The Routledge Handbook of Social and Political Philosophy of Language 9781138602434, 9780367759575, 9781003164869

292 97 7MB

English Pages [425] Year 2021

Report DMCA / Copyright


Polecaj historie

The Routledge Handbook of Social and Political Philosophy of Language
 9781138602434, 9780367759575, 9781003164869

Table of contents :
Half Title
Series Page
Title Page
Copyright Page
Table of Contents
Notes on Contributors
PART I: Social and Political Language: Methodological and Foundational Issues
1 Conceptual Engineering in Philosophy
2 Social Ontology
3 An Invitation to Social and Political Metasemantics
4 Linguistic Prescriptivism
5 Speech-Act Theory: Social and Political Applications
6 On the Uselessness of the Distinction Between Ideal and Non-Ideal Theory (At Least in the Philosophy of Language)
PART II: Non-Ideal Semantics and Pragmatics
7 Lying, Deception, and Epistemic Advantage
8 Propaganda
9 Code Words
10 Racist and Sexist Figleaves
11 Protest and Speech Act Theory
12 Defective Contexts
PART III: Linguistic Harms
13 Varieties of Pejoratives
14 Microaggressions and the Problem of Attributional Ambiguity
15 Hermeneutical Injustice
16 Social and Political Aspects of Generic Language and Speech
17 Language Extinction
18 “Laxwalxwash Potamáay Súngaan ‘Áawq // To Be Between the Blind Snake’s Teeth”: Indigenous Language Reclamation Between the Fangs of A (Simulated) Dilemma
PART IV: Applications
19 Language and Free Speech
20 Language and Ideology
21 Language and Legitimation
22 How Much Gender is Too Much Gender?
23 On Language and Sexuality: Demisexuals, Polyamorous, Bambi Lesbians, and Other Queers
24 The Language of Mental Illness

Citation preview


This Handbook brings together philosophical work on how language shapes, and is shaped by, social and political factors. Its 24 chapters were written exclusively for this volume by an international team of leading researchers, and together they provide a broad expert introduction to the major issues currently under discussion in this area. The volume is divided into four parts: Part I: Methodological and Foundational Issues Part II: Non-ideal Semantics and Pragmatics Part III: Linguistic Harms Part IV: Applications The parts, and chapters in each part, are introduced in the volume’s General Introduction. A list of Works Cited concludes each chapter, pointing readers to further areas of study. The Handbook is the first major, multi-authored reference work in this growing area and essential reading for anyone interested in the nature of language and its relationship to social and political reality. Justin Khoo is Associate Professor of Philosophy at MIT. He works primarily at the intersection of philosophy of language and linguistic semantics, and has research interests in metaphysics and meta-ethics. Rachel Sterken is Associate Professor of Philosophy at the University of Hong Kong. She works primarily at the intersection of philosophy of language, semantics, ethics, and social philosophy. Most of her research focuses on the semantics of generic language and issues related to conceptual engineering.


Routledge Handbooks in Philosophy are state-of-the-art surveys of emerging, newly refreshed, and important fields in philosophy, providing accessible yet thorough assessments of key problems, themes, thinkers, and recent developments in research. All chapters for each volume are specially commissioned, and written by leading scholars in the field. Carefully edited and organized, Routledge Handbooks in Philosophy provide indispensable reference tools for students and researchers seeking a comprehensive overview of new and exciting topics in philosophy. They are also valuable teaching resources as accompaniments to textbooks, anthologies, and research-orientated publications. Also available: THE ROUTLEDGE HANDBOOK OF ANARCHY AND ANARCHIST THOUGHT Edited by Gary Chartier and Chad Van Schoelandt THE ROUTLEDGE HANDBOOK OF THE PHILOSOPHY OF ENGINEERING Edited by Diane P. Michelfelder and Neelke Doorn THE ROUTLEDGE HANDBOOK OF MODALITY Edited by Otávio Bueno and Scott A. Shalkowski THE ROUTLEDGE HANDBOOK OF PRACTICAL REASON Edited by Kurt Sylvan and Ruth Chang THE ROUTLEDGE HANDBOOK OF PHILOSOPHY OF EUROPE Edited by Darian Meacham and Nicolas de Warren THE ROUTLEDGE HANDBOOK OF SOCIAL AND POLITICAL PHILOSOPHY OF LANGUAGE Edited by Justin Khoo and Rachel Sterken THE ROUTLEDGE HANDBOOK OF POLITICAL EPISTEMOLOGY Edited by Michael Hannon and Jeroen de Ridder For more information about this series, please visit:


Edited by Justin Khoo and Rachel Sterken

First published 2021 by Routledge 52 Vanderbilt Avenue, New York, NY 10017 and by Routledge 2 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN Routledge is an imprint of the Taylor & Francis Group, an informa business © 2021 Taylor & Francis The right of Justin Khoo and Rachel Sterken to be identified as the authors of the editorial material, and of the authors for their individual chapters, has been asserted in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. Library of Congress Cataloging-in-Publication Data A catalog record for this title has been requested ISBN: 978-1-138-60243-4 (hbk) ISBN: 978-0-367-75957-5 (pbk) ISBN: 978-1-003-16486-9 (ebk) Typeset in Bembo by codeMantra


Notes on Contributors viii Introduction 1 PART I

Social and Political Language: Methodological and Foundational Issues13 1 Conceptual Engineering in Philosophy 15 Matti Eklund 2 Social Ontology 31 Mari Mikkola 3 An Invitation to Social and Political Metasemantics 42 Derek Ball 4 Linguistic Prescriptivism 56 Alex Barber and Robert J. Stainton 5 Speech-Act Theory: Social and Political Applications Daniel W. Harris and Rachel McKinney


6 On the Uselessness of the Distinction Between Ideal and Non-Ideal Theory (At Least in the Philosophy of Language) 91 Herman Cappelen and Josh Dever


Contents PART II

Non-Ideal Semantics and Pragmatics107 7 Lying, Deception, and Epistemic Advantage 109 Eliot Michaelson and Andreas Stokke 8 Propaganda 125 Anne Quaranto and Jason Stanley 9 Code Words 147 Justin Khoo 10 Racist and Sexist Figleaves 161 Jennifer Saul 11 Protest and Speech Act Theory 179 Matthew Chrisman and Graham Hubbs 12 Defective Contexts 193 Andrew Peet PART III

Linguistic Harms209 13 Varieties of Pejoratives 211 Robin Jeshion 14 Microaggressions and the Problem of Attributional Ambiguity 232 Christina Friedlaender 15 Hermeneutical Injustice 247 Rebecca Mason 16 Social and Political Aspects of Generic Language and Speech 259 Matthew McKeever and Rachel Sterken 17 Language Extinction 281 Ethan Nowak 18 “Laxwalxwash Potamáay Súngaan ‘Áawq // To Be Between the Blind Snake’s Teeth”: Indigenous Language Reclamation Between the Fangs of A (Simulated) Dilemma 298 Shelbi Nahwilet Meissner


Contents PART IV



19 Language and Free Speech Ishani Maitra and Mary Kate McGowan


20 Language and Ideology Eric Swanson


21 Language and Legitimation Robert Simpson


22 How Much Gender is Too Much Gender? Robin Dembrof and Daniel Wodak


23 On Language and Sexuality: Demisexuals, Polyamorous, Bambi Lesbians, and Other Queers Saray Ayala-López and E. Díaz-León


24 The Language of Mental Illness Renée Jorgensen Bolinger






Saray Ayala-López  is Assistant Professor of Philosophy at California State University, Sacramento, and is interested in explanations, and the absence of them, sex and gender in science, conversational dynamics and the many things we can do with words, cognitive externalism, and several questions within social ontology and epistemology. Derek Ball is Senior Lecturer in Philosophy at the University of St Andrews. He has written on a wide range of topics in philosophy of mind, philosophy of language, and psychology. Alex Barber is Senior Lecturer of Philosophy at the Open University. Renée Jorgensen Bolinger  is Assistant Professor in the Department of Politics and University Center for Human Values and an associated faulty member in the Department of Philosophy at Princeton University. She specializes in social and political philosophy and philosophy of language. Her research focuses on the ways that agents’ moral and civil rights ground moral constraints on informal social norms and communicative conventions. Herman Cappelen  is Chair Professor at the University of Hong Kong. His two most recent books are: Fixing Language: An Essay on Conceptual Engineering (Oxford UP, 2018) and Making AI Intelligible: Philosophical Foundations (with Josh Dever, Oxford UP, 2021). Matthew Chrisman  is Professor of Ethics and Epistemology in the Philosophy Department at the University of Edinburgh. He is the author of The Meaning of ‘Ought’ (Oxford UP, 2015) and What Is This Thing Called Metaethics? (Routledge, 2016). His research and teaching focuses on metaethics, social epistemology, philosophy of language, and political philosophy. Robin Dembrof is Assistant Professor of philosophy at Yale University. They specialize in metaphysics and feminist philosophy, with an emphasis on social systems and human classifcations. Josh Dever is Professor of Philosophy at the University of Texas at Austin, who works primarily in the philosophy of language and philosophical logic, with interests in the application of these felds to problems throughout core areas of philosophy.



E. Díaz-León  is Associate Professor of Philosophy at the University of Barcelona. She specializes in philosophy of mind and language, and philosophy of gender, race, and sexuality, and her current research focuses on social construction, conceptual ethics, and the metaphysics of gender, race, and sexual orientation. Matti Eklund  is Chair Professor of Theoretical Philosophy at Uppsala University. He works mainly on topics in metaphysics, philosophy of language, philosophy of logic, and metaethics. Christina Friedlaender  is Instructor of Philosophy at Seattle University. Her research focuses on the social ontology of shared agency and anti-oppression resistance, attending in particular to the intersection of methodology, value pluralism, and social knowledge. Daniel W. Harris  is Assistant Professor of Philosophy at Hunter College, CUNY. His work is mostly about semantics, speech acts, and the psychological capacities that make humans such efcient communicators. Graham Hubbs is Associate Professor in the Department of Politics and Philosophy, University of Idaho. He uses social ontological methods to study topics that lie at the intersection of philosophy, politics, and economics. Robin Jeshion  is Professor of Philosophy at University of Southern California, who specializes in the philosophy of language, especially in its intersections with philosophy of mind and social-political philosophy. She is also keenly interested in animal cognition, aesthetics, feminism, and poetry. Justin Khoo is Associate Professor of Philosophy at MIT. He works on philosophy of language, metaphysics, and metaethics, and has written about conditionals, modality, and political speech. Ishani Maitra is Associate Professor of Philosophy at the University of Michigan. Her areas of research specialization include philosophy of language, feminist philosophy, and philosophy of law. Rebecca Mason is Assistant Professor of Philosophy at the University of San Francisco. She specializes in metaphysics (especially social metaphysics), feminist philosophy, and social and political philosophy. She also has interests in epistemology and philosophy of language (especially where these intersect with her areas of specialization). Mary Kate McGowan is the Margaret Clapp ‘30 Distinguished Alumna Professor of Philosophy at Wellesley College. She works in metaphysics, philosophy of language, philosophy of law, and feminism. She is the author of Just Words: On Speech and Hidden Harm (Oxford UP, 2019) and coeditor, with Ishani Maitra, of Speech and Harm: Controversies Over Free Speech (Oxford UP, 2009). Matthew McKeever is Executive Associate Editor at Inquiry and Research Assistant at ConceptLab at the University of Oslo. Rachel McKinney is an Assistant Professor of Philosophy and the Program Director of Politics, Philosophy, and Economics at Sufolk University in Boston. She writes on philosophy of language, social/political philosophy, and feminism.



Shelbi Nahwilet Meissner (Luiseño and Cupeño)  is Assistant Professor of Philosophy at Georgetown University. She researches and teaches American Indian and Indigenous philosophy. Eliot Michaelson is Senior Lecturer at King’s College London. His work on the philosophy of language has focused primarily on non-ideal approaches to understanding notions like meaning and communication. Together with Jessica Pepp and Rachel Sterken, he is also engaged in a longterm inquiry into what consideration of online contexts and artificial agents can tell us about the nature of language use. He typically prefers to be reading science fiction novels. Mari Mikkola is the Chair of Metaphysics at the University of Amsterdam (The Netherlands). She is the author of two books: The Wrong of Injustice: Dehumanization and Its Role in Feminist Philosophy (Oxford UP, 2016) and Pornography: A Philosophical Introduction (Oxford UP, 2019) and of several articles on feminist philosophy, social ontology, and pornography. Her current work is focused on philosophical methodology with a monograph on this topic under contract with Oxford UP. Ethan Nowak is Associate Professor in Philosophy at Umeå University. Andrew Peet is Postdoctoral Research Fellow in the Department of Philosophy at the University of Leeds. He specializes in epistemology and the philosophy of language. His recent research has focused on, among other things, miscommunication, testimony, and the theory of knowledge. Anne Quaranto is a graduate student in the Department of Philosophy at the University of Texas at Austin, specializing in social and political issues in the philosophy of language. Her research on dog whistles and coded speech focuses on how communicated content is shaped by temporally extended and sociopolitically structured linguistic practices. Jennifer Saul is Waterloo Chair in Social and Political Philosophy of Language at the University of Waterloo. She works on racism and sexism in political speech, and the world has sadly been keeping her very busy. Right now (October 2020) her main hobby is helping US citizens overseas to vote. She also enjoys cooking and bad movies. Robert Simpson  is Associate Professor of Philosophy at University College London. His research interests mainly lie in social and political philosophy, with a particular focus on arguments around free speech. Robert J. Stainton is Distinguished University Professor in the Department of Philosophy and Graduate Program in Linguistics at the University of Western Ontario. Generally speaking, his research draws on empirical linguistics to address issues in philosophy and vice versa. Of late, he has been writing on conversational impairments, history of philosophy of language, and the syntax/semantics/pragmatics interfaces. Jason Stanley is the Jacob Urowsky Professor of Philosophy at Yale University. He is the author of five books, most recently How Fascism Works: The Politics of Us and Them (Random House, 2018). His sixth book, The Politics of Language: An Essay in Non-Ideal Theory of Meaning, with David Beaver, is forthcoming with Princeton UP. Rachel Sterken  is Associate Professor of philosophy at the University of Hong Kong. She works primarily at the intersection of philosophy of language, semantics, ethics, and social ­philosophy. Most of her research focuses on the semantics of generic language and issues related to x


conceptual engineering. Her more recent work explores the semantics, pragmatics, and metasemantics of online communication. Andreas Stokke is a docent and senior lecturer in philosophy at the Department of Philosophy, Uppsala University, and a Pro Futura Scientia Fellow at the Swedish Collegium for Advanced Study. Previously, he has been a visiting scholar at the Department of Philosophy, New York University, an academic visitor of the Faculty of Philosophy, University of Oxford, and a visiting fellow at Magdalen College, Oxford. Stokke’s research is mainly in the fields of philosophy of language and epistemology, but he has also worked on ethics and the philosophy of action. Eric Swanson is Professor of Philosophy and Linguistics at the University of Michigan. Most of his current work is in social and political philosophy of language and metaethics, and pays special attention to relationships between language, context, and ideology. He is especially interested in the ways in which the force of language goes beyond the information that is apparent to or shared amongst discourse participants. Daniel Wodak is Assistant Professor of Philosophy and of Law at the University of Pennsylvania. They mostly work on ethics, social and political philosophy, and philosophy of law.


INTRODUCTION Justin Khoo and Rachel Sterken

This Handbook brings together philosophical work on how language shapes, and is shaped by, social and political factors. The volume’s 24 chapters are divided into four parts:

Part I:  Social and Political Language: Methodological and Foundational Issues Some tools, methodologies, and concepts in philosophy of language, feminist philosophy, and critical theory are especially useful in intellectual projects that aim to understand the social and political dimensions of language, meaning, and communication of interest to philosophers. From conceptual engineering, to speech act theory, to metasemantics and non-ideal theory, the chapters from Part I consider methodological and foundational issues raised by the study of social and political philosophy of language. One of the central and most important distinctions in this literature is between descriptive and ameliorative projects. Roughly, a descriptive project aims to offer an analysis of our existing concepts, articulating their application-conditions. By contrast, an ameliorative project aims to assess and potentially revise our concepts in light of the uses to which we aim to put them. The distinction goes back at least to Nietzsche (Nietzsche 1901/68), but arises in analytic philosophy in Frege, Wittgenstein, Carnap, Quine, Brandom, and, more recently, in the work of Sally Haslanger and many others (see Burgess et al. 2020 for a recent overview). Of particular interest to the social/ political philosopher are ameliorative projects focused on socially significant concepts like race, gender, justice, freedom, marriage, and others. Such radical projects raise a number of pressing foundational issues. In his chapter, “Conceptual engineering in philosophy” (Chapter 1) Matti Eklund explores a range of questions regarding the nature of amelioration: what are these concepts which can be defective or improved upon? What purposes do concepts serve? How, if at all, do we replace one concept with another? Eklund surveys answers to these questions while discussing a range of case studies from the literature and presents some general worries about the enterprise. The possibility of conceptual engineering blurs the line between language and the world it is used to describe. In her chapter, “Social ontology” (Chapter 2), Mari Mikkola discusses recent work in feminist philosophy exploring how and to what extent language can serve to create parts of social reality. As an example, she discusses Judith Butler’s (1999) performative theory of gender, according to which an announcement by a doctor “It’s a boy[girl]” in the presence of a newborn is 1

Justin Khoo and Rachel Sterken

not a descriptive claim but one with “illocutionary, performative force”: it constitutes, rather than records, a social fact about the baby’s gender. Realizing that words have this existence-making power, it becomes important that our words are good ones, and Mikkola presents the influential work of Sally Haslanger on ameliorative analyses, the aim of which is to improve our words and concepts, the better to shape our social world. Intimately related to the issue of how language and the world are related are questions in metasemantics about what makes it the case that words mean what they mean. The meaning of “marriage,” today, is such that it includes couples of the same sex. That is clearly a socially relevant fact. But is there anything socially or morally important about how it is that “marriage” has come to mean what it means today? Derek Ball’s chapter, “An invitation to social and political metasemantics” (Chapter 3), presents a framework for approaching these questions and suggests some possible answers. By precisely distinguishing semantics from metasemantics, he shows that it might indeed be the case that metasemantics is socially and morally important. To use a famous example from Miranda Fricker (2007), there was, until recently, no word that meant sexual ­harassment—nothing made it the case that sexual harassment was denoted by some English phrase. Ball argues that this is a metasemantic fact of moral importance, and by presenting this familiar case through a metasemantic lens he gives us the resources to spot further socially and politically relevant aspects of metasemantics. Ameliorative projects in social and political philosophy of language do seem to bear several similarities to the more mainstream view known as linguistic prescriptivism. Both involve commitment to normative recommendations about how to talk, for instance. In their chapter, “Linguistic prescriptivism” (Chapter 4), Alex Barber and Robert Stainton grapple with the issue of whether there really is a plausible and non-trivial notion of linguistic prescriptivism that might characterize the attitudes of people commonly associated with the view. They argue that we should drop the notion in favor of “language norming” (which is intended to include paradigm instances of ameliorative projects as well as injunctions not to split infinitives) and then argue that some instances of language norming are defensible and others are not. Many of the unacceptable instances are those highlighted in criticism of so-called linguistic prescriptivists; however, Barber and Stainton argue that many other instances of language norming may be acceptable. Another foundational topic of central significance to social and political philosophy of language is the nature of speech acts. In their chapter on the topic, “Speech-act theory: social and political applications” (Chapter 5), Daniel Harris and Rachel McKinney articulate the theory of speech acts found in Austin (1962) and Searle (1969), and then discuss a range of cases where speech act theory can be usefully applied. Promising, for example, is a speech act that some early modern theorists held to be part of what grounds social contracts. Similarly, in making laws or, more informally, consenting to sexual acts or medical procedures, one performs illocutionary acts with success and failure conditions. In helping us understand the nature and wide application of speech act theory, their chapter sheds light on many socially and politically relevant features of our lives, showing us how language can create parts of the world or—on the other hand—deprive people of the capacity to speak. Having surveyed some of the foundational issues arising from the study of social and political language, we might then wonder whether the topic genuinely constitutes a distinct subfield of philosophy of language. In their chapter, “On the uselessness of the distinction between ideal and non-ideal theory (at least in the philosophy of language)” (Chapter 6), Herman Cappelen and Josh Dever consider whether we can distinguish between ideal and non-ideal philosophy of language in the way that a philosopher like Charles Mills does for political philosophy (Mills 2005). Rawls (1971), Mills argues, provides a theory of justice inappropriate for the deeply antecedently unfair and unequal world in which we find ourselves; similarly, we might think that ideal philosophy of language, with its focus on topics like reference, anaphora, implicature, and intensionality, should 2


be distinguished from the more “applied” philosophy of language work discussed in this volume. Cappelen and Dever answer in the negative, holding that there is no deep distinction between the sort of philosophy collected in this volume and the more mainstream material, and trying to draw such a distinction risks obscuring the deep connections between, to take but one example, speech act theory and recent feminist analyses of pornography.

Part II:  Non-ideal Semantics and Pragmatics It’s plausible that successful linguistic communication can occur between people sharing certain assumptions: perhaps some shared understanding of the language being used to communicate (Lewis 1975), or a shared understanding that each is intending to get the other to believe things by means of recognizing their intentions (Grice 1989), or a shared understanding of some body of information or what’s at issue or salient (Stalnaker 1978), or a shared understanding of the rule-­ governed activity they are engaged in (Searle 1969, Lewis 1979, Brandom 1994, Williamson 1996). It’s also plausible that agents can exploit such assumptions to promote selfish ends. This is perhaps most obvious in the case of lying, where an agent exploits the trust of her hearer to gain some advantage over them. But what exactly is a lie? Not every case of misleading is a case of lying. Alice asks Beth if she has seen her toy ball, and Beth, who accidentally let the dog eat it, remains quiet. Alice then infers that Beth doesn’t know, and is thus mislead by Beth’s silence (Alice perhaps reasons as follows: if Beth knew, she would say so, since that is the cooperative thing to do, given that our aim in communicating is to maximize our shared information; therefore, since she didn’t say anything, she must not know). Nonetheless, Beth doesn’t thereby lie to Alice—after all, Beth didn’t say anything at all! Reflection on cases like these suggests what Eliot Michaelson and Andreas Stokke call the traditional view of lying (which they trace back to St. Augustine): to lie is to say something you believe to be false with the intention to deceive your audience. This view correctly predicts that Beth didn’t lie to Alice, but if Beth had instead said, “I don’t know where the ball is” she would have lied. In their chapter, “Lying, deception, and epistemic advantage” (Chapter 7), Eliot Michaelson and Andreas Stokke consider whether lying essentially involves an intention to deceive. Famously, cases of bald-faced lies raise trouble for the view that lying always involves an intention to deceive (Carson 2006). In some situations where it is commonly known that P, someone confessing that ~P may carry additional consequences. In such a case, that person saying ~P still seems to be a lie, even though there is no intention to deceive, since they know their audience knows that P is true. Michaelson and Stokke argue that the usual strategies of response to such challenges to the traditional view of lying fail. In its place, they suggest that lies are an attempt to gain some advantage over one’s audience—sometimes an epistemic advantage (in the case of deception) and sometimes an emotional or political or even a nihilistic advantage (an attempt to undermine credibility and trust in certain contexts). Political propaganda provides further cases of speech-like behavior that aims to manipulate its audience by circumventing rational engagement. In their chapter, “Propaganda” (Chapter 8), Anne Quaranto and Jason Stanley highlight different forms propaganda might take, from framing devices (“The war on drugs”), to covert messaging (“There are Muslims among us”), to emotional charged slogans (“Make America Great Again”) and myths (“The American dream”), to seemingly objective bureaucratic reports (“Crime has risen 4.2%”). In each case, Quaranto and Stanley note that the aim is to circumvent the rational evaluation of a thought in order to legitimize or delegitimize it or to reify distinctions between in- and outgroup individuals. Often, this works by some kind of emotional appeal achieved by way of repetition and associations. Here, we have come a long way from standard possible worlds semantics. 3

Justin Khoo and Rachel Sterken

And, yet, we might wonder whether the resources from analytic philosophy of language may be useful in theorizing about the mechanisms by which propaganda, and political speech more generally, functions. Following Stanley (2015), Quaranto and Stanley suggest that at least some propaganda may work by way of certain words or expressions acquiring new not-at-issue linguistic meanings, which, when used, covertly communicate certain messages. The idea is similar to using loaded questions to humiliate someone, as in: “Have you stopped cheating on your spouse?” Because this question presupposes that the addressee has been cheating, both a “yes” and a “no” answer will commit them to this claim—hence, the humiliation. Quaranto and Stanley suggest that some propaganda, in particular, “Inner city crime is rampant,” may work like this. Here, the phrase “inner city” conveys the not-at-issue content that the crime under discussion is perpetrated by African Americans. Pushing back against this idea, in his chapter, “Code words” (Chapter 9), Justin Khoo argues that such a theory is both unmotivated and unnecessary. Instead, he argues that a very simple theory of code words may be enough to account for why it is that politicians often speak in code, and for the empirically verified effects of coded speech. According to Khoo’s simple theory, “Inner city crime is rampant” just means that “Crime occurring in impoverished urban areas is rampant.” Yet, hearers often have associations with or beliefs about impoverished urban areas—in particular, that they are often populated primarily by African Americans. Thus, they draw the inference that the crime under discussion is perpetuated by African Americans. Drawing on Mendelberg (2001) and Stanley (2015), Khoo suggests that this simple theory can make sense of why politicians speak in code—they do so to afford themselves plausible deniability for violating certain social norms against racism. This enables these individuals to in fact violate these norms by conveying, implicitly, the racist attitudes, without being held accountable to them. But coded speech is not the only way to subvert such social norms. In her chapter, “Racist and sexist figleaves” (Chapter 10), Jennifer Saul articulates the notion of a figleaf, which is an utterance that blocks the inference (that would otherwise be warranted) that the speaker has violated the relevant social norm. Examples include, “I have great respect for women, but…” or “I have many Black friends….” The success of these intended figleaves may vary across individuals, but Saul’s point is that, given that social norms are standards that we hold each other accountable to, if enough people think the figleaf is successful, this will amount to the figleaf-offerer avoiding sanction. Saul argues that the net result of such behavior may be the gradual erosion of the social norm itself. Thus, coded speech and figleaves provide ways to violate social norms without repercussion. Given that they are so easily subverted, is there any value to such norms? Often derided as “political correctness,” perhaps such social norms just serve to further entrench political divisions, and even pave the way for demagogues who openly flout them.1 Against this thought, Khoo argues that, even if they did little to curb expressions of racism, social norms against racism may play a vital role in providing assurance (in Waldron 2012’s sense) that members of such persecuted groups have an equal standing within the society. Another way in which one may communicate solidarity with a group is by protesting laws or behaviors that are oppressive toward that group. But, very often, the message of the protest is not put forward as a standard communicative exchange—that is, it is very often either already common ground among the audience, or it is clear it will not be accepted by the audience to whom it is directed. So, we might think that protests aren’t communication in any usual sense we are familiar with. In their chapter, “Protest and speech act theory” (Chapter 11), Matthew Chrisman and ­Graham Hubbs draw on J. L. Austin’s speech act theory to argue that protests are a distinctive type of speech act that aims to convey opposition to some object and prescribe redress related to that object. What distinguishes a protest from a complaint is that protests bring to bear a conception 4


of fairness or justice in a general way on the object of protest, and are addressed to someone with the requisite authority to affect the object of protest in the proposed way. Thus, a traveler who complains to a flight attendant about the plane’s tardiness merely expresses a complaint since they are not bringing any conception of justice to bear on the matter and the flight attendant is not in a position to rectify the situation. However, a person holding up a Black Lives Matter sign on the front lawn of the White House is protesting by conveying to politicians with the relevant political authority that Black Americans have been unjustly abused by police. Cases where the same speech act may be interpreted differently by different audience members give rise to defective contexts, in which some of the participants have different presuppositions (cf. Stalnaker 1978). For instance, suppose Jones attempts a racial figleaf: “I have nothing against Black people, but they wouldn’t be disproportionately imprisoned if they obeyed the law.” Some members in the audience may not accept the figleaf (and think others reject it as well) and conclude that Jones harbors racist attitudes toward Black people; others might accept the figleaf (and think others do so as well), and think that Jones is not racist and merely offering a reasonable explanation of the facts. Given these differing presuppositions, these individuals will be in a defective context, which may affect their ability to communicate as the conversation progresses. For instance, statements presupposing that Jones is racist—such as, “Has Jones reconsidered his racist views towards African Americans?”—will run into communicative difficulties. Stalnaker argued that speakers tend to easily figure out whether they’re in a defective context when the defectiveness concerns a topic of interest, and thus defective contexts don’t really pose any threat to ordinary communication. In his chapter, “Defective contexts” (Chapter 12), Andrew Peet argues that this is not the case—defective contexts (where the defectiveness concerns some matter of issue) are actually quite common. One of Peet’s examples is a case of a hermeneutical impasse, which occurs when a presupposition central to one person’s understanding of the world is rejected by an interlocutor. Luvell Anderson has argued that the slogan “Black Lives Matter” can only be understood against the background presupposition that there is systemic injustice against African Americans by the police (Anderson 2017). Without accepting this presupposition, many White Americans interpret the slogan exclusively, and react by pointing out, “All lives matter.” Such a hermeneutical impasse leads to a defective context and miscommunication, and is not easily correctable, since the presuppositions involved are so fundamental and entrenched. The chapters in this section each make the case for thinking that actual human speech situations diverge from the ideal theoretical conditions typically assumed in philosophical and semantic models of linguistic communication. Much like scientists looking to apply their theories outside of the lab, as theorists of language we should consider how our models make contact with the phenomena they aim to explain, which may involve interrogating our theoretical tools, or supplementing our theories with resources from neighboring fields.

Part III:  Linguistic Harms Just as language can be exploited as a communicative tool to deceive and subvert social norms, so too can it be used to directly or indirectly harm others. In recent years, the literature on linguistic harms has exploded, with researchers turning attention to how language can be used to derogate, stereotype, undermine credibility, and how the unjust distribution of interpretive resources can harm groups by limiting their abilities to both express themselves and make sense of their experiences. The chapters in this part of the book explore some of the mechanisms by which linguistic harms are perpetrated, as well as how we might respond and resist such attempts at harm. Perhaps most saliently, language can be used pejoratively, to derogate or dehumanize members of certain groups. In her chapter, “Varieties of pejoratives” (Chapter 13) Robin Jeshion 5

Justin Khoo and Rachel Sterken

distinguishes pejorative lexical items (including slurs like the n-word and suffixes like -tard) from pejorative uses of words (like boy used to refer to an African American man) and pejorative speech acts, which may occur independently of an utterance of pejorative lexical items or any words used pejoratively at all (as in Trump’s comment about Megyn Kelly: “blood coming out of her wherever”). Within the category of pejorative lexical items, Jeshion distinguishes four categories: canonical slurs, descriptive slurs (“Christ killer”), gendered slurs (“slut”), and stereotyping expressions (“Welfare queen”). Jeshion discusses the main theories of what is particularly toxic about such expressions—that such expressions semantically, pragmatically, or socio- or psycho-linguistically communicate certain derogatory content, and express (or direct hearers to adopt) derogatory attitudes. Jeshion ends with an open call for theorists of pejorative language to expand their horizons beyond canonical slurs. Indeed, the literature on linguistic harms does seem to be expanding beyond an initial interest in lexicalized pejoratives. One such area of new research is in uses of language that may, individually, be only minimally harmful, but which occurs with such frequency that they constitute larger harms to the individuals or groups in question. In her chapter, “Microaggressions and the problem of attributional ambiguity” (Chapter 14) Christina Friedlaender offers an analysis of recent work on microaggressions, the study of which promises to shed light on malign uses of language. Microaggressions are “micro social practices that reinforce or contribute to one or more forms of oppression” that, as norms against large-scale oppression have become increasingly widely prevalent, have more frequently been the tool of oppressors. Friedlaender points to the phenomena of “attributional ambiguity,” which they call an “adaptation” of microaggressions: such ambiguity involves the perpetration of a microaggression that gives the perpetrator some plausible deniability, rendering the target unsure whether they have been wronged, which, in turn, epistemically impoverishes both the individual in question and prejudiced-against groups to which they belong. By presenting a case study of how oppression adapts to changing societal conditions (from macro- to microaggressions, and then to ambiguous microaggressions), Friedlaender gives us a set of concepts that will aid us in understanding the evolution of oppressive behaviors, including linguistic ones. Other kinds of linguistic harms are invisible, literally, because the harm consists in a lack of conceptual or linguistic resources available for certain groups to use in interpreting their experiences. Miranda Fricker calls such harms hermeneutical injustices, perhaps the most famous being the case of women in the workplace prior to the introduction of the term “sexual harassment.” Lacking a word for sexual harassment, such women faced challenges interpreting and labeling the mistreatment they suffered from (mostly male) colleagues. In her chapter, “Hermeneutical injustice” (Chapter 15), Rebecca Mason introduces Fricker’s influential theory of hermeneutical injustice and argues that it is both unclear in places and fails to account for all cases. For example, someone who is date-raped might lack the conceptual tools adequately to understand that they’ve been seriously wronged, even though they don’t lack the term and concept of rape. Mason argues such cases involve hermeneutic distortion: “concepts that comprise the collective hermeneutical resource are associated in ways that are biased or discriminatory.” In the case at hand, the person’s concept of rape is distorted because it doesn’t include the possibility that one can be raped by an acquaintance on a date. Mason argues that appreciating how prejudices in collective hermeneutical resources can arise from distortions allows for a fruitful analysis of a wider range of phenomena, including contributory injustice and testimonial quieting and smothering. In making sense of ourselves and the world, we often appeal to generalizations and the underlying explanations which support them. If, for example, you move to Hong Kong, it’s very helpful to know that Hong Kongers speak Cantonese and drive on the left side of the road. There are of 6


course exceptions, but these kinds of generic generalizations—generalizations like Hong Kongers speak Cantonese and nurses are overworked and underpaid—are distinctively useful in many settings. If you’re a nurse, it can be useful to know that nurses are overworked and underpaid. That helps you understand yourself and act on that situation in various ways. If you’re a child, it’s good to know that in many jurisdictions, children are protected by law from abuse. In these and indefinitely many other cases, generic claims help us understand ourselves, the world, our place in the world, and to act on such knowledge. However, generics may often times lead us badly astray. For example, upon hearing generics about a social kind—e.g. women are nurturing, women wear lipstick, and women are less ambitious than men—subjects come to believe that women by their very nature are nurturing, wear lipstick, and are less ambitious than men, which are all false and/or problematic in their own right. However, this is not all: such utterances, even utterances of true and seemingly innocuous generics, can reinforce the belief that the given social kind has a natural essence and makes it easier to appeal to such an essence in other explanations—e.g. why should Jenny stay at home with the kids? Because she’s a woman and she’s better suited to do it than her husband. More generally, treating a social kind as having a natural essence will cause subjects to misdiagnose why kind members possess the properties and exhibit the behaviors they do. It leads them to overemphasize inter-group differences, underemphasize intra-group differences and treat properties and behaviors as stable, immutable and necessary, when in fact they’re not. In light of these kinds of observations, several philosophers and psychologists have suggested that speakers should limit their use of generic sentences involving social kinds, and that they should respond to generic assertions with counter-speech. The thought here is that limiting the use and force of generics will limit some of their pernicious effects. Matthew McKeever and ­Rachel ­Sterken’s chapter, “Social and political aspects of generic language and speech” (Chapter 16) provides an overview of the various concerns raised about generics, bringing together evidence and observations from philosophy, linguistics, and psychology to classify and articulate these concerns. In addition, they present some arguments against limiting generic language. Finally, they propose some novel prescriptions to address the concerns, drawing on literature in conceptual engineering to reorient our understanding of possible solutions. Ethan Nowak’s chapter, “Language extinction” (Chapter 17), discusses the moral and prudential value of preserving endangered languages. He presents reasons from both popular and philosophical sources for why we should care about the fact that minority languages go extinct—from the potential loss of scientific knowledge that might happen only to be captured in an endangered language to the role speaking a language plays in the self-conception of its speakers—before sketching his own view, according to which certain speech acts are only performable in a given language, and so language extinction can lead to silencing. Tying that up to an interpretativist theory of semantics, according to which speech, belief, and desire are interrelated, he concludes that in depriving a minority language speaker of their linguistic heritage we risk thereby making them uninterpretable to us as creatures with beliefs and desires—that is, as fellow members of the human community. We may distinguish two components in language preservation efforts: reclamation and revitalization. Language reclamation comprises the efforts of a community to claim its right to speak a language (Leonard 2012), while language revitalization is the process of ensuring the language lives on in new speakers. In her chapter on the topic (“‘laxwalxwash potamáay súngaan ‘áawq // to be between the blind snake’s teeth’: Indigenous Language Reclamation Between the Fangs of a (Simulated) Dilemma,” Chapter 18), Shelbi Nahwilet Meissner raises a dilemma for Indigenous communities engaged in language preservation efforts. On the one hand, it seems such efforts can benefit from material aid offered by non-Indigenous peoples (researchers) and institutions (governments, NGOs, and universities) that would otherwise be unavailable. However, on the 7

Justin Khoo and Rachel Sterken

other hand, often such partnerships come with strings attached, including submitting to stringent externally imposed evaluation criteria and relinquishing control of the Indigenous community’s language materials. It may thus seem as if language preservation efforts will require a loss of linguistic sovereignty on behalf of the language community. Against this dilemma, Meissner argues that exogenously partnered language reclamation can succeed without compromising linguistic autonomy, citing the radical nation building efforts of the Kumeyaay language reclamation project, which was funded in part by the Heard Museum. Crucial to the success of these efforts, Meissner points out, is that the radical work done by the Kumeyaay communities was largely unknown to their non-Indigenous funding partners. Similarly, on the grassroots side, Meissner argues that the radical community-building endeavors of Pu’uhonua camps in Hawaii have led to the successful reclamation and revitalization of the Kanaka Maoli language, even though such efforts may not count as succeeding by state curriculum standards. Meissner draws on these cases and others to argue that the very linguistic sovereignty dilemma articulated above is the result of settler logics and concepts that ought to be rejected by Indigenous communities.

Part IV:  Applications The chapters in the final part of the book explore how the analytic tools familiar from philosophy of language may yield fruitful results when applied in various domains. Consider the question of what kinds of speech ought to be protected from legal regulation. Soliciting a bribe clearly ought not to be protected speech since it constitutes a crime, while holding up a sign criticizing governmental policies while on public property clearly ought to be protected. A standard thought is that what distinguishes these two kinds of speech is that one is an ­action—solicitation of a bribe—while one is an expression of an idea. In their chapter, “Language and free speech” (Chapter 19), Ishani Maitra and Mary Kate McGowan argue against various ways of making precise this kind of view. To start, as J. L. Austin argued, all speech, including communicative speech, is action: bribing is an action one does by uttering certain words, but so is asserting or questioning. The Spence-Johnson Test aims to improve on the distinction by holding that protected speech intends to convey a particular message and it is likely that the message would be understood in that context. But some messages are intimidations, and not all intimidations are protected speech (consider cross burning with the intent to intimidate, which the Supreme Court ruled is not protected by the First Amendment in Virginia v. Black). Instead, Maitra and McGowan, drawing inspiration from J. S. Mill (Mill 1998), argue that whether speech ought to be regulated depends on whether it constitutes legally recognized harms in the right sort of way. Exactly when speech constitutes such harms in the right way is a matter of deep controversy, but Maitra and McGowan suggest some examples to help us see where they see the debate going. Contrast a case of a viral YouTube video that convinces people that women are disposed to criminal activities, and thus causes employers to discriminate against women in their hiring decisions, with one in which an employer states a new policy against hiring women. In both cases, the speech leads to women being discriminated against, but in the YouTube video the harm is caused—mediated through some cognitive activity on behalf of employers— while in the policy case the harm is constituted by the enacting of the discriminatory policy. Maitra and McGowan suggest that this distinction between causing and constituting a harm may helpfully underlie a more fruitful reframing of the debate over free speech protections. Another illustration of an area for fruitful collaboration between philosophers of language and social/political philosophers is provided by Eric Swanson in his chapter, “Language and ideology” (Chapter 20). The study of ideology and ideological critique has largely been pursued independently of the analytic projects of interest to most contemporary philosophers of language. 8


But, as Swanson points out, many of the classic debates in the philosophy of language may “be intimately intertwined with ideological disputes that are clearly socially, historically, and politically important.” Swanson gives the example of “personalism”—the view that the meanings of utterances are entirely determined by the intentions of their speakers—and shows how such a view may interact in complex ways with the nature of offensive uses of language. So, there is clearly much to be gained by philosophers of language expanding their horizons to include ideology in their sights. On the flip side, Swanson suggests that the study of ideology may benefit from explicit analytic procedures aiming to diagnose mechanisms of ideological change and perpetuation. Swanson’s conclusion is ecumenical, suggesting that we should be open to our methods falsifying our hypotheses and also be open to enriching our analytical methods if their applications prove independently unsatisfactory. One way language might reinforce ideologies is by legitimating them. For instance, Rae Langton and Caroline West argue that pornography legitimates sexual violence toward women (what has come to be known as “rape culture”; cf. Langton 1993, Langton and West 1999). But, exactly what it means to legitimate something, especially in cases where the notion of legitimacy is not legal and the person doing the legitimating lacks any relevant authority? In “Language and legitimation” (Chapter 21), Robert Simpson discusses whether (and if so how) things that are formally illegitimate according to some law or social norm can gain pseudo-legitimacy via the behavior of non-authoritative individuals. Simpson considers two models of pseudo-legitimation. One comes from Ishani Maitra according to which non-authoritative individuals can gain licensed authority to legitimate by way of acting as if and others going along with their behavior (Maitra 2012). One of Maitra’s central examples is that of a group deciding where to go on a hike—in such cases, although no one has authority to decide unilaterally for the group, one member may just start acting as if they have such authority, and then, if no one objects, their authority to direct the group will be legitimated. The other model comes from Rae Langton’s proposal that legitimation is the result of making the relevant practice seem commonplace and inevitable (Langton 1998). On the Langton model, it is a background assumption that ordinary, commonplace, practices are usually legitimate, and thus when a practice is represented as ordinary and commonplace, it thereby gains legitimacy. Drawing on Bauer 2015’s criticisms of Maitra’s model, Simpson argues in favor of Langton’s model, showing how it may be extended in various cases to avoid potential challenges. Regarding the language of socially significant categories, such as gender or sexual orientation, we should distinguish two kinds of questions. Descriptive projects aim to correctly characterize what the expressions of natural languages actually mean; while ameliorative projects aim to characterize the concepts and expressions we ought to use to think and talk about such subject matters (cf. Haslanger 2000 and Matti Eklund’s chapter, “Conceptual engineering in philosophy” in this volume). Consider the example of third-person singular pronouns in English, which are gendered. A descriptive question is what properties of individuals (gender or biological sex, for instance) correctly govern the use of such pronouns. A related normative question is what properties of individuals ought to govern the use of such pronouns. In “How much gender is too much gender?” (Chapter 22), Robin Dembroff and Daniel Wodak consider a further normative question: whether we ought to have gender-specific terms in natural language at all. They argue that natural languages ought to have only non-gender-specific pronouns, honorifics, suffixes, and generics on the grounds that gender-specific expressions tend to have stigmatizing and stereotyping effects on certain genders, exclude certain individuals who fall outside of its grammaticalized categories, and force speakers into a bind to either deceive or disclose gender information. Rather than proliferate gendered expressions, which may lead to misrepresentation in practice, they suggest eliminating them from natural languages altogether. 9

Justin Khoo and Rachel Sterken

Importantly, however, Dembroff and Wodak do not advocate eliminating from language every expression we might use to categorize individuals by gender. Rather, they emphasize that having expressions like man and woman is important to identify real social problems like misogyny. Instead, we might take their target to be expressions that talk about things which may be identified independently of their gender. For instance, one cannot identify women except by their gender, but one can identify individuals independently of their genders. Related to questions of what gendered expressions mean are questions about sexual orientation terminology (words like gay or straight): what such terminology means, what it should mean, and what the landscape of conceptual distinctions between sexual orientations ought to be. In their chapter, “On language and sexuality: demisexuals, polyamorous, bambi lesbians, and other queers” (Chapter 23), Saray Ayala-Lopez and E. Díaz-León consider both kinds of projects. They argue (pace Dembroff 2016) that sexual orientation language is relational—relating the sex/gender of the subject to the sex/gender of their preferred sexual partners—and dispositional—defined in terms of the dispositions to instantiate certain mental states (sexual desires, fantasies, and so on). Díaz-León and Ayala-Lopez discuss some of the moral and political considerations driving the ameliorative project, including sensitivity to hermeneutical injustices faced by individuals unable to make sense of their sexual desires in a positive way, the role such expressions can play in emancipatory movements fighting for equal rights, and whether expanding the range of conceptual distinctions could marginalize certain sexual orientations. Included in what’s at stake in this debate is whether we should expand (or even possibly contract) our existing conceptual repertoire to include perhaps sexual orientation categories around race, number of lovers, sexual act type, and so on. Having distinguished descriptive and ameliorative projects, we might now wonder whether we find such projects “in the wild” so to speak—that is, outside of the philosophy room. In “The language of mental illness” (Chapter 24), Renée Jorgensen Bolinger argues that the debate over the proper classification of mental illnesses is best understood as a debate in conceptual ethics. She identifies various interconnected interests that matter to the debate, including researchers looking for categories to set up experimental trials, clinicians and service-users looking for useful categories for diagnoses and treatments as well as reassurance that certain experienced ailments are real and not something they should be blamed for. Exactly how to balance these often-conflicting considerations is a moral and political issue, not a purely scientific one. Bolinger also discusses the harms of uses of mental illness pejoratives—expressions like “crazy”—arguing that they inflict two wrongs: a potential undermining of the perceived credibility of the immediate target of the expression as well as of individuals suffering from mental illness more generally. Bolinger argues that the fact that their uses inflict these harms is sufficient reason to avoid using mental illness terms pejoratively, but she does not conclude that we should censor or stigmatize the use of such expressions. Rather, she proposes insulating “the group from the harms of pejorative uses by rejecting the terms as descriptive predicates for the group” and uprooting the social stigma associated with mental illness. The aim of such efforts would be to allow such mental illness expressions to survive only as dead metaphors, largely devoid of their pejorative meaning.

Note 1 See Stanley, “Democracy and the Demagogue,” for an argument to this effect.

Bibliography and Further Reading Anderson, Luvell. 2017. Hermeneutical impasses. Philosophical Topics, 45(2), 1–20. Anderson, Luvell, Sally Haslanger and Rae Langton. 2012. Language and race. In Gillian Russell & Delia Graff Fara (Eds.), The Routledge Companion to Philosophy of Language (pp. 753–67). New York: Routledge.


