositionally structured declarative assertions with truth-values. Even though philosophers of language occasionally acknowledge and discuss the structure of imperatives, interrogatives, etc., they virtually always treat these as ‘special’ discursive phenomena that are marginal and derivative in comparison with declaratives. McDowell and Sellars, for instance, take it as an unshakable starting point that insofar as a state has a discursive or a conceptual structure, it has, or is directly derivative upon something that has, a declarative, propositional structure.15 Davidson shares a similar commitment, and he takes assertions—but not queries, requests, evocations, or hails—as the necessary starting point for interpretation.16 Even Austin, most famous for drawing the attention of the philosophical world to the variety of “things we do with words,” and Grice, who taught us the many “ways of words,” both take the declaratival assertion as the paradigm of ‘normal’ language and then examine various marginal and quirky uses of language by way of their departure from or permutation of this norm.17 First Person,” forthcoming in Bernhard Weiss and Jeremy Wanderer, eds., Reading Brandom: Making It Explicit (New York: Routledge, 2009). 15. See McDowell, Mind and World; Sellars, Empiricism and the Philosophy of Mind. 16. See the essays in Donald Davidson, Inquiries into Truth and Interpretation, 2nd ed. (Oxford: Clarendon Press, 2001), especially “Truth and Meaning” and “Radical Interpretation.” See Rebecca Kukla, “How to Get an Interpretivist Committed,” Protosociology 14 (2000): 180–221, for an extended argument that assertions are an insufficient basis for Davidsonian interpretation, and that Davidson needs to acknowledge a wider array of types of speech acts and varieties of performative force from the start. 17. J. L. Austin, How to Do Things with Words (Cambridge, Mass.: Harvard University Press, 1975). H. P. Grice, Studies in the Ways of Words (Cambridge, Mass.: Harvard University Press, 1967). Continental philosophers of language have shown less temptation to commit the declarative fallacy. Authors such as Heidegger, Derrida, Louis Althusser, Jacques Lacan, and Judith Butler (who is American but in conversation with Continental texts) have given extended accounts of language that begin elsewhere than with its declaratival functioning. See Heidegger, Being and Time, trans. Joan Stambaugh (Albany: SUNY Press, 1996); Jacques Derrida, Limited Inc, trans. Jeffrey Mehlman and Samuel Weber (Evanston, Ill.: Northwestern University Press, 1988); Louis Althusser, “Ideology and Ideological State Apparatuses,” in Althusser, Lenin and Philosophy (London: New Left Books, 1971); Judith Butler, Excitable Speech: A Politics of the

Pragmatism, Pragmatics, and Discourse

11

Nuel Belnap accused traditional philosophical semantics of committing the “declarative fallacy” insofar as it presumed that semantic content in general could be understood entirely in terms of declaratival content.18 We wish to adopt this term and broaden it beyond its original semantic application, to encompass any philosophy of language, including a pragmatic account, which takes the declarative as the privileged and paradigmatic speech act. R. M. Hare identified something very close to the declarative fallacy when he talked about the widespread feeling, among philosophers, that the declarative (or indicative) “is somehow above suspicion in a way that other sorts of sentence are not; and that therefore, in order to put these other sorts of sentence above suspicion, it is necessary to show that they are really indicatives.”19 In almost every part of this book we will show how the declarative fallacy has distorted understanding and clouded philosophical vision. Indeed, as we see it, much of the potential explanatory benefit of the so-called linguistic turn in philosophy—which was purportedly the turn to approaching classic problems in metaphysics and epistemology by way of an analysis of language20—has been thwarted by a pervasive assumption that the structure of declarative assertions is the privileged or sole dimension of language to which we should attend in order to illuminate key questions in metaphysics and epistemology. Or as Brandom baldly puts the commitment, “Asserting is the fundamental speech act.”21 Of course it is contentious of Belnap, and of us, to speak of a fallacy here. It isn’t that Brandom and others fail to notice the existence of imperatives, interrogatives, etc., but that they feel confident that these will Performative (New York: Routledge, 1997); Jacques Lacan, “The Function and Field of Speech and Language in Psychoanalysis” and “The Instance of the Letter in the Unconscious, or Reason Since Freud,” in Lacan, Ecrits, trans. Bruce Fink (New York: Norton, 2004). Heidegger (for instance in Being and Time, Division 1, chaps. 5 and 6) and Derrida (in Limited Inc in particular) launch explicit and rigorous attacks on the shortsightedness of and the philosophical damage done by the declarative fallacy. Derrida’s subversive reading of Searle and Austin in Limited Inc is especially amusing and perceptive. 18. Nuel Belnap, “Declaratives Are Not Enough,” Philosophical Studies 59 (1990): 1–30. 19. R. M. Hare, The Language of Morals (Oxford: Oxford University Press, 1968), 8. J. O. Urmson also pointed out the overemphasis on fact-stating language in analytic philosophy of language in his Philosophical Analysis: Its Development between the Wars (Oxford: Clarendon Press, 1956). 20. See, for instance, Richard Rorty, ed., The Linguistic Turn: Essays in Philosophical Method (Chicago: University of Chicago Press, 1992). 21. Making It Explicit, 173.

12

‘Yo!’ and ‘Lo!’

fall into (their secondary) place once the account of declaratives is completed. But it is just this confidence that we, like Belnap, find philosophically reckless. Not only do we agree with Belnap that there are important semantic phenomena that cannot be accounted for in terms of declaratival content, but we argue in what follows that there are deep metaphysical and epistemological issues that are left mysterious on a pragmatist account precisely because of the initial neglect of the full range of pragmatic possibilities. Thus, to recap, in the case of philosophical pragmatics, the very categorization of speech acts—paradigmatically in terms of performative force—is motivated by philosophical starting points that we do not share, and functions to preclude from the outset the sorts of explanations of broader philosophical issues that we purport to provide. In the case of Pittsburgh School Pragmatism, we find the apparent disconnect between social pragmatist philosophy of mind and actual practical discursive phenomena troubling. We embrace pragmatist methodology not because of a bevy of concrete arguments against competitors, but because of the elegance and power of the explanations one can muster for a range of phenomena once one has, as a backdrop, an appropriately spelled out account of the pragmatic structure of discursive performances. In this, our work will “feel” far more akin to that of Brandom than to that of Searle. But for all that, we believe that Brandomian social pragmatism has been led seriously astray by a failure to begin by mapping the whole pragmatic topography of discourse.

1.2 Two Distinctions among Normative Statuses In this book we loosely follow Brandom in understanding speech acts as performances constitutive of changes in normative status among various members of a discursive community. Thus, for instance, to assert that P involves undertaking a commitment to P, taking up the role of one at whom challenges of P may be directed, etc. To order someone to see to it that P, by contrast, involves undertaking to incur upon her a prima facie obligation to see to it that P. Further, the performance of any speech act is the sort of thing one can be entitled to, or not. And so on. All speech acts, we claim, strive to bring about certain normative changes: for example, assertions strive to impart beliefs and grant inference licenses, orders strive to impute responsibilities for action, and

Pragmatism, Pragmatics, and Discourse

13

so forth. As already mentioned, we are not committing ourselves, here, to any thesis about the systematic relationship between semantics and pragmatics. Rather, what we want to adopt from this strand of thought is the insight that speech acts can be productively analyzed in terms of the normative statuses that enable them and the normative changes they effect through their performative structure. Our primary conceptual tool in this work is a categorization of speech acts insofar as they have a particular kind of functional design qua linguistic performance within a discursive community. When we identify types or dimensions of speech acts, we will be doing so by way of differences and similarities between such normative functions. For this reason, we stipulate that our names for different types of speech acts— declaratives, imperatives, interrogatives, and several other “ives” that we will introduce along the way—demarcate pragmatic functional categories. When we speak of imperatives, for instance, we are directly speaking of a pragmatic category of speech acts that strive to serve a particular normative function within a discursive practice. It is usually, though not always, the case that sentences in the imperatival mood are used to issue imperatives, and vice versa. Typical sentences marked by linguists as imperatives—e.g., “Mark, revise this example!”—are such that, in typical circumstances, their production amounts to the performance of an imperative. But there is no one-to-one correspondence between normative pragmatic structures of acts and grammatical forms of English. The utterance “It’s cold in here” can function (at least) as an imperative or a declarative, depending on context. By stipulation we insist that what makes a speech act an imperative is its discursive function, rather than its syntactic structure. The function of a speech act should not be confused with either the intention of the speaker in uttering it or the standard use of that string of words in the community. Of course, it would be absurd to think that there could be a whole system of discourse that had a normative structure completely divorced from either speakers’ intentions or conventional uses; there has to be at least a defeasible concordance between function, intention, and standard use, and it is patterns of intention and use that serve to institute the contentful pragmatic structure of a language in the first place. But, just as there are philosophers who wish to reduce semantics to intentions or to conventional use and other philosophers who resist them, we recognize the possibility of philosophical at-

14

‘Yo!’ and ‘Lo!’

tempts to reduce pragmatic function to intention or convention, and we reject them. Along the way, we will have occasion to give examples of speech acts that manifest a disconnect between actual pragmatic structure and intended or conventional effect. For now, we just want to make clear that we reject any analytic identities between function, intention, and convention, even while acknowledging that there is a constitutive, defeasible connection between these things.22 Now once we understand speech acts in functional terms, it makes sense to think of them, like any functions, as having inputs and outputs. (Indeed, given the long tradition of thinking of discursive judgments as functions that traces back to Kant, it is remarkable that no one has previously tried to carefully articulate the inputs and outputs of such functions.) If speech acts function to bring about changes in normative status, then they take normative statuses as inputs and produce them as outputs. Specifically, we can distinguish between the norms governing the proper production of a speech act, which give rise to statuses that entitle its performance, and the changes in normative status that their proper production strives to make. For instance, on the input end, assertions are properly performed if they are, or can be, doxastically justified. Orders, on the other hand, are properly performed only if the speaker occupies the relevant sort of authoritative social position with respect to the person(s) to whom the order is issued. On the output end, the production of an entitled assertion is inferentially fecund; it entitles its speaker and others to draw conclusions from the claim asserted. In the case of an order, its proper production has normative effects such as a prima facie responsibility, on the part of the one ordered, to carry out the order. So, for example, consider the case of an imperatival speech act, 22. In many places throughout this book we mention, and leave to one side, ideas that rely on a notion of defeasibility. We say that various defeasible connections must exist, though the corresponding universal connection does not. While we quite consciously leave the deep issues regarding how to understand defeasibility for another time, we do not doubt that aspects of our account depend upon how one understands this important notion. See Mark Norris Lance and Margaret O. Little, “Defeasibility and the Normative Significance of Context,” Erkenntnis 61 (2004): 435–455; Lance and Little, “Defending Moral Particularism,” in James Dreier, ed., Debates in Moral Theory (London: Blackwell, 2005), 305–321; Lance and Little, “From Particularism to Defeasibility,” in Mark Lance, Matjaz Portc, and Vojko Strahovnik, eds., Challenging Moral Particularism (New York: Routledge, 2008); and Lance and Little, “Where the Laws Are,” in Russ Shafer-Landau, ed., Oxford Studies in Metaethics, vol. 2 (Oxford: Oxford University Press, 2007).

Pragmatism, Pragmatics, and Discourse

15

whereby Rebecca orders, “Mark, revise this example!” The normative input here is Rebecca’s entitlement to issue this imperative, something she has in virtue of facts such as that she and Mark are writing a book together, have agreed to collaborate in certain ways, etc. The primary output, or the primary normative status resulting from the input, is a normative burden upon Mark to revise the example (or to defend his refusal to do so). It is essential to the imperative being the sort of speech act that it is that this kind of output follows from this kind of input—that Mark’s obligation be consequent upon Rebecca’s entitlement. Now consider a declaratival speech act. Sarah says, “Bakhtin is the most important literary theorist of the twentieth century.” The input here is Sarah’s entitlement to utter this declarative, which she has—if she does have it—by way of her warrant for being committed to the content. What her declaration aims to do is to entitle beliefs, inferences, and reassertions for both Sarah and others, and this is its output.23 Finally, consider an Austinian example: the preacher pronounces a couple married. Here the input is the preacher’s entitlement to marry people, in virtue of her particular status in the community, the circumstances of the event, etc., whereas the output is the normative status provided by the marriage itself. Throughout this book we make heavy use of this distinction between input and output—that is, between the normative statuses constitutive of entitlement to a given speech act and the normative changes (in the status of the speaker, or of others in the discursive community) that the act strives to produce. Notice that what a speech act strives to accomplish, as part of its normative function, is not the same as what it does accomplish. Sarah’s assertion that Bakhtin is the most important literary theorist of the twentieth century may strive to impart beliefs or pass on inference licenses to others, but it may fail to do so if Sarah is not heard, believed, or understood. The department chair’s order that everyone in the department sign up for service on a university committee may fail if the department members do not acknowledge his authority to so order. To impute such a notion of ‘striving’ to speech acts is not to attribute any kind of spooky 23. This is its output qua declarative. Sarah may also seek, with this speech act, to annoy one of her poststructuralist colleagues, to help establish her status in the field, to baffle her mother, etc. The existence of multiple layers of normativity and function governing discourse does not undermine our ability to distinguish between and isolate these layers.

16

‘Yo!’ and ‘Lo!’

agency to them, nor to plump for any particular metaphysics of striving, but merely to play off of their having a functional structure in the first place: anything—a machine, a policy, an action, a vital organ—that has a function may be said to strive, by design, to fulfill this function. Likewise it might fail to perfectly fulfill what it strives to fulfill, and thereby count as defective to that extent. (Whether such strivings can be ‘naturalized’, or fully explained in non-teleological language, is a question that does not interest us here.)24 Crucially, then, the output of a speech act is the normative statuses the speech act strives, as part of its function, to bring about—not what it actually manages to bring about. Meanwhile, the input is what would entitle the performance of a speech act, if it were entitled, which of course it may not be. Hence inputs and outputs are themselves normatively defined. Our first key distinction among normative statuses—that between inputs and outputs—falls fairly automatically out of our casting of speech acts in functional terms. Our second distinction regarding discursive performances draws on a conceptual distinction and bit of terminology that we borrow from moral philosophy. In that context, it is common to distinguish between “agent-relative” and “agent-neutral” reasons—that is, between reasons whose force is indexed to particular agents with particular positions in normative space, and reasons that are not targeted at anyone in particular in this way. But although this language has up until now had philosophical life primarily within ethics, it does not seem to us that there is anything about this abstract distinction that should make it specific to the moral domain. So, for example, if there is a clear enough distinction between agent-neutral and agent-relative moral entitlements and obligations, we might well wonder whether there is a distinction between agent-neutral and agent-relative discursive or epistemic entitlements and obligations. We believe that there is such a distinction. Once we identify speech acts as functions on normative statuses, we can describe either the input or the output of these functions as agent-relative or agent-neutral. In general, when considering a normative status such as an entitle24. But for those who are interested, see Richard Manning, “Biological Function, Selection, and Reduction,” British Journal for the Philosophy of Science, 48 (1997): 69–82, for a compelling argument that they cannot be.

Pragmatism, Pragmatics, and Discourse

17

ment or a commitment, we can ask whether it is in virtue of its pragmatic structure (as opposed to in virtue of its content, for instance) indexed to specific people inhabiting specific normative positions, or whether it is ‘for everyone’, that is, structurally blind to distinctions among agents. This distinction will need a lot of clarification, but in rough terms, the distinction between agent-relative and agent-neutral normative statuses is one between those that by their very structure are ‘personalized’, and those that are ‘structurally public’—that is, inherently available to whoever is perceptive or lucky or interested enough to be in a position to occupy them. A pretty reliable ordinary-language test of agent-neutrality is whether the normative status can be ascribed to a generalized “we,” as in “we know that P,” or “we ought to treat one another with respect”—or as Heidegger would say, to Das Man. Agent-neutral commitments and entitlements need not be universally held; indeed it is almost never the case that everyone is in a position to take up every commitment and entitlement that she or he could or should take up in ideal circumstances. An agent-neutral reason will not grip everyone. But we can say that agent-neutral normative statuses are universal as a regulative ideal. It is only through ignorance or other defect—albeit, perhaps, a completely routine and exculpable defect—that anyone fails to have an agent-neutral normative status, since there is nothing about this status that is specific to anyone in particular. So far, this is highly abstract. To concretize and clarify, consider the difference between the imperative, “Drop and give me ten push-ups!” and the declarative, “Paris is the capital of France.” It is in virtue of one’s position within a structure of authority (as a teacher or foreman or colonel, say) that she is entitled to issue the imperative. In its normative pragmatic structure, the legitimacy it has is personal; it is the colonel who is entitled to give the order to the lieutenant. And while there may be lots of colonels, the entitlement is still inherently colonel-entitlement. Nothing about this entitlement even suggests a similar entitlement on the part of a private to issue the order to a lieutenant. Hence this entitlement is agent-relative. The declarative, on the other hand, has an agentneutral input—in virtue of the objective purport of the sentence, it is a speech act that finds grounding in the world in a way that is not specific to who is asserting it. In rough and ready terms, the input is agentneutral in the sense that what entitles it is “true for everyone”; it is im-

18

‘Yo!’ and ‘Lo!’

personally available, even if (though) not everyone will be in a position to take advantage of this availability, because of ignorance, conflicting false beliefs, inferential ineptitude, and the like. Our imperative also has an agent-relative output: it changes the normative status of those people to whom the order is directed. The imperative targets a specific, personalized audience. In reality, it may be that not everyone at whom the order is directed obeys it, or that some passerby ‘obeys’ the order inappropriately for fun, or because she misunderstood it as aimed at her. But regardless of how smoothly the concrete normative uptake of the order goes, it is part of the functional design of the speech act that it target specific people upon whom it makes a normative claim. There is always someone (or several someones) at whom the speaker is directing her order; the order has no ordering force whatsoever when it comes to those who are not targeted by it, even if they happen to overhear it, and—interestingly—even if they agree that the order was perfectly legitimate. In contrast, our declarative has an agent-neutral output: the assertion “Paris is the capital of France” seeks to impute the entitlement to assert this claim to the discursive community in general, and demands that others allow its content to constrain their inferences and beliefs. Regardless of who concretely hears it, believes it, or takes it up (which will typically be less than everyone), the pragmatic normative purport of the utterance does not in any way personalize its effects or demands. We recognize that this last claim is complex, and we will be expanding upon and defending it at length below.

1.3 A Typology of Speech Acts At this point, we have drawn two distinctions among normative statuses—between inputs and outputs, and between agent-relative and agent-neutral statuses—and thus we have the resources to categorize speech acts according to a two-by-two grid, as shown in Figure 1. Thus the colonel’s imperative, “Drop and give me ten push-ups!” belongs in box 4 of the grid, as it has both an agent-relative input and an agent-relative output. And the declarative, “Paris is the capital of France,” belongs in box 1, as it has both an agent-neutral input and an agent-neutral output. Indeed, we argue that imperatives always belong

Pragmatism, Pragmatics, and Discourse

19

in box 4, and that declaratives, properly understood and restricted, always belong in box 1 (keeping in mind that for us these mark out pragmatic rather than grammatical categories). The case of imperatives seems uncontentious: they are speech acts that are entitled by specific facts about a speaker’s normative position and relationship to the target of the imperative, and they serve to make a demand upon the specific person or persons at whom they are targeted. Hence they always belong in box 4. On the other hand, not all box-4 speech acts are imperatives. When we make a claim based in our particular normative position, upon another person insofar as she occupies a

Input Output

Agent-neutral

Agent-relative

Figure 1

Agent-neutral

Agent-relative

1 Neutral input Neutral output

2 Relative input Neutral output

3 Neutral input Relative output

4 Relative input Relative output

20

‘Yo!’ and ‘Lo!’

particular normative position, we perform a box-4 speech act. One way to do this is to issue an order, using an imperative. But we can also implore, apologize, promise, invite, and reproach, all of which are box-4 speech acts. Insofar as a declarative makes a claim about a public, democratically accessible truth, seeking thereby to make an entitlement to reassertion and inference generally available for public use, it will belong in box 1. Most of the everyday speech acts with a declarative surface grammar— “Cats like to sleep on mats,” “There are no analytic a posteriori truths,” etc.—function as box-1 declaratives in this sense. However, we will argue that not all truth-claims belong in box 1, and also (again) that surface grammar can never be a perfect indicator or guarantor of pragmatic structure (including location on the grid). We will reserve the term ‘declarative’ for sentences that have this thoroughly agent-neutral structure. Thus it is in effect an analytic truth, within our system, that declaratives belong in box 1—albeit an analytic truth that has its genesis in a substantive insight about the functioning of a broad class of speech acts. At the same time, it is not an analytic truth that all box-1 speech acts are declaratives. In Chapters 5 and 8, we tentatively suggest an alternative type of box-1 speech act. In general, each box will be inhabited by a variety of types of speech act. Although the examples we have given so far might suggest as much, we cannot in general assume that agent-relative outputs are requirements to act whereas agent-neutral outputs express general truths. The output of a speech act, for our purposes, is neither an act nor a truth, but a set of normative statuses. What makes the output of a speech act agent-neutral is that it applies, de jure, to everyone, in a way that is not indexed to particular features of anyone’s normative position. The right to assert a truth is just one such change, with no special pragmatic privilege at this level of analysis. So, consider a marriage ceremony: part of the very structure of the act conferring the status of marriage is that what is conferred is a status that demands universal recognition—now we must treat you as married, not (primarily) in the sense of asserting that it is the case, but in the sense of acting as if the status obtains (for purposes of taxes, dinner invitations, deathbed privileges, and the like). It is in fact insulting to take a marriage ceremony as having only agent-relative rather than agent-neutral significance. In normal, well-

Pragmatism, Pragmatics, and Discourse

21

functioning cases, then, the speech acts that constitute a marriage have some agent-neutral outputs (in addition to the blatantly agent-relative outputs that they have for the new spouses).25 On the other hand, the normative status of entitlement to a truthclaim is always agent-neutral.26 We will come back to this important fact at length below. For now, note that the shared and public character of truths—the fact that they are democratically available and hold in a way that is essentially not conditioned by our personal normative relationship to them—is part of what we mean by calling them truths. Missing this point is what is wrong with the undergraduate’s chant that something is “true for you, but not for me.” While we all are in different positions of epistemic access to the truth, a truth-claim, by its very structure, is not a claim for me or for you but for all of us. As Lynn Hankinson Nelson puts it, “I can only know what we know.”27 We do not mean to suggest that discursive performances will always exhibit only one normative transitional structure. Indeed, were we to be maximally precise, we would continually insist that our grid provides a system for categorizing normative functions that speech acts instantiate—always recognizing that any actual utterance will perform multiple functions—rather than a system for categorizing utterances. In later chapters we argue that speech acts necessarily incarnate multiple functions that belong in different boxes. Although for pedagogical reasons we tend to focus, in the beginning, on speech acts insofar as they centrally exemplify one or another of the normative patterns, this is a deliberate oversimplification. Not only do speech acts incarnate multiple functions, but further, once one has a range of acts instituted within a 25. Given its insufficient clarity in drawing this distinction, together with its tendency to run together issues of the normative source of entitlement with features of the nature of that entitlement, the discussion of these matters in chapter 3 of The Grammar of Meaning by Mark Norris Lance and John O’Leary-Hawthorne (New York: Cambridge University Press, 1997) should be seen merely as a suggestive precursor of the typology presented here. Not only is that account radically incomplete, at least one of its authors now considers it confused in important respects. 26. Some sentences that make truth-claims can only coherently be uttered by specific people, such as the sentence “I am the father of Emma.” But “I am the father of Emma,” spoken by Mark, declares exactly the same truth-claim as “Mark is the father of Emma,” spoken by anyone else. Hence the entitlement to the truth-claim is agent-neutral. See Chapter 2, section 4. 27. Lynn Hankinson Nelson, Who Knows? From Quine to a Feminist Empiricism (Philadelphia: Temple University Press, 1992).

22

‘Yo!’ and ‘Lo!’

discursive community, one will be able to combine them in complex ways. A declarative may also be an imperative (“It’s still cold in here!”). When we call it a declarative, we are focusing on a particular normative function that it serves—specifically one that takes an agent-neutral input and yields an agent-neutral output, by stating a public truth. When we call it an imperative, we are focusing on its function as an order to do something about the temperature, targeted at a specific person from someone in a specific position of authority. We leave this section, then, having provisionally offered examples of types, or dimensions, of speech acts that fit into two of our four boxes, as shown in Figure 2.

Input Output

Agent-neutral

Agent-neutral

Agent-relative

1 Neutral input Neutral output

2 Relative input Neutral output

Declaratives

3 Neutral input Relative output Agent-relative

Figure 2

4 Relative input Relative output Imperatives (promises, invitations, reproaches...)

Pragmatism, Pragmatics, and Discourse

23

1.4 More about Agent-Relativity and Agent-Neutrality The notions of agent-neutrality and agent-relativity we employ here are complex and utterly essential to our project, so it is worth spending time clarifying and exploring them. The distinction between agent-relativity and agent-neutrality is not simply one of extensional scope. An agent-relative output, for example, could apply de facto to everyone. This might be because of its semantic structure (“Everyone raise your right hand”) or because of empirical facts that determine its extension (“Those of you under nine feet tall, wear this badge”). The output of these orders is agent-relative, even though they in fact target everyone. The universality of their target is in neither case a function of their pragmatic structure. By the same token, we have not defined agent-neutrality in such a way that every being on the planet, or every Kantian agent, must be contained in the scope of an agent-neutral status. We have left open the size of ‘the’ discursive community, as well as its boundaries and its ontology, although we take up these issues in detail in Chapter 8. We have said nothing so far that determines whether ‘the’ discursive community, which provides the ‘we’ across which discursive functions may range, is singular or whether there might be different discursive communities in different contexts, bounded by lines of nation, language, expertise, or whatever else. If discursive communities are multiple, then someone outside the scope of the community to which a speech act is referenced is not part its functional universe of discourse. We can draw a distinction between two species of agent-relative normative statuses. On the one hand, there are “kind-relative” statuses. These are statuses that apply to people in virtue of their membership in some general kind. On the input end, these are kind-relative entitlements: all colonels have certain entitlements to issue orders to privates; Martha’s entitlement to officiate at marriage ceremonies issues from her status as a justice of the peace in a particular jurisdiction; and so forth. On the output end, these are kind-relative claims: a community might prohibit all felons from voting, or call upon all civilians to evacuate an area in the path of a hurricane. In each of these cases, the normative entitlement or claim attaches to the kind or property. If a law is passed that prohibits convicted felons from voting, this is an act that has, as an output consequence, an imposition of a prohibition on members of a kind.

24

‘Yo!’ and ‘Lo!’

By contrast, some normative statuses are “agent-specific.” Such statuses apply to a particular agent (or some particular agents) in light of concrete, particularized facts about her normative position, and have no implications even in the ideal for others. When Martha pronounces Bob and Jerry married, her entitlement to do so is agent-relative and kindrelative, but Bob and Jerry’s new status as married is agent-relative and agent-specific: the ceremony alters the normative statuses of Bob and Jerry, qua concrete individuals, rather than qua fungible instances of a larger category. If Mark promises Rebecca that he will revise their paper, this generates an entitlement on the part of Rebecca to expect him to revise the paper and a commitment on the part of Mark to do so. More generally, promising is an act that creates agent-specific obligations on the part of the promiser. There is no impetus in the pragmatic structure of promising that this normative status be inheritable by others in virtue of their sharing properties or kind memberships with the promiser. Agent-specific normative statuses need not be singular; they can be held by several concrete, particular people. (In the next chapter, we will see examples of statuses that must be singular.) I may make a promise to more than one person with a single speech act, for instance, in which case each of them will have a special claim on me.28 But since the difference between agent-specific and kind-relative statuses does not concern the number of people with the status, we must have some other means for telling the two types of status apart. Notice that kind-relative statuses have counterfactual import. If one passes a law that forbids felons to vote, it applies to all felons, and if additional felons were to exist, it would apply to them as well. Kind-relative statuses do not distinguish between actual and possible instances of the kind. Agent-specific statuses are different. One cannot make a promise to merely possible agents. One can use a kind to designate the range of people to which a promise applies—“Attention citizens of Gotham: I promise to rid your city of masked super-villains!”—but this is a promise to the actual people of Gotham. This restriction to actual rather than counterfactual marks a normative transaction as agent-specific rather than kind-relative. Agent-neutrality raises its own specter of especially tricky issues and 28. Though this book is a collaborative work, and though we undertake joint commitment to all the claims made throughout, the nature of much of what we discuss requires the use of the first-person singular. We trust that our shifts in this regard will not lead to confusion.

Pragmatism, Pragmatics, and Discourse

25

possible misunderstandings. To begin with, when we say that the input of a declarative is agent-neutral, the point is that there is nothing about the entitlement to a declarative speech act that structurally indexes that entitlement to any particular agent or agents with specific normative positions. Rather, what entitles a declarative speech act is the character of our shared, public world.29 Now in fact, it will rarely, if ever, be the case that everyone has an entitlement to perform a declarative. Contingencies of expertise, location, access to testimony, etc., will determine who can actually take up an agent-neutral entitlement and properly utter a declarative. If Jones justifiably declares, “Mitosis is a form of reproduction,” then Smith might still not be in a position to declare this—either in virtue of having committed himself to the Stork Theory of Reproduction, or because of simple ignorance of this fact. Indeed, surely there are properly performed declaratives that express knowledge that only a few people, or maybe even only one person, have the epistemic skills to discern. A particularly gifted physician, for instance, might be able to diagnose a rare disorder on the basis of an examination, and might be genuinely warranted in declaring that a patient suffers from this disorder. This might be so even though no one else would be entitled to make this declaration, even with access to the very same facts and sensory inputs. And yet, we want to claim, the input entitlement in these cases remains agent-neutral, because it purports to express entitlement to facts that are public and in no way agent-relative, even if the epistemic conditions that allow these facts to be used to warrant claims are themselves agent-relative.30 That is, the agent-neutrality or agent-relativity of an input is based on the nature of the entitlement itself, wherever it came from.31 29. Does this mean that we cannot declare anything about ‘private’ entities such as mental states? We accept a basically Wittgensteinian line here. We think that we do, in fact, have access to one another’s mental states. We can see and know that other people have various emotions, beliefs, etc. To the extent that there is some truly private element to our mental life—if such an idea is coherent—that would be just the kind of thing that we could not talk about in language. As we progress through this book, it will become clearer why such Wittgensteinian sympathies are required and justified by our project. 30. Most people—though not all!—are particularly gifted at discerning their own mental states. But this does not mean that what is true about a person’s mental states for that person is not true about that person for someone else. 31. Thus, our distinction does not align with Dummett’s distinction between the criteria and consequences of application, discussed, among many other places, in his Frege: Philosophy of Language, 2nd ed. (Cambridge, Mass.: Harvard University Press, 1993).

26

‘Yo!’ and ‘Lo!’

Furthermore, along with the agent-neutrality of the inputs, in such cases, goes the universality of these inputs as a regulative ideal. That is to say, to the extent that someone does not have access to an agent-neutral entitlement, he is defective, albeit perhaps in an extremely common and blameless way. If everyone had all the epistemic skills and evidence available to them that anyone had (as well as the same antecedent commitments, biases, level of effort, etc.), then everyone would be entitled to utter the same declaratives as everyone else (subject to some interesting qualifications and precisifications we discuss below). There is nothing about the entitlement that indexes it to any particular kind of agent. This is not true of imperatives, which have agent-relative inputs. It is in no sense a defect in a concert pianist, for instance, that she is not also a colonel, and hence not entitled to issue certain sorts of imperatives to privates that colonels are entitled to issue. Analogous points can be made about declaratives on the output end. When one puts forward a claim as a declarative—as a claim about the way things objectively are—one professes that the claim is true, not simply “true for me,” or some such. In Brandomian lingo, Jones’s entitled declaration, “Mitosis is a form of reproduction,” issues reassertion and inference licenses that are not indexed to any specific agent or kind of agents. The output of her declaration is agent-neutral, in the sense that nothing about Jones’s speech act, insofar as it is serving a declarative function, targets any particular kind of agent. Brandom would say that it issues such licenses universally, but this language is quite misleading. Because of ignorance of Jones’s claim, ignorance of enough about Jones to make trust in her word rational, incompatible beliefs, or any of a number of other reasons, many people— most people—are not suddenly entitled to reassert Jones’s claim and use it in inference just in virtue of her having made it. Thus the issuing of a “universal” reassertion license cannot mean that everyone will be in a position to make use of it. Rather, the actual agent-neutrality of the output goes along with its universality as a regulative ideal. It is, as it were, a claim for everyone, which strives to contribute to the bank of public knowledge shared by the discursive community. There is a practical point to uttering a declarative, qua declarative, as long as there is anyone left who has not yet taken up the agent-neutral entitlement it offers. Again, the contrast with imperatives is instructive. When the colonel issues an order to the private, her order is specifically targeted at that per-

Pragmatism, Pragmatics, and Discourse

27

son. Others are not bound by it, not because her order was defective, or because they are defective in the uptake of the act, but because they were not its structural targets, even in the ideal. Notice that in ordinary language we speak both of what ‘is known’, or what ‘we know’, as well as of what Jones knows or what I know. That is, when keeping track of what we know, we seem to keep two sets of books, as it were: those governing particular people, and those agentneutral facts about what is known. There is an important sense in which once Daniel Mazia discovered that mitosis is a form of reproduction, or once our skillful doctor discovered that Mr. Brown had rare disease x, it became true that we know these facts, even though not everyone in the community knows them. As a textbook might put it, “We have known since 1951 that mitosis is a form of reproduction” (a statement most assuredly not true of the authors of this book). Consider a useful analogy for these two sets of books. In typical team sports, we can look at a goal either from the point of view of its effect on the score of the game, or in terms of its effect on individual players’ statistics. In the latter sense, we can intelligibly say that the midfielder scored a goal in the eighty-ninth minute. In that sense, the accomplishment was agent-relative; it was her goal and not anyone else’s. But in the former sense, the team scored the goal. In terms of the primary scoring regime of soccer, it is quite incoherent to attribute a goal to any particular player. Indeed, the sense in which the goal is the midfielder’s is the sense in which we give the midfielder some sort of special credit for bringing it about that the team scored a goal. The midfielder (with or without help) accomplished or brought about the scoring of a goal-forReal-Madrid. The analogous point applies to declarative speech acts. When a logician proves that R has 3,088 Ackerman constants, she personally brings it about that we know this, that it is known, and the accomplishment has an agent-neutral status. In both cases, an individual makes a normative achievement for all of us, as a representative, as it were, of the whole. Agent-neutral inputs are ideally universal, in the sense that were all people to live up to all the normative ideals—including all the epistemic norms—that apply to them, they would all have access to all the agentneutral input entitlements. They would be able to know and do everything whose entitlement is not structurally agent-relative. Whereas it is a defect in an agent that he fail to be entitled to an agent-neutral entitle-

28

‘Yo!’ and ‘Lo!’

ment, it is a defect in the functioning of a speech act that it fail to impute its agent-neutral entitlement universally. Normally, both are unremarkable, thoroughly exculpable defects. A declarative speech act may fall short of being a successful performance because of problems with the speaker (not everyone trusts her, or she wasn’t in fact entitled to say what she said), the audience (some people were not bright enough to get the point, or have a false belief that is incompatible with her claim), or, most commonly, the performance itself (not everyone heard it, as is virtually always the case). All or almost all declaratives will be at least somewhat defective. There is nothing spooky about this virtually universal defectiveness: no one ever shoots a perfect game of golf either. There is a trivial sense in which simply not knowing something that is true is a defect. One is less than omniscient any time one fails to know something true. And yet, we want to claim that the proper performance of a declarative, at least the first time it is uttered, turns failure to be entitled to that declarative into a defect in a different and stronger sense. Otherwise, there would be no reason for us to claim that such an entitlement is part of the agent-neutral output of a declarative speech act, given that not everyone will be in a position to take up the entitlement. In uttering a justified declarative, a speaker offers a truth-claim up for public consumption, or adds it to the public bank of knowledge—her claim is now part of what we know, in the agent-neutral sense we described. An individual’s failure to know what her discursive community knows puts her in a position of discursive deficiency—susceptibility to legitimate correction by others—that is concretely different from a mere failure of omniscience. While the mere fact that P can never constitute someone’s grounds to correct your belief that not-P, our agent-neutral entitlement to P always does.32 Before ‘we’ discovered that there were planets orbiting stars other 32. Often, the one who achieves this agent-neutral status will also achieve personal justification, but this is not essential. Imagine a scientist asking her diligent but relatively uneducated research assistant to run a test in the lab and to report back to her, telling her ‘A’ if the test comes out one way and ‘B’ if it comes out another way. After properly running the test, the RA receives the result that she knows is to be reported as ‘A’, and hence she declares ‘A’ to her supervisor. Now in fact, ‘A’ might be a scientific result that is proven by the test results. Hence we now know that A. But the research assistant might not understand that ‘A’ describes a truth proven by the test. (For all she knows, the test just provides incremental evidence that A). Hence she would not be justified, personally, in believing what she declares at the time that she declares it.

Pragmatism, Pragmatics, and Discourse

29

than our sun, for instance, we were all in some sense defective for not knowing this fact; however, no one had a proper entitlement to demand that others believe it. But once ‘we’ discover these planets, anyone who claims that there are no such planets can properly be corrected. Thus the proper performance of a declarative has implications for the normative status of everyone in a discursive community—for example by turning some epistemic statuses into new sorts of social defects—even though it is unlikely to fulfill its ideal discursive function of passing on a universal reassertion and inference license. The achievement of an agent-neutral entitlement always precludes entitlement by anyone else to any claim incompatible with the claim in question: once we discover planets orbiting other stars, it can never be the case that anyone can be properly entitled to the belief that there are no such planets (although, given incomplete knowledge, someone may still have good reasons for such a belief). To summarize: in the case of a declarative, the entitlement that follows from its performance is the agent-neutral entitlement “our knowing that P.” An immediate normative upshot of this idea is that the achievement of the entitlement constitutes anyone’s failure to know as a socially significant sort of defect—ignorance or unjustified incompatible belief. On the other hand, no such agent-neutrality is built into an imperative, such as the colonel’s imperative “Raise your hand when your name is called.” Nothing in the structure of the colonel’s entitlement to issue this order suggests that everyone ought to be able to issue this order, or that everyone ought to respond to the order, even in the ideal. It is no defect, no matter how exculpable, not to follow the colonel’s order, if you are not the one to whom the order was issued.

1.5 Several Caveats Speech acts are embedded and embroiled within the elaborate normative structure of human practices, and as such, it is often tricky to focus our attention on a single dimension of their discursive normative structure. Here are some warnings concerning how our attention can be led astray. The function of a speech act insofar as it transforms discursive normative statuses needs to be distinguished from the various other normative functions and effects it might have. Just as a football player can perform actions that uphold or violate the rules of football reasonably independently from whether those same actions uphold or violate

30

‘Yo!’ and ‘Lo!’

norms of etiquette, morality, aesthetics, or good grammar, the norms governing communication are multilayered and not all of them specifically concern discursive pragmatics. An order may be designed not only to impute an obligation but also to insult, praise, humiliate, remind, collude, display power, etc., and it has such normative effects within a whole network of norms that are not particularly discursive— norms of etiquette, institutional structure, morality, and so on. Someone who interrupts a mathematics colloquium to snidely interject a counterexample is violating norms, but not necessarily mathematical norms. Someone who makes a dumb point in a meeting is violating norms of rationality, but not necessarily Robert’s Rules of Order. The norms of keeping secrets, being polite, sticking to parliamentary meeting or debating rules, etc., are all norms governing discourse that are independent from the layer of narrowly discursive normative functioning that we are isolating here. An order may fulfill its pragmatic discursive function perfectly while violating a myriad of other kinds of norms (including, perhaps, other narrowly linguistic norms: “You give that book to her and I right now!”). But we can still isolate (or do a reasonably good job of isolating) its pragmatic discursive function and its place in a network of discursive norms.33 It is perhaps impossible—and certainly beyond our capacity—to offer a clean definition demarcating which parts of the normative structure of a speech act properly belong to its discursive functional structure. However, we hope that the notion has intuitive appeal and will become clearer as we progress. Not only do we need to separate the norms governing the pragmatic discursive function of a speech act from the other norms that govern it, but we must also differentiate between the accidental, circumstantial effects of a speech act and those that are essential to its proper functioning. In the domain of semantics, at least some mild form of meaning ho33. This is not to say that such discursive norms and other social norms governing language won’t have a complex, mutually constitutive, intertwined relationship to one another. For instance, that a speech act occurs in the context of a meeting governed by Robert’s Rules might well have everything to do with the proper reading of its pragmatic structure and functioning as a speech act. Speech acts performed by the chair, for instance, are likely to have a different performative structure and force from behaviorally similar acts by the other meeting attendees, or from those by the person who comes in during the meeting to refill the coffee urn. That the functioning of a speech act within Robert’s Rules helps constitute its functioning within discursive communication proper does not mean that there is no distinction between these.

Pragmatism, Pragmatics, and Discourse

31

lism is accepted by the majority of philosophers; most would admit that the contours of the meaning of a sentence or concept are in some sense shaped by or subject to an elaborate and extensive network of semantic relationships. At the same time, only the most harebrained holistic extremist believes that every one of the meanings in this extended network is relevantly part of the meaning of any one term.34 We acknowledge that our beliefs about evolution, pet shop ethics, carrots, and good-luck charms are, in some sense, relevant to the contours of the meaning of ‘rabbit’, but at the same time we are (most of us) comfortable saying that nothing about pet shop ethics or good-luck charms is integrally part of what we mean when we talk about rabbits. In these postWittgensteinian days, it would be foolish to think that we could draw a hard and fast line of this sort or specify in advance a litmus test for something’s being integral to a meaning, but we negotiate the distinction mostly without difficulty all the same. In just the same way, we need to hold on to an intuitive, reasonably robust, yet not fully specifiable distinction between the normative effects that are integral to a given pragmatic structure and those that are its external normative fallout, as it were. When Rebecca orders Mark to revise his example, many changes in normative status occur: Maggie, who overheard the comment, now believes that Mark will change his example; Karen, who also overheard, now is entitled to order both Rebecca and Mark to close the door to keep the noise down; Mark, who is tired of Rebecca criticizing his examples, now feels slighted and irritated; Rebecca is now entitled to believe that she will find the example in its final form more compelling than it is in its current form; and so forth. But the change in normative status that Mark undergoes—namely, his new prima facie responsibility to revise the example—is integral to the imperatival structure of the original speech act in a way that the rest of this normative fallout is not. Similarly, any speech act with an agentneutral output—one that changes everyone’s status impersonally—will have all sorts of agent-relative fallout: Rebecca’s claim “Mark’s dog is hungry,” uttered within earshot of Mark, functions to make an impersonal truth-claim, even though, along the way, it also serves to give Mark the agent-relative commitment to feed his dog, and so forth. (But here is 34. For one non-harebrained interpretation of holism, see Lance and O’Leary-Hawthorne, The Grammar of Meaning, chapter 2, which also makes out a normative notion of the analytic/ synthetic distinction that we take to mesh rather nicely with what we say here.

32

‘Yo!’ and ‘Lo!’

an example of why surface grammar can at best be a guide to pragmatic structure; Rebecca might utter these very same words for the express and primary purpose of ordering Mark to feed his dog.) It is important to keep in mind the distinction between effects that are essential and those that are accidental to the discursive function of a speech act, lest every speech act look like just a messy mixture of agentneutral and agent-relative inputs and outputs. As this work progresses, it will become increasingly clear that most and perhaps all speech acts do have both agent-neutral and agent-relative inputs and outputs, and not only as a matter of accidental fallout, but essentially as a condition of their functioning at all. But the terrain of inputs and outputs that any speech act engages is not a jumble but a prioritized structure. We have seen that speech acts are embedded within layers of norms of various sorts, not all of which concern their essential discursive function. We have also seen that output statuses function as regulative ideals that a given speech act, qua discursive act, strives to—but may not— achieve. It follows that we cannot determine the agent-neutrality or agent-relativity of the output of a speech act by looking at whether it in fact has the same normative impact on everyone. Indeed, because of the intersection of competing layers of norms, we cannot even assume that a speech act with an agent-neutral output ideally or by intention affects everyone the same way. Consider the analogy of gift giving.35 Suppose that Mark makes a statue and presents it as a gift. Imagine three different ways in which he might offer this gift. • He makes it for his mother, Helen, as a Mother’s Day gift. It is for

her, properly having the agent-relative status of belonging to Helen, a status it would have even if she died, rejected it, destroyed it, or in any other way became unable to help realize this constitutive purpose. In this case Mark’s gift-giving performance has an agentrelative output. • He makes a public statue, as a gift to the community. In this case, the statue is for everyone, in the agent-neutral sense. It will be for everyone, in this sense, even if not everyone will see it, not everyone wants it, not everyone even knows that it exists. This giftgiving performance has an agent-neutral output. 35. With thanks to Maggie Little for suggesting the analogy.

Pragmatism, Pragmatics, and Discourse

33

• He makes a public statue, as a gift to the community. However,

since his work happens to be an aesthetically stunning representation of a naked homosexual couple, the city decides that it would not be appropriate to mount the statue on the public square in front of the local nursing home. Instead it is placed in a park in the midst of the bohemian district, where conservatives and the elderly rarely travel. In this case, the gift is still offered to the public and the act of gift giving still has an agent-neutral output. But there are social proprieties that determine that the statue will end up in a place where only some people will be in a position to take advantage of this gift. The statue belongs to the community, but because of norms external to those of gift giving, only some members of the community will make use of it—and appropriately so. Likewise, if someone utters a declarative by telling a secret, or in the context of a closed meeting, this does not detract from the agentneutrality of the output of this declarative. Other social norms will govern who is in a position to hear and use the claim. In that sense, the declarative will fall short of living up to the discursive regulative ideal of universality, because of the conflict between this ideal and other social norms. Though only some will receive the claim, nonetheless qua declarative—qua knowledge claim—it is for everyone. Like the statue, it is public property, even though only some will have practical access to this property. Though there may be good non-epistemic reasons for those not in on the secret not to know the truth, they are nonetheless ignorant, which is an epistemic defect. It is true that for a secret telling to function as such, it must not fully succeed in executing its declaratival function. But this should neither be puzzling nor lead one to doubt that it still has this declaratival function. Consider an intentional walk in baseball. Here, the pitcher is trying to pitch balls, so as to, for example, load the bases. But the idea of an intentional walk is essentially derivative on the basic idea of a pitch, the constitutive goal of which is to throw a strike. It would not be a pitch at all without participating in that goal, and this despite the fact that its derivative function requires that it partially fail in its more fundamental one. This same tension between fundamental and derivative structure is operative in such linguistic phenomena as secret telling, sarcasm, rhetorical questions, etc. There is no conceptual puzzle built into the fact that

34

‘Yo!’ and ‘Lo!’

acts of speaking are determined by multiple systems of norms, which are often not fully consistent with one another.

1.6 Entitlement and Epistemic Responsibility We have argued that any properly performed speech act with an agentneutral output transforms anyone’s failure to have access to entitlement to that output into a kind of a defect. Once it is known that P, then anyone who doesn’t know P is substantively ignorant, by the standards of the discursive community, and one who believes not-P is unjustified, and in a status of disagreeing with something known. However, as we mentioned, there is an important sense in which the mere fact that something is the case is sufficient to make failure to know it—and hence failure to be entitled to declare it—into a defect, namely the defect of non-omniscience. But this might lead us to wonder: What difference does it make, normatively speaking, if someone actually performs a declarative, from the point of view of the normative status of everyone else? If it was already a defect not to know something true, how does the fact that this truth has now been declared make any agentneutral normative difference? We argued that the declaration puts the claim into social space in a new way. But one might object that there could be nothing agent-neutral about this achievement. Uttering a claim, including a properly entitled one, isn’t going to make all the ignorant people know it, so the most such an utterance can do is create some new beliefs in some people, and this effect is a matter of degree. Some declaratives are taken up by nearly everyone, some by a special few who are in the know, and some by nobody. Presumably, the claim expresses something that was already, agent-neutrally true, and the effect of the claim seems to be to help some people know this truth, rather than to change everyone’s status agent-neutrally.36 Taking up this important challenge in detail will allow us to explore some of the contours and complexities of epistemic responsibility. On the one hand, we are not responsible for knowing everything that is true, even though we are in some sense defective in virtue of our lack of omniscience. None of us can be held accountable for knowing the number 36. We are grateful to Richard Manning for raising this worry, in his “Comments on Kukla and Lance,” presented at Georgetown University, April 2005.

Pragmatism, Pragmatics, and Discourse

35

of rocks on Neptune or what Julius Caesar had for breakfast the morning before the Ides of March. Even more strongly, there are many things that we perfectly well could find out if we wanted to, but in which no one has a stake, nor any reason why she should have one, and we’re not answerable for these things either. We are not responsible for knowing how many leaves are on the birch tree behind the old schoolhouse, although we could become responsible for knowing such a thing if it came to matter for some reason (and one can always dream up such reasons). In other words, there seems to be a special class of epistemic defects that matter to us, insofar as we are actually situated within a concrete epistemic community, in which members have concrete epistemic positions and concrete concerns. While we are not responsible for knowing everything, we are—singly and collectively—always responsible for knowing more than we actually do know. The mere fact that we haven’t bothered to notice something or find something out doesn’t mean that we are not epistemically responsible for doing so. It is indeed an epistemic failure, on my part, if I don’t notice that my son is afraid to go to school, and it is a failure on our part, as a community, that we don’t know how to dose various life-saving drugs for female patients because we have only tested them on male subjects. I am responsible for knowing commonly cited facts, noticing the brute features of my environment, and drawing straightforward inferences; if I don’t do these things, I am not living up to the epistemic norms that bind me. Though there is less that plays a role in the space of reasons than all that is the case, there is nonetheless more than all that has been recognized or justified. When we are defective in the sense of failing to live up to our epistemic responsibilities (as we all always are to some extent), we are defective in a stronger sense than that of mere failure of omniscience. We have positive duties to observe, investigate, and think—positive duties that imply that there is such a thing as culpable epistemic negligence. While you may be afraid to find out what your child is doing with her evenings, you may nonetheless be responsible for recognizing the fact that she is a drug addict. Though scientists may not have bothered to put women in their clinical drug trials, they are nonetheless unjustified in drawing generalizations that fly in the face of the relevant (unknown) facts about the effects of the drugs on women, because they are responsible for knowing these facts. Some events in the world, given how they are positioned within a social world of epistemic

36

‘Yo!’ and ‘Lo!’

agents with concerns and stakes in how things are, have the normative significance of being such that you ought to attend to them.37 So the boundaries around what we are responsible for knowing carve out an area larger than the known and smaller than the knowable. How exactly these boundaries should be drawn depends on an endlessly complex cocktail of our collective capacities, skills, projects, interests, values, and environment. If we did not have epistemic responsibilities to know that could outstrip our actual knowledge, then the notion of failing in our epistemic responsibilities would be meaningless, and inquiry would cease to be a normative activity. There has to be a possible gap between what we are responsible for knowing and what we know. And if we had epistemic responsibilities only for drawing inferences from what was already known, rather than for seeking out and attending to new empirical facts through skilled observation, then the empirical world would not serve as a tribunal to which we hold ourselves accountable in inquiry. This means not only that we can exercise our receptive capacities, but that we are under an epistemic injunction to do so in specific ways.38 Now of course some kinds of epistemic responsibilities are agentrelative. I am responsible for noticing my son’s fear in a way that you are not. This is a moral difference between us. But there is also a sense in which we can talk about agent-neutral epistemic responsibilities. Insofar as a fact is or should be something that ‘we’ know, to that extent we are all answerable to it, and we each fail to meet our individual epistemic responsibilities, however minimally or exculpably, if we don’t know it. While it may not be morally required that you know about my son’s fear, if you don’t know about it, you don’t know something that matters. No one can blame me for not knowing arcane facts about muons published in specialized physics journals. But clearly this lack of knowledge is an epistemic defect in me in a way that my lack of knowledge of the number of rocks on Neptune is not. We know about muons because muons matter to us, and I know less than is known. Further, the fact that something is known, while it may not provide a 37. We take our account of normative accountability in this section to be deeply sympathetic with that given, in much more detail, in Rouse, How Scientific Practices Matter, from which we have inherited the language of ‘stakes’ in particular. 38. Many epistemological theories ignore this requirement, explaining at most why the empirical beliefs we do have can be justified, but leaving one the option of defending one’s beliefs by simply locking oneself in a sound- and light-proof room.