Introduction Augustine. 1952 [395]. Lying. In R. Deferrari (Ed.), Treaties on Various Subjects (Vol. 16, pp. 53–120). Washington, DC: Catholic University of America Press. Austin, J. L. 1962. How to Do Things with Words. Oxford: The Clarendon Press. Bauer, Nancy. 2015. How to Do Things with Pornography. Cambridge, MA: Harvard University Press. Brandom, Robert. 1994. Making it Explicit. Cambridge, MA: Harvard University Press. Burgess, Alexis, Herman Cappelen and David Plunkett (Eds.). 2020. Conceptual Engineering and Conceptual Ethics. Oxford: Oxford University Press. Butler, Judith. 1999. Gender Trouble: Feminism and the Subversion of Identity. 2nd edition, New York & London: Routledge. Carson, Thomas L. 2006. The definition of lying. Noûs, 40(2), 284–306. Dembroff, Robin. 2016. What is sexual orientation? Philosophers’ Imprint, 16(3), 1–27. Fricker, Miranda. 2007. Epistemic Injustice: Power and the Ethics of Knowing. Oxford: Oxford University Press. Grice, Paul. 1989. Studies in the Way of Words. Cambridge, MA: Harvard University Press. Haslanger, Sally. 2000. Gender and race: (What) are they? (What) do we want them to be? Noûs, 34(1), 31–55. Haslanger, Sally. 2011. Ideology, generics, and common ground. In Charlotte Witt (ed.), Feminist Metaphysics (pp. 179–207). Dordrecht: Springer Verlag. Khoo, Justin. 2017. Code words in political discourse. Philosophical Topics, 45(2), 33–64. Langton, Rae. 1993. Speech acts and unspeakable acts. Philosophy & Public Affairs, 22(4), 293–330. Langton, Rae. 1998. Subordination, silence, and pornography’s authority. In Robert C. Post (ed.), Censorship and Silencing: Practices of Cultural Regulation (pp. 261–84). Los Angeles, CA: Getty Research Institute for the History of Art and the Humanities. Langton, Rae and Caroline West. 1999. Scorekeeping in a pornographic language game. Australasian Journal of Philosophy, 77(3), 303–19. Leonard, Wesley. 2012. Framing language reclamation programmes for everybody’s empowerment. Gender and Language, 6(2), 339–67. Leslie, Sarah-Jane. 2017. The original sin of cognition: Fear, prejudice, and generalization. Journal of Philosophy, 114(8), 393–421. Lewis, David. 1975. Languages and language. Minnesota Studies in the Philosophy of Science, 7, 3–35. Lewis, David. 1979. Scorekeeping in a language game. Journal of Philosophical Logic, 8, 339–59. Maitra, Ishani. 2012. Subordinating speech. In Ishani Maitra and Mary Kate McGowan (eds.), Speech and Harm: Controversies Over Free Speech (pp. 94–120). Oxford: Oxford University Press. Maitra, Ishani and Mary Kate McGowan. 2012. Speech and Harm: Controversies Over Free Speech. Oxford: Oxford University Press. Mendelberg, Tali. 2001. The Race Card: Campaign Strategy, Implicit Messages, and the Norm of Equality. Princeton, NJ: Princeton University Press. Mill, John Stuart. 1998. On Liberty and Other Essays. Oxford: Oxford University Press. Mills, Charles W. 2005. “Ideal theory” as ideology. Hypatia, 20(3), 165–84. Nietzsche, Friedrich. 1901/68. The Will to Power. Translated by W. Kaufmann. New York: Random House. Rawls, John. 1971. A Theory of Justice. Cambridge, MA: Harvard University Press. Saul, Jennifer. 2012. Lying, Misleading, and What Is Said: An Exploration in Philosophy of Language and in Ethics. Oxford: Oxford University Press. Saul, Jennifer. 2017. Racial figleaves, the shifting boundaries of the permissible, and the rise of Donald Trump. Philosophical Topics, 45(2), 97–116. Searle, John. 1969. Speech Acts: An Essay in the Philosophy of Language. Cambridge: Cambridge University Press. Stalnaker, Robert. 1978. Assertion. In P. Cole (Ed.), Syntax and Semantics 9: Pragmatics (pp. 315–32). New York: Academic Press. Stanley, Jason. 2015. How Propaganda Works. Princeton, NJ: Princeton University Press. Waldron, Jeremy. 2012. The Harm in Hate Speech. Cambridge, MA: Harvard University Press. Williamson, Timothy. 1996. Knowing and Asserting. The Philosophical Review, 105(4), 489–523.



Social and Political Language Methodological and Foundational Issues


1 Introduction A good place to start when explaining and motivating what is variously called conceptual engineering, conceptual ethics, and ameliorative projects is the following passage from Nietzsche: What dawns on philosophers last of all: they must no longer accept concepts as a gift, nor merely purify and polish them, but frst make and create them, present them and make them convincing. Hitherto one has generally trusted one’s concepts as if they were a wonderful dowry from some sort of wonderland: but they are, after all, the inheritance from our most remote, most foolish as well as most intelligent ancestors. […] What is needed above all is an absolute skepticism toward all inherited concepts…1 Nietzsche puts this in characteristically dramatic terms. Put less dramatically, the basic idea is that there are concepts other than those we actually have and employ, and some of those concepts may be better suited for various purposes than our actual ones are. We can and should question our concepts, and consider whether there are better ones out there. And insofar as there are better ones out there, we should try to change our ways so that we come to employ the better concepts. There is then a philosophical project of critically assessing, and potentially revising or replacing, concepts. I will later say more about how to understand the talk of concepts here. But there is one thing it is important to be clear on right away. The project is not just one of critically assessing our theories of various subject matters. It is one of critically assessing the concepts themselves, where “concepts” are the basic representational building blocks of thought. Sometimes in ordinary language, “concept” is used loosely, in such a way that mere changes in theory trivially count as changing one’s “concept”. Here I will employ a concept/conception distinction, such that while such changes trivially count as changing one’s conception of the matter at hand, it is a further question whether this counts as a revision of the concept itself. Maybe in some cases it does, but it is a theoretically contentious matter if and when it is so.2 This project described is not a new one. Philosophers have often (or always?) to some extent critically assessed concepts, coming up with their own technical concepts where the old concepts have not seemed up to the task, as well as revise and criticize existing concepts. Within the analytic philosophy tradition, Carnap (1950, ch. 1) and Quine (1960) both prominently stress explication. Within the continental tradition, theorists like Heidegger make up their own concepts 15

Matti Eklund

for describing the world. Brandom (2002) describes Leibniz, Spinoza, and Hegel as conceptual engineers. 3 As displayed by the passage quoted above, conceptual engineering is a central theme in Nietzsche. And these are just a few examples. But what arguably is new is systematically studying conceptual engineering, where that involves fnding features common to seemingly quite disparate conceptual engineering projects and delving into general questions raised by such projects. For example, it can be claimed that the revision and replacement of concepts isn’t always self-conscious, and by making explicit what we are doing and being more systematic we can do it better; and by comparing conceptual engineering projects from diferent parts of philosophy we can shed useful light on them. Much of the discussion below will concern the study of conceptual engineering – the newer development – rather than the activity of conceptual engineering itself. There has been a furry of work on conceptual engineering in recent years. There is too much to usefully list or summarize, but among more important work can be mentioned Burgess and Plunkett’s pioneering articles (2013) and (2013a), Cappelen (2018), Scharp (2013), the essays collected in Haslanger (2012), and the essays collected in Burgess, Cappelen, and Plunkett (2020). In the discussion below I will keep returning to paradigm examples of conceptual engineering from the recent literature: Sally Haslanger on the concept woman and Kevin Scharp on the concept truth.4 This is partly for the sake of convenience. These examples are often discussed, and well-known. But it should be kept in mind that these are just some of many examples. Diferent labels have recently been used for the kind of project I will describe – “conceptual engineering”, “conceptual ethics”, and “ameliorative project”. For the purposes of my discussion here I will not be concerned with diferences between these labels, even though distinctions can be drawn.5

2 Examples Here, in brief, are Haslanger’s and Scharp’s ideas. Haslanger thinks that we should replace the actually used concept woman by the following one: S is a woman if S is systematically subordinated along some dimension (economic, political, legal, social, etc.), and S is “marked” as a target for this treatment by observed or imagined bodily features presumed to be evidence of a female’s biological role in reproduction.6 The motivation, put briefy, is that there is oppression of the kind described, and employing a concept that captures that is a way of raising the fact to salience, in a way that helps combat this oppression. I should perhaps stress that Haslanger at least sometimes clearly speaks of changing our conception of the concept woman rather than changing the concept woman itself: she emphasizes that how the concept seems to us (“the manifest concept”) difers from how the concept really works (“the operative concept”).7 However, in the discussion to follow my focus will be on Haslanger qua conceptual engineer. Scharp argues that our concept truth is in a certain sense inconsistent, in a way that comes to light through semantic paradoxes such as the liar paradox. He then goes on to argue that because the concept truth is inconsistent we should replace it, and he proposes as replacement a pair of concepts which together can do the work that, in theoretical contexts, we want the concept truth to do for us.8 The diferences between Haslanger’s and Scharp’s projects are stark, which is one reason for juxtaposing them in the context of discussing conceptual engineering. The concepts they focus on are obviously very diferent, and the considerations they bring to bear are also diferent. 16

Conceptual Engineering in Philosophy

3 Preliminary Remarks Armed with these very brief sketches of two prominent conceptual engineering projects, let me now address some questions which likely arise. What are concepts? There is no uncontroversial answer to this question. At a frst stab, they are the basic building blocks of thoughts, or of the thoughts that have something like linguistic structure. Perhaps one can say that concepts are the meanings of (subsentential) expressions.9 Needless to say, these characterizations carry theoretical baggage, and can be questioned. And even if the latter gloss can be defended, it is only moderately helpful, for example because there are diferent theories of what meanings are. One important distinction between views on meaning is that between views on the meaning of an expression is its intension, and views on which meaning is something more fne-grained which determines its intension. Some authors in the conceptual engineering literature, e.g. Scharp, approach meaning and concepts via the question of what it is to be competent with an expression or concept. One might for example say that to be competent with a concept is to be disposed to accept such-and-such principles involving it, or (Scharp’s preferred view) entitled to accept such-and-such governing principles it. This approach allows for meanings or concepts to be more fne-grained than intensions: for diferent sets of governing principles can correspond to one and the same intension. While as it is often discussed conceptual engineering concerns, precisely, concepts, this need not be taken very seriously.10 First, somewhat more neutrally one can speak of representational devices. And some may prefer to speak of words and their reference and extension, instead of speaking of concepts. Instead of asking which concepts we should have, one can ask which things we should have words for. Second, at least for some purposes one might sidestep representational devices altogether and instead directly speak in terms of that which such devices refer to or ascribe. Instead of saying, for example, “we should employ the concept green and not the concept grue” one can say: we should focus on the property greenness and not the property grueness. (Note that questions analogous to those of the individuation of concepts arise also in the case of properties: can there be two properties which are necessarily co-instantiated yet are distinct?) I say “for some purposes”, both because there are not always corresponding worldly items (consider e.g. nonrepresentational language and thought), and because sometimes what is at issue is not just what is represented but the manner in which it is represented. Who are “we”? In the above characterizations of Haslanger’s and Scharp’s projects, I have spoken of what “we” should do. But who are “we” here? Answers can difer depending on the project concerned. Scharp is at pains to emphasize that his proposal that the concept truth be replaced concerns only what certain kinds of theorists should do, in the context of their theoretical investigations. He is not at all concerned with everyday uses of the concept. By contrast, Haslanger is at least in part concerned with what concept woman should be used more broadly, even if her focus is on feminist theorizing. Retail or wholesale? The previous remarks are relevant also another natural question. Is the conceptual replacement or revision supposed to be retail – the proposal concerns only some uses of the concept – or wholesale – concern all uses of the concept? As already the above remarks on Scharp indicate, Scharp’s revision is clearly retail. He only means to address the question of which truth concept should be employed in certain theoretical contexts. Whereas I don’t see Haslanger committing herself to a wholesale view, her view is at least closer to such a view. What is it for a concept to be better than, or preferable to, another? First, it can be claimed that there are some possible outright defects that a concept can have and that a concept is generally better than another if the former is defect-free and the latter has one of these defects. Among such defects can be inconsistency and indeterminacy. Another defect might be that the use of the concept in some sense presupposes the truth of a false view, for example a false normative view.11 (Perhaps the use 17

Matti Eklund

of a concept expressed by a racial slur presupposes the truth of some false racist belief. Perhaps the use of concepts like those expressed by “promiscuous” or “chaste” presupposes the truth of certain false moral beliefs.12) Second, there can be reasons that are more specifc to the task at hand. A concept can be better than another in the sense that employing the former is more likely to generate positive social change. A concept can be better than another because it better tracks what is explanatorily powerful or inductively useful. A concept can be better than another simply because it is easier for creatures like us to work with. And so on. There will be many relevant dimensions along which a concept can be better than another. What is it to revise a concept? Sometimes in discussions of conceptual engineering, the focus is on revising a particular concept: before the changes the concept has a particular intension and maybe some particular governing principles, and after the changes the concept has a diferent intension and maybe diferent governing principles. Such talk invites questions about what changes a concept can undergo and still remain the same concept. Can a concept really survive a change in intension or governing principles? My own view is that such questions are not very important. There are simply diferent ways of individuating concepts, and so long as we are clear on which method of individuation we choose, nothing of interest hinges on the choice. If we are strict about concept identity then what some people speak of as one concept undergoing changes is in fact the replacement of a concept by another, related one. I myself gravitate toward such strict individuation; but I do not see my favoring that as my taking a stance on a theoretically signifcant issue. At any rate, not all changes to our overall conceptual repertoire in the spirit of the Nietzsche passage above would be instances of the kind of concept change just discussed, even for those happy with the idea that there is such concept change. Sometimes there is arguably genuine conceptual innovation, where a concept is added such that there is no earlier counterpart of the concept. Sometimes there may be cause for simple concept deletion: some concept is dropped, and no concept is added specifcally to replace it. Having and employing concepts. Independently of the issues of conceptual change mentioned above there are good questions about what it is to have and to employ a concept. Consider the concept which Haslanger proposes to identify as the concept woman. There is a sense in which we had this concept even before Haslanger entered the scene. We had the concepts subordination, etc., and we grasped the relevant means of forming complex concepts from simpler ones. So in what sense is what Haslanger proposes a conceptual novelty? There are distinctions to be drawn here. The following gestures toward what one relevant distinction might be: although we had all the constituent concepts and grasped their mode of combination, and so in a sense had the complex concept all along, we have not so far actually employed these complex concept, and starting to employ it is a change in our practices. Needless to say, more can be said about how to draw the relevant distinction. These distinctions are relevant to some stated motivations for conceptual engineering. Burgess and Plunkett (2013) say that one central motivation for conceptual engineering (or “conceptual ethics”, as they put it) is that which concepts we have “fxes what thoughts we can think” – our conceptual repertoire determines not only what beliefs we can have but also what hypotheses we can entertain, what desires we can form, what plans we can make on the basis of such mental states, and accordingly constrains what we can hope to accomplish in the world. But to the extent that the engineering concerns which concepts we more easily or readily employ, and not which concepts we have, some of this rhetoric should arguably be tempered. Already the point that some concepts are more easily and readily used than others is relevant – there may be good reasons to modify what concepts are easily and readily used – but it shouldn’t be confated with points about which concepts we have in the frst place. 18

Conceptual Engineering in Philosophy

4 Concepts and Words Above I mentioned that instead of asking which concepts we should employ, one can ask about reference and extension, and ask which things we should have words for. Now, even two theorists who agree about which concepts to employ and about things we should have words for can sharply disagree about which words should be used for what. It is important to distinguish between diferent kinds of questions that arise. Some questions concern what to have labels and descriptions for; others concern what labels and descriptions to use. While questions of the former kind arguably are more central to conceptual engineering, questions of the latter kind also come up.13 Cappelen (2018) prominently discusses what he calls lexical efects in connection with conceptual engineering. Examples he mentions to illustrate the phenomenon are that of naming a child “Hitler”, the use and importance of brand names, and the conception of slurs given which their slurring efects are due not to semantic features but simply due to which labels are being used.14 Especially in connection with Haslanger’s project, the question has come up of whether one should retain the old word for the revised concept. Should we use “woman” for the concept that Haslanger devises or not? Haslanger herself says, …if our goal is to identify a concept that serves our broader purposes, then the question of terminology is primarily a pragmatic and sometimes a political one: should we employ the terms of ordinary discourse to refer to our theoretical categories, or instead make up new terms? The issue of terminological appropriation is especially important, and especially sensitive, when the terms in question designate categories of social identity such as “race” and “gender”. …by appropriating the everyday terminology of race and gender, the analyses I’ve ofered invite us to acknowledge the force of oppressive systems in framing our personal and political identities. Each of us has some investment in our race and gender: I am a White woman. On my accounts [of race and gender], this claim locates me within social systems that in some respects privilege and in some respects subordinate me. Because gender and racial inequality are not simply a matter of public policy but implicate each of us at the heart of our selfunderstandings, the terminological shift calls us to reconsider who we think we are.15 No doubt terminology is important, for example in cases like those of race and gender. But there is still something that might seem problematic about what Haslanger says: “Each of us has some investment in our race and gender: I am a White woman. On my accounts, this claim locates me within social systems that in some respects privilege and in some respects subordinate me”. Here one starts with a particular claim, in the case of Haslanger that expressed by the sentence “I am a white woman”. What are claims? It is natural to identify claims as propositions. If I say “snow is white” and a German speaker says “Schnee ist weiss” we are making the same claim; that is, we assert the same proposition. If claims are propositions, then what Haslanger speaks of is what a particular proposition does on her accounts of gender and race. But a diferent reaction is to say that before Haslanger’s proposed conceptual and terminological revision, “I am a white woman” expresses one proposition and after the revision has been implemented, it expresses another. One may then be concerned that there is no one proposition such that we are invested in that proposition pre-revision, and that then that very proposition is such that given a revisionary account of it, that proposition locates the speaker in a social system in the way described.16 Some prominent criticisms of Haslanger have focused on terminology (see, e.g., Katharine Jenkins (2016)). Critics have fastened on the fact that Haslanger’s defnition of “woman” seems to exclude trans women, and they have objected to this consequence of the view. This criticism, even 19

Matti Eklund

if on target, has to do primarily with the label associated with the concept. If Haslanger instead had used a made-up label for the concept she characterizes, this would not have been an issue – at least not so long as Haslanger also allowed for another concept which would be labeled “woman”, under which trans women would fall.17 Jenkins proposes a revision of the concept woman such that in the sense of the proposed concept woman, trans women are “women”. This is meant as an improvement over Haslanger’s proposal. But one may worry that insofar the proposition regarding trans women that one seeks to defend is that trans woman indeed are women, one does not defend that proposition by proposing a new concept woman such that “trans women are women” expresses a true proposition when “women” expresses the new concept. To believe otherwise is tantamount to think that if we call a dog’s tail a leg, a dog will have fve legs. To defend the original claim one must defend that trans women are women in the ordinary, pre-revision sense of “woman”. Maybe that can be done – I am not weighing in on that issue – but concept revision does not make that case.18

5 Doubts about Conceptual Engineering as a Philosophical Project When motivating attention to conceptual engineering, Plunkett and Cappelen (2020) stress the ubiquity of concept use. Whatever cognitive activities we engage in, we employ concepts, and which concepts we employ matters to how we perform the activity in question. Plunkett and Cappelen say that facts such as these “help underscore the potential signifcance of philosophical work done on conceptual engineering”.19 Now, they do say “potential”, but it is anyway important to stress that the central importance of concepts by no means immediately shows that there is a signifcant philosophical project of studying conceptual engineering. In this section I bring up three possible reasons for skepticism. 1


Not unifed enough? Even if individual projects of conceptual engineering, such as Haslanger’s and Scharp’s, are of interest, one can think that these projects are dissimilar, in such a way that nothing is gained by treating these projects as parts of a bigger, unifed project, conceptual engineering. The projects may be no more unifed than the areas of philosophy of gender and race on the one hand and philosophy of logic, on the other, are unifed. To be sure, in both cases, we talk about revising and replacing concepts, but the means of assessing the proposals are so diferent that there is no unifed philosophical project there. There are, of course, general questions about concepts and their identity conditions (as well as about the legitimacy of concept talk in the frst place), but these questions are so general that they have nothing to do with conceptual engineering per se. Plunkett and Cappelen suggest that the importance of questions about which concepts we use means that “work done on conceptual engineering” may be signifcant. But, relating back to the distinction drawn early on, even if individual projects involving engineering concepts may be signifcant, the general study of conceptual engineering is a diferent thing, and it is in regards to that study that one may be concerned about lack of unity. Not philosophical enough? A second cause for doubt about conceptual engineering as a philosophical project concerns how philosophical the project properly is. Might not the question of how good or bad a concept is be a question of the actual consequences of employing it, as compared to employing other concepts – an empirical matter? And are not questions about how a concept is best implemented questions to be resolved empirically? Perhaps in each case one can to some extent speculate reasonably from the armchair; still, such speculation is no substitute for empirical work. It may be silly to police how the label “philosophy” should be used, but it still stands that the typical training of philosophers makes philosophers better equipped to address certain questions than to address others. 20

Conceptual Engineering in Philosophy


Which concept woman is best for particular feminist political purposes – whether we think of a concept woman for working feminist theorists to use or we think of a concept woman for the society at large to use – seems an eminently empirical matter. A project like Scharp’s can seem more non-empirical in nature, for the problem identifed in the case of the concept truth is inconsistency, and it can be thought that empirical research is not needed to establish that inconsistency is a defect. But frst, empirical work could still be needed to properly investigate which replacement concept or concepts are the most useful to employ. And second, while inconsistency certainly sounds like a defect, whether it is depends on what it is for a concept to be inconsistent – and there are diferent accounts of that. The way Scharp characterizes this, a concept is inconsistent exactly if competence with the concept entails being (defeasibly) entitled to accept principles which, in fact, lead to inconsistency.20 It is not obvious that this is cause for replacement. What is so bad if I employ a concept like that? My employing such a concept does not in any way mean that I believe contradictions, or am disposed to believe contradictions, or am in some sense committed to believing contradictions. But if the inconsistency of a concept is not generally a defect, the question of whether it is in a particular instance can be thought to turn on whether the inconsistency in fact leads to problems, and that is again an empirical issue. I will return to this below. A diferent kind of reason for thinking that questions about what concepts should be revised are at bottom empirical is that much of what we use words to convey – and arguably much of what we use concepts to think – goes beyond what is literally expressed by those words and what is strictly part of the contents of those concepts. But then much of a word’s or concept’s value or disvalue can have to do with what else the word or concept is being used to do; and the question of what else it is used to do is empirical. Maybe some conceptual engineering projects are more philosophical in nature. What makes Haslanger’s and Scharp’s motivations for conceptual engineering seem empirical is the focus on purported practical benefts of a given kind. A completely diferent kind of “conceptual engineering” project focuses not on practical matters but on what a kind of ideal language would be like – the language our theory is stated in at the end of inquiry, or the language that most perspicuously represents what reality is fundamentally like. Such projects may be more philosophical, for an obvious reason: designing such a language is not very diferent from the project of considering what reality is fundamentally like, and that latter question is one where philosophy is traditionally seen as having contributions to make.21 No distinct subject matter? A third reason for doubt is that conceptual engineering is not a distinct subject matter. I already alluded to the fact that some relevant questions, about concepts and their identity conditions, are more general and do not concern conceptual engineering per se. Also some other questions that have come up in connection with conceptual engineering seem clearly just to be applications of more general points. For example, Simion (2018) argues that it is not permissible to revise concepts when there is “epistemic loss”, and she says that what she identifes is a wrong-kind-of-reasons problem familiar from elsewhere in the literature.22 Beliefs are to be assessed on their own, epistemic terms rather than in terms of all-things-considered considerations. Whatever the fate of Simion’s argument, it is clear that it turns on how epistemic value relates to other sorts of normative considerations.23

6 Conceptual Engineering Questions One way for the friend of conceptual engineering to combat skepticism about conceptual engineering as a philosophical project is to, constructively, point to some philosophical questions that concern conceptual engineering as a general philosophical project, and emphasize the signifcance of these questions. Here are some questions she might point to. 21

Matti Eklund

First, there is what has come to be called Strawson’s challenge.24 This is discussed in, e.g., Prinzing (2018), Cappelen (2018), and Dutilh Novaes (2020). Here is Cappelen’s (2018) statement of the challenge: Change of extension and intension (and so also change of sense, if you believe in those) is a change of topic, so [projects involving conceptual revisions] are bound to fail. Even if the revisions succeed, they do not provide us with a better way to talk about what we were talking about; they simply change the topic.25 The concern is this. When we engage in conceptual engineering, we revise or replace concepts, and when we do so, we change the topic. But if we change the topic, then we no longer address the questions we originally were asking. This is a general concern, applying equally to projects like Haslanger’s and to projects like Scharp’s. And it applies specifcally to conceptual engineering, and isn’t equally an issue for all theories in some sense about concepts, for it concerns precisely conceptual change. Cappelen’s own response to the challenge involves defending a liberal notion of sameness of topic given which the topic is not changed. My own view on the challenge is that it is a non-issue. Whether it matters that the topic is changed depends on what our purposes are. If the concept revision is supposed to be justifed by moral and political aims, then the chief question is whether those aims are better served by the new concept. If it is supposed to be justifed by concern about explanatory and predictive success, then the chief question is whether the new concept better serves that aim. Change of topic is a bad thing exactly insofar as, among other purposes we do or should have, there is the purpose of saying something about the old topic. But why should that always be among our purposes? Maybe we are not answering the original questions asked. But maybe the old questions weren’t the right ones to ask anyway. When motivating Strawson’s challenge, Cappelen quotes, among others, Peter Ludlow (2005). What Ludlow actually says is: “[F]irst, and most obviously, any investigation into the nature of knowledge which did not conform to some signifcant degree with the semantics of the term ‘knows’ would simply be missing the point”.26 Of course, if an investigation really is an investigation into the nature of knowledge (to stick with the case Ludlow focuses on), then we had better make sure we not stray from the target, the nature of knowledge. But a conceptual engineer concerned with knowledge could well say: what we ought to be concerned with, e.g. for the purposes of the relevant kind of epistemic evaluation, isn’t knowledge but knowledge*. Then we ought not tie our investigation to the nature of knowledge but should shift the focus to the nature of knowledge* – at least so long as what we are interested in is what is best for the purpose of the relevant kind of epistemic evaluation.27 Second, recently (e.g. in Cappelen (2018), Koch (forthcoming)) there has been some work on whether externalism, the view that the meaning of an expression as used by a speaker is constitutively determined in part by factors outside her, makes the project of actively working for conceptual change unfeasible: speakers have no control over conceptual change. This is a question of the compatibility of externalism with conceptual engineering. Of potential relevance here is that there are diferent kinds of externalism. Sometimes externalists stress the causal environment; sometimes it is the speaker’s social environment that matters. Some externalists think some facts about meaning and content can be determined by future facts. Some externalists may think it is in the nature of meaning and content that it is determined in part by external factors where others may hold the more moderate view that it is in virtue of our implicit decisions and conventions that external factors turn out to be relevant (we have decided to let our concept water work in such a way that with respect to every possible world, it is true of the actual watery stuf ). The problems caused by externalism are at least more dramatic given the frst kind of view. 22

Conceptual Engineering in Philosophy

Again, is it not largely an empirical issue to what extent externalism presents problems? One possibility is that, for example, persuasive arguments for conceptual revision can have such impact on the community that even by externalist standards conceptual change is taking place, and since argumentation is the driving force, this is under our control. How likely it is that argumentation has that efect is an empirical issue.28 Third, maybe, despite diferences between conceptual engineering projects, there can be some informative principles governing how to evaluate concepts, and when to revise or replace concepts. For example, Alexis Burgess and David Plunkett have discussed the question of under what conditions a given concept can itself be legitimately be used in discussions of whether that very concept should be replaced or abandoned.29 If a concept is employed in an argument for retaining that concept, that may be seen as problematic: it is, so to speak, the concept itself that vouches for its innocence (or that is the concern). And if the concept is employed in an argument to the efect that the concept ought to be revised or replaced, that too might be seen as a problem: if the concept itself cannot be trusted, why should an argument that employs the concept? I am stating these questions only briefy. As stated they face the concern that a lot hinges on the specifc supposed reasons for why the concept should be replaced, and what the argument for replacement looks like. For example, if what is at issue is only that some other concept is more explanatory than C or the fact that C is more vague than relevant alternatives, there is no immediate reason at all to suspect that there is something amiss with using C in an argument that C should be retained. Fourth, there are questions about the limits of conceptual engineering. In his 2011 work, David Chalmers introduced the notion of a bedrock concept. As Cappelen (2018) discusses, this can be thought to have implications regarding conceptual engineering. The main topic of Chalmers (2011) is that of verbal disputes. As Chalmers notes, a strategy for deciding whether a given dispute is verbal is to state the dispute with the key term replaced by other terms, and see if the dispute remains. If the dispute goes away, it was verbal. If it remains, the dispute wasn’t verbal. But having stated this “method of elimination”, Chalmers notes an apparent limitation: in some cases it seems that the dispute cannot be faithfully restated without the use of the key term. Here he illustrates this in the case of moral disputes: …we can bar “ought” and introduce “ought1” and “ought2”, which are stipulated to apply to acts with the deontological and the consequence property respectively. Is there a residual disagreement now? As we proceed, the disagreement gets harder and harder to state. It is plausible that once all moral terms are gone, no disagreement can be stated. We might agree on all the nonmoral properties of the relevant actions but still disagree on whether it is right. In the case of “semantics”, “physicalism”, and so on, this situation suggested a verbal dispute. Should we likewise diagnose a verbal dispute here? Intuitively, the answer is no. For all we have said, moral disputes are substantive disputes. Instead, we have simply exhausted the relevant vocabulary. It appears that at a certain point (perhaps once we have fxed on the appropriate moral “ought”), we have reached bedrock: a substantive dispute involving a concept so basic that there is no hope of clarifying the dispute in more basic terms.30 Chalmers characterizes a “bedrock dispute” as “a substantive dispute for which no underlying dispute can be found by the method of elimination”, and “bedrock expressions” are such that “some disputes are bedrock with respect to those expressions” (where “a dispute is bedrock with respect to E when the dispute is substantive and there is no dispute not involving E (and analogous in other respects) that underlies the original dispute”). There is much to say about what Chalmers says about bedrock. Here I confne myself to discussing relevance to conceptual engineering. Cappelen (2018) brings up Chalmers’ notion of 23

Matti Eklund

bedrock as relevant to conceptual engineering. He says that in his terminology, bedrock expressions are “expressions that cannot be engineered”; they are “conceptual foundations where there’s no option of moving to a neighboring property”.31 He then goes on to say that he disagrees with Chalmers about whether there are any bedrock expressions. I think Cappelen at least overstates the relevance to conceptual engineering. If E is a bedrock expression then one cannot eliminate the expression without a sense of loss: one cannot faithfully state what was at issue in theses involving E when E is not used. But one can certainly still in principle argue that even so, E should be eliminated or revised. The loss in question is a loss we can and should live with. I have myself presented arguments regarding the limits of conceptual engineering, both suggesting that it becomes problematic to pose the conceptual engineer’s questions when it comes to foundationally important normative concepts, and that there is a sense in which there are no alternatives to the notions of truth and existence – there are, in a certain sense, no alternative suitably truth-like and existence-like concepts.32 In brief, the considerations are these. When it comes to truth, the question is what it would be for a community to employ a concept which plays the role of our concept truth but still is not coextensive with it. In the case of existence, the point is that diferences in what existence-like concept a community uses has ripple efects for the rest of that community’s system of concepts, so that selectively replacing existence by some other existence-like concept is not possible. The case of foundationally important normative concepts is the one I have discussed in the most detail – see my book Eklund (2017) – and there the point is, in brief, that it is unclear what the conceptual engineering could reasonably concern. When we ask, “which ought-like concept ought we to use, our actual concept ought, or an alternative?”, the obvious problem with the question is that the answer to the question will likely depend on what concept we use the “ought” to express. Cappelen brings up my arguments alongside those of Chalmers as arguments to the efect that some concepts cannot be engineered. Without getting into detail regarding those arguments of mine and their worth, let me just emphasize that the same sorts of remarks apply to them as applied to Chalmers’ arguments. Even if my arguments are completely correct as far as they go, and even if they do establish certain limits to conceptual engineering, certain forms of conceptual engineering are possible also when it comes to these concepts. There are still questions about whether one should employ a truth-like or existence-like concept at all. And in the case of normative concepts, my argument does not even purport to establish that we could not replace ought by some alternative, ought*: the problem rather has to do with what kind of comparison we are engaged in when proposing the replacement.

7 Diferent Conceptions of Meaning and Content Here is a rough distinction between two diferent ways of looking at concepts and meaning. On a referentialist view, there is nothing more to (descriptive) meaning than intensions. If two expressions have the same intension then they have the same meaning. On an inferentialist view, some inferences are in part and somehow constitutive of concepts or meaning. (The terminology is from Williamson (2009).) The inferentialist cuts things more fnely than the referentialist, for two expressions can have the same intension while being associated with diferent constitutive inferences. In the conceptual engineering discussion, Cappelen and Haslanger are referentialists.33 Cappelen even expresses doubts about the propriety of talk about concepts. Scharp is a prominent inferentialist (and I myself am attracted to inferentialism).34 These diferences, in turn, generate diferences in how to think of conceptual engineering. For the referentialist, the engineering has to do with changes in what we talk about (we shift from using an expression with one intension to using that or another expression with another intension), and with things that aren’t strictly on the meaning or content side at all – what Cappelen calls 24

Conceptual Engineering in Philosophy

lexical efects. For the inferentialist, there are more options: the inferentialist can also focus on what inferences are constitutive of an expression’s meaning or a concept. Whether one spells out constitutivity in psychological or epistemic terms, there is a possibility that constitutive inferences are not always necessarily truth-preserving, and they may even engender contradiction. Inferences we are disposed to accept may not be truth-preserving. Inferences we are entitled to make may not be truth-preserving. Both Scharp and myself hold that the liar paradox is generated by constitutive principles (together with undeniable truths). It is natural to think that when a concept’s constitutive principles are false or contradictory, the concepts needs to be revised or replaced.35 Haslanger’s referentialism may seem to sit ill with some possible motivations for her proposed ameliorative analysis of “woman”. Mark Richard (2020) summarizes Haslanger’s underlying thought as follows: The, or an important, purpose of the concept woman is to subordinate people on the basis of their (perceived) female properties. We shouldn’t be subordinating people on this basis; indeed, we should be fghting against such subordination. One way to do this is to reformulate the concept so that, so to speak, its noxious purpose is part of its defnition. This will put the purposes for which the concept is actually being used front and center, allowing us to fght gender subordination. We should therefore understand what it is to be a woman as being someone who is systematically subordinated on the basis of (perceived) female properties.36 Richard’s summary seems generally apt to me. But what does it mean for a referentialist to say that a purpose, or function,37 is part of the defnition of a concept? If Haslanger were an inferentialist, at least of the psychological kind, then she could in principle say that it is part of competence with (her revised concept) “woman” to have, or be disposed to have, certain beliefs about oppression; and the facts of oppression are in that way rendered salient by the concept. But given that all there is to the concept expressed by “woman” is its intension there is no way that the descriptive information that Haslanger uses to describe the intension of the concept is itself part of the concept, or such that it is part of competence with the concept to believe or even know of the descriptive information in question.38 Actually, there are complications regarding the connection between concepts and the beliefs of concept users even given inferentialism. On prominent versions of inferentialism, the connection between concept possession and beliefs is complex. If constitutivity is spelled out in epistemic terms  – a principle is constitutive of a concept if competence with the concept entails that one is entitled to believe the principle – then it doesn’t follow from the fact that women are oppressed is constitutive of the concept woman that anyone competent with the concept must believe that women are oppressed. And the most plausible version of the psychological view says only that competence with a concept involves having a disposition to believe the constitutive principles and not that competence involves actually believing these principles. In the background of the discussion of what is the best version of inferentialism are Timothy Williamson’s arguments (see, e.g., his (2007)), to the efect that for every principle that can be regarded as concept-constitutive one can imagine users seemingly competent with the principle who do not believe the principle in question. The proposed versions of inferentialism are intended to get around Williamson’s objections.39 It might be useful to compare Haslanger’s proposal that the concept woman should be revised with the diferent proposal that one should instill belief in the relevant sort of claim, without necessarily revising concepts. What about simply instilling beliefs to the efect that those people with observed or imagined bodily features presumed to be evidence of a female’s biological role in reproduction tend to be systematically subordinated along some dimension (economic, political, legal, social, etc.), they are “marked” by the dominant ideology (in C) as a target for this treatment? This belief can, in principle, be salient independently of what concept one uses “woman” 25

Matti Eklund

or any other label for. An opponent of the strategy of revising concepts can say that one should instead focus directly on beliefs like this. The project of revising concepts is of less importance. Of course, it can be that using Haslanger’s concept can play an important role when it comes to instilling this belief and raising it to salience. It is not clear to me to what extent what I have called attention to is a problem for Haslanger: she may be perfectly fne with the claim that what is of main importance is beliefs and their salience, and concept change is just an indirect, fallible means to that end. Still it is worth stressing that the connection between, on the one hand, concepts used and, on the other, beliefs and their salience is by no means straightforward. Let me also briefy consider how Scharp’s project looks in light of the discussion of referentialism and inferentialism. Scharp’s view is inconsistent with referentialism: since referentialism does not countenance constitutive principles, referentialism does not countenance inconsistent constitutive principles. And as earlier noted, Scharp favors a particular form of inferentialism. Inferentialism in either form discussed allows that there can be false and even inconsistent constitutive principles, and it can be thought that if some constitutive principles for a concept are false  – or even inconsistent – then the concept in question must be revised. Conceptual engineering enters the stage. But the connection is not immediate or obvious. If competence required actual belief in constitutive principles the connection would be at least somewhat stronger: but what is so bad about being entitled to believe constitutive principles, or about merely having dispositions to believe these principles?40

8 Concluding Remarks My aim here has been to summarize some main theoretical issues regarding conceptual engineering. Some of my remarks have been critical. I have doubts about conceptual engineering as an (even near) unifed philosophical project, and I have doubts about the philosophical signifcance of some questions that prominently have come up in connection with conceptual engineering. But doubts about a unifed philosophical project of conceptual engineering are compatible with enthusiasm about individual projects that concern conceptual engineering. Moreover, there is a distinction to be drawn. There is the prescriptive project of determining which concepts actually to employ, for such-and-such purposes, and the evaluative project of evaluating how good certain concepts are along diferent dimensions. Empirical issues arise most centrally in connection with the prescriptive projects. What are the actual consequences of employing this concept instead of that? How feasible is it to ensure that we employ a given concept rather than another? When it comes to the evaluative project, there are diferent kinds of evaluations. Some certainly turn on empirical issues. But many dimensions of evaluation are such that the evaluations need not turn on such issues. The evaluations can be independent of the questions about implementation. Some of the doubts expressed have concerned the prescriptive project specifcally; some more purely evaluative projects emerge relatively unscathed. The doubts should also be seen in light of the fact that in other ways I am very positively inclined toward the trend toward conceptual engineering. What concepts we use is important, and the concepts we have may not be the best concepts along all relevant dimensions or for all relevant purposes. Philosophy focusing only on our actual concepts and the things in the world they stand for threatens to be an unintentionally parochial enterprise.

Notes 1 Nietzsche (1901/68), Section 409. 2 On concept/conception distinctions similar to that drawn here, see, e.g., Rawls (1971), Higginbotham (1998), and Ezcurdia (1998). 3 Brandom (2002), p. 116.


Conceptual Engineering in Philosophy 4 See primarily Haslanger (2000) and (2020a), and Scharp (2013). 5 The expression “conceptual engineering” was apparently frst used by Creath (1991). Other early uses are found in Blackburn (1999) and Brandom (2001). “Ameliorative project” is Haslanger’s expression (in earlier work she used “analytic” rather than “ameliorative”). “Conceptual ethics” is the preferred label of Burgess and Plunkett (2013, 2013a). 6 Haslanger (2000), p. 39. In her more recent work (2020a), Haslanger includes a relativization to contexts. 7 See Haslanger (2006, p. 110) and Haslanger (2012, p. 15). For discussion, see Section 7.6 of Cappelen (2018). 8 Scharp develops the idea in many places, but see primarily his 2013 work. 9 Where presumably one should specify: expressions that are not context-sensitive. Other questions may be asked about this gloss. Should we here include all kinds of subsentential expressions – including things like “on” and “to”? It may sound odd to speak of “the concept to”; on the other hand, “to” makes a distinctive contribution to thoughts and one can evaluate this contribution as well as others, so the conceptual engineering questions arise here too. 10 For example, Cappelen (2018) expresses skepticism about concepts while defending conceptual engineering. 11 In an intuitive sense of “presupposes”. The suggestion is not necessarily that this is presupposition in any of the technical senses standardly employed in philosophy of language and linguistics. 12 Needless to say, these are just brief illustrations. There are alternative views on the examples mentioned. For example, maybe the concept expressed by a slur word is the same as that expressed by the corresponding neutral word: the racism expressed through the use of a racial slur is expressed through other mechanisms than through the concept used. (Maybe – see, e.g., Nunberg (2018) and Bolinger (2017) – the use conveys racism because it is common knowledge that this is the racists’ preferred word.) 13 In addition to what is discussed in the main text, see, e.g., Sterken (2020). 14 Cappelen (2018), p. 123f. 15 Haslanger (2000), p. 47. 16 Above I mentioned, only to set aside, the possibility of diferent individuations of concepts. There are parallel questions about the individuations of propositions. The formulation in the text presupposes that one and the same proposition cannot undergo changes in content. The underlying concern can be expressed even given a diferent individuation of propositions. Wasn’t what we were invested in the proposition with the content it had? (As opposed to: the proposition itself, whatever its content might come to be.) 17 Haslanger does stress that the question of the usefulness of the concept she describes and the question of whether to use “woman” for that concept are independent questions. See, e.g., Haslanger (2000, p. 47) and (2020a). 18 Again (see previous footnote) the formulation in the text presupposes that one and the same proposition cannot undergo changes in content. Again the underlying concern can be expressed even given a diferent individuation of propositions. 19 Plunkett and Cappelen (2020). 20 Scharp (2013), p. 46. 21 If we focus on the ideal scenario kind of case, another concern arises. Ought we not, at the ideal stage, to employ all concepts? At the end of inquiry we will want to have cataloged all truths, so if there are any truths at all that we fail to include, even fairly useless ones, we have not done as well as we could. In an idealized scenario we do not need to choose which among all the truths to focus on. Dever (2020) argues for “conceptual maximalism”, the view that we should employ all concepts; and, relatedly, for the view that we should work in “all languages”. He realizes that we of course cannot do everything at once, so that there will be choices about what to focus on here and now. The way I see it, his global maximalism is then a view about a kind of ideal theorizing. 22 See, e.g., Rabinowicz and Rønnow-Rasmussen (2004). 23 For discussion of Simion’s argument, see Catapang Podosky (forthcoming). 24 So called because it arguably echoes themes from Strawson’s (1963) discussion of Carnap. 25 Cappelen (2018), p. 100. 26 Ludlow (2005), p. 13; quoted in Cappelen (2018), p. 98. 27 In fact, even Strawson himself says, [I]t seems prima facie evident that to ofer formal explanations [i.e. Carnap-style explications] of key terms of scientifc theories to one who seeks philosophical illumination of essential concepts of non-scientifc discourse, is to do something utterly irrelevant – is a sheer misunderstanding. (1963, p. 505)


Matti Eklund


29 30 31 32 33

Notice the relativization: the point essentially concerns what serves the purposes of someone who “seeks philosophical illumination of essential concepts of non-scientifc discourse”. What if one simply does not have that aim? Cappelen (2018, p. 82) also points out that problems arise even given internalism. We need not have control over what concepts we employ even given internalism. As in the case of externalism, it is an empirical matter to what extent we do have such control. It is not a given that we have control over the internal factors that the internalist focuses on. See Burgess and Plunkett (2013a), Burgess (2014), and Burgess (2020). Chalmers (2011), p. 543. Cappelen (2018), p. 194. See, e.g., Eklund (2015), (2017) and (2020). See, e.g., Haslanger (2020a) and Cappelen (2018). Haslanger emphasizes that the content of a concept is a “partition of logical space”, and genuine conceptual revision amounts to changing what partition of logical space the term or concept represents (p. 242). (Haslanger contrasts genuine revision – what she calls informational or semantic amelioration – with epistemic amelioration, where we simply “improve our understanding of the informational content of the concept” (p. 242).) She emphasizes: There are no “core commitments” associated with words that cannot be overturned or negotiated. Although in some sense we represent the world – propositions are abstract entities that carry information and are, to that extent, representational – the “mode of representation” is not part of the informational content of what we say and think. (p. 238)

34 35

36 37 38



In her (2020b), Haslanger speaks of concepts as “clusters of dispositions” (p. 239), and distinguishes concepts from their contents. But at the same time, she – both in that work and other works – expresses skepticism about the idea that all users of a given concept have the same dispositions. For example, in her (2020b), she says, “possession of the concept may occur by virtue of diferent cognitive mechanisms and give rise to very diferent dispositions in diferent individuals” (p. 236). The claims seem hard to square with each other. It is tempting to assume that her real view is only that to have some suitable cluster of dispositions (and this is also how she puts it in her (2020a). But maybe the idea is that even if diferent individuals possessing the concept have diferent dispositions, “cluster of dispositions” is so general that the individuals count as possessing the same disposition cluster: that would make consistent sense of how Haslanger puts things in diferent recent papers. Despite possible complications, I will speak of Haslanger as a referentialist, reserving further discussion of her precise view to footnotes. I use Williamson’s terminology because I fnd it useful, and it has caught on. However, the label “inferentialism” is not perfect. In the relevant sense, Frege is supposed to be an inferentialist: Frege’s senses cut more fnely that reference. But Frege does not strictly focus on inferences. And for anyone who thinks this it might also be natural to go further and hold that whenever a concept’s constitutive principles are immodest in the sense that they require something substantive of reality then the concept needs to be revised or replaced, even if what is required happens to be true. The supposed thought would be that substantive matters should not be settled conceptually. (Naturally, there is a question of how best to understand “substantive”.) Richard (2020), p. 359. “Function” is arguably better in the context than Richard’s “purpose”. Haslanger (2020b) says that concepts are dispositions (see above, fn33), and given that view there is more to play with. Descriptive information can then be associated with the concept, by virtue of being the descriptive information that it is part of competence with the concept to accept. But even in her (2020b), Haslanger repeats her point that there are no “core commitments” associated with words. When Haslanger says that there are no “core commitments” associated with words, what she holds is that there are no commitments that “cannot be overturned or negotiated”. An inferentialist of the kind we are now considering can cheerfully agree. In the context of her discussion, Haslanger’s remark still seems to support a referentialist interpretation, since the alternative, positive picture she ofers is one where “our utterances and our mental states do not have senses or concepts as their content “and what we believe are propositions, understood “as functions from possible circumstances to truth values, or equivalently, as sets of possible situations”. (Haslanger (2020a), p. 238; the latter passage Haslanger quotes from Stalnaker (1998), p. 343.) These remarks are also relevant to the issue raised in fn35 of whether it is a defect in a concept if its constitutive principles require something substantive of reality. The intuitive argument against this stated


Conceptual Engineering in Philosophy above was that we might not want substantive matters to be settled conceptually. But given what is now stressed regarding constitutivity we can see that the talk of what is “settled” is misguided. Even if a principle is constitutive in the sense at issue, a thinker possessing the concept can come to reject it as untrue.