Pragmatism, Pragmatics, and Discourse

37

positive duty for me to find it out, most certainly does provide a positive prohibition against my denying it. Were I to deny something that is known, I would be subject to social correction; that is, I would be liable to being held to the epistemic norms by others in the community. And even though it is clearly not my personal social responsibility (given my training and expertise) to find out how heart medication should be dosed for women, it is our failing, in which I participate, that we don’t know this. In this broad sense, anything that either is or should be known by anyone is something that ‘we’ should know, precisely because of the agent-neutral outputs of our epistemic activities. This is part of what is special about epistemic responsibilities as opposed to other sorts: the fact that truth itself is agent-neutral goes tightly hand in hand with the fact that our narrowly epistemic responsibilities to it are shared. In light of all this, let us return to the question of the difference that the performance of declaratives actually makes to the status of other members of a discursive community. Some declaratives create new epistemic responsibilities by adding to the body of collective knowledge; they report on a new piece of knowledge—be it an empirical truth-claim justified by an observation, or a complex inference, or whatever—for the first time and thereby enter it into social epistemic space. In this case their agent-neutral import seems clear. The real question, then, is what to do with those declaratives that reassert what is already known, most paradigmatically by telling it to someone who did not yet know it. Such declaratives can seem to have no agent-neutral outputs. The truth-claims they assert are already known ‘by us’, so they do not enter those claims into agent-neutral social space or create new agent-neutral epistemic responsibilities. Their pragmatic function is specifically to tell some people who don’t already know about something. How can this involve an agent-neutral output? The answer turns on remembering that speech acts, on our account, are normative functions that strive to accomplish something but will often fall short of doing so. We have argued that the agent-neutrality of the output of a declarative (or an observative, for that matter) goes along with an ideal of universal uptake. Such speech acts seek uptake from everyone, although they will rarely achieve this. Thus, when we offer up new knowledge in an observative or a declarative, that knowledge is for everyone but will only be taken up by some people. Universal uptake is part of the telos of these speech acts, but not part of what they typically accomplish. But this means that there is still a performative point to ut-

38

‘Yo!’ and ‘Lo!’

tering a declarative that states something that has already been stated, in a new context, at a new time, to new listeners, etc. It is true—and important—that such repetitive declaratives which anaphorically pick up on the content of a prior speech act do not in any way change the ideal normative statuses of anyone in the community, for the prior speech act already established that ideal, and turned failure to live up to that ideal into a social epistemic defect. Rather, they normally change the facts about who lives up to that ideal by actually taking up the claim. But this means that the output of the declarative—qua declarative—is agent-neutral, even though the practical point of uttering it is only to change the normative status of specific (formerly ignorant) people. Like its anaphoric predecessor, its output is a truth-claim on everyone. Indeed, it calls upon those who did not already accept the claim to do so, not in virtue of specific agent-relative facts about them, but agent-neutrally, as mere members of the discursive epistemic community. There is an important sense in which the first utterance of a declarative can effect a normative transformation that is different from what the subsequent utterances will accomplish. But this does not take away the practical point to reasserting declaratives, or the agent-neutrality of their output. And again, utterances will never be purely declarative—they will enact multiple functions and have multiple inputs and outputs, and they will also be caught up in various levels of social and ethical normativity that do not directly concern their functioning as truth-claims. So, of course, there will be times when I specifically want you to hear and accept a particular claim. In this case, my speech act strives for an agentrelative effect. (For detailed discussion of such tellings, see Chapter 7.) But insofar as I want you to accept my claim as a warranted truth (and not, for instance, as something that I am demanding that you say because I am your boss), I want you to accept it as making an agent-neutral claim on you and giving you an agent-neutral entitlement. I want you to accept it not as true for you and me, but as true tout court.

1.7 Where We Go from Here In this introductory chapter we have focused on two types of speech acts, namely declaratives and imperatives. We have done so both be-

Pragmatism, Pragmatics, and Discourse

39

cause their structure, with respect to our typology, is relatively straightforward, and, perhaps more to the point, because these are the pragmatic categories of speech acts most familiar to philosophers. However, the attention we so far have paid to speech acts that fall into boxes 1 and 4 of our grid naturally raises the question of what sorts of speech acts might fit into boxes 2 and 3. A speech act in box 2 would be one with an agent-relative input and an agent-neutral output. Thus, its entitlement would be indexed to a specific kind of agent, or a particular agent, but it would offer up an agent-neutral normative status that is not so indexed. Although we will argue in Chapter 4 that Austinian performatives are not the best example of box-2 speech acts, some Austinian performatives provide relatively clear and philosophically familiar examples of how a box-2 speech act would work. Consider an old philosophical favorite, namely a baptism: “I name this ship the Queen Gizelba.” This speech act has an agentrelative input: only someone with a specific authoritative position, not even ideally extendable to everyone, can baptize a ship. But the output of the baptism is agent-neutral: its main effect is to make it true for everyone that this is now the ship’s proper name, and to create a public entitlement to use this name, in conversations, legal documents, or whatever else. A box-3 speech act would primarily function to draw upon an agentneutral entitlement in order to impute an agent-relative status. That is, it would use publicly available features of the world in order to target a particular agent or kind of agent for special normative entitlements or commitments. In Chapter 5 we will argue that many deontic claims— those we will call ‘prescriptives’—are examples of speech acts with this structure. A claim such as “Jim ought to do a better job of taking care of his dog” does not have its performance entitlement indexed to any particular kind of agent. If it is properly performed by anyone, it is because it reflects agent-neutral (moral and empirical) features of a public world. On the other hand, the claim has special, nonfungible normative implications for Jim. So far we have defended or provisionally suggested the categorizations of speech acts shown in Figure 3. In Chapters 2 and 3 we will examine the contents of box 2 in detail. Chapter 4 provides a brief interlude in which we consider the status of Austinian performatives. Chapter 5 will be devoted to box 3. In

40

‘Yo!’ and ‘Lo!’

Chapters 6 and 7 we will turn to a different—and in our view fundamental—category of box-4 speech acts. Chapter 8 concerns the character and constitution of discursive communities and the agents that inhabit them. Our special concern with declaratives in this first chapter should not be read as another inscription of the primacy of the declarative, which would amount to an instance of the declarative fallacy. By the end of the book we will have argued that at least two other pragmatic functions of language are as fundamental as its declaratival function. These are (1) giving expression and calling attention to a speaker’s receptive recognition of an empirical state of affairs (“Lo, a rabbit!”) and (2) call-

Input Output

Agent-neutral

Agent-neutral

Agent-relative

1 Neutral input Neutral output

2 Relative input Neutral output

Declaratives

3 Neutral input Relative output Agent-relative

Figure 3

Prescriptives (i.e. ought-claims)

Baptisms

4 Relative input Relative output Imperatives (promises, invitations, reproaches...)

Pragmatism, Pragmatics, and Discourse

41

ing or hailing another person (“Yo, Fiona!”). In distinctively different ways, our abilities to perform Lo-utterances and Yo-utterances are transcendental conditions upon the possibility of speaking a language with which we communicate with one another about a shared objective world.

2

Observatives and the Pragmatics of Perception

Among categories of speech acts, declaratives and imperatives have received the lion’s share of attention from philosophers of language, and indeed we upheld this trend in Chapter 1. Declaratives and imperatives share a significant feature: the pragmatic scope of their input matches that of their output. That is, where declaratives are agent-neutral in both input and output, imperatives are agent-relative in both input and output. Historically, focusing on these two types of speech acts has, not surprisingly, obscured the whole distinction between inputs and outputs: philosophers are used to thinking in terms of agent-neutrality or agentrelativity, tout court, but they have not noticed that the question of the scope of a norm, reason, or speech act has to be asked separately with respect to its input and its output. However, some of the most philosophically interesting work that we do, as members of discursive communities, gets done by actions that instantiate ‘mixed’ normative functions, belonging in boxes 2 and 3 of our grid (that is, with agent-neutral inputs and agent-relative outputs, or with agent-relative inputs and agentneutral outputs). Our purpose in this chapter is to explore some inhabitants of box 2 and their philosophical significance. A successful box-2 speech act must be one that, as a matter of its pragmatic structure, has an agent-relative input. As in the case of an imperative, entitlement to its performance must be structurally indexed to an agent (or agents) with a particular normative position that is not generalizable even in the ideal. At the same time, it must have an agentneutral output. As in the case of a declarative, it must pass on entitlements and commitments that are not indexed to particular agents, but 42

Observatives and the Pragmatics of Perception

43

are rather public, whether or not everyone manages to take them up. It must make a claim that is not for anyone in particular, but rather has ‘universal validity’, as Kant would say. What might such speech acts be? Near the end of Chapter 1, we suggested that certain Austinian performatives, such as baptisms, seem to belong to box 2: only someone with a specific normative position can perform a baptism—taking a stroll through a neonatal nursery and shouting names at other people’s babies does not constitute baptizing them— but a successful baptism makes it the case for everyone that this thing or person has this name. In Chapter 4, however, we will argue that while such examples may be helpful for initial clarificatory purposes, most Austinian performatives are not the most compelling, paradigmatic, or philosophically interesting or important examples of box-2 speech acts as such. Kant is perhaps the only figure in the history of philosophy who has identified a type of speech act that belongs in box 2—that is, one with agent-relative input and agent-neutral output. In his discussion of the structure of aesthetic judgments of taste in the Third Critique, Kant argues that judgments of taste are essentially singular, by which he means not only that they are about a single, concrete particular rather than a category of objects (though this follows immediately from his analysis), but also that they require a personal encounter with the object of judgment. He writes: “There can . . . be no rule in accordance with which someone could be compelled to acknowledge something as beautiful . . . No one allows himself to be talked into his judgment about that by means of any grounds or fundamental principles. One wants to submit the object to his own eyes.”1 The point here is not that no generally valid rules can predict with perfect accuracy which objects will be beautiful, though this is likely to be true too, but that even if we had such rules, inferring the beauty of an object on their basis wouldn’t count as an aesthetic judgment of taste at all—what it is to make an aesthetic judgment is to aesthetically respond to a concrete, sensuous encounter with an object. As such, only my inherently agent-relative encounter with an object can be the ground for my judgment of taste, regardless of how confident I am on the basis of general principles that I would or wouldn’t find this ob1. Immanuel Kant, Critique of the Power of Judgment, trans. Paul Guyer and Eric Matthews (New York: Cambridge University Press, 2000), 215–216.

44

‘Yo!’ and ‘Lo!’

ject beautiful once I encountered it. A judgment of taste, therefore, does not pass on a reassertion license—or better, a rejudgment license. “One wants to submit the object to his own eyes,” Kant says, “and yet, if one then calls the object beautiful, one believes oneself to have a universal voice, and lays claim to the consent of everyone.” Kant argues at length that even though my judgment of taste requires a personal encounter, in so judging I impute this judgment agent-neutrally, to everyone. I cannot pass on an entitlement to the same judgment as mine, but in judging I demand universal agreement. Even though I am well aware that in fact not everyone may judge as I do, in judging beauty I judge that the object is beautiful for everyone, in the sense that it is a defect in others if they fail to judge as I do. The judge “does not count on the agreement of others with his judgment of satisfaction because he has frequently found them to be agreeable with his own, but rather demands it from them [and] rebukes them if they judge otherwise.” Although they cannot pass on universal reassertion licenses, they make a universal demand upon others to judge for themselves as I judge should they encounter the same object. In other words, my singular judgment of taste has an agent-neutral output, and it makes the normative claim that others are in error if they disagree with me, thereby ‘laying claim’ to universal agreement. Judgments of taste are not objective truth-claims, because of the agent-relativity of their input, but they have, as Kant puts it, subjective universal validity. Kantian judgments of taste are structurally agent-neutral in the validity they claim, even as they are structurally agent-relative in their entitlement; hence they belong in box 2.2 Kant’s claim is stronger than that the inputs of judgments of taste are agent-relative. For one thing—in the lingo of Chapter 1—he is claiming that these inputs are agent-specific. It is not merely some particular social category of person to whom the entitlement of an aesthetic judgment 2. Ibid., 215–216, 212–213. Because philosophers have not, in the past, distinguished between the inputs and outputs of judgments—which in retrospect is surprising, since Kant himself provided the framework for thinking of judgments as normative functions—interpreters have struggled to make sense of Kant’s talk of “singular judgments.” Indeed, Kant himself changes the meaning of the term over the course of the critical philosophy. In the first critique, the ‘singularity’ of a judgment concerns the extensional scope of its content—it is a judgment about only one thing. But in the third critique, the ‘singularity’ of a judgment is a feature of how the judgment is made—it is entitled only by a personal encounter. In the past, philosophers have not had good language available for making clear that this later Kantian singularity is a pragmatic category—it identifies a structural feature of the input of the judgment.

Observatives and the Pragmatics of Perception

45

applies. Though entitlement to give orders to privates may attach to lieutenants, and entitlement to name a child to any of its parents, aesthetic entitlement of this sort applies not to a specific category of person but to a specific person. But Kant’s claim is even stronger than that: for him, the input of a judgment of taste is essentially first-personal: it is essential to a judgment of taste that what I give expression to, in judgment, is my own response to the object I encounter. I can make no judgment of taste on behalf of another, no matter what our relationship or our similarities. While some have argued that aesthetic judgment in fact plays an essential role as a moment in regular objective assertion, for Kant, at least on the surface, Kantian judgments of taste form an esoteric category cut off from the main concerns of epistemology.3 His discussion of them provides us with a beautiful historical precedent, but we will not focus on them as our paradigm of box-2 speech acts. Rather, in the rest of this chapter we turn to a category of speech acts that we believe share important structural similarities with Kant’s judgments of taste. We argue that these speech acts play a pivotal role in discourse about and knowledge of an objective empirical world.

2.1 Observatives Consider the difference between two speech acts: • When my friend asks why I am crouching near a bush with a car-

rot, I declare, “There’s a rabbit in the bush.” • As I see a rabbit dart into a bush I call out, “Lo, a rabbit!”

We claim that while both speech acts directly imply the presence of a rabbit in the bush—both would be misspoken if there were no rabbit in the bush—they differ crucially in pragmatic structure, in such a way as to make the latter, a speech act of a type we will call the ‘observative’, inhabit box 2 instead of box 1. First let us introduce the term ‘recognitive’ for any speech act a function of which is to give expression to a speaker’s recognition of something. A recognitive does not, as such, assert a proposition about the content of 3. For further citations, see Rebecca Kukla, “Introduction: Placing the Aesthetic in Kant’s Critical Epistemology,” in Kukla, ed., Aesthetics and Cognition in Kant’s Critical Philosophy (Cambridge: Cambridge University Press, 2006), 1–34.

46

‘Yo!’ and ‘Lo!’

what the speaker recognizes—although many recognitives will also be declaratives and do this as well. Rather, the pragmatic function of the recognitive is to discursively mark and communicate the event of recognition itself. English contains a few words and constructions that are specifically designed to mark this recognitive function, although most of them sound rather archaic: ‘Lo!’, ‘Ho!’, and, as we will discuss in detail in Chapter 6, ‘Yo!’ and other hails (e.g. ‘Ahoy!’). Most often, however, we issue recognitives using utterances whose surface grammar is not distinctive. “There’s a rabbit in the bush!” can either be a mere declarative statement of a truth (for instance, a truth we already knew or were told by someone else), or it can be an expressive response to actually seeing a rabbit. Because recognitives are not routinely marked by their surface grammar, we will, as a matter of notational convention, follow what seems to be the loose ordinary-language practice of using exclamation points to indicate the recognitive function of an utterance. So “I see a rabbit.” is our way of writing a declarative statement about an observational state, whereas “I see a rabbit!” indicates that this utterance centrally gives expression to a recognitive event. Observatives, by stipulation, are those recognitives that give expression to our recognition of an empirical fact, object, or state of affairs in observation, and most paradigmatically in perception. (For more about other kinds of recognitives, see Chapters 6 and 7.) That is to say, observatives are much like what philosophers have called observation reports, except that the latter term does not distinguish between mere declarative reports on the content of observation and expressions of the event of observational recognition. The importance of this pragmatic distinction is a major thesis of this chapter. Opinions will differ as to the scope of the notion of observation. There are varieties of skillful recognition that will seem to many not to be instances of observation. We might think, for example, that someone can be in a position to ‘see’ that Rebecca is uncomfortable at parties with people she doesn’t know well, or perhaps that a particular mathematical proof is inelegant. That is, philosophers will differ with respect to how narrowly they wish to tie the notion of observation to sensation. We want to remain neutral on this issue, and to avoid taking a stance concerning the boundaries of the empirical, the co-extension of perception and observation, and so forth. Our concern is with observatives insofar as they give expression to recognitive episodes that provide direct, non-

Observatives and the Pragmatics of Perception

47

inferential, receptive knowledge of the empirical world, regardless of disagreements over the scope of such knowledge. The output of an observative is agent-neutral. My epistemic accomplishment in expressing my observation of a rabbit establishes a public fact—indeed, it establishes a set of public facts, such as that there is a rabbit present, and that I see it. Again, if there were no rabbit present, I would have misspoken (although we shall argue below that it’s not quite right to say I would have spoken falsely). Accordingly, my warranted speech act provides anyone who accepts my entitlement to it with entitlement to the claim that there is a rabbit present (and that I see it, etc.). In virtue of a successful observative—“Lo, a rabbit!” for example—“we” now know (or “one knows”) that there is a rabbit nearby. My observative directly and agent-neutrally licenses beliefs, inferences, and declarative speech acts concerning the presence of a rabbit (among other beliefs and declarations). Yet my observative does not seek, even as an ideal, to bring it about that anyone else has observed the rabbit, nor likewise that anyone else is entitled to utter “Lo, a rabbit!” Entitlement to the observative (the ‘Lo!’ speech act) is inherently one’s own. In uttering this speech act, I express my receptive recognition of a rabbit. This speech act does not merely make the declarative claim that a rabbit is present. Nor does it merely make the declarative claim that I see a rabbit. Rather, it serves a special recognitive function: it marks or expresses my detection of a rabbit. It is the recognizing, and not just what is recognized or who is recognizing, that is given expression in such a claim, and since what is expressed is the indexed recognition itself, this entitlement is not generalizable, even in the ideal. In this sense, observatives have a pragmatic structure analogous to Kantian judgments of taste. In this chapter’s discussion of observatives, we sometimes focus on Lo-utterances because our most familiar examples of observatives in the philosophical literature come from Quine, who was fond of the locution. However, Lo-utterances actually form only a proper subset of observatives, and indeed one that has an additional distinctive pragmatic function. Notice that Lo-claims do more than just express recognition; they also ostend. That is to say, they call upon some others to attend to and recognize that which I am currently recognizing. If I see a rabbit that no one else is in a position to see—perhaps everyone but me is in another room of the house—I might call out an observative: “A rabbit!”; or

48

‘Yo!’ and ‘Lo!’

“There goes a rabbit!” But it would be inappropriate for me to utter “Lo, a rabbit!” The ‘Lo!’ both expresses my observation and calls upon specific others to share in my attention and thereby to observe the same thing for themselves. Since this ostensive function is directed only at those around us who are in a position to re-create our receptive encounter, it is an agent-relative output of the Lo-utterance. This means that Lo-claims have a complex structure, involving both a box-2 observative function and a box-4 ostensive function. We will take up the importance of such special ostensive observatives in the next chapter, and again in our final chapter. Observatives—unlike declaratives—do not issue reassertion licenses. To perceive is to be uniquely placed, indexically, with respect to what I see. Perceptual episodes are inherently particular and non-fungible in just the way that the inferential and assertional entitlements to which they give rise are not. The inference and assertion licenses I ‘pass on’ when I express what I perceive meaningfully maintain their identity through their different incarnations in different speakers, but the original perceptual episodes do not. My utterance, “Lo, a rabbit!,” may commit you to the belief that there is a rabbit present. You may even declare that the rabbit is present on the basis of this belief, without having managed to see it: imagine, for example, that you are peering into a bush trying to see the rabbit that I have authoritatively sworn is in there, and when a passerby asks you why you are staring at the bush, you declare, “There is a rabbit in the bush.” But unless you see the bunny yourself you are not entitled to utter “Lo, a rabbit!” Indeed, it would be deceptive for you to do so. And if you do see it, the source of your entitlement is this recognition—your recognition—not your acceptance of my entitlement to my own observative. Even if we insist that you see the very same thing as I do on a given occasion—you see the bunny also—we have two perceptual episodes grounding two different receptive entitlements, and not one. Just as only taking aesthetic pleasure in an object yourself can constitute an aesthetic judgment of beauty, likewise only seeing the rabbit yourself entitles you to this speech act—as opposed to, for instance, making a warranted inferential claim that there is a rabbit present. Observatives, that is, are licensed only for a concrete individual, from her particular, first-personal point of view. Perception paradigmatically yields new entitlements. But Rebecca’s perception yields entitlements for her only insofar as she recognizes that she has perceived the bunny her-

Observatives and the Pragmatics of Perception

49

self, whereas Mark earns his entitlements through learning that someone else has perceived the bunny. These are not interchangeable sources of entitlement: my perception will only yield new entitlements for anyone if they originate with me—no one else but me can be the first to pass on my perceptual entitlements. To perceive, then—as opposed to just inheriting entitlement to a belief—is to be first-personally claimed by what I see. To express this firstpersonal episode in language is to take on a singular responsibility for correct observation in a way that is not an expression of any kind of shared, agent-neutral commitment. The receptivity of perception is one of the essential means by which my commitments and entitlements do not merely accrue to me, but make a claim on me. The perspectivally owned character of perception is not just a phenomenological fact that needs separate accommodation, but rather it is essential to the cash value of the game of giving and asking for reasons. This is a Kantian commitment on our part. Part of the point of Kant’s transcendental synthesis of apperception in the ‘I think’ is just this necessarily firstpersonal ownership of objective representations. Kant himself may well be caught in the declarative fallacy here, insofar as he tries to capture this first-personality by adding an extra bit of propositionally structured representational content—the judgment that one is thinking the representation—to the original representation. For us, this first-personality is an irreducible feature of the pragmatic structure of the receptive event itself.4 There are three distinct claims here. First, and most simply, the input entitlement to an observative is agent-relative. Second, just as with Kant’s aesthetic judgments, it is agent-specific: I am entitled to my observative utterances, not insofar as I am a fungible instantiation of some category of agents, but because they give expression to my unique and fully con4. This leaves open the interesting question of whether one must be able to make explicit one’s own first-personal ownership of a receptive episode in order to count as a genuine perceiver. An anonymous referee pointed out that it is dubious that animals, for instance, could have such an explicit grasp of their own relationship to their perceptual states, while at the same time it is hard to deny that they are perceivers. Everyone will agree that there is some important sense in which animals perceive; the open question here is whether there is some richer epistemic practice that those of us who can explicitly recognize our first-personal states are engaging in when we perceive. So, for instance, Sellars and McDowell deny that animals are perceivers in the full-blooded sense in which we are, for this reason, whereas Brandom and Haugeland have less strict prerequisites for such perception.

50

‘Yo!’ and ‘Lo!’

crete encounter with the world. Third, the observative expresses my first-personal uptake of this agent-specific entitlement. Any speech act, including a declarative speech act, calls for a firstpersonal uptake of entitlement. Knowing that Mark is entitled to endorse the Kondo-Addison theorem provides me with no inferential guidance unless I know that I am Mark, and hence that I am entitled to endorse the theorem. But it does not follow from this that all speech acts—or even all speech acts with agent-specific entitlements—serve to express this first-personal uptake. When Rebecca says, “You are Eli,” in the right context, she does so out of an agent-specific entitlement to baptize her child (and her speech act falls into box 2). But there is nothing in this sort of baptizing speech act that expresses her first-personal uptake of this entitlement.5 In contrast, observatives have the essential function of giving expression to this first-personal uptake, and hence the first-person voice is a structural feature of their defining pragmatic function. At this point we can make explicit a final feature of the input entitlements of observatives: since they are both agent-specific and firstpersonal, they are also—like Kantian judgments of taste—essentially unshareable. That is, they express a first-person singular perspective, and never a first-person plural perspective. We come to be entitled to a belief on the basis of observation because we encounter the world in a certain way. But such events of receptive encountering are essentially individuating, in a very specific sense. As Heidegger would put it, they are in each case mine. The point is that such concrete receptive encounters by their very nature build in essentially singular first-personal ownership of that encounter; observational episodes cannot even conceptually be understood as floating free of being someone’s observational episode in particular. Whatever our metaphysics of subjectivity happens to be, and wherever we think that the boundaries of the individual may lie, it is built into the structure of certain events that only individuals can be the subjects of such events. Heidegger argued that only I can die my death: this 5. Conversely, it seems that there are speech acts that do serve—as part of their pragmatic structure—to express first-personal uptake of agent-neutral entitlements. Imagine that I claim to understand a well-known but difficult theorem. You doubt it. I respond by saying “Watch!” and then writing down the proof. Here, it seems, I am expressing my first-personal uptake of agent-neutral entitlement to the theorem. If the term were not already a name for a grammatical type, we would be happy to label such performances “demonstratives.”

Observatives and the Pragmatics of Perception

51

death cannot be shared and no one can die for me (though someone might die herself in order to spare my life, which is quite different). Just as essentially, no one but me can have my experiences or partake in my observations: they can be neither shared nor displaced. We might think that only human beings can be individual subjects, or that the status should be extended to animals, corporations, social groups, or whatever.6 We might think that individuals are in various ways ineliminably socially embedded, bound up in relationships, or historically constructed. But the structural point we are making here cuts across all of these views: whatever individual subjects turn out to be, only one of those can die a death, live a life, or recognize a rabbit in a bush.

2.2 Observatives and Occasion Sentences On the basis of our analysis in the last section, we can say that observatives, given their agent-relative input and their agent-neutral output, belong in box 2 of our grid. At this point, we have defended the classifications indicated in Figure 4. Contemporary philosophers of language have not identified observatives as speech acts with a distinctive pragmatic structure. This is odd, since observation reports have frequently been pressed into distinctive pragmatic service. Davidson, for instance, uses occasion sentences as the starting point for triangulation from the language of the interpretee to the language of the interpreter to the world itself.7 Indeed, he takes our capacity to recognize speech acts as observation reports—or, more precisely, as assents to sentences whose truth fluctuates with the passing scene, as he puts it—as a primitive condition for the possibility of interpretation. It is odd, given this special pragmatic place that Davidson assigns, not just to observation, but specifically to the special speech acts that report on observations, that he did not concern himself with giving any kind of pragmatic analysis of these acts. For example, Davidson proposes no mechanism for differentiating between proper observatives, in our sense, and declarative speech acts 6. We (Mark and Rebecca) disagree as to whether a group or a corporation could potentially count as an individual subject in the relevant senses. 7. Davidson’s triangulation argument shows up in multiple papers, but the classic source is “Three Varieties of Knowledge,” reprinted in Davidson, Subjective, Intersubjective, Objective (Oxford: Clarendon Press, 2001).

52

‘Yo!’ and ‘Lo!’

with which I describe my own observational episodes—both of these simply count, for him, as observation reports. But it’s hard to imagine how declarative descriptions of the contents of observation could anchor interpretation in any special way that other objective truth-claims could not. A Cartesian might try to mark out the speech acts that play a privileged role in anchoring interpretation at the level of semantic content, claiming that descriptions of the contents of our own experience (“I see a rabbit”) are epistemologically prior to claims about the world. But such a view is not open to Davidson, who, like all post-Kantian neopragmatists, holds that the ability to have beliefs about the world is as primordial as the ability to have beliefs about the contents of one’s own

Input Output

Agent-neutral

Agent-neutral

Agent-relative

1 Neutral input Neutral output

2 Relative input Neutral output

Declaratives

3 Neutral input Relative output Agent-relative

Figure 4

Kantian judgments of taste, baptisms, some recognitives, i.e. observatives

4 Relative input Relative output Imperatives (promises, invitations, reproaches...), ostensions

Observatives and the Pragmatics of Perception

53

experience. So there is no way to find the special role of observation in the semantics of claims like “I see that P.” And semantics will not help us if we turn simply to speech acts such as “There is a rabbit in the bush,” for the content of such acts is just what is shared between observational and inferential episodes. A Davidsonian might well respond that what makes occasion sentences distinctively useful in anchoring interpretation is the way they are causally connected to the world, rather than their semantics or their pragmatics. But all speech acts are causally connected to the world somehow or other, and Davidson would need some non-question-begging way of making the causal origin of occasion sentences distinctively important. It certainly seems that these causes are important precisely because they give rise to speech acts that are recognitive responses. But such responses need to be pragmatically distinguished from other speech acts if we are to demarcate which causes give rise to them, rather than the other way around. Hence, we believe, Davidson implicitly depends upon our ability to sort observatives from other inferentially fecund speech acts in virtue of their pragmatic function.8 More generally, we suggest that observatives have been playing an important theoretical role in philosophy of language for some time now, and our analysis should be understood as making precise a notion upon which we were already dependent.

2.3 Observing-That and the Declarative Fallacy Even though they are not themselves assertions, observatives are firmly planted within the conceptually articulated space of reasons. “Lo, a rabbit!” is an utterance that makes use of the concept ‘rabbit’; although it expresses a receptive encounter, what it captures is not a preconceptual ‘given’ but a conceptually articulated experience. (For discussion of the conceptually rich character of the content of observatives, see Chapter 3.) When philosophers have considered non-declarative, expressive lan8. Because Davidson focuses on sentences rather than speech acts, it would be hard for him to draw this distinction. For after all, one and the same sentence—such as “I see a rabbit”—can serve as either an observative or a declarative report on one’s observational state. Davidson and Quine, despite their lack of pragmatic analyses, often use explicitly observative locutions such as “Lo, a rabbit!” for their examples of occasion sentences. This perhaps enables them to focus in on speech acts that tend to do the pragmatic work they want without having explicitly identified their distinctive pragmatic structure.

54

‘Yo!’ and ‘Lo!’

guage in the past, they have focused on inarticulate expletives like ‘boo’ and ‘yay’. Indeed, expressive language has been taken as almost synonymous with ‘non-cognitivist’ language.9 However, observatives function differently from such traditional examples of expressive language. They ground beliefs and public knowledge, they facilitate discursive communication, and they allow us to give voice to our encounters with the objective world. Yet we maintain that observatives are not truth-claims, even though they license truth-claims. Truth is inherently public, and for that reason access to the truth is agent-neutral. (Although, again, not everyone will be equally able to access various truths.) While we are not proposing any kind of full-blown theory of truth in this book, we have suggested that the agent-neutral accessibility of truth goes right to the heart of what is distinctive about it; truth is never for you or for me. Likewise, truth-claims essentially have agent-neutral inputs. While the content of what I observe is a matter of public truth, the event of my observation is inherently mine, and when and insofar as I give expression to that event in an observative I do not assert a truth, any more than I do when I shout “Ouch!” or give expression to other first-personal events. A revealing piece of evidence that observatives are not truth-claims is that although you might accept my entitlement to my Lo-utterance, and indeed accept on its basis that there is a rabbit present, it is pragmatically inappropriate for you to respond, “That’s true!” Typical assertions of “That’s true,” whatever other function they may have, behave as “prosentences,” which pick up their semantic content anaphorically, from an antecedent declarative. In response to a declarative, the claim “That is true” functions both as the assertion with the same content— semantic anaphora—and also as the re-performance of the very same speech act with the same declarative structure.10 But if we acknowledge an observative—perhaps by saying “You’re right”—we thereby claim the semantic content of the observative but without reinstantiating its pragmatic structure. This new speech act has the same output as the observative, but a very different input. Standard discussions of anaphoric prosentences, then, are ambiguous, failing as they do to distinguish the 9. Allan Gibbard, Thinking How to Live (Cambridge, Mass.: Harvard University Press, 2003); C. L. Stevenson, Ethics and Language (New Haven: Yale University Press, 1944).

Observatives and the Pragmatics of Perception

55

reproduction of a speech act type from the production of a new speech act with the same semantic content.11 Observatives often lack explicit propositional content. “Lo, a rabbit!” is perfectly idiomatic (at least insofar as anyone is pretentious enough to say “Lo!” at all). And “Land, ho!” is certainly idiomatic among Hollywood pirates. If we insist upon understanding observatives as a subcategory of declaratives, then we must read them as expressing their propositional content elliptically. But this introduces apparent arbitrariness when it comes to filling in the ellipsis. Is “Lo, a rabbit!” equivalent to “Lo, there is a rabbit present!,” “Lo, there is a rabbit in the bush!,” or perhaps “Lo, a rabbit is near enough to me for that to be remarkable!”? There seems no principled way to choose among the numerous declaratives that serve as output to the observative, and this makes the declaratival analysis seem decidedly ad hoc. We claim instead that these observatives are complete, well-formed utterances that imply propositional truths with each of these contents (as well as many more), but that they are not themselves propositional in form.12 Now consider an observative that does not immediately raise the problem of ellipsis: “Willard is on the mat!” Unlike “Lo, a rabbit!,” or even just “A rabbit!,” this observative utterance takes the form of a proposition that can function as a purely declaratival expression. Clearly, there is an important sense in which these two utterances—“Willard is on the mat!” and “Willard is on the mat.”—have the same seman10. For philosophers, the most important discussions of prosentential anaphora are Robert Brandom, Making It Explicit: Reasoning, Representing, and Discursive Commitment (Cambridge, Mass.: Harvard University Press, 1994), chap. 5, and Dorothy Grover, Joseph Camp Jr., and Nuel Belnap, “A Prosentential Theory of Truth,” Philosophical Studies 27 (1975): 73–124. For an argument that an anaphoric account of the semantics of truth talk is quite independent of the broader inferentialist semantic framework in which Brandom places it, see Mark Lance, “The Significance of Anaphoric Theories of Truth and Reference,” in Bradley Armour-Garb and J. C. Beall, eds., Deflationary Truth (Chicago: Open Court Press, 2001). 11. If truth-talk functions only to relate anaphorically to the content of an antecedent utterance, one wonders whether there is a corresponding pragmatic pro-form, a speech type that functions systematically as a re-performance, drawing not only its content but its pragmatic significance from an antecedent. Such vocabulary exists in colloquial English, we think. One can utter “You can say that again!” or “Indeed!” (or, in reasonably current street lingo, “Word!”) as a way of picking up the pragmatic force of a speech act. 12. See Robert J. Stainton, Words and Thoughts: Subsentences, Ellipsis, and the Philosophy of Language (New York: Oxford, 2006), for a compelling argument against reading nonpropositional utterances as elliptical.

56

‘Yo!’ and ‘Lo!’

tic content. But unlike its declaratival half-sister, the observative still has an agent-relative input, and it still functions to express recognition of something. The fact that these two utterances in some sense share semantic content can make it easy to think that the observative is just a funny version of the declarative—one that just happens to also mark its own causal genesis in an observation, through tone of voice, or perhaps through the addition of a ‘Lo!’ At this point we might be tempted by the thought that this observative is really just a short form for two declaratives shoved together: “Willard is on the mat.” and “I am seeing that Willard is on the mat.” or something of the sort. But again there does not seem to be a good way of choosing between the various ‘declarative translations’ of the observative. Does it really mean “Willard is on the mat and I see him there”? Or perhaps, “Willard is on the mat, and the reason I know this is that I am seeing him there”? Or maybe, “I see something. Willard is on the mat.” Indeed, it seems that all of these express declarative commitments are immediately licensed by entitlement to the observative “Willard is on the mat!” At the same time, there seems to be no reason to pick one of them as the ‘proper’ analysis of the observative. Though not a declarative, an observative licenses moves to many declaratives. But the fact that these declaratives are licensed by the observative does not show that they are identical to it, and the fact that there seems to be no good reason to choose one of these declarative translations over the others strongly suggests that none of them in fact exhausts or nails down the import of the original. We submit that any such ‘reduction’ will be either arbitrary or driven by a theoretical commitment that begs the question in favor of the primacy of declaratives. In our view, it should not be surprising that the difference between the propositional observative and its declaratival counterpart cannot be analyzed in terms of a difference in propositional content. What interests us here is specifically that the difference between these two utterances seems to be one of pragmatic function, and not one of semantics. If we analyze “Willard is on the mat!” as some cluster of declaratives, we do not capture its crucial function of expressing (rather than asserting) the speaker’s observational recognition of the fact that Willard is on the mat. The temptation to try to capture the difference at the level of propositional content, we think, is motivated by the tendency of philosophers to privilege semantics over pragmatics, and declarative pragmatic struc-

Observatives and the Pragmatics of Perception

57

tures over others, with the result that all differences between utterances tend to be understood, if at all possible, as differences in semantic content between declaratives. In other words, it is an instance of the declarative fallacy. In rejecting the declarative fallacy, we reject first and foremost a methodological orientation that assumes that all speech acts have a declaratival structure made up of propositional contents until proven otherwise; we pointedly shift the burden of proof in the other direction, thereby pushing against a great deal of philosophical inertia. Observatives may express recognition of propositionally structured facts, or they may express recognition of phenomena or objects. In each case they commit their speakers to believing in the propositions that follow from what they observe. But if we do not buy into the presumption that all speech acts that justify truth-claims have the pragmatic structure of a declarative, then it seems clear that what observatives do, whether or not they happen to embed propositions, is something distinct from merely asserting their content; instead, they express recognitive uptake of their content, and this expression is of something essentially agentspecific and individuating in a way that a declarative assertion essentially isn’t. Recognizing something, including even recognizing the fact that a proposition obtains, is simply distinct—pragmatically distinct— from asserting that something is the case. Thus observatives are not just modifications or transformations of declaratives. Not everything propositional is declarative, and hence it remains possible to agree that observatives are pragmatically distinct from declaratives, while still insisting that observatives such as “Lo, a rabbit!” are elliptically propositional. However, we believe that by now we have undercut the motivation for this move. In arguing that we need not read non-propositional observatives as implicit declaratives, we are in fact resting on the idea that they need not have an implicit propositional structure either.13 We are content to say that the conclusion that there is a rabbit present follows directly from the acceptance of an utterance of “Lo, a rabbit!” as properly entitled, without a mediating translation into a propositionally formed premise. One simply cannot see a rabbit without it’s being the case that there is a rabbit present, and so the former event implies the later truth. Sometimes we express our recognition of 13. Thus we disagree with Sellars, who counts observation reports as a special “level of propositions,” and with McDowell, who follows him in doing so (for instance during his keynote presentation at the Space of Reasons Conference, Cape Town, South Africa, July 2004).

58

‘Yo!’ and ‘Lo!’

a propositionally structured fact—for example of our observation that there is a rabbit in the bush. At other times we express our recognition of an object, or an event—our observation of a rabbit, or of a rabbit darting into a bush. Debates rage over whether we must see-that in order to see at all. Some defend the strong claim that every seeing involves a seeing-that,14 although one could also hold the weaker thesis that being able to seethat, generally speaking, is a transcendental condition for the ability to see objects or anything else. It is not our primary purpose here to enter into such debates on the nature of seeing. Our inclination is to assert the weaker and deny the stronger thesis. But surely it is possible to see a rabbit or a rabbit hiding, in addition to whatever other propositionally structured facts we see, and whatever the dependence relations are between such seeings. We are urging that such non-propositional observations can be expressed in observatives. We are also arguing for the more substantive and contentious thesis that such non-propositional observations and their expressions ground justified declaratives. Now there will be those in the grip of a strong perceptual rationalism, driven perhaps by a Sellarsian critique of the given, who will find this thesis absurd. They will claim that nothing without propositional form could ground another proposition, because it wouldn’t have the right kind of structure to serve as a premise in an inference. But such an objection seems to rest on a suspiciously narrow conception of inference. After all, in the domain of practical inference, reasons ground actions that are not themselves propositional. Of course, only propositions can be inferentially related according to the usual rules of propositional logic, but to presume that propositional logic is, or is structurally analogous to, the only inferential game in town—or that it is the inferential game that must govern observation and its expression—is to beg the question. Indeed, this particular form of question-begging can be understood as a form of the declaratival fallacy, as it presumes that the only discursive logic is the propositional logic of the declarative, and that everything that isn’t propositionally structured must be somehow mute or inarticulate. We are not attached to calling the licensing moves from observatives 14. This is the position taken by McDowell in John McDowell, Mind and World (Cambridge, Mass.: Harvard University Press, 1994), and by Sellars in Empiricism and the Philosophy of Mind (Cambridge, Mass.: Harvard University Press, 1997).

Observatives and the Pragmatics of Perception

59

to declaratives “inference.” One reason to deny that it is possible to talk of an “inference” between observatives and declaratives is that our only well-developed accounts of good inference proceed in terms of truthpreservation, which concerns relations among truth-claims. In the Appendix we argue that one can understand the idea of an inference in terms of a more basic idea of licensed normative moves between types of actions (some of which are speech acts). This would include, for instance, practical inferences from beliefs and desires to actions. Within this framework, we can represent licensed transitions between speech acts as particular cases of the general phenomenon of pragmatically licensed act-transitions. But in any case, little turns on whether or not we attach the label ‘inference’ to the justificatory relation between observatives and declaratives. We could say that knowledge depends both on inference and on non-inferential warranted transitions from observatives to declaratives, or that there are two types of inference. For ease of terminology, we use ‘inference’ in the broader sense in this work.

2.4 The Ineliminability of the First-Person Voice Declarative truth-claims are not essentially indexed to any particular speaker or audience—they are inherently “impersonal” rather than structurally bound to a first-, second-, or third-person voice. A declarative such as “Ottawa is the capital of Canada” has no personal voice. Many declaratives do have a voice: “I am sick of crappy Mexican food”; “You have schmutz on your face”; “Louise thinks that Toronto is the capital of Canada.” However, to the extent that what we are interested in when it comes to the pragmatic force of declaratives is their status as assertions or truth-claims, any declarative can be translated from one personal voice to another without its force being changed in the least—it works the same way regardless of who says it, and to whom. Thus, qua truth-conditional assertion, “I am sick of crappy Mexican food” (spoken by Mark) is just the same as “Mark is sick of crappy Mexican food,” and “You have schmutz on your face” (directed at Rebecca) is just the same as “Rebecca has schmutz on her face,” or (uttered by Rebecca) “I have schmutz on my face.” The fundamental impersonality of the declarative is deeply linked to the agent-neutrality of its input and output. We have pointed out several times that contemporary philosophers of language take the declarative assertion as the fundamental build-

60

‘Yo!’ and ‘Lo!’

ing block of language. But now notice an interesting effect of this starting point: if we assume that the essence of discourse is captured by the functioning of declaratives, then, since declaratives are essentially impersonal and agent-neutral, we exclude from the start the possibility of discovering that some agent-relative, voiced dimensions of language play an important role in constituting or enabling meaningful discursive practices within a linguistic community. Why should we think that this restriction is important? Throughout this chapter we have argued that speech acts that give expression to perceptual episodes—observatives—are structurally firstpersonal. If perceptual episodes had a normative pragmatic structure analogous to acts of declaring, then they would likewise inherit the structural impersonality or agent-neutrality of declaratives. Perceptual episodes could then be ‘passed on’ or transferred between agents without loss of identity. Indeed, Brandom apparently understands perceptual episodes as funny kinds of assertions.15 However, we have argued that perceptual episodes are inherently individuating and unshareable. It is a correlate of this analysis that perceptual episodes themselves cannot be understood as analogous to ‘inner assertions’, but rather share the firstpersonal, agent-relative input structure of the speech acts that express them. Brandom acknowledges a certain perspectivality of entitlement at the level of the content of our intentional states: the content of an agent’s perceptual judgment, for instance, will depend on the orientation of her body and her visual perspective, or the specific way she is embedded in the environment.16 But these differences at the level of content go no distance toward getting a hold on the first-personal ownership of perspective that is essential to perception: no array of different perceptual contents inflected by different orientations will mark one of them as mine. 15. See Making It Explicit, 236 and 243; see also the excellent discussion of this point in Joseph Rouse, How Scientific Practices Matter: Explaining Philosophical Naturalism (Chicago: University of Chicago Press, 2002), 216ff. 16. Brandom, Making It Explicit, 590. Rouse comments: “To talk about sameness of content [for Brandom] is thus to bracket the pervasive and ineliminable differences in conceptual perspective that result from the inferential significance of differing collateral commitments and different embodied locations. It might be more natural to say that, on Brandom’s account, one could only inherit a perspectivally shifted conceptual content from others’ observation reports.” How Scientific Practices Matter, 216–217.

Observatives and the Pragmatics of Perception

61

It is not enough to emphasize that our perceptual judgments are inflected and constituted by our practical bodily relationship to what we observe, although this is true too. This engaged body must furthermore be my body in order for its entitlements and commitments to have any normative bite, and no mere enrichment of bodily and social details will get this in. All these details, no matter how embodied, would not help me be gripped by my situation if they belonged to someone else. An account of perception as an assertion-like episode that is perspectivally marked only by its content is insufficient. Rather, perceptual episodes are first-personally structured, agent-relative events. Voice, as we are characterizing it, is a pragmatic rather than a grammatical feature of a speech act (though it is of course often closely tracked by grammar). The voice of a speech act concerns the manner in which the agent takes up her entitlement to the speech act and strives to assign statuses to others. We argue just below that the capacity of a language to express such perceptual episodes in a first-person voice is a constitutive condition for its existence. To put the point much more simply, any language must allow its users to articulate observatives in order for it to allow its users to articulate anything at all. But if this is right, then there could be no such thing as a language that traded only in impersonal, agent-neutral speech acts. The capacity of language users to express observatives is a constitutive condition for the possibility of their sharing a language that enables its users to pass around empirical truth-claims: no declaratives, then, without observatives. Hence any functioning language must include means for speaking in the first-person voice, and for allowing speakers to perform speech acts with agentrelative entitlements. If we are right, then any philosophical account that commits the declarative fallacy, and considers discourse only as a series of agent-neutral, declarative speech acts, will go wrong. Someone might argue, in contrast, that this feature of language is contingent. Even granting that we are right about the ineliminable agentrelativity of some of our speech, the objection would go, there could perfectly well be a coherent, functioning language that contained the pragmatic resources only for impersonal declarative speech. So, for example, imperatives are inherently second-personal: an imperative must be issued to someone in order for it to count as an imperative at all. The idea of “translating” an imperative into the third or first person while retain-

62

‘Yo!’ and ‘Lo!’

ing its meaning or force does not even get any conceptual traction. “Close the door!” makes a specific demand upon someone in particular (or upon several particular people) by addressing the target of this demand. A “translation” into the third person, such as “Mark ought to close the door,” is a speech act with an irreducibly different pragmatic structure and function; it does not constitute an order at all. And yet, one might acknowledge this but also think that a discourse without imperatives (not just without a distinctive imperative syntax, but without any pragmatic resources for making second-personal demands upon others) would be inconvenient, but not impossible. Later we argue in detail that second-personal speech is not eliminable in this way, and that the capacity to make second-personal demands in language is as fundamental as the capacity to assert truth-claims. But we will not take up that argument now. Instead, we will focus on the ineliminability of the first person. We can legitimately inherit entitlements to declaratives in any of several ways: by having them passed on to us from someone else, by having them follow inferentially from our other commitments and entitlements, or, crucially, through direct experience. However, to the extent that our discourse as a whole counts as about and accountable to the concrete empirical world—rather than just being an elaborate syntactic, non-referential game—our declarative entitlements must be traceable, through chains of entitlement, back to direct experiences, whether ours or someone else’s. As McDowell has made vivid, it is this termination of inference in receptive experience that gives our thinking and talking the external constraint that it needs to count as objective claim-making, as opposed to mere frictionless spinning in the void.17 This much seems fairly uncontentious. But here’s the point: the edifice of empirical knowledge, and with it our justified ability to make empirical assertions, depends upon there being chains of commitments and entitlements that terminate in someone’s first-personal experiences. The necessary termination of empirical claims in experience means that whenever we make an empirical assertion, we are committing ourselves to someone having had an experience—a receptive encounter with concrete features of the world—that grounds this assertion. (Again, we are certainly not saying that these chains must terminate in a preconceptual ‘given’. Experiences, 17. McDowell, Mind and World, 67 and elsewhere.

Observatives and the Pragmatics of Perception

63

as we have described them, are conceptually articulate affairs.) And although we needn’t have any particular idea of who served as the origin of this bit of empirical knowledge, we are committed to thinking it was someone (or several people) in particular who did. Now we have seen that these perceptual episodes themselves have an inherently individuating and first-personal input structure. They are not experiences that merely happen to be linked with some agent: if I am the original perceiver, then I must experience them as irreducibly mine, and they are not shareable. If there were nobody who could claim a piece of experiential knowledge from this first-personal perspective, then it would not count as knowledge that we share and can make assertions about at all. From an epistemological point of view, this means that empirical assertions can be justified only to the extent that their asserters have reason to be committed to the claim that these assertions terminate in some particular speaker’s perceptual episodes. But if we could never recognize, in language, other people’s discursive expressions of their own, first-personal experiences, then such expressions could not function as discursive reasons in the game of giving and asking for reasons. In that case, we would be left without any way to attach our edifice of declarative assertions to the empirical world about which it is supposed to make claims. In turn, this means that empirical discourse gets to count as properly open to justification only to the extent that it has the capacity to recognize discursive expressions of such first-personal experiences. To put the point another way, it is through observatives that individual experiences contribute to public discursive space. Observatives create new epistemic responsibilities for everyone by expanding the space of the known. Indeed, an entitled empirical claim can be thought of as one that can trace its warrant to the kind of recognitive episodes expressed in observatives. Although individual empirical claims can be warranted on grounds that don’t trace to an observative, such as abductive grounds, the totality of our empirical knowledge must all rest, globally and holistically, on observatives. Yet it is not enough that people be able to express the content of perception in language in order for our claim-making practices to be grounded in linguistically expressed reasons. We must be able to distinguish, within language, between those empirical claims that are merely inherited through the passing on of an inference or reassertion licenses, and those that function as the termina-

64

‘Yo!’ and ‘Lo!’

tion of a set of claims in someone’s receptive contact with the external world—between inferential commitments to the presence of a rabbit in a bush and first-personal observations of the rabbit, for instance. This is part and parcel of being able to critically assess empirical claims within the space of reasons. Otherwise, we might just have an edifice of circulating claims, without any linguistic means of accessing and marking the essentially first-personal point of receptive contact between language and the world. But this means that we must be able to use language to give first-personal expression to perceptual episodes. In other words, our language, insofar as it is used to make empirical claims, requires the capacity to utter recognitives, because recognitives are the speech acts that make explicit our first-personal experiential encounters with the world. It may seem that it is enough that we be able to declaratively report upon our experiences, rather than expressing them with a recognitive. But if a declarative such as “I see a rabbit” truly had no recognitive, pragmatically first-personal component, then in effect we would be taking a third-personal stance toward ourselves, and reporting on an experience “from the outside.” Such an assertion is the kind of thing that can be translated into the third person and reasserted—or in Brandomian terms, it issues a reassertion license. You can now assert, “Mark sees a rabbit,” and this (again, to the extent that it is a pure declarative) will be the same assertion. But if this were all we could do in language—that is, make assertions that were agent-neutral in this way—then making such assertions would not give expression to the termination of empirical knowledge in a first-personal, owned experience: an experience that is practically grasped as mine. To the extent that I give expression to that when I say “I see a rabbit!,” I am not merely making a declarative, agentneutral report but uttering a recognitive, at least implicitly. And it is because we can recognize this pragmatic move in discourse that we count as a discursive community whose members are speaking in a way that is held accountable to the world. But why, we might wonder, must we be able to give discursive expression to our recognitive episodes at all? Could we not anchor assertional entitlement in acts of perceiving without giving the recognitive aspect of that perception public expression? Not without relying upon the myth of the given that Sellars notoriously exposed in Empiricism and the Philosophy of Mind. Minimally, avoiding this classic Sellarsian demon means

Observatives and the Pragmatics of Perception

65

not treating any empirical epistemic accomplishment as in principle immune from discursive examination, challenge, and discussion. But if recognitive entitlement—the entitlement that comes from a receptive first-personal encounter with the empirical world—plays a distinctive role in justification, then this distinctiveness must itself be able to be brought into the space of explicit critical examination within the game of giving and asking for reasons. For instance, we must, at least sometimes, be able to challenge a claimed receptive entitlement. This requires that our language have the resources for making recognitive discourse explicit. We conclude that any satisfactory account of language that takes pragmatics as fundamental cannot be built upon narrow attention to assertions. Indeed, the very practices of passing around commitments and entitlements to assertions cannot themselves exist except in the context of a richer set of linguistic practices. Specifically, we conclude that any philosophical account of language that attends only to agent-neutral assertions that have no essential voice will be insufficient. Recognitive discourse is a kind of discourse that is voiced and agent-relative in its input, and it is constitutive of any language with the expressive capacity to make meaningful empirical assertions. The subject who participates in discourse concerning a public, empirical world is one who can speak in and recognize the entitlement of the first-person voice.18 18. We have only argued that a language expressing empirical content—a language responsive to an objective empirical world—must contain observatives. One might think there could be, say, a language of pure mathematics. We don’t think so for several reasons. For one, we are not convinced that any talk is genuinely contentful in holistic abstraction from empirical language. For another, we are not sure that there could be such a thing as justification in mathematics without the recognitive uptake of such facts as that this is a genuine proof.

3

The Pragmatic Structure of Objectivity

3.1 Observatives, Observation, and Answerability to the World At the beginning of this book, we claimed that what we were ultimately interested in was the pragmatic topography of the space of reasons, and we suggested that the space of reasons could have a structure no less rich than that of the space of discourse. Up to now we have kept our focus on language. However, we believe that our analysis of discursive pragmatics in general, and of the observative in particular, has fairly direct implications for analyses of reasoning and observation themselves. There is a tight relationship between reasons and claims. Spelling out an exact theory of this relationship would require a book unto itself, but surely something like this is true: both reasons and claims can have conceptual articulation and propositional structure, and both have normative force. Successful claims (whether truth-claims or claims upon our actions, loyalties, or whatever) give us reasons (to believe, act, etc.), and in turn, reasons make claims upon us. Spoken claims succeed in claiming only insofar as they provide reasons, and reasons are the kinds of things that we can express in claims. Perhaps not all reasons are of a sort that could be translated into discursive claims, but surely lots of them are, particularly including reasons for belief and inference. Many mental events that provide reasons, including observing, coming to believe, inferring, deciding, and so forth, are of a sort that can be expressed in discursive claims. Indeed, post-Kantian philosophers such as Sellars, Davidson, Brandom, and McDowell are quite comfortable understanding our capacity for conceptually articulate mental activity as 66

The Pragmatic Structure of Objectivity

67

essentially parasitic upon our capacity to engage in similarly conceptually articulated discourse. Some claims must be spoken in order to do the work they are supposed to do. There is no such thing as issuing a mental imperative, at least not to someone other than oneself. But it seems plausible to suppose that some reason-giving mental events can be understood as having a pragmatic structure that mirrors the structure that their expression would have. In these cases, we can use the same name for the mental event as for its spoken correlate. A declarative mental event would be an occurrent doxastic commitment to a declarative truth-claim, for instance. When I come to believe that there is a rabbit in the bush (as I often but not always do coincidentally with seeing the rabbit), I achieve a normative status that is implicitly agent-neutral. Even if I do not share my new belief with anyone, and hence don’t actually try to transform anyone else’s beliefs, I am still committed to a public truth whose normative force has nothing to do with me personally—a truth whose denial would be a mistake on anyone’s part. In other words, I engage in a mental act with a structure analogous to the act of declaring. At the end of the last chapter we argued that perceptual episodes— like the observatives that gave expression to them—have an inherently first-personal and agent-relative input structure. Our working hypothesis is that the activity of observation should be understood as having a pragmatic structure analogous to that of the observative that expresses it; more generally, that episodes of recognition should be understood as having a recognitive structure, as opposed to a declarative structure. We think that understanding observation as a normatively structured activity, which takes an agent-relative, individuating, first-personal input and achieves an agent-neutral status, can help clarify a set of important philosophical puzzles and issues concerning the epistemological status and function of this activity—that is, the activity of recognizing features of the world that are sensuously received through our encounter with them, in a way that grounds empirical belief and inference. If observations have a recognitive structure analogous to that of observatives, then we can understand them as fully discursively and conceptually articulated participants in the space of reasons, and yet as not having the structure of declarative propositions. On the one hand, observations and their reports have full-fledged rational relations to beliefs, which they can license or contradict. Observations and observative

68

‘Yo!’ and ‘Lo!’

speech acts are, on our account, properly governed by the so-called constitutive ideal of rationality1—they are held to the tribunal of the world and embedded in the discursive structure of the space of reasons. Yet at the same time, observations are not beliefs, and observatives are not belief-reports; observations are not themselves adoptions of commitments to declarative propositions, though they directly license such commitments. We all exhibit a solid practical skill at determining what follows rationally when we see a rabbit. To acknowledge this skill, we need not insist on interpreting this event of seeing as one of becoming committed to some particular declarative belief. Moreover, this skill is one that draws essentially upon the conceptually articulated structure of our observation—the relationship between my recognition, “A rabbit!”, and my belief that there is a rabbit in the bush is not mutely causal, but rather clearly and intimately linked to the way my recognitional episode invokes my use of discursive concepts such as ‘rabbit’. In Empiricism and the Philosophy of Mind, Sellars dealt the death blow (in our opinion) to the so-called myth of the given by showing that experiential states that are not themselves conceptually articulate cannot ground inference and hence cannot function articulately within the space of reasons. In the wake of this argument, many philosophers have treated experience insofar as it can justify belief as having the form of full-on propositional belief itself. For instance, Richard Manning claims that “it is irrational to draw inferences from what one does not believe.”2 Likewise, Davidson famously claimed, “Nothing can count as a reason for holding a belief except another belief.”3 Sellars himself encouraged this commitment; he takes his arguments against the possibility of a non-conceptually-articulate given that can ground inference as arguments for the claim that only something with propositional form can ground inference to a declarative proposition. Although our goal here is not to give a close reading of Sellars’s texts, we claim that none of his arguments against the given support this stronger conclusion. The famil1. See Donald Davidson, “Mental Events,” in Davidson, Essays on Actions and Events, 2nd ed. (Oxford: Clarendon Press, 2001). 2. Richard N. Manning, “Interpretations, Reasons and Facts,” Inquiry 46, no. 3 (2003): 346–376, 371. See also Barry Stroud, “Sense Experience and the Grounding of Thought,” in Nicholas H. Smith, ed., Reading McDowell on Mind and World (New York: Routledge, 2002). 3. Donald Davidson, “A Coherence Theory of Truth and Knowledge,” in Davidson, Subjective, Intersubjective, Objective (Oxford: Clarendon Press, 2001), 137–158.