References Blackburn, Simon: 1999, Think: A Compelling Introduction to Philosophy, Oxford University Press. Bolinger, Renee: 2017, “The Pragmatics of Slurs”, Noûs 51: 439–62. Brandom, Robert: 2001, “Modality, Normativity, and Intentionality”, Philosophy and Phenomenological Research 63: 611–23. Brandom, Robert: 2002, Tales of the Mighty Dead, Harvard University Press. Burgess, Alexis: 2014, “Keeping ‘True’: A Case Study in Conceptual Ethics”, Inquiry 57: 580–606. Burgess, Alexis: 2020, “Never Say ‘Never Say ‘Never’’”, in Burgess, Cappelen and Plunkett (2020), pp. 125–31. Burgess, Alexis and David Plunkett: 2013, “Conceptual Ethics I”, Philosophy Compass 8: 1091–101. Burgess, Alexis and David Plunkett: 2013a, “Conceptual Ethics II”, Philosophy Compass 8: 1102–110. Burgess, Alexis, Herman Cappelen and David Plunkett (eds.): 2020, Conceptual Engineering and Conceptual Ethics, Oxford University Press. Cappelen, Herman: 2018, Fixing Language, Oxford University Press. Carnap, Rudolf: 1950, The Logical Foundations of Probability, University of Chicago Press. Catapang Podosky, Paul-Mikhail: forthcoming, “Ideology and Normativity: Constraints on Conceptual Engineering”, Inquiry. Chalmers, David: 2011, “Verbal Disputes”, Philosophical Review 120: 515–66. Creath, Richard (ed.): 1991, Dear Carnap, Dear Van, University of California Press. Dever, Josh: 2020, “Preliminary Scouting Reports from the Outer Limits of Conceptual Engineering”, in Burgess, Cappelen and Plunkett (2020), pp. 152–69. Dutilh Novaes, Catarina: 2020, “Carnapian Explication and Ameliorative Analysis: A Systematic Comparison”, Synthese 197: 1011–34. Eklund, Matti: 2015, “Intuitions, Conceptual Engineering, and Conceptual Fixed Points”, in Chris Daly (ed.), The Palgrave Handbook of Philosophical Methods, Palgrave Macmillan, pp. 363–85. Eklund, Matti: 2017, Choosing Normative Concepts, Oxford University Press. Eklund, Matti: 2020, “Variance Theses in Ontology and Metaethics”, in Burgess, Cappelen and Plunkett (2020), pp. 187–204. Ezcurdia, Maite: 1998, “The Concept-Conception Distinction”, Philosophical Issues 9, Concepts: 187–92. Haslanger, Sally: 2000, “Gender and Race: (What) Are They? (What) Do We Want Them To Be?”, Noûs 34: 31–55. Haslanger, Sally: 2006, “What Good Are Our Intuitions? Philosophical Analysis and Social Kinds”, Aristotelian Society Supplementary Volume 80: 89–118. Haslanger, Sally: 2012, Resisting Reality: Social Construction and Social Critique, Oxford University Press. Haslanger, Sally: 2020a, “Going On, Not in the Same Way”, in Burgess, Cappelen and Plunkett (2020), pp. 230–57. Haslanger, Sally: (2020b), “How Not to Change the Subject”, in Teresa Marques and Åsa Wikforss (eds.), Shifting Concepts, Oxford University Press, pp. 235–59. Higginbotham, James: 1998, “Conceptual Competence”, Philosophical Issues 9, Concepts: 149–62. Jenkins, Katharine: 2016, “Amelioration and Inclusion: Gender Identity and the Concept of Woman”, Ethics 126: 394–421. Koch, Stefen: forthcoming, “The Externalist Challenge to Conceptual Engineering”, Synthese 198: 327–48. Ludlow, Peter: 2005, “Contextualism and the New Linguistic Turn in Epistemology”, in Gerhard Preyer and Georg Peter (eds.), Contextualism in Philosophy: Knowledge, Meaning, and Truth, Oxford University Press, pp. 11–50. Nietzsche, Friedrich. (1901/68). The Will to Power. Translated by Walter Kaufmann, Random House. Nunberg, Geofrey: 2018, “The Social Life of Slurs”, in Daniel Fogal, Daniel Harris and Matt Moss (eds.), New Work on Speech Acts, Oxford University Press, pp. 237–94. Plunkett, David and Herman Cappelen: 2020, “Conceptual Engineering and Conceptual Ethics: An Introduction”, in Burgess, Cappelen and Plunkett (2020), pp. 1–34. Prinzing, Michael: 2018, “The Revisionist’s Rubric: Conceptual Engineering and the Discontinuity Objection”, Inquiry 61: 854–80. Quine, W.V.: 1960, Word and Object, MIT Press.


Matti Eklund Rabinowicz, Wlodek and Toni Rønnow-Rasmussen: 2004, “The Strike of the Demon: On Fitting ProAttitudes and Value”, Ethics 114: 391–423. Rawls, John: 1971, A Theory of Justice, Harvard University Press. Richard, Mark: 2020, “The A-Project and the B-Project”, in Burgess, Cappelen and Plunkett (2020), pp. 358–78. Scharp, Kevin: 2013, Replacing Truth, Oxford University Press. Simion, Mona: 2018, “The ‘Should’ in Conceptual Engineering”, Inquiry 61: 914–28. Stalnaker, Robert: 1998, “What Might Nonconceptual Content Be?”, Philosophical Issues 9: 339–52. Sterken, Rachel: 2020, “Linguistic Intervention and Meaning Change: Transformative Communicative Disruptions”, in Burgess, Cappelen and Plunkett (2020), pp. 417–33. Strawson, P.F.: 1963, “Carnap’s Views on Conceptual Systems versus Natural Languages in Analytic Philosophy”, in Paul Arthur Schilpp (ed.), The Philosophy of Rudolf Carnap, Open Court, pp. 503–18. Williamson, Timothy: 2007, The Philosophy of Philosophy, Blackwell. Williamson, Timothy: 2009, “Reference, Inference, and the Semantics of Pejoratives”, in Joseph Almog and Paolo Leonardi (eds.), The Philosophy of David Kaplan. Oxford University Press, pp. 137–59.



1 Introduction Social ontology is the philosophical study of the nature and properties of the social world. As a feld of philosophy, it investigates the nature of social groups and collective intentionality, as well as the nature and existence of money, corporations, institutions, race, gender, artifacts, and the law (among other things). Many paradigm objects that social ontology investigates are said to be socially constructed in some sense. Bluntly put: they are said to exist and have the natures that they do by virtue of what we – social agents – ascribe and attribute to those objects. For instance, many feminist philosophers take gender to be constitutively constructed: in defning what it is (for example) to be a woman, we must make reference to social factors rather than anatomy, such as one’s position in a social hierarchy or self-identifcation. There are many ways in which social construction works and how the social world ontologically depends on us. Prima facie it seems that mental states (both individual and collective) play a role in setting up social kinds or providing the ‘glue’ necessary to bind entities into social kinds. This may include forms of collective agreement and institutionalized social conventions. Alternatively, some hold that social kinds, entities, and practices are produced and reproduced in virtue of some functions that they putatively serve. For instance, a functional explanation of a social practice explains why that practice exists by making reference to the purpose or needs served by the practice (Kincaid 2006). In many social ontological accounts, language plays a seemingly key role in explicating how social kinds come into being and are set up. For instance, crudely put, in defning the term ‘landlord’ as “a person who rents a property owned”, we set up the conditions under which someone counts as a landlord – that is, with ‘our’ conceptual schemes and linguistic usage we fx what it is to be a landlord. Although much of social ontology takes language somehow to be central to how the social world is set up, surprisingly little has been written about the precise role of language in social ontology and how we should understand the constructive force of language. Here I will address this issue. More specifcally, although the idea that social categories ontologically depend on language in some sense is widely accepted, the precise explication of how social objects metaphysically depend on language has received much less focused attention – and it is this issue that I aim to address here. In order to do so, we must distinguish two more detailed questions: (1) How is it that language fgures in explicating the existence of social entities? (2) How is it that language shapes the nature of social entities? This chapter ofers some tentative answers to these questions hence providing a clarifcation of the connection between language and sociality. Doing so will 31

Mari Mikkola

hopefully advance our understanding of philosophical methods that social ontologists can fruitfully make use of. In what follows, I will largely focus on feminist discussions in social ontology that deal with gender terms and kinds by way of illustration.

2 Te Existence of Social Kinds Prima facie the idea that the existence of social entities and kinds depends on ‘our’ conceptual schemes, language use, or ‘discourse’ is puzzling: how is it that we can (bluntly put) call things into being? Just saying that something is the case does not make it the case. For instance, consider the example of being a landlord. Just stipulating the conditions of landlordy-ness does not make it the case that landlords magically come into being. Radical linguistic constructivism – the view that everything is socially constructed all the way down via language or discourse – looks highly implausible and questionable. Instead, it seems that the role language does and ought to play is more descriptive: it is a tool with which to articulate and make explicit ways in which reality is arranged thereby providing apt and accurate depictions of it. Consider this analogously to the task of (non-social) metaphysics and ontology: on some prominent contemporary accounts, it is about limning the structure of reality by “fguring out the right categories [carved in reality’s joints] for describing the world” (Sider 2011, 1). The role of language is to articulate how we discover the world to be, not to construct it to our liking. However, much of social ontology takes language (broadly understood) to have precisely such a constructive role. How should we understand this? I want to suggest that there may be a rather easy way to spell this out, contra frst appearances, at least relative to existence claims pertaining to paradigm social entities. But in order to see why the existence of social categories such as gender might be disputed, consider frst some postmodern critiques and how to deal with the term ‘woman’. It is a widely accepted feminist claim that gender injustice is not incidental and individual, but systematic and structural – it targets women as women. Feminism thus seemingly lends itself to identity politics: a form of political mobilization based on membership in women’s social kind, where shared experiences or traits delimit kind membership (Heyes 2000, 2012). However, the past few decades have allegedly witnessed a feminist identity crisis (Alcof 1988). Feminist politics presumes the existence of a women’s social kind founded on some category-wide common traits or experiences. But as many feminists have noted, no such transcultural and transhistorical commonality exists because our axes of identity (e.g. gender, race, ability, class) are not discrete and separable. In her classic work, Gender Trouble, Judith Butler takes issue with feminist identity politics (to name but one issue in a rich work). It appears as if the term ‘woman’ has some unitary cross-cultural and transhistorical meaning and that the term picks out some determinate group of people with an identity-defning feature in common. However, this picture is mistaken and the concept woman has no stable meaning. Instead, ‘woman’ is “a term in process, a becoming, a constructing that cannot rightfully be said to originate or end. As an ongoing discursive practice, it is open to intervention and resignifcation” (Butler 1999, 43). (For a similar argument that rejects ‘woman’ has some fxed and invariant meaning, see Cornell 1993.) The feminist picture of gender (for Butler) in no meaningful sense describes reality; rather, it is an unwitting product of feminist politics in its efort to represent the interests of certain political subjects (namely, of women). In aiming to represent women’s interests, feminism constructs its own political subject via accepted conceptual schemes and linguistic ‘discourse’. Hence, any notion of womanhood that is used to capture the class of women unhelpfully masks women’s diversity. It “necessarily produce[s] factionalization…Identity categories are never merely descriptive, but always normative, and as such, exclusionary” (Butler 1991, 160). Gender concepts articulated by feminist theorists turn out to articulate a set of “unspoken


Social Ontology

normative requirements” (Butler 1999, 9) that those hoping to gain feminist political representation should satisfy. Thus, they prescribe and produce a supposedly correct picture of womanhood despite aiming to ofer non-normative descriptions of womanhood – if you like, aiming to limn the structure of our gendered reality. Butler takes this to be a feature of terms denoting social identity categories. The underlying presumption appears to be that such terms can never be used in a non-ideological way (Moi 1999, 43). They will always prescribe some conditions that ought to be satisfed since all processes of drawing categorical distinctions involve normative commitments that involve the exercise of social power (Witt 1995). Those who do not conform to the normative picture of womanhood risk being alienated and excluded from feminist politics altogether. Nicholson captures this thought nicely: “the idea of [the term] ‘woman’ as unitary operates as a policing force which generates and legitimizes certain practices, experiences, etc., and curtails and delegitimizes others” (1998, 293). Butler’s aim is not, however, merely to critique prevalent conceptions of woman. Her argument is stronger than this: every defnition of woman will be insidiously normative and thus politically problematic. This is because, as noted above, terms denoting social identity categories cannot be used in non-normative and merely descriptive ways according to Butler. The mistake is not that feminists provided the incorrect defnition of woman. Rather, their mistake was to attempt to defne womanhood in a putatively descriptive manner to begin with – a task that Butler takes to be linguistically impossible since normativity is (for her) built into terms denoting identity categories. In fact, Butler thinks it is also metaphysically impossible to provide a defnition of woman because (for her) gender is performative. It is not “a stable identity or locus of agency from which various acts follow”; rather, gender comes into being through “a stylized repetition of [habitual] acts” (Butler 1999, 179). Gender is something that one does in wearing gender-coded clothing, in walking and sitting in gender-coded ways, in styling one’s hair in gender-coded manner and in desiring sexually the opposite sex/gender. Repeatedly engaging in ‘feminizing’ and ‘masculinizing’ acts congeals gender, thereby making people falsely think of gender as an identity that they somehow ‘naturally’ possess. This opens up the possibility to undermine gender dualism by subverting the way one ‘does’ one’s gender. Subsequently, feminists should actively resist defning womanhood, thereby opening it up for new, more emancipatory conceptions. Moreover, there is no ‘doer’ behind the deed as Butler puts it. Strictly speaking, there are no women or men; only habitually performed gendering acts.1 To think that gender is an identity that we possess and something that we are (rather something that we do) is a view that earlier feminists simply got wrong and it created a problematically exclusionary essentialism about gender. This type of challenge to the existence of gender has been immensely infuential in feminist theory. Until the early- to mid-2000s it was not uncommon to hear philosophers proclaim that gender isn’t real or to be asked as a feminist philosopher whether one thought gender really exists. To challenge this type of gender skepticism, or error theory about gender, feminist philosophers typically argue that manifest gender diferences are real in the sense of being part of our social realities, though not carved in nature’s joints. One task of feminist social ontology, then, is to show that classifcations central to feminist concerns are in important ways constitutively constructed (Haslanger 1995): for instance, in defning womanhood we must make reference to social factors, not anatomy. Of course, which social factors fx gender is an ongoing debate within feminist philosophy (for an overview, see Mikkola 2017). Alternatively, one could appeal to a form of causal construction, discursive construction: gendered beings are the way they are, at least to a substantial extent, because of what is attributed to them on the basis of ‘our’ underlying gendered conceptual schemes (for more on discursive construction, see Haslanger 1995, 99). In fact, this is close to how Butler understands the social constructedness of sex. In addition to arguing against identity


Mari Mikkola

politics and for gender performativity, Butler holds that distinguishing biological sex from social gender is unintelligible. For her, both are socially constructed: If the immutable character of sex is contested, perhaps this construct called ‘sex’ is as culturally constructed as gender; indeed, perhaps it was always already gender, with the consequence that the distinction between sex and gender turns out to be no distinction at all. (Butler 1999, 10–11) The idea that sex is a social construct, for Butler, boils down to the view that our sexed bodies are also performative and, so, they have “no ontological status apart from the various acts which constitute [their] reality” (1999, 173). Prima facie, this implausibly implies that sexed bodies do not have independent existence and that if gendering activities ceased, so would physical bodies. This is not Butler’s claim; rather, her position is that bodies viewed as the material foundations on which gender is constructed are themselves constructed as if they provide such material foundations (Butler 1993). Cultural conceptions about gender fgure in “the very apparatus of production whereby sexes themselves are established” (Butler 1999, 11). For Butler, sexed bodies never exist outside of social meanings and how we understand gender shapes how we understand sex (1999, 139). Sexed bodies are not empty matter on which gender is constructed and sex categories are not picked out on the basis of objective features of the world. Instead, our sexed bodies are discursively constructed: they are the way they are, at least to a substantial extent, because of what is attributed to sexed bodies and how they are classifed. Sex assignment – calling someone female or male – is normative (Butler 1993, 1). As Butler sees it, when a doctor proclaims at birth ‘It’s a boy/ girl’, they are not making a descriptive claim but a normative pronouncement with illocutionary, performative force. In efect, the doctor’s utterance makes infants into girls or boys – it counts as an instance of sexing. This is a normative activity in the sense of imposing a (normative) order and framework at birth onto children. So the proclamation does not report or describe some existing state of afairs, but rather constructs a state of afairs, because certain norms come to be ascribed to and imposed on people. We, then, engage in activities that make it seem as if sexes naturally come in two and that being female or male is an objective feature of reality, rather than being a consequence of certain constitutive acts (that is, rather than being performative). And this is what Butler means in saying that physical bodies never exist outside cultural and social meanings, and that sex is as socially constructed as gender. She does not deny that physical bodies exist in a material sense. But, she takes our understanding of this existence to be a product of social conditioning: social conditioning makes the existence of physical bodies intelligible to us by discursively constructing sexed bodies through certain constitutive acts, like illocutionary ‘sexing’ speech acts. The above elaborated some prominent social constructivist accounts of sex and gender, where language plays a performative role. In other words, some social agents have the power to bring social entities and states of afairs into being with language (for instance) given their institutional roles; just think of the doctor’s power to ‘sex’ new-born babies that positions them in gendered structures with wide-ranging and long-term efects on self-identities, self-presentations, and available social roles.2 Discursive construction works in similar ways in that ‘our’ classifcatory schemes may do more than just map preexisting [sic] groups of individuals; rather our attributions have the power to both establish and reinforce groupings which may eventually come to ‘ft’ the classifcation. In such cases, classifcatory schemes function more like a script than a map. (Haslanger 1995, 99) These sorts of performative accounts, however, are on the face of it difcult to square with the idea that language can more or less aptly represent or describe social reality. In a sense, it becomes 34

Social Ontology

difcult to critique the doctor’s proclamation that sexes infants as being simply false, if in saying that x is the case the doctor is able to make x the case. The worry with social ontological accounts that aford language too much constructive power is that it becomes impossible to challenge some views as being simply mistaken. Although language use and conceptual schemes conceivably have much performative power, social ontologists should retain some more descriptive and declarative role for language as well. With this in mind, there may be a more straightforward alternative way to fx the existence of many social entities through language. I have in mind a more recent idea of ‘easy ontology’ as articulated by Amie Thomasson. The basic point is that fully meaningful and well-specifed existence questions are straightforwardly answerable by “making use of our conceptual competence” (Thomasson 2015, 20). On this approach, Answers to certain disputed ontological questions can be reached easily by starting from an uncontroversial truth (e.g., ‘the cups and saucers are equinumerous’ or ‘snow is white’) and reasoning by what seem like trivial steps (to ‘the number of cups equals the number of saucers’ or ‘the proposition that snow is white is true’) to reach ontological conclusions (‘there are numbers’; ‘there are propositions’). (Thomasson 2015, 21) This allows us to resolve existence questions of the kind ‘Do numbers exist?’ conceptually by starting with a conceptual truth and making some easy inferences. The role of philosophy is to engage in conceptual analysis and to make explicit how conceptual truths license inferences, where this type of analysis involves removing conceptual confusion and proposing discursive revisions. The background framework for this type of analysis comes from Carnap’s view that there are two ways to ask existence questions. Internal question are about whether certain entities or kinds exist within ‘our’ language or linguistic framework. This includes questions such as ‘Are there any prime numbers between 780 and 790?’, which can be answered analytically given ‘our’ linguistic framework – they can be answered using the rules of use for number terms within the framework of arithmetic. External existence questions by contrast are formulated in highly general ways, such as ‘Are there numbers?’. Questions of this form, however, are ill-formed pseudo-questions and rather should be reinterpreted as implicitly pragmatic questions about which linguistic framework to accept as relevant: “we have to make the choice whether or not to accept and use the forms of expression in the framework in question” (Carnap 1956, 207). To put the point somewhat diferently, internal questions are asked within some relevant framework, whereas external questions pertain to “the existence or reality of the system of entities as a whole” (Carnap 1956, 206). Subsequently, answering external questions hinges on pragmatic considerations about which linguistic framework we ought to adopt (if any). Philosophers’ work consists in “constructing linguistic frameworks and making practical decisions about which framework to adopt for which purposes” (Thomasson 2015, 33), rather than undertaking an examination that supposedly afords them with some special ontological insight into reality. To sum up: internal existence questions may be answered analytically, while external existence questions – in being ill-formed pseudo-questions – can be made intelligible only if we understand them pragmatically as being questions about what linguistic framework we ought to adopt. If metaphysical questions are considered to be ‘hard’ (rather than easily resolvable by trivial analytical means), they are to be considered external questions. But then we should understand metaphysical existence questions as pragmatic questions if they are to be sensible at all. With respect to gender, these considerations yield the following. Depending on how we frame our existence questions about gender, they may be either internal, in which case we can answer them easily. For instance, start with the uncontroversial truth ‘Women are human beings’ and the trivial inference ‘The 35

Mari Mikkola

proposition that women are human beings is true’. This then on the ‘easy’ approach yields an ontological conclusion: there are women. We may be able to deal with many existence questions in social ontology in this manner, which demystifes them. Alternatively, existence questions about gender may be treated as external, in which case those questions will be ill-formed pseudo-questions. Given the Carnapian distinction, a question of the form ‘Are there women?’ will be an external one and hence a pseudo-question. In order to render the question sensible, we ought to conclude that this question cannot sensibly be answered by purely metaphysical considerations and rather is a pragmatic question about which linguistic framework (if any) to adopt. In this case, resolving the existence question about gender turns on a choice between diferent conceptual schemes and linguistic frameworks – for instance, given feminist political considerations we have prima facie good grounds to reject frameworks advanced by radical linguistic constructivists, which end up being skeptical about the existence or ‘materiality’ of gender. After all, material gendered conditions that are oppressive to women are not linguistic constructs, and politically efective feminism cannot alter those conditions merely by afecting extant conceptual schemes or ‘discourse’. If we understand existence questions as external, answering them hinges on pragmatic considerations about which linguistic framework is the most fruitful one. In order to engage in this type of inquiry, social ontologists need methods that can explicate the nature of social kinds. In other words, the ‘easy’ gender ontology tells us nothing about what women (or other social entities) are like – it just tells us that they exist. This conclusion undercuts the skeptical move of holding that since social entities are not ‘carved in nature’s joints’ and because they are essentially derivative, they do not really exist or they exist in some mode diferent to (say) numbers. ‘Easy’ ontology ofers a way for social ontologists to resist these moves; but in order to say something about the nature of social entities the easy approach cannot help us further. For this task, I will turn next to metalinguistic negotiation.

3 Te Nature of Social Kinds Social ontology typically accepts that conceptual schemes or ‘discourse’ can shape the social world in non-trivial and signifcant ways. Language and our linguistic repertoire are important parts of the social world. Social power relations shape meanings, how we (can) think and talk, and prevalent conceptual schemes, which, in turn, afect the way social entities and kinds are set up. Some ways to think and talk about reality are clearly defective. These include: cognitive defects (that undermine our ability to reason properly), moral or political defects (that undermine moral or political values of various sorts), theoretical defects (that undermine progress within some theoretical feld), or semantic defects (where the semantic value is incoherent, incomplete, or missing). (Cappelen and Plunkett 2020, 3; see also Cappelen 2018) For instance, take the concept of family: traditionally conceived, the content of this concept includes a mother, a father, heterosexual marriage contract, and children who are biologically related to the parents. But this conception is defective at least in cognitive, political, and semantic ways and generates a failure to recognize less traditional arrangements: e.g. same-sex unions, adoptive parents, parenting one’s partner’s biological children, non-married couples, childfree unions, etc. In order to improve our linguistic resources and repertoire to better (so to speak) limn the structure of the social reality, metalinguistic negotiation has more recently been advanced as an important methodological tool that focuses on the pragmatics of what is being said. Most basically, (normative) metalinguistic negotiation denotes a dispute “in which speakers each use (rather than mention) a term to advocate for a normative view about how that term should be used” (Plunkett 2015, 832). 36

Social Ontology

This is not equivalent to speakers merely talking past one another. In such disputes the interlocutors are not genuinely expressing a disagreement, unlike in cases of metalinguistic negotiation. A simple case of equivocation is where speakers use the same term to refer to two diferent concepts (e.g. ‘bank’ to denote a river bank and a fnancial institution). Another case that isn’t metalinguistic negotiation in the normative sense would be a dispute about whether someone counts as tall. For instance, given the national height averages, Sally who is 170 cm might be tall for a Finn but short for a Dutch. However, if Jo proclaims “Sally is tall” and Jane responds claiming “Sally is not tall”, we do not have a normative metalinguistic disagreement about how one ought to employ the term ‘tall’ – rather, for Plunkett (2015) this would be a descriptive metalinguistic dispute and issue about word usage or meaning. However, metalinguistic negotiations proper are not descriptive disputes. They are about what a word should mean, or how it should be used (Plunkett 2015, 838). Or as Thomasson puts it: “unlike paradigmatic verbal disputes, many cases of metalinguistic negotiation involve disputes very much worth having” (2017, 12). By way of example, she notes the following (among others): whether the Oklahoma City bombing counts as terrorism, whether waterboarding counts as torture, and whether certain extreme performances count as art. The point here is that we are not merely engaging in disputes about the terms ‘terrorism’, ‘torture’, or ‘art’. These are worldly disputes, where the interlocutors are not merely ofering competing descriptions of the world, but are rather “advocating for diferent conceptual schemes with diferent impact on our way of life” (Thomasson 2017, 20). Thinking back to easy ontology then: when we are dealing with external existence questions that essentially hinge on pragmatically choosing a conceptual scheme, we are engaged in pragmatically advocating for one conceptual scheme that refects some way of life over another scheme and way of life. Such a debate is not merely a superfcial verbal one, but a substantive dispute. This is because normative issues about concept choices being important do not “depend on the way in which speakers argue about those issues. Rather, it depends on the content of those issues” (Plunkett 2015, 844). In order to address the issue of which conceptual schemes we ought to advance, social ontologists could engage in ‘conceptual engineering’ and ‘conceptual ethics’ (Burgess and Plunkett 2013; Burgess, Cappelen, and Plunkett 2020). These projects can be characterized as follows: Conceptual engineering is the business of changing existing concepts and devising new ones. Conceptual ethics is the business of evaluating existing concepts and ways of talking, along with newly engineered ones, and making normative judgments as to whether they are ft for purpose. Jointly they promise reform through innovation and selection of how we think and talk in the pursuit of various ends – whether moral ends, practical ends, or alethic ends. (Braddon-Mitchell 2020, 79) The idea is to engage in some conceptual innovation so that we can meaningfully revise what we mean in order to gain better tools with which to think and talk. Conceptual ethics is not just about how we should understand some concept, but also about whether we should use some concept at all: whether we should be eliminativists or nihilists about some concept. If our conceptual repertoire aims to track some non-existent entities, we should no longer make use of that repertoire (nihilism). This might be the case with gender terms if approaches like Butler’s are to be preferred. But as I suggested above, ‘easy’ ontology ofers prima facie ways to dispel nihilism in the social realm, at least relative to gender. Then again, even if our conceptual repertoire putatively tracks something that exists, we may have good normative grounds to stop talking in those terms and be eliminativists. Within feminist social ontology, Sally Haslanger’s ameliorative approach to gender is conceivably the most well-known putative instance of conceptual engineering. Haslanger (2012) spells out diferent ways to answer questions of the kind ‘What is x?’. First, a conceptual analysis aims 37

Mari Mikkola

to articulate ‘our’ ordinary concepts by consulting native speakers’ intuitions – this spells out ‘our’ manifest concepts. This type is introspective in character and can prima facie be done from the philosopher’s armchair. Second, a descriptive approach focuses on our terms’ extensions, and which kinds our language use tracks – it aims to clarify ‘our’ operative concepts by looking at how language users in fact use certain terms, which should tell us something about their intension or meaning. This type of analysis should be conducted typically via empirical means. Neither conceptual nor descriptive projects, however, tackle more normative issues: how one ought to think and talk; and what role the speaker’s social position plays in generating meanings about the social world. This is what amelioration aims to do. It is about elucidating ‘our’ legitimate purposes and what concept of F-ness (if any) would serve them best – to explicate our target concepts. For instance, if our linguistic repertoire contains a morally and politically defective notion of family (one that fails to acknowledge non-traditional arrangements), amelioration has the aim of revising that notion in a manner that does away with the defects. In this case, it would involve engineering and tweaking the constitutive conditions of what counts as a family in a way that does capture non-traditional arrangements as well. Now, relative to gender straightforward conceptual and descriptive analyses of woman are insufcient in being theoretically unhelpful. Both approaches aim to explain, articulate, and refne our ordinary woman-concept. But either everyday gender vocabulary falsely encodes anatomical features (as with the manifest concept) or it is not specifc enough to do the necessary theoretical work (as with the operative concept): ordinary gender talk is too vague and idiosyncratic to be theoretically and politically helpful. Moreover, ordinary speakers tend to be confused about the conditions that make someone a woman. And so, they do not seem to have any introspective privileged access to the content of gender concepts. These problems show the ordinary woman-concept to be defective (at least) semantically, theoretically, and politically. In order to overcome the defects, Haslanger advocates for her ameliorative approach to gender. Of course this sort of project generates many difcult questions about how we decide which conceptual choices to appropriate and endorse: in other words, how do we evaluate which concepts are the most ftting for our purposes. Unfortunately, there are few easy and straightforward answers here. In one sense, the situation does not look so complicated. Given that amelioration is a goal-oriented activity, we appropriate the conceptual tools that best serve our goals. But there may be various aims we might want the same conceptual apparatus to satisfy: for instance, in choosing which conception of family to endorse we may have in mind moral/ political goals (e.g. to recognize the rights of adoptive parents), as well as theoretical ones (e.g. to help forge good social scientifc theories that can usefully be employed in policy making). These goals may confict. Perhaps the traditional conception is the most apt on some theoretical grounds giving us a simple and straightforward way to device laws and policies. This generates disputes about the legitimacy and primacy of our goals, which are also enormously difcult to settle (and something I cannot settle here). As Thomasson (2020, 440) puts it, conceptual ethics works on two levels. First, we ask what function our concepts should serve relative to some goals, where our goals are fxed. Second, we can ask deeper questions about what goals we ought to adopt in order to decide what concepts to employ, all things considered. At both levels, empirical realities and people’s lived experiences conceivably play important roles (among others). For instance, think of same-sex couples in longterm partnerships being denied hospital visitation rights on the grounds that they are supposedly not family. In order to ameliorate our conception of family, philosophy of language and linguistic intuitions can only go so far; what also matters are personal stories of how ‘our’ social reality arranged according to some linguistic resources (rather than others) harms and hurts us, or unduly constrains and limits our lives. Those who live (so to speak) traditionally gendered lives that conform to typical normative expectations may not realize how lives of others are unduly constrained by prevalent conceptual schemes and linguistic resources – and they may do so without any ill 38

Social Ontology

will or prejudice. Consciousness raising and activism may be necessary in order to engineer and ameliorate ‘our’ ways of thinking and speaking. Clearly there is no guarantee that social actors will reach a consensus is any meaningful sense. We may have to agree to disagree if faced with ‘bad faith’ interlocutors, who are simply unwilling to entertain the legitimacy of diferent ways of life. Or we may be forced to endorse a plurality of (say) family-conceptions for diferent goals and aims if no consensus is forthcoming (for such “strategic conceptual engineering”, see Brigandt and Rosario 2020). Ethics is hard, and unfortunately conceptual ethics is no diferent. That said, despite its clear normative and pragmatic focus, Diaz-Leon (2020) has recently argued for two possible ways to understand amelioration, where one is more descriptive. First, where a term determinately refers to an entity but we ought to change the referent; second, where a term’s referent is indeterminate but it should become determinate, given normative considerations. Many projects usually taken to be purely descriptive are actually ameliorative or contain important ameliorative elements (i.e. the former sorts of projects). Hence, Diaz-Leon holds that normative considerations are not only pertinent to ameliorative projects as traditionally conceived, but also to descriptive projects. This relates back to the discussion above in couple of ways. First, consider how to deal with existence questions that hinge on pragmatic considerations about which linguistic framework is the most fruitful one. A descriptive analysis may deliver multiple frameworks with which to elucidate what there is and how gender might be real. However, in order to decide among those frameworks, we will have to rely on normative considerations that are to do with signifcance and relevance given the questions we are aiming to answer and the political purposes that we are hoping to achieve. Second, consider the normative disagreements alluded to above. Settling conceptual questions in an evaluative manner – or deciding which conceptual schemes we should adopt – may be less of a fraught exercise if it turns out that the disagreement isn’t entirely normative. Perhaps interlocutors falsely think that they are involved in a purely normative disagreement, and demonstrating that the disagreement is in fact (at least partly) descriptive may help settle some disputes about how we ought to think and talk. And so, even though social ontological accounts may involve normatively driven pragmatic decisions about which linguistic framework to adopt, this does not undercut those accounts being descriptive and more or less aptly able to limn the structure of social reality. In other words, there is a way in which normatively driven linguistic approaches like conceptual ethics and amelioration can meaningfully discover and construct the nature of the social world. Descriptive analysis can elucidate candidate conceptual schemes ‘out there’, while normative considerations guide our choice of preferred conceptual scheme among the relevant candidates in the service of advocating for some way of life. This enables us to elucidate how language and ‘our’ linguistic repertoire can reveal something signifcant about the nature of social entities without implausibly buying into radical linguistic constructivism that renders social entities somehow solely dependent on such repertoire.

4 Conclusion: Why Care? One might wonder whether all of the above is merely insulated philosophical refection that has no real efects on sociality and our social relations. I wish to close the chapter by ofering some reasons to hold that thinking carefully about our conceptual choices is deeply important for social beings like us. To begin with, our conceptual repertoire enables and determines what we can think and say, which frames and constrains action in various ways in both mundane and sophisticated ways. Conceptual availability infuences the nature and form of our social and political institutions; it makes certain institutions (un)thinkable and (un)speakable. Think back to the example of family. As I have already discussed above, if we only have the traditionally conceived conception in our repertoire, less traditional conceptions simply won’t be articulable. Expanding ‘our’ conceptual repertoire will have profound extra-linguistic and non-conceptual consequences 39

Mari Mikkola

of making alternative family-models and conceptions possible – for instance, enabling simple everyday actions like being able to visit one’s same-sex spouse in hospital qua family member. Moreover, conceptual choices and possibilities shape what we can do and who we can be. Traditional conceptions of motherhood and what it is to be a mother have included putting one’s children frst and prioritizing family life over career aspirations. Due to such conceptions of motherhood, many women self-report struggling with conficting work life and family demands, and feeling a great deal of guilt in putting their children into daycare. And many are outright scorned for not living up to role expectations as mothers who sacrifce everything for their children. Consider yet another example: in her discussion of hermeneutical injustice, Miranda Fricker (2007) considers Edmund White’s autobiographical novel, A Boy’s Own Story. As the protagonist grows up, he is confronted with diverse bogeymen constructions of what it is to be “a Homosexual”. These constructions have signifcant power to forge not only the subject’s experiences (experiencing one’s sexual desires as shameful and guilt-inducing, for instance), but also his very social being. And so, the available conceptual choices can obscure our knowledge and understanding of our own experiences in harmful and disadvantageous ways. The lack of available hermeneutical resources can misconstrue a conception of ourselves: it can defectively construct self-conceptions that we would not have chosen had diferent resources been available to us. Again, we can see how language can both discover and construct sociality without this being particularly mysterious. We can discover existing linguistic frameworks to be defective and harmful when they ill ft our experiences and produce jarring efects in our social existence. But as Rachel Sterken holds, such jarring efects may be benefcial: Changing language, while necessary, is difcult… it can often lead to miscommunication and confusion… [But] these supposed problems can actually be benefcial. It’s good that changing language leads to miscommunication and confusion, because that can cause speakers to refect on their language, and that will lead them to focus on its faws and ways to improve them. (2020, 433) This can lead to a process that Sterken calls ‘transformative communicative disruption’. When faced with such disruption, we can subsequently (and hopefully) fnd better and more apt ways to account for social entities, kinds, and structures where this will benefcially impact how we understand and shape ourselves as social agents.

Notes 1 Butler is not alone in making these sorts of deconstructive claims and the idea that ‘really’ there are no genders is commonplace in postmodern feminism. For instance, Denise Riley (1988) claims that feminists should fght against attempts to classify women since this is always going to be misguided and dangerous. In fact, this is essential to feminism. And Julia Kristeva claims that the notion of woman must be deconstructed and cannot be reconstructed: “In [woman] I see something that cannot be represented, something that is not said, something above and beyond nomenclatures and ideologies” (1980, 137). 2 Of course this power isn’t absolute in any robust sense, and it is possible to resist and subvert sexing in the performative sense. My point is simply to illustrate how language not only describes some prior state of afairs, but can bring those states of afairs into being.

Bibliography Alcof, Linda. (1988). Cultural Feminism versus Poststructuralism: The Identity Crisis in Feminist Theory. Signs, 13, 405–36. Braddon-Mitchell, D. (2020). Reactive Concepts: Engineering the Concept CONCEPT. In A. Burgess, H. Cappelen and D. Plunkett (eds.) Conceptual Engineering and Conceptual Ethics. (79–99) Oxford: OUP.


Social Ontology Brigandt, I. and Rosario, E. (2020). Strategic Conceptual Engineering for Epistemic and Social Aims. In A. Burgess, H. Cappelen and D. Plunkett (eds.) Conceptual Engineering and Conceptual Ethics. (100–24) Oxford: OUP. Burgess, Alexis and Plunkett, David. (2013). Conceptual Ethics I. Philosophy Compass, 8, 1091–101. doi:10.1111/phc3.12086. Burgess, A., Cappelen, H., and Plunkett, D. (eds.) (2020). Conceptual Engineering and Conceptual Ethics. Oxford: Oxford University Press. Butler, Judith. (1991). Contingent Foundations: Feminism and the Question of ‘Postmodernism’. Praxis International, 11, 150–65. Butler, J. (1993). Bodies that Matter. London: Routledge. Butler, Judith. (1999). Gender Trouble: Feminism and the Subversion of Identity. 2nd edition, New York and London: Routledge. Cappelen, Herman. (2018). Fixing Language: An Essay on Conceptual Engineering. Oxford: OUP. Cappelen, H. and Plunkett, D. (2020) Introduction: A Guided Tour of Conceptual Engineering and Conceptual Ethics. In A. Burgess, H. Cappelen and D. Plunkett (eds.) Conceptual Engineering and Conceptual Ethics. (1–34) Oxford: OUP. Carnap, R. (1956). Meaning and Necessity: A Study in Semantics and Modal Logic. Chicago: Chicago University Press. Cornell, Drucilla. (1993). Transformations: Recollective Imagination and Sexual Diference. New York: Routledge. Diaz-Leon, E. (2020). Descriptive vs. Ameliorative Projects: The Role of Normative Considerations. In A. Burgess, H. Cappelen and D. Plunkett (eds.) Conceptual Engineering and Conceptual Ethics. (170–86) Oxford: OUP. Fricker, Miranda. (2007). Epistemic Injustice. Oxford: OUP. Haslanger, S. (1995). Ontology and Social Construction. Philosophical Topics, 23, 95–125. Haslanger, Sally. (2012). Resisting Reality. New York: Oxford University Press. Heyes, Cressida. (2000). Line Drawings. Ithaca, NY & London: Cornell University Press. Heyes, Cressida. (2012). Identity Politics. The Stanford Encyclopedia of Philosophy (Spring 2012 Edition), E. Zalta (ed.), Kincaid, Harold. 2006. Functional Explanation. In M. Risjord and S. Turner (eds.) Handbook for the Philosophy of Science: Philosophy of Anthropology and Sociology. (205–39). Amsterdam: Elsevier. Kristeva, Julia. (1980). Woman Can Never be Defned. In Elaine Marks and Isabelle de Courtivron (eds.) New French Feminisms. Amherst: University of Massachusetts Press. Mikkola, Mari. (2017). Feminist Perspectives on Sex and Gender. Stanford Encyclopedia of Philosophy (Winter 2017 Edition), Edward N. Zalta (ed.), feminism-gender. Moi, Toril. (1999). What Is a Woman? Oxford: OUP. Nicholson, Linda. (1998). Gender. In Alison Jaggar and Iris Marion Young (eds.) A Companion to Feminist Philosophy. Malden, MA: Blackwell Publishers Inc. Plunkett, D. (2015). Which Concepts Should We Use?: Metalinguistic Negotiations and the Methodology of Philosophy. Inquiry, 58, 828–74. Riley, Denise. (1988). Am I that Name? London: The Macmillan Press Ltd. Sider, T. (2009). Ontological Realism. In D. Chalmers, D. Manley and R. Wasserman (eds.) Metametaphysics: New Essays on the Foundations of Ontology. (384–423) Oxford: OUP Sider, T. (2011). Writing the Book of the World. Oxford: OUP. Sterken. R. K. (2020) Linguistic Intervention and Transformative CommunicativeDisruptions. In A.  Burgess, H. Cappelen and D. Plunkett (eds.) Conceptual Engineering and Conceptual Ethics. (417–34) Oxford: OUP. Thomasson, Amie. (2015). Ontology Made Easy. New York: OUP. Thomasson, Amie. (2017). Metaphysical Disputes and Metalinguistic Negotiation. Analytic Philosophy, 58, 1–28. Thomasson, Amie. (2020). A Pragmatic Method for Normative Conceptual Work. In A. Burgess, H. Cappelen and D. Plunkett (eds.) Conceptual Engineering and Conceptual Ethics. (435–58) Oxford: OUP. Witt, Charlotte. (1995). Anti-Essentialism in Feminist Theory. Philosophical Topics, 23, 321–44.



In an appendix to 1984, George Orwell described Newspeak, a language designed to advance the social and political ends of the totalitarian Ingsoc regime. Orwell discusses various linguistic features of Newspeak, including its syntax; but the most important feature of Newspeak is the meanings of its words: “Its vocabulary was so constructed as to give exact and often very subtle expression to every meaning that a Party member could properly wish to express, while excluding all other meanings” (1949/2000: 312–3). In short, the idea behind Newspeak is that a political group can bring about its political ends by making it the case that the language contains words with certain meanings, and lacks words with other meanings. And Big Brother isn’t the only one who has thought meaning-making an important tool for social and political change; those with more admirable social and political aims have shared this idea. For example, the anonymous authors of the 1990 pamphlet Queers Read This emphasise the political need for a word which “unlike GAY, doesn’t mean male”, because “when spoken to other gays and lesbians it’s a way of suggesting we close ranks, and forget (temporarily) our individual diferences because we face a more insidious common enemy” (1990). Are they right? Can making it the case that our words have some meanings (or that our language lacks words with other meanings) play a role in bringing about social and political ends? If so, exactly how does this work – what relations obtain between the facts that determine what we mean and the social and political facts? These are the questions that I engage with in this chapter. I begin in Section 1 by introducing some terminology and setting out the scope of discussion in the chapter. In Section 2, I consider ways in which facts relevant to the determination of meaning might be thought to be socially and politically signifcant. Section 3 considers the role of normative facts in meaning determination, and Section 4 concludes by laying out some broadly ethical questions about meaning-making – questions mostly neglected in the literature that deserve further attention. There is a great deal of work across a range of philosophical traditions that is relevant in some ways to these issues, but there has been to my knowledge no attempt to systematise these issues in the context of contemporary theorising about the foundations of meaning. My main aim in this chapter, therefore, is not so much to defend a particular view as to provide a systematic framework for thinking about the points at which the foundations of meaning might be socially and politically signifcant, and ways in which the social and political facts might bear on the foundations of meaning. My hope is that doing so will bring some neglected issues of interest both to philosophers of language and to theorists engaged with social and political issues to the fore. 42

Invitation to Political Metasemantics

1 Background It is commonplace in discussions of linguistic meaning to distinguish what words (and sentences, and other representational entities) mean, from what makes it the case that words mean what they do (e.g. Kaplan 1989: 573–4). We all know that the name “Barack Obama” refers to a particular person; that is a semantic fact, a fact about the meaning of the name. But semantic facts of this kind call out for explanation. Words mean what they do only in virtue of various attitudes and activities of speakers, and their relations to their environments; and it is a substantive question in the philosophy of language which attitudes, relations, and activities matter. Does “Obama” mean what it does because of certain beliefs that we associate with the name? Because of historical facts, for example, about the intentions of Obama’s parents at his christening ceremony or the signing of his birth certifcate? Because of facts about the dispositions of speakers (or perhaps some privileged group of expert speakers) to apply the name? Or something else entirely? Semantics is the study of what words (and other expressions) mean, and metasemantics is the study of what makes it the case that words have the meanings they do. Thus, it is a semantic fact that “Barack Obama” refers to Barack Obama; and if “Barack Obama” refers to Barack Obama in part because of the intentions of Obama’s parents in flling out his birth certifcate, then that is a metasemantic fact. Among the semantic facts, we can distinguish those that concern particular utterances or patterns of usage, or conventional meanings of particular words in a community – facts like, that when I said “the man with the martini”, I was referring to Jones, or that “dog” in English is correctly applied to dogs. If semantic facts are not primitive, each such particular semantic fact will have an explanation – as it might be, that when I said “the man with the martini”, I was referring to Jones because I intended to refer to Jones, or that “dog” in English is correctly applied to dogs because there exist conventions among English-speakers according to which utterances of sentences involving “dog” are systematically correlated with beliefs about dogs. Call facts of this kind particular metasemantic facts. The facts that Orwell points to as politically important have a somewhat diferent form; he emphasises that in Newspeak, there are words with certain meanings, and no words with others. These are quantifed claims, not claims about particular words. But we can nonetheless look for explanations of these quantifed semantic facts: for example, it might be that there is no word in Newspeak that means freedom because there is no word that is conventionally used by speakers of Newspeak to express attitudes about freedom, and this, in turn, might be true because the government has banned the expression of such attitudes. If that is so, then it is a quantifed metasemantic fact. Several points of clarifcation are in order. First, what are we talking about when we talk about meaning? The word “meaning” and related words such as “defnition” are used in a number of ways, both in ordinary English and in philosophy – especially those areas of philosophy that are concerned with the social and political signifcance of language.1 There are many worthwhile intellectual projects exploring the explanations of why representational entities have the meanings they do, in many diferent senses of “meaning”; exploring any signifcant proportion of them would be the project of a lifetime, rather than a chapter. I therefore restrict my attention to notions of meaning in the tradition of truth-conditional semantics (i.e. the tradition that traces its roots to Frege, Tarski, and Carnap, and that has been developed by Montague, Kaplan, Kripke, Lewis, Davidson, Partee, Heim, and Kratzer, among many others). These include notions of truth and truth-conditions (at a point of evaluation), of reference, of intension and extensions, of correct application, and of compositional semantic value. (See Rabern and Ball (2018) for discussion.) Second, when we say that a certain semantic fact obtains because of a certain other fact, we might have diferent kinds of relation in mind. For example, one could imagine that after a blow to the head, I begin to call cows “dogs”; and some theorists might think that if my disposition to use the word in this way is suitably robust, “dog” as I use it has taken on a new meaning, one on which it is correctly applied to cows. If that’s right, it is true both that (i) “dog” (as I use it) refers 43

Derek Ball

to cows because I was hit on the head, and (ii) that “dog” (as I use it) refers to cows because I have robust dispositions to apply call cows “dogs”. But (i) and (ii) are pointing to very diferent kinds of explanation of the same semantic fact. Similar examples are familiar: the water is hot because (i) I lit a fre underneath it or (ii) its molecular kinetic energy is high; the prisoner is a thief because (i) she grew up in difcult socioeconomic circumstances or (ii) she stole the diamonds. Very roughly, in each case (ii) is telling us something about the underlying nature of the fact, while (i) is telling us about its causal history. Call the (i)-claims causal explanations and the (ii)-claims constitutive explanations. I will not attempt to make this distinction precise, but will take for granted that it is reasonably clear in the cases of interest. (See Dasgupta 2017 for discussion.) In general, metasemantic theory is primarily interested in constitutive explanations, and that will be our focus in what follows, though we will turn our attention in one or two places to causal claims about how one might bring it about that a particular constitutive explanation obtains. Third, discussions of these issues often take it for granted that meaning is fxed by widespread beliefs or patterns of use, so that (for example) if (at a certain time) every English speaker is disposed to apply the word “fsh” to whales, or if every English speaker believes that whales are in the extension of “fsh”, then it is correct to apply the word “fsh” to whales. For example, Bettcher (2012) claims that in many uses in the “dominant cultural context”, the expression “trans woman” means “a man who lives as a woman”, on the grounds that “that meaning is accepted by many people and, indeed, often by the media, law enforcement agencies, domestic violence and homeless shelters, and so forth” (2012: 285), and goes on to defend the claim that “trans woman” as it is used in the trans community has a diferent meaning. Bettcher’s view is intriguing and may be right; but if we are concerned with meaning in the truth-conditional sense just described, the metasemantic principle on which her argument relies (viz. that if a particular claim about meaning is accepted by many people including relevant authorities, it is true), it is neither obviously correct nor something we can take for granted without argument. For example, one metasemantic view has it that meaning is determined in part by metaphysical “naturalness”; another has it that words mean what a fully informed and rational judge would take them to mean, taking into account the history of usage and empirical facts as well as present beliefs and patterns of use (Bigelow and Schroeter 2009, Schroeter and Schroeter 2009); another has it that meaning is determined by our judgements at the end of inquiry, once all the evidence is in (Ball 2020). On any of these views, meaning can come apart from what speakers at a particular time believe or accept, and from their patterns of usage (so that, for example, on these views it would be possible that speakers in the dominant cultural context are simply mistaken about the meaning of “trans woman” even as they use it, and that in fact speakers in the dominant cultural context use “trans woman” with the meaning Bettcher claims it has in the trans community); and this is a theoretical possibility that we will keep open for the sake of discussion.2 Fourth, I began this discussion by suggesting that the semantic facts are not fundamental; they “call out for explanation” in terms of the actions of speakers, their relation to their environment, and so on. But it does not follow from this that every good explanation of a semantic fact will appeal exclusively to non-semantic facts. Syntactically complex expressions provide one simple class of examples: plausibly, the fact that “Dogs bark” means that dogs bark is to be explained by appeal to further semantic facts (i.e. facts about the meaning of “dogs” and the meaning of “bark”). Likewise, many theorists maintain that facts about the meanings of words are ultimately to be explained in terms of facts about attitudes such as belief. But on one plausible view, beliefs are themselves mental representations, and as such will have semantic properties (e.g. Fodor 1975); and though many theorists fnd it plausible that the semantic properties of mental representations are ultimately explicable in non-semantic terms, this remains the subject of some controversy (e.g. Bealer 1997). I will take it that explanations of facts about word meaning in terms of facts about mental representations can count as good metasemantic explanations. Many of the mechanisms 44

Invitation to Political Metasemantics

that we will discuss may also be relevant to the metasemantics of mental representations, but our discussion will focus on language. Fifth, I have spoken freely of “facts”; but the nature of facts is the subject of much metaphysical disputes. I have no heavy-weight notion of fact in mind; I take facts to be true propositions, and I make no particular assumptions about the nature of propositions other than that they can be true and false. Sixth, the topic of this chapter is the social and political. I will leave these notions intuitive; I hope that it will be clear that the issues discussed are social and/or political in some sense. With these clarifcations in mind, we can distinguish three ways in which metasemantics might be relevant to the social and political. •

First, the obtaining of a particular metasemantic fact might have certain social and political consequences. (This seems to be what Orwell and the authors of Queers Read This had in mind; the idea was that by making it the case that words have certain meanings, we could bring about particular political ends.) Second, we can ask what role social and political considerations can play in metasemantics. For example, suppose that it would be just if a word had a certain meaning. Could this play a role in making it the case that it does have that meaning, and if so, how? Third, we can ask normative questions about metasemantic projects: would it be right to try to make it the case that a particular word has a particular meaning? Suppose (as many views have it) that some people have power to determine the meanings of words – power that others lack. Under what circumstances is it permissible to exercise this power? If we can shape the meanings of others’ words, do we need to get their consent before doing so?