The Pragmatic Structure of Objectivity

69

iar fact that practical reason always involves moves from the appropriateness of a claim to the appropriateness of an action should already cast into doubt the idea that the particular pragmatic and semantic form of a normative consequence must be the same as that of its antecedent. In casting doubt on the assumption that only commitments to declaratives can serve as reasons for commitments to declaratives (to translate Davidson’s dictum into language more helpful for us), we are merely opening a space for our argument about the structure of observation and its relationship to empirical knowledge. We are certainly not claiming that we have a full-blown theory of the rationality of such moves within the space of reasons—not that philosophers have offered much in the way of theories of material inference in any case. Despite this theoretical gap, we think that our pragmatic placement of observation reports helps us coherently occupy a much-sought-after philosophical ground between notoriously problematic positions. Philosophers who believe that only propositionally structured beliefs can serve as reasons for declarative claims face a dilemma: either the actual causal interactions between our sense organs and the world have no rational relation whatsoever to the kind of experience that is caught up in the space of reasons—in which case, as McDowell has often charged, our beliefs are left “spinning in the void” without making proper contact with the world that they are about—or we must say that somehow the world is itself already prepackaged in the form of propositions, and that perception is just a kind of reception of those propositions from the world. Perception, on this account, somehow involves the absorption from the world of fully formed declarative propositions, which seems to commit us to a level of rationalist excess that makes many of us uncomfortable.4 Several philosophers have tried to find a middle ground between “spinning in the void” and a propositionally prepackaged world by seeking some sort of intermediate status for perceptual episodes. Frequently this search turns into a quest to articulate a level of “nonconceptual content” taken in through perception.5 Such nonconceptual content is sup4. Indeed, in his repeated urging that the world is “what is the case,” McDowell comes close to this sort of position. See Manning, “Interpretations, Reasons, and Facts.” 5. See for instance Tim Crane, “The Non-Conceptual Content of Experience,” in Crane, ed., The Contents of Experience: Essays on Perception (Cambridge: Cambridge University Press, 1992), 136–157; D. W. Hamlyn, “Perception, Sensation and Non-Conceptual Content,” Philo-

70

‘Yo!’ and ‘Lo!’

posed to serve as the bridge between the causal impact of the world and the propositional attitudes, especially doxastic attitudes, which we form on the basis of this impact; it is meant to provide the transition into the space of reasons. As Michael Luntley puts it, “The idea of nonconceptual content is required in order to make sense of the thought that experience is an openness to the world; that the world is delivered in experience and thereby impinges on the operation of concepts within the space of reasons.”6 Here the content of perception is treated as transitional or “proto-conceptual,” making contact with our conceptual capacities but falling short of having conceptual structure. A major point to introducing such a layer of nonconceptual content is to take account of the rationally relevant and yet receptive character of perception—the sense in which it is “openness to the world”—that seems lost if we insist with Davidson that nothing could be a reason for a belief except another belief. Such accounts of nonconceptual content suffer from at least two problems, however. First, they are always at risk of raising the specter of a “third man.” Given that it is difficult to understand how conceptual judgment can be accountable to a world that is not itself already conceptually structured, it is equally difficult to understand how such conceptual judgment can be accountable to the nonconceptual contents of perception. For that matter, it is also unclear how causal impacts on our sense organs yield content that is “proto-conceptual.” Thus appeals to nonconceptual content seem simply to double the original explanatory conundrum.7 Second, these accounts generally leave open large quessophical Quarterly 44 (1994): 139–153; and Christopher Peacocke, “Does Perception Have a Nonconceptual Content?” Journal of Philosophy 98 (2001): 239–264. 6. Michael Luntley, abstract of “The World Delivered,” presented at the Space of Reasons Conference, Cape Town, South Africa, July 2004. 7. The person who most directly and extensively grappled with this specific type of thirdman problem was Kant, who posited schemata as the ‘bridge’ between the brute manifold of intuition and spontaneous discursive judgment in the Critique of Pure Reason, ed. Paul Guyer and Allen Wood (New York: Cambridge University Press, 1998). Where in the chain from the world to judgment Kant first wants to introduce conceptual structure, and whether he indeed sticks by his commitment to a level of content that serves as a bridge between the world and our judgments, are questions that receive vigorous debate in the secondary literature. See for example McDowell, “Having the World in View: Kant, Sellars, and Intentionality,” Journal of Philosophy 65 (1998): 431–450; and Richard N. Manning, “The Necessity of Receptivity,” in Rebecca Kukla, ed., Aesthetics and Cognition in Kant’s Critical Philosophy (Cambridge: Cambridge University Press, 2006), 61–84.

The Pragmatic Structure of Objectivity

71

tions about what kind of content they are pointing to. We know that it is not conceptual content, but since we use concepts whenever we talk, theorists of nonconceptual content understandably have a hard time explaining in any positive way what such content might be like. Thus these accounts are often unsatisfying in remaining almost exclusively negative. (The Stanford Encyclopedia of Philosophy opens its sympathetic entry on nonconceptual mental content by commenting, “The notion of nonconceptual content is fundamentally contrastive.”)8 Such accounts motivate themselves by correctly pointing at an important gap in our philosophical story of the path from the world to judgment, but they often do no more than posit a ‘something’ that will close that gap.9 McDowell has also tried to understand perception as ‘in between’ the brutely causal impact of the world on our sense organs and full-fledged doxastic judgment. For him, perception directly engages our conceptual faculties but falls short of belief. Like advocates of nonconceptual content, he too is motivated, not only by the gap that seemingly troubles Davidson and similar figures, but by the need to capture the receptive character of perception in contrast to the spontaneous character of committed judgment. However, for McDowell, perception is not protoconceptual but proto-doxastic. This is different from the appeal to nonconceptual content; his percepts are indeed conceptually articulated, and indeed propositionally structured. Thus he does not have a thirdman problem, in the sense that it is clear how such percepts hook up with the space of reasons. He holds on to the sensible idea that only conceptually articulated experience could bear rational relations to conceptually articulated beliefs, while getting rid of the assumption that everything conceptually articulated is a belief. We think that so far this is a promising approach to a solution to the problem of rational responsiveness to the world. Unfortunately, many commentators have been thwarted in their attempts to make sense of these quasi-doxastic percepts—these mysterious half-breeds—in McDowell.10 He often seems to just assert the coher8. José Bermúdez, plato.stanford.edu/entries/content-nonconceptual/, accessed 10/10/07. 9. For example see William P. Alston, “Sellars and ‘The Myth of the Given’,” Philosophy and Phenomenological Research 65 (2002): 69–86; and Alston, “Back to the Theory of Appearing,” Philosophical Perspectives 13 (1999): 181–203. 10. See for example Davidson, “Response to McDowell,” in The Philosophy of Donald Davidson, ed. Lewis E. Hahn (Chicago: Open Court, 1999); Manning, “Interpretations, Reasons, and

72

‘Yo!’ and ‘Lo!’

ence of such an option without giving a positive explanation of what perception is and how it works. Davidson is among those who have been frustrated by this move, writing that “McDowell holds that what is caused [by features of the world in perception] is not a belief, but a propositional attitude for which we have no word,” and that it is “entirely mysterious” what kind of ‘taking in’ of the world this constitutes. He complains that McDowell “gives no explanation of why features of the world cause the particular propositional attitude they do, nor of why an attitude which has no subjective probability whatever can provide a reason for a positive belief.”11 In return, McDowell has repeatedly punted the burden of proof.12 We agree that McDowell has not said enough about how perception works to satisfy those who cannot share his sanguine silence. However, we also think that the reason his proposal has seemed so baffling is that everyone, including McDowell himself, has presumed that if what is absorbed in perception is not a full declarative judgment, but still engages our concepts, then it must somehow be a proto-declarative. Indeed, like Sellars, McDowell often simply equates the propositional and the conceptual, presuming that insofar as our concepts are exercised in perception, the form that perception takes must be propositional form.13 McDowell interprets perception as having the form of a declarative judgment, only without the judging part. For McDowell, we perceive that the cat is on the mat, etc.: “That things are thus and so is the conceptual content of an experience.”14 Perceptual episodes must have propositional structure, and be “on their way” to being declarative commitments to that propositional structure. All we add, somehow, when we commit ourselves to our percepts in belief, is the commitment. But it is not clear how to understand a belief as decomposable into a propositionally structured part and a separate commitment part, and it is hard to understand what either of these parts could be like on its own without the other. This leaves it quite mysterious what kind of state McDowell thinks we are in when we have taken in something that parFacts”; Stroud, “Sense Experience and the Grounding of Thought”; and Brandom, “Placing McDowell’s Empiricism,” in Smith, ed., Reading McDowell. 11. Davidson, “Response to McDowell,” 107. 12. See for instance his responses to Brandom, Stroud, and others in Reading McDowell. 13. He has implied as much repeatedly, and he insisted on this particular commitment in conversation at the Space of Reasons Conference, Cape Town, South Africa, July 2004. 14. McDowell, Mind and World (Cambridge, Mass.: Harvard University Press, 1994), 26.

The Pragmatic Structure of Objectivity

73

ticipates in the space of reasons and has propositional and conceptual structure, but have not yet issued a judgment about it. One of Kant’s most powerful and influential moves in the Critique of Pure Reason was to understand concepts in terms of their possible roles in judgments, rather than understanding judgments as concatenations of concepts. His Table of Judgments gives the possible logical forms that the activity of propositional judging can take, and his Table of Categories is derived from the Table of Judgments. For Kant, empirical concepts are really rules for activities that essentially form a part of judgmental activities, which in turn take one of the forms outlined by the Table of Judgments. McDowell gives no hint that he wishes to depart from this core Kantian commitment to the priority of judgments over concepts. In fact this move is one of the banners of those who share a commitment to the usefulness of the space of reasons imagery, for all their differences. Yet McDowell offers us no tools for understanding what it means for the spontaneous activity of the concepts to be engaged in perception, without this activity constituting judgment. He does not provide any pragmatic story about what we can do with a proposition, in the course of our epistemic inquiries, other than either declaring it (in speech or thought) or refraining from declaring it, where the latter is only negatively defined.15 Once we cease to presuppose that everything that has any conceptual structure has the form of a declarative—that is, once we avoid this particular manifestation of the declarative fallacy—we open new room for making an account like McDowell’s satisfying. If we understand observations as having the pragmatic structure given expression in observatives, then we can hold on to several key results at once. Observatives directly express our receptive contact with the world, and yet they are thoroughly embedded within the rational, discursive, inferential structure of the space of reasons. Hence they can provide reasons for belief without being beliefs. Observations are, as McDowell insists, both conceptually structured and non-inferentially acquired, and thus they do 15. This is not to say that we do not accept a fundamental insight behind Kant’s understanding of concepts in terms of their possible role in judgments rather than judgments as a concatenation of concepts. The key here is Goethe’s point: in the beginning is the act. One cannot understand a conceptually significant action by putting it together with the right glue, taking concepts as functionally divorced from actions. The mistake is to assume that the only discursively significant type of action is judging.

74

‘Yo!’ and ‘Lo!’

not fall prey to the Sellarsian critique of the given. They are not the same as doxastic commitments, even if they generally accompany such commitments. To observe is not just to inherit entitlement to a belief, but rather to recognize how things show up to me. I recognize what I see with my concepts, and hence such recognitions already bear articulate rational relations to the rest of the space of reasons, including beliefs. (I see a rabbit as a rabbit, embedded within my web of beliefs about rabbits.) But my observation is not itself the production of a propositional judgment, nor does my expression of this observation in an observative give voice to one. (My observation has the form “A rabbit!,” is not grounded in but rather entails beliefs such as “There is a rabbit present.”) Unlike McDowell, we do not claim that observations somehow involve less commitment than do declaratively structured beliefs. So we can avoid the objection that there is no way that something to which we are not yet committed—for which we have “no subjective probability,” as Davidson put it—could provide a reason for a belief. Observations, on our account, are not beliefs that are missing something, but different pragmatic events that are not themselves declarative judgments, even though they commit us to truths. As they are for McDowell, observations are for us entries into declaratival judgment. However, their status as such entries derives not from their being mere “petitions for judgment,” as Brandom (glossing McDowell) puts it,16 but instead from the fact that they serve as the direct point of receptive contact between us and the world that is the tribunal of such judgments. Observations are not “on their way” to being beliefs, and they are no less firmly planted within the conceptually articulated space of reasons than are beliefs— they have a different pragmatic structure altogether. McDowell has accused Davidson, and by extension others who make beliefs the only ground for beliefs, of descending into a coherentism that leaves our beliefs without “friction” from the world that they are about and to which they are accountable. In granting the world only causal, arational efficacy in constituting our standing in the space of reasons, Davidson does not seem to allow any moment at which that world can show us what we must think about it. Davidson’s own insistence 16. Brandom, “Placing McDowell’s Empiricism,” 94–95. McDowell’s view here seems prima facie implausible. Perhaps when it merely appears to one that such and so we can see this as a sort of petition. But when one sees, one would think the petition had already been granted.

The Pragmatic Structure of Objectivity

75

upon the pivotal role of occasion sentences in anchoring interpretation, and with it the space of reasons as a whole, marks his own recognition of the necessary role of receptivity in accountable knowledge. But McDowell has plausibly argued that, if these receptive encounters are merely causal, they cannot serve to add the necessary friction. If the world merely causally produces full-fledged belief in us, then the causal origin of such belief in the world seems to be neither here nor there, epistemically speaking; the resulting beliefs, once we have them, seem indistinguishable from any other beliefs and without any special features that would let them serve as touchstones of empirical accountability. In order to avoid coherentism, we need not only receptive contact with the world, but special recognitive events: events that are conceptually structured and bear articulate rational relations to belief, but that bear their recognitive character on their sleeve, as part of their normative structure and not merely as their causal origin. This is the analogue, at the level of mental events, to the point we made about language at the end of the last chapter. There we pointed out that in order for observative expressions to play their necessary role in the discursive game of giving and asking for reasons, they had to express in language their distinctive, first-personal expressive structure. Here, the point is that in order for observations themselves to ground inference and play a role in rational judgment, they must likewise display their recognitive structure. We can put this McDowellian point in Kantian terms. Kant argued that our concepts, which we spontaneously apply in judgment, are “empty” without receptive intuition (just as intuition will be “blind” except as synthesized under concepts).17 But if there were nothing inside the space of reasons that marked when our conceptual judgments had directly receptive content, then the receptivity of intuition could play no rational role in filling in our concepts, which would in effect remain empty. The mere fact that the content of a conceptual judgment has its causal origin in the world does not make our receptive responsiveness to the world accessible to reason itself. What keeps our concepts from being empty is not just that we are impacted by the world, but that this impacting itself constitutes our recognizing when and how this is so. Hence this recognitive function of reason is essential to its empirical meaningfulness and accountability. 17. Kant, Critique of Pure Reason, A51/B75.

76

‘Yo!’ and ‘Lo!’

We saw earlier that this recognitive character cannot be borne by the semantics or the mere causal origin of a speech act or a propositional attitude, which is why Davidson’s assents to occasion sentences won’t do the trick. McDowell’s proto-doxastic perceptual takings-in can be read as his attempt to identify such a layer of recognitive events, and to demarcate them not by their semantics (which they share with the judgments they become) but by their pragmatics (they are missing spontaneous assent). Understanding this layer of recognitive events as observations with the normative structure of observatives—that is, with essentially singular and receptive agent-relative inputs, and producing agent-neutral outputs in the form of justifications for truth-claims and beliefs—does a better job of filling this McDowellian niche. Such events are essentially conceptually articulated and distinct from beliefs, and they bear their receptivity on their sleeve: their receptive connection to the encountered world plays a direct role in their pragmatic structure, rather than needing a separate account. Like McDowell’s takings-in, they are distinguished from beliefs by their pragmatics rather than their semantics. But we find it implausible to claim that observation doesn’t involve immediate (fallible, overridable) commitment to the truth or existence of its contents, and we think that McDowell has mislocated the pragmatic difference he is looking for, perhaps because of his failure to imagine a positive, non-declarative pragmatic structure for an event that grounds empirical judgment. Sellars was right to insist, in Empiricism and the Philosophy of Mind, that observations and observation reports “do not rest on other propositions in the same way as other propositions rest on them.”18 Observatives cannot, for instance, ever be inferences from declaratives (or from anything else); their receptive structure rules this out. But Sellars was also right that this does not imply that we can have them, or the knowledge they yield, prior to or independent of all sorts of declarative, propositional knowledge and commitments that we have in place. It is only once we inhabit the conceptually articulated space of reasons, which surely includes being properly in the grip of all sorts of propositional beliefs, that we can be concept-users of the sort whose normative entitle18. Empiricism and the Philosophy of Mind (Cambridge, Mass.: Harvard University Press, 1997), §38 (emphasis added).

The Pragmatic Structure of Objectivity

77

ments can be affected by interaction with the world, and hence who can issue observatives.19 The activity of observing, or perceiving, is a substantive skill.20 Though falling short of offering a full theory of perception, we can say a bit about this skill: an event occurs in the world and we perform an action of looking at that event or at the objects involved in it. As many neo-pragmatists have argued in detail, perceiving is no mere passive mirroring or being-pushed-about by the world. One deploys concepts, focuses attention, ignores much, highlights some, and articulates the result.21 But for all this complexity, observation is a matter of events in the world licensing the actions of epistemic agents. One’s entitlement to an observative arises, if it does, as a result of her normative achievement in the act of perceiving. Such a story requires that events in the world have normative significance. But there is nothing in principle puzzling about events in the natural world holding normative significance within our practices. That a natural resource is in a particular place means, in the right context, that I am entitled to take it; the violence of the storm constitutes a ground for not allowing my daughter to play outside; the edges of the pool table constitute the limits of a legal shot, etc. In all these cases, we enable features of the world to have normative significance—to matter to us in various ways—through our own engagement in normative practices that are essentially embedded in and responsible to various features of their environments. The normative involvement of the world in our practices is built into our running and driving on the world’s fields and 19. In Section I of Empiricism and the Philosophy of Mind, Sellars says: “I presume that no philosopher who has attacked the philosophical idea of givenness or, to use the Hegelian term, immediacy, has intended to deny that there is a difference between inferring that something is the case and, for example, seeing it to be the case. If the term ‘given’ referred merely to what is observed as being observed, or, perhaps, to a proper subset of the things we are said to determine by observation, the existence of ‘data’ would be as non-controversial as the existence of philosophical perplexities.” Sellars’s detailed discussion of the variety of types of interlinguistic dependence and his argument that these distinctions make it possible, without regress, to believe that observation reports are dependent for their authority on the existence of other warranted commitments occur in §§32–38. 20. For an excellent extended account of perception as a skill, which is reasonably compatible with our views, see Alva Noë, Action in Perception (Cambridge, Mass.: MIT Press, 2005). 21. For a nice discussion of this point see Joseph Rouse, How Scientific Practices Matter: Explaining Philosophical Naturalism (Chicago: University of Chicago Press, 2002).

78

‘Yo!’ and ‘Lo!’

roads, sailing on its oceans, trading its objects, and so forth. We use worldly objects directly in the practices to which they matter, including in the game of giving and asking for reasons. The rabbit darting into the bush, when related to my skillful perceptual activities in the right way, is what legitimates my belief that there is a rabbit in the bush. But if Davidson is right that only a belief can justify another belief, or more generally if the space of reasons is no more than the space of declaratives and their contents, then we have a serious problem—for the world does not have beliefs, nor make assertions, nor contain propositions, whatever metaphors we might like to invoke. However, by making room for observatives, we can claim that what we are open to, in observation, is the normatively significant events in the world that are caught up in our perceptual activities. Observations are the activities through which we engage with the elements of the world, the complex transitions from the competent and interactive moving about of our body— focusing eyes, picking things up, all the rest—to normative output. But if it is true that our perceptions are recognitions and the observatives that express them express recognitions, then we build receptivity right into the structure of observation. Perceptual episodes are different from judgments, but not because they are less conceptually articulated or inferentially fecund. Rather, they do something different: they take up, acknowledge, or recognize the normative significance of worldly events and objects. They put us into singular, first-personal receptive contact with the world and thereby render us answerable to it.

3.2 Intersubjectivity Consider what the structure of knowledge would look like if agentrelative entitlements could not give rise to agent-neutral commitments and entitlements. (In fact, it is impossible to properly imagine such a thing; some willing suspension of disbelief is required for this mental exercise.) In this case, Jones’s agent-relative entitlement to P, based in his observation that P, would have no normative implications for Smith’s relationship to P. Smith could know that Jones has this status, and happily go on either failing to believe P, or even believing not-P. Imagine Smith asserting not-P to Jones’s face. Jones replies: “But I am entitled to P.” Smith could then rightly return: “But this implies nothing for me.” To agree that no social constraint arises from an agent-relative entitlement

The Pragmatic Structure of Objectivity

79

would be to patently deny the socially shared nature of our epistemic projects. We might instead suppose that the only way in which other people’s observations can have normative epistemic consequences for me is by way of theoretical inference: Smith observes that Jones is entitled to P, and she believes that Jones is reliable. She concludes that P is true, and that is her reason to believe P. But this analysis makes no room for the crucial fact that, whatever other normative significance it has, Jones’s commitment to the truth of P has the import of standing in opposition to anyone’s not believing P, or believing not-P. Two agents in these positions are disagreeing, and argument or a change of belief is called for. For this reason, the beliefs of others cannot simply serve as ‘natural’ evidence for our own beliefs; rather, we must understand others’ beliefs as making a normative claim upon us. What one says, in saying P, is that anyone who will not accept P is wrong. It is precisely the agent-neutrality of beliefs that allows such disagreements to be a coherent possibility. Conversely, a practice in which agent-neutral entitlements were not holistically rooted in agent-relative entitlements would be equally distant from language as we find it. Insofar as we constitute a discursive community that empirically investigates a common world, we must meet at least two criteria: First, each of our individual observational episodes must have the sort of content that someone else could also take in through observation. Though my recognition is mine, essentially, and though what I recognize might be something that no one else ever in fact recognizes, if it is a genuine empirical observation it is of something that others could in principle observe as well.22 Second, our collective agentneutral entitlements must rest on a sufficient range of such reproducible observations. In particular cases we can come to know something on the basis of just one person’s having observed that it is the case. But the ability of such an unreproducible event to constitute agent-neutral knowledge depends upon the existence of a vast range of other knowledge 22. One might worry that “internal” mental states are an important counterexample. For just the sorts of reasons we are in the midst of discussing, we are committed to the view that we can indeed observe one another’s mental states. Of course, there are probably important qualitative differences between the way that Rebecca observes that Mark believes that the policies of the Bush administration are ill-considered, and the way that Mark observes that Mark believes this. However, we are happy to accept the weak behaviorist thesis that we have non-inferential access to one another’s mental states. See Rebecca Kukla, “How to Get an Interpretivist Committed,” Protosociology 14, (2000): 180–22, for a defense of this view.

80

‘Yo!’ and ‘Lo!’

based in reproducible empirical observation. Or, to put the point succinctly, our public body of empirical knowledge rests not just on a foundation of agent-neutral declaratives, but of widespread intersubjective agreement. Intersubjectivity is sometimes understood simply as universal agreement. But this kind of intersubjectivity cannot bear any interesting constitutive connection to objectivity—understood loosely, for the moment, as answerability to a public world—for at least two reasons. First of all, a claim can be objective even if not everyone assents to it—an objective claim must demand universal acceptance, but this demand will rarely be satisfied. Second, everyone may be wrong: we need to be able to forge a meaningful distance between what everyone agrees to and what is the case, in order for objectivity to have any bite at all. Surely, for example, debates about the objectivity of a rash of UFO sightings are not capable of being settled by a survey. In short, we want the relationship between objectivity and universal agreement to be structural and normative, rather than merely extensional. However, we have the resources now for a richer account of intersubjectivity, which shows more promise for playing a constitutive role in objectivity. Observative entitlements are intersubjective, not in the sense that everyone has them (not everyone does), but in the sense that they have essentially first-personal, singular, agent-relative inputs, yet their output is public, such that others can come to be entitled to an observative with the same content. Later we will see that not all recognitives share this intersubjectivity; it is distinctive of observatives, which voice recognition of features of the public world. This is a kind of intersubjectivity that, in our grid, has a distinctive box-2 pragmatic structure. There is a widespread sense that understanding objectivity as dependent upon intersubjectivity smacks of unsavory idealism or social constructivism. If intersubjectivity merely involves universal agreement, then this worry is justified. But the kind of intersubjectivity that we have built into the pragmatic possibility of objective claim-making—claimmaking that owes allegiance to the world—does not appear to raise any such concerns. Indeed, it earns its subjectivity precisely by involving a receptive encounter with the empirical world. We pointed out earlier that lo-claims are never pure observatives. ‘Lo!’ serves not only to mark my own recognition, but to ostend that which I recognize. “Lo, a rabbit!” involves deixis: in making such an utterance, I

The Pragmatic Structure of Objectivity

81

call upon or direct certain people—those to whom I am speaking—to observe the rabbit as well. The appropriate response to the utterance “Lo, a rabbit!” is not, then, merely to believe the consequent declarative, but to look and see the rabbit for yourself. Thus lo-claims call people into just those intersubjective practices of observation that constitute the necessary framework supporting declarative truth-claiming and epistemic inquiry. In a lo-claim, we explicitly mark the intersubjective character of observatives by calling others to shared attention to a public world. In Chapter 8 we return to this kind of intersubjective practice, and argue that our ability to use speech acts like lo-claims, which call others to direct their attention to a shared object, is essential to the constitution of discursive communities.

3.3 Objectivity Both declaratives and observatives must display fidelity to the objective world if they are to be legitimate: our declaratives must be answerable to this world, and our observatives must be responsive to it. The notion of objectivity, however, is one of the most slippery and most multivalent in the philosophical canon. Many philosophers have approached the philosophical problem of the nature of objectivity by offering a metaphysical account, in which they describe or enumerate the kinds of things that count as objective, and define objective claims as those that are about objective things.23 Such metaphysical accounts have generally taken one of two forms. Sometimes they have tried to distinguish levels of reality, in an effort to pin down the “really real.” So, perhaps, one might claim that values or so23. Such accounts show up in several philosophical domains. One version is the determined universal naturalistic reductionism of people like Peter Railton in ethics, David M. Armstrong in metaphysics and epistemology, and Hartry Field in science; see Railton, “Moral Realism,” Philosophical Review 95 (1986): 163–207; Armstrong, A Materialist Theory of Mind (New York: Routledge, 1993); Armstrong, Truth and Truthmakers (Cambridge: Cambridge University Press, 2004); and Field, Science without Numbers: A Defense of Nominalism (Princeton: Princeton University Press, 1980). Others take an eliminativist approach following on the example of Quine. Yet others acknowledge different metaphysical realms but assign them different grades of objectivity, either by claiming that there are different senses of truth applying to them, as Crispin Wright does in Truth and Objectivity (Cambridge, Mass.: Harvard University Press, 1992), or by simply offering alternative metaphysical accounts of the subject matter of non-objective discourse, as dualists do, for example David Chalmers in The Conscious Mind: In Search of a Fundamental Theory (New York: Oxford University Press, 1996).

82

‘Yo!’ and ‘Lo!’

cially emergent properties are not just figments of our imagination, but that they are not the “really real” stuff either. We will assert, dogmatically, that no one has been able to make this notion of the “really real” remotely coherent or compelling, except perhaps via eliminativist or reductionist moves. But if the really real is just the real, then the objective is just everything, and there is nothing subjective, and no interesting distinction has been drawn. Other times, metaphysical accounts identify the subjective with what’s “in the head” and the objective with what’s “out there.” This is a fair enough distinction, as long as we are willing to make all empirical psychology subjective and to admit that lots of things are neither objective nor subjective since they are not located at all (functions, waves of civic unrest, etc.). But it is certainly not satisfying as our only account of the distinction, for surely there is an important sense in which we can perfectly well make objective (universally valid, empirical, suitably independent, etc.) claims about psychological phenomena, and likewise various important senses in which we want to be able to challenge the objectivity of claims that have external objects as their topic. We propose to try to understand objectivity the other way around, by beginning with a pragmatic story about the nature of objective and subjective claims rather than with a metaphysical story about the nature of the referents of those claims. Lorraine Daston and Peter Galison have shown how our notion of objectivity actually incorporates various competing strands that do not fit together neatly, with different versions of objectivity having greater grip at different historical moments.24 We agree with Daston and Galison (though without committing ourselves here to her particular historical analysis) that our common philosophical notion of objective claim-making is neither neat nor unified, but rather is made up of sedimented layers of mismatched ideas. Pragmatic analysis turns out to be useful in sorting some of these out. Perhaps the most familiar first stab at a distinction between the two kinds of claims is the one still ringing in our ears from when we teach our introductory classes: an objective claim is supposed to be somehow “true for everyone” or “universally true,” whereas subjective claims are 24. See Lorraine Daston and Peter Galison, Objectivity (Cambridge, Mass.: Zone Books, 2007). See also Lorraine Daston, “Objectivity and the Escape from Perspective,” Social Studies of Science 22 (1992): 597–613; and Daston and Galison, “The Image of Objectivity,” Representations 40 (1992): 81–128.

The Pragmatic Structure of Objectivity

83

“true for me” or “true for you.” Along with this stab goes the idea that we can ask whether various domains of claims—especially moral or other normative claims—are objective and “universal” or subjective and just “true for the individual.” We have argued that truth-claims are by their very nature claims with agent-neutral outputs—that part of what is involved in making a truth-claim (and there are lots of other kinds of claims to make, as we have already seen) is making a claim that creates an agent-neutral entitlement, demanding of everyone, regardless of normative position, that he or she take up, reiterate, and use that claim. Hence we would argue that the notion of a claim that is only “true for the individual” is simply incoherent, and should be scrapped altogether. Truth by its nature claims us indiscriminately, and truth-claims build in a universal demand for acknowledgment, so if this is what we mean by objectivity (and it is one thing we could mean), then all truth is objective. The association of objective claims with agent-neutrality of output and subjective claims with agent-relativity of output is one elegant way of cashing out the intuition that the objective is the public: objective claims make normative statuses and entitlements available to everyone. Indeed, they ask, as a regulative ideal, that everyone take up these normative statuses and use these entitlements. All legitimate declaratives and observatives are objective, in this sense, which is also the sense in which they are about a public world—and notice that observatives have this publicity and “aboutness” despite the agent-relativity of their inputs. Imperatives, on the other hand, are not public in this sense. But we can equally articulate a different notion of objectivity that attaches it to agent-neutrality of input rather than output. In this sense, the objectivity of a claim resides in something like the democratic accessibility of its appropriate production—anyone has a claim on this claim, for the entitlement does not amount to any special normative feature of the speaker. Clearly some such notion is presupposed by the fundamental methodological assumption of science that all results are reproducible by any rational inquirer with the relevant equipment. In contrast, a speech act is ‘subjective’ if its entitlement is indexed to its speaker. In this case, the status of a speech act as objective or subjective does not turn on the truth being claimed, but on the sort of entitlement one can have to it. On this understanding, declaratives are again objective, but now observatives are not. Although observatives make a public truth ac-

84

‘Yo!’ and ‘Lo!’

cessible, and are therefore objective in the first sense, they do not do so by making their input entitlement publicly accessible. They are expressions of essentially private, first-personal recognitions that cannot, even in principle, be democratically shared. An observative is my speech act, my accomplishment, in a way that a declarative is not, and it is, in that sense, subjective. Similarly, the fact—expressed by a declarative—that two people are legally married is objective, while the performative—“I hereby pronounce you husband and wife”—is subjective, even though it licenses the former declarative. So far we have marshaled notions such as universality, publicity, and democratic accessibility in our discussion of different senses of objectivity. But equally time-honored is the attempt to understand objective claims as those that are accountable to, governed by, or responsive to the world (and indeed this is the prethematized notion of objectivity that we have appealed to in this book so far). Objective claims, we often say, express independent truths, or truths that are grounded in empirical experience, or held to the tribunal of the world, as McDowell would say. It is interesting to notice that this is prima facie a quite distinct sense of objectivity from either of the above. Although epistemology often presupposes that they go together, we would need to offer a specific argument to show that there is some essential link between agent-neutrality of either entitlement or import, on the one hand, and accountability to the world, on the other.25 Our conceptual framework allows us to give a succinct account of what it is for a claim to be empirical, or receptively responsive to the world: An empirical claim is one that can trace its warrant to or be invalidated by an observative. Although an individual empirical claim can be warranted without being traceable to an observative, the totality of our empirical knowledge must rest, globally and holistically, on observatives. Because we have left the scope of observatives open, this definition likewise leaves the scope of the empirical open. We do not come down on whether you can make empirical claims about values, meanings, abstract entities, or any of the many other domains that have come under contest in this regard. We see this as an advantage of our account. We are happy to say that we can define the notion of the empirical in advance of haggling over its particular contents. To associate objective claims with 25. Rebecca Kukla, in “Objectivity and Perspective in Empirical Knowledge,” Episteme 3 (2006): 80–95, makes the case against the existence of any such argument.

The Pragmatic Structure of Objectivity

85

empirical claims is to capture the intuition that objective claims are responsive to experience, which is another useful notion of objectivity. But the empiricality or responsiveness of a claim is not, we can now see, actually the same thing as the independence of the truth it asserts. We can say that the object of a claim is independent of that claim when there are no constitutive connections between the making of the claim and the correctness of what is claimed. Austinian performatives paradigmatically fail this particular independence test: in those cases it is the appropriate making of the claim that makes the claim true. More subtly, claims about social proprieties fail this particular independence test; while no one claim about a social propriety makes that social propriety hold or fail, it could not be the case that everyone’s claims about social proprieties were systematically wrong. There is a constitutive connection between the claims we make about social proprieties and the social proprieties themselves. Sometimes we understand a claim as objective insofar as its object is independent, in this sense of independence. There is a sense in which the rules of gravity are ‘objective’ while the rules of baseball are not. Hence we have found yet another locally reasonable sense of objectivity—and surely we could find yet more reasonable ways of defining objectivity, and independence too. Both objectivity-as-empiricality and objectivity-as-independence could be taken as (different) glosses on what we mean when we say that our objective claims “owe allegiance to the world,” as McDowell would say. McDowell takes care to provide a picture in which there is “external constraint” on thought. Yet the phrase “external constraint” seems to do double duty for him as a marker of both receptivity and independence. For instance, against those who would accuse him of an idealism that fails to give the world the proper independence, he argues that there is an inherently receptive character to the engagement of our spontaneous conceptual faculties. In order to counter the charge that we are condemned to “frictionless spinning in the void,” he reminds us that “to acknowledge the required external constraint, we need to appeal to receptivity.” Indeed, “The fact that experience is passive, a matter of receptivity in operation, should assure us that we have all the external constraint we can reasonably want. The constraint comes from outside thinking, but not from outside what is thinkable.”26 But there are, we can now see, many sorts of external constraint we 26. McDowell, Mind and World, 27–28, 50–51, 28.

86

‘Yo!’ and ‘Lo!’

could (and do) reasonably want. On the one hand, we can make the epistemological demand that our claims be justifiable in terms of empirical evidence. This is what McDowell apparently wants when he speaks of the essential role of receptivity in judgment. On the other hand, when he emphasizes the need for an external tribunal of our claims, he seems to be calling for some combination of the need for our claims to have universal output (and hence to be about a public world) and the need for them to be constitutively independent of linguistic propriety (and hence to be about an independent world). These are metaphysical rather than epistemological constraints. That Smith hit a home run yesterday can certainly be a deliverance of receptivity, but the existence of such deliverances will do nothing to assuage those social idealists who want to think of all of objective reality as constitutively dependent on social conventions in the way that this fact is. What McDowell really seems to be looking for is a multifaceted account of objectivity that illuminates how receptivity and independence are both compatible with the engagement of our spontaneous conceptual capacities in experience and judgment. Now there certainly are important categories of claims for which empiricality, intersubjective availability of warrant, universality of consequent, and objectivity-asindependence are ineliminably intertwined. As well, language and thought must be able to sustain both receptive, empirical claims and claims about independent states of affairs in order for any part of language or thinking to be contentful—and this is a point that McDowell has done a wonderful job of bringing home. But these dimensions of objectivity are nonetheless distinct, and play importantly different roles within the normative pragmatics of discourse. In this chapter we have explored a number of ways in which we distort the space of reasons if we understand it as in the first instance a space of inferentially articulated declarative claims. Instead, elements of this space have a rich and varied pragmatic structure. Authors like McDowell have worked hard to cure us of our attraction to any picture of the space of reasons that leaves it cut off from a world that can causally impinge upon us, deliver itself to us through our receptivity, or enjoy robust independence from our social norms. We suggest that the pragmatic framework that we have introduced can help alleviate the fear that we face a bridge between mind and world that philosophical explanation cannot cross.

4

Anticlimactic Interlude: Why Performatives Are Not That Important to Us

By ‘performatives’, we mean what Austin meant in the first part of How to Do Things with Words—that is, roughly, speech acts that in their very utterance serve to enact, institute, or make true what they assert. Consider, for example, a typical utterance of “The meeting is adjourned!” The fact that this speech act is performed—in the appropriate context, by someone of the relevant social position, etc.—constitutes the fact that the meeting is adjourned. There is no antecedent reality being described here. Rather, the speaker is creating the relevant reality through a discursive performance.1 Other examples that have become philosophical classics include acts of promising, baptizing, and marrying. Such performatives have been analytic philosophers’ favorite examples of utterances whose entitlement conditions are more normatively complex than the possession of epistemic warrant. The meaning of the term ‘performative’ notoriously drifts around in Austin’s classic work on the pragmatics of speech acts. In particular, he begins by defining ‘performatives’ as a pragmatic subclass of speech acts but ends by focusing on the fact that all speech acts have a performative dimension and force. Indeed, performative force has functioned as the 1. Of course on any account there will be matters of degree. A lower-court decision to the effect of “the law implies that such and so,” or a referee’s determination in a sport in which appeals are possible, does not in itself—even with the right background context—institute the truth. But it is nonetheless partially constitutive. It is, in some clear enough intuitive sense, productive of the truth that it asserts rather than merely reflective of some preexisting truth. Whatever other worries one has about the category, it is a simplification to think of the class of performatives as the class of utterances which by themselves constitute truths. The concerns we raise in this chapter apply independently of this complication.

87

88

‘Yo!’ and ‘Lo!’

main conceptual tool for categorizing speech acts by their pragmatic structure. Be that as it may, we reserve the term ‘Austinian performatives’ for the kind of speech acts that Austin marks out at the start of How to Do Things with Words, namely speech acts that constitute truths, not by causing a change in states of affairs, but by instituting new states of affairs in and through the very act of their utterance. Such speech acts—and the paradigmatic examples we have inherited from Austin, such as baptisms and commitment ceremonies—would appear on first blush to fit into box 2 of our grid. As Austin made clear, the entitlement to performatives depends on specificities of the normative position of their speaker, in the context of the social scenario in which the speech act occurs. Thus performatives seem to be classic examples of acts with agent-relative inputs. Only a religious or state authority can perform a marriage by pronouncing it; only the chair of a meeting can adjourn a meeting by asserting that it is adjourned. Meanwhile, it seems fairly natural to say that what these speech acts primarily do is to make true what they assert, and hence that they have agent-neutral outputs— in fact, it is this constitutive function that has drawn them together as a salient class of speech acts. Thus the functional effect of the utterance “The meeting is adjourned” seems to be to make what it pronounces— that the meeting is adjourned—true, not just for you, but simpliciter. Because they are so familiar to us as examples of speech acts with a different pragmatic structure from declaratives, one is tempted to think of them as the paradigmatic examples of box-2 speech acts. But there are at least two reasons why performatives aren’t as neat or illuminating as examples of box-2 speech acts as are observatives. First of all, as Austin makes explicit, performatives manage to function in the way that they do because of their function within social ritual or ceremony. A pronouncement of marriage does something only because we have established the social institution of marriage and a repertoire of ceremonies for initiating people into it, and what it does is establish a normative status defined within this social institution. Were there not established, conventionally agreed upon rituals and institutions of marriage, there would be nothing that it would be to be married, and nothing enacted by this speech act. Because performatives are so deeply dependent upon “convention” and “ceremony,” to use Austin’s favorite terms, they create a sense that such speech acts are somehow less

Anticlimactic Interlude

89

grounded in reality, less substantial, or less central to the core of epistemology and the functioning of language than are declaratives. Whether or not this intuition can be made precise while retaining plausibility, it is common in analytic philosophy to somehow feel that such speech acts are, if not outright gimmicks or tricks, at least not tied as firmly to the objective world as are “central” or “traditional” elements of language. Even though the truths performatives produce are agent-neutral, public truths, they are truths that are contingent upon our chosen conventions and practices in a special way—they seem to have “mere” conventions at their core. They are things we make true, not truths drawn forth by hard epistemic labor, or truths revealing the structure of independent reality. (Or, again, so we suspect the intuition runs.) Likewise, their distinctive pragmatic functioning is unlikely to unseat for us the intuition that basically, at its core, the pragmatics of declaratives is the pragmatics of language that we need to be worried about. One might well argue that no speech act—declarative, observative, or other—can in fact have any pragmatic function or performative force abstracted from the rituals and conventions that support the functioning of speech (even beyond the obviously conventional rules of syntax and semantics). But regardless of how successful such an argument would be if it were developed in detail, there seems to be no question but that the legitimacy of observatives is tied to objective facts that enjoy a kind of robust independence from social ceremonies and conventions of just the sort that performatives lack. Hence observatives cannot be written off as ‘gimmick sentences’, whose universal purport and status as truth-claims is dependent upon a trick of pragmatics. Observatives, which mark a receptive recognition, stand for their legitimacy before the tribunal of the independent world in as straightforward a way as any declarative; no plausible epistemology of our knowledge of the external world is going to be accomplished by assigning observation a second-class status. Attention to observatives, then, makes it clear(er) that what distinguishes box-2 from box-1 speech acts is not the distinction between noting and making, nor between objective truth and social convention, nor between receptivity and spontaneity, but rather the pragmatic structure of the entitlement itself. This point is merely pedagogical, of course. That they are less useful examples for the philosophical lessons we hope to draw from the func-

90

‘Yo!’ and ‘Lo!’

tion of box-2 structures is no point against the category of performatives per se, nor an argument that they are not genuine denizens of this box. But there is a second, deeper reason for rejecting performatives as the paradigmatic examples of box-2 speech acts. While it is true that performatives ‘make things true’, and in many cases that this making-true is their most characteristic performative function, it is not as easy to pin down the essential pragmatic output of a performative—that structure in virtue of which we classify it as a performative—as it is to pin down the output of a declarative, an imperative, or an observative. We have emphasized the fact that our boxes capture normative transitional structures, and that typical acts exhibit more than one such structure. And yet, declaratives, imperatives, and observatives are categories of speech acts brought together by a characteristic normative transition. Performatives do not seem to share any such single characteristic normative function. On the one hand, plenty of classic performatives arguably have agent-relative outputs that are more essential to their functioning than any making-true function. So, for instance, consider another Austinian chestnut, namely promising. When Rebecca says to Mark, “I promise to have a draft of this section done by Monday,” she indeed makes it true that a promise now has been made—the utterance does enact its own truth. But it is at least as central to the function of what she has done that she has undertaken an agent-relative commitment to Mark to produce the draft on time, and has entitled Mark specifically to expect her to do so and to hold her accountable for producing the draft. Nothing is a promise that doesn’t lead to (defeasible) obligations on the part of the promiser and entitlements on the part of the promisee, both of which are agent-relative normative outputs of the promise.2 2. It is essential to promises that they generate these particular agent-relative normative proprieties in a particular relation to one another. One has not understood promising if one merely understands that this person now is obliged to do x, and this other person entitled to hold them to doing x. These could be, as it were, free-standing norms. (That parents are obliged to feed their children and children are entitled to demand this of parents does not imply that a promise was made.) What makes something a promise is that the promiser is obliged to the promisee, and that the promisee is entitled, by and through recognizing the promise, to hold the promiser not merely to the act but to her promise. This dialogical, second-personal structure of promising goes missing in Gary Watson’s pragmatic analysis of the distinction between asserting (declaring) and promising, in which he explicitly analyzes promises as having (what we would call) an agent-neutral output; see Gary Watson, “Asserting and Promising,” Philosophical Studies 117 (2004): 57–77.

Anticlimactic Interlude

91

So such performatives arguably belong in box 4, along with imperatives, instead of in box 2. After all, all speech acts institute truths—if nothing else, the truth that they have been performed, and other inferentially and materially related truths. Hence, on a second pass, it seems that performatives do not form a category that maps neatly onto our grid, since some performatives reside most centrally in box 2 and some in box 4; performatives appear to be distinguished by the way they enact social statuses rather than by a distinctive pragmatic structure of the sort we are discussing. But once we notice this, we also notice that seemingly typical box-2 performatives, such as marriage pronouncements, have not only agentneutral but also agent-relative effects. In addition to making it true that two people are married, marriage pronouncements also impose clear, institutionally defined commitments and entitlements (as well as subtler, more implicit ones) upon these two people, as well as agent-relative requirements on others to recognize and respect the marriage in various, differential ways. (The normative force of the marriage for the spouses’ employers, who now have a new person to whom they owe benefits, is different from what it is for their parents, who now have a new member of the family, and so on.) Such agent-relative effects seem as essential to the function of the pronouncement as is the making-true of the public fact of marriage. Indeed, there wouldn’t be much point to all the fuss about marriage if engaging in it first and foremost added a truth to the universe. Likewise, box-4 performatives such as acts of promising produce agent-neutral outputs: not only do they make new truths, but, for instance, they make it the case that everyone is now entitled to treat the promiser as bound by her promise, even though the person(s) to whom the promise is made will also have a special entitlement to hold her to the promise. Thus it is starting to look as if all performatives, though uniformly agent-relative in input, will have a complex mix of agentrelative and agent-neutral outputs, and hence will not fit neatly into either box 2 or box 4. The best we can say, perhaps, is that a given performative may be primarily a box-2 or a box-4 utterance, depending on the centrality of its various outputs to its pragmatic functioning. Some performatives have no clear priority in either direction: arguably, it is equally central to the act of adjourning a meeting, for example, that we make it the case that

92

‘Yo!’ and ‘Lo!’

the meeting is over (an agent-neutral output), that no more motions can arise (also an agent-neutral output), that the chair no longer has the authority to call on speakers (an agent-relative output), and that the minute-taker need not continue to write down what people are saying (also an agent-relative output). All speech acts have some combination of agent-relative and agentneutral effects: Mark’s declaration that Emma gets out of school at 3:15 p.m. not only has agent-neutral outputs in virtue of being a truth-claim, but imposes an agent-relative burden on Rebecca to help finish this writing session in time for Mark to pick up Emma, and so on. Rebecca’s imperatival command that Mark revise his example has as a consequence the universally available truth that she commanded this, etc. In Chapter 1 we dealt with this problem by pointing out that we could distinguish, roughly but robustly, between the outputs that are part of the functional structure of a speech act and those that are accidental effects of it. So perhaps we should be no more disturbed by the apparently “mixed” character of performatives than we are in these other cases. The problem with performatives, though, is that the distinction between essential and accidental pragmatic effects does not seem to be nearly as intuitive or as vivid. It seems that performatives are especially messy in this way; that is, it is much harder than usual to distinguish which elements of their pragmatic functioning are essential to their structure and identity as speech acts. To return to the marriage example: Is it essential or accidental to the pragmatic functioning of the pronouncement of marriage that, for instance, the taxation status of partners now changes, or that the partners have new rights to make difficult decisions regarding one another’s health and welfare, or that the partners must be taken—by family members, and by the society at large—as genuine rights-and-responsibilities-bearing members of each other’s families? It is because the answers to these questions are not clear that there is so much to argue about in terms of what’s at stake in granting samesex couples the right to marry. Indeed, in the case of performatives, the pragmatic essences and consequences of speech acts are often up for active social negotiation.3 Perhaps there is a systematic reason for this inherent messiness, for it 3. See Judith Butler, Excitable Speech: A Politics of the Performative (New York: Routledge, 1997), for an excellent book-length philosophical analysis of such social negotiation.

Anticlimactic Interlude

93

seems that it is a fairly direct product of the dependence of performatives upon a wide variety of social conventions and ceremonies. A performative exists only insofar as it is placed as a move within a web of rituals and normative practices, and these webs are often thoroughly holistic, without neat centers or peripheries. There is no good answer to the question of how much of such a web must be in place in order to make possible a given type of speech act. Elaborate, multifaceted, politically charged institutions with vast numbers of implicit and explicit rules are just the sorts of things that support the possibility of performative speech acts, and they are also just the sorts of things whose boundaries, essences, and criteria of individuation are dynamic, contestable, and not determinate in advance of particular pragmatic and political struggles to establish or broaden them. There’s no fact of the matter as to which normative conditions, entitlements, and effects within such institutions are robust across context and which are changeable, prior to actual attempts to insist upon or change them. (Again, the current fight over the meaning and possibilities of same-sex marriages is an excellent case in point.) Thus it is in the nature of performatives not to be neatly parsable along the lines demarcated by our grid—and this is so because of the nature of social institutions and rituals themselves.4 The failure of these speech acts to fall neatly into one box or another is no criticism of the analytical framework we are advocating (nor, of course, is it a criticism of performatives themselves). We have already noted that individual speech acts can exhibit more than one of the normative pragmatic structures. In what follows we argue that all speech acts exhibit more than one. Thus, the point of the grid is not to sort speech acts into four mutually exclusive and exhaustive categories, but rather to isolate four distinct types of normative structure that can be present in, and constitutive of, a speech act. In some cases, one normative structure is more central to the nature of a particular act in particular ways, but others are always present. The case of Austinian performatives shows that in some utterances, multiple pragmatic normative 4. In contrast, many imperatives—“Shut the door,” “Please hand me the pen,” etc.—have well-defined functions that don’t rely essentially on the particularities of any such rich and contestable institutions, even though what they effect is a thoroughly social change in status. Likewise, observatives, like declaratives, function to note features of the world, and although they are enabled by various social rituals, this particular function is essential to them in a way that is robust across background social context.

94

‘Yo!’ and ‘Lo!’

dimensions are present, and there is no precise or determinate answer to the question of which are most central. But in every case, including the case of performatives, we can gain clarity about the nature of the speech act in question (in part) by becoming clear on which of these structures are present, in which specific ways, and in what relation to others.