These three sets of topics are the subjects of the three subsequent sections of this chapter.

2 Social and Political Consequences Can the metasemantic facts have social and political consequences? In answering this question, we need to be clear about exactly what “metasemantic facts” we have in mind. We have considered claims of the following form: s because e, where s is a semantic fact and e is the fact that explains s.3 We should distinguish three things: 1 2 3

The semantic explanandum (s) The metasemantic explanans (e) The metasemantic fact proper (i.e. the fact that the explanans explains the explanandum – that s because e).

Each of these three things might be thought to have social or political consequences, and we will examine each, in turn.

2.1 Te Semantic Explanandum On one reading, the examples with which we began – Orwell’s Newspeak and the use of “queer” – it is the semantic fact – the explanandum – that seems to be of primary interest. The idea behind Newspeak is that if Big Brother were able to bring about (for example) the quantifed semantic fact that there is no word for political freedom, people would become unable even to conceive of 45

Derek Ball

freedom and hence would be more easily politically subjugated; likewise, the authors of Queers Read This propose that by bringing about the particular semantic fact that “queer” picks out gay men and lesbians, one could make gay men and lesbians see themselves as a unifed group in a way that promotes social action. If that is the right way to understand these cases, they are concerned primarily with the semantic fact, and less (or not at all) with what makes it the case that these semantic facts obtain.4 Such claims about the signifcance of semantic facts are strictly speaking outside the scope of a chapter on metasemantics. However, it will be important to understand some of the reasons why certain semantic facts might be thought to have social and political signifcance, both to understand these cases better, and to facilitate discussion when we turn our attention to the metasemantic explanans. Several possible reasons of this kind are suggested by the passage on the word “queer” from Queers Read This: Well, yes, “gay” is great. It has its place. But when a lot of lesbians and gay men wake up in the morning we feel angry and disgusted, not gay. So we’ve chosen to call ourselves queer. Using “queer” is a way of reminding us how we are perceived by the rest of the world. It’s a way of telling ourselves we don’t have to be witty and charming people who keep our lives discreet and marginalized in the straight world. We use queer as gay men loving lesbians and lesbians loving being queer. Queer, unlike GAY, doesn’t mean MALE. And when spoken to other gays and lesbians it’s a way of suggesting we close ranks, and forget (temporarily) our individual diferences because we face a more insidious common enemy. Yeah, QUEER can be a rough word but it is also a sly and ironic weapon we can steal from the homophobe’s hands and use against him. (1990)s One thing to note about this passage is that several of the putative features of “gay” and “queer” are not matters of meaning or semantics in the sense that is the focus of this chapter (the truthconditional sense). For example, on one plausible way of individuating words, occurrences of “gay” which are roughly synonymous with “happy and carefree” are occurrences of a diferent word than the occurrences of “gay” which are the primary focus of the pamphlet. If the fact that these two words sound the same makes it the case that “gay” brings to mind being happy and carefree, and hence that being “gay” seems inconsistent with feeling angry and disgusted, that is not really a matter of semantics. Facts of this kind are certainly not be ignored; the association of “queer” with a homonym roughly synonymous with strange, and its history as a term of abuse and the memories that this is likely to trigger in the pamphlet’s audience, are crucial to the pamphlet’s case. But it seems that these do not bear directly on the contribution that these words make to the truth-conditions of occurrences of sentences in which they appear. I therefore set them aside. Why, then, might the fact that the word “queer” has a certain conventional meaning, or the fact that there is a word with that meaning, be important? Why, in particular, is it important to have a word that applies both to lesbians and to gay men? I think that pamphlet is best read as making a generalisation about human psychology: roughly, that regular use of a word causes one to pay attention to what everything in the extension of the word has in common, and what distinguishes entities in the extension of the word from others. (See Leslie 2017 for some related patterns of argument.) This may well be a truth about human psychology; and if it is, then one can see how semantic facts might have social and political efects: for example, making it the case that there are words that pick out particular groups might play a role in making those groups politically unifed blocs. Orwell seems to rely on a diferent claim about human psychology. He writes, “The purpose of Newspeak was not only to provide a medium of expression for the world-view and mental habits proper to the devotees of Ingsoc, but to make all other modes of thought impossible” (1949: 46

Invitation to Political Metasemantics

312). This suggests that the designers of Newspeak believe that one’s language determines one’s “world-view”, and makes certain “modes of thought” possible and others impossible. Orwell does not give a precise account of what he has in mind by “world-view” or “modes of thought”. One plausible reading would have Orwell presenting the designers of Newspeak as endorsing something in the vicinity of what 20th-century philosophers and linguists call the Sapir-Whorf hypothesis – itself a notoriously difcult idea to formulate precisely, but which can usefully be seen as combining two doctrines: •

Linguistic determinism – the claim that one’s language infuences or determines what one believes, in such a way that speakers of diferent languages will tend to possess diferent (and potentially incompatible) beliefs precisely because they speak diferent languages. Linguistic relativism – the claim that one’s language infuences or determines what concepts one possesses, and hence what thoughts one is capable of entertaining, in such a way that speakers of diferent languages frequently possess entirely diferent conceptual repertoires precisely because they speak diferent languages

Entering into the many empirical and philosophical debates about Sapir-Whorf would take us far beyond the scope of this chapter (see Swoyer 2014 for discussion). But we will pause to note that proponents of Sapir-Whorf have reason to think that semantic facts are socially signifcant, since the semantic facts shape our mental states and play a role in determining what we believe; and that some thought in this vicinity seems to be in the background of much thought about the social and political signifcance of language. We have so far focused on ways that semantic efects may impact on human minds. But some theorists have claimed that semantic facts have ontological consequences, because these facts are in part constitutive of social reality. To get a sense of why these claims are plausible, consider mortgages. Mortgages exist, but they do not exist independently of human activity; they are brought into being by things people do, and they can only be brought into being in a suitable social environment (i.e. one that includes conventions governing property ownership, money, contracts, and so on). Arguably, the existence of a way of talking about mortgages is crucial to making it the case that there are mortgages. (See Searle 1995 for one attempt to make this case, and see Haslanger 2012 for further analysis.) At least, it is certainly difcult to imagine a scenario in which there are mortgages but no way of talking about them. If this kind of claim is correct, then the existence of certain semantic facts plays a direct role in making it the case that certain phenomena – including, potentially, socially signifcant phenomena – exist. We have essayed a wide variety of possible social efects of semantic facts; let me therefore summarise the main categories of efect that we have discussed: 1

2 3

Psychological efects of grouping things together using a word (e.g. the idea that having a word that groups gay men and lesbians together will make salient to them what they have in common) Linguistic determinism and linguistic relativism (e.g. the idea that without a word that means “freedom” we could not entertain thoughts about freedom) Social ontology (e.g. the idea that words for mortgages are essential to making it the case that mortgages exist).

To be clear, I do not take this section to have established that all – or indeed, any – of these alleged phenomena are genuine. Instead, I have aimed to bring out some ideas that may be suggested by the idea that metasemantics may be socially and politically signifcant, but which are not really metasemantic phenomena at all (but should rather be thought of as semantic phenomena). 47

Derek Ball

This  was important for two reasons: frst, to the extent that these phenomena are real, we may want to exploit them (or to prevent others from doing so) to advance our social and political ends, and in order to do this we will need to understand metasemantics. (I return to this point in Section 2.3.) Second, I want to distinguish these (alleged) consequences of semantic facts from the consequences of metasemantic facts – the metasemantic explanans and the metasemantic fact proper – to which we now turn.

2.2 Te Metasemantic Explanans We have already noted that there are many metasemantic views in the literature. Exactly what one takes the social and political signifcance of the metasemantic facts to be will depend on exactly what one takes those facts to be. The sorts of facts supposed to be metasemantically signifcant on some views have relatively little, or relatively indirect, social, and political import. (For example, on Putnam’s well-known view, the fact that water is a natural kind plays a role in explaining the fact that “water” refers to water. But the fact that water is a natural kind has little direct social and political signifcance.) Other views will appeal to a wide range of facts, and some of these may be deeply signifcant. We may divide these into four broad classes: 1 2 3 4

Facts about speakers’ mental states and actions Facts about social structures Further (socially and politically signifcant) semantic facts Normative facts (e.g. facts about what is just and unjust).

We will return to the possible metasemantic role of normative facts in the next section. In this section, we will address the other three classes, in turn. Our aim is not to develop a comprehensive catalogue of metasemantic views that might have some social or political signifcance; rather it is to illustrate how the kinds of facts appealed to in metasemantics might be socially and politically signifcant by giving examples from prominent metasemantic views, and thereby to set the stage for the normative questions in metasemantics that will be the topic of Section 4.

2.2.1 Facts about Speakers’ Mental States and Actions The idea that the facts about the semantics of natural languages (such as Swahili or British Sign Language) are to be explained in terms of the mental states of speakers of these languages is familiar from the work of Grice (1989) and many others. Many controversies remain about the details of which mental states matter: Grice pointed to intentions, but there is dispute about exactly which intentions are relevant (e.g. Recanati 1986); and others might think that beliefs or other states play the crucial role. Now it is beyond doubt that our mental states in many typical cases have social and political consequences: our social and political structures are in large part a result of our actions, and our actions are in large part a result of our mental states. For example, if I believe that women are inferior to men, and intend to ensure that they are socially subordinated as a result of this belief, that may cause me to advocate for certain oppressive legislation. If facts like these play a role in metasemantic explanations, the metasemantic explanans will be socially signifcant in some cases. However, there is little reason in general to think that mental states that are so obviously socially signifcant are also metasemantically signifcant. (Misogynist attitudes and the like can play a role in causal metasemantic explanation – for example, such attitudes might cause us to introduce negative words for women – but on most views, they will not play a role in constitutive metasemantic explanation.) And in general, our views on the potential social signifcance of attitudes implicated 48

Invitation to Political Metasemantics

in the metasemantic explanans will depend on our views about what those attitudes are. However, it is plausible that on widely held metasemantic views, the mental states relevant to metasemantics will be potentially socially signifcant. For example, consider the Gricean view of speaker-meaning, on which meaning that p is a matter of having the intention to cause your interlocutor to form the belief that p on certain grounds. It is plausible that intending to cause others to believe certain propositions is morally problematic in part because its social and political efects: for example, one ought not form the intention to cause others to believe that people of a certain racial group are inferior. If that is right, then the mental states that play a role in metasemantics on at least one very prominent view are normatively evaluable in part at least because of their social and political signifcance. Of course, one might question whether it is the intention to cause others to form certain beliefs that is itself problematic, as opposed to the actions one might undertake as a result having that intention. And some metasemantic views may emphasise actions over the mental states that cause them. But as in the case of mental states, our views about whether the actions that are implicated in metasemantics are themselves particularly socially signifcant will depend on which actions we take to be metasemantically signifcant, and many of the most obviously socially and politically signifcant actions will have at most only indirect causal efects on metasemantics. Still, it is not hard to imagine potential social consequences of actions that play a constitutive metasemantic role according to prominent views in the literature: for example, “dubbing” or “baptising” something (i.e. introducing a name for it, which plays a central role in “causal chain” views of reference inspired by Kripke 1980) may bring that phenomenon to prominence, or indicate its importance.

2.2.2 Facts about Social Structures A variety of social facts might be thought to be metasemantically signifcant. For example, it is widely maintained that experts play a special role in metasemantics (e.g. Putnam 1975, Burge 1979). One version of the idea has it that the opinions or practices of experts determine what words mean, even in the mouths of non-experts. On this view, it might be that the word “molybdenum” refers to the chemical element with atomic number 42 because certain chemists systematically apply that word to that element; and that this determines what “molybdenum” means throughout the linguistic community, even as used by those non-experts who may be unaware that molybdenum is a chemical element, unable to tell molybdenum from aluminium, and so forth. There are numerous ways that one might try to make the idea more precise; but any view in the vicinity is going to need to say more about what makes one an expert. Plausibly, expertise in the relevant sense is not merely a matter of knowledge; on most views, one who knows a lot about chemistry but is completely socially isolated or ignored is not going to be in a position to fx the meaning of “molybdenum”. Instead, expertise in the relevant sense is at least in part a matter of enjoying a certain social status: for example, being such that other speakers are willing to accept correction from you and adjust their views in response to your testimony, or such that other speakers will intend to use words in the way that you use them. This makes social facts – facts about who is listened to, who is accorded status, authority, or power – relevant to metasemantics; on this style of metasemantic view, the metasemantic explanans will consist in part of facts about social structures.5

2.2.3 Conclusion on the Metasemantic Explanans We have noted that several widely held metasemantic views (Grice’s, Kripke’s, Burge’s) appeal to attitudes, actions, and phenomena that are socially and politically signifcant. This makes the normative evaluation of the metasemantic explanans potentially an important task, and we will begin to outline some of the relevant issues in Section 4. 49

Derek Ball

2.3 Te Metasemantic Fact Proper Consider again the claim that some semantic fact s is explained by e. We have now discussed the possible social signifcance of semantic explanandum, and the possible social signifcance of the metasemantic explanans e. What further signifcance can be accorded to the metasemantic fact proper – the fact s explains e? A natural response is: none at all. Once we have accounted for the social efects of the semantic explanandum and the social efects of the metasemantic explanans, there is nothing further that could be explained by the metasemantic fact proper. After all, the metasemantic fact proper (at least to the extent that we are construing it being about a constitutive explanation) is a fact about the obtaining of an obscure metaphysical relation between two more ordinary facts. A fact like that (it might be thought) could not possibly be socially signifcant. I want to consider two possible replies to the idea that the metasemantic fact proper cannot be socially signifcant. The frst reply contests it directly. Consider Fricker’s (2007) examples of hermeneutical epistemic injustice. The examples begin with thinkers who lack the cognitive resources necessary to make sense of their own experiences: for example, women in the mid-20th century who experienced sexual harassment, but lacked a word or conceptual category for this experience. Thinkers in such a position are unfortunate; but as Fricker observes, a lack of hermeneutical resources is not ipso facto an injustice Diferent groups can be hermeneutically disadvantaged for all sorts of reasons, as the changing social world frequently generates new sorts of experience of which our understanding may dawn only gradually; but only some of these cognitive disadvantages will strike one as unjust. (2007: 151) Fricker proposes that what distinguishes the cases of hermeneutical disadvantage that are unjust is that they involve “hermeneutical marginalization” – exclusion from the practices that play a role in putting us in a position to understand our experiences. Now it is not the case that every example of hermeneutical epistemic injustice (as Fricker presents it) directly involves metasemantics. (For example, some of Fricker’s cases involve subjects who have words that would apply to their situation, but are precluded by background facts and attitudes from applying those words to their own case.) But there are a range of cases in which the kind of fact that Fricker is pointing to is precisely a (quantifed) metasemantic fact. What was unjust about the situation in the mid-20th century is not the fact that there was no standard expression for “sexual harassment”; it is rather that there was no expression for sexual harassment precisely because the victims of sexual harassment were denied positions of authority that would have put them in a position to create such an expression – and that is a quantifed metasemantic fact (albeit at least in part a causal one). If Fricker is right, then the metasemantic facts are in part constitutive of certain facts about justice. The second reply to the idea that the metasemantic fact proper cannot be socially signifcant concedes the point that the metasemantic facts themselves, but insists that investigating the metasemantic facts proper is a task of critical importance to the socially concerned philosopher of language – and to anyone who wants to efect social change. This is because given the discussion above made it plausible that: A

The semantic facts have important social efects.

But given (A), and plausible assumptions (that diferent semantic facts have diferent efects, that the semantic facts can be changed, and so on), it follows that: 50

Invitation to Political Metasemantics


By changing the semantic facts, one can bring about social change (for good or ill).

(B) is what seems to have been presupposed by Orwell and the authors of Queers Read This. However, in general we are not in a position to change the metasemantic facts directly. We cannot just push a button and make it the case that a particular word has a particular meaning. If we want to change the semantic facts, we must do so by changing the facts on which they depend; and to be in a position reliably to change the semantic facts will therefore require knowing how the semantic facts depend on other facts (see e.g. Cappelen 2019). For example, an Orwellian totalitarian regime will only be in a position to try to make it the case that there is no word for freedom if they have some idea of what factors tend to make it the case that there is a word for freedom. In other words, they will need to know (or at least reasonably believe) metasemantic facts to the efect that “freedom” refers to freedom because ____ – and if we want to stop them, we need to know these same facts. In short, from (B) we can infer the following: The Social Utility of Knowledge of Metasemantic Fact: To the extent that semantic facts have social and political efects, those who wish to bring about (or to prevent) social change should aim to know the metasemantic facts. It might be objected at this point that what is useful is not knowledge of constitutive metasemantic facts, but causal metasemantic facts. (If Orwell’s totalitarian regime knew that they could make it the case that there is no word for “freedom” by hitting everyone on the head, that would serve their purposes perfectly well, even if the relation between blows on the head and meaning is merely causal and not constitutive.) This is strictly speaking correct, but it overlooks the fact that in many cases, one cannot know the causal metasemantic facts without knowing the constitutive metasemantic facts, at least in broad outline. If we cause a word to have a certain meaning, we do so by causing some fact to obtain that constitutively explains the word’s having that meaning; so if we want to cause a word to have a certain meaning, we will (in typical cases) need to know what facts would (if they obtained) be constitutive of its having that meaning.

3 Te Role of Social and Political Considerations in Metasemantics Consider the question of whether the word “marriage”, as used by typical English speakers in 1980, could correctly be applied to relationships between same-sex couples. (To be clear, the question is not whether any same-sex couples were then married (the legal facts prevented that, at least in most parts of the world); the question is whether the idea that a same-sex couple could be married even made any sense, given the meaning of “marriage” – whether, for example, enacting a law purporting to legalise same-sex marriage would involve changing the meaning of “marriage”.) There are familiar facts about speakers’ use at the time that suggest that “marriage” could not be correctly applied to same-sex couples: for example, the fact that virtually all speakers believed that it could not, the fact that they withheld application of “marriage” to same-sex relationships, and so on. But it is also familiar that many people now regard this situation as unjust: given the background facts, it is unfair if “marriage” cannot be correctly applied to same-sex relationships. The question I want to engage with in this section is: what is the metasemantic signifcance of facts like this? Can facts about justice (or related normative notions, such as fairness, right, and wrong), play a role in making it the case that a word has a particular meaning – and if so, what role? There are three main possibilities; in discussing them I will focus on the example of justice, but the conclusions should be generalisable to other normative notions. 51

Derek Ball

First, the facts about justice might play a role in explaining some semantic fact in a derivative way. The explanations in question may be derivative in one of two senses: (i) they may rely on some other semantic fact that links a particular word with justice, or (ii) they may be instances of a general metasemantic fact of which considerations of justice are one example among others. To take a trivial example of (i), suppose that the word “just” may correctly be applied to (say) a particular law in part because that law is just. That may be a correct metasemantic explanation; but if so, one natural view would have it that it is correct only because of an antecedent semantic fact: that the word “just” is correctly applied to things that are just. An example of (ii) would be the idea that “just” is correctly applied to things that are just because there is a reliable pattern of speakers applying it to things that are just. If that is correct, then the explanation of why “just” is correctly applied to things that are just will allude to things being just and speakers judging them so. But if a story like this is correct about “just”, it is presumably correct because it is an instance of a very general metasemantic fact that has nothing in particular to do with justice – for example, a fact to the efect that for any word w, what w is correctly applied to is determined in part by the facts about how speakers in fact apply w. These derivative metasemantic explanations are of relatively little interest. The second possibility is that the facts about justice play a causal role. Arguably, the facts about justice are not themselves causally efcacious. But it is extremely plausible that our beliefs about what is just and unjust are causally efcacious. And these beliefs may have an efect on our metasemantically explanatory attitudes or actions. For example, if we become convinced that it is unfair to withhold the word “marriage” from same-sex relationships, that may cause us to begin to apply the word to same-sex relationships; and this may, in turn, constitute a change in what we meant by “marriage”, or play a role in shaping what we meant by “marriage” all along. Such causal metasemantic explanations that allude to our beliefs about justice may be of practical interest, if we want to change the semantic facts; but it is not clear that they are particularly signifcant from the point of view of metasemantic theory. But there is a third possible way that facts about justice could be involved in metasemantics – a way that, to my knowledge, has been neglected in the literature. On this third possibility, there could be constitutive metasemantic explanations that appeal to facts about justice, which are neither derivative on other semantic facts, nor instances of general metasemantic rules. I can think of two broad ways this might go. The frst is if the facts about justice are constitutive of some other kind of fact on which meaning depends. As an example, consider the view that meaning is fxed by the views of experts. What makes someone an expert? One view is that it is simply being in a position of authority, a position in which one is listened to; and sad experience tells us that one can be in such a position no matter how little one cares for justice. But there are alternative ways of thinking about expertise: for example, one might think that the true expert must show a certain sensitivity to considerations of justice, so that the opinions of an unjust person could never do metasemantic work even if that person is in a position of authority. And one might likewise see justice as involved in the mechanisms postulated by other metasemantic views: for example, it might be that justice is in some cases constitutive of metaphysical naturalness. The second way that facts about justice might enter into substantive metasemantic explanations is not via a connection between justice and some other metasemantically signifcant notion, but directly: perhaps (for example) our judgements and opinions fx what we mean, but only if those judgements and opinions are just; or perhaps the facts about justice are one among many factors – alongside the views of experts, our patterns of usage, the facts about naturalness, and so on – that can play a role in making it the case that our words have a particular meaning. I know of no explicit attempt to defend a view of this kind in the literature.6 But it has some appealing features: for example, those of us who are tempted to think that the meaning of “marriage” never precluded its 52

Invitation to Political Metasemantics

application to same-sex couples (even if speakers in the past all confdently asserted that same-sex marriage was a contradiction in terms, ruled out by defnition) might fnd vindication in the idea that facts about justice play a meaning-fxing role.

4 Conclusion: Normative Refections on Metasemantics Consider again the Burge-inspired view that the linguistic usage of experts in the community plays a special role in fxing meaning (and set aside for the moment the possibility mentioned above that the relevant notion of expertise requires justice). Suppose a group of relevant experts enters into a conspiracy to make our views and utterances false by changing their own usage of certain words. For example, perhaps they decide to use the word “arthritis” in such a way that it cannot correctly be applied to any disease of the ankles (so that “arthritis” picks out swelling in non-ankle joints). Or worse, suppose that the experts shift their usage of some socially signifcant word: perhaps they all agree to use “marriage” in such a way that it cannot be correctly applied to same-sex couples. Now it is an interesting question whether, or under what circumstances, such a meaningmaking conspiracy would be possible. (The discussion in the previous section will obviously bear on the answer to this question.) But many metasemantic views allow that there are considerable asymmetries in the power of diferent speakers to contribute; so many theorists allow that something like this could happen. It seems that in that case the experts have done something wrong. Even considering a case like “arthritis” – a case that does not matter to people’s self-conceptions, and which has relatively few social consequences – there is something troubling about the experts’ arbitrarily choosing to make so many of our utterances and views false. And cases involving socially important words like “marriage” are more troubling still. What exactly is wrong with the experts’ actions here? Our answers to this question will turn in part on our views about deep issues in normative ethics – issues about what makes actions in general right and wrong. But several obvious possibilities raise interesting questions for further research. Perhaps the consequences of the experts’ actions are what matter – for example, that their actions make it the case that many of our utterances are false? Perhaps; but it seems that the same consequences might result if the experts change their views in the course of scientifc research on arthritis, or (perhaps per impossibile) the course of well-conducted, good-faith philosophical or legal inquiry into marriage, and adjusting their judgements accordingly, and it is pretty clear at least in the case of “arthritis” that we would not regard such change of view as morally problematic. (We may have a diferent view about the case of “marriage” – or perhaps still more, cases of terminology that people use to express deep aspects of their identities and self-conceptions (for example, terms related to gender, race, or sexuality).) Alternatively, perhaps it is that the experts are shaping what the rest of us mean without our consent? But again, it is plausible that this happens all of the time in the course of research; why is consent not required if experts change their judgements as a result of scientifc study? (See Ball 2020 for discussion of these and related issues.) This chapter aimed to develop a framework for thinking about social and political issues related to metasemantics. We have distinguished (1) ways that metasemantic facts may be socially and politically signifcant from (2) ways that socially and politically signifcant facts may bear on metasemantics. Our discussion of (1) distinguished the potential social signifcance of (1a) the semantic explanandum from the signifcance of the (1b) metasemantic explanans and (1c) the metasemantic fact proper. Our discussion of (2) set aside various relatively uninteresting ways in which considerations of justice might fgure into metasemantics before distinguishing (2a) views on which justice feeds in to some familiar factor that might play a metasemantic role (such as expertise or naturalness) from (2b) views on which justice is directly metasemantically relevant. And we concluded by considering (3) ethical issues associated with metasemantics. Many issues remain 53

Derek Ball

unresolved, and many relevant issues are entirely undiscussed in the context of contemporary thinking about metasemantics. The title of this chapter is meant to be taken seriously; I hope that it provides a framework on which future discussion can be built.

Notes 1 Just to mention one example germane to the topics at hand, Cameron and Kulick (2003) speak of words (as well as styles of dress, ways of speaking, and other phenomena) as “indexing” or “signalling” certain attitudes, identities, or other phenomena, where this means that those words (etc.) tend to be associated with those attitudes, identities, or phenomena; and they regard this as a kind of meaning. For example, they write, Although the two terms, “homosexual” and “gay”, have the same referential meaning – they identify the same group of people – their meanings in actual usage are not identical. The selection of one or the other can signify the diference between conceptualizing homosexuality as deviance or sickness, and conceptualizing it in other and more positive ways: as an alternative personal and/or political choice, for instance, or simply as one “natural” variant of human sexuality, less common than heterosexuality but not by that token deserving condemnation. (2003: 26)

2 3

4 5 6

This is certainly a diference worth exploring; but to the extent that it is true (contra the views of the authors of Queers Read This, quoted above) that “homosexual” and “gay” have the same “referential meaning” – are correctly applied to the same actual and possible individuals, make the same contribution to truth-conditions, etc. – exploring it is beyond the scope of this chapter. One might also be reminded here of more familiar externalist views: for example, Putnam’s (1975) famous claims that cats might turn out to be robots, and that pencils might turn out to be organisms rather than artefacts. One notion of explanation is factive; if it is the case that s because e, then it is also the case that s and the case that e. Defending particular semantic or metasemantic views is no aim of this chapter; but I need examples to discuss. I will therefore often mention various semantic claims as though they are true, and various metasemantic claims as though they were explanations, even though strictly speaking I do not mean to be taking a stand on whether they are genuine explanations or mere hypotheticals. Some philosophers doubt that semantic facts are explanatory in at least some contexts (e.g. Kim 1998, Rescorla 2012). I will not engage with these issues here, but will simply assume that explanatory appeals to semantic facts are in some cases legitimate. It is a consequence of this style of view that changing meanings may require changing social organisation. See Tirrell 1993 for some interesting discussion of related issues (which also bears on the issues to be discussed in Section 4). Saul (2012) discusses some relevant issues, but her main concerns are methodological, about what theorists should do – for example, how should semantic theorists respond to the claim that a given semantic view is unjust? If facts about justice play a metasemantic role, that would obviously have methodological consequences; but other considerations may also bear on methodology. (For example, a theorist who holds anti-realist views (in the sense that she denies that the aim of semantic theorising is to state semantic fact) might take considerations of justice into account even if they play no metasemantic role.)

References Anonymous. Queers Read This. Pamphlet distributed at Pride 1990. Text available online at http://www.qrd. org/qrd/misc/text/ Ball, D. (2020). Revisionary analysis without meaning change; or, could women be analytically oppressed? In A. Burgess, H. Cappelen, & D. Plunkett (Eds.), Conceptual Ethics and Conceptual Engineering (pp. 35–58). Oxford: Oxford University Press. Ball, D. (2020). Metasemantic ethics. Ratio, 33, 206–19. Bealer, G. (1997). Self-consciousness. Philosophical Review, 106, 69–117. Bettcher, T. M. (2012). Trans women and the meaning of ‘woman’. In N. Power, R. Halwani, & A. Soble (Eds.), The Philosophy of Sex: Contemporary Readings, Sixth Edition (pp. 233–50). Lanham, MD: Rowman and Littlefeld.


Invitation to Political Metasemantics Bigelow, J. & Schroeter, L. (2009). Jackson’s classical model of meaning. In I. Ravenscroft (Ed.), Minds, Ethics and Conditionals: Themes from the Philosophy of Frank Jackson (pp. 85–110). Oxford: Oxford University Press. Burge, T. (1979). Individualism and the mental. In P. Ludlow & N. Martin (Eds.), Individualism and SelfKnowledge (pp. 21–84). Stanford, CA: CSLI. Originally published in Midwest Studies in Philosophy, 4, 1979. Cameron, D. & Kulick, D. (2003). Language and Sexuality. Cambridge: Cambridge University Press. Cappelen, H. (2019) Fixing Language. Oxford: Oxford University Press. Dasgupta, S. (2017). Constitutive explanation. Philosophical Issues, 27, 74–97. Fodor, J. A. (1975). The Language of Thought. Cambridge, MA: Harvard University Press. Fricker, M. (2007). Epistemic Injustice: Power and the Ethics of Knowing. Oxford: Oxford University Press. Grice, P. (1989). Studies in the Way of Words. Cambridge, MA: Harvard University Press. Haslanger, S. (2012). Ontology and Social Construction. Reprinted in her Resisting Reality. Oxford: Oxford University Press. Kaplan, D. (1989). Afterthoughts. In J. Almog, J. Perry, & H. Wettstein (Eds.), Themes from Kaplan (pp. 565–614). New York: Oxford University Press. Kim, J. (1998). Mind in a Physical World. Cambridge, MA: The MIT Press. Kripke, S. (1980). Naming and Necessity. Cambridge, MA: Harvard University Press. Leslie, S. (2017). The original sin of cognition: Fear, prejudice, and generalization. Journal of Philosophy, 114, 393–421. Orwell, G. (1949/2000). Nineteen Eighty-Four. London: Penguin Classics. Putnam, H. (1975). The meaning of ‘meaning’. In Mind, Language, and Reality: Philosophical Papers Volume 2 (pp. 215–71). Cambridge: Cambridge University Press. Originally published in Minnesota Studies in the Philosophy of Science, 7, 1975. Rabern, B. & Ball, D. (2018). Introduction to the science of meaning. In D. Ball & B. Rabern (Eds.), The Science of Meaning (pp. 1–45). Oxford: Oxford University Press. Recanati, F. (1986). On defning communicative intentions. Mind and Language, 1, 213–42. Rescorla, M. (2012) Are computational transitions sensitive to semantics? Australasian Journal of Philosophy, 90, 703–21. Saul, J. M. (2012). Politically signifcant terms and philosophy of language. In S. L. Crason & A. M. Superson (Eds.), Out from the Shadows: Analytical Feminist Contributions to Traditional Philosophy (pp. 195–214). Oxford: Oxford University Press. Schroeter, F. & Schroeter, L. (2009). A third way in metaethics. Nous, 43, 1–30. Searle, J. (1995). The Construction of Social Reality. London: Allen Lane. Swoyer, C. (2014). The linguistic relativity hypothesis (supplement to Relativism). In E. N. Zalta (Ed.), The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab, Stanford University, Stanford, winter 2014 edition. Tirrell, L. (1993). Defnition and power: Toward authority without privilege. Hypatia, 8, 1–34.


4 LINGUISTIC PRESCRIPTIVISM Alex Barber and Robert J. Stainton

1 Introduction Linguistic prescriptivists have a bad name. To most academic linguists they display ignorance about language at best, and pernicious social or political attitudes at worst. Linguistics students are taught the shortcomings of prescriptivist grammars in their frst semester. Thereafter, their pronouncements about correctness are either ignored or relegated to data in sociolinguistic studies of (distressingly popular) attitudes towards language. Either way, linguistic prescriptivists are dismissed as having nothing useful to ofer. The non-academic public is more divided: while there is and always has been a market for their wares, many perceive language correctors as annoying busy bodies.1 In philosophy, curiously, it is hard to fnd much on linguistic prescriptivism beyond scattered remarks. Our plan is to make a start on changing this.2 For the avoidance of doubt (and perhaps disappointingly), we are not about to argue that it is wrong to split your infnitives. Our attitude here is in line with the orthodoxy in linguistics. Those who want you not to boldly go will need to fnd support elsewhere. Our plan instead is to pursue twin goals: Goal 1: Uncover a plausible interpretation of “linguistic prescriptivism”, ideally one under which it merits careful philosophical evaluation Goal 2: Consider whether any variant of linguistic prescriptivism, thus understood, is defensible Our tactic with respect to Goal 1 will be to run through a series of suggestions, at least some of which are obviously incorrect (Sections 2 and 3). We do this partly just to rule them out, but more positively so as to extract useful lessons regarding where each attempt goes wrong. Building on these lessons, we arrive at an understanding of linguistic prescriptivism sufcient to turn to Goal 2. Is linguistic prescriptivism as bad as it is said to be, whether for the reasons usually ofered or for others? Our conclusion: it is not (Section 4). This claim comes with a very important qualifcation, however. We are sufciently sympathetic to standard objections to stereotypical language prescriptions that we opt for a new, less freighted label. That done, however, we suspect most people reading this article will already agree with us, perhaps without realizing it.

2 First Approach: Linguistic Prescriptivism via Paradigm Examples Orthodox linguists perceive linguistic prescriptivists as pre-occupied with certain paradigm “bugbears”. We take these as our departure point, to give ourselves a kind of ostensive defnition: 56

Linguistic Prescriptivism

To be a linguistic prescriptivist is to feel passionately that such-and-such examples, and others like them, violate the rules of language, and that the correct rules must be followed. For this to become a proper defnition we need to fesh out both “such-and-such examples” and “others like them”. It’s easy enough to identify cases complained about by those Steven Pinker (1994) calls the “language mavens”: Pronunciation Word forms Word meanings Structure Confation Writing Euphemisms Informality Purity

E.g. “nuclear” as “new-kyu-lurr” E.g. backformations (“liaise” from “liaison”; “orientate” from “orientation”) E.g. “livid” as meaning red-faced E.g. preposition stranding (“Who’d you buy it for?”); case agreement (“John and me are married”); split infnitives (“to boldly go”) E.g. “disinterested” and “uninterested”; “shall” and “will”; “that” and “which”; “less” and “fewer” E.g. “Potatoe’s half price”; “naive” without the umlaut; the Oxford comma; starting a sentence with “But” E.g. “collateral damage” instead of “killed civilians”; “conscious uncoupling” instead of “deciding to break up” E.g. “ain’t”, and contractions in general (“doesn’t”) in writing E.g. so-called Americanisms (“done good” instead of “did well”)

We could go on (and on). In English-speaking countries, anxieties often centre on “bad grammar”. For many non-English language communities, prescriptions against creeping Anglicization are more central (“courriel” in place of “email” in French, say). But as the last example in our list shows, English speakers too can have anxieties about linguistic infection or domination (refecting, unsurprisingly, the wider socio-political context: Americans fretting over Briticisms is less common than the converse). Continuing with orthodox perceptions, we can now fesh out “and others like them”. By orthodox lights, all examples share something in common: they display an underlying muddle about language on the part of the prescriptivist. Two broad and overlapping failings stand out: being empirically ill-informed and having a nefarious social or political purpose. Let us take these, in turn. (Our intention at this point is not to press these charges home but to give a sympathetic airing to the standard take on prescriptivism.) As linguists like to point out, many prescriptions rest on evidence-free assertions about correct usage. The prescriptivist Michael Dummett, for example, rejects as “incorrect” the sentence “The Chinese economy is growing at 20% a year”. He simply states that “…is growing at the rate of…” and “…is growing by…”, are correct, while the amalgam is not (1993: 34). We have yet to fnd a native speaker who agrees. Another example Dummett ofers is the “dangling participle” in “Taking the broad view, Napoleon did more good than harm”. Using this would be “a grammatical mistake”, he tells us, unless one meant to say that Napoleon was taking the broad view (1993: 43). Again, to us and the other native speakers we have consulted, the more obvious reading (i.e. that Napoleon was a benign force overall) is perfectly acceptable, with context resolving the amphiboly in the usual way. (We don’t mean to pick on Dummett in particular: all such texts include similar cases.) Prescriptivist authors do occasionally attempt to justify their prescriptions, for example, by citing some general rule of “the language”. Thus, we come to a second kind of empirical misinformation: historical ignorance of how the rules arose. Many have roots in some arbitrary decision never consistently respected even in the past. John Dryden decided in the 17th century that preposition 57

Alex Barber and Robert J. Stainton

stranding was an “error” (1808 [1668]: 217–8). He chides one of England’s greatest playwrights (Ben Johnson) and one of its greatest poets (Dryden himself!) for the habit. David Crystal (2006: 111) suggests he was drawing parallels with Latin, in which prepositions must precede nouns (hence their name). But why should English follow Latin and why should we follow Dryden? The same applies for the rule that “disinterested” should mean impartial rather than uninterested/ indiferent. John Donne used it in the “wrong” way (1638 [1608]: 99), and many have since. This implicit prescriptivist appeal to a Golden Age is not only inaccurate, it has utterly implausible consequences, such as “nice” still being an insult – deriving from “nescius” via Old French, it once meant something like “stupid” – or modern Portuguese being a decadent form of Latin. Another example of historical obliviousness is the presumption that rapid language change is both novel and a sort of decay (Swift 1712; Truss 2003: 16–17). In fairness, prescriptivists do acknowledge the inevitability of language change, but their examples suggest this is lip-service. The British prescriptivist Simon Hefer, for example, begrudgingly acknowledges that “[w]ords do change their meaning over time”, (mis)describing this as a “democratic decision” that is “validated by authorities such as lexicographers caving in to abuse” (2011: 140). On the very next page, however, he bemoans “alibi” having lost its Latin meaning (“elsewhere”), describing this as a “rare case of an adverb having become a noun” and blaming contamination from overseas (“another thing for which we have to thank the Americans”, p. 141). Dummett, too, admits to the inevitability of language change, but argues that “it is we who change it, and it is up to us how fast it changes” and that “some people’s use of English is changing much too fast”, threatening to make Hume and Berkeley inaccessible (1993: 8–9). As any diachronic linguist recognizes, thanks to the arbitrariness of the sign, grammatical mutations and linguistic borrowings are and always have been ubiquitous, with new usages being coined as fast as old ones are lost. The changes are even semi-predictable in certain respects. Continuing with empirical ignorance, ill-founded charges about a loss of function are frequent. Dummett describes language change of the kind he is resisting as a “heinous crime” that is “harming generations to come”. Both he and Hefer ofer an A to Z of “important distinctions” being “lost to our language” (Hefer 2011: 140–62; cf. Dummett 1993: 89–96, 115). To use “begging the question” for raising a question allegedly risks “depriving the language of a useful phrase”, Dummett tells us (1993: 90). Likewise, he complains, the use of “alternately” as a synonym for alternatively will lead to the regrettable loss of its current meaning, which is to say switching back and forth between two states (1993: 89). Another popular prescriptivist distinction, supposedly functional, is the “shall/will” contrast. In American English, “shall” is nowadays largely confned to idioms; in British English, prescriptivists agree on the distinction’s importance but not on what it is supposed to be (compare Dummett 1993: 58–9 with Hefer 63–6). The result is that “shall” often serves merely as a nervous signal of formal register. Whatever the distinction once was, we seem able to communicate efectively without it now. We also now know we can get by without the “alternate/alternatively” distinction; and we can let go of “begging the question” since we still have “circular reasoning” and “petition principii”. Many claims about the cognitive function of certain vocabulary items are in any case empirically naive. Loss or change of vocabulary doesn’t have to weaken one’s ability to conceptualize and reason. Spanish speakers aren’t confused about the diference between toes and fngers just because they use “dedos” for both. Conversely, English-speaking biology students aren’t rendered less capable of absorbing information about the structural parallels between toes and fngers by their use of diferent words for each. The supposed direct and infexible infuence of the lexicon on psychology is overhyped: cognitive tools don’t always need to be refected in our language. Finally, complaints that vernacular language is illogical are generally ill-founded. Double negation (e.g. “That ain’t no bird”) is a common target, and Hefer singles it out as an “ofence against logic” (2011: 57). Unsurprisingly, no actual justifcation of this canard is forthcoming. (If 58

Linguistic Prescriptivism

English were some sort of regimented proof system, comparable to modal logic, he’d perhaps be right, but in that case “Jaime can’t run” would be readable as “Jaime is able to not run”.) Objections to “John and me are married” are likewise misguided. It is no good invoking the rule that any pronoun in subject position must have nominative case (giving “John and I are…”). Many other Indo-European languages do not require this for conjunctive subjects (e.g. in French it is “John et moi sommes mariés”, not “John et je sommes mariés”), suggesting there is no logical inconsistency when English speakers do the same. Besides being rife with empirical mistakes, heart-of-the-stereotype linguistic prescriptivism is commonly criticized (on the standard view of it among linguists) for expressing or being driven by dubious moral and political opinions. We’ll begin with some blatant exemplars, then note some more subtle ones. Some prescriptions look plain-old reactionary, antiquarian, backward-looking, or purityinvoking – in a way that chimes with a conservatism opposed to social progress. That is evidently the message behind Hefer’s description of himself as an advocate of “the highest standards of grammatical accuracy” in his dealings with “the apostles of political correctness” (2011: 199). Many prescriptions, far from being simple but much-needed assertions of what English really is, strike most linguists as simply snobbish, haughty or self-righteous. (The condescension is sometimes glaring, as with the name given to the so-called “greengrocer’s apostrophe”, i.e. a nonstandard apostrophe before an “s”.) Prescriptivists may also be criticized as apologists for sexist and racist language. Think here of the fact that there exists in English, only for women, a title specifc to a married person, namely, “Mrs.”. Linguistic prescriptivists may recognize this as a sexist feature of our language, yet insist that “such is proper English”, so that “Miss” and “Mrs.” must be retained over “Ms.”, as if we must live with the language bestowed upon us by history.3 The same applies to the racism apparently implicit in the verb “gyp”, used in American English to mean cheat or swindle and thought to be linked etymologically to “Gypsy”, i.e. Romany. Alternatively, stereotypical linguistic prescriptivism may unwittingly serve the cause of racism or sexism by denying that the language in question is problematic. Spanish usage, for example, defaults to the “-o” ending on nouns and adjectives when both genders are being talked about, which is also the ending for male-only groups. For instance, the term for a group of friends consisting of men is “amigos”; for a group of women, it is “amigas”; for a group of women and men it is, again, “amigos”. To “preserve the language”, prescriptivists may be led to deny that Spanish is thereby sexist. This brings us to more subtle considerations. Prescriptive rules are a hodgepodge of good and bad. Genuinely worthy advice can be found therein: style guides urge clarity, precision, brevity, etc. The call to obey their rules can therefore seem to be a mere plea for “standards” in a benign sense. Mixed in with these rules, however, are senseless ones. The only efective function of these is as shibboleths: one is expected to know and follow them in professional and even educational contexts. “Ignorance” marks one out; linguistic outsiders are disqualifed or disfavoured. Moreover, knowing these norms, precisely because they are pointless, typically requires growing up with the right sociolect, or efortful training, or expensive schooling. Consider, as one example among many, the British-English spelling rule that distinguishes “practice” (noun) from “practise” (verb). It is entirely redundant, since the grammatical category will always be evident from the position of the word in the sentence, as American spelling proves (always “c”). Yet ignorance of this “rule” and the hundreds like it can be a barrier to a good job, a higher grade, etc., because prescriptivists like Hefer include it among those that “have to be learned” (2011: 160). To object to enforcement of the “practice/practise” rule is to invite the charge of lacking standards, or of resisting the perfectly sensible advice that prescriptivists bundle up with the exclusionary foolishness. By such a mechanism, says our orthodox linguist, a malign socio-political consequence  – reinforced class hierarchies – is “masked” by sensible recommendations. 59

Alex Barber and Robert J. Stainton

A closely related and equally subtle way in which prescriptions can enforce unjust hierarchies is through the selection of norms that ensconce prestige dialects, on the specious grounds that they serve as a much-needed lingua franca. The fact that some are born into that “standard dialect” while others (speakers of African American English, say) have to adapt to pass as linguistically normal, is overlooked. Only those outside the linguistically and otherwise privileged group are called on to “remediate”. A discussion of the extent to which the descriptivists’ diagnosis of moral/political failings is justifed would take us too far afeld. We refer readers to the other articles in this volume for related discussion. We limit ourselves to this observation: “for the sake of the language” should not trump respect for the social and political well-being of oppressed groups. In the politics of gender and transgender language, for example, English pronouns are not a stakeholder. It’s reasonable to complain that linguistic prescriptivists downplay this: Dummett, for example, is concerned that by replacing “mankind”, etc., with “humankind”, etc., we risk “inficting damage on what has come down to us from the past” (1993: 110). If all this is along the right lines, we now have our “and others like them”. What the pronouncements of stereotypical prescriptivists have in common, it seems, is that they refect illinformed assumptions about language, often driven (intentionally or not) by a dubious political agenda. On the face of it, then, our two goals have been met. We have not only identifed what linguistic prescriptivism is, we have evaluated it – very negatively. Not so quick, we say. By unpacking both “such and such examples” and “others like them” in this way, what we have really sketched does indeed probably capture how anti-prescriptivists understand their target. Moreover, the criticisms typically made against this target – criticisms we have been sampling here – seem just (though a fuller discussion than we can give here would be needed to make that charge stick). But this whole approach fails to set up the debate satisfactorily. For a start, the view that we should follow ignorant and pernicious dictates about language is patently indefensible. It doesn’t merit careful philosophical evaluation. In addition, self-described prescriptivists would reject this characterization as inaccurate and unfair. If they were invited to say what “such-and-such examples and others like them” came to, they would point to a cluster of what they would describe as “useful linguistic distinctions” or “valuable grammatical rules”. On both fronts, this is hardly a promising platform for philosophical discussion. We also reject this frst conception of linguistic prescriptivism because it describes not so much a thesis as an attitude or tendency, or even an emotional disposition. A dismissive description of prescriptivists as annoyingly supercilious, as ruled by emotion rather than calm enquiry, seems to be about building a caricature rather than developing a thesis deserving of consideration. (We’d add that passion per se is somewhat of a red herring. Dispassionate prescriptivism is not a contradiction, and anti-prescriptivists can be passionate too.)