5

Prescriptives and the Metaphysics of Ought-Claims

Let us return to our original typology. When we last left it, we had filled it in as indicated in Figure 5. In this chapter we will explore the remaining empty box, namely box 3. A box-3 speech act would have an agent-neutral input and an agentrelative output: it would draw upon a public entitlement in order to effect a normative transition in someone (or some group of people) with a particular normative status. As we mentioned in Chapter 1, we will argue that a paradigmatic type of deontic claim belongs in box 3, namely, a claim that prescribes an action for a person or a group of people, such as “Rebecca ought to hurry up and finish a draft of this chapter,” or “Canadian voters should send Stephen Harper packing in the next election.” In order to distinguish such claims from other types of deontic claims (such as ought-to-be claims: “There ought to be universal health insurance in the United States”), and for purposes of terminological elegance, we will call such claims “prescriptives.” Although we don’t really make use of this, we understand ‘prescribe’ here in a broadened sense: a prescriptive can also say something about what someone is entitled to do, etc. (“Mark and Rebecca are allowed to spend another month working on this chapter.”) A prescriptive is simply a speech act that attributes a deontic status to someone or some group of people. In this chapter, we explore the pragmatic structure of prescriptives, and we argue that they belong (partially) in box 3. In essence, we will argue that prescriptives are a species of truthclaim; they articulate truths about the deontic status of some agent or 95

96

‘Yo!’ and ‘Lo!’

agents (or perhaps even all agents), and in doing so they prescribe actions for those agents. We think that it is a strength of our account that this proposal sounds anticlimactic. Once we delve into the pragmatic structure of such truth-claims, however, we will show that this deceptively simple analysis enables us to solve—or, perhaps more accurately, to dissolve—some persistent puzzles that have troubled metaethicists. We end this chapter with an analysis of a philosophically important subcategory of prescriptives, namely, categorical imperatives. Of course, we are not the first to defend the idea that we can solve or dissolve key problems in metaethics by attending to the pragmatics of

Input Output

Agent-neutral

Agent-neutral

Agent-relative

1 Neutral input Neutral output

2 Relative input Neutral output

Declaratives

3 Neutral input Relative output Agent-relative

Figure 5

Kantian judgments of taste, baptisms, some recognitives, i.e. observatives

4 Relative input Relative output Imperatives (promises, invitations, reproaches...), ostensions

Prescriptives and the Metaphysics of Ought-Claims

97

deontic claims, and in particular to their prescribing function. Indeed, this is the core tenet of ‘prescriptivism’, one of the time-honored metaethical positions. Prescriptivists (whom we will discuss in detail later in this chapter) share with us a central tenet: many of the problems that have appeared intractable in metaethics originate from the undefended assumption that deontic claims should be understood as declarative utterances, and that their philosophical distinctiveness is located entirely in their semantics; in other words, metaethicists have been crippled by the declarative fallacy. In contrast, prescriptivists propose—as we do— that we can only understand deontic claims by considering their pragmatic structure. Furthermore, as we will see, prescriptivists agree with us that part of what is distinctive about this pragmatic structure is its agent-relative component: prescriptives prescribe an action for someone, and hence surely their pragmatic import for that person is different from what it is for other people. Declarative claims have no analogous agentrelative import; therefore, prescriptivists conclude, prescriptives must not be declaratives. However, we will argue that classical prescriptivism has been an ultimately unsuccessful research program because the only framework available to the philosophical imagination for understanding a speech act as non-declarative and agent-relative has been that provided by imperative speech. Accordingly, prescriptivists have more or less identified prescriptives with funny kinds of imperatives. We will argue that this equation is unviable. Prescriptives are truth-claims, and not imperatives. But their distinctive structure could not be seen clearly until the distinction between inputs and outputs was articulated. With our typology in place, we can both accommodate the insights of prescriptivism and avoid its fatal pitfalls.

5.1 The Pragmatics of Prescriptives Our proposal, again, is to understand prescriptives as truth-claims that are about the commitments of those to whom they prescribe actions. That is, when I say, “Stephen Harper ought to replace his minister of the environment,” I am claiming that it is true of Stephen Harper that he has a commitment—whether or not he recognizes this commitment—to replace his minister of the environment. This commitment is among the

98

‘Yo!’ and ‘Lo!’

normative statuses that carve out his position in the space of reasons.1 My claim is true just in case Harper really has this commitment, and whether he does or not is a fact about the structure of the public world. The intuitive, commonsense appeal of this analysis is so strong that we do not think we need to defend it, as long as we can show that it yields helpful payoffs. Indeed, we think that the only reason to deny this obvious reading—and to produce the rather tortured accounts of oughttalk that have shaped much of the metaethics literature2—is if one believes, as many have, that any account that analyzes deontic claims as truth-claims is hopeless from the get-go. There have been two reasons, traditionally, why philosophers have thought this. The first is most classically and starkly summed up in J. L. Mackie’s “argument from queerness”: Mackie associates the idea that moral claims are truth-claims with the requirement that there be ‘objective values’ about which the truth-claims are made. But, he argues, “if there were objective values, then they would be entities or qualities or relations of a very strange sort, utterly different from anything else in the universe.”3 The second is that it has seemed to many that, if prescriptives were truth-claims, then they would be paradoxically severed from motivation. According to most philosophers, beliefs are distinct from desires, or motivating states, and truth-claims are the kinds of things we believe. Hence if prescriptives are truth-claims, it seems that we can believe them to be true without necessarily being motivated to act as they prescribe— we could, for instance, believe that it’s true that we ought not to swindle little old ladies out of their life savings, without this belief giving us any motivating reason not to do so. But this has seemed to misconstrue the essence of such claims, which appear to be inherently motivating.4 It is 1. Notice that our proposal draws no distinction between moral oughts, instrumental oughts, and any other kind that may come along; the distinctiveness of moral ought talk is mostly beyond the bounds of this book, although as we mentioned we will come to categorical imperatives later. Moral ought-claims, on our reading, are claims about moral commitments— about what it is morally good or right to do, whatever that turns out to mean. 2. For instance, see R. M. Hare, The Language of Morals (Oxford: Oxford University Press, 1968); and J. E. J. Altham, “The Legacy of Emotivism,” in Fact, Science, and Morality, ed. Graham MacDonald and Crispin Wright (Oxford: Blackwell, 1987). 3. J. L. Mackie, Ethics: Inventing Right and Wrong (New York: Penguin, 1977), 38. 4. Many authors have taken up this worry, which we discuss in detail later in this chapter. Important examples include Michael Smith, The Moral Problem (Oxford: Blackwell, 1994), and David McNaughton, Moral Vision (Oxford: Blackwell, 1989). Philippa Foot embraced the sev-

Prescriptives and the Metaphysics of Ought-Claims

99

precisely this problem that has led philosophers to seek to understand moral language as having a pragmatic form other than that of a truthclaim, and has given birth to expressivism, prescriptivism, and kindred positions. The problem of motivation is highly relevant for us, and we return to it later in this chapter. We believe that one of the most powerful payoffs of our conceptual apparatus is that it allows us to solve this metaethical dilemma, and show how a claim can both be a truth-claim and essentially provide reasons for action. In contrast, we can quickly dispense with Mackie’s version of the problem of queerness. In order for a claim to have a truth-value, it must be able to correspond, or fail to correspond, to some objective state of affairs in the public world. But Mackie’s argument presupposes that, in order for this to be so, the words in the claim must correspond to object-like entities in the world. This is the persistent and normally unarticulated philosophical intuition that Derrida, reading Heidegger, has termed the “metaphysics of presence.”5 But one need not delve into the moral domain to see that this assumption is ungrounded. We make perfectly reasonable truth-claims all the time about events and states of affairs that cannot easily be reduced to any component object-like entities that serve as the referents of our words: we speak about intentions, elections, waves of civic unrest, holidays, selection pressures, etc. Unless Mackie is promoting a radical reductionist revision of language that would deny truth-claim status to most of our folk psychological, scientific, and everyday talk along with our moral talk (which he never suggests), there seems to be no reason at all to assume that values in particular must correspond to object-like entities if we are to be able to make truth-claims about them. One might grant this general point and still be impressed by the argument from queerness if one thinks that it is particularly hard to see how deontic claims could describe nice, regular (though perhaps not radically naturalizable) states of affairs—if one feels that these states of affairs would need to have ‘spooky’ properties. As Mackie puts it, “Plato’s forms give a dramatic picture of what objective values would have to erance of moral belief from moral motivation in her classic “Morality as a System of Hypothetical Imperatives,” Philosophical Review 81 (1972): 305–316. 5. Jacques Derrida, Of Grammatology, trans. Gayatri C. Spivak (Baltimore: Johns Hopkins University Press, 1974).

100

‘Yo!’ and ‘Lo!’

be. The Form of the Good is such that knowledge of it provides the knower with both a direction and an overriding motive . . . similarly, if there were objective principles of right and wrong, any wrong (possible) course of action would have not-to-be-doneness somehow built into it.”6 But notice that our account already manages to avoid such worries. If prescriptives are about normative statuses such as commitments, then they require a metaphysics no richer or more mysterious than that required by any of our other claims about normative statuses, including our claims about what we are committed or entitled to believe or infer. Indeed, our everyday ontology is riddled with normative statuses; we appeal to them when we talk about legal contracts, the implications of a scientific result, the structure of a philosophical argument, and so forth. While there may be some who still dream of the eventual reduction of all normative talk to ‘naturalized’ talk, our equation of prescriptives with truth-claims makes them rest on entities and states no spookier or more immune from naturalization than the rest of such talk. And commitments and entitlements—the topic of prescriptives—do in some obvious but non-pernicious sense ‘have not-to-be-doneness (and to-bedoneness, and allowed-to-be-doneness) built into them’: this is just what a commitment is—it is what-is-to-be-done. Its to-be-doneness does not need to be understood as some extra property clinging to it, some sticky motivational jelly coating it; to think this would be to conceive commitments as objects differentiated by their predicates, rather than as states of affairs, and thereby to succumb, again, to the metaphysics of presence. Therefore there is no metaphysical barrier to our treating prescriptives as truth-claims.7 Thus the ground is cleared for us to begin our pragmatic analysis of these claims and their distinctive function. In Chapter 1 we argued that truth-claims are entitled by access to features of the public, shared world, and hence that their input is necessarily agent-neutral. Nothing can be true “for me” but not “for you”; this is essential to truth and to our ability to reason and communicate about a shared world. Prescriptives are no exception; if it is true that Stephen 6. Mackie, Ethics, 40. 7. Notice also that we can remain completely neutral on questions such as whether normative statuses are ‘socially constructed’. Our ought-claims have unproblematic references insofar as there really are normative statuses, regardless of how they got there.

Prescriptives and the Metaphysics of Ought-Claims

101

Harper ought to replace his minister of the environment, then it is true “for everyone” that this is so, and one’s entitlement to the claim is constituted by access to facts about the world, rather than on specific features of one’s normative standing. So prescriptives, like all truth-claims, have agent-neutral inputs. This is not to be confused with the thesis, defended by Hare and others, that oughts are based on universal principles. The reasons for an ought-claim may be as particularist and contextspecific as we like, but if it is a true claim, then it describes a public fact, and entitlement to it is agent-neutral. We also argued in Chapter 1 that the utterance of a legitimate truthclaim, by putting this agent-neutral fact into public social space, makes it a defect in anyone else in the discursive community not to share in this entitlement, albeit perhaps a completely exculpable defect of ignorance; such a person does not know what ‘we’ know. Again, prescriptives are no exception: like all truth-claims, they strive for universal recognition and uptake, and to the extent that they are legitimate, it is a defect not to give them this uptake. Since this effect is agent-neutral, we can say that prescriptives always and essentially have an important agentneutral output. At least one of their central functions belongs in box 1. Indeed, this all amounts to saying that prescriptives serve a perfectly healthy declarative function. However—and here is where things get interesting—they have another essential output as well. For when failing to acknowledge a commitment becomes a defect, this is a very different transformation for the person whose commitment this is than it is for everyone else. There is a world of difference between recognizing that Stephen Harper is committed to x (or ought to do x) and recognizing that I (who am thankfully not Stephen Harper) am committed to x (or ought to do x). Recognizing a truth about (someone else’s) commitments is like recognizing any other fact; it commits us to taking account of this fact in our beliefs, inferences, and behavior. We give proper uptake to this commitment by acknowledging this truth. But recognizing that I, first-personally, am committed to x is essentially a matter of practically acknowledging that I am bound to do x. Of course I still may not want to do x, or my reasons to do x may be trumped by reasons to do conflicting things, and so forth. But I literally have not understood the force of the truth-claim if I do not give uptake to the fact that I am committed to doing x. If I acknowledge the truth of a prescriptive that is targeted at me, but somehow deny that

102

‘Yo!’ and ‘Lo!’

it makes any claim on me to follow its prescription for action, then either I don’t understand what I am acknowledging, or I don’t identify, firstpersonally, with the person whose commitment it is.8 John Perry’s famous argument for the essential indexical showed that no set of categorical propositions could capture the practical inferential content of “I am here”—a content without which no knowledge is deployable.9 Likewise, no set of categorical propositions can capture our placement within the space of reasons, as bearers of particular commitments and entitlements. In order for my commitments and entitlements to exert governing force upon my practices, I must recognize that these are mine. I must grasp not only the shape of the normative web of commitments and entitlements, and not only the ways in which new speech acts change this web, but also where I am located within it—I must know which commitments and entitlements are mine and which new claims demand uptake from me. This perspectival grasp of the space of reasons is a logical condition for any of these statuses making a difference to me at all. I must not only recognize my commitments and entitlements, but also have practical, perspectival uptake of the fact that they are mine—that they commit and entitle me. When language is functioning ideally, then, the person whose commitments are targeted by a prescriptive will undergo an agent-relative transformation that others are not called upon to undergo: she will recognize herself as bound to do what the prescriptive prescribes (whether that recognition takes the form of doing it, making excuses for not doing it, feeling guilty for not doing it, etc.). No one else but she is called upon to do this by the prescriptive. The prescriptive calls to everyone, agentneutrally, to recognize the truth of the claim it makes, but it also calls to her to give first-personal practical recognition of the claims her commitments make upon her. This is an agent-relative output of the prescriptive, whose utterance is still agent-neutrally entitled. While only I am practically claimed by my commitments, anyone who has access to the truth is entitled to note them. Thus this function of the prescriptive has an agent-neutral input and an agent-relative output, and it belongs in box 3. Hence our grid now appears as in Figure 6. 8. For instance, I might acknowledge the truth of the claim that anyone who has a drinking problem should seek professional help, but not recognize that it is really I who have a drinking problem. 9. John Perry, “The Problem of the Essential Indexical,” Noûs 12 (1979): 3–21.

Prescriptives and the Metaphysics of Ought-Claims

103

Furthermore, the prescriptive always has this distinctive output for those whose normative status it describes. For as we saw in Chapter 1, truth-claims seek, in the ideal, to receive uptake from everyone. But this will always include the person whose commitments the prescriptive describes, and it will always be the case that appropriate uptake, in this person’s case, takes this special form. We have said that uttering a prescriptive turns it into a defect not to recognize the relevant commitments. Accordingly, when it is your commitments that are being described, it becomes a defect for you not to give recognitive uptake to the commitments you have. But this is a very different kind of defect from the kind that others suffer if they merely fail to acknowledge a truth-

Input Output

Agent-neutral

Agent-neutral

Agent-relative

1 Neutral input Neutral output

2 Relative input Neutral output

Declaratives

3 Neutral input Relative output Agent-relative

Figure 6

Prescriptives

Kantian judgments of taste, baptisms, some recognitives, i.e. observatives

4 Relative input Relative output Imperatives (promises, invitations, reproaches...), ostensions

104

‘Yo!’ and ‘Lo!’

claim. Others suffer an epistemic defect, and lack entitlement to beliefs to which they would ideally be entitled. You, on the other hand, suffer a practical defect: your actions are not governed in the right way by the commitments you have, because you fail to recognize these commitments. Merely uttering a prescriptive does not generally give someone a commitment that he did not already have. Normally, whether Mark’s commitment to x is a moral commitment to act rightly or a prudential commitment to act sensibly, the true claim that Mark ought to x reflects rather than constitutes the commitment he already has. We have all sorts of commitments and entitlements that we do not know about, and our not knowing about them does not lessen their reality, any more than it lessens my entitlement to the money left to me in a relative’s will if I don’t know about it. And here we can note a remarkable disanalogy between doxastic and practical normative statuses: if I do not know of good reasons for believing a claim, I am not committed or entitled to that belief, and this is so even if I am lacking those reasons only because of a defect, such as ignorance. Consider the claim that mitosis is a form of reproduction. The claim is true, and I ought to have reasons to believe it. It’s common knowledge, after all. But if I have managed to remain ignorant of these reasons, then I am not committed to the belief. It’s not as though I have the doxastic commitment sitting inside me but just don’t know about it—rather, I get the commitment (and the entitlement) once I grasp the reasons.10 Epistemic commitments are produced by epistemic events, such as discovering reasons. Although we could probably invent recherché Austinian counterexamples, practical commitments and entitlements generally don’t work this way. I have them regardless of my epistemic stance toward them. This gives us another way of getting at the difference between the agent-neutral and the agent-relative outputs of a prescriptive. Imagine that I say, truthfully, “Stephen Harper ought to replace his minister of the environment,” and imagine also that disturbingly few people hear and accept my claim, despite my backing it up with solid arguments and evidence. It now becomes a defect not to acknowledge that Stephen Harper 10. One reason that Brandom has confused some readers is that he often speaks as if inferential entitlements are automatically transmitted when someone makes a claim, even to people who have no way of knowing about the claim. But we have not been using the term this way in this book.

Prescriptives and the Metaphysics of Ought-Claims

105

ought to replace his minister of the environment. What sort of defect is this? For all persons other than Stephen Harper who fail to accept my claim, they fail to have a commitment that they should have. They ought to be committed to this claim, but they simply are not. The commitment is missing. Likewise, given their recalcitrance, they would be unjustified in using the claim in inference or acting on it in any way. Stephen Harper, in contrast, is defective in failing to give first-personal, practical recognition to a commitment that he really does have. As long as he does not fire his minister, he is falling short of living up to one of his practical commitments, whether or not he recognizes that he has this commitment. The prescriptive thus seeks to produce (doxastic) commitments or entitlements in third-party listeners, but to induce recognition of (practical) normative statuses in those to whom it applies.11 (Normally, as we discussed in Chapter 1, the form that the recognition of a practical commitment will take is simply performance of the action we are committed to performing. But this is not always the case: you might recognize that you have a commitment but also that you have a stronger one that conflicts with it, etc.) This difference between the output of prescriptives and that of other declaratives underscores the prescriptive’s box-3 status.

5.2 Four Ways of Telling Someone What to Do We have seen that prescriptives always have both agent-neutral and agent-relative outputs; they necessarily enact both box-1 and box-3 functions. Their box-3 function is to call upon someone to recognize her commitments, which involves practically, first-personally recognizing the normative claim they make on her actions. There are in fact several different ways in which you can be discursively called upon to act on your commitments. Consider first the distinction between prescriptives that are spoken in the third and in the second person, at the level of their surface grammar. For instance, consider the difference between the claims “Scott ought to lose some weight” and “You ought to lose some weight” (said to Scott). In the first case, I might utter the claim with no expectation that Scott 11. If, on the other hand, the prescriptive was about theoretical commitments to start with—i.e., if someone told me, “You ought to know that mitosis is a form of reproduction”— then the prescriptive might both create the commitment and demand recognition of it at the same time.

106

‘Yo!’ and ‘Lo!’

will ever know I did so. In this case, presumably, the primary purpose of the speech act is simply to declare a fact about what Scott ought to do. I hope to convince the person to whom I am speaking (presumably not Scott) that something is true of Scott. Now because it is in the nature of truth-claims to have agent-neutral outputs, and to strive, in the ideal, for universal uptake, my utterance also seeks uptake from Scott, in some attenuated sense. Furthermore, should my claim reach Scott, and should he acknowledge its truth, the shift in his normative status would be quite different from the shift in my conversational partner’s normative status, for the reasons we have seen: acknowledgment of this truth on Scott’s part would require a firstpersonal, practical recognition of his commitment to lose weight and its claim upon his actions. (Colloquially, he would have to recognize, “That’s me they are talking about! I’m the one who has to lose weight.”) Hence my third-personal utterance has both agent-neutral and agentrelative outputs. However, at a concrete communicative level, the importance of its box-1, declarative function far outstrips that of its peripheral (yet ineliminable) box-3 function. Although my claim has special normative implications for Scott, it is a stretch here to say that I am telling Scott what to do. Let’s turn to the second case. This utterance, “You should lose some weight,” makes no sense unless I am speaking to Scott and expect him to hear my claim. I am still making a truth-claim—I am purporting to say something true about Scott’s commitments—and so for Scott to accept my claim, he must take up an agent-neutral commitment to this truth. In this, he is like any other listener. Yet it is clear that in this case the primary social purpose of my utterance is not to convince Scott of the agent-neutral truth of this claim, but to make him recognize the distinctive weight (so to speak) that it has for him. I seek to induce in him practical, first-personal uptake of the force of his commitment. In this case, the box-3 function of the prescriptive takes center stage. In principle, this is not a deep difference. Both utterances are prescriptives with their characteristic box-1 and box-3 functions; the difference is one of social emphasis and expectations. But the second-personal prescriptives seem (in most cases) to share an ostensive function with the ‘Lo!’ utterances of Chapter 2 (“Lo, a rabbit!,” etc.). Remember that in such speech acts we not only utter an observative that gives recognitive expression to our first-personal experience, but we also ostend

Prescriptives and the Metaphysics of Ought-Claims

107

that which we observe. While observatives, we said, have agent-neutral outputs, the ostensive function that often accompanies them has an agent-relative output; it can be understood only as directed at those who are in a position to appropriately direct their attention, and it asks nothing of other people, even in the ideal. In the case of second-personal prescriptives, we rarely utter them for the mere purpose of declaring a truth, which incidentally happens to be about the person we are talking to.12 Rather, we speak in order to call that person’s attention to the norm that binds him. If I tell Scott, “You need to lose weight,” I do not order him to lose weight, but neither do I merely inform him of a truth—I ostend his normative status, which will make a practical claim on his actions once he recognizes it. So whereas my third-personal prescriptive had agent-relative normative implications for Scott, my second-personal prescriptive further demands of him that he attend to and appropriately recognize his normative situation. There is a subtle but from our point of view crucial distinction between such exhibitions of a norm, on the one hand, and the attempt to hold someone to following the norm, on the other hand. Surface grammar is exceptionally slippery and unreliable here, but in the most literal cases, “You ought to lose weight” is importantly different in its discursive function from “Please lose weight!,” which is our third and most direct way of telling someone what to do. The latter is a box-4 speech act such as an entreaty or an imperative, and it holds the other to a norm rather than merely ostending that norm.13 On our account, such a speech act had better be different from the first, since the first, being a prescriptive, belongs in box 3, and the second belongs in box 4. Since “You ought to lose weight” is a truth-claim, anyone who has access to the truth that Scott ought to lose weight is entitled to utter it to Scott. In other words, its input is agent-neutral. It may be very rude for various people to tell Scott this—as we discussed in Chapter 1, there are various levels of social normativity, and not every utterance that is dis12. When people excuse giving all sorts of tactless advice by saying “I am just telling the truth,” we take them to be either lying or lacking a fairly basic understanding of conversational pragmatics. 13. In Chapter 1 we pointed out that imperatives are not the only box-4 speech acts through which we make agent-relative claims upon one another: we can also invite, promise, entreat, etc. What is important for us here, however, is the distinction between agent-neutrally entitled exhibitions of a norm, on the one hand, and second-personal, agent-relative holdings, on the other.

108

‘Yo!’ and ‘Lo!’

cursively entitled is socially appropriate. However, qua truth-claim, I will be discursively entitled to this utterance as long as I am entitled to believe its truth. While I may well inappropriately hurt Scott’s feelings by calling his attention to this truth, or violate other social norms governing speech, my utterance cannot be accused of pragmatic misfire at the level of its structural discursive function. On the other hand, I may have no standing whatsoever to tell Scott to lose weight, and hence the utterance, “Please lose weight!” may well pragmatically misfire, no matter how politely I say it. It is contestable who has the standing to ask Scott to lose weight—his doctor and his spouse, probably, and perhaps his children. In the United States, some employers have taken themselves to be so entitled. But in any case, such an utterance, like all box-4 speech acts, is grounded in an agent-relative entitlement. I must be someone with the right sort of authority and standing in order to be entitled to ask this of Scott. My knowing that he should lose weight and my wishing that he would, even for his own sake, is not enough. This request not only directs Scott’s attention to his normative commitment, but it seeks to hold him to that commitment, and only someone with the right kind of normative relationship to Scott has any status as an entitled holder of this sort. In both cases, I say something in the hope that it will make Scott take himself as bound to lose weight. When I utter the prescriptive, I call Scott’s attention to his normative status, in the hope that this status itself will exert force over his actions. When I utter the request, I try to use my normative position in order to exert force over Scott’s actions. I am, in effect, asking Scott to lose weight for me, or out of recognition of my authority, although presumably a large part of my entitlement to make this request is based in facts about his actual need to lose weight. In The Second-Person Standpoint, Stephen Darwall opens with the distinction between telling someone that she ought to stop stepping on your foot by calling attention to the moral benefits of ending your pain, and telling her to please get off your foot, speaking as the one whose foot is being stepped on, to the one who is doing the stepping.14 While Darwall’s analysis of this distinction differs from ours (as will become apparent later in this chapter and in Chapters 7 and 8), the example works nicely for our 14. Stephen Darwall, The Second-Person Standpoint: Morality, Respect, and Accountability (Cambridge, Mass.: Harvard University Press, 2006).

Prescriptives and the Metaphysics of Ought-Claims

109

purposes. The first utterance is agent-neutral in its entitlement: anyone who witnessed the foot-stepping could have told her she ought to get off your foot, and their entitlement would have been just the same. But you have special normative standing as person being stepped on when it comes to asking her to get off your foot. In the first case you try to make the norm itself guide the stepper’s actions, and in the second case you try to guide her actions by holding her to the norm. In practice, it is often hard not to smuggle a subtle holding into a secondpersonal prescriptive; when I point out an ‘ought’ to someone I am almost inevitably heard as requesting that she obey it, rather than as merely exhibiting its salience so that it can do its own normative work. Conversely, when I entreat or command you to act as you ought, it would normally be very odd for me to do so without trying to direct your attention to the fact that you ought to act in this way. That is, when I hold you to your commitments, I do not usually just request that you do what you were bound to do anyhow; instead, I direct your attention to the norm that binds you and ask that you acknowledge its force. As we have seen repeatedly, speech acts frequently combine several normative functions. An utterance such as “Don’t you think you owe your father an apology?” will often serve to make a prescriptive truth-claim, direct your attention to the binding force of your commitment, and hold you to that commitment, all at the same time. Subtle as it may often be in practice, we think that the distinction between entitlements to prescriptives and entitlements to holdings is philosophically and ethically important. Margaret Little has given a careful philosophical analysis of how various kinds of intimacies enable us to make different sorts of claims upon one another.15 She has worked to separate the fact that someone ought to do x from another person’s entitlement to ask her to do x, and in turn, she has distinguished various modalities of holdings (commands, entreaties, etc.) and their distinctive entitlements. Especially in the domain of duties involving intimacies of the body, Little argues, our entitlements to hold others to their duties in various ways will depend on the specificity of our relationships to them. She offers the following sorts of examples: there may be cases where I ought to have sex with a stranger—when I find him perfectly attractive, 15. See especially Little’s forthcoming Intimate Assistance: Rethinking Abortion in Morality and Law.

110

‘Yo!’ and ‘Lo!’

there are condoms available, and it is his dying wish, perhaps. But even in such a case, the stranger has no standing to demand that I do as I ought. It is simply beyond the purview of his entitlement to order me to honor this duty, no matter how nicely he puts it (though he may invite me to do so). Little argues that abortion is a domain marked by such intimate duties and likewise by such shades of entitlement. There are cases in which my having an abortion would be a callous and morally inappropriate act, for instance if I get pregnant intentionally and have plenty of financial resources, but decide later that I want to have an abortion just for the spiteful satisfaction of disappointing my mother. But even in such a case, she claims, random people (not to mention the law) have no standing to hold me to gestating against my will. While they would state a truth if they said I ought not to have an abortion, they would overstep their entitlement if they ordered or even asked me not to. Close friends and family may have the standing needed to gently hold me to my duty to gestate, for instance through a Strawsonian reactive attitude that shames me; yet they still do not have the entitlement to order me to gestate.16 This distinction between prescriptive truth-claims and holdings with agent-relative entitlements is inchoately reflected in the common (and nearly incoherent) opinion that ‘abortion is wrong’ but ‘it’s nobody’s business to judge’ women who have abortions. If abortion is wrong, then in fact it is everybody’s “business” to judge that a woman who has one has done something wrong—this is an example of the agent-neutrality of the entitlements of truth-claims. Yet it might be nobody’s business (or at least no stranger’s or government’s business) to hold a woman to her commitment not to abort. We all know that the space of the legal should never extend as far as the space of the moral; we can now state precisely one reason why this is so. It is not just that people should be given some latitude to behave immorally in some ways. Laws hold us to acting in certain ways, and the mere fact that a prescriptive is true doesn’t entitle a corresponding state-issued holding. (What does entitle a state-issued holding is a difficult and important question indeed; much as we hope to delve into it eventually, we are not going to address it here.) 16. Such shades of holding are examined in detail, in ways that are relevant to our project, by Coleen Macnamara in “Beyond Praise and Blame: A Theory of Holding Others Responsible” (Ph.D. diss., Georgetown University, 2005).

Prescriptives and the Metaphysics of Ought-Claims

111

So far, we have noted three ways that an utterance may produce pressure on someone, say Ella, to act as she ought: 1. A third-personal prescriptive calls upon everyone to acknowledge a truth about Ella’s commitments, and thereby calls upon Ella to give first-personal uptake to these commitments. 2. A second-personal prescriptive seeks to induce such first-personal uptake in Ella by calling her attention to her commitments and displaying their force. 3. An imperative (or entreaty, etc.) holds Ella to acting in accordance with her commitments. Even though the imperative belongs in box 4, and has an agent-relative input, it is a distinctive kind of holding that inherits some of the grounding in facts about the world enjoyed by prescriptives. For notice that this imperative seeks to hold Ella to a commitment that she was bound by anyhow. The imperative, if it is legitimate, may make Ella especially beholden to the speaker for upholding her commitments, but it does not create the commitments in the first place. As someone entitled to the imperative—Ella’s mother or spiritual counselor, perhaps—I am in a special position that allows me to make her responsible to me for upholding her commitments, in addition to the impersonal responsibility she already has; to put this the other way around, if Ella fails to live up to her commitments, she will now have failed me and not just failed to do as she ought. (The stereotype has it that Jewish mothers are particularly good at marshaling this particular normative tool.) But no matter what my relation to Ella, the content of my imperative is justified by the facts about Ella’s commitments. My standing does not add to or create this content; it merely enables me to demand Ella’s uptake of it. Let’s call such an imperative an alethic imperative. An alethic imperative holds its target responsible for living up to commitments that she already has.17 That is, an alethic imperative is one that demands that someone do something that she is independently bound to do given true facts about the world. (We could, of course, similarly have alethic entreaties, alethic suggestions, alethic permittings, and other varieties of alethic holdings, in all of which we hold someone responsible for living up to 17. See Macnamara, “Beyond Praise and Blame,” for a development of this notion of holdingresponsible.

112

‘Yo!’ and ‘Lo!’

preexisting normative statuses. For purposes of parsimony, we will stick to the consideration of imperatives for the rest of this discussion, with the understanding that the generalization to other modalities of holding is trivial.) All imperatives whose deontic content is also the content of a true prescriptive will be alethic imperatives: if it’s true that you ought to do x, then an imperative of the form “Do x!” will be an alethic imperative. Conversely, for every alethic imperative, there is a corresponding prescriptive with the same deontic content: If, when I tell you to do x, it was already independently true that you were committed to doing x, then it was also already true that you ought to do x.18 (On the other hand, to repeat, the mere existence of a true prescriptive does not on its own guarantee anyone’s—not to mention everyone’s—entitlement to a corresponding alethic imperative. Special normative standing is required before one is entitled to an alethic imperative, regardless of the truth of the corresponding prescriptive.) The relationship between alethic holdings and the prescriptives that share their deontic content (“Please lose some weight!” and “You should lose some weight”) is analogous to the relationship between observatives and declaratives that commit us to the same sets of beliefs (“Lo, fat Scott!” and “Scott is fat.”): in both cases, the two speech acts share the same output, but they differ with respect to the agent-neutrality or agent-relativity of their input. Yet not all imperatives (entreaties, etc.) are alethic. Many imperatives are what we can call constative: they seek to create a new commitment through their utterance where none existed before. Normally, if a colonel orders a private to drop and give her twenty push-ups, the private thereby becomes committed to doing so—however, there is no sense in which the private already ought to do the push-ups in advance of the order. In such a case, the legitimate order constitutes a new duty. Constative holdings litter our interactions: when we appropriately ask or tell one another to pass the salt, practice the same piano piece one more time, report to duty at 0600 hours, or welcome a speaker with a round of applause, we use our normative standing to hold people to actions they were not otherwise committed to performing. If such an imperative is legitimate, it will make a new prescriptive true: once you have asked me to 18. This may be a good place to point out that our account does not preclude the existence of moral dilemmas. Perhaps we can be truly committed to conflicting actions, and hence subject to conflicting oughts.

Prescriptives and the Metaphysics of Ought-Claims

113

pass the salt, assuming it’s a reasonable request, then I ought to pass it. But your utterance here tells me what to do by making it so that I should do it, rather than by holding me responsible for living up to commitments I already have.19 Hence we have seen four pragmatically distinct kinds of speech acts that call upon someone to do something: 1. Third-personal prescriptives (which belong in box 1 and box 3, but in which the box-1 function dominates). 2. Second-personal prescriptives (which belong in box 1 and box 3, but in which the box-3 function dominates). 3. Alethic holdings (which belong in box 4 but are grounded in true prescriptives). 4. Constative holdings (which belong in box 4 but can make new prescriptives true).

5.3 Two Alternative Accounts In this section we turn to influential attempts to understand oughtclaims as having a distinctive pragmatic structure: R. M. Hare’s classic version of prescriptivism, in which he understands prescriptives as odd sorts of universalized imperatives; and J. E. J. Altham’s analysis of moral judgments as ‘besires’, especially as filtered through Michael Smith’s more widely read discussion of besires.20 Both these accounts share with ours the insight that deontic claims cannot be understood simply as declarative judgments because they inherently have a distinctive practical import for the person whose ‘oughts’ they concern. Both also share with us a refusal, in rejecting the idea that such judgments are declarative, to simply place them outside the space of articulate discourse, as did the ‘Boo/Yay’ emotivists such as A. J. Ayer and C. L. 19. We can rig an example in which an imperative seems to function constatively even though it demands that someone do something they ought to do anyhow. I might tell my son to go play at the neighbor’s house, not because I think he has an independent duty to do so, but just because I want him out of my hair for a couple of hours. As his parent, I have the authority to order him to do this. Unbeknownst to me, perhaps, my son has already promised that neighbor that he would come over and visit that evening. In this case, he already had a duty to visit. However, my order does not seek to hold him to that duty, but rather to his new duty to do as his parent commanded. In this case, my imperative seems best classed as constative, even though it holds him to do what he was obligated to do anyhow. 20. Hare, Language of Morals; Altham, “Legacy of Emotivism;” Smith, Moral Problem.

114

‘Yo!’ and ‘Lo!’

Stevenson.21 Here we hope to show that our analysis is more elegant than these two kindred accounts and better avoids classic metaethical paradoxes. R. M. Hare was impressed by the fact that, like traditional imperatives, moral language, including ought-claims, primarily guided action rather than belief. Because he presumed that truth-claims were to be equated with declaratives, and because analyses of moral judgments as declaratives seemed unable to capture their prescriptive force, he concluded that moral judgments could not be truth-claims. Instead, he sought to understand them as a peculiar, broadened form of imperatives, which for him formed the only familiar pragmatic category of speech acts with prescriptive force. Although Hare’s account is out of date and not really a player in contemporary debates, his idea that prescriptives should be understood as more similar to imperatives than to declaratives continues to have substantial influence in metaethics. From our point of view, Hare is exactly half right: prescriptives—or at least prescriptives in their distinctive box-3 function—share an agent-neutral input with declaratives, but an agent-relative output with imperatives. We hope to show that our account enables us to make sense, in a way that Hare was not in a position to do, of both the analogy between prescriptives and imperatives and the places where this analogy breaks down. For Hare, ought-claims, like imperatives, have prescriptive rather than declarative force: “Their primary function is not to give information; it is to prescribe or advise or instruct.” According to him, prescriptives manage to provide information only insofar as they depend upon our understanding which background facts must be true in order for them to make sense. However, he claims, they are distinguished from regular imperatives by their (implicit or explicit) appeal to a universal rule. In his view, whenever we say ‘ought’, we are “invoking some general principle . . . by uttering [an ought] we seem to imply (in a loose sense) that there is some principle that we are invoking—though it may not be at once clear, even to us, exactly what this principle is.” He claims that a truly universal imperative would just be an ought.22 In trying to explain how prescriptives are distinct from regular imper21. A. J. Ayer, Language, Truth, and Logic (New York: Penguin, 2001); C. L. Stevenson, Ethics and Language (New Haven: Yale University Press, 1944). 22. Hare, Language of Morals, 159, 156, 178.

Prescriptives and the Metaphysics of Ought-Claims

115

atives by appeal to some kind of ‘universality’ or ‘generality’, Hare appears to be using an attenuated conceptual toolbox to try to capture the agent-neutrality of the input of prescriptives, by finding this universality at the level of their justification rather than at the level of their pragmatic function. For otherwise his appeal to general principles is really quite mysterious. Especially given that Hare, like us, does not restrict himself to moral ‘oughts’, why in the world would he think that oughts are more grounded in universal principles than are imperatives? Hare’s only argument is that even when no such rule is explicit, we can always ask for the reason for an ought-claim in order to draw it out. To use his example, if I tell you, “You ought to use the starting handle,” you can always legitimately ask me, “Why ought I to use the starting handle?”23 Hare is certainly right that prescriptives are essentially such that it is always in order to call for reasons for their legitimacy. But he offers us no grounds for thinking that the legitimate reasons we might offer in response will be any more general than the original claim. (“Because the automatic start-up function is frozen,” we might reply.) And even if we have separate philosophical reasons to be committed to the idea that any chain of legitimate reasons of this sort will bottom out in general rules, the more important point is this: it seems we can ask for exactly the same sort of reasons in response to an imperative. (“Use the starting handle!”; “Why should I?”; “Because the automatic start-up function is frozen.”) But perhaps it is only alethic imperatives for which we can demand reasons of this sort, and perhaps Hare would claim that such imperatives are simply prescriptives phrased using a different surface grammar. Two points are necessary in response to this move. First, this seems to do a grave injustice to constative imperatives. If I tell you to close the door, or to drop and give me ten push-ups, but I can provide no reason at all for why my request is legitimate, then it seems to be a dubious request. Only in a few rarified situations of very unequal authority can one get away with the time-honored parental response “Because I said so!”— and even then one would hope that this is not the actual whole reason for the request. Second, we have shown that there is an important distinction between alethic imperatives and prescriptives, and if Hare has elided this distinction, then he has significantly misrepresented the pragmatic structure of one or the other or both. 23. Ibid., 156 (emphasis added).

116

‘Yo!’ and ‘Lo!’

Indeed, it seems that Hare is doomed to misrepresent both, given his failure to recognize two distinctions that have been central for us throughout this book. On the one hand, his failure to distinguish the inputs from the outputs of speech acts hides the existence of normative functions that fit into boxes 2 and 3 in our typography—that is, the ‘mixed’ forms where the input and the output have different pragmatic scopes. Without this distinction, Hare has no mechanism for retaining the agent-neutrality of the input of prescriptives while holding on to the agent-relative import they share with imperatives. Hence he is committed to insisting that prescriptives are ‘not truth-claims’ and that they ‘have no informational content’. But on our account, prescriptives have plenty of informational content—they give us information about people’s deontic statuses. Hare in effect leaves us with no discursive means for delivering such information, nor does he defend or seem to believe in any kind of radical elimitivism that would deny its existence. Deontic information becomes weirdly ineffable, on his account. On the other hand, Hare also fails to notice the structural distinction between the first-, second-, and third-person voices. He finds nothing inherently problematic about the idea of ‘translating’ imperatives, not only out of the second-person voice, but into a completely impersonal voice not marked by speaker or audience: “The imperative mood, therefore, has for our purposes to be enriched in order to make it possible to frame sentences in all persons and all tenses.” He suggests the example, “All mules being barren, please.”24 From everything we have said in this book, such an imperative is not merely unidiomatic; it is pragmatic gibberish and has no discernable functional structure. Yet Hare’s severance of the imperative from any particular voice or audience is essential for his purposes, for he needs imperatives to be entailed by prescriptives. While he recognizes that imperatives are not entailed by anything if we restrict entailment to propositional inference narrowly construed, he thinks the notion of entailment can and should be extended so as to allow practical and prescriptive inferences. Given this extension, it is important to him that imperatives and prescriptives be the kind of things that can bear entailment relations, so that normative judgments do not become mere noncognitive expressions that lie beyond the space of reasons, as they did for the emotivists. Sensitive to 24. Ibid., 188–189.

Prescriptives and the Metaphysics of Ought-Claims

117

the common worry that any form of prescriptivism will devolve into such noncognitivism, Hare wishes his ought-claims to be combinable into conditionals, able to serve as premises of arguments, and so forth. However, Hare can maintain that imperatives are entailed by prescriptives only by completely erasing the functionally essential role that voice plays in both prescriptive and imperative speech; only thus can he claim that a speech act with an agent-neutral input (to which anyone who is free of epistemic defect would be entitled) could entail one with an agent-relative input (to which only those with a specific normative position are entitled). As we have seen, entitlement to an imperative always requires a specific agent-relative normative status, which will necessarily exceed the generalized conditions for entitlement to a prescriptive with an agent-neutral input. On Hare’s account, entitlement to the prescriptive “Scott ought to lose weight” entails entitlement to the imperative (issued to Scott) “Lose weight!” But entitlement to the first speech act only requires knowledge of its truth, and this is insufficient to warrant just anyone to issue the imperative. Surely if a stranger, noticing the truth that Scott ought to lose weight, walked up to him and ordered him to lose weight, Scott would be right to accuse her not only of rudeness but of wildly overstepping her entitlement to make demands of him. Hare, though, erases the entire issue of who is entitled to an imperative on the basis of the truth of a prescriptive. Hare manages to hide the problem in his text by only giving an example in which someone is both the speaker and the target of the prescriptive and the imperative. In defending the claim that all evaluative claims necessarily entail their corresponding imperatives, he writes: “Valuejudgements, if they are action-guiding, must be held to entail imperatives . . . I propose to say that the test, whether someone is using the judgement ‘I ought to do x’ as a value judgement or not is ‘Does he or does he not recognize that if he assents to the judgement, he must also assent to the command “Let me do x”?’”25 Hare has maintained the plausibility of his general claim about entailment by switching to the case of a particular speaker, namely the target. Now the entailment principle may actually work in the first-personal case: it may indeed be that if you recognize, first-personally, that you ought to do x (lose weight, etc.), then in so recognizing, you are auto25. Ibid., 163, 168–9.

118

‘Yo!’ and ‘Lo!’

matically committed to acknowledging the legitimacy of commanding yourself to follow this prescription. This is an interesting claim, and it strikes us as plausible; first-personal practical recognition of your own commitment may intrinsically involve recognizing that you ought to hold yourself to that commitment. But this plausibility seems to come entirely from the uniquely first-personal structure of the uptake that we each give to prescriptives that apply to us—that is, from the output characteristic of their box-3 function, rather than from their general agentneutral output. And of course, it concerns only our own entitlement to the corresponding first-personal imperative; it gives us no reason to conclude that everyone is entitled to order me to uphold my commitments. Hence the entailment link that Hare has identified depends essentially on the very structure of voice and agent-relative statuses that he himself erases from his picture. Hare attempted to mark out the distinctiveness of evaluative language pragmatically, rather than semantically. Furthermore, he tried to do this while keeping evaluative language within the space of reasons. We have followed his lead in both respects. But he tried to give a voicefree, audience-neutral account of the pragmatics of prescriptives, and hence he did not have the conceptual resources he needed to identify the difference between prescriptives and imperatives. Notice that on our account, attention to voice and audience is what has enabled us to understand prescriptives as functioning as truth-claims and as action-guiding at one and the same time: the very same speech act that has declarative force for third parties has practical, action-guiding force for those for whom it has first-personal prescriptive significance. We have, with no need for metaphysical excess, cut ourselves free from what has been viewed as one of the central paradoxes of metaethics, which Michael Smith has perhaps extravagantly dubbed “the moral problem.” ‘The’ moral problem, according to Smith, is as follows: How can moral judgments be both objective, in the sense of being truth-claims about a public world, and inherently motivating or practical reason-giving? The problem in reconciling these two apparent features of moral judgments, in Smith’s analysis, comes from our widespread commitment to some form of belief-desire psychology—that is, an ontology of propositional attitudes such that they either seek to fit the world (be-

Prescriptives and the Metaphysics of Ought-Claims

119

liefs), or seek to make the world fit them (desires), but not both at once. Within the framework of such a psychology, if moral judgments function to describe the world, then they cannot at the same time inherently motivate actions that would change it, whereas if their function is to motivate actions, then they cannot at the same time describe it. Smith points out that his analysis leaves three options: we can reject the idea that moral claims are inherently motivating, we can reject the idea that they have descriptive content, or we can reject belief-desire psychology. The first position is associated with ‘cognitivists’ such as Philippa Foot, who take moral judgments to be simply declarative truthclaims without inherent motivating force. On such a view, there is no contradiction in being committed to the proposition “I ought to spend more time with my aging mother” while not recognizing that this commitment gives me any practical reason whatsoever to spend more time with my mother, however much the truth and the practical reason are contingently linked in practice.26 Here, the motivation to do what we judge that we ought to do is external to the judgment—hence the name ‘externalists’ for proponents of this type of cognitivism.27 The second position is associated with ‘non-cognitivists’, including prescriptivists and emotivists, who claim that moral judgments are inherently motivating, and that precisely for that reason they ought not to be understood as having any declarative content. We already saw Hare give this argument.28 The final position is perhaps the least popular: it seeks to retain both the declarative content and the motivational force of moral judgments by rejecting belief-desire psychology, in particular the idea that propositional attitudes cannot have both ‘directions of fit’ at the same time. John McDowell is the best-known proponent of this third view, arguing in “Virtue and Reason” and other classic essays that in perceiving moral facts, we perceive truths that directly exert practical norma26. Foot, “Morality as a System of Hypothetical Imperatives.” 27. See McNaughton, Moral Vision. 28. Another famous proponent of this argument is Gilbert Harman; see his The Nature of Morality (New York: Oxford University Press, 1977). Noncognitivists often find extra ammunition in the idea that cognitivists, in understanding moral judgments as truth-claims, are committed to an unsavory moral realism. However, we have argued that there are no special metaphysical barriers to treating prescriptives as truth-claims, so this worry need not concern us here.

120

‘Yo!’ and ‘Lo!’

tive force upon us—we literally ‘see reasons’ for action.29 Smith follows J. E. J. Altham in giving the name ‘besires’ to such double-edged attitudes, which share a direction of fit with both beliefs and desires. McDowell and his admirers aside, besires have been generally perceived as implausible and undesirable additions to a moral psychology, and not just because of their silly name. Our account retains the cognitivists’ commitment to the status of moral judgments as truth-claims (while denying that they have merely declarative force), as well as the noncognitivists’ commitment to their inherent practical import. Hence we are best understood as falling into McDowell’s camp and rejecting the dualistic underpinning of beliefdesire psychology. On our account, first-personal commitments to claims about one’s own commitments—that is, judgments of the form “I ought to x”—do indeed count as ‘besires’, if a besire is simply an attitude that has both directions of fit built into it. This commitment faces the tribunal of the world as beliefs do: if I discover facts that show it to be false, I should give it up. But it also makes a claim on action: if I have this commitment, it directs me to act in a certain way. We believe that this option looks vastly less mysterious when it is situated within the framework we have developed. We need make no claims about the features of the world having the capacity to reach out and grab onto our motivational structure as soon as we recognize truths about them. If we begin from impersonal evaluative claims stripped of any particular voice—such as the ‘it is right that. . .’ claims that Smith favors as examples—and portray them as truths that somehow, inherently, make a direct practical claim on whoever happens to note them, this can seem mysterious; it looks as if the practical import somehow rests in the truth and is waiting to be noticed. This idea of an inherent motivational property that can cling to states of the world is what most fundamentally struck Mackie as unacceptably ‘queer’. The apparent air of queerness surrounding ‘besires’ is dissolved once we distinguish carefully between first-personal and third-personal prescriptive judgments, taking this distinction in voice as a pragmatic distinction as opposed to a semantic or merely grammatical one. Recognizing that I have a commitment is essentially a matter of recognizing a claim over my actions, whereas recog29. See McDowell, “Virtue and Reason,” The Monist 62 (1979): 331–350; and McDowell, “Are Moral Requirements Hypothetical Imperatives?” Proceedings of the Aristotelian Society, supp. 52 (1978): 13–29.

Prescriptives and the Metaphysics of Ought-Claims

121

nizing third-personally that a commitment exists is a theoretical recognition of a truth. The practical force of my commitment comes from my recognition that it is I who am committed. When I recognize my own commitment, I both recognize a truth and, in so recognizing, acknowledge a claim on my actions, and hence this recognition has both directions of fit. Smith distinguishes between ‘motivating reasons’, which are the things that psychologically move us to act, and ‘normative reasons’, or actual rational claims upon us. He argues that the temptation to believe in ‘besires’ is based in a conflation of these two types of reasons. In his view, normative reasons are “best thought of as truths,”30 and they have no inherent motivating force; they need not exert psychological pressure on us, which is what motivating reasons do. The apparent existence of ‘besires’, according to Smith, comes from our conflating what are in fact two separate mental states: our acceptance of a normative reason (which is a belief) with our having a motivating reason (which is a desire). However, as we have described them, first-personal recognitions of our own commitments provide a kind of reason that does not fit neatly into either of Smith’s categories. It is surely true that we can recognize our commitments without this recognition creating in us a desire that psychologically motivates us to act. I might recognize that I ought to donate all of my spare wealth to charity or get my cholesterol level checked while experiencing no psychological feeling that I want to do these things, not even one that is trumped by competing desires. Yet when I recognize these things about myself, I don’t recognize a mere truth, as I would if I recognized them about someone else. Rather, I recognize that I have a commitment to act. This is one single mental act, not two separate mental states, and it is an act that recognizes a truth and a practical reason at the same time. There is no way of construing this commitment as devoid of practical import; its practical import is what I recognize. I cannot merely theoretically note a discrepancy between what I ought to do and what I actually do; by noting such a discrepancy, I acknowledge my own transgression. To put the point another way, in so recognizing, I must take my own inaction in the face of this commitment as a practical failing on my part. We should not confuse acknowledgment of a practical reason with a psychological feeling of wanting to act in accordance 30. Smith, Moral Problem, 95.

122

‘Yo!’ and ‘Lo!’

with it, any more than we should confuse acknowledgment of a theoretical reason with a psychological feeling of wanting to draw inferences in accordance with it. Thus our analysis of prescriptives, and in particular our attention to the special pragmatic structure of first-personal prescriptive judgments, makes several resilient problems of metaethics go away, with little fuss. It accommodates the intuition that there is a tight relationship between prescriptives and imperatives, without needing to struggle with the pitfalls that have attended versions of the claim that prescriptives just are imperatives. It retains the commonsense intuition motivating cognitivism, namely that moral judgments are truth-claims bearing articulate inferential relations to other claims, but without positing metaphysically questionable entities or properties. And it solves ‘the’ moral problem by taking the mystery out of the idea that a single moral judgment could have both descriptive content and practical import.

5.4 Reasons, Claims, and Addresses Stephen Darwall, in The Second-Person Standpoint, analyzes the structure of moral claims by taking their voice as essential to their form and function. Since Darwall seems to share several of our motivations and to be trying to get at distinctions that matter to us, it is worth exploring how his account diverges from ours. We have distinguished in this chapter between two kinds of claims that have practical import for the person(s) whose commitments they target: prescriptive claims, which are truth-claims with agent-neutral inputs that have special practical import when taken up first-personally, and imperatives and other holdings, including alethic imperatives, which have agent-relative inputs. Darwall opens his book with the example we cited earlier, which at first glance seems to track this distinction. He writes of a situation where someone is standing on your foot: Compare two different ways in which you might try to give someone a reason to stop causing you pain, say, to remove his foot from on top of yours. One would be to get him to feel sympathetic concern for you in your plight, thereby leading him to want you to be free of pain. Were he to have this desire, he would see your being in pain as a bad thing, a

Prescriptives and the Metaphysics of Ought-Claims

123

state of the world there is reason for him (or, indeed, for anyone who is able) to change. And he would most naturally see his desire that you be pain-free, not as the source of this reason, but as a form of access to a reason that is there anyway . . . Alternatively, you might lay a claim or address a purportedly valid demand. You might say something that asserts or implies your authority to claim or demand that he move his foot and that simultaneously expresses this demand . . . the reason you would address would be agent-relative rather than agent-neutral.31

According to Darwall, not only does the first transaction result in an ‘agent-neutral’ reason to act whereas the second results in an ‘agentrelative’ reason, but also, the second provides what he calls a ‘secondpersonal reason’. Although this sounds roughly like our distinction between prescriptives (“You ought to get off my foot”) and alethic imperatives (“Please get off my foot”), two distinctions that we have dwelled upon are conflated in this passage. First, like Darwall, we portray prescriptives as providing access to a reason that is there anyway, and we also insist that this gives them a kind of agent-neutrality, namely at the level of their input. As Darwall aptly puts it, when we show someone what he has a reason to do with a prescriptive, he accepts a “state-of-the-world-regarding” reason: “Qua this form of reason-giving, you would be asking him to agree, as it were, that there is a reason for him to do something rather than asking him to agree to do it.”32 You are not drawing on your agent-relative entitlement to make a claim on him, but pointing his attention to an agent-neutral fact. But this agent-neutrality of input does not entail the agent-neutrality of output that Darwall assumes when he says that what we show is merely a state of the world that anyone who is able has a reason to change. A prescriptive may be agent-neutral and state-of-the-world-regarding in its input while being utterly and non-interchangeably specific in its output: for instance, I might point out that because of the special commitments that you have taken on in adopting a child, it is now your obligation to see to it that this child gets the unexpected medical care she needs. Here I am not demanding or entreating that you do so, or drawing on the special relations of authority between us; indeed, I might point this out to a third party instead of to you. I am pointing out a truth that anyone in touch with the truth is (discursively) entitled to point out, namely that 31. Darwall, Second-Person Standpoint, 5, 7. 32. Ibid., 6–7.