3 Second Attempt: Linguistic Prescriptivism as a Tesis Ideally we want a thesis, one it will be instructive to evaluate philosophically but in any case one that serves as a marker of the “prescriptivist” side of this heated dispute. Let’s call this posited thesis “Linguistic Prescriptivism” (big L, big P). We now canvas a series of possible statements of it. As per our ongoing strategy, we begin with something not especially promising: Linguistic Prescriptivism1: There are norms governing language use, and they should be followed. In favour of this, prescriptivists will at least recognize themselves as advocates of it. That is a step up from where we were heading in Section 2. Unfortunately, both in common parlance and when 60

Linguistic Prescriptivism

it is being poopooed by linguistically informed academics, our target thesis cannot amount merely to “There’s a way language should be used”. Just about everyone (rightly, we suggest) agrees to the applicability of a wide range of norms on language use. These include a moral prohibition on at least some out of: lying, insinuating known falsehoods, speaking out of ignorance, ad hominem attacks, unwarranted rudeness, propaganda, bullshit, coercion, linguistic oppression, gratuitous insults, hate speech, and inciting genocidal violence. They also include the conversational norms without which linguistic exchange would be vastly more challenging, such as Grice’s “be relevant” (1989), or proscriptions on defective concepts. Perhaps what sets prescriptivists apart is a more specifc concern with the choice of linguistic “vehicle” (the wrong word, the wrong spelling, the wrong grammar, etc.) to express a given content. This yields: Linguistic Prescriptivism 2: There are norms governing not just what things you say but how you say them; these norms should be followed. This captures the bugbears. Nonetheless, it is still far too wide. There are many attitudes to language use that ft this defnition without constituting what we would ordinarily call “prescriptivism”. Some word choices, for example, are resisted as very generally ofensive by all reasonable parties and not only by prescriptivists. Slurring words come immediately to mind. Following Diaz-Legaspe et al. (2019), we take these to have the same propositional content as their neutral correlates: begging pardon for the nasty term, “chink” and “Chinese person” are true of the same set of individuals. The diference is a matter of sociolinguistic register: slurring words are typically marked as [+ vulgar, − polite, + derogatory], whereas their counterparts are not. By our lights, then, what is said with a slurring term need not be disrespectful, but how it is said will always be, setting aside the notorious problem cases of re-appropriation and mention. A non-prescriptivist can also hold that certain other forms of language are vaguely wrong most of the time, albeit not morally so: impressive sounding but obscure or vague terms, for example. Some forms of language are arguably simply unattractive, such as repetitive phrasing, cliché, and “management-speak”. A non-prescriptivist can surely lament, as violations of good style, these “ways of saying it”.4 Sociolinguistic register gives us another set of examples, this time highly context-sensitive: phrasings that, though not inevitably or even typically objectionable, are inappropriate in certain conversational situations by nearly anyone’s lights, prescriptivist or not. Consider pub-worthy register in an academic essay, or a medical professional asking if you wouldn’t mind “having some shit grubbed from your ass” rather than asking to “take a fecal swab”. (Notice that this last example has a moral tinge: use of [+ vulgar, + slang] terms is impolite and disrespectful to the interlocutor, given this discourse situation. We revisit the topic of rudeness and disrespect below.) Finally, several of Grice’s (1989) maxims of manner apply precisely to how something is said (choose unambiguous wording, be orderly, be brief, etc.). But not all Griceans – and not all conversants – are prescriptivists. Perhaps we need to recognize that the norms linguistic prescriptivists claim to champion are the rules of the language. They aren’t concerned merely with what and how we communicate, but with the “linguistic code” – with the tongue itself, rather than “speakings”. Just as moving a rook diagonally is wrong because it violates a rule of chess, those who object to split infnitives may claim to be upholding a rule of English. This thought gives us: Linguistic Prescriptivism3: The language itself, or some endorsed “standard” version of it, is constituted by specifable norms, and failure to acknowledge and respect them constitutes being mistaken. 61

Alex Barber and Robert J. Stainton

This frames the prescriptivism debate as a debate over whether the rules of proper English permit “Who’d you buy it for?”, “new-kyu-lurr”, etc. A strength of this suggestion is that it fts with both pro- and anti-prescriptivists’ understandings of what is at stake. Thus, it meshes with a familiar charge against prescriptivists: that they naively assume there is such a thing as “the language” (Spanish, German or what-have-you). It squares with prescriptivists’ customary insistence that some standard form must be learned and respected, or “that one variety of language [the ‘standard’ variety] has an inherently higher value than others, and that this ought to be imposed on the whole speech community” (Crystal 2010: 3). It also refects characteristically “prescriptivist” concerns over the alleged decay, unrestrained change, drift, or infection of the language. Frustratingly, though no longer obviously inadequate, this remains too lax. Defenders of our Linguistic Prescriptivism 3 would include many one would not ordinarily classify as prescriptivist. A bona fde non-prescriptivist might nonetheless argue, in a Davidsonian or Wittgensteinian spirit, that speech is a purposive, essentially rational activity, and that linguistic meaning is inherently normative. That would seem to satisfy Linguistic Prescriptivism 3. Indeed, nearly everyone reading this article, in their day-to-day practices, is signed up to this thesis in virtue of correcting the lexical and grammatical mistakes of non-native speakers, and commenting in student essays on spelling, word choice, and grammatical structure, or indeed in redrafting their own work. The whole point of such practices is to reinforce or at least conform with existing norms and standard forms. Prescriptivists caricature the descriptivist as holding that “anything goes” (cf. Dummett 1993: 9; Hefer 2011: xv; Truss 2003: 4) but their foe here is imaginary. Whatever they preach, descriptivists issue corrections all the time – just not the ones issued by prescriptivists. Could it be that, while everyone’s practice is to uphold present norms, only the prescriptivist insists loudly on the importance of doing so and expresses unwillingness to countenance language change, perhaps on grounds of maintaining traditions? Again no. This modifcation would not rescue Linguistic Prescriptivism 3 because many of the people loosely identifable as prescriptivist acknowledge the importance of language evolution. We already saw both Dummett and Heffer doing so, albeit a little disingenuously. Moreover, historical prescriptivists such as Jonathan Swift (1712) and the political radical William Godwin (1810) were as concerned with fxing or “optimizing” the rules of English as with ascertaining and maintaining the existing rules. The same goes for Dryden in his proposal for avoiding preposition stranding. And in modern times, reforming language has been a clarion for one of the stereotypical “bad guys” of prescriptivism: L’Académie Française regularly introduces and promotes neologisms to discourage borrowings from English (“baladeur” instead of “Walkman”, “ordinateur” instead of “computer”). Let us take stock. In pursuit of Goal 1 we drew attention in Section 2 to a range of stereotypical cases, a heated debate, and some suggestive societal attitudes (and name-calling). This helped us to recognize who the combatants are but didn’t give us a useful defnition. In this section we have been attempting to formulate a thesis, “Linguistic Prescriptivism”, advocacy for which serves to mark out those who self-identify as and are described by critics as linguistic prescriptivists – the “language champions” or the “language mavens”, as it may be. None of our three attempts has succeeded. Maybe the fault is ours. Maybe there is such a thesis, but we have yet to discover it. This isn’t wholly implausible given the additional options that come readily to mind. Mightn’t a linguistic prescriptivist be a person who holds that a purely descriptive science of Linguistics is neither possible nor desirable? Or mightn’t they be singled out as holding that it’s possible for a majority of native speakers to misapply a term (meaning that linguistic facts fail to supervene on the behaviour of the majority of speakers)? In both cases, probably not. Against the frst, we will see below that even non-prescriptivists might hope to glean facts relevant to language reform from the scientifc study of human languages. Against the second, consider a philosopher who thought (in the vein of Kripke/Putnam) that expertise or causation play a central role in fxing reference. An upshot 62

Linguistic Prescriptivism

of their position might seem to be that, outside the ken of its users, the majority practice may fail to determine meaning (as in e.g. “libel” vs. “slander”, or “quark”, or “aster” [“star”] among the Ancient Greeks). Such a philosopher is not ipso facto on the “prescriptivist” side. We incline towards a more radical stance: there simply is no proposition, Linguistic Prescriptivism, held by all and only those stigmatized/valorized as “prescriptivists” and ripe for philosophical evaluation. We incline towards this not simply on the egocentric grounds that we have failed to identify it. Our suspicion as based on principled reasons. One is that “linguistic prescriptivist” seems to be a loaded label for which no neutral characterization is possible (see Cameron 2012: 8). For critics, it functions as a boo-word while for supporters it is a hooray-word. (This would explain why normative proposals about language which strike orthodox linguists as acceptable are thereby discounted as not genuinely prescriptivist.) A second reason is that, as noted earlier, prescriptivists prescribe an unholy mixture of anodyne good sense and the highly specious. In one chapter one is being told to write clearly and respect the genre of your text; in the next, one is being told how important it is to use (or not use) the Oxford comma. Unsurprisingly, then, it has proven frustrating to isolate exactly the things prescriptivists do in a way that isn’t either circular, dismissive, or contentious. In this respect, our frst attempt (Section 2) was perhaps on the right track after all: rather than seeing prescriptivism as the expression of some univocal thesis, we should recognize it as a social phenomenon or practice, engaged in by some and bemoaned by others. A useful comparison might be made with “populism”. Diferent populist fgures, it is true, tend to adopt thematically similar stances – nationalism, victimhood, etc. – but one would not try to defne the term simply by distilling these postures into a unifying thesis. Populism is something certain people engage in under certain conditions with certain efects. Happily, there is a philosophically interesting thesis in the same neighbourhood. We encountered it above, as Linguistic Prescriptivism1 (“There are norms governing language use, and they should be followed”). Why did this fail to capture the prescriptivist viewpoint, pertaining as it does to prescriptions about language? Because the label is simply too freighted to admit of such a straightforward characterization. (Compare the fact that, nowadays, “holding politically accurate viewpoints” fails miserably to explicate “politically correct”.) Our suggested expedient is to hold fast to the defniens and abandon the defniendum. That is, we propose dropping the label “linguistic prescriptivism” and talking instead of “language norming”, by which we mean to include any attempt to evaluate language use on normative grounds. This includes the stereotypical prescriptivist attempts, certainly, but not only them. Making this shift takes us immediately to where we want to be: with an interesting evaluative question. When, if ever, is language norming a defensible pursuit? On the standard view in linguistics, the stereotypical examples of linguistic prescriptivism we opened with in Section 2 are not defensible instances of language norming. But that hardly shows there aren’t other instances that are defensible. Answering this question is (modulo the verbal shift) our earlier Goal 2. A comprehensive treatment is well beyond the reach of this article. As we announced at the outset, this article is an attempt to get a philosophical conversation started. In that same spirit, in Section 4 we survey a range of cases of what we regard as acceptable (or not obviously unacceptable) language norming. If nothing else, we hope to illustrate the sudden ease with which it becomes possible to discuss normative topics once we drop the “prescriptivism” label (and the invective) and move away from the paradigm cases.5

4 Is Language Norming Ever Defensible, and If So, When? We begin with two general worries about a blanket rejection of all language norming (i.e. not just of “traditional prescriptivism”). First, we note that if such a rejection were motivated by hostility to stereotypical prescriptivism it would be hypocritical, since even prescriptivism-bashers engage 63

Alex Barber and Robert J. Stainton

in norming to some degree. The imperative in the title of Robert Hall’s (1950) anti-prescriptivist lament, Leave Your Language Alone!, is itself an instance of language norming. The view that it is permissible to split infnitives is every bit as normative as the view that it is impermissible. Second, and more speculatively, we worry that self-avowedly “neutral descriptions” of language often mask judgements of superiority that tacitly underwrite discrimination. However unwittingly, even “descriptive” dictionaries may efectively denigrate non-standard dialects. To come at it via an analogy, one might repudiate allegedly “socio-politically neutral” sciences of criminology or economics – since, plausibly, it’s naive to suppose that there can exist such things. If norming really is unavoidable, it’s better to do it openly and properly. This is not to say there cannot be descriptive, scientifc aspects to linguistics; it is merely to warn that there will potentially always be more to it. We turn now to describing some potential instances of defensible language norming. All are subject to caveats and pleas for restraint, but each one has at least prima facie credibility. For expository ease rather than out of a desire to make deep distinctions, we divide them into aesthetic, practical, epistemic, moral, and fnally political. Beginning with aesthetic norms, many, and we include ourselves here, will recognize that there exist objectively good and bad styles. Rev. Martin Luther King’s verbal fourishes are striking and stirring. W.V.O. Quine’s prose stylings set him apart as a superior philosophical writer. In contrast, management-speak and cliché show how some language can be fatly ugly. What’s more, one may well cherish the code itself, not just the creative deployment of it: Welsh recitation trades on the distinctive prosody of that very language, and translations, particularly of verse or other artistic forms, typically miss something present in the original. Specifying what exactly is given up when aesthetically valuable language is foresworn is notoriously difcult. An analytical rendition of lines from a poem (e.g. from “Theirs not to make reply, / Theirs not to reason why, / Theirs but to do and die” in Tennyson’s poem, Charge of the Light Brigade, to “It wasn’t up to them to dispute their orders or to ask why they were given; their job was simply to act and to die”6) clearly loses something, but what? That question presumably overlaps substantially with parallel questions we might ask about dance, music, or design. For those who are sceptical about such considerations, we ofer a diferent kind of aesthetic example. We presume that there is nothing globally wrong with culturally specifc rules of etiquette, e.g. social norms about which knife to use with which dish (a curved and broad blade for fsh), what dress is appropriate for what occasion (avoid bright pastels at funerals and button-down shirts at the beach). Other options aren’t illogical or immoral: the rules that persist are just the ones that people happen to follow. Nor is every violation inexcusably wrong. One may, to make a point, wear a bikini at a thesis defence or blue jeans at a wedding, or eat peas of a knife (cf. Thackeray 1848: 5). But insofar as norms of manner are defensible in these other contexts, manners in language are too. Indeed, our earlier example of apt register in a medical setting seems to be a case in point. “We, like, seen ’em all, like, shitting n’ stuf” may not be normatively unacceptable across the board, but it is when composed by a solicitor in a legal brief or inserted in an academic essay or a letter to one’s employer. As with politeness in other contexts, linguistic politeness norms can potentially become shibboleths, a badge of membership and so a means of exclusion. But that only shows the need for a balance to be struck. Professional contexts should be neither a bear-pit nor like the court at Versailles. Turning to practical language norms, the broad thought here is obvious: it is useful to have a uniform linguistic standard for knowledge transmission and for the coordination of action. This justifcation has the same form as for any other common standard: we need a standard; current practice should serve as a default; current practice is X; so, all else equal, we should respect X. Indeed, attempts to establish a common linguistic standard are familiar and many seem perfectly unobjectionable. In the course of writing this very article we have had to verify American spellings 64

Linguistic Prescriptivism

in dictionaries; and we have been told that, for the volume as a whole, authors should adopt double quotes for mention and APA style for references. Defeasible standards of this kind can have a clear practical function. A consistent use of “billion” across the English-speaking world is hardly a “minor instance of cultural rape” just because the American-English meaning won out over the British-English one (Dummett’s description in 1993: 112). Linguistic subcultures – academic philosophy comes to mind – are full of terms with agreed-upon meanings, norms that are genuinely useful to those in that subculture and not merely as a kind of membership badge. Now for the caveats. Not all claims made in the name of usefulness stand up to scrutiny. We saw Hefer taking himself to be upholding the highest standards, and Dummett seemingly took it upon himself to invent a standard of his own. Recall too the dangers underscored by antiprescriptivists. Despite his admirably liberal views on other matters (2001), Dummett’s production of a heavily prescriptive style guide aimed at undergraduates at an elite establishment university arguably supplies nourishment to the British class system, disguised behind claims about having a responsibility to the language and to future generations (1993: 8). Thus, any given practical prescription demands credible and specifc arguments, with due concern for unintended efects. A second caveat is that that the scope for setting explicit standards for the entire language community is necessarily limited. These things tend, in the main, to take care of themselves. If a phrase is genuinely important it will typically survive or else be replaced by an equivalent (e.g. “circular reasoning” for “beg the question”). The relatively new term “proactive” annoys some, but its survival will probably turn on its utility (in forming a contrast with “reactive”, say), not on anyone’s lobbying for or against. When a usage fzzles out, that probably just means it has ceased to matter enough. This is why the loss of the distinction between “practical” and “practicable” is not the tragedy Hefer implies (2011: 160). A fnal caveat is that any practical standard needs to be change-tolerant. New modes of expression will arise as communication media evolve, for example, which they have at least since the invention of the telegraph. Indeed, as the world as a whole evolves, and humans continue to express their creativity, or to talk across linguistic boundaries, linguistic practice will force changes on any standard. As noted, the Académie Française has managed to root out some Anglicisms but the attempt to resist all borrowings has been doomed. (“La fn de semaine” for weekends is used in North America, but the standard in Europe is “le week-end”.) Practical standards, then, must be invoked with specifc justifcations and due modesty. We turn now to epistemology. Putting a word to something, as with “slut-shaming” or the “proactive/reactive” contrast, can sometimes help us understand our world more easily. So, there may be cases where reform, in the guise of introducing a novel usage, is knowledge-conducive. The repeated and overt choice of “partner” over “spouse”, for example, can draw conscious attention to the existence of equally valuable marriage-like relationships or to the non-availability of marriage as an option for homosexual couples. The caveat this time is that, as we suggested earlier, the cognitive impact of the lexicon has sometimes been overhyped (recall our “dedos” example). The Sapir-Whorf hypothesis in its strong form has been empirically falsifed. Thus, it’s a mistake to insist, without evidence, that losing “that/which” or “who/whom” will have a negative epistemic impact. Again, language norming can serve an epistemological end, but where it does so must be decided on a case-by-case basis. For moral norms governing language use, we suggest that, depending upon context, using the familiar second-person pronoun tu in French with an unfamiliar elder, as opposed to the more formal vous, may signal disrespect (likewise for du/Sie in German, tū/āp in Hindi, and ti/chi in Welsh). To the extent that it is sometimes morally wrong to signal disrespect, “Do not address unfamiliar elders with tu” is a legitimate prescription. English no longer contains the same distinction, but similar examples may be achievable through the example of titles. Consider the student who, having never met his professor, writes her an email with the salutation “Hey bud!” or “Dear 65

Alex Barber and Robert J. Stainton

Miss ——” despite awareness of the linguistic norm that anything short of “Dear Professor ——” would signal disrespect. Now for the tiresome caveats. The precise relation between etiquette and morality is controversial, and our understanding of it may be in fux as West meets East (Bell & Pei 2020: 1–8; Olberding 2016). These proofs-by-example therefore come with qualifcations. One is an assumption we are making but will not try to justify, namely, that while etiquette rules are not the same as moral rules (they have an arbitrariness that moral rules lack, for example), once they are in place and legitimately so, they have the power to generate moral prescriptions. Purposely disrespecting them is rude, impolite, and hence to some degree morally blameworthy. The second is an allowance: some etiquette rules lack legitimacy, in which case it may be acceptable to violate them as a way of signalling disagreement. A French speaker who dislikes the way the “tu/vous” distinction operates might reason in this way. (That same battle took place in English in the past. Quakers were mocked for insisting on the familiar “thou” form for everyone, irrespective of status. Ironically it was this familiar form that died out, in a kind of levelling-up that left just “ye” or “you”.) And among those who argue for the elimination of titles, some would include professional academic titles on the list. Finally, prescriptions of this kind are, like most moral rules, defeasible at best: in some circumstances, rudeness may be warranted. Linguistic appropriateness often depends on extra-linguistic considerations. Turning fnally to political cases, the defensibility of language norming has already emerged thanks to the example of progressive reforms: promoting gender-neutral language (avoiding “mankind” for humankind, permitting “they” as gender-neutral singular), for example, and excising slurring words. Discussion of these and other examples (e.g. applying conceptual engineering to defective concepts; decrying fg-leaves and essentializing generics) can be found throughout this volume. Such cases confrm that language norming is not a conservative phenomenon and is not a progressive phenomenon. It cuts across the political spectrum.7 Taking of from such examples, we note in passing that even an earnest descriptivist might reasonably hold that the science of linguistics, and not armchair linguistics, can and should serve as a departure point for informed social change. As an example, granting that certain forms of language should be excised, descriptive linguistics might tell the politically progressive how best to achieve this: e.g. feld work or tracking of corpora might tell the progressive reformer what, as a matter of fact, tends to rapidly and thoroughly induce change (e.g. avoidance by trend-setters) and what does not (e.g. top-down proselytization in academic settings). Compare criminology or epidemiology: the descriptive science might reveal what causally underpins racial disparities in incarceration or the rapid spread of viruses, and hence how governments that value the elimination of these efects might best react. There is nothing nefarious about understanding those disciplines as compatible with normative evaluation. Likewise, language norming is perfectly compatible with the practice of linguistics as a science. Another potential (and admittedly tentative) defence of language norming would see it as a prophylactic against an overreaching individualist libertarian view on language. Here we envisage someone taking a rabidly anti-prescriptivist stance, reasoning that no language is superior to any other, and that people should be allowed to choose whichever one they prefer: given the “arbitrariness of the sign”, choice of language per se is value neutral. Linguistic change, or linguistic uniformity for that matter, is thus always fne in itself as long as no coercion is involved. After making some obvious practical concessions (no rational agent will choose to use “dog” to refer to applesauce, or prefer “He sleep is quiet no” over “He is sleeping quietly”), this unfettered anti-prescriptivist will hold that no ofcial, paternalistic norming is permissible. Such an unqualifed stance would have unhappy consequences, at least by our lights. In particular, this anti-normative line of reasoning would preclude any laws aimed at promoting and preserving minority languages, or indeed treasuring other forms of linguistic diversity. 66

Linguistic Prescriptivism

More than that, consonant with extreme libertarian principles, it might prohibit the use of tax dollars towards such ends. This would jeopardize measures like those in Canada, where legislation has successfully re-established French as a lingua franca, halting an emerging hegemony of English. Quebec’s Law 101, for example, required children of immigrants to go to French-medium schools rather than to English-medium schools, and forced commercial signage to be in French or prominently in French. Softer measures in Ireland and Britain are helping preserve Welsh, Irish and Scottish Gaelic as living, evolving languages, though these measures are often resisted by an alliance of quasi-libertarian thinking and the anti-prescriptivist thought that, since there is nothing wrong with English, nature should be allowed to take its course. To come at the point another way, natural languages are, among other things, cultural products. So, it is possible to treasure and to want to protect a given tongue, without supposing that it is better than other languages – just as one can prize a distinctive cuisine. Diversity of cultural goods might in itself be valuable, and the persistence of a particular cultural tradition can be an aspect of the welfare of those whose culture it is. Likewise, diferent ways of speaking should be valued and respected, not nudged out of existence or into museums. Where necessary, diversity should be promoted on grounds of social justice – including measures to protect language environments. This, however, is a kind of language norming.

5 Summary and Conclusion Given the relative silence among philosophers on the topic of linguistic prescription, we began our exploration of the terrain with a modest goal: Goal 1: Uncover a plausible interpretation of “linguistic prescriptivism”, ideally one under which it merits careful philosophical evaluation. This proved surprisingly difcult. Our frst gambit involved identifying some paradigm examples: the ones self-avowed prescriptivists seem most exercised about. We then drew on standard criticisms by descriptive linguists to fnd a unifying feature of these cases. This frst attempt was manifestly fawed. It perhaps captured what prescriptivism’s foes have in mind as their target – a passionate, misguided promoter of the asinine – but it cannot serve as a neutral account of the label’s meaning. Nor does it yield a viewpoint worthy of philosophical assessment. This frst sally thus uncovered, at best, a social phenomenon, a practice stoutly defended by some and decried by others. Our second approach, then, was to consider whether there was any unifying thesis, Linguistic Prescriptivism, acceptance of which marks out those commonly agreed on all sides to be linguistic prescriptivists. We moved through a series of gradually more promising possibilities. Each one, however, turned out to be too broad (since denying that lying and inciting violence are acceptable uses of language, celebrating concision and clarity of expression, or insisting on correct spelling in journal articles, hardly makes one a prescriptivist in the normal sense) or else too restrictive (since prescriptivists do not resist all language change). At that stage we sort of gave up on Goal 1. We concluded that no thesis is endorsed by, and only by, those who ft the label “prescriptivist” as it is ordinarily used. The fundamental problem, we decided, lies with the term. It is valanced (negatively for some, positively for others); the recommendations associated with it intertwine the obviously unobjectionable with the silly and lamentable; and it captures a social activity more than a theory. Only “sort of ”, though, because there is a perfectly clear proposition in the neighbourhood: that there are norms governing language use, and they should be followed. This took us to where we wanted to be – with an interesting question – but required us to rechristen our topic (“language norming” instead of “linguistic prescriptivism”), if only to distance ourselves from the justly tarnished stereotype. 67

Alex Barber and Robert J. Stainton

We then proceeded to Goal 2, reworded in efect to: Goal 2: Consider whether any variant of linguistic prescriptivism, thus understood, language norming is defensible. We pursued two avenues. First, any all-purpose rejection of language norms, we tentatively suggested, would be both hypocritical, and underappreciative of the inevitability of norming. Second, again briefy and guardedly, we ofered initially plausible examples of appropriate norming: aesthetic, practical, epistemic, moral, and political. The overall lesson is that one should drop the standard perfunctory dismissal: there is a nonsilly thesis, a philosophical cousin of linguistic prescriptivism in the usual sense. Language norming is acceptable, and the interesting question is just: when? Rarely, it turns out, if we limit our attention to the stereotypical complaints of linguistic prescriptivists. But we suggested a range of cases where language norming looks attractive. A fuller defence of any of them must turn on extra-linguistic considerations (political outlook, views on etiquette, etc.). But on the face of it, and drawing on the expertise of descriptive linguists, philosophers can and should undertake language norming, albeit with due caution and on a case-by-case basis.

Acknowledgements For extremely helpful feedback on a presentation of draft material, our thanks go to: Derek Ball, Axel Barceló, Andrew Brook, John Collins, Sandy Goldberg, Steven Gross, Robyn Jeshion, Robert Matthews, Eliot Michaelson, Ethan Nowak, Jennifer Saul, Adam Sennet, and Nicole Wyatt. We are also grateful to the Philosophy Department, Dalhousie University, for inviting us to address their weekly colloquium in the summer of 2020. Their comments and questions also greatly improved the paper. We count ourselves very fortunate to have such clever and insightful friends and colleagues.

Notes 1 For a history of prescriptivist battles over English since the Anglo-Saxon period, see Crystal (2006). For a nuanced discussion of prescriptivism today, see Cameron (2012). 2 The topic is addressed a little more substantially by philosophically minded linguists (see, e.g., Huddleston & Pullum 2002; Liberman & Pullum 2006; Lyons 1981; McWhorter 2008; and Pinker 1994). Among philosophers, Michael Dummett has written a prescriptivist guide to English (1993), consonant with remarks he makes in his philosophy proper (1981: 142, 1991: 88–92). This guide fgures in our discussion. R.M. Hare (1952), we note, draws a contrast between descriptivism and prescriptivism in metaethics; the contrast of the same name in linguistics arose around the same time, but any link seems to end there. 3 All three titles, in fact, were originally variant abbreviations of the title “Mistress”, which was as neutral on marital status as “Master”. The “Mrs.” form frst mutated into a marker of social rather than marital status (Erickson 2014: 45). 4 Steven Pinker, while hardly a friend of prescriptivists (see his 1994), nonetheless tries to rescue the norms of “good writing” from them in his 2015 work. 5 Deborah Cameron, with whom we are in substantial agreement here, countenances a more vivid alternative, “language hygiene” (2012), but for our purposes this has an unwanted negative whif. The same goes for “language ideologies” (Schiefelin 1998). 6 The rendition is from, retrieved 03.08.20. 7 Anti-progressives, moreover, have increasingly co-opted the “linguistic prescriptivism” charge for use against proposals of this kind. We see this as further reason to drop what has become, in essence, a confusing boo term.


Linguistic Prescriptivism

References Bell, Daniel A. and Pei, Wang. (2020) Just Hierarchy: Why Social Hierarchies Matter in China and the Rest of the World. Princeton, NJ: Princeton University Press. Cameron, Deborah. (2012) Language Hygiene. Oxford: Oxford University Press. Crystal, David. (2006) The Fight for English. Oxford: Oxford University Press. Crystal, David. (2010) Cambridge Encyclopedia of Language, 3rd edn. Cambridge: Cambridge University Press. Diaz-Legaspe, J., Liu, C., and Stainton, R.J. (2019) ‘Slurs and register: A case study in meaning pluralism’. Mind and Language, 53(2): 156–82. Donne, John. (1638 [1608]) Biathanatos. London: Humphrey Moseley. Dryden, John. (1808 [1668]) The Works of John Dryden, Vol. 4. Ed. W. Scott. London: William Miller. Dummett, M. (1981) Frege: Philosophy of Language, 2nd edn. Cambridge, MA: Harvard University Press. Dummett, M. (1991) Logical Basis of Metaphysics, Cambridge, MA: Harvard University Press. Dummett, M. (1993) Grammar and Style: For Examination Candidates and Others. London: Duckworth. Dummett, M. (2001) On Immigration and Refugees. Abingdon: Routledge. Erickson, A. L. (2014) ‘Mistresses and Marriage: Or, a Short History of the Mrs’. History Workshop Journal, 78: 39–57. Godwin, W. (1810) Outlines of English Grammar. London: Juvenile Library. (Under the pseudonym Edward Baldwin.) Grice, H. P. (1989) ‘Logic and Conversation’. In his Studies in the Way of Words. Cambridge, MA: Harvard University Press. Pp. 22–40. Hall, R. (1950) Leave Your Language Alone! Ithaca, NY: Linguistica. Hare, R. M. (1952) The Language of Morals. Oxford: Clarendon Press. Hefer, Simon. (2011) Strictly English: The Correct Way to Write and Why it Matters. London: Windmill. Huddleston, Rodney. D., and Pullum, Geofrey. K. (2002) The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press. Liberman, Mark and Pullum, Geofrey. K. (2006) Far from the Madding Gerund. Wilsonville, OR: William, James & Co. Lyons, John. (1981) Language and Linguistics. Cambridge: Cambridge University Press. McWhorter, John. (2008) Our Magnifcent Bastard Tongue. New York: Penguin. Olberding, Amy. (2016) ‘Etiquette: A Confucian Contribution to Moral Philosophy’. Ethics, 126(2): 422–46. Pinker, S. (1994) The Language Instinct. New York: William Morrow. Pinker, S. (2015) The Sense of Style: The Thinking Person’s Guide to Writing in the 21st Century. New York: Penguin Books. Schiefelin, Bambi B., ed. (1998) Language Ideologies: Practice and Theory. Oxford: Oxford University Press. Swift, J. (1712) A Proposal for Correcting, Improving and Ascertaining the English Language. London: Benjamin Tooke. Thackeray, William Makepeace. (1848) The Book of Snobs. London: Punch. Truss, Lynn. (2003) Eats, Shoots and Leaves. London: Profle Books.


5 SPEECH-ACT THEORY Social and Political Applications Daniel W. Harris and Rachel McKinney

The motivating insight of speech-act theory is that we do many things by speaking, but one can’t normally tell what someone is doing merely from the expressions that they utter. In the terminology introduced by Austin (1962), the locutionary act of uttering a given sentence with a given meaning does not determine the illocutionary act that one thereby performs. For example, someone who utters (1) might be describing local bylaws or issuing a command. For that matter, they may be joking around, speaking sarcastically, or acting in a play. (1) You can’t park your car there. Austin further distinguished illocutionary acts from perlocutionary acts—acts performed by performing illocutionary acts and whose nature depends on the downstream consequences of those illocutionary acts. By commanding someone not to park next to the hydrant, one might also perform an act of annoying them and an act of making them move their car, for example. So what does it take to perform an illocutionary act, and what makes it an illocutionary act of one kind rather than another? The central task of a theory of speech acts is to answer these questions, and the main contenders in speech-act theory give diferent answers. In §1, we will describe an ongoing debate between three families of answers, which ground the nature of illocutionary acts in conventions, intentions, and normativity, respectively. When conducted wholly within the philosophy of language, the debate between these views can seem abstractly metaphysical. In §2, we will survey a range of issues in social and political philosophy that can be proftably understood as applications of speech-act theory. We will consider the roles of speech acts in the social contract (§2.1), the nature of laws (§2.2), the nature of social norms and practices (§2.3), the mechanisms by which speech may be silenced (§2.4), and freedom of speech (§2.5). It will turn out that debates about the roles of conventions, intentions, and norms show up in many of these venues. Our contention is that the intricacies of debates about speech acts are actually widely applicable to the social and political realms.

1 Speech-Act Teory The three theories of speech acts on which we will focus all emerged in the 1950s and 1960s, and remain infuential today. The frst, originally developed by J. L. Austin (1962, 1963, 1970), is that illocutionary acts are conventional procedures whose conditions of felicitous performance are defned by localized social conventions. Diferent illocutionary acts, on this view, are performed by acting in accordance with diferent linguistic or social conventions. The second view, originating in the work of Paul Grice (1957, 1968, 1969) and translated into the idiom of speech 70

Speech-Act Theory

acts by P. F. Strawson (1964), is that many illocutionary acts are performed by acting with overt, audience-directed intentions. Diferent illocutionary acts are distinguished by the fact that they are performed with intentions to change others’ minds in diferent ways. The third view, which originates in the work of Wilfrid Sellars (1954, 1969), is that speech acts (along with intentional mental states) should be understood in terms of their functional roles in broader patterns of “norm-conforming behavior”—activities that are constitutively governed by social norms (Sellars 1954, 204). On this view, diferent illocutionary acts are governed by diferent norms and give rise to normatively diferent outcomes.1 Rich traditions have followed in the wakes of these theories. Conventionalists have followed Austin in taking illocutionary acts to be primarily a matter of convention, though latter-day conventionalists have often placed greater emphasis on narrowly linguistic conventions rather than the broadly social conventions emphasized by Austin.2 Conventionalism has some obvious things going for it. We normally do perform speech acts by exploiting linguistic conventions. And many of the speech acts emphasized by Austin, such as the act of marrying someone or christening a ship, really could not be performed except against the backdrop of complex social conventions and institutions. It may be similarly tempting to think, as Searle (1965) infuentially argued, that, for example, asserting that you have a PhD in philosophy just isn’t something that could be done without the aid of convention-governed linguistic expressions. Intentionalists tend to agree that Austin was roughly right about the institutionalized acts on which he focused, but argue that Grice was right to think that the basic communicative acts, such as asserting, requesting, and questioning, are not essentially conventional. 3 They are, rather, exercises of basic human capacities for inferring and shaping others’ states of mind. They have ofered several infuential arguments against across-the-board conventionalism. One is that we often communicate indirectly and nonliterally, so that what we mean diverges from or goes beyond the conventional meaning of what we utter. When Romeo says to Juliet that death “hath sucked the honey of thy breath”, it is only by going beyond the conventional meanings of “sucked” and “honey” that Romeo means what he does.4 Moreover, even when we are speaking literally and directly, our conventions don’t fully determine the content and force of an illocutionary act. Our example (1) illustrates this point, but so do arguably all other sentences, given the ubiquity of context-sensitive expressions and the possibility of using any sentence in a normal interaction, in a joke, when acting in a play, and so on. 5 Finally, as Grice (1957) took pains to illustrate, we seem able to regularly perform communicative acts by nonlinguistic and unconventional means. Indeed, it would seem necessary to suppose that communication can happen in the absence of conventions in order to explain how those conventions are created and acquired by children in the frst place (Harris 2016). For these reasons, intentionalists tend to think that although linguistic convention plays an important role in allowing speakers to give evidence of their intentions, communicative acts are not inherently conventional. The third major family of theories tells us that illocutionary acts are fundamentally normative. One kind of theory in this family, which owes much of its popularity to Williamson (1996, 2000), analyzes illocutionary acts in terms of the norms that constitutively govern them. Although these theories are quite infuential, especially among epistemologists, only assertion has received detailed attention. Williamson argues that assertion is constitutively governed by the “knowledge norm”, according to which one should assert only what one knows. On Williamson’s view, this is not merely a norm that happens to govern assertions; being governed by this norm is what makes a speech act an assertion. His most infuential argument for this claim is that the account explains why it is normally infelicitous to assert something of the form, “p, but I don’t know that p”. Few conventionalists and intentionalists have been persuaded by this argument, however, since it is possible to accept that assertion is governed by the knowledge norm while insisting that this 71

Daniel W. Harris and Rachel McKinney

follows from the intentional or conventional nature of assertion together with broader norms that govern all sorts of social interactions.6 A second kind of normative theory has sought to analyze speech acts, along with intentional mental states, such as belief, desire, and intention itself, in terms of the commitments and entitlements that they involve.7 For example, we might think of asserting p as the undertaking of a commitment to the truth of p—a commitment that entitles the addressee to take the speaker at their word, to ask for supporting reasons in case of doubt, and to rebuke the speaker if p turns out to have been false. Again, the most obvious motivation for this view is that speech acts typically do have normative consequences like these. Of course, we might again try to explain how these commitments and entitlements arise by appealing to a non-normative theory of a given speech act, together with an account of how commitments and entitlements arise from conventions  or communication more generally (see, for example, Harris (2019)). Another motivation for —normative views is to avoid the assumption—common to intentionalism and many versions of —conventionalism—that intentional mental states are explanatorily prior to, or more fundamental than, illocutionary acts. Brandom (1994) has argued that the intentionality of language and thought are on an explanatory par, so that neither should be used to give an account of the other. In place of such a reductive strategy, Brandom argues that we should explain the intentionality of both speech and thought in terms of underlying normative concepts. This motivation itself has often been a target of criticism. One problem is that the approach seems to confict with theories that attribute intentional mental states to nonlinguistic creatures, such as animals and young children. For example, many models of language acquisition tell us that young children learn the meanings of words in part by forming beliefs about other language-users’ beliefs and intentions (see, e.g., Bloom (2000, ch. 3)). Some critics have also worried that normativity is a dubious unexplained explainer in the context of theories of these kinds (e.g. Rosen (1997)). Speech-act theorists have found ways of abstracting away from their foundational disagreements, at least for some purposes. The most widespread method involves the idea that conversations revolve around discourse contexts, which are databases of information that afect how speech acts are performed and that are, in turn, updated by the performance of speech acts. The most infuential model of this kind is due to Stalnaker (1978, 2002, 2014), who takes a discourse context to be the set of propositions that interlocutors are treating as common ground for the purposes of their conversation. To assert p, says Stalnaker, is to propose adding p to the common ground. Others have generalized this idea to other speech acts. For example, Roberts (2004, 2012) thinks of asking a question as a proposal to add a new “question under discussion” to the context—the question that it is the participants’ immediate conversational goal to answer. And Portner (2004, 2007, 2012) models directives and permissions as proposals to alter the “to-do list”, which he thinks of a component of the context that tracks the preferences on which the participants have coordinated. An infuential generalization of these ideas is due to Lewis (1979), who argues that we should think of the context of a conversation as being akin to the scoreboard of a baseball game, tracking all facts about a conversation that may play a role in determining how it will unfold and that may be altered by the performance of further speech acts. Following Lewis, it is common to refer to a conversation’s discourse context as its “conversational score”, or “scoreboard”.8 These ideas are abstract enough that they can be made to ft with each of the competing theories of speech acts discussed above. On the intentionalist construal, the discourse context consists of the participants’ shared states of mind, and illocutionary acts are understood in terms of their intended efects on these states (Roberts 2018; Stalnaker 2018; Thomason 1990). The conventionalist interpretation thinks of the context as a social construct whose state is determined by the conventions governing a conversation together with the objective facts about its history—facts that may trump the intentions of speakers and the shared attitudes of the interlocutors (DeVault and Stone 2006; Lepore and Stone 2015; McGowan 2018). Within the tradition that thinks of 72

Speech-Act Theory

speech acts in fundamentally normative terms, the context can be thought of as the sum total of commitments, entitlements, and other normative facts that are currently in play in a conversation (Brandom 1994; Geurts 2019; Nickel 2013). Some theorists have argued that we need more than one of these conceptions of discourse context to coexist within an adequate theory of speech acts (Camp 2018). More often, discussions of discourse context, common ground, and conversational score take place at a level of abstraction that ignores these diferent possible interpretations. This can facilitate interesting debates across theoretical boundaries, but it can also mask deep theoretical disputes in ways that confuse some issues.

2 Social and Political Applications 2.1 Speech Acts and the Social Contract One of the major issues in the history of political philosophy is a question about speech acts: how can agents make promises to each other? More specifcally: how can agents credibly make binding commitments to each other, such that they have trust or assurance enough to motivate cooperation for mutual beneft? This question lies at the heart of any inquiry into the nature of contracts, which the contractarian tradition has placed at the center of theories of the nature of both political institutions and morality. Hobbes gives us one classic answer 9: because each individual is such that they are better of when they are able to cooperate and coordinate with others, and because each individual knows of others that this is true of them too, each individual has a prima facie motivation to make commitments that another agent has reason to accept as binding—hence, a promise. However, it is also common knowledge that agents will have incentives to break promises. Hobbes argues that it is therefore rational for self-interested agents to develop some system of sanction to back up their commitments to make them binding on pain of harm to reputation or punishment. Thus Hobbes renders the speech act of promising (which can be irrational to break) as the frst step in a series that can eventually lead to contract (which can be but is not always punishable upon breaking). The transition from promise (that is, binding commitment) to promise with a sanction upon defection (that is, assurance) is key. Hobbes thus proposes to bootstrap a social contract (something like laws of a commonwealth) out of mutually enforced promises: A man is obligated by an agreement, i.e. he ought to perform because of his promise. But he is kept to his obligation by a law, i.e. he is compelled to performance by fear of the penalty laid down in the law. (Hobbes 1984, 14.2n) Note what is absent so far in this story: any appeal to practices, conventions, customs, or prior rules. Hobbes may have answered how agents can in principle form credible binding agreements and how—in principle—such agreements can serve as the rational ground for something like a social contract, but so far such agreements appear to be made through one-of promises, entered into on the basis of the recognition of an individual speaker’s intentions, and backed by (perhaps fairly ad hoc) social sanction—harm to reputation or punishment by a sovereign. He appears to think such a structure is possible, stable, and enforceable absent anything like a convention of promise-making (though perhaps not absent a regularity of promise-making). Another way of putting the point: Hobbes thinks agents in the state of nature can promise. Hume famously critiques Hobbes on this point, arguing that in order for promises to be both credible and binding, agents must be party to a prior convention according to which an utterance of “I promise…” counts as an instance of a type of action recognizable even by third parties as 73

Daniel W. Harris and Rachel McKinney

generating commitment—hence other parties will judge that an agent who breaks a promise has done something to violate a rule, not merely something that happens to have disadvantageous consequences. I say, frst, that a promise is not intelligible naturally, nor antecedent to human conventions; and that a man, unacquainted with society, could never enter into any engagements with another, even though they could perceive each other’s thoughts by intuition. (Hume 1738 T, SBN 516) Here we see an early instance of what would become the debate between intentionalism and conventionalism about speech acts. This debate about the nature of what has come to be called “promissory obligation” picked up steam in the late 20th century, among both political philosophers and philosophers of language alike.10 One infuential conventionalist about promising is Rawls (1955), who—foreshadowing Lewis’ use of the same metaphor—describes the making of a promise as akin to a move in a baseball game. He reasons as follows. Consider someone who makes a promise and then— failing to do what is promised—tells the promisee that actually, upon refection, it wasn’t all things considered the best thing to do after all. We would judge such a person to be confused or misinformed about the nature of promises—what it is to make one and what it requires one to do. Rawls says that we would have grounds for judging the promiser to have misunderstood the practice of promising. Similarly: consider the baseball player who asks the umpire “can I have four outs?”—because it would be better overall for him, or his team, if it was so. Such a player would be confused about what it is to play the game of baseball. Rawls argues that such a question constitutes a confusion between practical questions that are internal to a practice and questions about how to justify the practice itself. Promissory obligation, Rawls thinks, is itself a practice-internal notion. Building on Rawls’ ideas, Searle (1964) uses promissory obligation as his case study in how the conventional constitutive rules of a social practice can ground normative facts in non-normative facts. In later work, Searle builds his own theory of speech acts around this same notion of a constitutive rule, again taking the act of promising as his paradigm case (Searle 1965, 1969). On Searle’s view, to perform an illocutionary act is to speak in a way that conforms to a collection of constitutive rules, all of which ultimately apply in virtue of linguistic convention. To felicitously promise, for example, requires that the speaker express a proposition to the efect that they will perform some act, that the hearer would prefer the speaker to perform this act, that it is not obvious that the speaker would perform it anyway, that the speaker forms an intention to carry out the act and for his utterance to place him under an obligation to do so, that the speaker communicatively intends the hearer to recognize that the speaker has these intentions, and that all of this follows from the linguistic conventions governing the expressions that the speaker uses to perform their speech act. According to Searle, promising is not special in being made possible by conventional constitutive rules of this kind. An analogous suite of rules sets the parameters for the performance of any illocutionary act. Failure to keep a promise, on Searle’s view, is analogous to violating the felicity conditions of other illocutionary acts—for example, by asserting what one knows to be a falsehood.11 There are some reasons to doubt that promising is constituted by the conventional rules of a particular social or linguistic practice, however. One infuential argument, pressed by Scanlon (1998, ch.7), is that it really does seem possible to make promises outside of any shared conventional framework, in the state of nature. Scanlon imagines two hunters from tribes who have had no previous contact and do not share a language, and who meet on opposite sides of a river, having each accidentally thrown their hunting weapons to the other’s side. Scanlon argues that these two 74

Speech-Act Theory

individuals can enter into a promissory arrangement, despite lacking a common conventional framework in which to do so.12 Instead of thinking of promissory obligation as a conventional practice, Scanlon (1990, 1998) instead argues that we should think of it as arising from the expectations that a promiser instills in a promisee. To instill these expectations in someone and then fail to live up to them is to manipulate others in a way that is normally morally unacceptable. Scanlon explains this moral unacceptability by appeal to his own brand of contractualist ethics, while other expectation-based accounts of promissory obligation appeal to other background theories of moral obligation.13 On expectation-based views, promising is, frst and foremost, a communicative act whose point is to produce expectations in an addressee. Although Scanlon does not commit himself to any particular theory of communication, his view fts naturally with intentionalism, which is built to explain the possibility of communication in the absence of convention. And at least some intentionalists have ofered accounts of promising that would ft in nicely with expectation-based accounts of promissory obligation.14 A third school of thought takes promissory obligation to be a normative status that is more fundamental than any social convention but that is not reducible to the promisee’s expectations. Theories of this kind take the act of promising to be fundamentally normative in nature: they take promises to be possible by virtue of our powers to shape the normative statuses that govern us. Within this genre, theories difer with respect to the nature and sources of the normative status involved. Some examples15: Raz (1977) takes promissory obligation to be a kind of right enjoyed by the promisee and created by the promiser as a result of communicating their intention to undertake a commitment. Shifrin (2008) argues that promissory obligation and other related forms of commitment are “integral part[s] of the ability to engage in special relationships in a morally good way, under conditions of equal respect” (485). Owens (2006) argues that promising arises from moral agents’ legitimate interest in having the ability to have authority over others (such as when one has accepted their promise to do something). Again, we can fnd norm-theoretic accounts of speech acts that ft nicely with these views by social and political philosophers (see, e.g., Kukla and Lance (2009, 24, 90)). But by the same token, normative theories of promising are subject to some of the same criticisms that face normative theories of speech acts more generally, some of which we discussed in §1.