124

‘Yo!’ and ‘Lo!’

you have a special, agent-relative reason to act that is not transferable to anyone else, even if someone else is in a position to help.33 Other times, it may well be that my prescriptive gives you a reason to act only because you happen to be one of the people in a position to do something that ought to be done—for instance, turning off a running tap so as not to waste water. In both cases, the prescriptive has practical rather than theoretical import for you only in virtue of your taking up its import first-personally. Darwall need not and should not draw any conclusions about the generality or neutrality of the reason that a prescriptive reveals, in pointing out the neutrality of its entitlement, or the fact that the reason is ‘there anyway’, independent of my address. Second, here and elsewhere in the book, Darwall conflates the agentrelativity of a claim with its second-personality. He writes: “What makes a reason second-personal is that it is grounded in (de jure) authority relations that an addresser takes to hold between him and his addressee.” This definition actually includes two distinct elements: (1) the grounding of a reason in specific, agent-relative relations between people; and (2) the grounding of the force of a claim in a second-personal address. Darwall explicitly associates both of these with his second-personal reasons. For instance, he first introduces second-personal reasons as follows: “A command is a form of address that purports to give a person a distinctive kind of (normative) reason for acting, one I call a second-personal reason.” But shortly thereafter he claims that it would be a secondpersonal reason because it “would concern, most fundamentally, his relations to others . . . The reason would not be addressed to him as someone who is simply in a position to alter the regrettable state of someone’s pain . . . It would be addressed to him, rather, as the person causing gratuitous pain to another person.”34 These features ought to be kept quite distinct. Someone can perfectly well have an agent-relative (or even an agent-specific) reason to act—a reason that is grounded in her particular relations to others and specific to her particular normative position (as the one causing pain, as the one who adopted the child, etc.)—without that reason being presented to her in the form of a second-personal ad33. Indeed, even if I am the adoptive child herself, I still make a claim with an agent-neutral input if I point out your distinctive obligation to me. But as the adoptive child, I also have an agent-relative entitlement to ask you to live up to your commitment to me, which is a box-4 speech act. 34. Darwall, Second-Person Standpoint, 4, 3–4, 7.

Prescriptives and the Metaphysics of Ought-Claims

125

dress. For that matter, she can have agent-relative reasons of this sort that no one ever points out, or that are pointed out only between third parties and not to her at all. Indeed, she might come to have an agentrelative reason to act because of a transaction between two other people in which she is not involved: If her agent rents out her New York property to a tenant while she is living in Florida, she now has an agent-relative reason to pay taxes on that rental income. Meanwhile, I can be held to acting on an agent-neutral reason through a second-person address; my spouse may beseech me to return that ridiculous Hummer to the dealer and buy a Civic instead, for instance, because we should all stop burning unnecessary fossil fuels. Thus the agent-relativity or agent-neutrality of reasons seems to be a red herring for Darwall, who is (or at least ought to be) more interested in the addressing and holding functions of second-person transactions. Whether or not our reasons for acting are specific to us, and whether or not these reasons are dependent upon particular relations to others, are questions that are simply orthogonal to whether we are held to those reasons by a second-personal demand made by someone with the proper authority to so hold us. Part of the reason why Darwall may have difficulty nailing down the location of the second-personality he seeks is because he consistently talks in terms of the second-personality of reasons (and sometimes ‘perspectives’) rather than of claims. It makes perfect sense to separate agent-relative from agent-neutral reasons—that is, reasons that claim us in virtue of our special normative position and reasons with generic force. Such a distinction is part and parcel of our distinction between agent-neutral and agent-relative normative statuses. However, it is not at all clear what it means for a reason to be second-personal (or thirdpersonal, etc.). Second-personality, one would think, is a feature that a transaction such as a speech act can have. A second-personal claim is one that I make to you. Transactions can have such a second-person voice because they can have a direction and a transitive object. In this work, we have used the notions of first- and second-personality to describe how claims and speech acts are directed and received (and starting in the next chapter, the second-person voice will become vastly more central to our account). Granted, Darwall admits up front that he is stretching the notion of second-personality and bending it to his own ends, since his concern is

126

‘Yo!’ and ‘Lo!’

not with second-personality as a grammatical voice. But it is hard to see how to get any traction out of the use of the term unless we are talking about the kind of thing that can have a direction in this way. And reasons do not seem to be good contenders for such directionality. Reasons are essentially subject to public epistemic appraisal, and they bear publicly accessible rational relations to other reasons. Whether a reason has force only for me or for everyone, a statement of that reason will be an agentneutral truth-claim. One and the same agent-relative reason may be described in a third-personal truth-claim (“Since Mark is the one who is causing Rebecca pain, he really ought to get off her foot”), or imputed in a second-personal address (“Mark, please get off my foot!”). The voice attaches not to the reason but to the claim. While my demand may give you an extra reason to act as a by-product—wanting to do as I ask as a matter of politeness, for instance—this is clearly not the reason that Darwall has in mind. He is explicit that it is your causing me pain that provides the reason for action here, and this seems to be the same across the two cases. Why would Darwall think that the reasons themselves could be secondpersonal? We suspect that this mistake is rooted in his overinflation of the category of constative imperatives (and similar constative box-4 holdings), which are second-person addresses that create reasons that did not exist before. Darwall says of the second-personal case, “What is important for our purposes is that someone can sensibly accept this second reason for moving his foot, one embodied in your claim or demand, only if he also accepts your authority to demand this of him (secondpersonally).”35 But this seems manifestly false. Surely you can accept that you have an agent-relative reason to stop causing me pain whether or not you accept the authority of my demand, or even hear it at all. Indeed, we may accept all sorts of thoroughly agent-relative reasons for doing something for another person while openly denying that that person has the authority to demand this from us. I might accept that I have a reason to shower compliments upon my spouse when he is feeling unconfident, while denying that he has the right to demand compliments from me. (And I certainly wouldn’t think that everyone owes my spouse these compliments.) Darwall apparently believes that a second-person address always holds its target to act on reasons that have force only in35. Ibid., 8.

Prescriptives and the Metaphysics of Ought-Claims

127

sofar as this target accepts the authority of the speaker to hold him to the reason. This is a long-winded way of saying that Darwall apparently believes only in constative holdings, and not in alethic holdings. (This perhaps explains why he grants epigraphic status to Rawls’s statement “People are self-originating sources of claims.”) He then counts the reasons arising from all imperatives, entreaties, etc., as second-personal reasons—that is, as reasons that have force only because they were given that force by the second-person address. As we discussed earlier, there certainly are reasons that are constituted by addresses in this way. If, as your professor, I ask you to write a paper on a particular topic, you now have a new reason to write that paper, namely that I, with my proper authoritative position with respect to making such demands of you, have asked you to do so. Whether or not this is the only reason for you to write the paper, it is certainly a reason that was constituted by my demand. There could have been no true prescriptive describing your obligation to write this paper in advance of my imperative. Since such a reason owes its entire life to a second-person transaction between us—an address grounded in our proper normative relations—it is easy to think of the second-personality as somehow attaching to the reason itself. Fair enough.36 However, most of the reasons and addresses that Darwall discusses are not of this sort. When I request that you get off my foot, this is an alethic request: I am holding you, secondpersonally, to doing something you already had a reason to do, namely to stop causing me pain. I am holding you to getting off my foot for the very same reason that I (or a third party) would be impressing upon you if I (or she) uttered a prescriptive instead, pointing your attention to the fact that you ought to get off my foot. My special normative relationship to you gives me the right to hold you to your reason, but it does not create the reason. His belief that all second-person holdings are constative, and that they create the reasons that they demand be obeyed, may explain why Darwall believes that his account ultimately supports a contractarian theory of morality. He argues throughout the book that second-person ad36. But even here, the reason so created is perfectly describable by others. I may say, “Smith ought to write a paper, because her professor told her to,” and this statement will express the very same reason that was constituted by the second-personal interaction. So the reason itself does not, properly speaking, have any particular voice.

128

‘Yo!’ and ‘Lo!’

dresses and relations of holding are constitutive of the possibility of moral obligation, responsibility, and community, and hence that something like a Hegelian or Fichtean picture of the foundational role of mutual recognition in moral agency is correct. We agree, and we will argue this in detail ourselves in Chapter 8. However, he also takes this conclusion as sufficient to compel a contractarian account of morality. If you believe both that second-person transactions underwrite all moral commitments and that such transactions always function by creating new reasons, then it perhaps makes sense to conclude that all moral commitments are ultimately contracted into existence through second-person transactions. As we will argue soon, we agree that second-person addresses are the condition for the possibility of moral community and agency. But from all we have said in this chapter, this position allows us plenty of room for robust realism about people’s commitments and reasons. A second-person address can hold you to commitments you already had, rather than creating these commitments.

5.5 Coda: Categorical Imperatives Kant, as we know, based his moral theory around the functioning of a special type of ought-claim, the ‘categorical imperative’. The categorical imperative supposedly makes a claim on each of us, regardless of any of our individuating features or particular ends, sheerly in virtue of our potential for rationality. Now in fact, when we talk to one another about what we ought to do, using prescriptive discourse or second-personal holdings, almost all of our claims are context-specific and have agentrelative outputs. We tell people to call their grandmother, give them advice about how to negotiate a sticky situation at work, point out to them that they ought to have been more tactful or generous in a particular case, and so forth. Almost never do we make unconditional pronouncements about what commitments everyone has, merely in virtue of membership in the moral community or the space of reasons. Even if our moral duties are grounded in perfectly general, context-independent moral rules (maximize utility, etc.), we must acknowledge that virtually all of our contentful moral claims are applications of these rules in particular contexts, with agent-relative outputs, rather than categorical statements of principle. Indeed, many critics of Kant over the centuries have argued that there

Prescriptives and the Metaphysics of Ought-Claims

129

could be no such thing as contentful categorical imperatives—that is, they have argued that maximally general principles can yield contentful commitments with practical import only when situated within contexts that include the type of individuating ends and circumstances that Kant expelled from the domain of pure moral discourse.37 However, if we can formulate claims about unconditional commitments—about commitments that we have merely in virtue of our placement within the space of reasons—then the pragmatic structure of such claims will be quite distinctive. In particular, they seem, by definition, to be deontic claims that have (only) agent-neutral outputs, unlike the prescriptives that have concerned us in the rest of this chapter. Let us consider the pragmatic structure of Kant’s categorical imperative in particular, since Kant is the most explicit and important proponent of perfectly universal moral claims that target us in abstraction from any of our individuating features. Although he calls his moral judgments ‘imperatives’, Kant himself does not distinguish carefully between imperatives and prescriptives. In fact, his introduction of the notion of a categorical imperative in the Groundwork of the Metaphysics of Morals fluctuates at least twice within a single passage between these two types of speech acts. He writes: “The representation of an objective principle, insofar as it is necessitating for a will, is called a command (of reason), and the formula of the command is called an imperative. All imperatives are expressed by an ought . . . they say that to do or to omit something would be good.”38 The representation of a principle would normally take the form of an ought-claim, but he calls this representation a command. Yet he immediately claims that ‘all imperatives are expressed by an ought’, which of course they are not. Some authors, such as Foot, have simply concluded that Kant misused the term ‘imperative’ and was actually interested in deontic claims, or prescriptives.39 It seems more likely to us that Kant had both cognitiv37. Most famously, though certainly not most recently, see Hegel’s critique of Kant in Elements of the Philosophy of Right, ed. Allen Wood (Cambridge: Cambridge University Press, 1991). There has been a recent move to read Kantian morality as more contextually sensitive than this, but our interest here is not in the details of Kant’s account but in the basic idea of a categorical imperative. 38. Immanuel Kant, Groundwork of the Metaphysics of Morals, trans. Mary Gregor (New York: Cambridge University Press, 1998), 4:413. 39. Foot, “Morality as a System of Hypothetical Imperatives.” This reading was convenient for Foot, as it biased the text in favor of her cognitivist, externalist account of moral judgments.

130

‘Yo!’ and ‘Lo!’

ist and prescriptivist intuitions and had not clearly worked out the difference between prescriptive and imperative claims. In any case, if there are any moral judgments that make an unconditional claim on us simply in virtue of our rationality—as we will assume for purposes of argument for the rest of this section—then we can easily talk about either type of speech act: a prescriptive that states what we ought (unconditionally) to do, or a corresponding alethic imperative that (unconditionally) commands us to do it. Although they may be of special importance to moral philosophy, such Kantian alethic imperatives would not have a particularly unusual pragmatic structure for our purposes. Kant is clear that we can be bound by categorical imperatives only autonomously rather than heteronymously; famously, I must give myself the moral law in order for its claim to be legitimate. Although Kant thinks that we are to act as if we were willing our action as a universal law for everyone to follow, it is not this act of willing on our part that will bind others, but only their own autonomous subjection to the law. This means that the alethic imperative corresponding to the categorical prescriptive, for Kant, must be issued by each of us to our self. Each of us can command, “Let me obey the moral law,” and this command will bind only its speaker. Such an imperative—like any other—will have an agent-relative input and an agent-relative output, and will belong in box 4 of our grid. And if someone else undertakes to hold me to my Kantian duty through an alethic imperative or other such alethic second-person address, her speech act will likewise belong in box 4, along with all other holdings. Our pragmatic interest is instead in the prescriptive form of the ‘categorical imperative’—henceforth CI—and we will restrict our attention to it. Kant notoriously defines CI several times over, but his primary and most general formulation is that the CI “would be that which represented an action as objectively necessary of itself, without reference to another end.”40 The universally binding force of the CI comes from the fact that it presents an unconditional ought, which is in no way indexed to a particular agent. Thus Kant builds the agent-neutrality of its output into its very definition. What is special about CIs is not the universality of their extensional scope, but the agent-neutrality of their binding force. A true prescriptive that applies to all of us because of its content, such as ‘we all ought to try to minimize our use of fossil fuels’, does not count as 40. Kant, Groundwork, 4:414.

Prescriptives and the Metaphysics of Ought-Claims

131

a CI, for Kant; it binds each one of us only because of conditional facts about our supply of fossil fuels and the bad effects of burning them that happen to have a similar normative significance for all of us. Hence this is a thoroughly hypothetical imperative regardless of its scope. Indeed Kant is famously critical of those who try to ground CI in extensionally universal but ultimately contingent features of agents. He writes: “There is . . . only a single categorical imperative and it is this: act only in accordance with that maxim through which you can at the same time will that it become a universal law.”41 Since we are not, in fact, willing for anyone but ourselves here, Kant’s point is not that the output of CI is universal in its extensional scope, but rather that there is nothing in the maxim that marks its force as in any way specific to us and our normative position.42 In CI, the universality of its scope is in fact a product of its structural agent-neutrality, rather than the reverse. The output of the categorical imperative is thus genuinely different from the prescriptives we have considered in the rest of this chapter. Whereas normal prescriptives have agent-relative practical import for those whose commitments they identify, the categorical imperative has this practical import for everyone, not merely as a matter of fact, but specifically because this import is independent of any specific features of the agent. The categorical imperative speaks to each of us and calls upon us to act merely as inhabitants of the space of reasons, or as members of the most generalized possible ‘we’, rather than as individuals with a specific normative place in the space of reasons. (Indeed, to the extent that it speaks to us as specific, differentiated individuals, it does not function categorically.) It thus has agent-neutral instead of agent-relative outputs. At the same time, like all prescriptives, the categorical imperative has an agent-neutral input. Not only is it a truth-claim, but its grounding is supposed to be in reason itself, again specifically independent of any particular features of the speaking agent. We are each a subject of the 41. Ibid., 4:421. 42. It might, in some cases, be unclear whether the universality of a prescriptive is grounded in its structural agent-neutrality or merely in its extensional scope. Colleen Fulton (in private correspondence) suggested the example of the Christian call to repent for our sins. Surely this call is meant to be universal in its output. Is the idea that being a sinner who ought to repent is a transcendental condition for being a finite agent in the first place, or is it that as a matter of fact, everyone happens to be a sinner? In the first case, the claim that we are sinners who ought to repent is a CI with an agent-neutral output, but in the second it is a normal prescriptive with an agent-relative output. Which meaning is intended by Christian dogma is an interesting theological question lying far beyond our expertise.

132

‘Yo!’ and ‘Lo!’

moral law, but also, “every rational being . . . must regard himself as giving universal law.”43 Hence the categorical imperative is agent-neutral in both its input and its output, and—like declaratives, but unlike normal prescriptives—it belongs in box 1. And yet it is not just a declarative with a special subject matter. For its agent-neutrality in no way compromises its practical import. As with all prescriptives, first-personal recognition of the legitimacy of the categorical imperative reveals our commitments and thereby calls upon us to act, and not just to entertain a belief in its truth. The only difference here is that all of us, merely qua inhabitants of the space of reasons, have a first-personal relationship to the claim it makes. Kant is clear that the import of a CI is practical rather than theoretical: the representation of the moral law determines the will to act, as he puts it.44 In recognizing the legitimacy of the categorical imperative, we recognize that it binds our actions. This is what Kant tries to underscore by claiming that a CI commands, even though we have seen that, strictly speaking, the moral law is not the same as an imperatival command to follow that law. Yet the categorical imperative calls us to act, not on the basis of our particular situation and ends, but simply as an unmarked rational will. Since the output of a CI is both practical and agent-neutral, it will also, unlike the output of a declarative, be universal in practice and not just in the ideal. We saw earlier that one difference between practical and epistemic commitments is that you need not be in a position to recognize your practical commitment in order to really have it: if you are committed to paying $3,000 in taxes, say, your ignorance of this fact does not detract from your commitment, whereas you can be ignorant of the fact that mitosis is a form of reproduction and genuinely not be committed to that fact either, even though, ideally, you would be. If there exists such a thing as a categorical imperative that captures practical commitments that we have merely in virtue of being rational agents, then these commitments, unlike their epistemic counterparts, are not only agent-neutral but also universal. Thus, although this chapter has been devoted primarily to understanding the box-3 function of (most) prescriptives, we end with the in43. Groundwork, 4:443. 44. Ibid., 4:427 and throughout.

Prescriptives and the Metaphysics of Ought-Claims

133

teresting conclusion that if there can be unconditional prescriptives that impute contentful practical commitments, then declaratives are not the only inhabitants of box 1. Although we have not offered an argument for the existence of legitimate prescriptives of this sort, their prima facie coherence as a pragmatic category gives us good evidence that we should not take the declarative as the essential or definitional instance of a speech act with an agent-neutral input and output. One might have thought that such thoroughgoing agent-neutrality was the special purview of the declaratives, but we now to have reason to be suspicious of even this claim to their primacy or uniqueness. The current state of development or our typology is indicated in Figure 7.

Input Output

Agent-neutral

Agent-neutral

Agent-relative

1 Neutral input Neutral output

2 Relative input Neutral output

Declaratives Categorical Imperatives

3 Neutral input Relative output Agent-relative

Figure 7

Prescriptives

Kantian judgments of taste, baptisms, some recognitives, i.e. observatives

4 Relative input Relative output Imperatives (promises, invitations, reproaches...), ostensions

6

Vocatives, Acknowledgments, and the Pragmatics of Recognition

Vocative speech acts—that is, second-personal speech acts that hail or call (such as “Yo, Emma!” or “Hi, Eli!”)—form a distinct pragmatic category of utterances. They are not declaratives, prescriptives, or any other kind of truth-claim, for they have no truth-value. Nor do they seem to function straightforwardly as interrogatives, imperatives, or any of the other types of speech act that form the traditional canon of pragmatic analysis. Sentences in the vocative mood were recognized by the ancient classical languages as forming a distinct grammatical category, with its own declension. But contemporary philosophers of language and linguists have given vocatives next to no attention; the syntax, semantics, and pragmatics of the hail are all more or less uncharted territory. Some Continental philosophers, most notably Buber and Levinas, have given second-personal encounters special philosophical pride of place, but here too second-personal and vocatival discourse has received little rigorous attention.1 In this chapter we will tease out the pragmatic structure of vocatives. We began this book by pointing out that philosophers of language have tended to focus on impersonal, declarative speech that reports on public facts as the paradigm of language. In the wake of influential twentieth-century philosophers as diverse as Dewey, Wittgenstein, Hei1. An important exception—and an important inspiration for our analysis of vocatives and acknowledgments in this chapter and the next—is Louis Althusser’s discussion of interpellation in “Ideology and Ideological State Apparatuses,” in Althusser, Lenin and Philosophy (London: New Left Books, 1971).

134

Vocatives and Acknowledgments

135

degger, and Austin, however, many authors assign philosophical importance to the fact that language is essentially personal: it does not merely involve the abstract movement of information, but is always spoken by this or that concrete, embodied, context-bound person, to a particular audience. Over the course of this work we have seen several senses in which language can be ‘personal’. It may be agent-relative in its input, its output, or both. Furthermore, agent-relative inputs and outputs may or may not be agent-specific; some agent-relative statuses are distinctively indexed to particular concrete individuals, while others attach to whoever fills a certain slot or description. And some agent-relative statuses are singular (only one person can have them), whereas others can be held by several agents at once. A status that is both singular and specific is individuating; it essentially attaches to one and only one concrete agent. Finally, and most relevantly, some discursive claims are inherently voiced: they must be understood not just as entitled for or targeting this or that agent, but as from me or for you. For instance, we saw in Chapters 2 and 3 that observational episodes, and the observative speech acts that express them, are not only agent-relative and singular, but in each case mine. Their origination in a first-personal perspective is ineliminably built into the kind of speech act that they are. In focusing on vocatives and acknowledgments in this chapter, we will be examining speech acts that are not only agentrelative in both their input and their output (and hence belong in box 4 of our grid), but also inherently second-personal: they are in each case directed at you. They not only pick out a concrete target audience, but inherently address themselves to that audience. This indexing of speech acts to first and second persons cannot be reduced to semantic or structural features of these acts that could be third-personally available; such speech acts must be irreducibly heard and owned as mine and as yours— and such hearing and owning can make sense only from particular points of view that are actually taken up by living, embodied subjects who are capable of making and being bound by claims. A vocative is the purest form of an address in which one person calls out to another. The second-personal structure of the vocative seems clear even though there may be no second-person pronoun used in the hail. This second-personality does not reside in the surface grammar of

136

‘Yo!’ and ‘Lo!’

the hail, but in its pragmatic structure. As Hagi Kenaan puts it, such speech acts have a specific ‘directionality’: they are essentially from me to you.2 Oddly, although both Kenaan and Stephen Darwall have argued at length for the fundamental importance of second-personal speech acts in general, and addresses in particular, neither of them discusses vocatives per se, but rather they look for cases in which more philosophically familiar forms of speech such as declaratives and imperatives function as addresses. Both Kenaan and Darwall conflate the addressing function of certain speech acts with the generic fact that any speech act at all, including an impersonal declarative, can be addressed to a particular other person in the course of communication. This conflation makes it difficult for them to clearly and consistently demarcate those speech acts that count as second-personal in the sense that is important to them. For of course even a thoroughly agent-neutral declarative is always uttered by someone in particular, and usually uttered to someone in particular, for the purpose of informing or otherwise affecting that person; such perlocutionary effects, however, do not transform a perfectly decent impersonal declarative into a second-personal speech act. By focusing directly on the vocative, or the pure form of the address, we will be able to isolate and examine the pragmatic structure of the second-personal address more effectively. In the next chapter we will argue that, in fact, vocatives play an essential role in discourse; indeed, we claim that all meaningful, functional speech acts contain what we call a transcendental vocative—that is, they each have a vocative function, in addition to whatever other functions they might have, and this vocative function is a condition for the possibility of their being genuine speech acts at all. We will argue that the vocative is not only an example of the second-personal address, but the essence of its pragmatic form, and furthermore that all speech acts have a second-personal address built into their function. But before we get to that argument, which is one of the most important punch lines of this book, we need to devote some detailed attention to regular, everyday vocatives and the acknowledgments they call for. 2. Hagi Kenaan, The Present Personal: Philosophy and the Hidden Face of Language (New York: Columbia University Press, 2005).

Vocatives and Acknowledgments

137

6.1 Two Kinds of Recognitives Remember that a recognitive is a speech act that serves to express recognition of something that makes itself present to the receptive faculties of the speaker. We argued in Chapter 2 that an observative speech act such as “Lo, a rabbit!” does not merely make the declarative claim that a rabbit is present, nor that I see a rabbit. Rather, it serves a special recognitive function: it marks or expresses my detection of a rabbit. It is the recognizing, and not just what is recognized or who is recognizing, that is given expression in such a claim, and since what is expressed is the indexed recognition itself, this entitlement is not generalizable, even in the ideal. While others may well see the same thing as I do, and so while my entitlement to such a speech act may well not be unique, it is still the case that this reception and recognition are unshareably and specifically mine.3 Hence the entitlement that grounds my speech act is agent-relative and individuating. Since recognitives by definition express such recognitions, they have necessarily agent-relative inputs. In natural language we speak of ‘recognizing’ many kinds of things: we can recognize objects, facts, persons, nations, rights, and claims, to name just a few. But there is an important difference between the way that I use speech to recognize a state of affairs (that there is a rabbit in the bush, perhaps), and the way that I recognize a person by calling upon her in a meeting. While we might be tempted to interpret the first type of recognition as a mere passive noting, notice that both types of recognitive speech acts change the normative status of others in the discursive community, as well as making demands upon those others. When I call out “A rabbit!” my utterance embeds a demand that others accept that there is a rabbit present and that I saw it, use these beliefs appropriately in inference, and so on. Such a recognitive has an agentneutral output; I have secured public knowledge, even though my entitlement to the speech act is agent-relative. (If I utter “Lo, a rabbit!” my utterance also has the agent-relative output of ostending the rabbit— that is, calling upon others around me to direct their attention so that they too will recognize the rabbit in a certain way.) On the other hand, 3. This leaves open whether the agent, the ‘I’, that recognizes might be sometimes also a ‘we’ or a group agent.

138

‘Yo!’ and ‘Lo!’

when I recognize you as a speaker in a meeting, or when the provost at a university commencement ceremony recognizes the graduating class of 2008, the output of such a recognitive speech act is agent-relative: the recognition makes special demands upon and grants special entitlements to the one(s) recognized that are not shared even in the ideal. In recognizing you, I call upon you to speak; in recognizing a graduating class, the provost calls upon the members of that class to assume the duties and entitlements that attach to their degree (and, as collateral, I and the provost call upon others to recognize these normative claims). Hence all recognitives are agent-relative in their input, but some have agent-relative outputs and others have agent-neutral outputs; some recognitives belong in box 2 of our grid and others in box 4.

6.2 Vocatives Vocatives are hails. To utter a vocative is to call another person—in calling out “Hello, Eli!” I recognize the fact that that person there is Eli, and I do so by calling upon him to recognize that he has been properly recognized. Vocatives are thus recognitives with agent-relative outputs, and their pragmatic structure is rather complex. You cannot hail someone unless you recognize that he is there to be hailed, and part of what your hail expresses is this very recognition (where this recognition certainly need not involve direct perception—you can hail over the Internet, etc.). The vocatival demand that the one called appropriately acknowledge the call is not a separate pragmatic component of the hail over and above its recognitive function. Rather, it is how it carries out that recognitive function. This is why “Lo! Richard” is a very different speech act from “Yo! Richard”; both recognize Richard, but the first makes a claim with an agent-neutral output—it recognizes the publicly available, shared fact of Richard’s presence (and would be quite odd if directed at Richard)— whereas the second, in calling for a response specifically from Richard, has an agent-relative (and agent-specific) output. Many speech acts function to call upon an audience to recognize and respond appropriately to that speech act—indeed, we will soon argue that all of them do. Vocatives, however, in their pure form, isolate this function and elevate it to their central point; they recognize a person specifically in calling forth an appropriate recognition back from him that this recognition was itself appropriate and received. In the language

Vocatives and Acknowledgments

139

of Chapter 5, vocatives serve a constative function: in recognizing another, they hold that other to responding appropriately to that recognition. When I call out “Hello, Eli!” I hold Eli to saying “Hey, how’s it going!” back, or something of the sort, but of course my hail created rather than reflected this “duty” of Eli’s; he could have had no prior duty to respond to a recognition that hadn’t happened yet. (In Chapter 8 we will show that vocatives serve an alethic function as well.) Consider two cases of hailing: a teacher calls out my name during roll call in a class; a colleague greets me as she passes in the hall. The teacher and the colleague are discursively registering their recognition of who I am. However, these are not merely observative utterances, but rather they call for a response from me. In turn, my recognition that it is really I to whom my colleague is speaking is not just a matter of my recognizing the descriptive content of his claim; it is part and parcel of my recognition that I am the one who is being called upon to respond and uphold the norms of greeting behavior. The point is that my recognition of the hail and my recognition of the normative demand it makes on me to act so as to acknowledge the correctness of its recognition of me are one and the same thing. Recognizing the hail involves recognizing not just its presence but its target, its source, and its binding force, which is inseparable from taking it as really aimed at me—as making a real claim on me in virtue of having correctly identified me. This demand is not just a social nicety added to the discursive recognition, but part of the performative function of the call itself. The vocative hail can be a visceral, even an uncomfortable experience when it succeeds in grabbing its target. When the moderator of a panel asks, “Does anyone have any questions?”; when the leader of a support group asks, “Who would like to share their experiences with the group?”; or when a buddy I wasn’t expecting to run into calls out “Hey, Mark!” across the strip club,4 the feeling that it is really I who ought to respond to the hail may become a tangible weight. We have pointed out that vocatives are agent-relative in their output: they are intrinsically directed at an individual whom they identify or recognize. They are also agent-relative in their input. This is so, not only in virtue of their recognitive character, but also because they can only be entitled insofar as their speaker has the proper authority to legitimately make such a demand on the one she is hailing—and this authority, 4. As an anonymous referee was concerned might happen.

140

‘Yo!’ and ‘Lo!’

though it may be widely distributed, is always individually owned. The right to hail someone (in a particular fashion, in a particular circumstance) is not a publicly available entitlement but one that attaches to the normative position of the hailer as she is particularly situated within a structure of authority and normative relationships. Different people have the right to call one another in different ways and in different situations, and to call one another different things. A doctor’s patients may only be able to appropriately call her by her title and last name, whereas her friends can call her by her first name, and her parents can call her by an endearing albeit undignified nickname. Those same friends, patients, and parents cannot call her at all when she is in the middle of surgery or a public lecture, except perhaps in an emergency, which changes the normative structure of entitlements yet again. We can also have differential commitments to hail in various contexts: shopkeepers often ought to greet incoming customers, for instance. It may seem that a minimal hail—calling “hello” to someone as she passes by, for instance—requires no special authority or entitlements. But this is just because it requires an authority weak enough that almost everyone has it. It is easy to imagine a society so inegalitarian that even such minimal hails are not acceptable or acknowledged between certain kinds of people—perhaps hardly anyone can hail the king in even this minimal way, or perhaps men may not hail women in public. Indeed, all cultures have elaborate rules and rituals that constrain who can be hailed by whom and how and when, and the ‘neutral’ hail is an illusion sustainable only in a relatively lax and ritualistically flexible society such as ours.5 Because hailing someone always makes a demand upon her, asking for something in return, entitlement to a hail is never simply an agent-neutral given. It is essential to the functioning of vocatives that they establish a nor5. Notice that in the case of both imperatives and vocatives the normative structure of social ethics plays a large role in constituting discursive entitlements to speak. In such cases, the internal norms of discursive pragmatics are particularly intertwined with larger social normative structure governing discourse, and thus the two levels of normativity we distinguished in Chapter 1 are heavily interdependent. This does not mean that the distinction has been confused or undermined. If one issues an imperative or a vocative without the proper discursive entitlement, however constituted by social ethics this entitlement is, then the resulting speech act will not merely be rude or socially inappropriate; it will be a pragmatic misfire without appropriate performative force. Conversely, a vocative or an imperative can be rude and yet effective, because the norms governing these two layers of normativity are inextricably intertwined and mutually constitutive, with fuzzy boundaries between them, but still distinct.

Vocatives and Acknowledgments

141

mative relationship between caller and called and engage the one called in this relationship. This establishment and engagement places claims and burdens upon the one called, in part by demanding acknowledgment of this very establishment and engagement—the vocative does much more than note someone’s presence. It calls for a response and hence draws the one called into a direct and agent-relative normative relationship with the caller. The right to call others into such normative relationships is differentially distributed. It is against the background of the normative relationships that are already established that we sometimes, but not always, earn the right to engage others in new relationships through speaking to them.6 This is why vocatives can so easily be received as abusive, burdensome, or obtrusive: consider, for instance, how a man’s hailing of a woman he doesn’t know in a bar, or a homeless person’s attempt to hail me as I pass on the street, can be received as an uncomfortable or onerous demand for a response. Learning the theoretical fact that a homeless person needs money and wants you to give it to him is very different, in terms of its practical normative burden, from being called upon by that homeless person to give money. For example, it is only in response to the latter that our not giving money counts as a refusal. The homeless person’s call to you to give money, regardless of whether you do so, establishes a new normative relationship between you within which inaction is transformed into refusal, which is itself a second-personal, transitive action: I refuse you.7 In order to have standing as a person with normative commitments and entitlements in a community, others in that community must be able to recognize us as such and hold us responsible for that normative standing, and this in turn requires that we be hailable—an appropriate 6. James Bohman, in “The Importance of the Second Person: Interpretation, Practical Knowledge, and Normative Attitudes,” in Hans H. Kögler and Karsten Stueber, eds., Empathy and Agency: The Problem of Understanding in the Human Sciences (Boulder: Westview, 2000), 222– 242, points out that “competent interpreters and communicators thus possess the practical knowledge that is manifested in the ability to establish and maintain . . . normatively guided interactions and social relationships” (234). 7. Rousseau, in the Sixth Walk of his Reveries of a Solitary Walker, trans. Peter Frank (New York: Penguin, 1979), explores this normative transformation with the story of a little crippled beggar-boy, who, through the act of hailing Rousseau on his walk, turns the pleasure of charity into a resented duty to the boy. Rousseau argues that the former is somehow morally preferable, while the agent-relative duty imputed by the call is an unjustified interruption of his individual freedom, but of course we need not follow him in this particular romantic individualist fancy.

142

‘Yo!’ and ‘Lo!’

target for the vocative—for at least some people in some circumstances. One of Darwall’s central claims in The Second-Person Standpoint is that second-personal interactions necessarily involve mutual recognition— that is, two agents, each of whom recognizes the other as recognizing her. (Although Darwall’s main historical referent for this relationship is Fichte, in fact this is more familiar to most of us as Hegelian mutual recognition.) Vocatives have the characteristic function of calling another person into such a relationship of mutual recognition. They initiate that relationship by expressing recognition of the other, and they call for its continuation by asking for recognition in turn. They thus function to forge second-personal relationships of the sort Darwall describes. Darwall, like Hegel and presumably Fichte before him, points out that mutual recognition presupposes and acknowledges the personhood of each party; we recognize the other as one capable of recognizing in turn, and hence as a being with a standing and a perspective in normative space. Darwall does not talk about vocatives, but he does attach a great deal of importance to Strawsonian reactive attitudes. Reactive attitudes, he points out, recognize the one to whom they are a reaction as a person who can in turn recognize and respond to this reaction, by holding that person responsible. Vocatives are speech acts that crystallize and express this normative pragmatic structure. While not all vocatives reflect a judgment of moral responsibility, all reactive attitudes serve a vocative function. Vocatives, like speech acts in our other categories, need have no particular surface grammar: though terms such as ‘hey’ and ‘yo’ explicitly mark the vocative mood, a vocative can take the form of anything from a slight nod of the head, to a formal greeting, to a declarative utterance whose point, in context, is to recognize someone by calling him and demanding an appropriate acknowledgment (for instance, “I see a cute little boy in a red sweater!” directed at a toddler son by a parent who has just returned home from work). The semantic content of a vocative call may be more or less specific—we can call “Hey, you with the blue hat and the big nose!” or “Anyone who is willing to volunteer please raise your hand,” or just “Hi!”—But in terms of its pragmatic function, its job is to call upon you as this particular person who was recognized by the call. As Althusser puts it, in order for the call to be effective, its intended

Vocatives and Acknowledgments

143

target needs to recognize: “That’s really me being called.” Otherwise this audience cannot be effectively claimed by the normative demands that are differentially distributed by the call. A third person, who hears and notes the correctness of a call, is not claimed by it in the same way— there is no demand on her to recognize herself as having been called. This means that, in order for the call to reach its target, the one called has to recognize more than that the call occurred, and even more than that it has correctly identified her, in some funny third-personal sense. Rather, her response has to be essentially deictically indexed and firstpersonal so that she can be claimed by the norms that demand her appropriate acknowledgment in an inherently personalized way. The vocative has an agent-specific output rather than functioning as a generalized address. Because of the recognitive dimension of the hail, in order for it to succeed in reaching its target, it has to be heard as calling, not just anyone who happens to fall under a particular description, but me. In this sense, vocatives are inherently individuating; where perceptual episodes, we argued earlier, individuate at the level of input, vocatives serve an individuating function at the output end.8 Even when a vocative calls several people (“Those of you seated in rows ten to twenty may now board”) or even everyone (“Thou shalt not kill”), it is, as Heidegger might put it, “in each case yours.” Furthermore, not only is the call agent-specific and individuating, but its voice is inherently second-personal. A speech act can recognize and have differential normative implications for a particular person while functioning third-personally—for example, I may see Jim mistreating his dog, and on the basis of my recognition of this particular event of mistreatment, say, “Jim should stop mistreating his dog.” This is a prescriptive with an agent-relative output, and furthermore it is an output that individuates a particular target—it imputes a normative commitment to Jim, and not to anyone who happens to be mistreating his or her dog. But it is a third-personal speech act, and not a call to Jim.9 Here 8. In chapter I of division II of Being and Time, Heidegger is concerned with just this individuating character of calls, which is why his call of conscience is, despite its lack of content, a call to authenticity, or a call to the subject to ‘be herself’ and not some other self. 9. At least not at its most straightforward level of function. Since we will be arguing in the next chapter that all speech acts have a vocative dimension, this particular claim will have to be precisified later.

144

‘Yo!’ and ‘Lo!’

we can see that the audience of a speech act and the target of its output are not necessarily the same, which is part of why agent-specificity and second-personality are not the same property of speech acts. However, vocatives, unlike prescriptives, necessarily call to the person they recognize and in doing so call upon that person to recognize that recognition. Thus a speech act counts as a vocative only if it is a second-personal call to you (and, perhaps, you and you). Some quasi-vocatives involve variations on the paradigmatic, recognitive structure that we have described. Some are what we might call ‘conditional vocatives’; instead of responding to a person recognized through the receptive faculties of the speaker, these speech acts seek their target. When Mark comes home and calls out, “Hello? Are you home?” he is not hailing Amy by recognizing her, but seeking to discover if she is there to be hailed. If Amy is home and hears this call, then she will have been called upon to acknowledge herself as having been called in the traditional vocative fashion; furthermore, Mark’s call still recognizes Amy as the proper target of the call. Such conditional vocatives are fairly benign tweaks on the traditional vocative structure. A more interesting variation on the vocative is an interrogative that hails by seeking an as-yet-indeterminate target. When someone calls, “We need help! Is there a doctor on the airplane?” this is not a call to anyone in particular, even if it successfully targets someone who is in fact there. Yet this type of call is still clearly parasitic on the traditional vocative. It is not in the first instance a recognitive, but it still seeks the kind of acknowledgment appropriate to a vocative, and, interestingly, if successful it will function as though it had been a recognitive for the one called, who will respond out of the recognition that “that’s really me being called.” (“I am a doctor! They’re calling me!”) Even further from traditional vocatives are those that function not to discover a proper target but to create such a target. Much advertising has this structure. An advertisement that reads “Are you over thirty and worried about premature wrinkles?” or even “Now that you’re over thirty, you’re worried about premature wrinkles” seeks to marshal the “that’s really me being called” response, in order to constitute a proper audience upon whom it places a normative demand (to buy the wrinkle cream). In recognizing myself as targeted by the vocative, I become (really) someone over thirty who is (now!) worried about premature wrinkles. Such speech acts are not recognitives in any straightforward

Vocatives and Acknowledgments

145

sense—we might call them constitutive misrecognitives.10 However, once again, their functioning is parasitic upon the functioning of traditional recognitive vocatives, and when they successfully grab their target they will be received as recognitives that demand appropriate acknowledgment. We will return to such vocatives in detail in the final chapter, arguing that this constitutive function is extremely important to enabling the possibility of a discursive community and placing individuals within this community; we mention them here only for the sake of completeness.

6.3 Acknowledgments An acknowledgment, of the sort we are interested in here, is a (more or less) explicit taking on of the normative status and responsibilities demanded of one by a given speech act. In acknowledging a speech act, whether a vocative or some other kind of utterance, you give expression to your normative uptake of its output—you mark your claiming of the commitments and entitlements that it imputes to you. You can, for instance, acknowledge an imperative by uttering “okay” while carrying out whatever action was demanded of you. You can acknowledge a declarative by marking that you accept that you are committed to its truthclaim, for instance by uttering “You’re right”11 or “Oh!” You can acknowledge a vocative in various ways: by returning the hail, nodding your head, etc. Testifying provides a lovely example of acknowledgments at work. The preacher calls out “God is great,” and the congregation calls back “Amen!” or “I hear you!” The congregation’s speech acts do not (just) reassert the preacher’s claims—they give expression to the uptake of the claims. When a preacher calls out and demands an acknowledgment, 10. For detailed analyses of this type of constitutive vocative see Rebecca Kukla, “Myth, Memory and Misrecognition in Sellars’ ‘Empiricism and the Philosophy of Mind’,” Philosophical Studies 101 (2000): 161–211; Kukla, “Talking Back: Monstrosity, Mundanity and Cynicism in Television Talk Shows,” Rethinking Marxism 14 (2002): 67–96; and Kukla, “The Ontology and Temporality of Conscience,” Continental Philosophy Review 35 (2002): 1–34. 11. “You’re right” is a bit of colloquial English that nicely incorporates several features of declaratival acknowledgment. It functions as an anaphoric reassertion, in that it entails commitment to the content of the declarative (“You’re right,” that is, pragmatically entails “That’s true”). Furthermore, the phrase is usefully ambiguous between the normative recognition of justified performance and the normative undertaking of the content of the speech act. So, as colloquially used, “You’re right” implies also “You’re justified.”

146

‘Yo!’ and ‘Lo!’

this is an explicit call upon the members of the congregation, not only to endorse the position (consider how underwhelming it would be for the congregation to respond “That’s true, Minister”), but to express that the preacher’s words have made a claim on them, and to express and highlight (and celebrate) their (new or renewed) commitment to this claim and all it entails. Vocatives stand in a dialectical relation of mutual dependence with acknowledgments. Vocatives, in essence, recognize second-personally by calling for acknowledgment of that recognition, and hence a vocative that goes completely unacknowledged has failed in its function. Any speech act can be acknowledged, and many speech acts call for acknowledgment as part of their normative output. For instance, Darwall points out that when a sergeant issues an imperative, she “expects from her charges a ‘looking back’ that reciprocates her address to them.”12 That is, she demands more than mere compliance; she expects acknowledgment of the normative status of her address. However, not only do vocatives put the demand for acknowledgment front and center, but they are also peculiarly self-referential: what they call upon you to do is specifically to acknowledge them. Furthermore, what they call upon you to acknowledge when you acknowledge them is that you were properly so called. The acknowledgment of a vocative recognizes that the vocative was appropriate, and that it reached its target, and it meets the demand for recognition of that vocative recognition. While conversations consist (usually) of more than just a string of hails and acknowledgments, they are sustained as conversations (rather than as a mere succession of speech acts) by an ongoing fabric of vocative calls for recognition and acknowledgments of these calls, which sustain the normative engagement of the interlocutors with one another and with the conversational project. Acknowledgments of vocatives serve as vocatives in their own right; they are targeted at the hailer and call for her to recognize that her demand for recognition has been met. One might worry that this begins a pragmatic regress of calls for recognition. But perhaps this ought to give us insight rather than worry; remember all the meaningless conversations we have been stuck in, which have involved seemingly unending

12. Stephen Darwall, The Second-Person Standpoint: Morality, Respect, and Accountability (Cambridge, Mass.: Harvard University Press, 2006), 91.

Vocatives and Acknowledgments

147

cycles of response, just because at no point has it seemed allowable to fail to acknowledge what has just been said. Failure to respond to a vocative with an acknowledgment is not merely rude; it is a pragmatic subversion—the hail has not succeeded in carrying out its function if it is ignored. And ignoring a hail on purpose can constitute a shunning tantamount to acting as if the hailer did not have the authority to issue that hail in the first place. In thus denying the authoritative status of the hail, we might be resisting this person’s right to call us in this way, because of either the normative positioning implied by the hail, the normative position of the one hailing, or both. For instance, a married woman who has kept her maiden name might refuse to respond when she is called by her husband’s name, even though she knows that it is she who is being called. Or a supervisor might refuse to respond to subordinates who call her by her first name rather than a more formal title. Ignoring is a complex attitude. The paradigmatic failure of a hail is when its normative upshot is not taken up and hence not acknowledged. But refusing to take up a normative demand requires recognizing that the demand has been made. To refuse a hail is already to recognize and acknowledge it as a hail in refusing it. We can refuse a hail only from someone whom we ultimately recognize as capable of addressing us.13 For instance, if I walk past a talking toy bunny in a store that “hails” me, I am not ignoring or shunning it in “refusing” to answer it—I simply don’t recognize the sounds it makes as a hail at all. Hence ignoring a hail is a degenerate form of acknowledgment that recognizes the personhood of the speaker and the claim made by the hail through the very act of rejecting that claim, and with it the speaker’s attempt to forge a relationship of mutual recognition with us. Because acknowledging involves recognizing that a normative status has been imputed to me appropriately, along with expressing my acceptance of that status, the acknowledgment is itself another type of recogni13. Colleen Fulton (in private correspondence) has made the interesting point that certain vocatives can be oppressive or objectifying precisely because they call for being ignored, and hence there is no escape from them or way of rejecting their force. Her example is catcalls, for which not responding actually seems to be the proper, expected response. (After all, one who responded would be a “slut.”) This makes such calls unshunnable, in an important sense. If you ignore a catcall, you have in effect provided the acknowledgment the call asks for, and hence you have, whether you like it or not, acknowledged the authority of the caller to call you in this way. As Fulton puts it, there is no way for the person called to “win the exchange.”

148

‘Yo!’ and ‘Lo!’

tive speech act. Here, though, what is recognized is not a fact or an object or a subject, but, in the first instance, the force of a normative claim. The input of the acknowledgment, like that of all recognitives, is inherently agent-relative and individuating: what I acknowledge is the claim that a speech act makes on me given my particular position as the audience of the speech act—and this is still so even when everyone is targeted by the speech act, either in virtue of its extensional scope or its agentneutral pragmatic structure. Only I can express my normative uptake of a claim. On the other hand, the output of an acknowledgment can be either agent-neutral or agent-relative. When we acknowledge a speech act— that is, give expression to our normative uptake of the claim it makes on us—we may do so by acknowledging the speech act to the speaker, or through a public act. Sometimes we acknowledge the claim a speech act makes without the specific identity of the speaker being essential to the normative force of the claim, and our acknowledgment takes the form of an agent-neutral, public expression. I might publicly acknowledge my responsibility for a crime of which I stand accused, for instance, or I might publicly acknowledge the truth of a scientific theory in response to a rival’s evidence. Both acknowledgments are expressed recognitions of the force of particular speech acts, but they are not directed at particular agents or dependent upon the identity of the original speakers, and they have agent-neutral outputs. At other times we may acknowledge a speech act that succeeds in making a normative claim on us because of the specific identity of the speaker, even though our acknowledgment itself is public and agentneutral. I may affirm my commitment to a particular political cause, perhaps, because it was Nelson Mandela who called upon me to commit to it, and I may do so without directing my acknowledgment at Nelson Mandela (who doesn’t even know that I exist) himself. These types of public acknowledgments have recognitive, agent-relative inputs (because uptakes are agent-relative statuses) and agent-neutral outputs, and hence they are box-2 speech acts similar to observatives. They include rituals designed to show—in a public, agent-neutral way—that one has taken up a normative claim: accepting the wafer during mass, donning a garish frosh-week sweatshirt during hazing, etc. Indeed, ele-

Vocatives and Acknowledgments

149

gantly, in the religious context, we call such practices of publicly acting out our uptake of normative claims observance, or being observant. Such public acts of acknowledgment are quite different speech acts from those that acknowledge to a speaker that I have recognized and accepted the normative force of her claim. The latter type of acknowledgment is inherently second-personal, and it is agent-relative in its output; it serves to complete a relationship of mutual recognition.14 Any speech act can be given such second-personal acknowledgment. Some speech acts can be acknowledged in either way. For example, I can respond to a colleague’s compelling argument either by saying “You’re right,” and thereby offering acknowledgment to her, second-personally, or by publicly expressing my acceptance of the commitments entailed by her argument. In contrast, by their very nature, vocatives call for secondpersonal acknowledgment and fulfill their pragmatic function only if they get it. Since their normative demand on us is that we respond to our recognizer with recognition in turn, we cannot even make sense of the idea of acknowledging the vocative in a public way that is not directed to the one who called us. If Mark says “Hi, Rebecca!” and Rebecca says to Richard, or to no one in particular, “Yup, Mark recognized me,” then Rebecca has not taken up the normative claim that Mark’s hail made on her, which was a claim to a returned recognition. Hence she has not in fact acknowledged the vocative, publicly or otherwise. She has (rather snarkily) reported on its propriety, rather than expressing uptake of it. Thus the acknowledgment of a vocative has an agent-relative output along with its agent-relative input: it acknowledges your claim on me, to you. Thus acknowledgments are recognitives, and they can belong in box 2 or box 4. At this point, our grid is as filled in as it is going to get in the course of this book; see Figure 8. Some speech acts are first and foremost explicit acknowledgments— for instance, “Yes?” in response to a vocative, “Okay” in response to an 14. For completeness’s sake, we should also mention the possibility of acknowledging a speech act second-personally to someone other than the original speaker, as, for example, when one acknowledges the appropriateness of “You may now kiss your spouse” by kissing him. This case should be distinguished from giving a speech act public, agent-neutral acknowledgment in the course of a conversation with some particular person, since of course speech acts with agent-neutral outputs can be (and usually are) spoken to some particular audience in the course of a conversation.

150

‘Yo!’ and ‘Lo!’

imperative, or “You’re right” or “Amen” in response to a declarative. Others function as acknowledgments more or less implicitly while doing double duty as another kind of speech act, for instance when we respond to a vocative with a returned vocative, or to a truth-claim by declaring an inference from that truth-claim. Sometimes acknowledgment doesn’t require a speech act at all, because the recognition and uptake of the change in normative status is manifested directly in action, for instance when we follow an order. (However, we shouldn’t conflate acknowledging an order by following it with merely doing what was ordered. One could do something that was ordered without knowing that

Input Output

Agent-neutral

1 Neutral input Neutral output Agent-neutral

Declaratives Categorical Imperatives

3 Neutral input Relative output Prescriptives

Agent-relative

Figure 8

Agent-relative

2 Relative input Neutral output Kantian judgments of taste, baptisms, some recognitives, i.e. observatives, some acknowledgments

4 Relative input Relative output Imperatives (promises, invitations, reproaches...), ostensions, some recognitives, i.e. vocatives, some acknowledgments

Vocatives and Acknowledgments

151

the order was given, or without recognizing the appropriateness of the order. Even doing what was ordered while noting that the order was justified is not sufficient to constitute an acknowledgment of the order. You could, for instance, recognize that the colonel has entitlement to order you to raise your hand, and also raise your hand without doing it as an acknowledgment of this imperative—perhaps because someone else whose authority you respect more than the colonel’s gave you the same order.)15 An acknowledgment in our sense never merely notes a change in normative status—it has to enact or express the uptake of the claim made by a speech act. This is essential to the recognitive character of the acknowledgment. Again, I do not acknowledge a hail simply by noting its propriety—i.e., by commenting, “She notes correctly that I am Rebecca.” Instead, I say “Hi” back, or smile and grunt, or do any one of the myriad of other context-dependently socially acceptable things that count as expressing my uptake of the claim that has been made on me by the vocative. Notice, then, that no set of mere assertions of belief, even about normative statuses, could ever amount to an acknowledgment, for an acknowledgment performs the concrete acceptance, in practical and not just theoretical reason, of a normative status imputed by a speech act.16 Anyone, whether targeted by a speech act or not, might be in a po15. Nor do we think that what is missing is a causal relation. Clearly the mere fact that the evaluational recognition causes the arm to raise—say by startling you in such a way that you raise your hands in a reflexive gesture of surrender—is not sufficient, but we do not think the Davidsonian strategy of filling in the “right sort” of causal relation is on the right track. 16. Gettier notoriously spawned an industry producing cases in which, in response to a knowledge claim, someone is willing to grant justification and belief, and acknowledge truth, but not willing to take the agent’s justification as itself her own grounds for taking-true; see Edmund Gettier, “Is Justified True Belief Knowledge?” Analysis 23 (1963): 121–123. Our analysis of acknowledgments shows the phenomenon of the Gettier gap between knowledge and justified true belief to be a species of a general pragmatic phenomenon which includes, for example, the gap between, on the one hand, acknowledging that someone gave an order and that she did so appropriately, and doing the thing ordered, and, on the other hand, following the order. Acknowledgments express recognition not only of the entitlement to a speech act and of the adoption of the normative statuses imputed by the speech act, but also of the force of this imputation itself. Whether or not we make it explicit in speech, if we don’t acknowledge this, we haven’t acknowledged the speech act as fully successful. In the case of knowledge claims, this means that the acknowledgment of a knowledge claim requires more than the acknowledgment of justified true belief, as contemporary epistemology suspected—but that ‘something more’ is not an external or a semantic condition, but a kind of pragmatic success.