2.2 Speech Acts and the Law What are laws? An answer inspired by Thomas Hobbes and made infuential by the early legal positivists Jeremy Bentham (1970) and John Austin (1832) was that a law is an “assemblage of signs” whose meaning is a command addressed by a sovereign to their subjects and backed by the threat of force. “Imperativalism”, as this theory is sometimes called, is no longer a popular view.16 Its most infuential critic was H. L. A. Hart (1994, sec. 3), much of whose criticism draws on ideas about speech acts that resemble those of J. L. Austin.17 Hart argues that laws cannot be thought of as pieces of language, although they are in many cases put into efect by linguistic means. Ultimately, though, the fact that a given utterance—for example, a statute—constitutes law must be due to extra-linguistic social conventions that govern a society’s practices, institutions, and norms. Hart also argues that laws needn’t have the force of commands or even permissions, but may instead confer a range of other kinds of normative status, such as the power to enter into contracts or the right to marry. The speech acts by which laws are created thus form a sub-category of what Austin (1962, 152) called exercitives—acts that bring new and often normatively loaded facts into being. Moreover, in modern bureaucratic democracies, there is no one entity that plays the role of the sovereign—the “speaker” of these exercitives. Rather, city clerks, law enforcement, and regulators act as agents in relation to a principal—representing the interests of the people within a 75

Daniel W. Harris and Rachel McKinney

domain of state authority sharply indexed to an organizational role. In this respect laws resemble the conventional procedures on which Austin focuses: passing legislation, like a felicitous marriage ceremony, requires a structure that indexes rights and responsibilities to institutional roles occupied by agents empowered to act on those powers. Hart also points out that laws needn’t express any person’s will, may be misunderstood by the very legislators who vote on them, and may “bind the legislators themselves”; in these respects they resemble Austin’s conventional acts more than they do ordinary commands. Austin himself emphasized the roles of exercitives in lawmaking, but he also distinguishes them from a second category of speech acts, verdictives, that play a related legal role (Austin 1962, 150). Verdictives, he says, “consist in the delivering of a fnding, ofcial or unofcial, upon evidence or reasons as to value or fact” (152). A verdictive “is done in virtue of an ofcial position: but it still purports to be correct or incorrect, right or wrong, justifable or unjustifable on the evidence. It is not made as a decision in favour or against” (153). Paradigm cases include the decisions of juries, arbitrators, umpires, and judges. The question of how judicial decisions are justifed is one of the central issues in the philosophy of law. Consider the task faced by a judge when interpreting a statute or constitution in order to decide a case. As in ordinary communication, the precise content and force of a piece of law often isn’t made satisfactorily determinate by the words in which it is framed.18 Another way to put this is that legal texts typically display a kind of semantic underdetermination—a phenomenon that they presumably inherit from everyday language use, where semantic underdetermination is endemic and perhaps inevitable.19 Consider an example discussed by Scalia (1997, 23–4) and Neale (2007, 251–2). Smith v. United States20 turned on the interpretation of Title 18, Section 924(c)(1) of the US Code, which mandated a fve-year prison sentence for anyone who “uses…a frearm during and in relation to… [a] drug-trafcking ofense”. John Angus Smith had traded a bag containing an unloaded gun for two ounces of cocaine, but had not brandished or threatened to use the gun at any time. The Supreme Court was faced with the question of whether this counted as “using” the gun “during and in relation to” the trade. The answer apparently turns on the question of whether “using a gun” should be understood, for the purpose of this statute, as implicitly equivalent to “using a gun as a weapon”. How should judges overcome semantic underdetermination in cases like this one? What further information should they be seeking? Or, to approach the issue from a diferent direction: in virtue of what underlying facts does the law have the content and force that it does, such that judges should be seeking out those facts? This central question about jurisprudence turns out to be a question about the nature of speech acts. The debate over how to answer these questions has spawned a range of answers that resemble diferent positions on the nature of illocutionary acts. For example, intentionalists about the law (also known as “purposivists”) argue that the content of the law is fxed by the intentions of those who create it, and that judges should interpret a legal text by attempting to infer the legislative intent behind it. This sort of reasoning dominated the American courts until a few decades ago (Manning 2006), and is still common. There is some evidence that the Supreme Court reasoned this way in deciding Smith v. United States. Here is an excerpt from Justice O’Connor’s majority opinion, which upheld the lower courts’ opinion that trading a gun does constitute using it. Had Congress intended §924(c)(I) to require proof that the defendant not only used his frearm but used it in a specifc manner—as a weapon—it could have so indicated in the statute. However, Congress did not. One straightforward argument for intentionalism about the law takes intentionalism about ordinary communicative acts as a key premise. In ordinary communication, the argument goes, the 76

Speech-Act Theory

content of what is communicated is always determined by the speaker’s intentions; the language used (and the conventions governing the language, as well as information about the context in which it was uttered, and so on) only ever serves as partial and defeasible evidence of the speaker’s intentions. But legal texts are formulated in the same language as ordinary communicative acts, and so it would be bizarre if language is (or even could be) serving an entirely diferent, non-evidential role in that context. By analogy, then, we should take the content of the law to be what was asserted or stipulated by the legislators, and this is of necessity a matter of what they intended—something of which the texts they create can only serve as partial evidence.21 However, there are some important disanalogies between the law and most ordinary communicative acts with which any intentionalist about the law will have to contend. One problem is that the legislators who pass a statute can’t consider every possible eventuality, and so their intentions may not be determinate enough to settle some cases. Smith v. United States provides a plausible example: legislators may not have considered whether trading a gun for drugs should count as using it to commit a crime, and so knowing their intentions might not help to settle the case. This sort of thing happens in ordinary communication, which is not usually a big problem: we can ask for clarifcation, or just let some indeterminacy slide. But a judge who has to make a decision doesn’t have these options.22 A second problem is that constitutions, statutes, and contracts are generally created not by individual speakers but by groups of agents. This raises the question of what it takes for a group of agents to collectively perform a speech act, and for the intentionalist it raises the further question of what it takes for a group of agents to intend something.23 Moreover, the members of a legislative body often have diferent and even conficting goals when they write and pass a piece of legislation. For example, suppose we are able to determine that the creators of the above-discussed statute had the following intentions: one-third explicitly intended trading a gun to count as using it, one-third explicitly intended trading a gun not to count as using it, and the remaining third were aware of this confict, had no opinion on the matter, and intended to deliberately leave the law underdetermined on this point so that the other two factions would vote for the statute. What is the legislative intent in a case like this? An intentionalist about the law owes us a principled way of answering questions like this.24 A third problem is that legislators’ intentions aren’t always easy to discern. Legislators might even seek to hide their true intentions for political reasons, and the causal pathway by which the specifc wording of a statute is reached is usually intricate and impossible to accurately reconstruct (Manning 2003, 2005,). This issue poses a special problem in the legal context, since it is important for the law to be publicly accessible to those who are bound or empowered by it. These and other problems have led some legal theorists to recoil from legislative intentions, often under the banner of textualism—a theory of legal interpretation that has become highly infuential in recent decades, both among scholars and in American courts.25 Textualism is often associated with a skepticism about the existence or legal relevance of legislative intentions, and with the idea that the content of a legal text is the “plain meaning” that a reasonable language-user would extract from it, independent of the author’s intentions. This position resembles conventionalism about speech acts, which emphasizes the role of publicly accessible linguistic conventions in determining the content and force of an utterance.26 Scalia (1997, 24) tells us that this was the sort of textualist reasoning behind his dissent in Smith v. United States. One way to see the central appeal of textualism is to understand it as a response to the problem about how to aggregate the intentions or preferences of a group of legislators. Textualists point out that the legislative process itself is our method of accomplishing this aggregation, and that what survives this process is a legal text, not any of the many possibly conficting intentions with which it was assembled, and which are inaccessible to judges. It is the text supplemented with public aspects of its context that should be understood to constitute the law (Manning 2006). 77

Daniel W. Harris and Rachel McKinney

Intentionalists have responded that there is no escaping legislative intentions, because to interpret any utterance, including a legal text is, ipso facto, to infer the intentions with which it was produced (Alexander and Prakash 2004; Ekins 2012; Neale 2008). Nonetheless, contemporary intentionalists have tended to include constraints on legislative intentions to the efect that these intentions must be publicly accessible to a reasonable person (Ekins and Goldsworthy 2014; Neale 2008; Soames 2013). And contemporary textualists no longer deny that a text can be interpreted independently of context. Textualism and intentionalism have thus converged in some ways, and the debate between them has become fner-grained (Manning 2006)—much in the same way as the debate between intentionalists and conventionalists debate about speech acts.

3 Speech Acts and the Creation of Social Norms and Practices In the context of the right social institutions, speech acts can create laws. Other speech acts are similar in that they give rise to new normative statuses, but dissimilar in that they are embedded in informal social practices rather than formalized legal institutions. A parent might grant their children permission to use the family car. A person might consent to a sexual act. Children playing a game of hide and seek might decide to adopt the rule that a player can’t hide in the same spot twice. Informal exercitives of these kinds have been of major interest to socially and politically minded speech-act theorists in recent decades. Consider Catharine MacKinnon’s (1993) claim that the creation and distribution of pornography constitutes an act of subordinating women. MacKinnon begins from the idea, encouraged by the American courts, that pornography is a kind of speech. But, MacKinnon argues, it is a form of speech that must be understood not merely in terms of what it represents, but in terms of what it does. …But the way it works is not as a thought or through its ideas as such, at least not in the way thoughts and ideas are protected as speech. Its place in abuse requires understanding it more in active than in passive terms, as constructing and performative rather than as merely referential or connotative. (MacKinnon 1993, 21) “The message of these materials”, she continues, “is ‘get her,’ pointing at all women…” (21). MacKinnon defnes pornography as graphic sexually explicit materials that subordinate women through pictures or words. …This defnition includes the harm of what pornography says—its function as defamation or hate speech—but defnes it and it alone in terms of what it does—its role as subordination, as sex discrimination, including what it does through what it says. (MacKinnon 1993, 22) Several philosophers have attempted to defend MacKinnon’s claims about pornography by explicitly framing them in the idiom of speech-act theory.27 The central claim of this literature is that we should think of pornography as speech that subordinates women as a matter of its illocutionary force, and not merely as a locutionary act that depicts subordination or a perlocutionary act of causing subordination. Langton draws an analogy to acts of subordination in legislative contexts: Consider this utterance: “Blacks are not permitted to vote”. Imagine that it is uttered by a legislator in Praetoria in the context of enacting legislation that underpins apartheid. It is 78

Speech-Act Theory

a locutionary act: by “Blacks” it refers to blacks. It is a perlocutionary act: it will have the efect, among others, that blacks stay away from polling booths. But it is, frst and foremost, an illocutionary act: it makes it the case that blacks are nor permitted to vote. It—plausibly— subordinates blacks. (Langton 1993, 302) Langton argues that pornography is like the South African legislator’s utterance in that both acts unfairly rank some people lower than others, legitimate harmfully discriminatory behavior toward them, and unjustly deprive them of important powers (Langton 1993, 303–4). Like the act of passing a law, Langton argues that the creation and publication of pornography can itself bring into existence a new kind of normative status. Of course, there are disanalogies between pornography and the law. The legislator’s authority to alter the normative facts derives from their formal position in a legislative institution. Langton argues that pornography enjoys a kind of de facto authority, in virtue of the role it plays in shaping norms about sexual power among its consumers. As she puts it, pornography “shapes desire, eroticizing hierarchy” (Langton 2017). But it is important to Langton’s position that this shaping of desire is a matter of pornography’s illocutionary force, rather than merely an efect of repeated viewing (on this point, see Langton and West (1999)). Langton (2018a, 2018b) and McGowan (2003, 2004) have defended related theories of how pornography exerts its authority that draws on work in the philosophy of language about presupposition accommodation. If a speech act triggers a presupposition, the act’s felicity depends on whether the presupposed proposition is common ground (Stalnaker 1974, 2014). When someone presupposes something that is not common ground, it is possible to object, though this typically has the efect of derailing the conversation, since presuppositions are typically “not at issue”. If no one does object, then the presupposition is normally accommodated—silently added to the common ground, as if through the back door (Lewis 1979). McGowan and Langton both argue that accommodation is a typical process by which de facto authority arises: the act of publishing pornography would be felicitous only if certain social norms or practices are in place—it presupposes these norms and practices—and so pornography winds up being part of what makes these very norms operative.28 Perhaps the weakest link in this project is the idea that pornography is speech in a sense that invites a speech-act-theoretic treatment. Langton mostly does not try to defend this premise of her argument, instead merely repeating MacKinnon’s point that the American courts have deemed pornography to be speech for First-Amendment purposes. But this is not very convincing on its own: “speech” is one technical term in the context of the American legal system and another technical term in the context of a theory of illocutionary acts. It is not obvious that the extensions of these two technical terms are sufciently aligned for Langton’s purposes. Langton has sometimes responded to this worry by pointing out that it suits her broader purposes if pornography does not turn out to be speech. For example: Notice that if [pornography does not work as speech after all], then feminist arguments against pornography ought to have an easier time of it than they do, since it cannot be a right to free speech that protects pornography, if anything does. (Langton and West 1999, 305) But this does not address our point. Some things count as speech by the technical defnition of the American legal system—e.g. monetary contributions to political campaigns—that probably don’t count as speech by either ordinary or speech-act-theoretic standards. The question is whether pornography is one of these things. 79

Daniel W. Harris and Rachel McKinney

There are reasons for doubt. One is the problem of saying who is the “speaker” of pornography. This interacts with the authority question in interesting ways. No individual pornographic actor, director, producer, or publisher has authority over societal norms. It is more plausible that the industry as a whole has this power. But it is doubtful whether an industry is a well-enough organized entity to be capable of performing illocutionary acts.29 Pornography also seems to lack some of the essential characteristics of illocutionary acts by the standards of most theories of speech acts. An intentionalist can object that there is no communicative intention to subordinate behind pornography; most pornographers presumably intend merely to make money and perhaps also gratify themselves. A Searlean conventionalist might additionally object that pornography has no sincerity or preparatory conditions, and does not have straightforward propositional content. A Brandomian could object that someone who creates or publishes pornography needn’t thereby commit themselves to any content, any more than someone who publishes any other work of fction. Even Austin, whose work has had the most direct infuence on Langton, might object that it is difcult to articulate the felicity conditions of an act of subordinating women by creating or publishing pornography, such that failure to meet these conditions would result in a misfre or an abuse (see Austin (1962, 18)). Unless objections like these can be satisfactorily answered, one worries that the idea of pornography as an illocutionary act is crudely metaphorical. One attempt to answer some of these objections is due to McGowan (2004, 2009, 2018), who argues that exercitive speech acts that operate via a mechanism like presupposition accommodation lack many of the features that speech-act theorists have thought essential to speech acts, including communicative intentions, uptake requirements, and propositional content. McGowan goes on to argue that this is the right way to construe pornography as a kind of speech act, and draws on a normative theory of illocutionary acts in order to do so (see also McGowan (2003)). Some of the same theoretical moves that have been applied in these debates over pornography have also been used to understand other kinds of harmful, subordinating speech.30 Interestingly, these theories have often been built on conficting approaches to the nature of speech acts, and so have posited diferent mechanisms by means of which speech acts manipulate social norms. Langton’s infuential early work is framed in largely Austinian terms, though more recent work by Langton, McGowan, Maitra, and others also borrows from work by Stalnaker and Lewis on presupposition accommodation. Stanley (2015) develops a similar theory of how some forms of propaganda do their work by sneaking “not-at-issue content” into the common ground of a whole society (as it were).31 Maitra, McGowan, and Langton have developed analogous treatments of how hate speech inficts harm by covertly manipulating social norms.32 Meanwhile, Tirrell (2012) has developed an account of how hate speech creates the conditions for political oppression and violence that builds on a theory of speech acts inspired by the normative-functionalist views of Sellars and Brandom.

4 Silencing Speech Acts All theories of speech acts agree that successfully performing an illocutionary act requires more than performing a locutionary act—more than the utterance of meaningful expressions. This raises the possibility of manipulating speakers’ ability to meet these further conditions, either by undermining the conditions necessary for performing an illocutionary act or by creating the conditions for the performance of an illocutionary act against the will of the speaker. The former phenomenon has been dubbed “silencing” (Hornsby 1993; Langton 1993) and “illocutionary disablement” (Maitra 2009) and the latter has been called “extracted speech” (McKinney 2016). The notion of illocutionary silencing emerged from the same feminist critique of pornography discussed in §2.3.33 In addition to subordinating women, MacKinnon (1993) argues, pornography also silences them—a claim that forms the basis of an argument for regulating pornography on 80

Speech-Act Theory

free-speech grounds. In building their case for this claim, Langton (1993) and Hornsby (1993) both take inspiration from Austin’s view that performing an illocutionary act normally requires achieving “uptake”: I cannot be said to have warned an audience unless it hears what I say and takes what I say in a certain sense. An efect must be achieved on the audience if the illocutionary act is to be carried out. …Generally the efect amounts to bringing about the understanding of the meaning and of the force of the locution. So the performance of an illocutionary act involves the securing of uptake. (Austin 1962, 115–16) Langton and Hornsby claim that the act of felicitously refusing sex likewise requires securing uptake—that in order for someone to successfully perform the act of refusing sex, their addressee must understand them as doing so. By manipulating men’s expectations about female sexual desire, Langton and Hornsby argue, pornography lowers the chance that its consumers will understand acts of refusing sex, and so lowers the chance that women who attempt to refuse sex will secure uptake. Pornography thus systematically undermines the felicity conditions of refusing sex, thereby making the act impossible for some women to perform.34 As Maitra (2009), Unnsteinsson (2019), and others have pointed out, the phenomenon of silencing is interesting in ways that go far beyond the use of it to understand the efects of pornography. It is plausible that silencing is a mechanism of structural oppression across many domains, as prejudice prevents the members of oppressed groups from making themselves understood in the performance of all manner of illocutionary acts. Since the ability to perform some speech acts— those involved in political deliberation and protest, for example—may be considered constitutive of full citizenship in a democracy, silencing and the mechanisms by which it happens should be of paramount interest to political philosophers. There are interesting questions about how the notion of silencing can be taken up by speechact theorists who do not share the Austinian approach on which Langton and Hornsby’s approach is based. One matter of controversy is the claim that securing uptake is a necessary condition for performing an illocutionary act. Intentionalists typically draw fner-grained distinctions between ways that a communicative act can succeed. 35 Performing an act requires making an utterance with a communicative intention—(i) an intention to produce a response (such as a belief ) in an addressee, (ii) an intention for the addressee to recognize intention (i), and (iii) an intention for the addressee to satisfy (i) as a result of satisfying (ii) (Grice 1957, 1969). To successfully communicate requires that the addressee recognize which response the speaker intends to produce, thereby satisfying clause (ii). Actually producing this response is a further, extra-communicative achievement: it is possible to be understood without being persuasive. Uptake, on this view, is identifed with the fulfllment of clause (ii) of a communicative intention (Strawson 1964, 448). Although securing uptake is a necessary (and sufcient) condition for successful communication, it is not a necessary condition for performing an illocutionary act. On this view, someone who cannot secure uptake is not prevented from performing an illocutionary act, but they are prevented from communicating. Maitra (2009) argues that this sort of inability to communicate still deserves to be thought of as a pernicious form of silencing, since much of the social and political value of in our ability to perform communicative acts lies in our ability to communicate with them. What use is someone’s ability to state political opinions, to engage in protest, or to refuse sex, one might ask, if others are systematically incapable of understanding? Unnsteinsson (2019) points out a further mechanism by which silencing might work, given the intentionalist view. Most theorists of intention, including intentionalists about speech acts, hold that it is either impossible or irrational to intend what one believes can’t be done.36 Someone 81

Daniel W. Harris and Rachel McKinney

who is aware that they are in a communication-undermining predicament of the kind that Maitra describes will therefore be unable to (rationally) attempt to communicate, and so will be unable to (rationally) perform a communicative act at all.37 Unnsteinsson also argues that a speaker in this situation may believe that they are powerless to communicate in a situation like this, and yet be sufciently self-deceived about this belief that they go through the motions of performing a speech act without actually having the intentions necessary to genuinely perform it. These forms of silencing undermine speakers’ ability to perform illocutionary acts, and not merely their ability to communicate with them. Other theories of speech acts give us diferent ways of understanding how silencing works. Stanley (2011) adopts Williamson’s (2000) view that assertion is constitutively governed by the norm that one must assert only what one knows. It follows that in contexts where the knowledge norm is not operative, genuine assertion becomes impossible. Stanley argues that US political discourse is on the cusp of falling into this state, to the detriment of all those involved, because Americans have all but ceased to expect political actors to say only what they know, and have stopped holding them accountable when they fail to meet this standard. If Stanley is right, then genuine assertion is on the verge of becoming impossible in US political contexts. Kukla (2014) builds a theory of silencing using tools from the normative framework of Kukla and Lance (2009). On Kukla’s view, illocutionary acts are ways of altering normative statuses, such as commitments and entitlements. Crucially, though, Kukla and Lance argue that speech acts can be performed only when certain “input conditions” are met, and these input conditions themselves consist of normative statuses possessed by those involved. Facts about social power—for example, the fact that employees generally aren’t entitled to give commands to their boss—make certain illocutionary acts available to some speakers but not others. On Kukla’s view, if social practices are arranged in such a way that women’s social power is systematically undermined, then they too may simply be unable to perform speech acts, such as commands, that presuppose that power, even if they are nominally in positions that grant them the authority to do so. Finally, McKinney (2016) has argued that just as attempted illocutionary acts may be silenced, they may also be extracted from speakers against their interests or will. This can happen either as a result of a situation that incentivizes or tricks speakers into saying things that will come back to haunt them—as in McKinney’s example of unjust police confessions—or it may happen when speakers are manipulated into satisfying the felicity conditions of illocutionary acts with whose rules they are unfamiliar.

5 Freedom of Speech Many democracies accord their citizens a right to free speech, and protect this right by constitutionally limiting the ways in which the state can interfere with citizens’ speech. In this context, “speech” sometimes has an expansive defnition. In the United States, for example, speech may include the publication of pornography and donations to political action committees, and these activities enjoy considerable protection from government regulation. Still, there is much variation in what kinds of speech are protected in diferent jurisdictions and at diferent times. For example, defamation is subject to criminal or civil liability in many countries, though the details vary considerably. The Swedish Criminal Code (Ch. 18, Sec. 2) specifes a special punishment of up to six years for defamation of the King or another member of the Royal Family.38 Article 274 of the Danish Criminal Code specifes a punishment of up to four years for defaming the deceased.39 Hate speech is another variable category of exception to free-speech protections. Germany has strict laws against hate speech, including, for example, a prohibition on “denying or downplaying” the Holocaust—a crime that is punishable by up to fve years in prison (German Criminal Code, §130.3). By contrast, the First Amendment of the 82

Speech-Act Theory

United States Constitution has been interpreted so as to protect nearly all forms of hate speech. In the words of Schauer (2005), this is one thing that makes the United States’ expansive free-speech protection “a recalcitrant outlier to a growing international understanding of what the freedom of expression entails” (30). This variation makes salient the question of which speech protections are just, and why? Most traditional arguments for free-speech protections presuppose that the speech being protected is assertoric, and that its function is to express beliefs or opinions. For example, Mill (1859) defends broad speech protection on the ground that the airing of opinions is ultimately for the public good, whether the opinions are true or false. But the peculiar evil of silencing the expression of an opinion is, that it is robbing the human race; posterity as well as the existing generation; those who dissent from the opinion, still more than those who hold it. If the opinion is right, they are deprived of the opportunity of exchanging error for truth: if wrong, they lose, what is almost as great a beneft, the clearer perception and livelier impression of truth, produced by its collision with error. (Mill 1859, ch. 2) A defense of free speech framed in this way is open to an accusation of ignoring the fact that we do many things other than express beliefs when we speak. A basic awareness of speech-act theory thus raises deep and challenging questions about what forms of speech—which speech acts—should enjoy state protection. This line of thought has been the basis for several critiques of expansive, American-style freespeech protections. One implication of the feminist arguments that pornography silences women (see §2.4) is that pornography is a threat to free speech, since it undermines the possibility of performing illocutionary acts that, quite plausibly, ought to count as protected speech. Hornsby and Langton (1998) draw out this implication, arguing that laws that protect the publication of pornography undermine rather than support free speech. In a similar vein, McGowan (2012) argues that some acts of hate speech should be regulated on the grounds that they constitute illocutionary acts of discriminating against their target groups and don’t merely cause discrimination. Waldron (2012) defends the regulation of hate speech on the ground that its publication constitutes an act of undermining the basic dignity of its targets, and ultimately their status as equal citizens (4–5, 166). Although Tirrell (2012) does not draw out the implications of her analysis of hate speech for free-speech law, a similar upshot to that of Waldron is easy to anticipate. In the right conditions, Tirrell argues, hate speech enacts a normative framework that permits terrible acts of oppression and violence. Since to do this is to undermine basic human rights, Tirrell’s analysis would seem to ofer support for the regulation of hate speech in at least some contexts. Beyond ofering us these avenues to critique broad free-speech protections, speech-act theory forces us to confront fne-grained questions about how to interpret the free-speech law that already exists. To take just one example, the First Amendment to the United States’ constitution tells us that “congress shall make no law…abridging the freedom of speech”. How should we translate this directive into the idiom of speech-act theory, with its multifarious distinctions? Jacobson (1995) argues that the First Amendment protects only the freedom to perform locutionary acts, and not the freedom to perform illocutionary acts. After all: as Austin showed, many illocutionary acts, such as the act of performing a wedding ceremony or placing someone under arrest, are limited to the purview of speakers acting in particular institutional roles that grant them unique powers. Hornsby and Langton (1998) grant Jacobson’s point that we lack a blanket right to perform any illocutionary act whatsoever, but argue that freedom of speech does entail the freedom to perform communicative acts. Similarly, Maitra and McGowan (2007, 2010) argue 83

Daniel W. Harris and Rachel McKinney

that free-speech protections should not be understood to protect any illocutionary acts that enact obligations, including the exercitive acts that, they argue, are constituted by pornography and hate speech. Perhaps the most fully worked-out theory in this vein is due to Solum (1989), who develops a detailed theory on which freedom of speech should be equated with freedom of communicative action, in the sense theorized by Habermas (1981).

6 Loose Ends We have attempted to give an opinionated tour of some of the fruitful intersections of speech-act theory and social and political philosophy. Of course, there are various other overlaps that we would have liked to explore, given more space. One example is the role of speech acts in deliberative democracy—a topic that is at the core of Habermas’ (1981, 1998) theory of communicative action. A second example is the analogy between the roles played by cooperativity and trust in communication, on the one hand, and in human social organization more generally, on the other hand (see, for example, Williams (2002)). A third is the infuence of speech-act theory on ideas about gender performativity (Butler 1990; Salih 2007). Although the foregoing survey has not been exhaustive, however, we hope to have shown of some of the ways in which the theory of speech acts has proven illuminating outside of the narrow confnes of the philosophy of language.

Acknowledgments Thanks to Justin Khoo and Matthew McKeever for feedback on an earlier draft.

Notes 1 There are several other theories of speech acts that we won’t discuss in detail here. One view, originally stemming from some remarks by Wittgenstein (1960, 67–74), is that at least some speech acts are direct expressions of the speaker’s states of mind—see Bar-On (2004, 2013); Davis (2003); Green (2007); Pagin (2011). Another, which takes inspiration from Sellars’ functionalism while eschewing his appeal to normativity, understands illocutionary acts in terms of the efects on addressees that it is their proper function to produce (Harms 2004; Millikan 1998; Skyrms 2010; Zollman 2011). For a more comprehensive survey of recent work on speech acts, see Harris, Fogal, and Moss (2018); and for the history of these views, with emphasis on the infuence of Wittgenstein, see Harris (2020). 2 Alston (2000); Lepore and Stone (2010, 2018, 2015); Searle (1965, 1968, 1975, 1969); Searle and Vanderveken (1985). 3 Bach and Harnish (1979); Harris (2016); Schifer (1972); Strawson (1964). 4 The most notable conventionalist response to this line of argument is to deny that what we do when speaking metaphorically or indirectly is best understood as communicating or performing illocutionary acts at all. See, for example, Lepore and Stone (2010, 2015). For criticisms of this argument, see Camp (2006); Harris (2016). 5 Carston (2002); Condoravdi and Lauer (2012); Davidson (1979); Neale (2007, 2004); Recanati (2004); Searle (1978); Sperber and Wilson (1995); Travis (2008); Wilson and Sperber (1988). The most promising conventionalist strategy in response to this line of argument has been to argue that intentionalists have failed to appreciate the richness of the discourse-level conventions that govern our exchanges. See, for example, Lepore and Stone (2010) and Harris (2016) for a response. 6 On this and related issues, see Ball (2014); Benton (2011); Sosa (2009). 7 Brandom (1983, 1994, 2000); Geurts (2019); Kukla (2014); Kukla and Lance (2009); Lance and Kukla (2013); Peregrin (2014). 8 Another early statement of a similar view is due to Gazdar (1981). 9 Hobbes’ interest in and application of speech-act-theoretic ideas is not limited to his investigations into the social contract. We will discuss his theory of the nature of laws as imperatives in §2.2. Hobbes also takes speech acts to be of enough inherent interest that he ofers an extended attempt at a taxonomy in Chapter 4 of Leviathan (1651).


Speech-Act Theory 10 For a more comprehensive overview of this debate than we can ofer here, see Habib (2018). 11 Aside from Rawls and Searle, other conventionalists about promising include Fried (2015); Gauthier (1986); Kolodny and Wallace (2015). 12 Similar arguments have been made by intentionalists to show that other speech acts, such as asserting, requesting, and asking questions are likewise not conventional acts. See, for example, Harris (2016, 2019). 13 See, for example, Atiyah (1981); Norcross (2011). 14 For example, Schifer (1972, 108–9); Bach and Harnish (1979, 50). 15 For others, see Hart (1955); Rosati (2011); Watson (2004). 16 Though, for a contemporary defense, see Landenson (1980). 17 This collision of jurisprudence and speech-act theory was no coincidence. Hart was part of the same postwar-Oxford scene that Austin and Grice inhabited, regularly attended Austin’s Saturday-morning reading group, co-taught seminars with Austin, and later described Austin as his greatest philosophical infuence (Sugarman and Hart 2005, 273–75). Hart also infuenced Austin, and is one of the three philosophers cited by name in How to do Things with Words, where he is given credit for helping Austin to understand the relationship between performatives and their felicity conditions by analogy with the relationship between the “operative” part of a statute and “the rest of the document [which] merely ‘recites’ the circumstances in which the transaction is to be efected” (Austin 1962, 7). 18 Indeed the problem generalizes beyond judges: parties to a contract cannot specify in advance every possible contingency that may occur. This—the problem of incomplete contracts—creates space for disagreement, arbitrage, and opportunistic renegotiation. See, e.g. Grossman and Hart (1986); Hart (1995). 19 On semantic underdetermination in the law, see Marmor (2008); Neale (2007, 2008); Soames (2008). 20 508 US 223 (1993). 21 For this and related arguments, see Neale (2007, 2008); Soames (2013, 2008). For other defenses of intentionalism about the law, see Ekins (2012); Ekins and Goldsworthy (2014); Fish (2008); Goldsworthy (2010); Solan (2005). Marmor (2014) defends a modifed version of this view on which legal interpreters should rely on the plain meaning of a text up to the point at which that fails to fully determine the content of the law, at which point intentions enter the picture. Matczak (2017) argues that the intentions grounding the illocutionary aspects of laws, and not merely their locutionary aspects, should be considered the proper object of legal interpretation. 22 Soames (2013, 2017) argues that in such cases, the content of the law is itself indeterminate, and it is the role of the judicial authority to make new law that is in keeping with the “original rationale for the law”. If Soames is right, then judicial decisions are sometimes better understood as exercitives rather than as verdictives. 23 There is a considerable philosophical literature on collective intentionality. For major recent works on group agency and intentions in particular, see Bratman (2014); Gilbert (2013); Ludwig (2016, 2017); Tuomela (2014). For work on collective intentionality in general, see the chapters in Jankovic and Ludwig (2017), and in particular Yafe (2017) on applications to the law. Note that the problem about collective speech acts is not specifc to intentionalists. For work on collective speech acts in general, see Hancher (1979); Lackey (2017); Ludwig (2020). 24 For attempts to overcome this problem, see Hurd (1990, 968–75); Ekins (2012, 47–76). 25 Infuential defenses include Easterbrook (1988, 1990); Eskridge, Jr. (1990); Manning (2005, 2006, 2003); Scalia (1997, 1989). 26 Matczak (2016) explicitly argues that Lepore and Stone’s recent push for conventionalism about speech acts should be taken up by legal theorists who wish to do away with legislative intentions. 27 Hornsby (1993); Hornsby and Langton (1998); Langton (1993, 2018); Langton and West (1999); Maitra (2012); McGowan (2003, 2004). 28 For another accounts of how pornography may achieve authority, see McGlynn (2016). 29 For a diferent argument that pornography is not authoritative, see Green (1998). See also Langton’s (2009, ch. 4) response. 30 For more detailed discussions of this work, see Maitra and McGowan’s chapter on “Language and Free Speech” in this volume, as well as the chapters in Maitra and McGowan (2012). 31 See the chapter on propaganda by Stanley and Quaranto in this volume for discussion of this and other work on propaganda. 32 Langton (2018a, 2018b); Maitra (2012); McGowan (2004, 2012, 2018). 33 A precursor is due to Frye (1983, 89), who uses the notion to theorize situations in which women are unable to express anger due to structural oppression.


Daniel W. Harris and Rachel McKinney 34 It is worth pointing out that Austinian speech-act theory gives us several other ways to diagnose what is going wrong in cases of silencing, since Austin takes speech acts to possess a variety of felicity conditions. Hesni (2018) explores some of these other options, arguing that an act of refusing sex may misfre (or may be treated as misfring) because the speaker lacks (or is treated as lacking) standing, which is (roughly) the social position one needs in order to authoritatively perform a given illocutionary act. 35 See, for example, (Strawson 1964) and (Harris 2020). 36 See, for example, Bratman (1987); Broome (2013); Donnellan (1968); Grice (1971); Neale (2016). 37 This is closely related to what Dotson (2011) calls “illocutionary smothering”. 38 Laws against insulting heads of state remain quite common. See Grifen (2017, 15–18) for further examples. 39 For similar examples in other countries, see Grifen (2017, 26–7).

Bibliography Alexander, Larry, and Sikrishna Prakash. 2004. “Is That English You’re Speaking? Some Arguments for the Primacy of Intent in Interpretation.” San Diego Law Review 41: 967–96. Alston, William P. 2000. Illocutionary Acts and Sentence Meaning. Ithica: Cornell University Press. Atiyah, P. S. 1981. Promises, Morals, and Law. Oxford: Oxford University Press. Austin, J. L. 1962. How to Do Things with Words. Oxford: The Clarendon Press. ———. 1963. “Performative-Constative.” In Philosophy and Ordinary Language, edited by Charles E. Caton, 22–54. Urbana: University of Illinois Press. ———. 1970. “Performative Utterances.” In Philosophical Papers, edited by J. O. Urmson and G. J. Warnock, Second, 233–52. Oxford: Oxford University Press. Austin, John. 1832. The Province of Jurisprudence Determined. London: John Murray. Bach, Kent, and Robert M. Harnish. 1979. Linguistic Communication and Speech Acts. Cambridge, MA: MIT Press. Ball, Brian. 2014. “Speech Acts: Natural or Normative Kinds.” Mind & Language 29 (3): 336–50. Bar-On, Dorit. 2004. Speaking My Mind: Expression and Self-Knowledge. Oxford: Oxford University Press. ———. 2013. “Origins of Meaning: Must We ‘Go Gricean’?” Mind and Language 28 (3): 342–75. Bentham, Jeremy. 1970. Of Laws in General, edited by H. L. A Hart. London: Athlone Press. Benton, Matthew. 2011. “Two More for the Knowledge Account of Assertion.” Analysis 72: 682–87. Bloom, Paul. 2000. How Children Learn the Meanings of Words. Cambridge, MA: MIT Press. Brandom, Robert. 1983. “Asserting.” Noûs 17 (4): 637–50. ———. 1994. Making It Explicit: Reasoning, Representing, and Discursive Commitment. Cambridge: Harvard University Press. ———. 2000. Articulating Reasons. Harvard University Press. Bratman, Michael. 1987. Intention, Plans, and Practical Reason. Cambridge, MA: Harvard University Press. ———. 2014. Shared Agency: A Planning Theory of Acting Together. Oxford: Oxford University Press. Broome, John. 2013. Rationality through Reasoning. West Sussex: Wiley-Blackwell. Butler, Judith. 1990. Gender Trouble. London: Routledge. Camp, Elisabeth. 2006. “Metaphor and That Certain ‘Je Ne Sais Quoi’.” Philosophical Studies 129 (1): 1–25. ———. 2018. “Insinuation, Common Ground, and the Conversational Record.” In New Work on Speech Acts, edited by Daniel Fogal, Daniel Harris, and Matt Moss, 40–66. Oxford: Oxford University Press. Carston, Robyn. 2002. Thoughts and Utterances: The Pragmatics of Explicit Communication. Oxford: Blackwell. Condoravdi, Cleo, and Sven Lauer. 2012. “Imperatives: Meaning and Illocutionary Force.” In Empirical Issues in Syntax and Semantics 9: Papers from the Colloque de Syntaxe et Sémantique à Paris 2011, edited by C. Pinõn, 37–58. New York: Springer Verlag. Davidson, Donald. 1979. “Moods and Performances.” In Meaning and Use, edited by A. Margalit, 9–20. Dordrecht: D. Reidel. Davis, Wayne. 2003. Meaning, Expression, and Thought. Cambridge: Cambridge University Press. DeVault, David, and Matthew Stone. 2006. “Scorekeeping in an Uncertain Language Game.” In Brandial 2006: Proceedings of the 10th Workshop on the Semantics and Pragmatics of Dialogue, edited by David Schlangen and Raquel Fernández, 10:139–46. SemDial; University of Potsdam. Donnellan, Keith S. 1968. “Putting Humpty Dumpty Together Again.” Philosophical Review 77 (2): 203–15. Dotson, Kristie. 2011. “Tracking Epistemic Violence, Tracking Practices of Silencing.” Hypatia 26 (2): 236–57. Easterbrook, Frank H. 1988. “The Role of Original Intent in Statutory Construction.” Harvard Journal of Law and Public Policy 11 (1): 59–66.


Speech-Act Theory ———. 1990. “What Does Legislative History Tell Us?” Chicago-Kent Law Review 66: 441–50. Ekins, Richard. 2012. The Nature of Legislative Intent. Oxford: Oxford University Press. Ekins, Richard, and Jefrey Goldsworthy. 2014. “The Reality and Indispensability of Legislative Intentions.” The Sydney Law Review 36 (1): 39–68. Eskridge, Jr., William N. 1990. “The New Textualism.” U.C.L.A. Law Review 37: 621–91. Fish, Stanley. 2008. “Intention Is All There Is: A Critical Analysis of Aharon Barak’s Purposive Interpretation in Law.” Cardozo Law Review 29 (3): 1109–46. Fried, Charles. 2015. Contract as Promise: A Theory of Contractual Obligation. Second. Oxford: Oxford University Press. Frye, Marilyn. 1983. The Politics of Reality: Essays in Feminist Theory. Berkeley, CA: Crossing Press. Gauthier, David. 1986. Morals by Agreement. Oxford: Oxford University Press. Gazdar, Gerald. 1981. “Speech Act Assignment.” In Elements of Discourse Understanding, edited by Aravind K. Joshi, Bonnie L. Weber, and Ivan A. Sag, 64–83. Cambridge: Cambridge University Press. Geurts, Bart. 2019. “Communication as Commitment Sharing.” Theoretical Linguistics 45 (1–2): 1–30. Gilbert, Margaret. 2013. Joint Commitment: How We Make the Social World. Oxford: Oxford University Press. Goldsworthy, Jefrey. 2010. Parliamentary Sovereignty: Contemporary Debates. Cambridge: Cambridge University Press. Green, Leslie. 1998. “Pornographizing, Subordinating, and Silencing.” In Censorship and Silencing: Practices of Cultural Regulation, edited by Robert C. Post, 285–311. Los Angeles, CA: Getty Publications. Green, Mitchell S. 2007. Self-Expression. Oxford: Oxford University Press. Grice, H. P. 1957. “Meaning.” The Philosophical Review 66 (3): 377–88. ———. 1968. “Utterer’s Meaning, Sentence-Meaning, and Word-Meaning.” Foundations of Language 4 (3): 225–42. ———. 1969. “Utterer’s Meaning and Intention.” The Philosophical Review 78 (2): 147–77. ———. 1971. “Intention and Uncertainty.” Proceedings of the British Academy 57: 263–79. Grifen, Scott. 2017. Defamation and Insult Laws in the OSCE Region: A Comparative Study. Organization for Security; Co-operation in Europe. Grossman, Sanford J., and Oliver D. Hart. 1986. “The Costs and Benefts of Ownership: A Theory of Vertical and Lateral Integration.” Journal of Political Economy 94 (4): 691–719. Habermas, Jürgen. 1981. Theorie Des Kommunikativen Handelns, Band I: Handlungsrationalität Und Gesellschaftliche Rationalisierung. Frankfurt am Main: Suhrkamp Verlag. ———. 1998. On the Pragmatics of Communication. Edited by Maeve Cooke. Cambridge, MA: MIT Press. Habib, Allen. 2018. “Promises.” In Stanford Encyclopedia of Philosophy (Spring 2018 Edition), edited by Edward N. Zalta. Hancher, Michael. 1979. “The Classifcation of Cooperative Illocutionary Acts.” Language in Society 8 (1): 1–14. Harms, William F. 2004. Information and Meaning in Evolutionary Processes. Cambridge: Cambridge University Press. Harris, Daniel W. 2016. “Intentionalism versus the New Conventionalism.” Croatian Journal of Philosophy 16 (47): 173–201. ———. 2019. “Intention and Commitment in Speech Acts.” Theoretical Linguistics 45 (1–2): 53–67. ———. 2020. “We Talk to People, Not Contexts.” Philosophical Studies 177 (9): 2713–33. Harris, Daniel W. 2020. Semantics without Semantic Content. Mind & Language (published online): 1–25. Harris, Daniel W., Daniel Fogal, and Matt Moss. 2018. “Speech Acts: The Contemporary Theoretical Landscape.” In New Work on Speech Acts, edited by Daniel Fogal, Daniel W. Harris, and Matt Moss, 1–39. Oxford: Oxford University Press. Hart, H. L. A. 1955. “Are There Any Natural Rights?” The Philosophical Review 64 (2): 175–91. ———. 1994. The Concept of Law. Second. Oxford: Oxford University Press. Hart, Oliver. 1995. Firms, Contracts, and Financial Structure. Clarendon Lectures in Economics. Oxford: Oxford University Press. Hesni, Samia. 2018. “Illocutionary Frustration.” Mind 127 (508): 947–76. Hobbes, Thomas. 1651. Leviathan. ———. 1984. De Cive: The English Version. Edited by Howard Warrender. Critical Edition by Howard Warrender. Oxford: Oxford University Press. Hornsby, Jennifer. 1993. “Speech Acts and Pornography.” Women’s Philosophical Review 10: 38–45. Hornsby, Jennifer, and Rae Langton. 1998. “Free Speech and Illocution.” Legal Theory 4 (1): 21–37. Hume, David. 1738. A Treatise of Human Nature. London. Hurd, Heidi M. 1990. “Sovereignty in Silence.” The Yale Law Journal 99 (5): 945–1028.