152

‘Yo!’ and ‘Lo!’

sition to note that the speech act was entitled and successfully performed—the declarative was true, the imperative was appropriate, the vocative properly recognized the one it hailed—and to note this evaluation in speech. While we can evaluate the propriety of a speech act with a third-personal declarative (“Sue had every right to tell Joe not to buy that car”; “Sarah’s version of the proof was correct”), we can acknowledge our uptake of the normative claim made by a speech act only by way of a speech act with an agent-relative input. (Only Joe can acknowledge Sue’s telling him not to buy that car, even while we can stand by and note her propriety in doing so.) The necessarily agent-relative, individuating input of acknowledgments will be significant in the next chapter. What it shows is that while many speech acts have agent-neutral outputs, all speech acts, insofar as they call for normative uptake and acknowledgment, must call upon particular individuals to acknowledge them, and therefore must to that extent have agent-relative outputs. So, for instance, even if a declarative makes an agent-neutral truth-claim that belongs in public space, the way it calls upon me to give it normative uptake is never interchangeable with the way it calls upon you to do so. Only I can acknowledge its call upon me; only you can acknowledge its call upon you. Any kind of speech act can be acknowledged. However, acknowledgments bear a special relationship to vocatives. Acknowledgments of vocatives crystallize and purify the acknowledging function of discourse, because acknowledgment just is what vocatives call for. The acknowledgment of a vocative is self-referential, in the specific sense that the normative uptake to which it gives expression just is the uptake of the demand for this expression. Indeed, the second-personal vocativeacknowledgment exchange, in its pure form, just is the discursive expression of mutual recognition, in its pure form. This equivalence will be important to us in our final chapter, when we argue that, as Hegel thought, mutual recognition is a constitutive condition for the possibility of a community of agents subject to the claims of the world and of one another. First, in the following chapter, we argue that the secondpersonal vocative call for acknowledgment, whose function is to forge such mutual recognition, is an essential pragmatic component of all discourse.

7

The Essential Second Person Riddle: “What do you call a boomerang that doesn’t come back to you?” Answer: “A stick.” —Anonymous

In The Dialogic Imagination, Mikhail Bakhtin argues for what he calls the “internal dialogism” of speech: “The word in living conversation is directly, blatantly, oriented toward a future answer-word: it provokes an answer, anticipates it, and structures itself in the answer’s direction.”1 This structural call for an answer is the essence of the vocative. In this chapter we argue that in order for a speech act to perform any normative function—that is, in order for it to count as a speech act at all—it must have a vocative function, in addition to whatever other functions it has. This vocative function is a condition for the possibility of the speech act’s doing or meaning whatever else it does or means—or as we shall put it, all speech acts contain a transcendental vocative. Speech acts not only strive to make normative claims upon those whom they target, but they call second-personally upon those to whom they speak to recognize themselves as bound by these normative claims, and to acknowledge this uptake. That is to say, to speak is to hail. Because vocative speech is necessarily second-personal, this means that, on our account, language has an essential second-personal pragmatic dimension. In Chapter 2 we argued that the ability to mark the first-personal perspective in language in a way that deictically attaches speech acts to points of view is essential to the possibility of discourse. 1. Mikhail Bakhtin, “Discourse in the Novel,” in The Dialogic Imagination: Four Essays, trans. Vadim Liapunov (Austin: University of Texas Press, 1982), 280.

153

154

‘Yo!’ and ‘Lo!’

We also claimed that Brandom did not have the resources to make sense of the pragmatic structure of this first-personal ownership of speech acts, because he treated normative commitments and entitlements exclusively as impersonal scorecards rather than as inherently owned by embodied subjects with points of view. In this chapter we make a parallel claim, not about the owner of a speech act, but about its target. The ability to mark the direction of a speech act within language is essential to the possibility of discourse. The direction of a speech act cannot be reduced to or cashed out in terms of any combination of its impersonal and contextual features; rather, this direction is a basic structural feature of the speech act itself. Any account of the pragmatics of discourse must build in from the beginning the fact that speech acts are directed (though not necessarily direct) transactions between agents, as opposed to just shifts in abstract scorecards of commitments and entitlements. Our argument in Chapter 2 was that discursive communities have to have the capacity to express their first-personal uptake in recognitives, and hence that the ability to express the first-personal perspective is essential to discourse. There, we did not draw the conclusion that every speech act has to give expression to this first-personal perspective, but only that it has to be possible to give such expression in speech. At the same time, as Austin and others have made clear, any particular speech act will depend for its felicity and success upon a host of commitments and statuses of various kinds, and so our capacities to perform speech acts have a generally holistic, interdependent character. Our claim in this chapter is quite a bit stronger in form than either of these: here we claim not just that any language must have the resources to enable directed, second-person speech, nor even just that the ability to produce some vocatives is a holistically necessary condition upon the ability to engage in discourse at all. Rather, we claim that each felicitous speech act contains a vocative, second-personal call. Although we will consider some partial exceptions to this universal claim later in the chapter, the exceptions will turn out to be derivative variants on the basic vocative structure of speech; they will not detract from the fundamental role that the vocative call plays in making it possible for a speech act to execute any discursive pragmatic function. Now this is not, of course, to make the absurd claim that the secondperson voice is the only voice in which we can speak. Many declaratives, prescriptives, etc., are clearly spoken in the third-person or impersonal

The Essential Second Person

155

voice. Our claim will be that speech acts have a second-personal, vocative function in addition to whatever other function they have. There is, as we have seen, nothing odd about a single utterance serving more than one pragmatic function. The utterance “That music is awfully loud” can function simultaneously as a declarative description of the sound level (in a third-person voice, with an agent-neutral input and output) and as an imperatival order to turn it down (in a second-person voice, with an agent-relative input and output). “How’s it going?” can serve as an interrogative, a vocative, and an acknowledgment of a vocative all at once. We seek to uncover the transcendental place of the second-person voice and its vocative call within language, not to exclude or diminish the many other voices and functions that make up discourse.

7.1 Concrete Habitation of the Space of Reasons Throughout this book we have taken it as a core principle that what a speech act—as a material act performed by and among agents within a discursive community—does is to draw upon the normative entitlements of its speaker in striving to change the normative commitments and entitlements of others. As such, in speaking, we make normative claims upon others. A genuine speech act can have an agent-relative or an agent-neutral output but it cannot have a null output, which is to say that it must seek to make claims on someone (though it may not succeed). Once we situate speech acts within a social context, the insight that they must seek to make a claim on someone can appear fairly trivial. But this, in and of itself, does not yet show that this claim-making is essentially second-personal. For, we might ask, why can’t it just be true that a speech act by person A makes a claim upon person B, without the speech act being second-personal? Even once we recognize that all speech is structured so as make a claim on someone, we might still think that it can be an objective fact that such a claim has been made, without the speaker attempting to transmit the claim through a second-personal call. When the government changes the tax laws, the claim on me to pay taxes changes, whether or not I receive a second-personal communication informing me of this. And every legal play in football changes the normative status of every other player, but the plays are not directed at anyone. The hard part will be arguing that, in speech, claims must be issued in an essentially second-person voice, and that they must seek up-

156

‘Yo!’ and ‘Lo!’

take and recognitive acknowledgement of that uptake. We must show that speech acts not only make claims on someone, but are essentially directed claims made by me, on you. Since the beginning of this book we have portrayed speech acts as striving to change the normative status of others. Performing an utterance does not magically transform the normative status of others in the community without any mechanism or possibility of failure; rather, it is through concrete communicative interactions that involve holding one another responsible, granting entitlements to one another, and making demands upon one another that our speech acts strive to effect their normative functions. Normative statuses are not abstract entities that shift around in some kind of ideal space—on some great, abstractly characterizable Platonic scoreboard, as it were. To think of language in this way would undercut the point of putting pragmatics front and center, which requires recognizing language as a concrete normative phenomenon, grasped in the first instance as a body of skillful interactions among speakers. But in order for my performances to constitute discursive speech acts that are entitled by my normative positions and that make normative claims upon others, at least two conditions must be met. On the one hand, I must have a determinate normative position within the space of reasons; I must be located, not just inside the space of reasons, but at some particular place inside it. On the other hand, my speech acts must constitute interactions with particular other people upon whom I make claims. No normative scorecards will actually shift except through the material efforts of determinately located speakers making claims upon other determinately located speakers. Yet when authors such as McDowell and Sellars speak of our habitation of the space of reasons, there is something oddly missing in their use of the metaphor. While the space of reasons is richly articulate in the sense that it displays normative and rational structure (by definition), it doesn’t seem, in the work of these authors, to provide much by way of articulate locations for the people who inhabit it. We get the sense from their writing that one is in this space only by having access to the reasons that give it its structure; one is either in or out, but in contrast with typical spaces, one does not occupy any particular location within this one. However, we inhabit the space of reasons not just by being able to rec-

The Essential Second Person

157

ognize it, but by negotiating it—we use reasons, are claimed by them, are thwarted by them, and attribute them to others, for instance, and these negotiations help fix our particular normative position within this space. Each of us will occupy a unique, non-fungible place in this space, if only because we are open to at least slightly different deliverances of sensibility, and hence entitled to different observatives (though our place will also be unique because it is articulated by every interpersonal relationship, every social role, etc., that makes a normative claim upon us). Brandom gives us language for talking about individual locations within the space of reasons, by introducing his ‘scorecards’ of commitments and entitlements, which mark the social distinction between undertaking and attribution. Out of all the normative statuses that there are, Brandom suggests, some particular set of them attaches to each agent, and this set defines her place in normative space. Furthermore, we can ‘move’ others around in this space by altering their scorecards through the claims we ourselves make—we can pass entitlements on, impute commitments, and so forth. But what does it take for a normative status to belong to me? As far as Brandom’s account goes, my scorecard simply attaches to me like legal property. He tells us nothing about the pragmatic relationship between me and my scorecard that makes the scorecard meaningfully mine, rather than just somehow correlated with me. Nor does he explain the “motion” of statuses: how your utterance of a declarative, for instance, manages in practice to pass on an entitlement to me and thereby change my place in normative space, by making a claim upon me and demanding my uptake of a normative status. In Chapter 5 we extended John Perry’s argument for the essential indexical, arguing that in order for my commitments and entitlements to exert governing force upon my practices, I must have a practical, firstpersonal grasp that they are mine. This perspectival grasp of the space of reasons is a logical condition for any of my statuses making a difference to me at all. In other words, normative space requires essentially indexical knowledge for its negotiation analogous to what Perry showed was required for the negotiation of material space. Recognizing all the commitments and entitlements that attach to subject position x (including the commitment to be bound by one’s commitments, etc.), and recognizing that Mark is the inhabitant of subject position x, gets me,

158

‘Yo!’ and ‘Lo!’

Mark, no closer to genuine normative commitment. It is only when I first-personally recognize that I am the one who is so committed that I enter practical normative space. Hence I must not only recognize my commitments and entitlements but also have practical, perspectival uptake of the fact that they are mine—they commit and entitle me. Nor is our practical grasp of our first-personal relationship to our own normative statuses a kind of knowledge that is a required accompaniment to these statuses. Rather, our normative statuses are not in a position to have a hold over us at all, and cannot be said to make a normative claim on us, except insofar as we have such a first-personal grasp of our own place in normative space. Thus the point here is not the reasonably familiar one that normative commitments must be materially enacted and not just theoretically attached to us, but the more specific one that such material enactment requires first-personal ownership of our place in normative space in order to be so enacted. Yet the ability to locate ourselves first-personally in normative—or for that matter material—space is not sufficient to render our knowledge of such space practically deployable. For crucially—although Perry does not make this point—in order to usefully place ourselves in any space we need to be relationally anchored. For all the theoretical knowledge I want plus the first-personal knowledge of where I am on a theoretical map will still leave this “I” an empty point with no substantive or usable relationships to the rest of the world. That is, knowing my place on the map provides me with neither scale nor orientation in a form deployable in practical inference. I am here now, and there is a bottle three feet off to my right. But how much is three feet and which way is right? What I need is both a place on the map and some practical ability to orient myself from this place to some other place, in order to have any practical knowledge at all. In the case of my practical placing in a world of things, then, I need both first-personal knowledge of where I am on the map and also practical knowledge of how to find something else from here. I must understand not only that “I am here,” but also that that is the lake, she is over here, this is how long a foot is. I must grasp how other things are located and locatable relative to me. Indeed, I need a great deal of such relational and demonstrative knowledge in order to give any usable bite to my theoretical and indexical knowledge. But we can run a parallel argument for the ability to negotiate normative space. In order to have a place in the space of discursive reasons, we

The Essential Second Person

159

have seen, I need to understand not just some abstract list of normative statuses, but that some of these statuses are mine and have a governing hold over me. What does it take to have such practical, first-personal grasp of a normative status? Among other things, it requires understanding how the claims of others make claims on me, and likewise, the kinds of claims that I can make on others. In order to grasp the import of an imperative or a vocative, for instance, I must understand when such a speech act is directed at me. Likewise, to understand my position in normative space is to understand, in part, to whom I can legitimately address various speech acts and what sorts of claims I can make upon which others. I must know not only that I am entitled to speak differently to my children than to others’ children, but also that this is my child. Given the inherently communicative nature of discourse, without such a relational understanding of others’ positions in normative space, my understanding of my own position is reduced to something empty and undeployable. Even when we are primarily dealing in declaratives, negotiating a conversation involves grasping that I disagree with Rebecca by saying P, I agree with Mark by saying Q, and so forth. Hence the practical, deictic understanding that is a condition for habitation in the space of reasons and participation in discourse includes understanding of my own position in this space, and, interdependently, understanding of others’ positions in relation to me. From all this it follows not only that my grasp of the space of reasons must be both first-personal and relational, but also that I must have an understanding—as both user and recipient—of the directedness of speech. For in order to deploy all this perspectival and relational knowledge, I must be able to direct my speech acts to others and to recognize when they have been directed at me. Knowing that a speech act can be directed at me, or at others, is not equivalent to knowing how to direct it at others or how to hear that it is directed at me. In Chapter 2 we argued that Brandom’s account of the pragmatics of language, with the primacy it accords to declarative utterances, did not have the resources to make sense of how commitments and entitlements could be first-personally owned, nor of how such ownership could be expressed in language. For all of the same reasons, his account also cannot make sense of the second-personal, directed force of speech acts. His speech acts make a difference to others, but they do so agent-neutrally and automatically—for Brandom, shifts in normative commitments and

160

‘Yo!’ and ‘Lo!’

entitlements simply happen because of speech acts, and while he surely grants that this ‘happening’ is incarnated in a material process, he leaves it as a separate question for others to worry about how this material process might go. He realizes that this process occurs through communicative acts that we direct at one another, but for him, this is contingent. It strictly makes no difference to my normative position, for Brandom, whether I am told things or whether they are simply ‘said’, in some impersonal way (see section 3 of this chapter for a discussion of this distinction). What we have uncovered, however, is that the normative discursive practices that Brandom describes cannot exist at all except insofar as they occur between speakers with a rich, relational, deictic grasp of their position in the space of reasons, as well as a grasp of first- and secondperson voice and the directedness of utterances. We concretely inhabit the space of reasons first- and second-personally, and not just as spectators. So far we have offered no argument that each speech act has a secondpersonal component, let alone a specifically vocative component. What we have shown, rather, is that it is a global condition upon being a member of a discursive community that one be able to locate oneself firstpersonally, address others second-personally, and understand oneself as the second-personal target of others’ speech acts. Thus competence at second-personal discursive interactions is a necessary condition for linguistic competence in general.

7.2 Second-Person Speech Earlier we explained that the second-person voice that interests us is not that which is marked in the surface grammar of a sentence. Sentences with a second-personal grammatical structure can function, at the level of pragmatics, as impersonal declaratives (“You are wearing a red sweater”), and utterances that do not include the word ‘you’ or its cognates can function second-personally, by having an address to another person built into their pragmatic function (“Please close the door!”). Not all speech acts that are, in fact, directed at another person are primarily spoken in the second-person voice, in this pragmatic sense. If they were, then given the communicative nature of language, there would be no room left for third-personal speech. Rather, some but

The Essential Second Person

161

not all speech acts primarily execute a normative function that cannot be coherently understood except as a directed address from a speaker to a particular audience. For example, all imperatives are second-personal in this sense. As we saw earlier, an imperative must be issued to someone in order for it to count as an imperative at all. The idea of “translating” an imperative into the third or first person while retaining its meaning or force does not get any conceptual traction. “Close the door!” makes a specific demand upon someone in particular (or upon several particular people) by addressing the target of this demand. A “translation” into the third person, such as “Mark ought to close the door,” is a prescriptive with a different pragmatic structure and function; it does not constitute an order at all. A speech act can have normative implications for someone even if it is not structurally directed to that person as an address. For instance, declaratives, we have argued, have normative implications for everyone. If I take a speaker to be authoritative, I can recognize myself as having new commitments and entitlements simply by overhearing a declarative utterance of hers, even if it was not particularly addressed to me. Prescriptives have agent-relative normative implications for the person about whom they make an ought-claim, whether or not they are uttered to that person. But the way in which an imperative (for instance) makes a claim on someone by addressing her is not simply reducible to the fact that it has normative implications for her, nor even that it has an agentrelative output for her. If I tell Richard “Mark should close the door,” my utterance (assuming it is properly entitled) has normative implications for Mark, but it does not address Mark. But if I tell Mark “Close the door!” I make a claim on Mark by addressing him (assuming, again, that my utterance is properly entitled). Indeed, I make a constative claim on him, imputing a commitment produced by the act of address itself. The imperative is second-personal because its normative function is carried by a directed address. More generally, second-personal speech acts are those in which the act of addressing is central to the normative function of the speech act. But what is it to address someone in speech, as opposed to simply speaking in a way that makes a normative claim upon her? An address does not merely shift the normative status of its target; rather, it makes a demand. It calls upon its target, not only to recognize the force of the normative claims made upon her by the speech act, but also to acknowledge

162

‘Yo!’ and ‘Lo!’

her uptake of these claims to the speaker. As Darwall puts it, the secondpersonal address is a summons to respond by recognizing the force of the speech act.2 When we address someone, our speech has misfired if that person does not acknowledge the address in turn. Of course, not every address calls for an acknowledgment in the form of an explicit speech act. As we pointed out in Chapter 6, acknowledgments often take the form of actions; one typically (though not always) acknowledges an imperative by simply following it, for instance. All the same, a suitable acknowledgment of an address must be a recognitive that is itself directed at the addressing speaker. Second-personal speech forges a transactional normative relationship with the target of that speech, and asks that target to participate appropriately in that relationship. Part (though usually not all) of the participation it demands is reciprocal acknowledgment of the normative uptake of that speech. Speech that does not call for such acknowledgment fails to be second-personal. For example, you do not actually issue an imperative if you whisper “Close the door” either to no one in particular or “to” someone from whom you could in no way hope to receive acknowledgment. Rather, in issuing an imperative, you must recognize an appropriate target for your imperative, and demand something of him, and in demanding you must call upon him to recognize himself as the one targeted by your demand, and to respond appropriately, expressing his normative uptake of the demand. His response to your imperative, whether it is compliance, pointed refusal, or anything in between, serves as a second-personal acknowledgment of your speech act. But this is just to say that the addressing function of language is interchangeable with its vocative function. The vocative moment in secondpersonal speech is what makes the difference between a speech act directed at you and a speech act that merely has normative implications for the person who happens to be you. If you understand the semantics of a second-personal speech act such as an imperative, but do not practically recognize that you are its target, you have not in fact understood its import. As Darwall repeatedly points out, the address is a call for mutual recognition. The vocative, as we saw in the last chapter, is the discursive distillation of such a call. Hence all second-personal discourse per2. Stephen Darwall, The Second-Person Standpoint: Morality, Respect, and Accountability (Cambridge, Mass.: Harvard University Press, 2006), 161.

The Essential Second Person

163

forms a vocative function, albeit usually (or perhaps always) an impure one; indeed, it is this vocative dimension that makes it second-personal. All structurally second-personal speech acts—imperatives, ostensions, promises, entreaties, invitations, and so forth—therefore have a vocative function. Imperatives include a vocative call that seeks acknowledgment in the form of uptake of that which the imperative demands. This is true even of alethic imperatives, wherein I call you to do something you were committed to doing anyhow. In this case, if you fail to honor your commitment, you now fail me, and not just the commitment itself. Since it turns out that the second-personal address and the vocative call are one and the same thing, the claim that all speech has a vocative function and the claim that all speech has a second-personal dimension are in fact equivalent. Our task, then, is to show that even speech acts whose primary voice is not second-personal—for example, declaratives, observatives, and prescriptives—must at the same time function as second-personal addresses.

7.3 Tellings, Holdings, and Transcendental Vocatives Brandom offers us a framework for thinking about speech acts as normative functions: what a speech act does is transform the commitments and entitlements of those who fall under the scope of its output. We have filled in this framework by arguing that such functions must be incarnated within a concrete discursive space structured by first-personallyowned normative positions and second-personally-directed speech acts. A concrete speech act is a performance by an agent who is positioned first-personally in discursive space, and who strives, at least partly through directed speech, to change the status of others who are also so positioned. Furthermore, competent speakers must be skilled at recognizing the direction of various speech acts, and concomitantly, at recognizing when they are themselves the second-personal target of a speech act. In such a concrete space thus articulated by owned subject positions, a speech act succeeds in having any pragmatic force at all only when (1) it is the kind of thing that can be concretely recognized and taken up as having such force, and (2) it is part of the structure of the act that it seeks to be recognized and taken up in this way. If it is not part of the structural aim of a speech act to make a claim on someone and demand recognition of this claim, then that speech act fails to have any actual,

164

‘Yo!’ and ‘Lo!’

lived pragmatic force at all; part of what makes a speech act a claim is that it seeks normative uptake from agents capable of recognizing normative claims. This seeking attaches to the performance of the speech act itself, regardless of whether the speech act in fact gains all the recognition it ideally seeks. This is to say that the speech act strives to change the normative statuses of various agents, even though many of them will never be in a position to recognize its claim. (As we saw, this is obvious in the case of a warranted declarative, which rarely receives universal uptake.) But how, then, can we maintain our core pragmatic insight, that successful normative changes can never be mere shifts in Platonic status, but must rather be registered in the embodied uptake of concrete agents? The only possibility is that any given output of the speech act—whether or not this output is recognized by everyone who falls under its scope— must be constituted by someone’s concrete recognition. Interestingly, the person who gives the speech act concrete recognition need not always be the person most directly affected by its claim. For instance, if I tell my lawyer that Richard is the beneficiary of my estate, then Richard’s normative status changes in an agent-relative way, even if he is in no position to give my speech act recognitive uptake. However, I have not succeeded in changing Richard’s normative status if nobody, including my lawyer, gives my speech act this uptake. (At other times, the person from whom a speech act seeks concrete recognition must coincide with the person targeted by the speech act, for instance in the case of imperatives.) A speech act, that is, succeeds in having real normative consequences only if someone successfully recognizes it. Therefore, speech can structurally seek to alter the normative status of anyone by actually holding someone accountable for its uptake. (Plenty of speech acts are unsuccessful, and strive for recognition while being taken up by no one and changing no one’s normative status—speech acts may go unheard, their legitimacy may be rejected, etc. But most speech acts must be at least partially successful in order for discursive practices as a whole to get off the ground.)3 3. Prayer, for instance, is a type of second-person speech act that always fails to receive the acknowledgment it seeks. If someone prays silently, or by herself with no one there to hear, then her speech act will totally fail to be recognized. But it still strives for recognition and acknowledgment. Working out the pragmatics of prayer as a distinctive type of speech would be interesting, for either a theist or an atheist. It is not clear if prayer is best understood as a regu-

The Essential Second Person

165

But we saw in the previous chapter that holding accountable is itself a second-personal normative act that establishes and engages others in a normative relationship. Indeed, it is just the act that constitutes the vocative call. I can alter someone’s normative status third-personally, but I can hold her accountable for this status only second-personally, by recognizing her as a target of my speech act and calling her to recognize herself as targeted by it. Abstract ‘scores’ become genuine normative statuses only when they are taken up in practice. In speaking in a way that actually has pragmatic force within a discursive space structured by first-personal positions and second-personal relations, in other words, we speak by calling to others to recognize and take up the force of our words for them. Whatever else speech does, including drawing upon agent-neutral entitlements and making agent-neutral claims, it does it by seeking to forge such a relationship of mutual recognition between speaker and target audience through a vocative call. This is the transcendental vocative moment that underwrites all other pragmatic functions of speech. In some sense, QED. This conclusion is hardest to accept in the case of declarative speech acts with agent-neutral inputs and outputs, which seem to be paradigmatically impersonal rather than second-personal. The pragmatic function of a declarative is by definition not anchored in entitlements that are specific to any particular agent, nor is the claim it makes specific to any particular audience. Hence it seems a stretch to claim that any concrete, functional declarative must involve a second-personal call. The trick is to show how even a speech act with an agent-neutral input and output can have a built-in vocative function. In other words, we need to show that the vocative function of a speech act is necessary, regardless of the primary structure of its input and output. If we can prove this necessity, then that ought to be sufficient to convince us that other nonsecond-personal speech acts such as observatives and prescriptives also have an analogous vocative, second-personal dimension. In “Getting Told and Being Believed,” Richard Moran explores the pragmatics of telling. When I tell you something, I do not merely issue a declarative within the range of your hearing. Rather, I offer you my word lar attempt to converse with a (nonexistent) being, or whether it has a different structure, given that the purported target of the speech act is omniscient, not materially embodied, etc. We would like to take up this analysis, whose interest was pointed out by an anonymous referee, in a later work.

166

‘Yo!’ and ‘Lo!’

that something is true. I take responsibility for my words and give you the entitlement to hold me to this responsibility. In telling you something, then, I invite you into a relationship of trust and normative responsibility. Edward Hinchman, arguing similarly in “Telling as Inviting to Trust,” distinguishes the way that we merely make an entitlement available in asserting, from the way that we offer an entitlement in telling. Brandom claims that assertions pass on entitlements, but he is concerned only with assertions that, as it were, are offered up impersonally into public space. Moran and Hinchman suggest instead that entitlements to declaratives are actively passed on through tellings, and not just impersonal assertions (although an entitlement can certainly be picked up from an impersonal assertion, without it’s having been specifically offered).4 Moran points out that often my reasons for believing what someone tells me can be internal to the second-personal relationship between us. I may accept the call to trust that is built into her act of telling something to me. When someone tells me something, Moran argues, I can take her words merely as evidence for the truth of what she tells, or I can do something quite different: I can accept the normative commitment that she makes in her act of telling. In doing so, I perform an act that is not just belief-formation but the establishment of a normative relationship of commitments and responsibilities. We “place ourselves in another’s hands,” as he puts it, rather than just adding to our evidence base. “It is the special relations of telling someone, being told, and accepting or refusing another’s word that are the home of the network of beliefs we acquire through human testimony. And these relations . . . provide a kind of reason for belief that is categorically different from that provided by evidence.”5 Brandom and others recognize that any issuing of a declarative (or any other speech act) requires taking responsibility for what I say. But in Brandom’s account of assertion, we do not particularly take on a responsibility to anyone in particular in asserting. In contrast, a telling, in Moran’s sense, involves taking on a specific normative responsibility to the person we address in speech. Telling forges a distinctive, second4. Richard Moran, “Getting Told and Being Believed,” Philosophers’ Imprint 5 (2005): 1–29. Edward Hinchman, “Telling as Inviting to Trust,” Philosophy and Phenomenological Research 70 (2005): 256–287. 5. Moran, “Getting Told,” 4.

The Essential Second Person

167

personal relationship that is different from the mere transfer of information—a transfer that may perfectly well occur merely because we overhear someone say something, though not to us. Tellings are pragmatically structured discursive acts that presuppose and constitute a relationship based not only on each party’s recognition of background norms and authority but also on the agent-specific acts of inviting and offering trust. Unlike pure declaratives, tellings have both agent-neutral and agent-relative inputs and outputs: only the speaker can offer his words as a telling, and only the person he addresses can take the address as a telling. Some third party can overhear the telling and take it as a perfectly good reassertable declarative, but she is not invited into the second-personal normative relationship forged by the act of telling.6 Such acts of telling are second-personal speech acts that call for theoretical belief. As such, they must at the same time have agent-neutral outputs. However, their function is to call for the uptake of these agentneutral outputs, and as we have seen, such uptake is always agentspecific. Hence insofar as the teller calls upon the told to put her trust in this particular directed speech act and the normative relationship of trust and responsibility in which it is embedded, it also has agent-relative outputs; my friend may call on me to trust her telling—to take myself as committed by her words in virtue of our relationship to one another—in a way that she would not expect to transfer to another listener. In turn, I may trust her because of our distinctive relationship, but this cannot make what she tells me true for me but not for others. Indeed, I demean and dismiss her act of telling if I claim to accept it as making a claim on me but not on others; to do so is to fail to take it as a truth-claim. In the language of Chapter 5, such tellings, while agent-relative in their input and outputs, have alethic rather than constative entitlements. Like holdings based on oughts, they call for recognition within the context of a second-personal holding, but they do so on the basis of the speaker’s entitlement to agent-neutral public facts. To treat the movement of normative commitments and entitlements to belief from speaker to speaker as rooted in such acts of telling is to understand second-personal normative relationships as a primary medium by which discourse—notably including declarative discourse—succeeds in changing normative statuses and thereby fulfills its structural func6. See ibid., 27, for a similar point not couched in our lingo.

168

‘Yo!’ and ‘Lo!’

tion. In contrast, Moran argues, in most contemporary philosophy of mind and language, “speech is seen as a kind of interpretable human behavior like any other,” whose normative implications are accessible from a third-personal, outsider’s perspective. Likewise, Hagi Kenaan claims that most philosophers employ a language “whose essence is ‘instilling information’, . . . a language indifferent to whom one speaks of, to whom one listens.”7 In contrast with Moran and with the picture we have just drawn, Darwall draws a sharp distinction between ‘moral’ and ‘theoretical’ claims, and flatly denies that the latter make second-personal claims. He insists that “moral obligations are . . . to others in a much more robust way than those of logic are.”8 But even if this is right, it does not follow that we hold one another second-personally only to moral obligations, and not to logical or theoretical obligations. Indeed, we seem to do the latter regularly. From our point of view, it is not the moral/theoretical distinction that cuts the difference between the obligations that are and are not grounded in second-personal transactions. Ostensive speech acts, for instance, can call upon you second-personally to attend to and recognize theoretical facts. When I address another second-personally, I call upon her to express normative uptake of my speech act through what she does and says. Whether this is a call to take up an epistemic obligation—to look, to infer, to reconsider, to believe—or to do something else seems irrelevant to the basic second-personal structure of the address. Darwall points out correctly that “by its very nature, belief is responsible to an independent order of fact, which it aims to represent in a believer-neutral way,” and thus that epistemic “authority is not secondpersonal all the way down,” in contrast to the kind of moral authority he is interested in, which “derives from normative relations that reciprocally recognizing persons assume to exist between them.”9 But the publicity of the facts to which beliefs are accountable is a separate issue from the pragmatic structure of the various kinds of speech acts that hold us to these facts. We saw in Chapter 5 that moral claims, including secondpersonal moral holdings, are themselves normally grounded in entitlements to public facts, namely facts about normative statuses. Darwall 7. Hagi Kenaan, The Present Personal: Philosophy and the Hidden Face of Language (New York: Columbia University Press, 2005), 47. 8. Darwall, Second-Person Standpoint, 27. 9. Ibid., 56, 57, 60.

The Essential Second Person

169

is certainly right that our entitlement to utter declaratives ultimately derives from relation-independent facts—or as we would put it, from agent-neutral facts; indeed this has been a major theme for us in this book. But this does not mean that in each speech act of giving reasons for belief, the second-personal dimensions of the act can be stripped away from its core normative function. As Moran makes clear, the call to trust and the offer of responsibility, or the invitation to let someone put herself in your hands with respect to her beliefs, is not reducible to the presentation of impersonal evidence for one’s trustworthiness or epistemic reliability. Tellings, like promises and apologies, are speech acts that are directed to another person, inviting him into a normative relationship.10 Speech acts that function primarily as second-personal tellings, in Moran’s and Hinchman’s sense, are invitations to trust based on a specific relationship between speaker and audience, even though what they invite their audience to trust is the agent-neutral entitlements of the speaker—entitlements that would be available to everyone, were it not for epistemic defect. This is the kind of speaking and listening that can go on between doctors and patients, teachers and students, parents and children, and even political representatives and their constituents. Here the second-personal dimension of the speech act is reasonably manifest. But at other times, we engage in purer versions of declarative speech: we draw upon our agent-neutral entitlements to speak not as one who has a special normative relationship to her listeners, but as a representative of the ‘we’ who has taken up a normative entitlement that, short of epistemic defect, is available to all. Likewise, even though in practice we can at most expect a few people to hear us speak, we may speak to others not as particular others bound to us by a particular second-personal relationship, but as representatives of the ‘we’ who are inheriting public entitlements. In such cases, we are speaking as anyone, to anyone. This is the kind of speech that we find in newspaper reporting, formal expert testimony, and academic writing. It is also the kind of speech that brings us closest to pure declarative speech that is agentneutral in both input and output. In such cases, it is particularly tempting to deny the presence of any second-personal, vocative dimension at all. (It is worth pausing to notice that, at this point, such speech seems 10. Moran, “Getting Told,” 23–24.

170

‘Yo!’ and ‘Lo!’

rather rarified and not particularly typical of everyday conversation, and yet it has consumed the lion’s share of philosophical attention.) Even in maximally impersonal declarative speech, it’s not quite right to say that I speak as anyone, to anyone. Rather, I speak as someone who has already taken up a normative entitlement that would be available to anyone under the right epistemic conditions—for why else would I have the authority to speak and be listened to at all? And my speech act strives to impart this entitlement to anyone who has not yet taken up this entitlement—for why else should I bother to speak at all? That is, when I engage in such impersonal, declarative speech, I issue a generalized call to take up the normative entitlement passed on by my speech. I recognize my audience, not as holders of a distinctive normative position, but as generic members of the ‘we’ who serve as legitimate recipients of the public normative statuses I pass on. As such, I call for normative uptake from them. I ask them to acknowledge my call, not by acknowledging my distinctive normative position, but by acknowledging my entitlement to speak as one who has already taken up my entitlement to public reasons. In such maximally impersonal declaratives, I speak as a representative of the ‘we’, to you as a representative of the ‘we’. This is a generalized version of the vocative structure that is common to all speech. Thus even in this most impersonal form of speech there exists an attenuated but genuine vocative dimension. We must be careful not to confuse the structural second-personal dimension of speech in general with the (occasional) intimacy of the speaker/audience relationship that enables it. For instance, in explaining the purported essential second-personal dimension of a particular speech act, Kenaan writes: “Your speech was not part of a neutral exchange between an unspecified pair of addresser and addressee who, in this specific case, happen to be us. Rather, it grew out of a meeting whose essence was precisely the meeting between you and I. It was born of the singular encounter between us.”11 But this kind of intimate particularity is structurally irrelevant to the issue of voice. I can speak secondpersonally in the course of neutral exchanges with total strangers (“Do you have the time?”), to people who I have never directly encountered (“If you have high cholesterol, try Lipitor”), and, as we have just seen, as a generic speaker who just happens in this case to be me to a generic au11. Kenaan, Present Personal, 144.

The Essential Second Person

171

dience member who just happens in this case to be you. In addition, I can utter speech acts that make no sense at all outside the context of my particular, concrete relation with another person and yet that are not second-personally directed at that person—for instance when I delight lovingly and protectively in my child’s quirks when describing her to someone else. Our argument here has not been about the intimacy or context-bound character of (some) speech, but rather about its inherent directedness and its structural call for recognition and uptake.

7.4 Speech as Communication and as Calling Given our analysis of addressing and telling, all of this is just to say that in order to actually incarnate their defining normative functions, speech acts must be addressed to a concrete audience, and as such they must have a vocative, second-personal dimension. Speech is inherently directed and responsive, as Bakhtin put it. But it is perilously easy to confuse our claim that all speech acts must function as second-personal addresses with an importantly different, far weaker claim, namely the completely obvious, uninteresting point that language is primarily used for communication. That language is fundamentally communicative was in fact our starting point; it would be rather anticlimactic if it turned out to be one of our main conclusions as well. Philosophers of language, of course, never deny that language is communicative, in that we use it in order to have effects on particular other people, through our interactions with them. However, according to traditional, impersonalist pictures of language, this function is strictly speaking external to the linguistic act itself. On such a picture, bits of discourse have their identities independent from their use in conversation, but we happen to use them almost exclusively for conversational purposes, and perhaps wouldn’t have bothered inventing language if we didn’t want to communicate (or, if you prefer, language evolved because of the selectional advantages of communication). Bits of language are almost always addressed to other people, but the address is not generally integral to their structure, on this view—it’s something we do with them. Consider an analogy: a house is the house it is whether or not anyone lives in it. However, we wouldn’t bother building houses did we not intend for them to be lived in, and the overwhelming majority of houses are lived in. Most philosophy of language treats conversation as

172

‘Yo!’ and ‘Lo!’

something—indeed the main thing—that we “do with” utterances, just as living in them is the main thing we do with houses. That the use of language in second-personal, conversational address is external to its structural identity, in most philosophy of language, is clear from the fact that almost all analyses of language proceed by bracketing this secondpersonal function entirely.12 In the case of philosophers who privilege semantics or syntax over pragmatics, the meaning or syntactic structure of a speech act is taken as self-sufficient and independent of its normal use in conversation. But notice that even for someone like Brandom, who privileges pragmatics and identifies speech acts by their pragmatic functions, the addressing function of speech is external to its structure. Brandom’s pragmatic analysis of speech acts not only focuses entirely on speech acts with agentneutral inputs and outputs, but also employs an ideal sense of commitments and entitlements; for his purposes, a (declarative) speech act shifts the (ideal) commitments and entitlements of everyone in the discursive community automatically. In this sense, it doesn’t matter at all, for Brandom, whether the speech act was actually addressed to or heard by the people it targets—it achieves its function and shifts the normative status of everyone in the community in the relevant way simply by being uttered. Such an account requires no directionality of discourse, nor any particular normative engagement between speakers. Brandom does not deny that we generally do care who in particular hears and can make use of our speech acts—it’s just that this actual uptake is not part of his story of the pragmatic individuation and analysis of speech acts. Likewise, the addressing function of language makes no appearance in the work of other pragmatists such as McDowell or Sellars. Interestingly, even Davidson, who entitles one of this papers “The Second Person” and privileges interpretive encounters between individuals in his analysis of language, restricts himself entirely to an observer’s perspective on the speech of another person, rather than discussing addresses to another.13 In contrast, we have argued that the addressing function of lan12. It’s hard to imagine how one could bracket the second-personal addressing function of vocative speech; perhaps tellingly, vocative speech barely shows up on the radar for philosophers of language. 13. Donald Davidson, “The Second Person,” Midwest Studies in Philosophy 17 (1992): 255– 267. What these views have in common is a conception of the normativity of linguistic perfor-

The Essential Second Person

173

guage is built into its defining normative structure, as a precondition for its having whatever other function it has. Addressing others secondpersonally is not something that we happen to do with language, even all the time, nor is it even just the reason we bother to have language— it is a transcendental condition of discourse as such. We have analyzed speech acts as holdings and demands for normative uptake, which themselves have a second-personal, vocative structure. The inputs and outputs that constitute the vocative dimension of speech are part of its defining form. To see the difference, it is perhaps helpful to look at some other acts that we do for and with others. I throw a ball so that another may try to catch it; I take off my clothes in order to entice someone to have sex with me; I drop my pants in the direction of the president in order that he might feel ashamed of his foreign-policy decisions. These are acts that are directed and second-personal. It would be stretching the boundaries of the notion to call these acts discursive, but they are certainly communicative. Now of course, I can throw a ball without another person there to catch it, and I regularly take off some or all of my clothes without trying to get anyone to do or feel anything. As such, these are actions that can be directed second-personally, and can be used to make normative claims on another that demand recognition, but they need not be. There is nothing about the action of throwing a ball that requires that it seek normative uptake from someone to whom the ball is thrown. The traditional picture of language treats speech acts as analogous to such inherently impersonal, contingently communicative acts. However, we can also think of these as examples of different, thicker actions: throwing a ball to someone (passing), undressing for someone (seducing), and dropping my pants at someone (mooning). We claim that throwing a pass is not just the same action as throwing a ball plus the extra intention that someone catch it, but rather a distinct activity. In the throwing of a pass, the recognition and response of the other are intrinsic goals of the action and are key to its success. At least as plausibly, undressing for someone, as an act of seduction, is a distinct action that is mances from the point of view of a “referee.” For Davidson, it is the interpreter who gathers data so as to postulate a truth theory, and if we do it ourselves, it is by taking on the strange stance of external theorist toward those we talk to (or sometimes even to ourselves, for Quine). Brandom focuses less on the figure of the external referee, but nonetheless on the product— universally changing scores—that would seem to be accessible only to such a referee.

174

‘Yo!’ and ‘Lo!’

not just the act of undressing plus the intention that this act have the effect of producing a desire for sex in someone. And so forth. These are actions that are intrinsically second-personal in their pragmatic structure. Indeed, insofar as we consider these acts in their thicker, second-personal sense—as addresses—they are incarnating normative functions, making claims upon and seeking recognition from a target audience, just as speech does. We claim that discourse must be understood as second-personal in this second, more robust sense; any act that can count as a discursive act at all must have this thicker structure. Now that we have argued that the vocative function is a universal, transcendental component of speech acts, we need to consider some possible exceptions and determine whether any backpedaling is called for. The most obvious type of apparent exception is a speech act that the speaker does not intend anyone else to know about, such as an insult or a command that I mutter under my breath, or an entry in my diary. We do not think that these are actually exceptions, and this is precisely because the intrinsic addressing function of language is not reducible to its extrinsic communicative function—speaking is not making noise along with some Gricean intention to get others to react. Such speech acts are ‘perverse’, in the sense that we design them to fail at their own structural function—but this is a common perversity we have seen before. In Chapter 1, when we first introduced declaratives, we separated the intrinsic structural function of the speech act from its external social use. We pointed out that the fact that we can utter a declarative as a secret, for instance, does not detract from the agent-neutrality of its output (though, we can now add, the act of telling a secret also forges a different, special kind of second-personal normative relationship in the act of telling). Likewise, the output of a muttered or private speech act is just the same as it would be if it were intended to be heard. So for instance, if I write in my diary “I found Waldo,” then my statement makes a standard-issue, agent-neutral truth-claim upon anyone who manages to read my diary, while if I mutter under my breath “Just leave him already!” when I am listening to my friend tell me about her horrible boyfriend, then this utterance would make an agent-specific claim on her if (contrary to my intention) she managed to hear me. In the latter case, it seems clear that this second-personal speech act is still addressed to my friend, even if (perversely) I go out of my way to prevent

The Essential Second Person

175

the address from succeeding in making a claim on its target. In the former case, it seems that I am just using myself as the generic but concrete representative of the ‘we’ to whom my truth-claim is structurally addressed. We have to distinguish such ‘secret’ speech acts from a different type of act that has no structural communicative function, even in the ideal. We sometimes make sounds—including sounds that in other contexts would constitute speech, and sounds that serve various social purposes—that simply do not have the structural function of transforming normative statuses. These can range from involuntary expletives (“Ouch!”; “Oh crap!”), to singing, to practicing a tongue twister. These may have no vocative component at all. Sometimes such acts (for example singing) are parts of social activities, but they are not addressed to the other participants. We claim that these are not exceptions to the transcendental necessity of the vocative, because they are not speech acts at all. As one of our referees put it, they are acts that may “exploit one’s speech capacities.” Some sufferers of Tourette’s syndrome curse, and only in their native tongue, for instance. But we do not think that this makes such sounds into speech, any more than the fact that drawing someone a map exploits my artistic capacities makes my map a work of art. These actions do not seek to make a claim on anyone, and they have no normative output of the sort characteristic of speech. Because they don’t have any output, they also don’t seek to realize that output through a second-personal address. We can, of course, worry about borderline cases—it will not always be clear whether or not I am singing a song in order to execute a discursive function (calling you to shared attention, reminding you of a lost love, conveying a coded message that the revolution is about to begin). It will also be difficult to determine where the merely causal impact of an act lets off and the properly discursive function begins; when does a curse, for instance, affect others merely causally, and when does it serve a discursive purpose? But the difficulty in determining where speech begins and ends is beside the point; our claim is that to the extent that an act seeks to execute a discursive function, to that extent it must include a second-personal address—though, as we just saw, not necessarily an actual intention to communicate. We also should not be distracted by another class of speech acts that appear to lack an audience, namely, those in which we cast about for an

176

‘Yo!’ and ‘Lo!’

audience—for instance when we speak into a dark room or broadcast over a ham radio, unsure whether there’s anyone there to hear us. Earlier we pointed out that vocatives themselves can have this non-recognitive, casting-about structure, but that this is a derivative form of speech act that can be understood only as a variation on the traditional recognitive vocative that retains its essentially vocative function. Similarly, we may grant now that these are cases that lack the traditional second-personal structure of an address, since there isn’t yet a determinate answer to who it is we are addressing, if we’re managing to address anyone at all. But such exceptions seem harmless, given their clear structural dependence upon and kinship with standard addresses. Finally, we will just mention here a class of complicated cases that we also mentioned in Chapter 6 and will return to in great detail in Chapter 8. These are speech acts that do not simply address a recognizable subject, but rather play a role in constituting that individual as an addressable subject. The most intuitive contenders are ‘speech acts’ ‘addressed’ to babies. We ‘address’ babies before they can really serve as the targets of second-personal speech acts, and in fact, such ‘addresses’ are among the most important tools that we use to induct them into the discursive community and to help them develop into addressable subjects. In Chapter 8 we will claim that to the extent that babies are not yet addressable second-personally, things we ‘say’ to them do not actually function as speech, but merely as causal tools for such induction and development. In any case, as above, such quasi-addresses are clearly tight variants on vocatives that exist and function only because they are riffs on the general structure of addresses, and as such, we do not think they challenge our core claim about the universality of the transcendental vocative. Language, then, is spoken by agents who own their normative positions first-personally, to other agents who are called upon to take up the normative import of speech. This can seem like a conclusion that ought to need much less argumentative buildup than we have given it. The point may seem to verge on the trivial now, given that we have been governed throughout this book by a typology of speech acts that began by asking who is entitled to speech acts and at whom they are targeted. However, the point remains invisible if we think about speech acts merely as shifting scorecards of commitments and entitlements in abstract space (or,

The Essential Second Person

177

perhaps even less promisingly, as mere physical productions backed by independent hopes that certain causal consequences will result). Given the typology with which we began, it is not surprising that we have ended up with an analysis that makes vivid the essential role of the speaker-audience relationship in discourse, and likewise ended up understanding speech acts in a way that shows off their vocative component. Once we understand discourse as essentially incarnated in a space of first-personally-owned positions, agent-relative statuses, secondpersonal and third-personal relations to others, and directed speech acts, the status of speech acts as addresses becomes no big surprise. In contrast, if our philosophy of discursive pragmatics doesn’t have the resources to articulate these deictic, agent-relative, and owned dimensions of speech—for instance if we believe that the primacy of the declarative is secure enough that we can base our theory only on this type of speech act—then all of these dimensions of speech disappear from view. The difference between first-, second-, and third-personal dimensions of speech acts will not show up as salient, and there will then be no cash value to the sense in which these are my commitments and entitlements and those are yours, and this speech act is directed from me to you. Indeed, so deep goes the declaratival bias in contemporary analytic philosophy that—as we have seen—even an author like Darwall, who is specifically interested in the structure of the second-personal address, is sure that his analysis applies only to moral claims rather than to speech acts in general. But we have shown that speech acts essentially place agents in normative relationships structured by the claims we make upon one another. We must speak to one another as a ‘me’ and a ‘you’ as opposed to just a collection of normative-status-trading engines. The way in which we make these claims upon one another is through calls to mutual recognition: we recognize others as appropriate targets of our claims, and we call to them to in turn recognize the force of our claims upon them.

Salvator Rosa, A Philosopher pointing to the ground; a youth beside him (1652). Image © The Fitzwilliam Museum, Cambridge. Reproduced by permission of the Syndics of the Fitzwilliam Museum, Cambridge.

8

Sharing a World We called each other Yo. —Jack Kerouac, On the Road

Throughout this book we have presupposed the existence of agents who can negotiate and be claimed by norms, as well as the discursive community within which such agents reside. We have portrayed speech acts as functional transformations of the normative statuses of such agents, and we have assumed—as we believe any full-throated pragmatist must—that such normative statuses exist only insofar as they are concretely incarnated within such a discursive community and among such agents. We have not, however, turned our attention to the ontology or the genesis of such agents and communities themselves. In this final chapter we draw on the picture of language, discursive performance, and voice that we have developed in order to sketch a picture of the nature and limits of normative agency and discursive communities, and the process by which fleshy bodies develop into normative agents and are inducted into such communities. We do not tell an historical story, pseudo- or otherwise, about how a world without language-users or normative agents is transformed into a world with these things. Rather, we begin in medias res and ask what is involved in becoming and being a normative agent who belongs to our discursive community. In a famous passage in “Philosophy and the Scientific Image of Man,” Sellars claims that our recognition of someone as a person in a particular normative position should not in fact be understood as having (merely) the performative force of a description: “One does, indeed, describe him, but one does something more . . . to recognize a featherless biped or dolphin or Martian as a person requires that one think thoughts of the 179

180

‘Yo!’ and ‘Lo!’

form, ‘We (one) shall do (or abstain from doing) actions of kind A in circumstances of kind C’. To think thoughts of this kind is not to classify or explain, but to rehearse an intention.”1 This recognition, for Sellars, has the force of both a description and a normative prescription for how this individual should be treated. In focusing on how the recognizer intends to treat the recognized, Sellars gives at most half of the normative picture, for taking someone as a (particular) person clearly involves holding her to specific behaviors, rather than just intending to treat her a certain way. With the concepts of recognitives and vocatives available to us, however, Sellars’s point can be made more clearly. We have distinguished between the kind of recognition we give in an observative, which expresses our receptive experience, and the type of recognition that engages the recognized in a normative relationship and demands a response. Sellars is getting at the point that recognizing someone as a person is not merely an observative act, but also a practical act of the second kind. We need not debate here the primacy that Sellars gives to intentions as opposed to other kinds of normative statuses and performance; the important claim, for our purposes, is that such acts of recognizing forge practical normative relationships as much as they note the character of something. Furthermore, they must do so, because—we will argue here—it is only by being drawn into a network of such practical normative relationships that we become persons with specific normative positions at all.

8.1 Interpellation and Induction into Normative Space A person or agent of the sort who can participate in a discursive community is one that has a particular, concrete location in normative space— a particular set of agent-relative and agent-neutral commitments and entitlements, a particular set of normatively inflected relationships with other people, a complicated set of normatively demarcated social roles, and a first-person perspective on the space of reasons, as well as a nonoptional fidelity to the tribunal of the world.2 Of course, many other 1. Wilfrid Sellars, “Philosophy and the Scientific Image of Man,” in Robert Colodny, ed., Frontiers of Science and Philosophy (Pittsburgh: University of Pittsburgh Press, 1962), 35– 78, 39. 2. That fidelity to the tribunal of the world is a non-optional ideal is built into the agentneutrality of declarative truth-claims, as we argued in Chapter 3. The agent-neutrality of the declarative goes hand in hand with the inherent publicity of truth, and likewise with the fact

Sharing a World

181

definitions of personhood or agency are possible, and might be appealing in other philosophical contexts.3 For our purposes, however, this type of discursive, socially situated agency is the type we are interested in, and it is what we will mean by ‘personhood’ here. It is only such an agent who can be the subject or the target of the various practices we have examined in this book. In Chapters 6 and 7 we argued that concrete normative relations among people are established and sustained through vocatives—that is, through the Yo-claims that hold us in place in social space. We become and remain the types of beings that have specific, agent-relative engagements with others through an ongoing network of hails and acknowledgments. Hence to be a person, in the rich sense we just described, requires that we be recognized as one, repeatedly and specifically. This suggests a claim about the foundational role of vocatives that goes beyond what we argued in Chapter 7. There we claimed that the vocative function is essential to all discourse. Now, we are suggesting that vocative discourse plays a crucial role in constituting individuals as particular, normatively positioned persons. Louis Althusser gave the name ‘interpellation’ to what he claimed was the process by which people are constituted as persons with particular locations in social normative space through vocative calls and the acknowledgments they demand. He argues that the vocative, or the ‘hail’ as he calls it, is the central pragmatic mechanism by which subjects are produced: subjects are recruited “by that very precise operation which I have called interpellation or hailing, and which can be imagined along the lines of the most commonplace everyday . . . hailing: ‘Hey, you there!’ . . . The hailed individual will turn round. By this mere onehundred-and-eighty-degree physical conversion, he becomes a subject. Why? Because he has recognized that the hail was ‘really’ addressed to him, and that ‘it was really he who was hailed (and not someone else).”4 In this parable, a hailer with the right kind of entitlement uses a hail that truth claims are not ‘for’ anyone in particular. We will return to this issue later in this chapter, but for now we simply refer the reader back to Chapter 3. 3. For another approach, which we take to be compatible with the discussion of this book, see Mark Lance and W. Heath White, “Stereoscopic Vision: Reasons, Causes, and Two Spaces of Material Inference,” Philosophers’ Imprint 7 (2007): 1–21. That account, which focuses on the issue of free action, examines the structure of material inferential propriety within which discursive agents function. 4. Louis Althusser, “Ideology and Ideological State Apparatuses,” in Althusser, Lenin and Philosophy (London: New Left Books, 1971), 174.