Daniel W. Harris and Rachel McKinney Jacobson, Daniel. 1995. “Freedom of Speech Acts? A Response to Langton.” Philosophy and Public Afairs 24 (1): 64–79. Jankovic, Marija, and Kirk Ludwig, eds. 2017. The Routledge Handbook of Collective Intentionality. London: Routledge. Kolodny, Niko, and R. Jay Wallace. 2015. “Promises and Practices Revisited.” Philosophy and Public Afairs 31: 119–54. Kukla, Rebecca. 2014. “Performative Force, Convention, and Discursive Injustice.” Hypatia 29 (2): 440–57. Kukla, Rebecca, and Mark Lance. 2009. “Yo!” And “Lo!”: The Pragmatic Topography of the Space of Reasons. Cambridge: Harvard University Press. Lackey, Jennifer. 2017. “Group Assertion.” Erkenntnis 83 (1): 21–42. Lance, Mark, and Rebecca Kukla. 2013. “‘Leave the Gun; Take the Cannoli’: The Pragmatic Topography of Second-Person Calls.” Ethics 123 (3): 456–78. Landenson, Robert. 1980. “In Defense of a Hobbesian Conception of Law.” Philosophy and Public Afairs 9 (2): 134–59. Langton, Rae. 1993. “Speech Acts and Unspeakable Acts.” Philosophy and Public Afairs 22: 305–30. ———. 2009. Sexual Solipsism. Oxford: Oxford University Press. ———. 2017. “Is Pornography Like the Law?” In Beyond Speech: Pornography and Analytic Feminist Philosophy, edited by Mari Mikkola, 23–38. Oxford: Oxford University Press. ———. 2018a. “Blocking as Counter-Speech.” In New Work on Speech Acts, edited by Daniel Fogal, Daniel W. Harris, and Matt Moss, 144–64. Oxford: Oxford University Press. ———. 2018b. “The Authority of Hate Speech.” In Oxford Studies in Philosophy of Law, edited by John Gardner, Leslie Green, and Brian Leiter, 3:123–52. Oxford: Oxford University Press. Langton, Rae, and Caroline West. 1999. “Scorekeeping in a Pornographic Language Game.” Australasian Journal of Philosophy 77 (3): 303–19. Lepore, Ernie, and Matthew Stone. 2010. “Against Metaphorical Meaning.” Topoi 29 (2): 165–80. ———. 2015. Imagination and Convention. Oxford: Oxford University Press. ———. 2018. “Explicit Indirection.” In New Work on Speech Acts, edited by Daniel Fogal, Daniel Harris, and Matt Moss. Oxford: Oxford University Press. Lewis, David K. 1979. “Scorekeeping in a Language Game.” Journal of Philosophical Logic 8 (3): 339–59. Ludwig, Kirk. 2016. From Individual to Plural Agency: Collective Action I. Oxford: Oxford University Press. ———. 2017. From Plural to Institutional Agency: Collective Action Ii. Oxford: Oxford University Press. ———. 2020. “What Are Group Speech Acts?” Language & Communication 70: 46–58. MacKinnon, Catherine. 1993. Only Words. Cambridge: Harvard University Press. Maitra, Ishani. 2009. “Silencing Speech.” Canadian Journal of Philosophy 39 (2): 309–38. ———. 2012. “Subordinating Speech.” In Speech and Harm, edited by Ishani Maitra and Mary Kate McGowan, 94–120. Oxford: Oxford University Press. Maitra, Ishani, and Mary Kate McGowan. 2007. “The Limits of Free Speech: Pornography and the Question of Coverage.” Legal Theory 13: 41–68. ———. 2010. “On Racist Hate Speech and the Scope of a Free Speech Principle.” Canadian Journal of Law and Jurisprudence 23 (2): 343–72. ———, eds. 2012. Speech and Harm: Controversies over Free Speech. Oxford: Oxford University Press. Manning, John F. 2003. “The Absurdity Doctrine.” Harvard Law Review 116 (8): 2387–486. ———. 2005. “Textualism and Legislative Intent.” Virginia Law Review 91: 419–50. ———. 2006. “What Divides Textualists from Purposivists.” Columbia Law Review 106 (1): 70–111. Marmor, Andrei. 2008. “The Pragmatics of Legal Language.” Ratio Juris 21: 423–52. ———. 2014. The Language of Law. Oxford: Oxford University Press. Matczak, Marcin. 2016. “Does Legal Interpretation Need Paul Grice? Refections on Lepore and Stone’s Imagination and Convention.” Polish Journal of Philosophy X (1): 67–87. ———. 2017. “Three Kinds of Intention in Lawmaking.” Law and Philosophy 36: 651–74. McGlynn, Aidan. 2016. “Propaganda and the Authority of Pornography.” Theoria 31 (3): 329–43. McGowan, Mary Kate. 2003. “Conversational Exercitives and the Force of Pornography.” Philosophy & Public Afairs 31 (2): 155–89. ———. 2004. “Conversational Exercitives: Something Else We Do with Our Words.” Linguistics and Philosophy 27 (1): 93–111. ———. 2009. “Oppressive Speech.” Australasian Journal of Philosophy 87 (3): 389–407. ———. 2012. “On ‘Whites Only’ Signs and Racist Hate Speech: Verbal Acts of Racial Discrimination.” In Speech and Harm, edited by Ishani Maitra and Mary Kate McGowan, 121–47. Oxford: Oxford University Press.


Speech-Act Theory ———. 2018. “On Covert Exercitives: Speech and the Social World.” In New Work on Speech Acts, edited by Daniel Fogal, Daniel Harris, and Matt Moss. Oxford: Oxford University Press. McKinney, Rachel. 2016. “Extracted Speech.” Social Theory and Practice 42 (2): 258–84. Mill, John Stuart. 1859. On Liberty. London: John W. Parker; Son. Millikan, Ruth Garrett. 1998. “Proper Function and Convention in Speech Acts.” In The Philosophy of Peter F. Strawson, edited by Lewis E. Hahn, 25–43. The Library of Living Philosophers. LaSalle, IL: Open Court. Neale, Stephen. 2004. “This, That, and the Other.” In Descriptions and Beyond, edited by Anne Bezuidenhout and Marga Reimer, 68–182. Oxford: Oxford University Press. ———. 2007. “On Location.” In Situating Semantics: Essays on the Philosophy of John Perry, edited by Michael O’Rourke and Corey Washington, 251–393. Cambridge, MA: MIT Press. ———. 2008. “Textualism with Intent.” Unpublished Manuscript. ———. 2016. “Silent Reference.” In Meanings and Other Things: Essays in Honor of Stephen Schiffer, edited by Gary Ostertag, 229–342. Oxford: Oxford University Press. Nickel, Bernhard. 2013. “Dynamics, Brandom-Style.” Philosophical Studies 162 (2): 333–54. Norcross, Alisdair. 2011. “Act-Utilitarianism and Promissory Obligation.” In Promises and Agreements: Philosophical Essays, edited by Hanoch Scheinman, 217–36. Oxford: Oxford University Press. Owens, David. 2006. “A Simple Theory of Promising.” The Philosophical Review 115 (1): 51–77. Pagin, Peter. 2011. “Information and Assertoric Force.” In Assertion, edited by Jessica Brown and Herman Cappelen, 97–136. Oxford: Oxford University Press. Peregrin, Jaroslav. 2014. Inferentialism: Why Rules Matter. Basingstoke; New York: Palgrave Macmillan. Portner, Paul. 2004. “The Semantics of Imperatives within a Theory of Clause-Types.” Edited by K. ­Watanabe and R. Young. CLC Publications. ———. 2007. “Imperatives and Modals.” Natural Language Semantics 15: 351–83. ———. 2012. “Permission and Choice.” In Discourse and Grammar: From Sentence Types to Lexical Categories, edited by G. Grewendorf and T. Zimmermann, 43–68. Berlin: Mouton de Gruyter. Rawls, John. 1955. “Two Concepts of Rules.” The Philosophical Review 64 (1): 3–32. Raz, Joseph. 1977. “Promises and Obligations.” In Law, Morality, and Society, edited by P. M. S. Hacker and Joseph Raz, 210–28. Oxford: Oxford University Press. Recanati, François. 2004. Literal Meaning. Cambridge: Cambridge University Press. Roberts, Craige. 2004. “Context in Dynamic Interpretation.” In The Handbook of Pragmatics, edited by Lawrence Horn and Gregory Ward, 197–220. Oxford: Blackwell. ———. 2012. “Information Structure in Discourse: Toward an Integrated Formal Theory of Pragmatics.” Semantics and Pragmatics 5: 1–69. ———. 2018. “Speech Acts in Discourse Context.” In New Work on Speech Acts, edited by Daniel Fogal, Daniel Harris, and Matt Moss, 317–59. Oxford: Oxford University Press. Rosati, Connie S. 2011. “The Importance of Self-Promises.” In Promises and Agreements: Philosophical Essays, edited by Hanoch Scheinman, 124–55. Oxford: Oxford University Press. Rosen, Gideon. 1997. “Who Makes the Rules Around Here?” Philosophy and Phenomenological Research 57 (1): 163–71. Salih, Sarah. 2007. “On Judith Butler and Performativity.” In Sexualities and Communication in Everyday Life, edited by Karen E. Lovaas and Mercilee M. Jenkins, 55–68. New York: Sage. Scalia, Antonin. 1989. “Judicial Deference to Administrative Interpretations of Law.” Duke Law Review (3): 511–21. ———. 1997. A Matter of Interpretation: Federal Courts and the Law. Princeton, NJ: Princeton University Press. Scanlon, T. M. 1990. “Promises and Practices.” Philosophy and Public Affairs 19 (3): 199–226. ———. 1998. What We Owe Each Other. Cambridge University Press. Schauer, Frederick. 2005. “The Exceptional First Amendment.” In American Exceptionalism and Human Rights, edited by Michael Ignatieff, 29–56. Princeton, NJ: Princeton University Press. Schiffer, Stephen. 1972. Meaning. Oxford: Oxford University Press. Searle, John. 1964. “How to Derive ‘Ought’ from ‘Is’.” The Philosophical Review 73 (1): 43–58. ———. 1965. “What Is a Speech Act?” In Philosophy in America, edited by Maurice Black, 221–39. London: Allen; Unwin. ———. 1968. “Austin on Locutionary and Illocutionary Acts.” The Philosophical Review 77 (4): 405–24. ———. 1969. Speech Acts. London: Cambridge University Press. ———. 1975. “A Taxonomy of Illocutionary Acts.” In Language, Mind, and Knowledge, edited by Keith Gunderson, VII:344–69. Minnesota Studies in the Philosophy of Science. Minneapolis: University of Minnesota Press. ———. 1978. “Literal Meaning.” Erkenntnis 13 (1): 207–24.


Daniel W. Harris and Rachel McKinney Searle, John, and Daniel Vanderveken. 1985. Foundations of Illocutionary Logic. Cambridge: Cambridge University Press. Sellars, Wilfrid. 1954. “Some Refections on Language Games.” Philosophy of Science 21 (3): 204–28. ———. 1969. “Language as Thought and as Communication.” Philosophy and Phenomenological Research 29 (4): 506–27. Shifrin, Seana. 2008. “Promising, Intimate Relationships, and Conventionalism.” The Philosophical Review 117 (4): 481–524. Skyrms, Brian. 2010. Signals. Oxford: Oxford University Press. Soames, Scott. 2008. “Interpreting Legal Texts: What Is, and What Is Not, Special about the Law.” In Philosophical Essays, Volume 1, 403–24. Princeton, NJ: Princeton University Press. ———. 2013. “Deferentialism: A Post–Originalist Theory of Legal Interpretation.” Fordham Law Review 82 (2): 597–617. ———. 2017. “Deferentialism, Living Originalism, and the Constitution.” In The Nature of Legal Interpretation: What Jurists Can Learn about Legal Interpretation from Linguistics and Philosophy, edited by Brian G. Slocum, 218–40. Chicago, IL: University of Chicago Press. Solan, Lawrence M. 2005. “Private Language, Public Laws: The Central Role of Legislative Intent in Statutory Interpretation.” Georgetown Law Journal 93 (2): 427–86. Solum, Lawrence Byard. 1989. “Freedom of Communicative Action: A Theory of the First Amendment Freedom of Speech.” Northwestern University Law Review 83 (1–2): 54–135. Sosa, David. 2009. “Dubious Assertions.” Philosophical Studies 146: 269–72. Sperber, Dan, and Deirdre Wilson. 1995. Relevance: Communication and Cognition. Oxford: Blackwell. Stalnaker, Robert. 1974. “Pragmatic Presuppositions.” In Semantics and Philosophy, edited by Milton Munitz and Peter Unger, 197–214. New York: New York University Press. ———. 1978. “Assertion.” In Syntax and Semantics 9, edited by Peter Cole, 315–32. New York: Academic Press. ———. 2002. “Common Ground.” Linguistics and Philosophy 25: 701–21. ———. 2014. Context. Context and Content. Oxford: Oxford University Press. ———. 2018. “Dynamic Pragmatics.” In New Work on Speech Acts, edited by Daniel Fogal, Daniel W. Harris, and Matt Moss, 384–99. Oxford: Oxford University Press. Stanley, Jason. 2011. “The Ways of Silencing.” The New York Times. ———. 2015. How Propaganda Works. Princeton, NJ: Princeton University Press. Strawson, P. F. 1964. “Intention and Convention in Speech Acts.” The Philosophical Review 73 (4): 439–60. Sugarman, David, and H. L. A. Hart. 2005. “Hart Interviewed: H.L.A. Hart in Conversation with David Sugarman.” Journal of Law and Society 32 (2): 267–93. Thomason, Richmond H. 1990. “Accommodation, Meaning, and Implicature: Interdisciplinary Foundations for Pragmatics.” In Intentions in Communication, edited by Philip. R. Cohen Jerry Morgan and Martha E. Pollack, 325–64. Cambridge, MA: MIT Press. Tirrell, Lynne. 2012. “Genocidal Language Games.” In Speech and Harm, edited by Ishani Maitra and Mary Kate McGowan, 174–221. Oxford: Oxford University Press. Travis, Charles. 2008. Occasion-Sensitivity: Selected Essays. Oxford University Press. Tuomela, Raimo. 2014. Social Ontology: Collective Intentionality and Group Agents. Oxford: Oxford University Press. Unnsteinsson, Elmar. 2019. “Silencing without Convention.” Pacifc Philosophical Quarterly 100 (2): 573–98. Waldron, Jeremy. 2012. The Harm in Hate Speech. Cambridge, MA: Harvard University Press. Watson, Gary. 2004. “Asserting and Promising.” Philosophical Studies 117: 57–77. Williams, Bernard. 2002. Truth and Truthfulness. Princeton: Princeton University Press. Williamson, Timothy. 1996. “Knowing and Asserting.” Philosophical Review 105: 489–523. ———. 2000. Knowledge and Its Limits. Oxford: Oxford University Press. Wilson, Deirdre, and Dan Sperber. 1988. “Mood and the Analysis of Non-Declarative Sentences.” In Human Agency: Language, Duty, and Value, edited by Jonathan Dancy, J. M. E. Moravcsik, and C. C. W. Taylor, 210–29. Stanford, CA: Stanford University Press. Wittgenstein, Ludwig. 1960. The Blue and Brown Books: Preliminary Studies for the “Philosophical Investigations”. Second. New York: Harper Torchbooks. Yafe, Gideon. 2017. “Collective Intentionality in the Law.” In The Routledge Handbook of Collective Intentionality, edited by Marija Jankovic and Kirk Ludwig, 378–88. London: Routledge. Zollman, Kevin J. S. 2011. “Separating Directives and Assertions Using Simple Signaling Games.” The Journal of Philosophy 108 (3): 158–69.


6 ON THE USELESSNESS OF THE DISTINCTION BETWEEN IDEAL AND NON-IDEAL THEORY (AT LEAST IN THE PHILOSOPHY OF LANGUAGE) Herman Cappelen and Josh Dever There’s an interesting debate in moral and political philosophy about the nature of, and relationship between, ideal and non-ideal theory. In this chapter, we discuss whether an analogous distinction can be drawn in philosophy of language. Our conclusion is negative: even if you think that distinction can be put to work within moral and political philosophy, there’s no useful way to extend it to work that has been done in the philosophy of language. Here is the plan: • •

In Section 1 we present a sketch of how theorizing in philosophy of language works, so that we can highlight the ways in which that theorizing can go wrong. In Section 2, we outline four ways in which ideal and non-ideal theory have been understood in moral and political philosophy and ask whether any of them can provide frameworks to help us pick out interesting subsets of work in philosophy of language. We conclude that none of them can. In Section 4 we turn to the question of whether the expression “social and political philosophy of language” (the title of this volume) can be used to pick out an interesting subdiscipline. Again, we conclude that it cannot. In Sections 5 we give a diagnosis of why the eforts in Sections 2 and 4 fail and draw some general lessons from those failures.

Before we get to all that, some context to make clear why this an interesting issue. Some might have in mind the following little caricature of philosophy of language: The philosopher of language sits in his armchair, never venturing into the outside world. He thinks of a few sentences, and then starts throwing mathematical machinery at those sentences. The result is a theory that handles some tiny fragment of the language (probably mostly mathematical language), giving a bloodless picture of how some hypothetical perfectly rational, dispassionate, and cooperative agents might use that language to pass back and forth packets of information. Confronted with this caricature, one might think that philosophy of language has lost track of “the real world”. Where is the language of metaphor, hyperbole, propaganda, and threats? Where 91

Herman Cappelen and Josh Dever

are the liars, the bullshitters, the hucksters, and the rabble-rousers? Someone familiar with the debate in moral and political philosophy might think the right way to frame this concern is by appeal to the ideal/non-ideal distinction: have philosophers of language been too focused on Ideal Philosophy of Language and ignored non-ideal philosophy language? We’re going to answer that question in the negative. In the end, we think that philosophy of language properly develops theoretical tools by starting with the study of simple cases which, in fact, make up the bulk of ordinary language use, and then systematically extending those theoretical tools to cover a wider and wider range of harder and more complicated cases that often involve navigating tricky waters where questions about specifcally linguistic phenomena start to merge into broader questions about overall human behaviour. That’s not to say, of course, that philosophy of language never goes wrong. It does, as all theoretical projects do. But we will argue that attempts to import a dispute from normative theory about the relationship between ideal and non-ideal theory don’t fnd a distinctive failing in philosophy of language to latch onto. There are, for example, many underexplored topics in philosophy of language. Philosophers of language have had relatively little to say about the persuasive functioning of language in advertisement, and they have had relatively little to say about the semantics of prepositions, for example.1 But there is no evidence of any systematic patterns in what topics are underexplored. It’s hard, of course, to get any good count of stuf we haven’t been paying attention to—but to the extent that looking around is a helpful guide, unexplored topics seem to be scattered about at random.

1 Teory Building: A Partial Sketch We begin by observing one thing that goes under the name “idealization” that will not yield a helpful distinction between ideal and non-ideal theory. Consider a toy sketch of how building a theory of language goes. We begin by collecting some data. Data collection is a messy matter. If we just go out and start writing down what’s happening as people use language, we’re going to get an enormous collection of heterogeneous information. We’ll have details about people’s linguistic output; people’s intentions, plans, hopes, fears, desires, and anxieties; people’s professions, social status, clothing, fnancial difculties, and mathematical profciency; the social conventions governing politeness, greetings and departings, meals, and gift-giving; the specifcs of the physical environment including the weather, the local geography, and what objects are positioned where. No theory of language can incorporate all of that data—if it tried, it wouldn’t be a theory, but just a data dump. The theorist must extract from the data what look to be interesting patterns, and then try to explain those patterns. Doing this involves both deciding what data to keep and what data to discard and deciding what categories to use in characterizing the patterns we fnd in the data. All of this can go under the heading of idealizing, although it could just as well be called abstracting or simplifying.2 But idealization in this sense is not going to produce a distinction between ideal and non-ideal theories. That’s because all theory involves idealization in this sense, so there is no possibility of a “non-ideal” theory. To be non-ideal would be not to idealize, and thus not to abstract and extract patterns from the data. That wouldn’t be a theory at all. However, this sense of idealization does have something to do with our opening caricature of a philosophy of language that has lost engagement with the real world. When we abstract patterns from the data, we make choices about what to pay theoretical attention to. For example: •

We have to decide where to get our data (what parts of the total possible data to take into account). We could look at data restricted to our own introspective judgements on sentence types, or we could look at cross-linguistic feld data drawn from language users in a variety of social settings. 92

Ideal and Non-Ideal Theory

We have to decide what aspects of the data to try to explain. We could focus on the parts of the data concerning truth conditions of and inferential relations among sentences, or we could focus on the parts of the data that concern the forms of emotional, social, and fnancial manipulation that people are attempting in their language use. We have to decide what kinds of patterns to extract from the data. We could characterize patterns using categories like rational and cooperative speaker, thus perhaps fnding a simpler pattern that applies in fewer cases, or we could characterize patterns using categories like deceptive and uncooperative speaker, thus perhaps fnding a more complicated pattern that applies in more cases.

These choices about how to idealize/abstract/simplify our data aren’t choices that will make the diference between right and wrong theories. Whichever way we choose to idealize, we will (assuming we then do our theory building well) end up with a correct theory. But the choices will make a diference in what our theory ends up being about. Idealize in one way, and we end up with a theory of cross-sentential anaphora. Idealize in another way, and we end up with a theory of the role of presuppositions in creating plausible linguistic deniability. So while we don’t get, on this picture of idealization, a distinction between ideal and non-ideal theory, we do get a distinction between explored and unexplored topics. How we idealize shapes what topics we end up theorizing about, and the result is that some topics might end up being neglected. Concerns that important topics are being neglected are legitimate concerns. Work in philosophy of language inevitably sometimes doesn’t gather data that it ought to gather (thus the increased interest in experimental methods and cross-linguistic work, for example). It inevitably sometimes neglects interesting patterns in the data that would have led to interesting projects (the curious behaviour of “or” in free choice permission claims was neglected during decades of work on the semantics of logical vocabulary). And it inevitably sometimes provides explanations that restrict the scope of theories unnecessarily (we start by building theories that focus on the semantics of declarative sentences, when it turns out that equally robust theories could have incorporated the semantic features of interrogative and imperative sentences as well). None of this is meant to be startling news. We don’t always get the right data, ask the right questions, or use the right tools in answering the questions we do ask. Hopefully “don’t always” falls reasonably far short of “frequently don’t”, and hopefully there are mechanisms in the feld that help fx the problems when they arise. (Note that the specifc instances cited in the previous paragraph have more-or-less-satisfactorily been addressed.) But we should always have our eye on what topics have been unjustly neglected in the feld.

2 “Ideal vs Non-Ideal Teory” We turn now to the various ways in which the terms “ideal” and “non-ideal” have been used in describing theories in moral and political philosophy. In these felds, there is a lively debate about how to balance ideal vs non-ideal theory, with many theorists arguing that the alleged dominance of ideal theory has had various pernicious consequences. The debates internal to moral and political philosophy will not be our concern here. Instead, we will focus on the question of whether this distinction and the accompanying debate can be applied to philosophy of language and communication. Is there a distinction between ideal and Non-Ideal Philosophy of Language? If yes, has ideal theory dominated? And has that had negative consequences? To answer these questions, we frst need to understand the distinction as it has been drawn in moral and political philosophy. As with most kinds of fashionable jargon, we fnd that the terminology is used in many conficting ways. That in itself is a serious problem (and a strong prima 93

Herman Cappelen and Josh Dever

facie reason for avoiding the spread of the terminology to other felds). In what follows we outline some of the uses of these terms in moral and political philosophy and the reasons why they don’t illuminate anything useful about work done in philosophy of language. The conclusion will be this: no version of the ideal/non-ideal distinction terminology helps delineate anything important about linguistic theorizing. There are no good reasons to let that piece of confused terminology spread to philosophy of language. Before we turn to the use of “[non]ideal theory” in moral and political philosophy, let’s reemphasize one central point made in the frst section and give it a name: Underexplored: In any feld, there will always be topics that are under-explored. However, to say that you don’t need to import terminology from a contentious debate in moral and political philosophy. You just say: this topic is underexplored. Below, we’ll argue that a lot of what people aim to capture with the ideal/ non-ideal distinction is unimportant or incoherent, and we suggest that the remaining sensible parts reduce to various instances of Underexplored.

Use 1: Full-Compliance vs Partial-Compliance (Rawls) Much of the terminological confusions surrounding “ideal/non-ideal theory” can be traced back to the political philosopher John Rawls. In his book A Theory of Justice, Rawls uses these terms to mark two very diferent distinctions. The frst distinction is between theories that assume full compliance with the demands of justice and those that don’t make that assumption.3 Our question is whether this distinction is usefully applied to work in philosophy of language. First note the most obvious discrepancy: semantic, syntactic, and pragmatic theories are not theories of justice and so there’s no distinction there between those semantic, syntactic, or pragmatic theories that assume full compliance with the demands of justice and those that don’t. Whether people are in full or partial compliance with the demands of justice in Rawls’s sense is just irrelevant. To make the distinction between ideal and non-ideal theory useful when thinking about philosophy of language, we would need move away from how Rawls uses those terms. We have to fnd a substitute for “compliance with the demands of justice” that’s relevant to the study of language. It’s entirely non-trivial to come up with that. Here a possibility: a theory of meaning will make assumptions about how speakers behave and what semantic and pragmatic rules they follow. Maybe the analogue of Rawls’s “full compliance” is the assumption that everyone always follows all these rules fully. But the details of this are extremely hard to work out: “salmon” denotes salmons in English. Is “full compliance” then to be understood as the claim that all English speakers always use “salmon” to denote salmon? Are all English speakers in full compliance with the rigidity of proper names? Insofar as these questions make sense at all, the answers are not useful for dividing up work done in philosophy of language. We have debates of about whether there are natural languages or just micro languages, how much of language is context sensitive, how context sensitivity works, etc. Adding the “ideal/ non-ideal” distinction will do nothing to illuminate or move those debates forward.4 Maybe the more plausible illustration would be variations on Grice’s cooperative principles. Grice says: We might then formulate a rough general principle which participants will be expected (ceteris paribus) to observe, namely: Make your conversational contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged. One might label this the Cooperative Principle. (1967: 45) 94

Ideal and Non-Ideal Theory

Now we could create an analogue of Rawls’s principle as follows: Griceans assume full compliance with the demands of the cooperative principle. Hence they’re doing ideal philosophy of language. The problem with this way of extending the terminology is that there’s no one in the Gricean tradition who ever assumed full compliance with this or related principles. One of the core points of Gricean pragmatics is that people both fout and violate these maxims all the time. No one in the Gricean tradition thought that there were no liars or that people don’t use language for deceptive and evil ends. Griceans try to describe what happens when speakers follow the cooperative principle and they think some people sometimes do that. The Gricean also predict that if we’re in a situation in which it’s known that speakers are being uncooperative, a lot of the usual implicatures will go away. And in fact this seems right. (Scalar implicatures, for example, can go away in confrontational courtroom situations.) So the Gricean not only doesn’t assume that the normative principle is always being followed—they actually have a theory that says the right thing about cases in which the normative principle isn’t being followed. More generally, if we take “full compliance” to mean that we’re theorizing only about people with certain (desirable) features—people who are acting well, following the rules, being cooperative, knowing the meanings of words, etc., then (a) it’s not clear what the “desirable” features are in the linguistic case5 and (b) the theories aren’t assuming that everyone has the desirable features. Minimally they’re saying what happens when language is used by people with the desirable features. More expansively, they might predict that when language is used by people without those features, some of those things don’t happen (as with the implicatures). Or they might say that when language is used by people without those features, there are no systematic consequences to characterize. Now you might think that’s all well and good, but also think that not enough efort has been spent exploring what happens when people aren’t in full compliance. While it might be true that Grice has something to say about non-cooperative behaviour, maybe he and the tradition after him didn’t focus enough on, say, lying, misleading, bullshitting, or various other manipulative uses of language. Below we will show that this complaint is unfounded, but let’s grant it for the sake of argument. Note that what we then have is an instance of Underexplored. We have already established that in order to make Underexplored complaints, we don’t need the ideal/ non-ideal terminology.6

Use 2: Ideal Teory Assumes Favourable Circumstances (Rawls) The second way in which Rawls uses the ideal/non-ideal distinction is to contrast theories of justice that assume that the natural and historical conditions are favourable with those that don’t. According to Rawls, the conditions are favourable if the economic and social conditions are suffciently developed to realize justice. Again, the political version of this doesn’t apply in any immediate way to work in philosophy of language. Consider the claim that proper names are rigid designators. That claim makes no assumptions about justice at all. We would need a broader sense of “favourable”. Maybe the relevant distinction would be between theories that assume “favourable linguistic conditions” and those that don’t (where the former are Ideal Philosophy of Language and the latter Non-Ideal). Does Kripke assume that linguistic conditions are favourable for achieving rigidity? For much the same reasons we gave in the previous section, it’s hard to know what that question means. If proper names are rigid, then, yes, in some sense the linguistic conditions are favourable towards 95

Herman Cappelen and Josh Dever

rigidity. Again, that’s just to say that names are rigid and so the world is such that it makes proper names rigid. To contrast theories that assume “favourable” conditions with those that don’t is just to contrast theories that accept rigidity and those that don’t. As in the case of the “full compliance” reading of “ideal”, one might suspect that the Gricean principles of cooperation are better candidates for spelling out some linguistic analogue of “favourable” conditions. Does the Gricean tradition (with its focus on e.g. the Cooperative Principle) assume favourable linguistic conditions? Well, Grice thinks that actual people often cooperate in conversations and he has a theory of what happens when they do and what happens when they don’t. That’s just to say that he thinks he gives a correct description of the activity some speakers engage in some of the time. Of course, for activity A to take place, the world has to be such that A can take place in it. It is hard to see that an appeal to “favourable conditions” will do much to illuminate that work or help to contrast it with an alternative, Non-Ideal Philosophy of Language. The non-favourable conditions would be when people violate or fout the maxims, but that is something Grice talks extensively about. Again, maybe there some kinds of non-cooperation (e.g. Frankfurtian bullshitting (Frankfurt 2005)) that are underexplored, but that’s just an instance of Underexplored—not something we need a new distinction to characterize. Here is another illustration of why assumptions about “favourable conditions” aren’t useful for characterizing work in philosophy of language. Imagine someone noticing that Grice and others who theorize about communication never talk about how much noise there is in the real world. In the examples Griceans discuss, speakers are never next to jackhammers or other things that disrupt communication. Griceans tend to just ignore noise levels. Now ask: is that a sense in which the Gricean tradition in philosophy of language has been assuming “favourable” conditions (without even mentioning it)? Are they involved in theorizing that idealizes away from the real world (flled as it is with noise)? Is that a sense in which we can say that philosophy of language is ideal theory and that we could introduce a non-ideal theory that’s more sensitive to auditory interference (and illegible handwriting etc.)? Surely the answer here is no. That would just be silly. Names are still rigid even when spoken next to jackhammers. Speakers try to generate implicatures in noisy rooms, it’s just that they sometimes fail because they can’t hear each other. Then there is another theory—the theory of auditory perception—that will explain when people can hear each other and when they cannot. That turns out to be a good division of labour because those with expertise in syntax, semantics and pragmatics don’t know very much about ears and jackhammers. In sum: we don’t need the distinction between ideal and non-ideal theory to explain why we abstract from noise levels. This second Rawlsian way of distinguishing between ideal and non-ideal theory doesn’t apply in any interesting way to work in philosophy of language.

Use 3: Utopian vs (More or Less) Realistic Teory (Valentini) Laura Valentini (2012) articulates a third use of “ideal vs non-ideal theory” (albeit it one very close to Use 2). She says that it is sometimes used to distinguish: … the types of feasibility constraints we might take into account when designing normative principles…. First, we need to distinguish between “fully utopian” theories, which altogether reject the need to place feasibility constraints on principles of justice, and “realistic” theories, which accept some such constraints. Second, among the latter group of theories, we need to distinguish between more or less realistic ones, depending on what kinds of real-world constraints they factor into the design of normative principles. (2012: 656–7) 96

Ideal and Non-Ideal Theory

Valentini uses G.A. Cohen to illustrate the Utopian view according to which justice is “timeless (fact-free) value, perhaps akin to a Platonic ideal” (2012: 657). Whether we can achieve it or not is irrelevant and fguring out what justice is, won’t tell us what to do to get there, according to Cohen. What would be the linguistic analogue of this kind of ideal theory? What would be a linguistic utopia? It would have to be a theory that doesn’t place “feasibility constraints” on a theory of meaning and communication. But no such theory has ever been proposed, so it’s a pointless category. No philosopher of language has tried to describe an ideal language unconstrained by any facts about how people can use such a language. The various theories that spring out the kind of process described in Section 1 are all constrained by facts about real languages and their speakers. Even theorists who constructed improved languages (e.g. Frege and Carnap) did it in order to help people—real people—think and reason better. Turning to the distinction between the more or less realistic theories, it is also hard to see how that adds anything to what we already have in Section 1. Theories of the meaning and efects of utterances are constrained by whatever factors the theorist chooses to include. She will always abstract from some features of the real world (that is in the nature of theorizing) and focus on some at the expense of others (again, that’s in the nature of theorizing). As above, we don’t need the “ideal/non-ideal theory” distinction to point that out.

Use 4: Idealized Models: “An Ideal—In the Sense of an Exemplar—Model of How P Should Work” (Mills) The fnal attempt to distinguish ideal from non-ideal theory comes from an infuential paper by Charles Mills (2005). He distinguishes between two ways of constructing models to describe the behaviour of a given phenomenon P: Ideal-descriptive models. Such models purport to be descriptive of P’s crucial aspects (its essential nature) and how it actually works (its basic dynamics). An ideal-descriptive model has to abstract away from certain features of P: one will make simplifying assumptions, based on what one takes the most important features of P to be, and include certain features while omitting others. Ideal-idealized models: such models provide an exemplar of what an ideal P should be like. The “should” here will in general not necessarily be a moral “should,” but may involve norms of a technical functionalist kind (an ideal vacuum cleaner, an ideal factory farm, an ideal digestive system, and so on) or just limiting assumptions convenient for the purposes of mathematization and calculation (an ideal gas, a perfect vacuum, a frictionless plane, a resistance-free conductor). Is this distinction useful for categorizing work in philosophy of language? First, we should give Mills credit for trying to articulate a more general distinction—it’s not meant to be applicable just to work in moral and political philosophy. As a result, this is potentially more directly applicable to, for example, philosophy of language. That said, we also think that the attempted generalization is problematic because the distinction between the two kinds of theorizing is unclear. Here is what we have in mind: ideal-descriptive models abstract away from various features of the world. Which features? In order to not be a subset of ideal-ideal-models, these features must not be delineated in the broad sense of “ideal” or “normative” that is characteristic of the ideal-ideal models. It is hard to see how to avoid that (i.e. it is hard to avoid appealing to something normative when picking out the features that ideal-descriptive theories abstract away from). If we start classifying properties of, say, vacuum cleaners, as important and not so important (as we are doing in an ideal-descriptive model), the natural way to do that is to 97

Herman Cappelen and Josh Dever

think about what’s required for a well-functioning vacuum cleaner. The natural thought is to proceed as follows: We abstract away from lots of the defciencies that plague most vacuum cleaners. The idealdescriptive model of a vacuum cleaner will exclude those that have got gum stuck inside because they are defective. For the same reason, they will exclude those that are cracked or have any number of potential defects that are found in real world vacuum cleaners. Abstraction is typically towards non-defective, non-functioning exemplars. That, however, but that can’t be right because we are then introducing the normative component that’s supposed to be distinctive of the ideal-ideal models (and distinguish those from the ideal-descriptive models). The challenge, therefore, is to make systematic sense of the form of abstraction involved in ideal-descriptive models, without appeal to the broad sense of “should” the ideal-descriptive models involve. We are not claiming this cannot be done, but if it is done, we suspect there are very few real cases left of pure ideal-descriptive models: it’s an abstraction with few real instantiations. That’s our frst concern with Mills picture, but it won’t play a central role in what follows. Mills goes on to argue that in moral and political philosophy, ideal-idealized models are deeply problematic. In what follows, we’ll spend a bit of time outlining the arguments that Mills gives for this. We do that because we suspect some readers are inclined to reason as follows: Mills is right about the bad efects of ideal theory in moral and political philosophy and “idealized philosophy of language” is problematic for some of the same reasons. Our conclusion will be as before: it’s pointless to classify philosophy of language as ideal or non-ideal in this sense, and so there’s nothing problematic here. The problem, according to Mills, with ideal-idealized models is, “…the exploration of the ideal as an end in itself without ever turning to the question of what is morally required in the context of the radically deviant non-ideal actuality” (Pateman and Mills 2007: 118). Mills says: If we start from what is presumably the uncontroversial premise that the ultimate point of ethics is to guide our actions and make ourselves better people and the world a better place, then the framework above will not only be unhelpful, but will in certain respects be deeply antithetical to the proper goal of theoretical ethics as an enterprise. In modeling humans, human capacities, human interaction, human institutions, and human society on ideal-asidealized-models, in never exploring how deeply diferent this is from ideal-as-descriptivemodels, we are abstracting away from realities crucial to our comprehension of the actual workings of injustice in human interactions and social institutions, and thereby guaranteeing that the ideal-as-idealized-model will never be achieved. (2005: 170) Okay, back to philosophy of language. Let’s try to mimic Mills’s rhetorical fourishes applying them to theories of meaning and communication. You might try the following: If we start from what is presumably the uncontroversial premise that the ultimate point of philosophy of language and communication is to guide our speech and make ourselves better speakers, then the ideal-idealized philosophy of language will not only be unhelpful, but will in certain respects be deeply antithetical to the proper goal of philosophy of language as an enterprise. In modeling human speech on ideal-as-idealized-models and in never exploring how deeply diferent this is from ideal-as-descriptive-models, we are abstracting away from realities crucial to our comprehension of the actual workings of speech and communication and thereby guaranteeing that the ideal-as-idealized-model will never be achieved. 98

Ideal and Non-Ideal Theory

This type of argument is deeply fawed. First, the goal of philosophy of language is not to make us better speakers or communicators.7 That much is, we take it, common ground here. Second, as we have seen, philosophers of language have not presented ideal-idealized models of how language ought to be. They’ve tried to describe how languages actually are and how speakers actually behave. In so doing they by necessity abstract from certain feature of real speakers, but that’s essential to all theorizing (as Mills recognizes). Russell doesn’t say that descriptions ought to be quantifers. He says that they are. Kripke doesn’t claim that names ought to be rigid designators. He claims that they are. Grice doesn’t claim that people ought to follow the cooperative principle. He claims that some people sometimes do (and that if you want to exchange information, the maxims of conversation are useful). So (putting aside the problems with understanding Mills’s distinction between ideal-descriptive and ideal-idealized models) the objection to ideal-idealized models doesn’t work when applied to philosophy of language.8 Mills goes further and argues that ideal-idealized models in political philosophy are ideologies that refect the interests of middle-to-upper-class white males (i.e. people such as John Rawls). He says: Ideal theory, I would contend, is really an ideology, a distortional complex of ideas, values, norms, and beliefs that refects the nonrepresentative interests and experiences of a small minority of the national population—middle-to-upper-class white males—who are hugely over-represented in the professional philosophical population. (2005: 172) You could try to mimic this as applied to theories of meaning and communication. Question: Are theories of meaning and communication a distortional complex of ideas, values, norms, and beliefs that refects the nonrepresentative interests and experiences of middle-toupper-class white males who are hugely over-represented among philosophers of language and linguists? Again, we take it the answer here is fairly non-controversial: Answer: No. People who are not middle-to-upper-class white males also use names as rigid designators, defnite descriptions as quantifers, and when they talk to each other they try to be cooperative. Before we leave Mills behind, we’ll pause for a moment and consider a kind of objection that might easily be misunderstood as being aligned with Mills’s. Possible objection: You keep saying that it’s not just middle-to-upper-class white males who use names as rigid designators—all English speakers do. But experimental philosophers have given us powerful evidence that non-western cultures don’t use names as rigid designators (Machery et al. 2004). Isn’t this exactly the kind of thing that Mills had in mind? It shows that the nonrepresentative interests and experiences of western academics has distorted core elements of philosophy of language. Doesn’t this show that mainstream philosophy of language is idealized in exactly Mills’s sense? Our reply will summarize core points from above: the kind of concern outlined in the objection is exactly the kind of data that our central principle Underexplored is perfectly placed to account for. The complaint is that there’s important data that some theorists have overlooked and that the 99

Herman Cappelen and Josh Dever

reason they have overlooked it is because the majority of theorists have certain socio-cultural biases. That happens all the time. Often people will be reluctant to take those kinds of objections seriously because it challenges orthodoxy. That’s just an obvious corollary of the fact that researchers are human beings with all the accompanying defects. We really don’t need to import terminology from moral/political philosophy to describe that.

3 Talking Stock: Philosophy of Language and the Distinction between Ideal vs Non-Ideal Teory We have looked at four ways the ideal/non-ideal theory distinction has been drawn by infuential moral and political philosophers. We have tried to see whether the distinction can be extended to illuminate an interesting divide in philosophy of language (between ideal theories and non-ideal theories). Our conclusion, in each case, has been that there is no useful extension. We suspend judgement about whether it’s a useful distinction in moral and political philosophy (though we have to admit we have our doubts), but if it is, that has do with peculiarities of those disciplines. With this in mind, let’s consider some authors who explicitly invoke the ideal/non-ideal distinction when talking about philosophy of language. The authors of this paper provide a good illustration. In the frst chapter of Bad Language, for example, Cappelen and Dever identify what they call “some idealizations that have guided many of the eforts to theorize about natural language, communication, and speakers” (Cappelen and Dever 2019: 1), and contrast these idealizations with “tools needed to deal with a decidedly non-ideal world” (1). Their list of idealizations includes items such as: • • •

Conversation and communication are fundamentally cooperative exercises. Speakers say only what they know. A common language has words with stable meanings known to all conversational participants.

But “lists of idealizations” of this sort are confused in multiple ways. •

First, as the discussion of the previous section has shown, there is no helpful sense in which these are idealizations. Speakers that cooperate or who say what they know, or languages with stable word meanings, aren’t in any interesting sense better or more properly functioning than speakers or languages that don’t. These are, if anything, simplifying assumptions. But simplifying assumptions are all over the place, and don’t follow any systematic pattern. Sometimes we simplify by assuming that a body of information is common knowledge for the conversational participants—known by all the speakers, and known to be known by all the speakers, and so on. At other times we simplify by assuming that adjectives are all intersective. And at other times we simplify radically by considering a language that contains only a single modal of interest and some uninterpreted sentence letters. Second, it’s not clear that work in philosophy of language really is making those particular simplifying assumptions. It’s certainly true that many formal tools are developed to deal with simplifed cases—an intersective semantics for adjectives really will have trouble dealing with “tall”, and a material conditional semantics for “if ” really will have trouble dealing with almost all natural language conditionals. But it’s much less clear which pieces of theoretical machinery are going to break or misfre when speakers aren’t being cooperative or when contextual information isn’t common knowledge. In fact, as we’ve noted above, some of the standard tools look like they make exactly the right predictions in these cases (no Gricean 100

Ideal and Non-Ideal Theory

conversational implicatures result from violations of the maxim of quantity when speakers aren’t cooperative; the update efects of epistemic modals become non-trivial when contextual information has a non-S5 structure9). And third, even where the simplifying assumptions are being made, they are typically being made as steppingstones towards a more sophisticated theory capable of handling less simplifed data.

4 Is the Category “Social and Political Philosophy of Language” More Helpful? Our job for this volume was to clarify the distinction between Ideal and Non-Ideal Philosophy of Language. We’ve ended up rejecting the distinction. However, maybe there’s another, closely related distinction that is useful. This volume has in its title the phrase “social and political philosophy of language”. What does that category pick out and what is it contrasted with? At the risk of appearing to be incurable curmudgeons, we are going to also reject that category as fairly useless. What could it possibly pick out? Maybe it’s an attempt to pick out speech by people who are politicians or people who talk about political/social topics. If that’s the remit of social-political philosophy of language, then it should be concerned with sentences like: • •

“We should increase the sale tax on cigarettes because that would give the city more money for schools and it could reduce cancer rates”. “The price of parking on city streets should be increased because it is not fair that public property should be rented to car owners for very little money”.

That’s what the vast majority of “political speech” is like. It’s just “ordinary speech”. It’s not special in any way. It doesn’t constitute a distinct subset of speech. The above are just normal sentences uttered by normal speakers who have certain interests or have certain jobs. The same point applies if the claim is that “social and political philosophy of language” picks out words or constructions or speech acts that are politically relevant. The act of saying: “The meeting starts on Thursday at 5pm” is politically relevant, if the meeting is about politics, but there’s nothing semantically distinctive about the sentence or the act.

5 A Final Proposal: Two Lists It turns out to be impossible to fnd defnitions of “ideal/non-ideal” or “social and political” that will do useful work in carving out interesting subsets of work in philosophy of language. One way to bypass that problem is to just have a list of what people think of as falling into these categories— without asking them to justify or defne them. Here’s a conjecture: if you were to ask a group of contemporary philosopher of language to produce two lists, one with topics from “social and political or non-ideal philosophy of language” and one from “ideal or not-social-or-political philosophy of language”, they would probably know what to do and they would end up with fairly similar lists: List 1 • • • • • •

Slurs Bullshitting Lying Insincerity Silencing Manipulation and propaganda 101

Herman Cappelen and Josh Dever

• • • • •

The language of race, gender, and disability Lexical efects Code words Conceptual engineering Generics

List 2 • • • • • • • • • • •

Rigidity of proper names and natural kind terms Gricean maxims and the production of conversational implicatures De re and de dicto readings of attitude reports Externalism and naturalness in metasemantics Generalized quantifers Presupposition projection Update conditions of embedded questions Donkey anaphora Indeterminacy of translation and underdetermination of meaning Semantics of vague expressions Lexical semantics.

Suppose our conjecture is true: there would, for example, be somewhat broad consensus that the topics on List 1 belong in this volume and that the topics on List 2 don’t.10 Suppose also that we are right that List 1 topics have no unity. In particular: they don’t share characteristics that List 2 topics lack. What are we to conclude from this? At least this: List 1 is a more or less random collection of topics with no internal unity that for various hard to understand sociological reasons tend to be lumped together. This might seem surprising (and disappointing for those who have bought this volume expecting the topics covered to have a certain kind of unity). However, it’s not all that unusual. We freely operate with distinctions like “analytic vs continental philosophy” where there’s also no unity, but whereby some random convention philosophers would know how to make lists. Maybe even “philosophy” itself is like that: a list of topics with no thematic or methodological unity. More generally, our classifcations and divisions are often random and lack internal coherence in this way and it’s important to emphasize that this doesn’t make topics on the lists less interesting or less worthy of being explored. However, divisions created in these random ways can sometimes have the efect of blinding us to important interconnections. In the case of List 1 and List 2, that would be very unfortunate because the topics on the two lists are so intimately intertwined. We end with some brief case studies that illustrate this.

Case Study 1: From Kripke to Scorekeeping in a Pornographic Language Game Consider one historical line of development in philosophy of language. In the 1960s, work by Saul Kripke on modal logic and by Arthur Prior on tense logic led to the development of a formal toolkit of methods involving truth at an index of evaluation. In the late 1960s, Hans Kamp, working with that toolkit, uses data about the temporal and modal indexicals “now” and “actually” to motivate extending these methods to a two-dimensional notion of truth relative to a pair of indices. David Kaplan then in the early 1970s provides a philosophical setting for the formal methods by introducing the character/content distinction and thus


Ideal and Non-Ideal Theory

allowing for careful semantic analysis of context-sensitivity and careful representation of contexts as agent-time-position-world tuples. Then in the late 1970s, David Lewis in “Scorekeeping in a Language Game” generalizes the Kaplanian notion of a context to include such features as commonly shared bodies of information, conversationally licenced permissions, and plans. Lewis’s “scorekeeping” picture then gets deployed, in the 1990s and early 2000s, by Langton and West in “Scorekeeping in a Pornographic Language Game” and by Mary Kate McGowan in her work on oppressive speech to give detailed theoretical stories about particular ways in which the structures of language interact with power dynamics in society.