182

‘Yo!’ and ‘Lo!’

to recognize someone as a person—in general, one who can be properly claimed by a hail and thereby bound to acknowledge the hail, and specifically as a particular individual having the right properties and bearing the right relationship to the hailer to be so hailed. The hail, as we have seen, has a double structure: it recognizes the fact that the person is there to be hailed, and at the same time it functions as a demand that the subject acknowledge in some appropriate way that she has been appropriately hailed, that it is really she who has been hailed. In being recognized, the interpellated subject is called upon to respond to that recognition. In turn, in recognizing that she really is the one being called upon to respond, she recognizes and accepts the authority and binding force of the norms that mark out her subject position. When my colleague calls out “hello” to me in the hallway, my recognition that it is really I to whom he is speaking is not just a matter of my recognizing the descriptive content of his recognition; rather, I also recognize the pragmatic force of that recognition—the demand it makes upon me to respond by upholding the norms of greeting behavior. If I hear him say “hello” but do not respond, then I am acting as though I was not the one who was being called, or as if the call was not entitled. The point is that recognizing the hail involves recognizing not just its presence but its binding force, which is inseparable from taking it as really aimed at me—as making a claim on me in virtue of having correctly identified me. Of course, as we saw, vocatives need not have any particular surface grammar. Any speech act addressed to a second person, asking that person to recognize herself as having been recognized, is a vocative, from a shopkeeper saying to a young girl “I bet you love pink, don’t you?” to a police officer flashing her lights at a driver to get him to pull over. Remember that every vocative, whether it is a simple “hello” or a semantically complex utterance, calls for an acknowledgment that falls within some specific range of acceptable behavior. When the teacher calls out a student’s name at the start of class, the student properly acknowledges the call only by, for instance, calling out “Here!,” but not by starting a small-talk conversation about the health of the teacher’s parents. It is because of the specificity of the response demanded by a hail that it can play the role it does in constituting personhood, on Althusser’s account. It is through responding to countless such hails, and thereby recognizing the appropriateness of the recognitions that they represent, that an individual becomes the particular, normatively

Sharing a World

183

articulated person that she is. In acknowledging the hail that recognizes him as a student in the class, the student practically takes up the social role of student. It is by acting appropriately in response to the various interpellative demands of studenthood (handing in papers that have been assigned, showing up for office hours, etc.) that the student counts as a student at all. These are relationally defined activities sustained by mutual recognition. Engaging in them is what being a student consists in; there is no prior, inherent, or “natural” property of studenthood that the teacher recognizes in the student when she acknowledges him during roll call. On a much larger scale, we can argue that it is by responding to interpellative hails that recognize us in terms of our gender, our class, etc., as well as our more particularized positions such as Sarah’s friend, Amy’s spouse, etc., that we have these identities at all. In acknowledging herself as the little girl identified as likely to prefer pink, the girl practically takes up a gender identity. In turning around when your child calls “Daddy!” you practically take up your place as her father. When I recognize you as a student, or as a man, or as a friend, my recognition takes you as already having these identities. But at the same time, I call forth an appropriate response from you that contributes to this identity.5 It is because you respond as a student, a man, and a friend should that you are these things. Thus the vocative does not merely recognize us on the basis of preexisting features of our identity, but rather it calls forth behaviors that constitute this identity. The interpellative hail has a peculiar performative structure: it outstrips its own recognitive content, recognizing someone as already being a particular person with a normatively defined identity, but at the same time helping to constitute and solidify this identity. Despite Althusser’s hyperbolic talk of a “one-hundred-and-eighty-degree turn,” this identity is solidified by way of a vast number of little interpellative moments that make up our mundane negotiations of the social world. Most of them have no special personal or political interest and do next to no constitu5. Slavoj ÑiÒek writes: “‘Being a king’ is an effect of a network of social relations between a ‘king’ and his ‘subjects’, but . . . to the participants of this social bond, the relationship appears necessarily in inverse form: they think that they are subjects giving the king royal treatment because the king is already in himself, outside of the relation to his subjects, a king; as if the determination of ‘being a king’ were a ‘natural’ property of the person of a king.” ÑiÒek, The Sublime Object of Ideology (London: Verso, 1989), 25.

184

‘Yo!’ and ‘Lo!’

tive work on their own. Our selves are not slammed together; rather these small interpellative moments slowly form and contour our normatively defined, socially embedded identities. In fact, every vocative plays at least a tiny interpellative role. This is because each vocative, no matter how mundane, is both alethic and constative, to draw on our terminology from Chapter 5. The vocative is alethic insofar as it calls upon us to uphold norms that already bind us, by calling us to recognize ourselves as really the one subject to those norms—as a father, a friend, a doctor, the person who checked out this particular library book, or the person capable of answering this question. Even a simple “hello” calls upon me to recognize myself (1) as the type of person bound by the norms of greeting behavior and (2) as bearing the right normative relationship to this hailer to entitle the hail and commit me to the acknowledgment it demands. Other, more loaded hails—a cosmetic salesperson asking “Do you worry about premature wrinkles?” for instance—call upon me to recognize my placement within a cluster of norms that target my gender, age group, class, stress level, and more, and hold me to these norms. When we hail someone, part of what we are doing is calling her attention to the norms that already bind her in virtue of her position in normative space. Interestingly, because vocatives have this alethic component, every ‘Yo!’ contains a kind of ‘Lo!’ (Later we will argue that every ‘Lo!’ also contains a ‘Yo!’, and not merely in the generic sense we discussed in Chapter 7.) At the same time, the vocative is constative, at least in the minimal sense that the speech act itself places a new normative demand on its target, namely the demand for acknowledgment. It thereby shifts the total normative position or scorecard of the person it hails. The process of interpellation draws on both the alethic and the constative dimensions of the vocative: only because of the alethic dimension is the recognition legitimate—this dimension is what allows it to be “really me” who has been recognized. But only in virtue of the constative dimension can the vocative do its constitutive work. In more dramatic cases of interpellation, the vocative not only shifts the normative status of the person hailed by demanding acknowledgment of a correct recognition, but it shifts the identity of the person recognized by demanding that she recognize herself in a new way and act appropriately. As she does so, this new self-recognition becomes more appropriate. For instance, consider the practice of many childbirth edu-

Sharing a World

185

cators of referring to the people in their classes as “moms” and “dads.” Any pregnant woman in the class knows that it is “really she” who is being addressed when the teacher says, “Okay, moms, now lie down on the floor.” Acknowledging this call plays a part in her beginning to take on an identity as a mom, with all of the normative baggage that comes with it. Thus the interpellation is constative, not only in giving her a new burden to respond, but in helping to give her the commitments and entitlements that come along with motherhood more generally. Such interpellations can be used strategically and manipulatively (as this one often is), but they need not be; recognizing one another as inhabiting certain normative positions—such as that of an expert in a particular field, that of a loyal friend, etc.—more completely than we do, and thereby solidifying these positions, is a common currency of everyday social interaction. Normally, such interpellative moments happen without conscious purpose or instrumental motivations, simply as part of the ongoing stream of social transactions that engage us in normative relationships and hold us in normative place. None of this is to say that in being interpellated as inhabiting a certain normative position, we are automatically constituted as obedient to these norms. Calling a pregnant woman a mom doesn’t make her one, and sales pitches for wrinkle cream do not determine that I will begin to worry about premature wrinkles. Indeed, the possibility of refusing or rejecting a hail is inherent to its vocative character. We cannot engage in normative practices at all unless we are the sorts of beings who can recognize the claims of norms, but this is not possible unless we can transgress or fail to live up to these claims, because the binding force of norms makes sense only in the face of a possible gap between what we do and what we ought to do. Further, we are not truly responsive to norms if we are merely subject them and cannot author or resist them. Norms make claims on us only insofar as they are legitimate, and hence we must be capable, not only of violating them, but of challenging their legitimacy in order to count as able to recognize their binding force. Since a vocative seeks to make a normative claim on the one called, it builds in the presupposition that the one called may refuse or fail to take up that claim. Even if I accept that a hail identifies me correctly and legitimately demands acknowledgment, this can’t mean that I am stuck accepting all of the norms that make a claim on me in virtue of that identity, without resistance or cri-

186

‘Yo!’ and ‘Lo!’

tique. Not only can I refuse to be hailed as a mother (at least while I am still pregnant), but I can accept the mantle of motherhood while resisting various norms of motherhood, and so forth. On the other hand, as we’ve seen previously, vocatives can have a certain kind of inescapability. Rejecting or resisting an identity or a norm is not the same as completely failing to recognize the claim it makes. When I resist being interpellated by the label of “mom” in a prenatal class, the name doesn’t merely bounce off me and leave me unscathed.6 Rather, it puts me into a tense, partially antagonistic relationship with a particular identity and set of norms. When I challenge a hail—by denying its appropriateness altogether or by denying the specificities of what it demands from me—I still acknowledge its attempt to make a claim on me. I treat it as a second-personal speech act that calls for some acknowledgment out of a range of possible responses from me; given the nature of norms, contestation and refusal always count as part of this range. Thus by recognizing the hail as having targeted me, rightly or wrongly, I already give it an acknowledgment that grants some legitimacy to the original recognition. Individuals are interpellated long before they are in fact self-recognizing subjects capable of properly acknowledging a hail. Althusser points out that, indeed, babies are ‘expected’ even before birth; if nothing else, they already have a (typically patrilineal) name that comes with a cluster of normative positionings.7 With the advent of prenatal testing, not only do we gender-code babies’ rooms before they are born, but we often “notice” personality traits on ultrasound screens, and in fact it is standard for ultrasound technicians to talk “to” the fetus as they scan, commenting upon these “traits” (“My, you’re quick—you’re going to be an athlete like Daddy, aren’t you?”).8 On a massive scale, interpellation is one of the main tools we use to bring babies through the transition from sentient animals outside the space of reasons to normatively positioned beings with first-person perspectives who are capable of second-person dis6. The children’s retort “I’m rubber and you’re glue. . .” and the perniciously dismissive parenting chestnut “Sticks and stones may break your bones, but words will never hurt you” are thus profoundly wrong. 7. Althusser, “Ideology and Ideological State Apparatuses,” 132. 8. Lisa Mitchell, Baby’s First Picture: Ultrasound and the Politics of Fetal Subjects (Toronto: University of Toronto Press, 2001), 93.

Sharing a World

187

course and are responsive to normative claims. We induct babies and young children into various identities by “recognizing” them as already having them (by painting a baby boy’s room blue, telling a baby that he is “a real lover” or a “funny little joker,” etc.), and we teach them responsiveness to the claims of various norms—moral, social, rational, etc.— by treating them as if they already are responsive to them (“We don’t want to hurt your sister, do we?”). Early on, these can at best count as quasi-vocatives. A true vocative seeks acknowledgment based in self-recognition, but when we first start talking to babies and young children, they are in no position to have such self-recognition. We are faking it, as it were, for the purpose of bringing it about that such discourse will no longer be a pretense. Slowly, retroactively, children come to find themselves (partially, with resistance, and so forth) in these recognitions of them, precisely because we demand appropriate acknowledgment from them in advance of their ability to give it. When a vocative plays an interpellative role in constituting an adult’s identity—when I am called to recognize myself as a mother during a prenatal class, for instance—the adult is normally fully capable of recognizing herself as bound by various norms. Even if I do not yet actually or fully occupy the position that the hail recognizes me as occupying, I can recognize that I have been recognized, and put into question the normative statuses that are imputed to me in that recognition. Interpellation can succeed in constituting me when I recognize myself in how I have been recognized. In the case of very young children, on the other hand, we use the vocative as a causal tool for creating an identity and set of normative commitments out of creatures not properly in normative space at all. Part of what our quasi-vocatives do is help them to develop the capacity to recognize themselves as recognizable subjects bound by norms in the first place. Until this happens, they cannot properly count as the targets of hails—or, for that matter, of any speech acts, given that speech acts are in the business of seeking to make normative claims that are taken up by others. Hence these quasi-vocatives are not only not real vocatives; insofar as they are directed at the baby, they are not really speech at all, because they are not targeted at the kind of being who can enter into the relationship of mutual recognition called for by the vocative dimension

188

‘Yo!’ and ‘Lo!’

of speech. We often say things “to” babies for the benefit of others. I can say “Well, aren’t you cute!” to a baby first and foremost as a means of flattering her beaming father, for instance, or I may say “We don’t hit the dog!” as much to demonstrate responsible parenthood to others as to train my baby not to hit the dog. These are speech acts—they just aren’t addresses to the baby. While they are perfectly real interactions with the baby, they are not discursive interactions with him. At least at the start, we should see these quasi-vocatives as playing a merely causal role in producing a hailable subject. Later, when they serve to hone and solidify a child’s subject position, they will come to have genuinely vocative as well as causal elements. At some point after infancy, we begin genuinely demanding self-recognizing acknowledgment from our children, even when we know they are not yet able to give it (“You know better than to judge your friends by the color of their skin, don’t you?”). Only after the fact will the child be able to recognize herself properly in such hails, if they are successful. The transition from the quasi-vocative to the full vocative is surely gradual and incomplete. As we have seen, vocative discourse has an avant la lettre constitutive character even among adults. Furthermore, interpellation is an ongoing process that helps sustain the concrete currency of social life; it is not a process that inducts us into a determinate form of personhood and then stops. As we move through life, we are ushered into new identities and claimed by new sets of norms at every turn. This is not only because the hails that apply to us change as we change—we become rightly recognized as parents, friends, leading figures in our field, alcoholics, etc.—but also because normative space itself changes. The norms that claim me in virtue of my gender identity (as a man, let’s say) are constantly developing and under negotiation. This means that vocatives that recognize me on the basis of this identity (my colleague asking me which team I like for the Super Bowl, for instance) are constantly calling me to recognize myself as (already) bound by subtly different, evolving norms. In each case my acknowledgment will contain at least a small creative moment, as I accept, resist, or transform the expectations embedded in the hail.9 9. Judith Butler has developed this point about the ongoing and creative character of our responses to interpellation in detail. See in particular Butler, Excitable Speech: A Politics of the Performative (New York: Routledge, 1997).

Sharing a World

189

Although our negotiations of our practical social identities serve to provide (what we hope are) juicy and vivid examples of interpellation at work, we should not forget that our position in normative space includes our position in theoretical space: our mastery of concepts, our commitments to truths, and so forth. Accordingly, we can often witness the process of interpellation at work as we master and negotiate and sometimes expand or change this space. One of the main ways in which we come to grasp a new concept is by plunging in and using it as if we had already mastered it; one of the main ways in which others teach us new concepts is by treating us as though we are already capable of using them. Think of John, Sellars’s hapless necktie salesman in Empiricism and the Philosophy of Mind; his fellow salesmen point out to him that the ties he is calling “green” don’t look green in natural light, as if he were already a master-user of the concept ‘green’ (see section 4 of this chapter for an extended discussion of this point). We ask our student a question about Davidson’s critique of Quine, thereby demanding a demonstration of her theoretical mastery of the issues involved. And for a very long time with such questions—perhaps forever—we do not suppose that the student is actually capable of adopting a coherent position within the normative space of analytic post-empiricism. But just as with the baby’s induction into discourse in general, so induction into particular corners of conceptual space operates at this constative level. We assign statuses to others so as to make them capable of deserving those statuses. As Wittgenstein famously made clear, concept-mastery cannot be exhausted by memorization of rules; instead, an ineliminable part of grasping a concept is developing the practical skill of putting it to use. For this reason, using a concept before we are fully ready, as if we already can, is often the only way to learn it. Being a member of a discursive community, having a normative position that gives us a distinct roster of agent-relative commitments and entitlements, and being hailable—that is, recognizable by others with a vocative—are equiprimordial phenomena. Later we will argue that being responsive to an objective world also belongs on this list. Agent-relative commitments and entitlements attach themselves to us in virtue of our normative position, including our relations to others, and these relations are forged during second-person interactions with others in our

190

‘Yo!’ and ‘Lo!’

discursive community—they do not somehow inhere in us waiting to be expressed in our interactions.10 At the same time, when we recognize someone with a hail, we presuppose that she can be claimed by our speech act, and hence that she belongs to our discursive community and has a normative position within it. Furthermore, if someone were not recognizable as a distinct person with a normative position within the discursive community, and thus could not be hailed, then he could not engage in the kind of second-person transactions that establish our position in normative space. Hailability is therefore a precondition for being a person (in our sense) and a member of the discursive community. Interpellation—including the quasi-interpellation that begins before we are properly hailable—is a primary mechanism through which individuals become hailable discursive community members. As we just saw, we must be hailable in order to be normatively positioned subjects, while we must be normatively positioned subjects in order to be hailable. The constitutive structure of interpellation is what saves this apparent chicken-and-egg problem from being pernicious; functioning both alethically and constatively, it calls us to recognize ourselves in recognitions of us that (slightly) exceed the reality at the time of the hail.11 It is through the many interpellative exchanges of mutual recognition that slowly position us in social space that we come to be a person with agent-relative commitments and entitlements.

8.2 Membership in the Discursive Community We have argued that membership in a discursive community is a precondition for normative agency, and we have depended heavily upon the presupposition of such membership throughout this book. For example, we cannot understand speech acts with agent-neutral outputs, such as truth-claims and observatives, without appeal to the notion of membership in a discursive community, for such speech acts, we have argued, have ‘public’ output and are ideally ‘for everyone’. But the public or the 10. Though of course it is quite likely that various sorts of capacities to successfully enter into such relations are part of our biological inheritance. 11. Jacques Lacan argued repeatedly that we always and necessarily recognize ourselves as having a more unified and solidified identity than we actually have, and that these (mis)recognitions are part of the process by which we come to have an identity that is (imperfectly) unified and solidified. See, for instance, “The Mirror Stage,” in Lacan, Ecrits, trans. Bruce. Fink (New York: Norton, 2004).

Sharing a World

191

‘everyone’ here is the discursive community, and one is the target of such a speech act if and only if one is a member of that community. We have deferred the question of what makes something a discursive community and what it takes to belong to one, but we turn to it now. In one sense the answer is straightforward. A discursive community, in the sense useful to this book, is a community that serves as the (ideal, functional) domain for the inputs and outputs of speech acts by community members. For instance, a speech act with an agent-neutral output targets everyone. The ‘everyone’ here is the discursive community. Who is ‘everyone’? When I declare that Florida is hot in July, I am not attempting to change the commitments and entitlements of my pet ferret, even in the ideal. However, the ‘everyone’ extends well beyond the scope of my fellow English-speakers who could understand the declaration directly; it is as true for a Norwegian that Florida is hot in July as it is for me. The scope of my speech acts—and of my discursive community— extends as far as the network of people to whom I belong who can transform one another’s normative statuses through speech, either directly or via intermediate speakers. One is a member of the community if one can do this and have it done to one by the others. In order for this to be possible, one must meet at least two criteria. First, one must be hailable—that is, recognizable through vocative speech acts (“Yo!”) by (at least some) other members of the community. If one cannot be recognized in this way, then one cannot be targeted by speech acts, as we saw in Chapter 7. Second, one must be capable of observing the same public objects and being sensitive to the same public facts as others, and of being called by others to shared attention to these facts and objects (“Lo!”). We will argue for this second criterion in detail later. For now, just notice that in order to participate in declarative discourse with others—the giving and asking for reasons, in the traditional Brandomian sense—one must be able to call others’ attention and have one’s own attention called to empirical evidence for or against various truth-claims. Without this ability to invoke shared sensitivity to the testimony of the world, meaningful declarative agreement and disagreement, and with them meaningful declarative discourse, are impossible. (See Chapter 3 for our argument that declarative discourse must ‘terminate’ in observatives.) Individuals do not share a public, intersubjectively available world to which their discourse refers without this ability, and hence they need it in order to share a discursive community.

192

‘Yo!’ and ‘Lo!’

But this leads to the question, What sort of beings can in fact be recognized as members of a discursive community? Or, in other words, who can be successfully hailed? This is a version of a question that has received a great deal of attention from both ethicists and philosophers of mind. Ethicists have asked what features an entity must have in order to count as member of the moral community, worthy of the specific form of respect due to persons. Philosophers of mind have asked which features an entity must have in order to count as having a mind, or representational or conceptual capacities. For both groups, arguments over borderline cases—animals, babies, people in a persistent vegetative state, etc.—have ensued. However, an advantage of our normative pragmatic account is that we can answer our version of this question the other way around: we need not look for any inherent features or properties that are the markers or conditions of personhood. Rather, someone is a person (in the relevant sense) if he is in fact caught up in a network of discursive holdings—that is, if others successfully recognize him, through speech acts containing vocative moments, as a user and receiver of speech acts, or as the kind of being who can transform the normative status of others and have his own normative status transformed through discourse with other community members. Likewise, to be a member of the community is not, in the first instance, to have some feature in common with other community members. Rather, the ‘we’ is constituted and sustained through the transactions among the various mutually recognizing subjects who make it up. The community is not a predefined space into which candidates may fit or fail to fit; it is a space created and given its character and its boundaries by the discursively interacting individuals who make it up—individuals who can speak from a first-person perspective to others in a second-person voice.12 In order to determine whether someone is a person or a member of 12. In the Phenomenology of Spirit, trans. A. V. Miller (New York: Oxford University Press, 1977), Hegel famously develops such a vision of the community as constituted through mutual recognitions among subjects who can make normative claims upon one another in discourse. On the one hand, he contrasts such a picture with understandings of the community as defined by shared blood, shared cultural traditions or ethical commitments, or other such contingent unifying features. On the other hand, he contrasts it with a merely formal conception of a community as an association of autonomous individuals. Our understanding of the discursive community constituted in and through concrete second-person interactions that are responsive to reasons is explicitly intended to be Hegelian, although we do not mean it as a reading of Hegel, and hence are not concerned if it differs from Hegel’s account in the details.

Sharing a World

193

the discursive community, then, we need not look for some special cognitive, bodily, or other characteristic that makes him an appropriate interlocutor. Rather, success at hailing him and involving him in discourse is proof of personhood, and global failure at doing this places him outside the discursive community by definition. We can constitute personhood through interpellation where none exists to begin with, as we do with infants, but whether our interpellations are constitutive of personhood or merely play-acting will depend on our eventual success or failure. When we are confronted with a case of marginal personhood, the best we can do, both morally and epistemically, is to do our very best to recognize and engage that individual as a person in our discursive community, at least minimally or intermittently. By this standard, people in persistent vegetative states and fetuses are not persons, but most patients with dementia are. Animals may or may not be. In deciding such cases, it is our pragmatic criterion of personhood—success at recognizing someone as a specific person and receiving acknowledgment—that is ultimately definitive, as opposed to any theoretical story about the conditions of personhood that we seek to apply. Of course, there is nothing philosophically or empirically illegitimate about trying to figure out which capacities or features a creature must have in order to be a candidate for induction into personhood. However, we should not confuse the causal prerequisites for personhood with its defining criteria. Although all of this sounds almost analytically obvious, it commits us to the rather strong thesis that only beings who are recognized as members of a discursive community count as normative subjects capable of having agent-relative commitments and entitlements. Such normative subjectivity cannot exist merely in the form of potential recognizability. For if a being has never engaged in the relevant type of normative transactions with others, then she isn’t a locus of normative activity, bound up in a concrete network of holdings, and she has no particular, concrete normative position. And since, as we have discussed repeatedly, all such transactions contain a vocative moment of mutual recognition, a being who is not recognized as a normatively positioned person simply is not one. Once we give up the “Platonic scorecard” vision of normative space as an abstract network of commitments and entitlements, we can be beings with normative statuses only if we are treated as such and act as such in practice, and this requires that we participate in exchanges of mutual recognition.

194

‘Yo!’ and ‘Lo!’

Now, someone may have the potential to be such an agent, but because of her community’s extreme bigotry, neglect, or some other reason, this potential may never have received the interpellative uptake it needs in order to be actualized. We can count such a being as a moral victim; she has been prevented from having access to the community of persons, and this is a terrible loss. Likewise, those of us who are already moral persons commit a terrible moral wrong by neglecting a potential person in this way. However, we cannot take her as a moral person who has been disrespected by not having this personhood acknowledged. Consider, for instance, what used to be our standard practice of consigning infants with Down syndrome to institutions where they received minimal physical care and virtually no stimulation. These children generally failed to develop any recognizable discursive or social capacities. We now know that most children with Down syndrome are capable of quite sophisticated cognitive development if they are given the right kind of stimulation and normal, loving social interaction. Thus it turns out that our old practices were deplorable, and that we inflicted a terrible harm upon these children. Yet one should bite the bullet and say that (in general) the harm that was done was that of not giving them what they needed in order to be inducted into normative personhood, rather than the harm of failing to acknowledge the personhood they already had, “on the inside,” as it were. On the other hand, refusing to hail or acknowledge the hails of someone who is already a person—shunning her, or casting her out of the community—is a quite different kind of moral harm. As we discussed in Chapter 6, shunning and intentionally ignoring are complicated actions that in some sense must acknowledge or recognize the very normative status they seek to deny. Just as refusing or resisting an interpellative hail is still a complicated kind of engagement with it, likewise a refusal to hail involves a kind of normative engagement with the object of that refusal. In claiming that normative status is constituted in second-person transactions involving mutual recognition, we are in agreement with Darwall’s account in The Second-Person Standpoint. However, there are important differences between our pragmatic account and Darwall’s contractarianism. Contractarianism requires subjects who can enter into contracts and forge commitments and entitlements prior to being nor-

Sharing a World

195

matively positioned beings. In contrast, we claim that one can be a subject capable of such transactions only once one already inhabits normative space. Furthermore, we are not saying that each commitment or entitlement must be recognized in order to exist. Rather, our point is that, globally speaking, one can be a subject with agent-relative commitments and entitlements only if one is actually recognized as such a subject by others in one’s discursive community (we will return to the question of whether such social placement is required for having agentneutral commitments and entitlements, such as entitlements to declaratives and commitments to truths, later in this chapter). Once a structure of mutual second-personal recognition is up and running, surely the normative space that it institutes will far outstrip the set of normative claims that we actually make upon one another.

8.3 How Many Discursive Communities Are There? Throughout this book we have spoken of ‘the’ discursive community, remaining intentionally vague on the question of whether there is only one. Now in an obvious sense there are of course many, many discursive communities—communities of people who share a language or a dialect and recognize one another as doing so. Davidson had to insist upon a narrow and rarified definition of a language in order to plausibly deny the obvious fact that various ones exist.13 But for our purposes, the scope of the discursive community is broader than this, as we have seen; it includes everyone who is capable of entering into direct or indirect mutually recognizing discursive interactions with other community members, and is capable of responsive sensitivity to the objects and events in the public world that community members share. Thus the real question concerning the number of discursive communities is this: Is it possible for there to be bounded clusters of mutually recognizing agents who share a world, such that recognition and shared attention (‘Yo’s and ‘Lo’s) 13. “I conclude that there is no such thing as a language, not if a language is anything like what many philosophers and linguists have supposed. There is therefore no such thing to be learned, mastered, or born with. We must give up the idea of a clearly defined shared structure which language-users acquire and then apply to cases.” Davidson, “A Nice Derangement of Epitaphs,” in Lepore, ed., Truth and Interpretation: Perspectives on the Philosophy of Donald Davidson (New York: Blackwell, 1986), 446.

196

‘Yo!’ and ‘Lo!’

are not possible across the boundaries of these clusters? Or, to put the point another way, can there be speech acts with agent-neutral inputs or outputs that range over different sets of agents? On the face of it, it seems that there are three possibilities: 1. There are multiple discursive communities. Putting aside the possibility of one individual having membership in more than one community, members of one community cannot enter into relationships of mutual recognition with members of other communities, and the claims of one community have no import for the members of the other. Discursive agents are separated into groups that inhabit “different worlds.” 2. There is only one discursive community. To be a normative, discursive agent at all is to be responsive to a single public world and to be able to enter into mutually recognizing normative transactions, at least indirectly, with any other such agent. All agent-neutral inputs and outputs have exactly the same, completely universal scope. Surely the boundaries of this community are negotiable and not sharply demarcated, and of course full participation in this community—sensitivity to every truth and reason, the capacity to enter into mutually recognizing transactions with all other community members, etc.—is a regulative ideal that agents will rarely if ever meet. But the discursive community, like the space of reasons and norms this community negotiates, is essentially unitary. 3. There is one, fundamental, discursive community, but there can be provisional and derivative discursive communities within that. All discursive agents must share an Ur-discursive community that allows us to recognize our common personhood and refer to a single objective world. However, against this background, a subdiscourse can be indexed to a subcommunity. Only members of the subcommunity can hail one another in the subdiscourse, and so forth. We will dispense with option 1 fairly quickly. The interesting question is whether option 2 or option 3 is correct. Option 3 is substantially more complex than the others and will need clarification. Option 1 leads to an unacceptable form of extreme relativism. If discursive communities can be cut off from one another, then, by defini-

Sharing a World

197

tion, all the entitled agent-neutral declaratives in each community are targeted only at members of that community. In other words, the truthclaims of one community have no import at all for another community, which is just to say that the communities do not share a world about which agreement and disagreement is possible. Nelson Goodman explicitly defended this idea that different communities inhabit different worlds.14 Some ‘standpoint epistemologists’, who argue that different groups have access to different warrants, also defend such a line.15 Ian Hacking approvingly describes the kind of discursive segregation that accompanies this view: We cannot reason as to whether alternative systems of reasoning are better or worse than ours, because the propositions to which we reason get their sense only from the method of reasoning being employed. The propositions have no existence independent of ways of reasoning toward them.16

But we have already committed ourselves to the view that truth is absolutely public and not indexed to a perspective in this way. The indexical view seems to fundamentally undermine any possibility of objectivity, because it implies that any purportedly objective claim we might make would be neither compatible nor incompatible with any of the purportedly meaningful things said by people in the alternative “world.” So it seems inescapable that we would not in fact be claiming truth for our assertions, but merely some sort of entitlement-for-us. It has been a central presupposition of this work that negotiation of the space of reasons essentially involves responsiveness and fidelity to the claims of a genuinely public world that sets a tribunal of truth independent of our varying perspectives and discursive positions. Hence we insist that truthclaims, at a minimum, must seek to make the same normative claim on 14. Nelson Goodman, Ways of Worldmaking (Indianapolis: Hackett, 1978). 15. See for example Susan Hekman, “Truth and Method: Feminist Standpoint Theory Revisited,” Signs: A Journal of Women in Culture and Society 22 (1997): 341–365. One of us has argued at length that standpoint theory need not lead to such relativist, anti-objectivist conclusions. See Rebecca Kukla and Laura Ruetsche, “Contingent Natures and Virtuous Knowers: Could Epistemology be ‘Gendered’?” Canadian Journal of Philosophy 32 (2002): 389–418; and Kukla, “Objectivity and Perspective in Empirical Knowledge,” Episteme 3 (2006): 80–95. 16. Ian Hacking, “Language, Truth, and Reason,” in Steven Lukes and Martin Hollis, eds., Rationality and Relativism (Oxford: Blackwell, 1982), 48–66, 65.

198

‘Yo!’ and ‘Lo!’

all rational, discursive agents; they must not only be agent-neutral in their output, but the domain of this output must include all inhabitants of the space of reasons. This is sufficient to rule out option 1. We will consider one more reason to reject option 1. If option 1 were correct, then there would be normatively responsive, discursive agents from whom ‘we’ were utterly discursively cut off. This would mean that we could not hail these agents with vocatives, nor receive acknowledgments from them. It seems we should take it as a defect in our discursive practices if we cannot use our language to acknowledge and engage people who are, in other contexts—or other worlds, perhaps?—perfectly acknowledgeable and engageable. The defect would appear to be both moral and linguistic: a language that cannot be used to hail everyone who is hailable seems to have fewer resources than it ideally should; more important, the speakers of such a language seem to be missing a morally significant capacity to respond to some agents who are in fact genuine, normatively responsive, discursively able moral persons, as such. These persons would have to inhabit a different world from ours. But presumably, if they are actual, they have to show up somehow as material objects in our world. And if this is so, it seems that—morally and epistemically—we ought to be able to recognize and respond to their real features and normative statuses. But if this is a defect in our discursive practices, it is because our discourse ideally strives to include these others in its scope, in which case option 1 is false. This may count only as an argument from intuition, but it strikes us as powerful nonetheless.17 Let us move on, then, to options 2 and 3. If there exists one and only one discursive community, then the “we” that provides the scope of discourse is simply all inhabitants of the space of reasons. It seems that there are two main, related reasons why one might worry about this option. 17. To insist that option one is false is to extract a core lesson from Davidson’s paper “On the Very Idea of a Conceptual Scheme,” Proceedings and Addresses of the American Philosophical Association 47 (1973): 5–20, by insisting on the unity of truth and the possibility in principle of discursive interaction and mutual recognition among all agents. It is not, however, to accept Davidson’s strong conclusion that all conceptual schemes are intertranslatable. Davidson has twigged onto the ideal universality of truth-claims, as well as the impossibility of discursive agency that receives no uptake from other speakers. But he has no basis for his move from there to the conclusion that, in actuality, all discourse is intertranslatable and everyone is responsive to the same features of the empirical world.

Sharing a World

199

First, there are many contexts in which a community—based on class, ethnicity, profession, friendship, etc.—will address a claim to ‘everyone’ or make a claim about ‘everyone’ that includes only the members of that community. “Everyone these days sends their kids to Kaplan prep classes before the SATs” and “Everyone goes to the Copa Cabana on Fridays” are clearly not claims whose scope is all rational agents. When the chair of a faculty meeting asks if “everyone” is ready to vote, she clearly does not mean to include the staff member refilling the coffee urn at the back of the room. These are cases, one might think, in which the implied domain of our speech acts is a community smaller than that of the inhabitants of the space of reasons. Second, we have learned from Kuhn and similar writers that different subcommunities can share practices, standards, and techniques that afford sensitivity to different objects, events, and features of the empirical world. We might think that discourse about such specialized entities targets only the members of that particular subcommunity. Assuming we have rejected option 1, such thoughts may lead us to option 3, namely, the idea that there can be provisional or context-specific discursive communities that provide the scope of some of the agent-neutral claims made within them, even though such communities exist only within the frame of our most general, universal discursive community and its shared world. At first blush, we can accommodate claims about or to ‘everyone’ that seem to have a restricted scope much more straightforwardly. We can take declarative claims about ‘everyone’ such as “Everyone goes to the Copa Cabana on Fridays” (or, as Yogi Berra purportedly once said, “That place is so crowded nobody goes there anymore”) as simply elliptical for more restricted claims, such as “Everyone [who lives in this city and is able-bodied and cool] goes to the Copa Cabana on Fridays” or perhaps “Everyone [who is worth hanging out with at all] goes to the Copa Cabana on Fridays.”18 At most, we might say, this phrasing indicates a selfcentered lack of interest in what other groups of people do, not a literal exclusion of them from the discursive community, even provisionally. 18. Neither need we assume that such uses of ‘everyone’ are explicitly definable in terms of other predicates. It may well be that the proper understanding of a given restricted use of ‘everyone’ is something that must be picked up in context and that cannot be defined in other terms. But still, this is merely a semantic phenomenon governing certain uses of ‘everyone.’ Even if the ‘everyone’ in “Everyone goes to the Copa Cabana on Fridays” is not precisely synonymous with any other phrase, the function of the declarative is nonetheless to claim of a certain class of people that they go to the Copa.

200

‘Yo!’ and ‘Lo!’

Claims addressed to a restricted ‘everyone’, such as “Everyone be here by 8:00 tomorrow!,” simply have kind-relative agent-relative outputs: “For those of you who are members of this committee, be here by 8:00 tomorrow!” None of this challenges the singularity of the discursive community. As we have seen, various people can hail one another in various ways, so the existence of insiders’ ways of hailing one another doesn’t prove much. Furthermore, the mere fact that members of a group tend, in certain contexts, to acknowledge and address one another without acknowledging that anyone has been left out does not show that their discourse cannot accommodate a wider range of acknowledgment and address. What of the worry that people involved in different sets of practices, such as different scientific paradigms, have access to different kinds of objects and facts? In contrast to option 1, here the claim is not that people in different paradigms “inhabit different worlds” and cannot speak to one another, but the more limited claim that within our shared public world and shared discourse, there are clusters of people who can speak to one another about some limited objects and events that others do not have access to. Limiting this kind of relativism to a particular domain of discourse does not help with the fundamental problems that we raised when we rejected option 1. It is a central tenet of this book that observatives and truth-claims must be absolutely (rather than provisionally) agentneutral in their output, because we—all of us rational, normatively responsive agents—must be able to share one public world about which we can agree or disagree in ways that can be held to the tribunal of that world. Even though people differ in their sensitivity to facts and features of the world, this does not mean that their claims about these facts and features are only for others who can detect them. If claims about objects visible to the community had import only for community members, then these would not be objective parts of the public world at all. Rather, such claims are for everyone, but most of us are deficient observers in various ways and will not be able to see or understand or claim entitlement to what some others can see or understand. Whether or not I understand the claim, whether or not I’m capable of knowing one way or another, and whether or not I have the relevant perceptual abilities, it is easy enough for me to foolishly deny a true claim made by a theoretical physicist. And if I do so, I’ve said something false, made an error, chal-

Sharing a World

201

lenged the claim of the physicist without grounds. All of this demonstrates that I’m in the scope of the normative output of the claim by the physicist. We have said all along that the universal reach of agent-neutral discourse is an ideal, and that not everyone will be in a position to take up every entitlement that discourse seeks to impute to her. While agentneutral claims within such scientific communities are ideally for everyone, the social reality is that only those with specialized training are in a position to be sensitive to the phenomena in question or to understand claims about them. Remember John, Sellars’s perceptually challenged necktie salesman. John did not have the ability to see that objects were green, because he had not learned how to separate standard from misleading conditions for viewing colors, and his co-workers helped educate him. But before he developed this ability, green neckties were green, agent-neutrally, whether he knew it or not. His inability was a defect in him, and it did not compromise the universal validity of the declarative “This tie is green.” The defects of some perceivers damage neither the absolute agent-neutrality of observatives or declaratives nor the universality of their ideal reach. Adding restricted discursive communities to our ontology is complicated. There would be strict limits on the type of discursive practices that they could support. We have already argued that all declarative truth-claims must take the universal community of all rational agents as their domain, for otherwise they cannot be hooking on to an objective world. If a claim made within a subcommunity is applicable only to members of that community, then it is not a truth-claim. In our example “Everyone goes to the Copa Cabana on Fridays,” the restricted sense of ‘everyone’ occurs in the subject position. Presumably it is still true, for those of us sitting at home with our children and our computers on Friday evenings, that ‘everyone’ named by the claim is at the Copa Cabana. Hence the scope of the output of the claim is still universal. If there are genuine discursive subcommunities, then their discursive practices do not include issuing declaratives; insofar as members of the community make declarative claims, they thereby revert to speaking as members of the larger community (albeit, perhaps, members who are only socially interested in conversing with fellow subcommunity members). Mutatis mutandis, the same goes for observatives: in expressing recognition of public facts, objects, etc., and thereby grounding infer-

202

‘Yo!’ and ‘Lo!’

ences to declarative truth-claims, observatives must be referenced to a world shared by all negotiators of the space of reasons. You can’t recognize a rabbit in the bush unless it is true, for the unrestricted everyone, that there is a rabbit in the bush. At the same time, it is difficult to see what would compel us to index any speech acts with agent-relative outputs to a discursive subcommunity. Since it is in the nature of such speech acts (imperatives, vocatives, etc.) to be specific about whom they address, the fact that not everyone is addressed by them seems benign. For speech acts with agent-relative outputs to be examples of the discursive practices of a subcommunity, we would need them not only to target a restricted audience, but somehow to structurally require a background reference to a restricted community of ‘everyone’ standing behind this restricted audience. We cannot think of any examples of speech acts that compel such a reading, nor can we imagine what they might look like, although we have no principled reason for rejecting their possibility. Hence we have reason to think that discourse indexed to a subcommunity, if it exists, does not include declaratives, observatives, or speech acts with agent-relative outputs. Instead, such discourse would have to consist of speech acts with agent-neutral outputs (belonging to box 1 or box 2 of our grid) that are not in the business of making or grounding truth-claims, but rather serve some other pragmatic function. This would have to be a function that we can understand only by construing the output of the speech act as genuinely agent-neutral, even though its scope is restricted—as opposed, for instance, to merely having a kind-relative output. Such a discursive practice would have to appeal to a pragmatic, structural sense of ‘everyone’—rather than a semantic sense of ‘everyone’—that is, a selfcontained ‘we’ within the larger, background discursive community. Proper use of Occam’s razor seems to require that we find a compelling example of such a type of speech before we add discursive subcommunities to our ontology and start figuring out how to track their complicated dependence relationships with the larger discursive community, and how to understand them as having any kind of self-contained existence, given that their members would have to constantly ‘exit’ the subcommunity whenever they made declarative assertions and other such routine utterances. Without such an example, we have no reason to plump for option 3 as opposed to option 2. Although we are not at all convinced that the idea of provisional dis-

Sharing a World

203

cursive subcommunities is ultimately useful or coherent, we can think of two sorts of discursive practice that might seem to meet the requirements we just described and to be best understood as the practice of such a community. Consider, first, a predominantly white cotillion in the southeastern United States, at which the hostess taps her glass and calls out, “Let’s everyone raise our glasses for a toast!” Now, it is clear that the predominantly black staff members who are waiting the tables are not addressed by this superficially universal request. Why not just interpret this, again, as elliptical for a kind-relative box-4 speech act? Well, we might think that this social context sets up a provisional situation in which the black waitstaff are not merely not included in the request, but are rather not hailable or recognizable as persons at all. The hostess and guests really see ‘everyone’ at the party as white, ‘everyone’ as there to celebrate, etc. Of course, the hostess can perfectly well step out of this context in order to address declaratives, imperatives, or other speech acts to the staff, and if a fire breaks out and she yells “Everyone out!” her imperative will presumably include them. She is not incapable of recognizing and addressing them as persons. Yet we might think that for some purposes within that context, the waitstaff are not merely excluded but invisible. The call for a toast isn’t intended to target some people at the party and not others; rather (the tentative intuition goes), the toast occurs within a background context in which the members of the staff do not show up as persons at all. In an importantly restricted sense, we might think, the request is structurally agent-neutral; it is addressed to everyone who is recognizable as an agent within this limited and easily exitable practical context. And yet, it is difficult not to interpret such a phenomenon as reflecting either a mean-spirited bigotry or a kind of blindness or ignorance on the part of the hostess and guests. In the first case, they intentionally exclude from their addresses and practices of recognition agents who they understand could be included. But to do so is already to acknowledge these agents, if only in rejecting or shunning them, and hence it does not meet our criterion for constituting a subcommunity. In the second, more interesting case, the hostess and guests are guilty of somehow failing to see agents in their midst, at least for certain purposes. (It is not uncommon to see people become blind to the personhood of those in the service industry in this way.) They take the waitstaff for granted to such an

204

‘Yo!’ and ‘Lo!’

extent that they are incapable of understanding that their practices are exclusionary and deny recognition. In this case, we would have to say that the speakers misunderstand their own speech acts and their pragmatic import, and miss the fact that the domain to which those speech acts are referenced is universal. Since all speech acts strive to realize an ideal function but may fall short of doing so, this kind of misunderstanding does not detract from the true universality of the discursive community in these cases. Consider another, more challenging case. What are we to make of legal pronouncements concerning “everyone’s” rights and responsibilities? Think, in particular, of laws that try to formulate basic principles of justice, such as constitutional laws. Let us take as an example Item 2 of the Canadian Charter of Rights and Freedoms: 2. Everyone has the following fundamental freedoms:

a) Freedom of conscience and religion; b) Freedom of thought, belief, opinion and expression, including freedom of the press and other media of communication; c) Freedom of peaceful assembly; and d) Freedom of association. Now on the one hand, laws explicitly bind and grant rights only to the citizens that fall under their scope. A non-Canadian cannot use this text in order to claim that his legal rights have been violated. On the other hand, the intended force of the text seems to be inherently universal. The point of the passage is to specify what people are owed simply as agents, not as particular agents within a specific social context, and in particular, not as Canadians. There seems to be an agent-neutrality built into the very structure of a document that establishes such basic rights. The text says what’s right, not what normative status this or that agent, or kind of agent, has. There is no interesting tension or pragmatic structure here if we interpret this bit of text as a declarative assertion. The text can tell us (everyone, all rational agents) a truth about what freedoms should be protected for all rational agents, while at the same time we can all understand that the Canadian laws apply to and protect only the Canadian people. But this seems to flatten the pragmatic texture of the speech act. A founding legal document like this does not merely describe; it establishes legally protected rights. Arguably, the speech act that establishes

Sharing a World

205

them does so for its own citizens, by ostending the agent-neutral force of its own normative claims. The voice of constitutional law, perhaps, is inherently agent-neutral—the law is “inherently general,” we often say— while the scope of the output of that voice is inherently restricted to the citizens under its jurisdiction. If so, then such a founding legal pronouncement may have the structure of a categorical imperative, albeit one whose effective scope is restricted to a particular group of citizens. We would need to do vastly more work on the performative structure of the legislative voice in order to convince ourselves or our readers that this is the correct analysis, or that such speech is genuinely best understood as indexed to a discursive subcommunity. However, we are tantalized by this direction of exploration. Ultimately, we are not compelled enough by such possible recherché counterexamples, at this point, to be convinced that adding discursive subcommunities to our ontology is worth the complications. We thus suggest, tentatively, that there is really only one discursive community— only one ‘we’ made up of rational, mutually recognizing agents who are sensitive to the claims of a public world.

8.4 Sharing a World and Learning to See We saw in Chapter 3 that observatives are essential to discourse, because they are the necessary points of contact with the empirical world in our truth-directed game of giving and asking for reasons. But in order for observatives to serve as anchors to the world in this way, speakers must be able to use Lo-utterances as well. Observatives anchor our discourse to a public world because others can also be brought to see what we see; we can not only express that we have seen x, but also call others’ attention to x so that they can see it for themselves. Now, of course, observation can anchor entitled declaratives even if not everyone can, in fact, see for herself. This is the whole point of passing on declarative entitlements. Mark can legitimately declare that it is ridiculously hot in Florida in July on the basis of Rebecca’s observative expression of this fact—he need not experience the heat for himself. However, if someone purportedly observed x but could never call shared attention to x with a “Lo!”— if x were somehow just not shareable in this way—then we would have to say that x was not observable. And if a declarative x could not trace its

206

‘Yo!’ and ‘Lo!’

justification (in the way discussed in Chapter 3) to observatives sharable with Lo-claims, then it could not be functioning to make an objective truth-claim about the public world. Correspondingly, it is through successful calls to shared attention that we establish, in practice, that we do share an intersubjective world that serves as the tribunal of our empirical claims, and were this not established in practice, then we would not count as a discursive community that shared such a world. Hence the “Lo!,” while not a necessary component of every observative utterance, is a globally necessary element of discourse, and anything that is observable must also (sometimes) be Lo-able. Throughout this book we have made the point that agents generally suffer various epistemic defects that make them unable to take up entitlements they would ideally have, and that this does not hurt the universal scope of claims with agent-neutral outputs. But in the previous section we saw a kind of epistemic defect that we have not discussed before. While we have repeatedly discussed how agents may fail to be entitled to a public fact because of ignorance, poor reasoning, etc., in that section we introduced the literal inability to observe certain public, objective facts or objects or events, because of an inadequate set of cognitive or perceptual resources. To the extent that some of us cannot see x when others can, those of us who cannot are defective in our normative responsiveness to the empirical world. Furthermore, we are deficient participants in the discursive community, because participating essentially involves negotiating and responding to the empirical world, as well as being the successful target of Lo-claims. The kind of ignorance under consideration here is not the sort where someone has merely failed to hear a public fact or to draw an available inference. Rather, even if the defective observer encounters the relevant state of affairs with her own eyes, she will not be in a position to recognize it. There are many examples of this phenomenon, both recherché and mundane: only those trained in reading ultrasounds or MRIs can see the meaningful information they contain; people with autism often cannot perceive common social cues; music experts detect features of musical performances that others cannot; many people are unable to recognize various forms of sexual harassment, class bias, racism, etc.; some people cannot detect happiness or anger in Scandinavians. Changes in our own bodies or even the bodies of loved ones will change our perceptual dispositions and sensitivities—for example, new impairments in

Sharing a World

207

our own body or that of someone we care for can develop our ability to see which spaces are and are not accessible to bodies with these impairments. Thus not all of us are capable of the same observatives, and when we call one another’s attention to something in the world, sometimes these Lo-utterances will fail because the person whose attention they call does not have the capacity to observe that to which she is being called to attend. This signals a defect in that person, and not a compromise of the objectivity or publicity of the object. Differences exist in our ability to observe because our observational capacities are, at least in part, skills that take work to develop. Sellars offers an account in which all observational capacities are skills that must be developed. He argues that if perception is to be able to provide any warrant, its contents have to have conceptual structure sufficient to allow them to bear rational relationships to other conceptually articulated judgments. We must be able to perceive that x is F, rather than just taking in brute sense data. But, he contends, our ability to perceive that some perceptual fact of the form ‘x is F’ holds requires that we grasp the conditions for the appropriate application of the concept F. That is, we must understand the conditions under which things that appear to be F are F, and vice versa. To use his example, I cannot see that a necktie is green unless I understand facts such as that green things look green under natural lighting, that they don’t look green when seen on a black and white television, and so forth.19 Now, grasping such conditions for property recognition involves understanding under what conditions various inferences (such as the inference, in a certain context, to x’s actually being F) are or are not licensed by appearances. Without this normative and inferential mastery, we cannot distinguish between seeing that x is F and it merely looking as though x is F, in which case, according to Sellars, we cannot drive the crucial wedge between appearance and reality that is necessary for our perceptual states to count as properly epistemic states. Hence for him, the ability to recognize a piece of evidence cannot be neatly separated from the ability to use it in inference, and thus perception cannot be taken as a capacity for discovery that lies outside the context of justification. In Sellars’s terms, perceiving that x is F requires that our recognitional epi19. Sellars, Empiricism and the Philosophy of Mind (Cambridge, Mass.: Harvard University Press, 1997), §18.

208

‘Yo!’ and ‘Lo!’

sode be placed “in the logical space of reasons, of justifying and being able to justify what one says.”20 For Sellars, perceptual capacities are inculcated through our contingent histories, “involving a long history of acquiring piecemeal habits of response to various objects in various circumstances.”21 This is the history of our mastery of the ability to recognize instantiations of various concepts, in and through our mastery of their normative and inferential relationships. Thus if x is indeed perceptibly F for a particular agent, this fact is dependent upon the agent’s contingent past. Only if she has the right history will she have developed the capacity to perceive that x is F, and only then will her empirical confrontation with x warrant beliefs and inferences based on the fact that x is F. An agent’s particular history of observational situations and learned responses will inflect the topography of the recognitional concepts she brings to bear in perception, by giving these concepts their life and hence their content within differently inflected spaces of reasons. Our contingent history of concerns, experiences, and conditions of observation helps determine which facts and properties can show up for us and what counts as normal and aberrant behavior for objects of different sorts. Thus these contingent histories will help constitute what evidence is available and which inferences are warranted in the face of worldly objects and events. Sellars writes: “For we now recognize that instead of coming to have a concept of something because we have noticed that sort of thing, to have the ability to notice a sort of thing is already to have the concept of that sort of thing, and cannot account for it.”22 Developing the ability to observe takes work and experience, as Aristotle taught us. We must learn to see features of the world—patterns in ultrasound images, happiness in Norwegians, it seems even green on neckties. How do we do this? Sellars has provided a partial answer: our conceptual understanding of what we see and its inferential relations to various facts (and, we might add, its practical relations to action) cannot come after our perceptual capacity; it must predate it or at least develop in tandem with it. But at the same time, a conceptual understanding of 20. Ibid., §36. 21. Ibid., §19. 22. Ibid., §45, italics in the original. Sellars’s use of the language of recognition here suggests that, in his view, this kind of theoretical insight can also be thought of as a kind of observation.

Sharing a World

209

something does not suffice to give us the ability to see it. I might understand the significance of various kinds of information without being able to detect this information myself.23 We are taught to put our conceptual knowledge into observational practice by other community members, who direct our attention in ways that train up our perceptual capacities.24 Thus, just as “Yo!” not only recognizes someone with a settled identity and normative position but also serves as a tool to constitute this identity and normative position, likewise “Lo!” not only calls others to attend and observe, but also serves to constitute others’ capacity to attend and observe.25 We cannot teach one another to see through explanation alone, as Kant argued in the Schematism chapter of the First Critique, and then again in section 8 of the Third Critique; rather, we can only guide others in their use of their own senses, helping them confront something in the right way with their own eyes. While all discursive agents share the same public world, sensitivity to this world and its normative claims comes in different degrees and forms. For beings who are subject to the normative claims of truth, becoming more accurate, more complete perceivers is a built-in ideal of rationality. Qua members of the discursive community, full and competent participation in this community, including the ability to respond appropriately to any legitimate Lo-claim or other second-person call, is also a built-in ideal. Regardless of the fact that none of us are such perfect observers or interlocutors, our discourse functionally presupposes a potential community of ideal participants, who can take up every entitlement, respond appropriately to every feature of the world, and universally recognize and acknowledge one another. Luckily, discourse itself can play a constitutive role in bringing us closer to this ideal. 23. In this regard, Sellars’s focus on detecting colors may be misleading. Since color is inherently something we see, it’s hard to know what it could mean to understand the concept of a color without having the skill of detecting it. However, this doesn’t generalize. We can understand the concept of a twelve-bar blues chord progression without being able to notice one, or understand the concept of fetal nuchal thickening without being able to detect it on an ultrasound screen. 24. And of course more brutely physical capacities differ as well. Wade Boggs famously claimed to be able to see the direction of rotation of a baseball’s seams while it was traveling from the pitcher toward him at ninety miles per hour. We assume that, if this is true, it signals a difference in hardware from that of an ordinary human. 25. For a detailed comparison of these two constitutive processes see Rebecca Kukla, “Myth, Memory and Misrecognition in Sellars’ ‘Empiricism and the Philosophy of Mind’,” Philosophical Studies 101 (2000): 161–211.

210

‘Yo!’ and ‘Lo!’

8.5 On the Equiprimordiality and Entanglement of ‘Yo!’ and ‘Lo!’ At this point we have argued for the fundamental importance of both vocatives—‘Yo!’s—that draw us into normative relations with one another and place us in social space, and ostensive observatives—‘Lo!’s— that establish our responsiveness to a shared empirical world and make possible reasonable debate about that world. In fact, however, we can tie these two forms of speech act tightly together. Every hail involves an ostension, and vice versa. Unlike the straight observative, an ostensive speech act has an agentrelative output: The ‘Lo!’ calls a particular person to attend to something in a second-person voice, and demands acknowledgment from that person in the form of an observative (“Lo, a rabbit!”; “Oh, yeah, there he is!”). Thus any ‘Lo!’ also functions as a vocative—not just in the generic sense that every speech act has a vocative moment, as we have argued, but in the much more direct sense that it recognizes someone secondpersonally as a potential observer of something, and calls for that person to acknowledge this recognition by attending. The shift of attention is the kind of acknowledgment this hail demands, just as raising a hand and saying “Here!” is the kind of acknowledgment that having one’s name called during roll call demands. At the same time, hails are material events, and people (even in their virtual incarnations on an instant-messaging screen) are perceptible objects. Whether a hail occurred, what its character was, and whether it was legitimate are all empirical questions that have to be answerable and debatable with reference to the testimony of the senses. The recognitive call of the vocative is a call for recognition in return. Hence any vocative functions to call the attention of the target of the hail to the one doing the hailing. In other words, the hailer must ostend himself, qua material, observable entity, in hailing. Thus every ‘Yo!’ functions at the same time as a ‘Lo!’26 We can generalize the point: we have already argued that every speech act contains a vocative moment. Correspondingly, given that speech is made up of concrete, material transactions among observable agents, every speech act calls the attention of the agents to whom it is 26. We saw earlier in this chapter that the hail also calls its target’s attention to the norms that govern her response. This, as we pointed out, is also a kind of a ‘Lo!’, although one that calls for a less concrete form of observational attention.

Sharing a World

211

spoken to the speaker.27 We cannot speak to one another unless we can perceive one another (where, again, this perception may be mediated in various ways); the recognitions that we have argued are fundamental to all discursive transactions are literal receptive encounters.28 Thus our capacities to utter and to be the target of ‘Lo!’s and ‘Yo!’s are equiprimordial and fundamentally entangled. More generally, we can now see that our social placement as agents within the discursive community and our responsiveness to the normative claims of the empirical world are interdependent. I can have a place within social space and enter into discursive transactions with others only if I can recognize those others as specific material beings and their speech acts as specific material events—the kinds of things about which I can stand corrected by the empirical testimony of my senses. But at the same time, we have seen that the notion of truth is an inherently intersubjective notion, in that it makes reference to a public world—the type that we share with one another and can disagree about. To be normatively responsive to the truth—and hence to inhabit the space of reasons—I must be able to judge what belongs to the public world and what does not, or in other words, I must be able to tell which things can properly be ostended (Lo!-ed) for other members of the discursive community, assuming that their perceptual capacities are up to snuff. This is another way of putting Sellars’s point about the relation between looks and seeings-that, which he in turn inherits from Kant’s notion of objective validity: in order to be a truth-discerner, we must be able to distinguish between x looking F to me and my seeing that x is F, and this requires my ability to determine whether x and its F-ness belong to a public world and can serve as targets of shared attention. We are far from the first philosophers to argue that genuine normative agency requires both placement within social space and sensitivity to the claims of the empirical world. Kant, Heidegger, Wittgenstein, Sellars, Haugeland, and others have argued for the interdependence and necessity of these two forms of normative placement. Heidegger in par27. In Chapter 7 we discussed a variety of uninteresting qualifications and exceptions— speaking to one’s self, calling out to see if anyone is listening, etc. 28. See Joseph Rouse, How Scientific Practices Matter: Explaining Philosophical Naturalism (Chicago: University of Chicago Press, 2002), for an excellent discussion of the embodied character of discourse and the corresponding fundamental inseparability of discourse and perception.

212

‘Yo!’ and ‘Lo!’

ticular has identified Dasein with discursively articulated being-in-theworld, where such being is essentially both with-others and responsive to empirical phenomena; Mitda-sein, or being-with-others, is “existentially constitutive for being-in-the-world.”29 However, we believe that the second-personal, vocative dimension of normative agency has been lost in all of these accounts. While Heidegger paid careful attention to the role of the first-person perspective (remember that Dasein is “in each case mine”), his description of Mitda-sein in Being and Time is oddly passive. Others are “there with” Dasein, and one is “among” them and “encounters” them, and is concerned for them, but there is no talk, in this text, either of vocative discourse or of the phenomenological form of the second-person encounter. In reading Being and Time, one could get the impression that we necessarily speak and act surrounded by one another, but not to one another. A similar criticism can be leveled against all the other authors we just mentioned. But we have argued in this book that it is only through second-person discourse that we establish a place in normative space and are engaged in responsive normative relations to others. Vocatives call us into relations of mutual recognition, while ostensions establish our shared normative sensitivity to the features of a public world. An account that leaves out this second-personal dimension of our being-with-others will have no tools for explaining how we make real, concrete claims upon one another and speak to one another about a shared world.

8.6 Fugue In the end, almost all the work we have done in this book has been a matter of spelling out the consequences of clarifying three distinctions: between the input and the output of speech acts, between agent-relative and agent-neutral normative statuses, and between the different voices and directions (first-personal, second-personal, impersonal) that form part of the pragmatic structure and import of speech acts. We have cycled through these three distinctions over and over again and put them together in various ways. The speech acts that have been central for us, but have played little or no role in mainstream philosophy of language— 29. Heidegger, Being and Time, trans. Joan Stambaugh (Albany: SUNY Press, 1996), paragraph 121.

Sharing a World

213

such as observatives, prescriptives, vocatives, and acknowledgments— are functions that are built out of combinations of these different scopes and voices. In turn, these functions have been the basis for our accounts of various phenomena such as holding, calling attention, and observing. We hope we have shown that once we have these three distinctions clearly in view, various formerly resilient philosophical puzzles in epistemology, philosophy of mind, and metaethics dissolve, while our capacity to describe and understand the pragmatic texture and topography of speech is enhanced. We hope in particular that seeing these distinctions can make our capacity to respond to normative claims and engage one another in discourse that is held to the testimony of a shared empirical world seem less philosophically opaque. Certainly our typography and distinctions are not uniquely useful in teasing out the pragmatic topography of the space of reasons, and we make nothing like a claim to have captured the subtlety and character of this space exhaustively. Yet we are surprised at the extent to which one can articulate and appreciate its complexity with just this small repertoire of conceptual tools. We began by charging the philosophical tradition with a pervasive and tenacious propensity to commit the declarative fallacy. This fallacy single-handedly masked all three of these distinctions quite effectively. In a legitimate declarative speech act, the input and the output are both agent-neutral, and they are identical to one another: a speaker is entitled agent-neutrally to a truth claim, and the speech act strives to pass on entitlement to this same claim agent-neutrally. As long as we philosophers focused almost all our attention on such speech acts, it is not surprising that we did not think to worry about the distinction between inputs and outputs, since in these cases they coincide. Nor is it surprising that we failed to focus on the distinction between agent-neutral and agentrelative statuses, since we encountered only the former in the course of our philosophical work. Finally, declarative speech is distinctive in its impersonality. It is part of the character of the declarative that the first-person perspective and the second-person target drop out as functionally irrelevant. Truth-claims are not voiced or indexed to perspectives. Our discipline’s insistent and nearly exclusive focus on declarative speech has not only masked the significance of voice and perspective to discursive pragmatics, but even made these considerations almost unseemly, as they threaten to sully the objectivity of such speech. Indeed, in writing this book, we have struggled to find ways of articulating these

214

‘Yo!’ and ‘Lo!’

dimensions of language. For example, linguistics and philosophy of language have not provided us with resources for discussing the direction and the pragmatic voice of speech, as opposed to its grammatical voice. When we avoid the declarative fallacy, it becomes undeniable that speech varies in its voice, direction, and scope, and that its entitlement can be different from its import. Ultimately, we are less invested in convincing you that our particular arguments and conclusions are correct than we are in saying: Lo, the wide variety, importance, and rich texture of non-declarative speech!

Appendix Index

Appendix: Toward a Formal Pragmatics of Normative Statuses With Greg Restall In this book we have focused almost exclusively on the pragmatics of language, and resolutely on a pragmatist approach to pragmatics—that is, an approach that attends always to the concrete embodied character of language. This is not to indicate that we think semantics uninteresting, or that we think no insight into language can be had from accounts that look at it from an abstract point of view, even from the point of view we have uncharitably called a scorecard changing in Platonic space. Our objection was to the supposition that language could be understood entirely in terms of some abstract normative structure, and not meant to imply that such a structure could give no illumination of anything. With this in mind, we turn in this Appendix to the preliminary development of a scorekeeping semantics that begins with the broader field of vision that is opened up when we eschew the declarative fallacy and make use of abstract versions of the pragmatic distinctions developed in this book. In so doing, we take on in many ways the point of view that we have been at pains to resist, that of the detached theorist characterizing features of the abstract significance of performances by agents in a community in which we do not participate. Our primary concern here is to reassure. Many in the pragmatist tradition, often following Brandom’s work in this regard, are keen to develop formal inferentialist accounts, and we insist that nothing we do here should be seen as preventing such investigation. Indeed, we claim that such projects look much more promising against the broader field of possibilities opened up by our more systematic pragmatics. That is, we show that a notion of propositional content and inference is straightforwardly definable in terms of 217

218

Appendix

the normative transitions between acts that we discuss throughout the book, and we give one example of how to develop other types of speech acts. The underlying idea throughout this book has been to understand language as a prescriptively constrained social practice. Thus, in laying out its structure, one naturally begins with a representation of the kinds of prescriptive significance that can be practically conferred upon actions, and then attempts to understand language in general, and semantics in particular, as a special sort of structure of instituted relations arising out of such a generic system. The most detailed such program is Brandom’s, and it is his approach that we aim to make contact with here. Our account will arrive at the point at which we can cut and paste the semantic work of Making It Explicit. But our account has far more general application. The pragmatic back-story could, with fairly minor modification, be adapted to generate truth-conditional contents, verification conditions, or what have you. Our back-story gives an account of how a practice in which agents evaluate one another’s actions can be seen to include actions such as declarings and prescribings, and we give an account of the normative structure of such a practice. We do not prescribe how you might go on to account for the “semantics” of these declarations and prescriptions. The basic picture within which we place our formalism is this: we are the theorists; those creatures over there, engaged in social practice in their natural environment, are what we are interpreting. We interpret them by taking them to be scorekeepers, that is, creatures who both engage in actions and assign prescriptive statuses to their own acts and those of their fellow creatures. Of course we don’t think they do this using our vocabulary, or any vocabulary at all. Rather, we interpret their practice as implicitly involving such attributions and undertakings of prescriptive statuses. This picture, we note again, looks very different from the typical context of communicative speech that we discussed in the book; it requires taking up an impersonal, outsiders’ stance for the purposes of interpretation. This is not to go back on anything we said earlier in the book. Instead it is to acknowledge a difference between a stance that may be productive for the purposes of doing inter-

Appendix

219

pretive semantics and a stance that enables actual communicative encounters.1 We offer no reductive account of these takings in terms of behavior. Indeed, for present purposes we simply help ourselves to notions such as, roughly, “a practice such that x is committed to taking y to be entitled to z” and the like. But we can say various things about typical manifestations of assignments, defeasible connections between behavior and status, and so on. Thus, a creature who takes another to be committed to doing something will generally expect the other to take that action, and will be surprised if he doesn’t. She will also typically engage in some form of censoriousness toward one she takes to be committed to something that he doesn’t do, and perhaps offer encouragement or reinforcement to one who discharges his commitments. Similarly, acts that are taken to be performed without entitlement will receive some kind of negative sanction, while attempts by others to sanction an entitled act will be resisted. (Think of such rules of thumb as akin to the sorts of things Davidson points to as relevant evidence regarding the assignment of occasion sentences to speakers in radical translation.) In the end, one would like an epistemological account of how one comes to exhibit the understanding constitutive of such implicit assignments of significance, but for purposes of developing the semantic framework, all this is simply assumed. Now what sorts of things are on scorecards? Well, there are various agents and actions they perform. Acts by agents, actual or potential, are the things one assigns status to. And the scorecards will assign to these acts-by-agents various prescriptive statuses. Which statuses? Here we are faced with a choice, for there are clearly a wide range of prescriptive statuses that agents can and do assign to one another’s actions. The framework we develop puts no constraints on which of these could be modeled, but we aim to begin simply, taking on board a few selective examples, including the two prescriptive flavors employed by Brandom— commitment and entitlement. So Jones’s prescriptive position in practice will be represented by a scorecard which lists things such as that Smith 1. Other philosophers who have described interpretive contexts, such as Brandom and Davidson, have not made this distinction; however, their descriptions of the attitude appropriate to doing interpretive semantics are deeply unsatisfying as descriptions of the attitude appropriate to speaking to and with others.

220

Appendix

is entitled to collect acorns today, and that Williams is committed to cleaning the nest. Several points need to be made about these scorecards at the outset. • As we said, for us interpreters to assign such a score to Jones—say

that he assigns a certain commitment to Smith—is not for us to suppose that Jones has the conceptual ability to form beliefs about Smith. We do not begin our account by postulating anything with the structure of a speech act. • The set of prescriptive statuses that can be assigned to acts must include one more than the pair of commitment and entitlement. As we noted, a paradigmatic exhibition of taking someone to be committed to something is to sanction him if he doesn’t do what he is committed to. But this highlights the fact that there is little sense to a practice in which we keep track of one another’s commitments, but don’t keep track of which commitments are acted upon. Neither does entitlement have a point if we don’t keep track of when someone has performed an entitled action, or one to which he was not entitled. So we include the status of “done,” emphasizing that it is a normative status, rather like having scored in a game (succeeded according to the rules). • Scorecards should be coherent. This is not to deny that scorekeepers can be incoherent (locally, and to a certain extent). But for us to have an interpretation that makes sense of what they are doing—even if what they are doing is incoherent—we must be coherent. Sometimes, for example, writing down that Jones has scored things one way requires us to write down that he has scored things another. (Think of assigning to Jones the status of being married to Smith without assigning to Smith the status of being married to Jones.) And sometimes there will be two scores which it makes no sense to write down on the same card.2 • A word is in order about how we are understanding the notion of entitlement here. One could certainly define this status in such a way that it applies only to actual performances, that is, to things that are done. On this definition, an actual act is either entitled or 2. Of course once we have a recursive specification of an infinite range of actions, the implication relation requires us to cease taking “write down” literally. Then we think of implication as telling us what we have implicitly written down in virtue of an explicit writing.

Appendix

221

not, but potential acts simply are not at issue. This is not the only thing one could mean by entitlement. One could think of agents as having entitlements that they keep in their pockets, as it were, a sort of performance coupon. In this sense, entitlement, like commitment, applies to potential or actual acts. This understanding, which is the one we employ, raises a further question of whether the coupons are “combinable with any other offer.” That is, we need to ask whether entitlement to A and entitlement to B implies entitlement to do both. Though either answer can be stipulated, generating a coherent normative status, “no” is the more intuitive answer and the one we follow. Mark can be entitled, in the usual sense of this word, to marry Sam and entitled to marry Kelly, but not entitled to marry both. (This question does not arise for the notion of entitled performances, when both actions have been performed.) Another issue will then arise for our definition of entitlement incompatibility below. One could stipulate that A is incompatible with B iff one cannot be entitled to A and entitled to B. On this reading, ‘Mark marries Kelly’ is compatible with ‘Mark marries Sam’. Since it seems more intuitive to treat these as incompatible—in part because we generally would like to say that if two things cannot be coherently assigned the status of “done,” then they can’t be compatible in the entitlement sense, but more importantly because this is the notion that is useful in developing semantic inferential relations—we define incompatibility in terms of the impossibility of entitlement to perform the joint act A and B. That is, we say that two acts are incompatible if it cannot be the case that one performs both and is entitled to both. • We note that though normative statuses apply to act-types, these can be as specific as one likes. One can be committed to taking out the trash, to taking out the trash between 8:00 and 8:05 with a shovel, etc. But however specific, we think of these as one-off commitments rather than as “standing commitments.” That is, we think of a commitment as the sort of thing that is taken care of, or discharged, when one does what one is committed to doing. This is not true of all commitments, of course. Greg’s commitment to be collegial toward Rebecca is not something that is “discharged” when he performs some collegial act. It is an ongoing commitment, demanding frequent vigilance against uncollegial behavior. How

222

Appendix

one would represent these standing commitments in a more complicated system—perhaps as a set of conditional commitments of the form “if this situation arises, one is committed to acting collegially there,” or no doubt something more complicated—is not something we take up in this Appendix. • We see these statuses as applying to acts in general. We neither make nor assume any sort of distinction between speech acts and other acts. That distinction will be constructed in due course. But there is a second form of generality that requires more explanation. That is, no serious distinction is made in the abstract practice between “agents” and “objects.” We begin our characterization of social practice at a level rather like what Sellars refers to as the “Original Image,” an image in which there are not as yet conceptual or practical resources to distinguish agents from mere objects. This is the point of view we begin with, allowing in the formalism that one can attribute commitment to the clouds to bring forth rain and of course one then also keeps track of which of these commitments have been carried out. Just as with assignments to humans, such attributions are implicit in our expectations, exhortations, (allegedly) punitive or censorious responses, etc.

That events in the world play a role in determining the normative significance of our own acts is a crucial element of a genuinely embodied practice, and this is something that we want to have represented in the formalism. Of course this is not to deny that the prescriptive involvements of non-agents in our linguistic practice ought to be circumscribed. The rising of the sun plays a role in generating my entitlement to the speech act “Lo, the sunrise!” but is not something the sun can genuinely be entitled to perform. But the fact that the sun does not perform actions is a substantive fact. It is something to be argued for within the space of reasons, not something to be legislated in our definition of that space. Further, the difficult task, as it happens, is not to circumscribe non-actions, but to get them into the language game to begin with. We should draw the agent/object and act/event distinctions, to be sure, but only from within a space of normative significance in which each has a logically antecedent function. Just as Sellars speaks of the Manifest Image as arising out of the Original Image as we strip away agential content from our conception of non-

Appendix

223

human entities, so a notion of agent—and non-agent—will emerge here. But the stripping away of these statuses is something one can only do within a practice that has sufficient conceptual resources to utter speech acts, specifically prescriptives, and to make explicit underlying normative statuses with that speech. Thus, one will say things like “Stars don’t have commitments.” But although stars are not agents, their inclusion in the practice of normative attribution leaves a trace on the prescriptive role of happenings involving them in two ways. First, relations between statuses give content to act types. And those relations hold whether or not the entity in question is actually capable of having that status at all. So commitment to picking something up is commitment to moving it. And this leaves a trace in the fact that to say that the wind picks up our trailer is to say that the wind moves it, even when we develop the conceptual wherewithal to distinguish the wind’s picking something up from a genuine doing. Second, we continue right to the end to assign statuses of doing to non-agents. To say of the cloud that it carried out (did) its commitment to rain upon the ground need not, however, imply that it was literally committed to doing this or to its being a genuine agent. It is, rather, merely to take the cloud to have, as if were, done what would count as that commitment were it to have it, or to have changed in a way that, were it an agent, would be a doing of what would be a carrying out of a commitment to rain, were there such a commitment. In short, a space of doing arises as a space articulated by the roles happenings are fit to play in relation to commitments and entitlements, but roles they need not always actually be playing. Taking something to be a mere happening is taking it to be a doing minus the agentive commitment. Our aim, then, is to provide a formal vocabulary in which we can discuss prescriptive pragmatics of action in a rigorous way. This requires three basic kinds of things: • A non-empty set of agents (AGENT), • A non-empty set of action-types (ACTION), • and a non-empty set of prescriptive statuses (STATUS).

We will call a triple consisting of a choice of sets AGENT, ACTION and STATUS a field of play. We will use lowercase Greek letters , , , , etc., as variables ranging over the class AGENT of agents. When we wish to use concrete examples of agents, we will use Mark, Rebecca, and Greg. F,

224

Appendix

G, H, etc., range over the class ACTION of act-types, and s, t, etc., are statuses—the canonical examples are commitment (c), doing (d), and entitlement (e), though the formalism leaves room for any number of other prescriptive statuses should they prove useful. Here is how the different aspects of a field of play may combine. Statuses are used to evaluate act-types relative to agents: given any action in ACTION it makes sense to evaluate it with respect to each person. This makes sense whether or not the action is one that is, or even could be, carried out. So the basic unit of our analysis is a triple of the form s 〈F, 〉 where s ∈ STATUS, F ∈ ACTION and ∈ AGENT. We will call such triples prescriptive assignments. When our formal syntax allows, we will abbreviate this language: instead of always writing s 〈F, 〉 we may write sF if this expression is not ambiguous.3 So a field of play is the basic raw material that a theorist may use in giving an account of the prescriptive behavior of some community of agents. A scorecard appropriate for a field of play will be some collection of prescriptive assignments. However, not every collection of assignments 3. Here are some examples of prescriptive assignments: c 〈help the homeless, Greg〉 This records Greg’s commitment to help the homeless (which is a commitment that Greg may or may not manage to meet, or even to recognize). e 〈assert that R has 3,088 Ackermann constants, Rebecca〉 This records Rebecca’s entitlement to assert that the relevant logic R has 3,088 Ackermann constants. d 〈ask for directions to the Baillieu Library, Mark〉 This records that Mark has asked for directions to the Baillieu Library. Any kind of action, whether a straightforward bodily movement or a complex multi-stage project, whether an inarticulate doing or a highly structured verbal speech act, may count as an action in a field of play, since it makes sense to judge agents to be committed to or entitled to each of these different kinds of act-types. Notice that nothing in our picture so far makes reference to any restriction of the capacity of an agent to carry out each act-type in ACTION. If the class ACTION contains suitable actions, it makes sense to ask the question of whether or not c 〈square a circle with ruler and compass, Mark〉 even though this commitment is one that he will never, and can never, discharge.

Appendix

225

counts as a scorecard: there are a number of constraints governing the interaction of different statuses. The general form of a structural constraint on scorecards has the form X dY where X and Y are sets of assignments. A scorecard respects the constraint X d Y if and only if it contains each assignment in X only when it contains some assignment in Y. That is, it does not contain every assignment in X and no assignment in Y. Notice that scorecard constraints satisfy the usual structural rules governing consequence relations. Every scorecard respects the constraint sF d sF , and more generally, any constraint X d Y in which some assignment sF is in both X and Y. This is an identity constraint. Similarly, if a scorecard respects X d sF ,Y and it respects X, sF d Y then X d Y as you can check. This is a transitivity condition, also called the cut rule. An important special case of such a constraint is the form X d, where the right-hand side is empty. A scorecard violates this constraint when it contains every assignment in X. It respects this constraint iff there is some evaluation in X that it avoids. So to say X d is to say that the assignments in X are jointly incompatible, since no scorecard containing every element of X is correct. We will also write “z X” for this. In the very special case where X = {A, B}, we write “A z B.” Incompatibility notions will play a central role in the rest of this Appendix. Given our original notion of incompatibility z between assignments we may define a weaker, but equally important, incompatibility relation. Consider two actions F and G , both of which can be done, but which cannot be jointly entitled. As noted above, we mean by this not that one cannot be entitled to the one and also entitled to the other, but that one cannot be entitled to both. For example, it is a good thing to conserve natural resources for future generations. Rebecca and Greg might be trawling the ocean for fish. Rebecca takes a load of fish, and so does Greg. Jointly, their haul endangers the survival of this species, and so we take it that Rebecca’s action and Greg’s action are not both entitled. Though it may well be that either of them would have been entitled if the other had not done what s/he did, it is impossible that both perform their actions with entitlement. Given what both did, one of them (if not both) has overstepped the mark of her or his entitlement. (Perhaps the

226

Appendix

first to get the catch was entitled but the second wasn’t. On the other hand, they may both have been in error.) Regardless of the details, even though we have both d 〈catch n fish, Rebecca〉, d 〈catch m fish, Greg〉 the two actions are incompatible in another sense: they are entitlement incompatible, since the following assignments are incompatible: e 〈catch n fish, Rebecca〉, d 〈catch n fish, Rebecca〉, e 〈catch m fish, Greg〉, d 〈catch m fish, Greg〉 In general, we will say that a set X of d-evaluations are entitlement incompatible if and only if the union of X and the set X[d := e] of corresponding e-evaluations (found by replacing each d by e in X) is incompatible in our original sense. That is, for any set X of d-evaluations, ze X iff z X[d := e] ∪ X We may do the same thing for commitments. A collection of commitments may also be jointly incompatible, this time in different ways. Firstly, commitments may be undercut because “the world fails to cooperate.” To return to our fishing example, if flushed with success, Rebecca and Greg return to sea with plans to fish more, c 〈catch 2n fish, Rebecca〉

c 〈catch 2m fish, Greg〉

but in waters where there are fewer than 2n + 2m fish, then their commitments will not be jointly discharged. We can say that these commitments are d-incompatible, since d 〈catch 2n fish, Rebecca〉 z d 〈catch 2m fish, Greg〉. In general, given a set X of c-evaluations, zd X iff z X[c := d] In a similar fashion, it makes sense to extend the definition of e-incompatibility to cover arbitrary sets of evaluations (d, e, and c). Such a set X is e-incompatible if and only iff X[c := e, d := e] ∪ X[c := d, e := d] is ordinarily incompatible. A set of commitments, entitlements, and doings is e-incompatible in this sense just when (joint) entitlement to all those actions is not available. For evaluation to be genuinely social we not only need to be able to take creatures to be taking each other to be committed, doing what they are

Appendix

227

committed to, and being entitled in their actions (or not), but we also need to take these creatures as acting in their assignments of prescriptive statuses to actions. If these assignments are not themselves seen as actions, then they will not themselves be the sorts of things that can be entitled or not within the practice, and so not the sort of thing that could be brought within the space of reasons as fodder for argument. That is, once we think of a creature as taking to have been committed to F, then we can think of this very taking as yet another act-type, to which it can be entitled or not. So, let’s call a family ACTION of act-types in a field of play 〈AGENT, ACTION, STATUS〉 an assignment-rich set of actions if for each α ∈ AGENT, F ∈ ACTION, and s ∈ STATUS, there is an act-type sF ∈ ACTION. This is the act-type of ‘taking ’s F-ing to be s’. In the cases that concern us, the relevant actions are taking ’s F-ing to be a commitment, or to be done, or to be entitled. Given an assignment-rich set of actions in a field of play, it makes sense to consider what kinds of constraints are appropriate. It seems quite appropriate to hold agents to exactly the same criteria to which we hold ourselves in theorizing about them. In other words, if we take it that X d Y for some sets X, Y of assignments, then this should not only constrain our account of the statuses of creatures—it also constrains theirs. If I cannot take F to be an entitled doing, then neither should or any other creature in AGENT. To state this in its generality, we add a constraint of ‘sociality of assignments’, which requires the simple idea of a ‘lifting’ of an assignment. Given a set X of assignments its ‘d-lifting to the agent ’ is the following set of assignments: X dβ = {d 〈sF , 〉 : sF ∈

}

In other words, given a set of assignments, its d-lifting to is the set of accomplished actions performed by to make those assignments. Now the sociality constraint is easily stated: If X d Y then X dβ d Y βd So now, we consider a field of play with an assignment-rich set of actions, satisfying the sociality constraint. Not only do we the interpreters take the agents to be taking each other to be committed, entitled to actions, and to be doing them, but now we have the resources to describe the agents doing the same. * * *

228

Appendix

We are now in a position to characterize speech acts formally. As one would expect, our characterization of the declarative will depend crucially on the agent-neutrality of input and output. As it happens, the formal property that corresponds to agent-neutrality is quite simple, namely that some incompatibilities are ideally universal. There is some picking and choosing at work here, both in terms of how abstract a characterization to give of the various elements of concrete practice, and in terms of what particular elements to represent in the formalism. In Chapter 1 we noted that when one utters a declarative, one takes up a normative position of opposition to anyone who utters a declarative incompatible with it. This normative stance is, in some ways, an abstraction. One certainly won’t be aware of everyone who denies what one says. One may not even recognize that certain declaratives are incompatible with what one says. But nonetheless, the universality of one’s stance, in this sense, is a crucial constitutive goal of the act of declaring. Another important dimension of speech acts with agent-neutral output is that they strive to make entitlement available universally. Here, as we argued, the distance between practical import and structural ideal is even greater. Rebecca may be entitled to believe P, while Greg is committed to denying P, and hence not in a position to take up that entitlement, and Mark may simply be unaware of Rebecca’s entitlement. Nonetheless, we urged that it is a constitutive feature of declarative assertion that it strive for universal uptake of entitlement. In our formal theory below, we make use only of the former abstraction, for a reason no more principled than that this is the feature that is needed for the construction of the relevant sort of propositional content. With this in mind, we now characterize a declarative speech act, an act type D(F, ) of a declarative: saying of that it F-ed.

Declaratives An act type D(F, ) is a declarative (saying that following three conditions:

did F) if it satisfies the

d 〈D(F, ), 〉 d d 〈dF , 〉 e 〈D(F, ), 〉 d e 〈dF , 〉 If ze {d 〈D(F, ), 〉, d 〈D(G, ), 〉} then ze {d 〈D(F, ), 〉, d 〈D(G, ), 〉}

Appendix

229

Consider these constraints in turn. The first two connect declaring that something is the case with practically taking it to be the case. For to perform the declarative D(F, ) is, in part, for to take to have F-ed. Indeed, it is in many ways the central function of a declarative to mark explicitly and publicly the taking of something to be the case.4 Similarly, entitlement to the declarative implies entitlement to the corresponding practical attitude. The third constraint is our notion of agent-neutrality. Since our goal is to reproduce incompatibility and entailment relations, the crucial aspect of agent-neutrality is that, in declaring something, we take up a stance of rhetorical opposition to anyone who has endorsed a claim incompatible with ours. If Rebecca would contradict herself in uttering two declaratives, then she and Greg have disagreed if they each assert one. Recall that it is just these entitlement incompatibilities that Brandom makes use of in his semantic constructions. Though an assertion, for Brandom, is the undertaking of a justificatory commitment, the incompatibilities between such undertakings are not commitment incompatibilities. The idea of incompatibility he is working with does not preclude asserting both that an object is red and that it is colorless. Obviously one can assert both. It precludes, rather, the idea that one could be entitled to both, in the sense that one could assert both with entitlement.5 The reader may wonder why our second condition above is not strengthened to a biconditional. The answer draws once again upon the agent-neutral nature of declaratives. When someone says something, she places herself in a normative position of opposition to anyone who 4. Of course one can do this disingenuously. (For example, by lying.) And in this case there is a psychological sense in which one who says that something is the case doesn’t “take it” that it is. But what is at stake here is normative status, not psychological description. And regardless of whether or not the action was sincere, if we take to have performed the declarative, we are thereby ascribing to that taking. 5. Brandom himself often defines incompatibility as a normatively mixed statement: incompatibility of P and Q is defined as commitment to P precludes entitlement to Q. But it is hard to see how this relation holds generally among even paradigmatically incompatible claims. Suppose someone is looking at a red can and so asserts that the can is red. Suppose she also has endorsed a complicated and systematic quantum theory of color, that unbeknownst to anyone implies that no object can be red. Given this, she is also (implicitly) committed to the claim that the can is not red. But does the latter commitment mean that she isn’t entitled to claim that the can is red? Certainly not. It is the theory to which she isn’t entitled. So commitment to the claim that a can is not red does not (always) preclude entitlement to the claim that it is. Thus, on Brandom’s definition, the two propositions are not incompatible.

230

Appendix

denies what she says. In representing the ideal to which declaratives strive, we must see them as normatively universal. Entitlement to the claim is “our” knowing, or “one’s” knowing. But we cannot assume that such a requirement applies to all mere takings. Recall that the notion of a taking with which we began did not depend upon the institution of any structure of public speech acts whatsoever. Thus, there are prelinguistic takings that do not constitute issuances of a universal re-taking license. So not all entitlements are ideally universal. Hence mere entitlement to a taking is weaker than entitlement to a declarative. Finally, we note that the first and third conditions imply the following condition: If dF z dG then ze {d 〈D(F, ), 〉, d 〈D(G, ),

〉}

So the third condition above means that we have an incompatibility role that is independent of agent, and hence amounts to the reproducible content of an act type. Since the incompatibility of declaratives is invariant under who is doing the declaring, we can talk simply of the declaratives themselves as being incompatible. This means that we can begin to look simply at the incompatibility role of these declaratives as itself a kind of inferential content (defining inference in the familiar ways in terms of normative incompatibility). And now, as promised, we can cut and paste! That is, at this point we simply can take on board the Brandomian semantic project with only minor revisions. Given incompatibility roles, we define inferential relations. Given inferential relations, we then develop substitution inferences, (de)compositionally combinable subsentential content, conceptions of predicates, singular terms, logical vocabulary, and quantification.6 We have shown how practice can institute linguistic meaning. But the meaning instituted by properly constrained practice does not stop with 6. For substitution inferences, subsentential content, singular terms, predication, and logical vocabulary, see the relevant sections of Making It Explicit. For more details on the logical structure of the formal inference relation, see Mark Lance, “The Logical Structure of Linguistic Commitment III: Brandomian Scorekeeping and Incompatibility,” Journal of Philosophical Logic 30 (2001): 439–464; and for quantification see Lance, “Quantification, Substitution, and Conceptual Content,” Nous 30 (1996): 481–507. Readers may suspect that we have only instituted propositional contents for one-place predicates. Though we do not pursue the matter here, it is not hard to define relations in the terms of this Appendix. This and many other technical issues will be pursued in a future work.

Appendix

231

the declarative. The advantage of the current framework is not that it rederives familiar semantic constructions starting a step earlier, but that it does so from a broader context that allows a richer range of constructions. Thus, for example, the structure of imperatives can be clarified with only a small alteration on the account of declaratives. The main change is that we do not require that incompatibility relations be agentneutral, either regarding entitlement to the imperative or regarding the commitments that follow from warranted imperatives. That is, e 〈I(F, ), 〉 d cF Though we do not pursue the issue here, the latter agent-relative commitment generates, via agent-relative incompatibilities, an agent-relative consequence relation as well—one that is quite distinct from the relation of declaratival consequences. Thus, if Rebecca orders Greg to help homeless people, and does so with entitlement, Greg is committed to helping homeless people. But whereas the declarative content ‘Greg helps homeless’ people entails that there are homeless people, Rebecca’s imperative was not an order that homeless people should exist, that Greg should see to it that they do, etc. If we develop the “content” of an imperative in terms of what is normatively licensed by the output commitment, we can expect this to fall out quite naturally.7 Before closing, we work the analysis through one more of our pragmatic speech act types. Consider prescriptives, which include ought7. The fact that we build in normative agent-relative transitional appropriatenesses from the beginning is a huge formal advantage in developing logics of action, deontic logics, and the like. Even Belnap, we would claim, falls prey to a version of the declarative fallacy in his STIT work. See, for instance, Nuel Belnap, Michael Perloff, and Ming Xu, Facing the Future: Agents and Choices in Our Indeterminist World (New York: Oxford University Press, 2001). Though the embedding within a temporal structure makes for a sophisticated treatment, each element of the structure is given by the set of declarative facts true there. This has unfortunate consequences in various cases. Imagine that Jones hits a deep fly ball. Smith decides to run in, realizing too late that he cannot reach the ball. Had he run out, he could have made a leaping catch of the ball above the wall, but now it falls over the wall for a home run. Any nontheoretical characterization of this case will hold that Jones hit a home run. But the STIT analysis holds that he didn’t, since his act did not make it determinately true that a home run would result. Only the corporate body of Jones and Smith hit the home run. We would diagnose this problem as arising from an attempt to build normative attachments to particular agents out of agentneutral facts and causal significances of those facts. The normative individuation of action is taken up in ways that are broadly in line with the account of this book in Chauncey Maher, “Counting What We Do: A Normative Functionalist Approach to Action Individuation” (Ph.D. diss., Georgetown University, 2008).

232

Appendix

claims and similar deontic claims. (Since we are working at the moment only with commitment and entitlement, the ought-claim will be our only example of a prescriptive. If we were to introduce weaker deontic statuses, we could characterize other prescriptives.) As we argued in Chapter 5, prescriptives are characterized by agent-neutral inputs and agent-relative outputs. Thus, prescriptives share the agent-relativity of their output with imperatives. If Mark is justified in saying that Greg ought to help the homeless, then Greg is committed to doing so, just as with an imperative. But Mark’s entitlement to the speech act is agentneutral, like a declarative and unlike an imperative. This is the sense in which one must justify an ought-claim via reason and receptivity, rather than secure social status. Mark saying that Greg ought to do x and Rebecca saying that Greg ought not to do x are incompatible in the typical (agent-neutral) declarative sense. More precisely:

Prescriptives An act type P(F, ) is a prescriptive (such as saying that satisfies the following three conditions:

ought to F) if it

d 〈P(F, ), 〉 d d 〈cF , 〉 e 〈P(F, ), 〉, d 〈P(F, ), 〉 d cF If ze {d 〈P(F, ), 〉, d 〈P(G, ), 〉} then ze {d 〈P(F, ), 〉, d 〈P(G, ), 〉} The possibility of a prescriptive act—one that draws entitlement from the socially instituted space of giving and asking for reasons in order to justify commitments on one or more agents—is essential to our account, and to the idea of building genuine semantically significant normativity from an “original image” form of interaction with the world. So far, we have spoken as if the normative statuses undertaken by various agents in a field of play, together with the constraints on scorecards, are simple sociological facts. But if one builds semantic content out of such normative statuses, such an understanding will not do, since semantic content must be rationally and empirically assessable. But assessing the normative statuses that institute semantic content requires the ability to distinguish between declarative judgments to the effect that, say, everyone takes it that an act is entitled, and claims that such entitlement is genu-

Appendix

233

ine. And it requires as well the ability to draw justification for the latter from the space of reasons, and ultimately from the world. We have built the idea of a prescriptive around the assumption that entitlement to it arises in generically the same way as does entitlement to a declarative. It is entitled agent-neutrally in virtue of moves within the common linguistic practice—moves constrained by the agentneutral incompatibility and inferential significances of the practice. But the difference between a prescriptive and a declarative is that the former does not amount to a performance that is merely attempting to achieve entitlement to a position in inferential space, but rather it attempts to place substantive commitments on others. To successfully defend a claim that x ought to do y is to place a social commitment upon x. In formal terms, we distinguish the two contents as follows: d 〈D(cF , We), 〉 d 〈P(F, ), 〉 In the first of these, we use ‘We’ to indicate a quantification over everyone or perhaps a reference to the community as a whole. Given this, we record the score that declares that everyone scores as committed to doing F. The second involves prescribing that perform F. Entitlement to the former is an agent-neutral achievement in the space of reasons, a linguistic accomplishment, from which other declarative entitlements can be drawn inferentially. Entitlement to the second actually places a social onus upon . If we combine the pragmatic category of the prescriptive with the ability to make linguistic statuses explicit, we can throw away the “agentive” ladder of the original image. Doing so again involves borrowing from Brandom. Once declarative and prescriptive contents are on board, the ability to make substantive normative distinctions between agents and non-agents should involve “merely” inferential complexity. A practice complex enough, that is, to include both prescriptives and the semantic resources for making explicit its own structural elements will have the resources to claim that one is not entitled to assign commitments to clouds, or entitlement to the sun. One justifies the agent/object and act/ event distinctions within the practice which is a particular and particularly rich instance of a general framework—the original image—that fails to make such distinctions.

234

Appendix

We have left an enormous amount of work to do in pursuing logical semantics from the perspective of normative pragmatics. This work is exciting since we have given ourselves an apparently much richer set of resources with which to account for semantic phenomena than is available to an inferentialist hampered by the declarative fallacy. But all that is far too much to address here. For now, we content ourselves with having shown that nothing is lost with respect to standard accounts that begin with propositions—that propositional contents standing in inferential and incompatibility relations are constructible out of the resources of normative proprieties applying to acts in general.

Index

Acknowledgments, 145–152, 161–162, 182– 189; defined, 145; recognitive character of, 138; relationship to vocatives, 138–152. See also Vocatives Addresses/addressing, 124–128, 160–163, 171–174, 176, 177; relationship to vocatives, 162–163. See also Tellings; Vocatives; Voice, second-person Agents/agency: constitution of, 141–142, 181–190, 194; dependence on discursive communities, 190–195, 209; and interpellation, 142, 180–190; marginal, 192–194; in the ‘original image’, 223, 232–233 Agent-neutrality: defined, 16–18, 20, 25; of reasons, 16, 17, 122, 123, 125, 126; relationship to universality, 17, 25–29, 37–38, 228. See also Declaratives, agent-neutrality of Agent-relativity: defined, 16–18, 20; of reasons, 16, 122–128 Agent-specificity, 23–24, 44, 124, 135; vs. kind-relativity, 23–24 Alston, William, 4, 71 Altham, J. E. J., 113, 120–121. See also Besires Althusser, Louis, 10n17, 134n1, 142–143, 181–182, 183, 186. See also Interpellation Animals, 49, 193 Assertion. See Declaratives Austin, John L., 4, 10, 87, 88, 135, 154. See also Austinean performatives Austinean performatives, 15, 39, 43, 84–85, 87–94 Ayer, Alfred J., 113 Babies, 176, 186–189, 193 Bach, Kent, 5

Bakhtin, Mikhail, 153, 171 Belief-desire psychology, 98–99, 118–122. See also Besires; Noncognitivism in ethics; Objectivity, of values and moral judgments Beliefs. See Mental states/mental events Belnap, Nuel, 11–12, 231n7. See also Declarative fallacy Besires, 113, 120 Bohman, James, 141n6 Brandom, Robert, 1, 4, 5–6, 8–11, 9n13, 12, 49n4, 55n10, 60, 66, 74, 104n10, 154, 157, 159–160, 163, 166, 172–173, 217, 218, 219, 229, 230, 233. See also Game of giving and asking for reasons; Scorecards/ score keeping Butler, Judith, 10n17, 92n3, 188n9 Categorical imperatives, 128–133. See also Imperatives; Kant, Immanuel Cognitivism in ethics, 119, 120, 122, 129–130 Communication, 29–30, 156, 171–174. See also Speech acts Community. See Discursive community/communities Concepts, 73, 76–77, 189. See also Kant, Immanuel, discursive judgments Content, 11–12, 52–58, 60–61, 65n18, 234; nonconceptual, 69–71. See also Mental states/mental events; Semantics Darwall, Stephen, 108, 122–128, 136, 146, 162, 168–169, 177, 194–195 Daston, Lorraine, 81 Davidson, Donald, 4, 5, 10, 51–53, 66, 68– 72, 74–75, 76, 78, 151n15, 172–173, 195n13, 198n17, 219. See also Observation reports/occasion sentences

235

236

Index

Declarative fallacy, 11–12, 40, 42, 49, 53–59, 61, 73, 97, 177, 213, 217, 231n7, 234 Declaratives, 7, 9–12, 14, 15, 17–18, 20–22, 25–29, 33–38, 40, 42, 59–60, 69, 72, 73, 76, 78, 81, 83–84, 89, 93n4, 97, 112, 113, 132, 133, 161, 165–167, 169–170, 177, 201, 228–232, 233; agent-neutrality of, 17–18, 20–22, 25–29, 37–38, 201; and anaphora, 38, 54–55; formal characterization of, 228–230; impersonality of, 17–18, 26–27, 33, 165–167; reperformance of, reassertion licenses, 15, 20, 26, 29, 37–38. See also Declarative fallacy; Truth claims Deixis. See Ostension Derrida, Jacques, 10n17, 99 Discursive community/communities, 12, 13, 23, 37–38, 64, 79, 145, 154, 160, 191– 205; induction into, 176; membership in, 190–195, 211; number of, 23, 195–205 Dummett, Michael, 25n31 Embodiment/embodied practices, 3–4, 5, 8– 9, 77–78, 156, 179 Emotivism, 113–114, 116, 119 Empirical world, the, 36–37, 77–78; accountability to, 36–37, 62–64, 68–78, 80, 81, 180, 197, 211. See also McDowell, John Entitlement, types of, 220–221 Experience. See Empirical world, the; Observation(s); Receptivity/receptive capacities First-personality, structural, 45, 50, 65, 101– 105, 117–118, 121–122, 124, 132, 153– 154, 157–160, 163, 177. See also Voice, first-person Foot, Philippa, 98n4, 119, 129 Fulton, Colleen, 131n42, 147n13 Galison, Peter, 81 Game of giving and asking for reasons, 5–6, 8, 191. See also Brandom, Robert; Sellars, Wilfrid; Space of reasons Gettier, Edmund, 151n16 Given, the, 62, 74; myth of the given, 58, 64–65, 68. See also Sellars, Wilfrid Grammar/grammatical form, 7, 13–14; grammatical mood, 7, 13, 142; surface

grammar, 13, 20, 31–32, 46, 107, 115, 142, 160, 182 Grice, Paul, 4, 7n10, 10, 174 Hacking, Ian, 197 Hails. See Interpellation; Vocatives Hare, R. M., 11, 113–118, 199 Haugeland, John, 4, 49n4, 211 Hegel, Georg Wilhelm Friedrich, 6, 142, 152, 192n12, 212. See also Recognition, mutual recognition Heidegger, Martin, 10, 17, 50, 99, 134–135, 143, 211–212 Hinchman, Edward, 166, 169 Holdings, 107–113, 122–123, 126–128, 130, 164–165, 168; vs. prescriptives, 107–110, 112, 123. See also Imperatives; Prescriptives Ignorance, 17, 18, 25, 26, 29, 33, 34–37; and epistemic responsibility, 34–37 Ignoring. See Shunning Imperatives, 10, 11, 13, 14–15, 17–20, 22, 26–27, 29, 31, 61–62, 83, 97, 107, 111– 118, 122, 129, 150–151, 161, 162, 163, 231; acknowledgment of, 150–151; alethic imperatives/holdings, 111–113, 115, 122, 123, 127, 130, 163, 167, 184; constative imperatives/holdings, 112–113, 126–127, 139, 161, 167; formal characterization of, 231; vs. prescriptives, 114–118, 122, 129, 161. See also Categorical imperatives; Holdings; Prescriptives Indexicals: essential indexical, 102, 157–159. See also Perry, John Inference, 8–9, 12, 15, 26, 29, 35–36, 48, 58–59, 62–63, 68, 69, 75, 76, 79, 105, 207–208, 230, 234; broadened conception of, 58–59, 116–117; practical, 58–59, 158. See also Justification Inferentialism, 9n13, 55n10, 217–218, 234 Input and output: defined, 14–16 Intentions, 7, 13–14, 173–174 Interpellation, 134n1, 181–190, 193. See also Agents/agency; Vocatives; ‘Yo!’ Intersubjectivity, 80–81, 206–207. See also Objectivity

Index Justification, 63–65; vs. entitlement, 28n32. See also Inference Kant, Immanuel, 43–45, 47, 49–50, 70n7, 73, 75, 128–133, 209, 211; concepts, 73, 75; discursive judgments, 14, 73; intuition, 70n7, 75; judgments of taste, 43–45, 47, 49–50; schemata, 70n7; singular judgments, 45; transcendental synthesis of apperception, 49; universal validity, 43–44. See also Categorical imperatives Kenaan, Hagi, 136, 168, 170 Lacan, Jacques, 190n11 Law/legal norms, 23–24, 84, 87n1, 110, 204– 205 Little, Margaret, 32n35, 109–110 ‘Lo!’, 40–41, 47–48, 80–81, 106, 191, 205– 206, 209, 210–212; essential to language, 205–206. See also Observatives; Ostension; ‘Yo!’ Luntley, Michael, 70 Mackie, John L., 98–100, 120 Manning, Richard, 34n36, 68 McDowell, John, 1–2, 4, 10, 49n4, 57n13, 58n14, 62, 66, 69, 71–76, 84–86, 119– 120, 156, 172; on objectivity, 84–86; proto-doxastic judgments in, 71–73; on receptivity to the world, perception, 71– 76, 85–86. See also Empirical world, the; Space of reasons Mental states/mental events, 6–7, 66–75, 79n22, 118–122. See also Content; Intentions Mood. See Grammar/grammatical form Mooning, 173 Moran, Richard, 165–169 Nelson, Lynn Hankinson, 21 Noncognitivism in ethics, 113–114, 116– 117, 120, 199 Objectivity, 80–86, 197, 206; of values and moral judgments, 81–82, 98–100, 118– 122. See also Mackie, John L.; Truth Observation reports/occasion sentences, 46,

237

51–53, 64, 69, 76. See also Observation(s); Observatives Observation(s), 48–49, 57–59, 60–64, 66– 79, 207–208; accountable to the world, 62, 63–64, 68–78, 197, 211; as conceptually articulated, 57–59, 63, 67, 68, 73–74, 207; propositional content and, 55, 57–58, 74; relation to action, 77–78. See also Observation reports/occasion sentences; Observatives Observatives, 45–51, 54–65, 67–68, 73–74, 76–78, 80, 81, 83–84, 89, 93n4, 106–107, 112, 137, 157, 201–202; vs. declaratives, 45–47, 48, 55, 56, 74; defined, 46; essential role of, 59–65; propositional vs. nonpropositional, 55, 57; relationship to observation, 67–68, 73–74. See also Declaratives; First-personality, structural; ‘Lo!’; Observation reports/occasion sentences; Observation(s) Orders. See Imperatives Ostension, 47–48, 80–81, 106–107, 137, 163, 168, 210–212. See also ‘Lo!’ Perception/perceptual episodes. See Observation(s) Performatives. See Austinean performatives Perry, John, 102, 157. See also Indexicals Persons/personhood. See Agents/agency Pragmatics, formal, 4, 5–7, 218–234; of normative statuses, 223–234 Pragmatism: classical American pragmatism, 3–4; Pittsburgh School pragmatism, 4, 5– 6, 8–10, 12; priority of the pragmatic, 2, 5–9, 82, 118, 156, 172, 217; varieties of, 3–12, 217 Prayer, 164–165n3 Prescriptives, 39, 95–124, 127, 128–133, 161, 223, 231–233; first-personal, 101– 103, 105–106, 117–118, 120–122; formal characterization of, 231–233; secondpersonal vs. third-personal, 106–111, 113; as truth claims, 95–96, 97–101, 107–108, 120, 131. See also Categorical imperatives; Holdings; Imperatives Prescriptivism, 96–97, 113–118, 119, 129– 130. See also Hare, R. M.

238

Index

Promises, 24, 90, 90n2, 91, 163, 169. See also Austinean performatives; Holdings Quine, Willard Van Orman, 53n8, 172– 173n13. See also ‘Lo!’; Observation reports/occasion sentences; Observatives Reactive attitudes, 110, 142 Receptivity/receptive capacities, 36, 45–49, 50, 62, 64, 65, 70–76, 78, 80, 85–86, 210– 211. See also Empirical world, the Recognition, 75–78, 79, 101–102, 121–122, 137–139, 163–164, 180, 182–190; constitutive misrecognition, 144–145, 183, 187, 190n11; mutual recognition, 128, 142, 147, 149, 152, 162, 165, 177, 183–190, 192n12, 193, 212. See also Acknowledgments; Interpellation; Vocatives; ‘Yo!’ Recognitives, 45–47, 53, 57, 64–65, 75, 80, 137–139, 143–145, 148, 149, 154, 162, 180. See also Acknowledgments; Observatives; Vocatives Representationalism and antirepresentationalism, 4, 5, 6–7 Rorty, Richard, 4 Rouse, Joseph, 5n9, 36n38, 60n15, 60n16, 211n28 Rousseau, Jean-Jacques, 141n7 Sanctions, 219 Scorecards/score keeping, 8, 9n13, 154, 156, 157, 165, 172–173n13, 176, 184, 193, 217–222, 224–225, 232–233; formal constraints on, 224–227, 232. See also Game of giving and asking for reasons Searle, John, 4, 7, 12 Second-personality, structural, 90n2, 135– 136, 122–128, 153–155, 159–177. See also Vocatives; Voice, second-person Sellars, Wilfrid, 1–2, 4, 5–6, 7–9, 10, 49n4, 57n13, 64–65, 66, 68, 74–78, 156, 172, 179–180, 189, 201, 207–208, 211, 222. See also Game of giving and asking for reasons; Given, the; Space of reasons Semantics, 4–5, 6–12, 13, 30–31, 52–57, 76, 218–219, 229, 230–234 Shunning, 147, 194, 203 Smith, Michael, 113, 118–122 Space of reasons, 1–2, 8–10, 12, 35, 66, 68,

69, 70, 75, 128–129, 132, 156–160, 189, 191, 196, 222, 232–233; initiation into, 189 (see also Babies); location within, 156–160, 189, 191; pragmatic structure/ topography of, 2, 8–10, 12, 66. See also Brandom, Robert; Game of giving and asking for reasons; McDowell, John; Sellars, Wilfrid Speech acts: defective, 16, 28, 33; formal characterization of, 228–233; functional structure of, 13–16, 18, 21–22, 29–34; nature of, 2–3, 12–13; private/secret, 33–34, 174–175; speech act theory, 4; typology of, 3, 18–22, 38–41, 52, 95–96, 102–103, 133, 150, 176–177. See also Communication; Speech acts, types of; Voice Speech acts, types of. See Acknowledgments; Addresses/addressing; Austinean performatives; Declaratives; Imperatives; Observatives; Performatives; Promises; Recognitives; Tellings; Vocatives Stevenson, Charles L., 113–114 Syntax. See Grammar/grammatical form Tellings, 165–169 Testifying, 145–146 Third man problem, 70–71. See also Content, nonconceptual; Kant, Immanuel, schemata Truth: agent-neutrality of, 21, 26, 28–29, 37, 38, 200–201; prosentential theory of, 55n10, 55n11; publicity of, 28–29, 34, 101, 197, 211; relativism about, 21, 186– 187. See also Objectivity Truth claims, 20, 21, 26, 28, 29, 37, 54–55, 57, 59, 83, 97–100. See also Declaratives; Observatives; Truth Vocatives, 134–136, 138–152, 153–155, 159, 162–163, 165, 170, 174–177, 181–190, 191, 212; as burdensome, 141, 185; constitutive role in personhood, 141–142, 176, 180–190; differential entitlement to, 140–141; quasivocatives, 144–145, 175–176, 187–188, 190; rejection of, 147, 185–186. See also Acknowledgments; Addresses/addressing; Interpellation; Second-personality, structural; Voice, second-person; ‘Yo!’ Vocatives, transcendental. See Second-personality, structural

Index Voice, 59–60, 61, 116–118, 120–121, 122; directionality of speech, 18, 59, 80–81, 125–126, 141, 153–154, 156, 158–160, 161, 163, 166–171, 172, 177, 182; firstperson, 50, 59–65, 84, 135, 161, 192, 211; impersonal, 31, 59–61, 116, 120, 154– 155, 165–166, 170, 213; pragmatic, not grammatical (vs. mood), 46, 61, 125–126, 142, 182; second-person, 116, 124–126, 135, 141, 143, 153–155, 159–174, 189, 190, 192, 210, 212; third-person, 61–62, 64, 105–107, 111, 113, 126, 143–144, 152, 154–155, 160, 161. See also Speech acts

239

Watson, Gary, 90n2 Wittgenstein, Ludwig, 4, 6, 25n30, 31, 135, 189, 211 ‘Yo!’, 40–41, 134–152, 191, 205–212; equiprimordiality with ‘Lo!’, 184, 205– 206, 209, 210–212. See also Interpellation; ‘Lo!’; Vocatives; Voice, second-person ÑiÒek, Slavoj, 183n5