Case Study 2: Slurs and Race/Gender Terms In the 1940s and 1950s, work by W.V.O. Quine and Ruth Barcan Marcus on the interaction between modal operators and quantifers led especially Marcus to an interest in the way that modality shapes the best theory of reference. Further considerations along these lines led Kripke in Naming and Necessity in 1970 to defend a broadly direct reference view of names via a metasemantic mechanism that placed reference fxing in the hands of external causal mechanisms. This initial externalist idea in semantics is further developed by Hilary Putnam in the early 1970s, and is then given a specifcally social development by Burge’s work in the late 1970s and early 1980s emphasizing the way that the semantic burden of meaning determination is distributed across a speech community, rather than being individualized. The resulting externalist picture, with its easy accommodation of a speaker alienated from their own conceptual repertoire, is then used by Sally Haslanger in “What Are We Talking About? The Semantics and Politics of Social Kinds” to give an account of the meaning of gender and race terms that fts with her ameliorative project and by Chris Hom in “The Semantics of Racial Epithets” to produce a theory of the normative content of slurs.

Case Study 3: Te Essentializing Efects of Generics Bertrand Russell’s investigation of type theory as a solution to the semantic paradoxes leads to the development of categorical grammar by Kazimierz Ajdukiewicz and the lambda calculus by Alonzo Church. These formal tools are then taken up by Richard Montague in the late 1960s and very early 1970s. In “The Proper Treatment of Quantifcation in English”, he uses a rich-type hierarchy in giving a semantic theory for a fragment of English. The use of higher types in Montague semantics then leads to Greg Carlson investigating the semantics of reference to kinds in the late 1970s, sparking work on the semantics of generics. Simultaneously, early work on the semantics of “donkey pronouns” by Peter Geach and Gareth Evans leads to David Lewis’s treatment of adverbs of quantifcation, which is then put to work in giving quantifcational analyses of generics. The link between adverbs of quantifcation and modal adverbs then leads to “normal worlds” modal analyses of generics such as Nicholas Asher and Michael Morreau’s work in the 1990s. The combination of the modal element and the reference to kinds in generics then leads to work by Sally Haslanger and Sarah-Jane Leslie exploring the tendency of generic language to give rise to problematic essentializing inferences with bad social consequences.


Herman Cappelen and Josh Dever

We conclude from these case studies that the divide between List 1 and List 2 doesn’t represent a substantial or theoretically interesting divide in the feld. There is of course a general movement from theories built to handle simple cases to theories that extend those earlier theories to handle more complicated cases. The items on List 1 are perhaps in general more complicated than the items on List 2 (although the comparative complexity of, for example, slurs and donkey pronouns doesn’t seem like a matter than can be easily evaluated), so there is perhaps a general tendency for theoretical work to progress from tools crafted in addressing List 2 topics to the use and extension of those tools in addressing List 1 topics. But all of this theoretical work is being done together, with tools, distinctions, and data being freely passed from one topic to another. In philosophy of language, not only does tomorrow not never come—it is already here. Of course, this is not to say that there are not unjustly neglected topics in philosophy of language. Surely there are. Philosophy of language is, it’s worth remembering, a young feld. A bit more than a century old at the most; little more than 50 years old by slightly stricter standards. It’s far too early for us even to have the slightest idea what all the important topics are, much less for us to have adequately addressed all the topics. But “unjustly neglected” doesn’t track anything like a putative Ideal/Non-Ideal Philosophy of Language distinction, and indeed doesn’t look it tracks any interesting distinction (other than perhaps the “hard but maybe tractable/just impossibly hard” distinction). Back to work, then, everyone—there’s much to be done.

Notes 1 Others have had much to say about both of these topics. People have looked, for example, at how syntactic complexity afects viewers’ comprehension and recall of advertisements (Bradley and Meeds (2002)), and there is work about prepositions by semanticists such as Joost Zwarts (e.g. 2017). 2 Anticipating some of our discussion of the next section, this form of idealization maps reasonably well onto what Mills calls building ideal-as-descriptive models, although we will suggest that Mills’s further distinction between ideal-as-descriptive and ideal-as-idealization models is not helpful. 3 Stemplowska and Swift points out that Rawls gives the impression that the distinction between ideal and non-ideal theory simply reduces to whether a theory assumes strict or only partial compliance: “Thus, he speaks of ‘[i]deal, or strict compliance, theory…’ (2001: 13) and, even more strikingly, introduces the concept of ideal theory for the frst time exclusively with reference to the problem of compliance (1999a: 7–8)”. 4 We also have the performance/competence distinction; thus, we have compliance at the level of performance (something we never achieve, even for idiolects) and at the level of competence. 5 Is it “desirable” to apply a word only to things described by its satisfaction conditions, so that metaphorical speakers are acting undesirably? 6 Note that the explored-underexplored distinction doesn’t even map accidentally onto the ideal/nonideal distinction: free choice permission in ability modals is underexplored—that’s got no connection with how “ideal” anyone or anything is. 7 Just as the goal of ethics is not to guide our actions or to make us better people. It’s to understand the nature of morality and justice. 8 We should add that even if there were ideal-idealized models in philosophy of language, they would have no impact on whether we achieve the ideal because theories in philosophy of language and linguistics have absolutely no impact on how speech develops ( just as political philosophers have no impact whatsoever on whether societies become more just, or, more cautiously, no more impact than a butterfy fapping its little wings). 9 “To say that it is unclear what the contextual information is, is to say that diferent epistemically possible worlds, diferent sets of worlds represent the contextual information. Thus not all worlds in the context set agree on what worlds are in the context set, and the accessibility relation is not S5. Lewis’s example ‘You can put the public interest frst for once!’ (Lewis 1979) is the classic example of an existential (here, practical) modal having a non-trivial update efect because of lack of common knowledge of the contextual options. (Willer 2013) gives a general model of update efects of epistemic modals under conditions


Ideal and Non-Ideal Theory of contextual uncertainty and shows how that model captures many standard conversational uses of those modals; the tools of Dynamic Epistemic Logic (van Benthem 2011) allow modeling of epistemic updates in a wide range of non-S5 situations involving epistemic interactions of agents under conditions of limited shared knowledge”. 10 The conjecture is true: the topics on List 1 are in this volume and it has gone through a thorough refereeing process.

Bibliography Bradley, S. D., & Meeds, R. (2002). Surface-structure transformations and advertising slogans: The case for moderate syntactic complexity. Psychology & Marketing 19 (7–8):595–619. Cappelen, Herman & Dever, Josh (2019). Bad Language. OUP. Frankfurt (2005). On Bullshit. Princeton: Princeton University Press. Grice, Herbert Paul (1967). Logic and conversation. In Paul Grice (ed.), Studies in the Way of Words. Harvard University Press. pp. 41–58. Lewis, David (1979). Scorekeeping in a language game. Journal of Philosophical Logic 8 (1):339–59. Machery, Edouard; Mallon, Ron; Nichols, Shaun & Stich, Stephen (2004). Semantics, cross-cultural style. Cognition 92 (3):1–12. Mills, Charles W. (2005). “Ideal theory” as ideology. Hypatia 20 (3):165–84. Pateman, Carole & Mills, Charles (2007). The Contract and Domination. Polity. Rawls, John (1971). A Theory of Justice. Harvard University Press. Stemplowska, Z., & Swift, A. (2012). Ideal and non ideal theory. In David Estlund (ed.), The Oxford Handbook of Political Philosophy. pp. 373–89. Oxford: Oxford University Press. Valentini, Laura (2012). Ideal vs. non-ideal theory: A conceptual map. Philosophy Compass 7 (9):654–64. van Benthem, Johan (2011). Logical Dynamics of Information and Interaction. Cambridge University Press. Willer, Malte (2013). Dynamics of epistemic modality. Philosophical Review 122 (1): 45–92. Zwarts, J. (2017). Spatial semantics: Modeling the meaning of prepositions. Language and Linguistics Compass 11 (5):e12241.



Non-Ideal Semantics and Pragmatics

7 LYING, DECEPTION, AND EPISTEMIC ADVANTAGE Eliot Michaelson and Andreas Stokke1

The Bishop. Can Alexander explain why one lies? Alexander. One lies to win an advantage. -I. Bergman, Fanny and Alexander

1 Introduction Lying is a paradigmatic way of deceiving with words. But there are other ways of using language deceptively that don’t amount to lies. Bernard Williams gives the following example: “Someone has been opening your mail,” she helpfully says, and you, trusting her, take it that it was not the speaker herself. If you discover that it was the speaker, you will have to agree (if through clenched teeth) that what she said was true. So, you must also agree, she did not tell you a lie. (Williams, 2002, 96) With respect to Williams’s example, the contrast between lying and speaking deceptively without lying—often referred to as “merely misleading”—might be summarized as the diference between uttering (1a) or (1b). 1

a I haven’t been opening your mail. b Someone has been opening your mail.

In this situation, uttering (1a) is to lie while uttering (1b) is not. Crucially, (1a) is something the speaker believes (indeed knows) to be false, whereas (1b) is something she believes to be true. Yet both utterances are aimed at the same thing: deceiving the hearer into believing that the speaker has not been opening the mail. Based on this sort of contrast, philosophers throughout history have commonly held that to lie is to say something one believes to be false with the intent to deceive one’s listeners into believing it. You lie only if you say something you believe to be false, as in (1a), and as opposed to (1b). But moreover, according to this tradition, the liar is centrally concerned with instilling a false belief in the hearer. So, on this broad view, lying is a form of deception that is characterized by explicitly saying what one wants to deceive one’s listener about. 109

Eliot Michaelson and Andreas Stokke

This conception of lying can be further justifed via consideration of a wide range of commonplace lies. Most lying fts the bill of saying something one believes to be false with the intention to make one’s addressee believe it. In an attempt to impress your friends, you tell them that you went camel-riding during your summer holidays while knowing full well that you stood by and watched. You tell your housemate that the market was out of butter, even though you simply forgot to buy it. Or perhaps you fnd yourself telling your colleague that you enjoyed reading his or her recent book, even though you did not care for it. Unsurprisingly, given all this, the intention to deceive has long been viewed as a central— perhaps even the central—characteristic of lying. In the 4th century Augustine wrote, In reality, the fault of a person who tells a lie consists in his desire to deceive in expressing his thought. (Augustine, (1952 [395]), 55–6) Many, probably even most, philosophers since have followed Augustine in understanding the intention to deceive as a central—perhaps even the central—moral problem with lying. In the last two decades, however, this philosophical package has been called into question. In light of examples involving what are commonly known as “bald-faced lies,” many have found the traditional view wanting. That has served to put the question of what it is that makes an utterance a lie at the center of recent debates regarding the morality of lying. This chapter frst rehearses this dialectic, including some recent responses on behalf of traditionalists. We argue that none of these responses look very promising. Second, we argue that neo-traditionalist attempts to save the traditionalist account of lying ultimately threaten to hollow out the notion of deception, divorcing it from its most interesting features. By our reckoning, the central feature of an intention to deceive is that it aims to give the deceiver an epistemic advantage over the deceived. There is, however, no way for neo-traditionalists to endorse this claim about deception. That, in turn, threatens to undermine one of the main advantages of the view: its purported ability to explain the wrongness of lying. We close with what we hope will be a note of reconciliation: whether or not one accepts the Augustinian picture, one can accept that one of the central motivations for lying is to gain an advantage over one’s audience. This raises the possibility that what makes lying wrong, when it is wrong, is that this advantage is sought out unjustly.

2 Bald-Faced Lies 2.1 Te Augustinian Defnition Augustine’s work on lying spawned a tradition which defned lying as a species of deception. More precisely this view can be spelled out as follows: The Augustinian Defnition of Lying A lies to B if and only if there is a proposition p such that: (LD1) A states that p to B, and (LD2) A believes that p is false, and (LD3) By saying that p to B, A intends to deceive B into believing that p. Against this traditional view, a number of philosophers have recently argued that lying does not necessarily involve intentions to deceive.2 That is, even if you do not aim at deceiving your 110

Lying, Deception, and Epistemic Advantage

listener, you might still be lying. One central type of counterexample to the traditional view involves what Roy Sorensen (2007) has called “bald-faced lies.” Here is (a version of ) a widely discussed example originally introduced by Thomas Carson (2006, 290)3: The Cheating Student The Dean is terrifed of lawsuits. Accordingly, she will only punish students for cheating if those students admit to having cheated. The Dean’s policy is well-known to all parties. Professor X has impeccable video evidence of a student cheating during the exam. The student, Professor X, and the Dean have all just watched this video, and are all aware that the others watched as well. So it is common knowledge that the student cheated on the exam. The Dean asks the student if she cheated on the exam and the student replies “I did not cheat on the exam.” Carson points out that, in cases like this one, condition (LD3) of the Augustinian Defnition is not satisfed. Since it is common knowledge that the student cheated on the exam, it would seem that the student cannot intend to deceive the Dean into believing that she did not cheat on the exam. Not only does the student know that the Dean knows that she cheated, she also knows that the Dean knows that she knows that the Dean knows that she cheated. So how could she hope to convince the Dean otherwise? If this line of reasoning is right, (LD3) fails and the Augustinian Defnition predicts that the student was not lying. And yet, most take it to be abundantly clear that she was lying. Two main strategies have emerged for defending the Augustinian Defnition of Lying in response to cases like the Cheating Student. Some writers have responded by denying that putative bald-faced lies, like the student’s utterance, really are lies. Others have accepted that bald-faced lies are bona fde lies while arguing that bald-faced lies are deceptive, initial appearances to the contrary. In the rest of this section we consider the frst of these strategies. We devote Section 3 to the second of them.

2.2 Bald-Faced Lies and Assertion There are diferent ways of arguing that bald-faced lies are not genuine lies.4 Here we focus on a particular type of argument according to which utterances like the student’s utterance to the Dean are not assertions. There is broad agreement in the literature that you lie only if you make an assertion—though not everyone uses that particular terminology. According to this line of thought, since bald-faced lies are not assertions, they are not lies. Therefore such examples do not show that you can lie without intending to deceive. One example of this strategy can be found in work by Jörg Meibauer (2014). His main argument for the claim that bald-faced lies are not assertion is that the bald-faced liar does not really present p as true in the context since he lets shine through that p is false. He would not feel committed to the truth of p, and he would not be ready to provide further evidence. (Meibauer, 2014, 140) In other words, Meibauer proposes that (LD1) should be replaced with something stronger, something like: A assertively commits herself to B regarding the truth of p. The cheating student does 111

Eliot Michaelson and Andreas Stokke

not assertively commit herself to the truth of p in this way—so she isn’t lying. Hence, the case fails to constitute a counterexample to the revised Augustinian Defnition.5 The problem with this line of argument is that it subtly shifts the original example. Even though the student knows that the Dean knows that what she is saying is false, the student in no way signals this mutual knowledge in making her utterance. So there is no reason to think that the student “lets shine through” that her statement is false. This stands in contrast to uttering something while winking or speaking in an ironic tone of voice. Moreover, the student here is arguably committed to her statement in the sense that, subsequently, she is in no position to claim that it wasn’t uttered in earnest. And, fnally, while the student may not be in a position to provide further evidence to support her claim, because there is none, she would surely do so if she could. A more promising defense of the claim that utterances like the student’s are not assertions, and hence are not lies, is ofered by Ishani Maitra (2018). Her argument is founded on a view of assertion according to which assertion is a speech act governed by the following constitutive rule: Evidence-Responsiveness Rule: If a speaker S’s utterance of U is not sufciently responsive to her (total) evidence that bears on p, she does not assert p via uttering U. (Maitra, 2018, 72) Most philosophers agree that, as Timothy Williamson puts it, Constitutive rules do not lay down necessary conditions for performing the constituted act. When one breaks a rule of a game, one does not thereby cease to be playing that game. (Williamson, 1996, 491) Similarly, on Maitra’s view, one can fail to conform to the Evidence-Responsiveness Rule (henceforth, ERR) and nevertheless succeed in making an assertion. Yet, according to Maitra, some violations of ERR do thereby constitute failures to make an assertion. In particular, such failures must be fagrant, in a particular sense. Hence, on Maitra’s account, fagrant failures to conform to this rule are incompatible with asserting, while both conformity to the rule, as well as non-fagrant failures to conform, are compatible with asserting. (Maitra, 2018, 72) So, fagrant violations of ERR entail that the speaker is not asserting at all, but mere violations of ERR, in contrast, do not. Having clarifed this, we can now see how Maitra will explain why bald-faced lies like that of Carson’s cheating student constitute merely apparent counterexamples to the Augustinian Defnition. Namely, the cheating student is not making an assertion. Hence, pace our initial reaction to the case, it should not ultimately be viewed as a counterexample; we just had to get clearer on what it takes to assert something in a context. Supposing for the moment that ERR is well-motivated, the persuasiveness of this response clearly depends on how convincingly one can make the case that apparent bald-faced lies involve fagrant violations of ERR. So we should ask: what makes a violation of ERR fagrant? Maitra repeatedly says that the failure must be “intentional and sufciently marked” (Maitra, 2018, 72 et passim). With respect to bald-faced liars in particular, Maitra claims that, rather than make assertions, “they do something much more like what an actor does” (Maitra, 2018, 65). By way of motivating this claim, Maitra mentions four points of comparison between the cheating student and an actor on stage: 112

Lying, Deception, and Epistemic Advantage


The student knows that his utterance (“I didn’t cheat on the exam”) is false. Moreover, he also knows that it is not at all supported by his evidence. After all, he knows that he cheated. The Dean knows these things too. (Maitra, 2018, 66) ii [C]ome what may, he will continue to insist that he didn’t cheat on the exam. (Maitra, 2018, 76) iii [T]he student’s performance is only partly intended for his immediate interlocutor, the Dean. Crucially, it’s also intended for others beyond his immediate audience, such as future readers of any transcript or summary of the proceedings. (Maitra, 2018, 77) iv Fourth, and fnally, each of the student and the Dean is aware that the other is aware of the features of the student’s performance enumerated above. (Maitra, 2018, 77) We note that, of these, only (iv) and the observation that “The Dean knows these things too” in (i) are distinctive of the bald-faced liar as opposed to ordinary, deceptive liars. Uncontroversially, ordinary deceptive liars know (or at least believe) that what they say is false and is not supported by their evidence. Likewise, ordinary liars will typically continue to insist on the truth of what they say. But moreover, there are arguably many instances of ordinary, deceptive lying that are intended for an audience beyond the immediate interlocutor—such as, politicians lying in meetings attended by reporters, and so on. In other words, Maitra’s argument that bald-faced lies are not assertions rests on the claim that the bald-faced liar and her audience are mutually aware that the speaker is not responding to evidence, in Maitra’s sense. This is what makes the violation of ERR fagrant, and hence what makes the utterance fall short of assertion. However, even if one is sympathetic to a view of assertion as governed by one or more constitutive rules, there are reasons to doubt the suggestion that asserting is incompatible with fagrantly violating such a rule, in this sense. That is, there are reasons to reject the idea that if the speaker and hearer are mutually aware that the speaker is violating the relevant constitutive rule, the utterance no longer counts as an assertion. In general, given that constitutive rules ought not be understood as necessary conditions for playing the relevant games, it is unclear why instances of breaking such rules with the added feature of mutual awareness would amount to opting out of the game. It is a constitutive rule of soccer that deliberately handling the ball (by any player who is not goalkeeper) is a foul. Consider the following story6: Flagrant Handling Lisa is deemed to be ofside during an attack. She is upset and angry with the referee who she thinks made a mistake. In her rage she looks the referee straight in the eye and in plain view of everyone on the feld and all the spectators she picks up the ball and throws it away. Lisa is clearly violating the no-handling rule, and she is clearly doing so fagrantly, in Maitra’s sense. But it does not seem to us that she is thereby ceasing to play the game. What will happen as a result of Lisa’s violation of the rule? Most likely, she will get a yellow card and a free kick will be awarded the opposing team. But the game has not stopped. She has not placed herself outside the rules by breaking the no-handling rule fagrantly. This (not unrealistic) situation is signifcantly diferent from situations that would be clear instances of opting out of the game. Suppose, for instance, that instead of picking up the ball, Lisa in her rage just walks of the feld, and out of the stadium, refusing to listen to anyone or take any further part in the playing of the match. Or imagine that she sits down on the feld, refusing to get up or move or do anything. These are clear cases of opting out of the game. Importantly though, neither of them involves violating any constitutive rules. Rather, opting out of a game is most commonly a matter of no longer performing actions that typically constitute playing it. 113

Eliot Michaelson and Andreas Stokke

Similarly, if one thinks that assertion is governed by one or more constitutive rules, and one moreover thinks that liars like the cheating student, along with ordinary liars, are violating such rules, one should not think that doing so fagrantly is tantamount to not making an assertion. There are any number of ways of uttering a declarative without making an assertion—winking, for instance, or impersonating someone else—but, as best we can tell, stating something that is not responsive to one’s evidence is not one of these ways, even when this non-responsiveness is apparent to everyone involved, and known to be so. Put slightly differently: if one isn’t asserting p in saying it, then it won’t be appropriate to hold one responsible for putting forward that content in the ordinary way. But we can see no reason to think that one’s failing to be properly responsive to the evidence—even when that is mutually evident to all involved—should prove exculpatory in this manner. On the contrary, these strike us as some of the most important instances in which to be able to hold each other to account for asserting this or that.

3 Lying and Intentions to Deceive We noted two main strategies for defending the Augustinian Defnition of Lying in the face of bald-faced lies. In the previous section we considered a version of the frst of these, centering on the claim that bald-faced lies are not assertions and hence are not lies. We now turn to the second general strategy, according to which bald-faced lies are indeed genuine lies, but should nevertheless be seen as deceptive in some way.

3.1 Deception and Deceit The main proponent of this line of argument is Jennifer Lackey (2013), who attempts to defend a neo-traditionalist picture of lying by substituting a broader notion of “deception” for the more traditional “deceit.” She ofers the following sufciency claims for each of these notions: Deceit: A deceives B with respect to whether p if and only if A aims to bring about a false belief in B regarding whether p. Deception: A is deceptive to B with respect to whether p if A aims to conceal information from B regarding whether p. (Lackey, 2013, 241) The former notion of deceit is the one involved in the Augustinian Defnition, more precisely in (LD3). So, since Lackey agrees that bald-faced lies are lies, she rejects (LD3) and proposes that lying instead be characterized in terms of her notion of deception. Here is Lackey’s (2013, 237) defnition of lying: Lackey’s Defnition of Lying A lies to B if and only if (LL1) A states that p to B, and (LL2) A believes that p is false, and (LL3) A intends to be deceptive to B in stating that p. Lackey’s condition (LL3) contrasts with the traditional (LD3), according to which lying necessarily involves the intention to induce false beliefs in the audience. Hence, on Lackey’s view, lying requires saying something you believe to be false and thereby intending to be deceptive to your listener. But it does not require intending to deceive her. 114

Lying, Deception, and Epistemic Advantage

Returning to the case of the cheating student, Lackey’s key claim is that, even if the student cannot intend to deceive the Dean in the sense of making the Dean believe that she did not cheat, she can still intend to conceal information from the Dean regarding whether she did. More specifcally, the student can intend to conceal her confession that she cheated on the exam. Assuming that such a confession, were there one, would count as evidence, and assuming further that evidence counts as information, Lackey’s neo-traditionalist defnition of lying will classify this case as a lie. The student intends to conceal her confession, and thus (LL3) is fulflled.

3.2 Concealment Don Fallis (2015) and Andreas Stokke (2018) have each criticized Lackey’s position by arguing that the cheating student cannot be said to aim at concealing information from the Dean, and hence does not satisfy (LL3). Lackey specifes that Concealing information regarding whether p can be understood broadly here, so that it subsumes, among other phenomena, concealing evidence regarding whether p. (Lackey, 2013, 241) Further, Lackey argues that while he does not intend to deceive the Dean into falsely believing that he did not cheat, he does intend to conceal crucial evidence from the Dean that is needed for punishment from the university—namely, an admission of wrongdoing. (Lackey, 2013, 241–2) Yet, as Fallis (2015) argues, this argument relies on the questionable assumption that one can conceal x even if x does not exist: [T]he student does not aim to conceal his confession. There is no confession to be concealed since he has not confessed. It is not like when a criminal conceals or destroys existing evidence of his crime. (Fallis, 2015, 90) Similarly, if Jones does not have, or have access to, the blood-stained dagger, she cannot conceal it from the police. And, even more clearly, if there is no blood-stained dagger at all, no one can conceal it, or even intend to do so, as long as they are aware of its non-existence. Many will conclude that the cheating student cannot be said to intend to conceal her confession from the Dean, since there is no confession to conceal. So to defend the view that the student is deceptive, in Lackey’s sense of (LL3), one might instead try to argue that the student intends to conceal information from the Dean in some other sense. There are two main candidates for which information the student might be said to intend to conceal: 2

a That the student cheated. b That the student believes she cheated.

As suggested earlier, however, there are plausible versions of Carson’s original example on which both (2a) and (2b) are common knowledge. But if both these are common knowledge, then it is unclear how the speaker might intend to conceal either of them. In general, it is plausible to think 115

Eliot Michaelson and Andreas Stokke

that if p is common knowledge between A and B, then, since A knows that B already knows p, A cannot rationally believe that she can conceal p from B. If Jones knows that the police know that she knows that they know (etc.) that she has the dagger, she cannot rationally believe that she can conceal the fact that she has the dagger from the police. From this, Fallis and Stokke both conclude that, assuming that the student is a rational agent, she cannot intend to conceal either (2a) or (2b) from the Dean. This argument relies on the assumption that, roughly, one cannot intend to do what one believes one cannot do. That is, on some version of the following widely endorsed principle7: Belief-Intention Constraint A intends to φ only if A believes that she can φ. At this point, if one wants to defend the kind of neo-traditionalist view embodied by (LL3), there are two main options. First, one could opt to broaden the notion of deception further still, so that the student counts as being deceptive after all. Second, one could opt to broaden the notion of intending beyond the Belief-Intention Constraint, so that the student counts as intending to conceal information in spite of her believing that she will inevitably fail. We will consider each in turn.

3.3 Withholding Information and Deception Secundum Quid One way of modifying the neo-traditionalist notion of deception is to focus on withholding information instead of concealment. This might be spelled out as follows: Deception*: A is deceptive to B with respect to whether p if A aims to withhold information from B regarding whether p. The frst thing to note about this suggestion is that, in order to succeed, it would need to be argued that causing false beliefs is a form of withholding information. If not, the neo-traditionalist view will fail to capture the sort of ordinary deceptive lies that originally motivated the Augustinian view. In other words, if the neo-traditionalist wants to appeal to Deception* in feshing out her claim that lying necessarily involves an intention to be deceptive, she needs to argue that whenever A aims to cause a false belief in B, A necessarily aims to withhold information from B. To make good on this challenge, it might be suggested that when causing someone to falsely believe that p, one is thereby preventing them from acquiring the true belief that not-p. Yet to argue that all cases of causing false beliefs are cases of preventing acquisition of true beliefs is arguably a tall order. Within their category of commissive deception, Chisholm and Feehan (1977) distinguished between what they called positive deception simpliciter and negative deception secundum quid. We spell out this distinction as follows: Positive commissive deception simpliciter: A contributes causally to B’s acquiring the false belief that p. Negative commissive deception secundum quid: A contributes causally to preventing B from acquiring the true belief that not-p. At this point, then, the neo-traditionalist is forced to argue that the former entails the latter. Unsurprisingly, though, there are counterexamples to this proposal. Consider the following elaboration on Williams’s example involving (1a): 116

Lying, Deception, and Epistemic Advantage

Mail Opener Anna has been opening Sheryl’s private mail while she’s been away. Sheryl’s back in the ofce, and she sees that someone has been opening her mail. She asks Anna who it was. As it happens, Anna has just quit her job in the ofce that very morning, and she is on her way to leave for good and will, in all likelihood, never see Sheryl or anyone else from the frm again. She also knows that their colleague, Emily, will appear any second now and reveal the truth to Sheryl. So, Anna says, “I haven’t been opening your mail.” Sheryl thanks her and wishes her well in her future career, and Anna immediately leaves the ofce. A short while later Emily reveals the truth to Sheryl, as Anna predicted. Anna’s utterance is clearly a lie. Moreover, her aim is clearly to make Sheryl believe the falsehood that she has not been opening her mail so that she can leave the ofce quickly. But she is clearly not preventing Sheryl from learning the truth. And she knows full well that she is not doing so. Hence, as per the Belief-Intention Constraint, she cannot be said to intend to contribute casually to preventing Sheryl from acquiring the relevant true belief. There is more room to maneuver here. For instance, the neo-traditionalist might fnd some way of making good on the claim that intending to cause a false belief is invariably a way of intending to withhold information. Yet, as we will see next, there are further problems that arise if one opts for a defnition that runs of of Deception*, either in total or just in part.

3.4 Minimal Deception and Epistemic Transparency Even if a conception of deception along the lines of withholding information can be made to capture the traditional causal category, the question remains whether bald-faced liars can be said to have intentions to withhold information. Lackey explicitly glosses withholding information as distinct from concealment: To withhold information is to fail to provide it, rather than to hide or keep it secret. (Lackey, 2013, 241) Fallis argues that the cheating student does not ft into this category: But the student does not aim to conceal, or withhold, any information that is relevant to his guilt. If an object (or an event) might convey some information about some topic, but you already have that information, I cannot withhold the information from you by withholding the object (or the event). (Fallis, 2015, 90) This is right only to the extent that withholding information necessarily involves preventing someone from acquiring it. By contrast, as quoted above, Lackey suggests that to withhold information is “to fail to provide it.” There is arguably a sense in which the student fails to provide the relevant information. In particular, she fails to provide the information in (2a), that is, that she cheated. To be sure, she is not preventing the Dean from acquiring (2a). The Dean already has that information; indeed, the Dean knows (2a). Rather, by failing to provide it, she is interfering with the Dean’s ability to make use of the information they both already have. Hence, there is arguably a sense in which the cheating student does withhold information from the Dean (and intends to do so). So, if the neo-traditionalist wants to construe the kind of deception involved in lying as Deception*, it is plausible to think that (at least many) bald-faced liars can be characterized as deceptive, after all. 117

Eliot Michaelson and Andreas Stokke

Yet this only vindicates the neo-traditionalist claim to reinstate the traditional view of lying as a species of deception to the extent that Deception* is a plausible notion of deception. As we will now argue, there are reasons to be skeptical about this. If the cheating student can be seen as deceptive even in the case where the relevant information is common knowledge between her and the Dean, this means that you can intend to deceive someone without intending to tamper with their epistemic state regarding the information in question. If this strikes one as implausible, it might be because one fnds the following more general idea attractive: Minimal Deception In φ’ing, A is deceptive to B regarding p only if, in φ’ing, either A contributes causally to B being wrong about p or A allows B to continue being wrong about p. By being wrong about p we mean, roughly, having bad beliefs or credences regarding p. In terms of full belief, this means having false beliefs. In terms of credences, roughly, we have in mind cases in which one’s actual credences do not match probability on one’s evidence, or rational credence. For instance, if p is .6 probable on one’s evidence, having a credence of .4 or .8 are both ways of being wrong about p, on this view. If one does not intend to be minimally deceptive in this sense, then one does not have an intention to deceive.8 This is arguably a way of capturing the general thought behind Chisholm and Feehan’s taxonomy of deception and deceptive intentions. To deny Minimal Deception is to accept that deception can be epistemically transparent. That is, on such a view, A can deceive B simply by telling B something she thinks is false, even though they both know that they both know (etc.) that it is false, and even though A knows full well that telling B the falsehood will have no efect whatsoever on B’s beliefs or credences regarding what she says. This, it seems to us, is to efectively abandon the concept of deception. (We return to some further ramifcations of this thought in Section 4.)

3.5 Trying We noted that there are two main ways of defending a neo-traditional view based on (LL3) in order to argue that bald-faced lies involve intentions to deceive. We have seen that a challenge for the frst of these is to avoid giving up Minimal Deception. We now turn to the second strategy, that of modifying the notion of intending. The arguments considered so far have relied on the Belief-Intention Constraint. The general thought is that, since (2a-b) are common knowledge between the student and the Dean, the student cannot (rationally) believe that she can conceal them from the Dean. Hence, given the Belief-Intention Constraint, she cannot intend to do so, and therefore the student does not satisfy the neo-traditionalist proposal as embodied in Lackey’s notion of deception. One strategy, then, would be for the neo-traditionalist to reject the Belief-Intention Constraint. This, however, comes at a substantial cost for the view. Efectively, one has saved the view that lying necessarily involves intentions to deceive by rejecting a central tenet of most theories of intentional action. A more attractive route would be to look for another conative category that is not subject to a constraint like the Belief-Intention Constraint. Consider, for instance, the following modifed version of (LL3): (LL3*) A tries to be deceptive to B with respect to whether p by stating that p. 118

Lying, Deception, and Epistemic Advantage

For the sake of argument, we assume that trying is not constrained by beliefs in the way that intentions plausibly are. That is, we assume that someone can try to φ even if she does not believe she can φ, and hence does not intend to φ.9 So, the claim would be that, in lying, the cheating student tries to conceal information from the Dean, even if she does not intend to do so. However, as we explain below, there are lies that will incorrectly be ruled out by (LL3*).

3.6 Self-Realizing Lies Consider a new example : 10

Inbox Betty, the department secretary, is a master at navigating the university’s byzantine administration. Graduate students regularly come to her for help with various obscure forms, which she is generally happy to render if asked politely. One day, one of the less competent graduate students, A, stops by with a form that he has been trying to fgure out. Rather than asking politely for help, he thrusts the form at Betty and demands “Betty, you need to fll out this form for me!” Betty calmly takes the form and says to A, “I’ll put this at the top of my pile.” As she says this, she carefully, deliberately, and in full view of A places the form at the bottom of her overfowing inbox.11 Betty’s utterance is a bald-faced lie. Signifcantly, it is made false by the very gesture that Betty makes simultaneous with her utterance. Betty’s utterance therefore an instance of what we might call a “self-realizing” lie. Such lies do not satisfy (LL3*). First, note that Betty’s lie does not satisfy (LL3*) with respect to Deception—that is, the notion according to which deception is a matter of concealing information. With her gesture Betty is actively trying to reveal that she is not going to put the form at the top of her inbox. She is not trying to conceal that information. Hence, she is not being deceptive. Second, note that Betty’s lie does not satisfy (LL3*) with respect to Deception*—that is, the notion according to which deception is a matter of withholding information. To say that it does is to say that Betty is trying to withhold information regarding whether she will put the form at the top. But again, by gesturing the way she does, Betty is actively trying to reveal that she is not going to do so. Hence, to say that Betty satisfes (LL3*) with respect to Deception* is to say that Betty’s mental state (or perhaps her act itself ) is incoherent. However, it is clearly implausible to say that Betty’s mental state (or her act) are incoherent. Both her mental state and the act whereby she expresses that state are perfectly comprehensible. In other words, lies like Betty’s do not satisfy (LL3*). By her utterance, Betty is not even trying to be either Deceptive or Deceptive*. One potential response here would be to argue that Betty’s utterance is an instance of irony. In that case, her words would serve to communicate something other than their literal meaning— presumably something true—and thus she would not be lying. We are skeptical about such a response, however. Whether one is speaking ironically is typically understood as a matter to be decided by appeal to the speaker’s mental state. Speakers speak ironically when they have certain sorts of intentions—typically, intentions for their literal meaning to be disregarded in favor of some other communicated content. We are free to simply stipulate that the version of the case in which we are interested is one where Betty lacks whatever intentions would be required to speak ironically; rather, she intends to speak plainly and for her utterance to be plainly and obviously false; she wants the utterance to be heard unironically, as she aims to be slightly rude to this graduate student. Even stipulating all this, we can see nothing incoherent about either Betty’s mental state or her act. 119

Eliot Michaelson and Andreas Stokke

Cases like Inbox demonstrate that it is possible to lie while simultaneously pointing to or otherwise making evident the very falsehood of what one is saying. Even if the neo-traditionalist could deal with the Cheating Student and other instances of bald-faced lying, it is harder to see a way for her to deal with self-realizing lies like these. The basic problem is that, on any version of the neo-traditionalist theory, self-realizing lies will incorrectly be classifed as incoherent. That, however, hardly seems an accurate description of them.

4 Lying, Deception, and Advantage 4.1 Minimal Deception and Epistemic Advantage Earlier, we suggested that it is a bedrock fact about deception that it cannot be epistemically transparent. One cannot deceive someone without either causing or helping maintain that person’s being wrong about the relevant information. We put this in terms of a condition on deception: Minimal Deception In φ’ing, A is deceptive to B regarding p only if, in φ’ing, either A contributes causally to B being wrong about p or A allows B to continue being wrong about p. We want to highlight a corollary to Minimal Deception: Epistemic Advantage If, in φ’ing, A is deceptive to B regarding p, then A either gains or maintains an epistemic advantage over B with regard to p. A holds an epistemic advantage over B regarding p just in case A’s credence in p matches the probability of p on her evidence better than B’s credence matches the probability of p on B’s evidence, and A’s evidence is at least as good as B’s. Given what we said earlier, this amounts to, roughly, A being “less wrong” about p. Of course, deceptive acts will not necessarily succeed in bringing about an epistemic advantage; the deception might not work, or the agent herself may be signifcantly confused about how the world is. Rather, when one aims to be deceptive, one aims to gain an epistemic advantage over another; one takes oneself to be in a relatively good doxastic state regarding p, and one aims to leverage that to put one’s interlocutor in a worse state regarding the very same proposition. To illustrate, consider the following situation. The probability of p on both A’s and B’s evidence is .5. A has .2 credence in p and B has .1 credence in p. A tells B that p. As a result, B’s credence increases to .2. It might seem that in this case A is deceptive toward B because she tells her something she regards as unlikely (perhaps false) and yet she causes B to be less wrong about p. However, we take this to be a case in which A aimed at being deceptive toward B, and also thereby aimed at gaining an epistemic advantage over B, but failed at both. Had A realized what B’s initial credal situation looked like, we would expect, given her aims, for A not to have said p here. Otherwise, A would correctly be described as aiming to make B epistemically better of, which is arguably not a way of being deceptive. Since we do not think all lies are deceptive, we do not think that all lies aim at epistemic advantage. The lies told by both Betty and the cheating student do not aim at epistemic advantage. Still, we take it that these lies do generate or maintain a sort of positional advantage—just not an epistemic one. What sort of advantages are these, exactly, and what in general can be said about why lies are potentially appealing ways of obtaining those sorts of advantages? 120

Lying, Deception, and Epistemic Advantage

4.2 Positional Advantage We characterized epistemic advantage as having a perspective on the world that is, in a certain sense, better or more accurate than someone else’s. Intentionally bringing it about that one has an epistemic advantage over another, then, should put one in a position to take advantage of that gap. Lying is a potentially efective tool for bringing about such an advantage because it in no way afects one’s own epistemic state, and yet it has the potential to infuence one’s interlocutor’s state for the worse—relative to one’s own take on things, at least. But lies do not just have efects on epistemic states. As speech acts, they can afect all manner of other sorts of mental and social states as well. Even if we were to assume that only assertions of declarative sentences can constitute lies, that form of sentence can be used to do a great deal more than just to trade information back and forth: one can use such sentences to make contracts and commitments (“I’ll see you at the airport”), to name things (“I dub thee the Furzey Gorse”), to make believe (“You enter a dimly lit cavern. In the distance, you hear the mufed voices of orcs.”), etc. And in addition to all this, we can of course use declaratives to prompt people to feel all manner of things, like grief, anger, pride, and so forth. This broadens the range of things about which one might try to gain an advantage by means of a lie. So, for instance, one might try to obtain an emotional advantage over one’s chess opponent by informing her that you slept with her partner a few days prior. Even if you expect your opponent not to believe you, the ruse may well take up some of her mental energy during the match—potentially leaving her play distracted. Epistemic advantage, then, is only one sort of advantage that can be obtained via lying. Our earlier cases provide us with some further examples. In the case of the cheating student, for instance, lying deprives the Dean of certain potential courses of action, given the Dean’s unwillingness to act in the absence of a confession. This constitutes a certain sort of positional advantage within the bureaucratic procedures of the university. Betty, on the other hand, blocks A from asking for further assistance without outright telling A that she is unwilling to help him. If A tries to follow up with someone else, Betty can fnd A’s form in her pile and say something to the efect of “Look, it’s right here. I just haven’t had a chance to get to it yet.”

4.3 Political Lies and Nihilistic Advantage One rather perplexing question in contemporary politics is why politicians lie so much— particularly about things where there is little to no chance of their being believed. One possibility is that they are trying to have other sorts of doxastic efects, like defeating our knowledge on certain topics. Another is that they are trying to have emotional efects on their listeners—one might even go so far as to say “manipulate the emotions” of their listeners. Surely, both of these are part of the story of lying in recent political life. But there is another sort of advantage that we think is likely more specifc to our contemporary political milieu, and one that we take to be worth drawing attention to. Suppose that one belongs to a party which professes to stand for a range of goals: lowering crime, strengthening the middle-class, and ensuring access to high-quality public education and health care. Suppose further that many of these outcomes overlap with the professed goals of one’s political opponents; the diferences between the parties (with respect to this range of goals, at least) amounts largely to diferences in strategies for obtaining these goods rather than diferences over what the relevant goods are. Finally, suppose that there is a large and growing body of research suggesting that one’s policy prescriptions for obtaining these goals are bunk. In contrast, your opponent’s prescriptions are generally well-supported by the evidence. Strategically, in order to take or remain in power, what should one do? One could of course shift one’s policy proposals. But then these would fail to distinguish you from your opponent. 121

Eliot Michaelson and Andreas Stokke

Alternatively, one could shift the goals toward which one is working. But supposing that these are widely accepted in the society in which you live, that might well prove costly. Or one could lie. The more standard calculation is that you might manage to convince some of those listening to you that the available evidence is bogus, or does not actually support your opponent. But there is an even more insidious strategy to be played here. For suppose that it is already prevalent in your society to think that politicians are apt to lie. Then, when you disagree with your opponent about some basic facts, your audience is likely to think that one of you must be lying—regardless of whether they have an opinion on who. That, in turn, should prompt them to raise their credence in the proposition that politicians often, or even always, lie even more. In fact, it doesn’t even matter if you expect that members of the public are going to brand you a liar. So long as that raises their credence in propositions like politicians are liars, that serves to hurt your opponent. For one of your opponent’s main advantages is that she actually has better policy prescriptions than you do, at least as far as one can tell given the evidence. But if informed debate on that is of the table—if both are lying, then what’s the point of trying to fgure out the truth?—then the election will have to hinge on other things. Since one was peddling make-believe from the get-go, this is purely to your advantage. We propose to call this “nihilistic advantage”: a sort of positional advantage that obtains when one’s goals are aided by others coming to believe that everyone is lying to them, or perhaps even that there is no truth at all. Sadly, given a certain set of background credences and social conditions that might very well obtain in our own world, one can easily see how there is going to be a fairly direct way of playing for such advantage by lying, and blatantly lying, to the public. When you have nothing but fantasy to sell, the truth—and, in particular, the thought that the truth might be readily obtained by either politicians themselves or at least the media—becomes the enemy. A widespread distrust that truth can be obtained, in turn, becomes a positional advantage. Sometimes, then, one can play for positional advantage by trying to convince one’s audience that everyone—even, possibly, oneself—is a liar.

5 Conclusion We began by considering the traditional, Augustinian account of lying, along with some recent challenges to that account. Next, we considered two broad strategies for responding to those challenges. Finding both of those strategies wanting, we urged caution with respect to neotraditionalist theories of lying. What most interested us, however, was the worry that, in trying to defend a neo-traditionalist account, one might be forced to give up on the plausible claim that deception aims at generating epistemic advantage. Finally, we suggested that, if we are willing to accept a wider notion of “positional advantage” that subsumes epistemic advantage but is by no means equivalent to it, then we might well derive a fairly unifed account of what makes lies wrong: lies are wrong when one uses them, unjustly, to seek out some positional advantage. So far as we can tell, this is both an attractive and underexplored thesis. This latter fact is somewhat surprising, particularly given how ecumenical this thesis turns out to be; depending on how hard a line we take regarding when one might be justifed in seeking positional advantage, the thesis should be equally amenable to those who think it is never justifed to lie and those who think it is often justifed. What might have seemed like a slugfest of intuitions about lying is revealed to hinge on a deeper disagreement about what might justify seeking epistemic or positional advantage vis-a-vis one’s interlocutor. In fact, even without flling in such further details, we take it that there are signifcant advantages to be had by adopting this sort of Bergmanian thesis: that lying is wrong, when it is wrong, in virtue of the liar unjustly seeking either epistemic or a broader positional advantage. For one thing, it points to a way of explaining why political lies like Trump’s are wrong even when we all 122

Lying, Deception, and Epistemic Advantage

know that they are lies: because someone like Trump gains a positional advantage if he succeeds in increasing our credence that all politicians are liars. And, plausibly, it is unjustifed to seek such an advantage in a modern democracy in general—particularly when your opponent actually is not (or is not much of ) a liar.

Acknowledgments We are grateful to Don Fallis, Ethan Nowak, Quinn White, Torfnn Huvenes, and Vlad Krstić for helpful comments and discussion.

Notes 1 2 3 4 5 6 7 8 9 10


This work is entirely collaborative and names appear in alphabetical order. For example, Carson (2006), (2010) Sorensen (2007), Fallis (2009), Stokke (2013). Carson (2010, 21) ofers a slightly diferent version of the example. See e.g. Mahon (2011), Meibauer (2011), (2014), Faulkner (2013), Keiser (2016). Cf. Stokke (2018, ch. 1). Thanks to Torfnn Huvenes for this example. See e.g. Audi (1973), Grice (1973), (1989, 98), Davidson (1989 [1978]), Velleman (1989), Neale (2005), Mele (2010), Stokke (2010), (2018), Kissine (2013, ch. 2), Fallis (2015). See also Velleman (1989, 113–5) for discussion and more references. Note that one can intend to be minimally deceptive with respect to a content other than the one asserted—for instance, an implicature. Here, we are following a line of thought on trying developed in Ludwig (1992). This example bears some resemblance to the “tell-tale” variety of bald-faced lies that have been discussed by Krstić (2019). The diference is that, in the latter cases, there is a feature (typically a behavioral feature of the liar) which makes it common knowledge that the speaker is lying, whereas in cases like Inbox there is a feature that makes it common knowledge that what the speaker says is false. One of the authors would like to take the opportunity to express his profound thanks to Betty White for all of her generous help navigating UCLA’s generally Kaf kaesque forms and regulations during his time as a PhD student there.

References Audi, R. (1973). Intending. Journal of Philosophy, 70(13), 387–403. Augustine. (1952 [395]). Lying. In R. Deferrari (Ed.), Treaties on various subjects (Vol. 16, pp. 53–120). Washington, DC: Catholic University of America Press. Carson, T. (2006). The defnition of lying. Noûs, 40, 284–306. Carson, T. (2010). Lying and deception: Theory and practice. Oxford and New York: Oxford University Press. Chisholm, R., & Feehan, T. (1977). The intent to deceive. Journal of Philosophy, 74, 143–59. Davidson, D. (1989 [1978]). Intending. In Essays on actions and events (pp. 83–102). Oxford: Clarendon Press. Fallis, D. (2009). What is lying? Journal of Philosophy, 106, 29–56. Fallis, D. (2015). Are bald-faced lies deceptive after all? Ratio, 28, 81–96. Faulkner, P. (2013). Lying and deceit. In H. LaFollette (Ed.), International encyclopedia of ethics. Hoboken, NJ: Wiley. Grice, H. (1973). Intention and uncertainty. Proceedings of the British Academy, 57, 263–79. Grice, H. (1989). Studies in the way of words. Cambridge, MA: Harvard University Press. Keiser, J. (2016). Bald-faced lies: How to make a move in a language game without making a move in a conversation. Philosophical Studies, 173, 461–477. Kissine, M. (2013). From utterances to speech acts. Cambridge and New York: Cambridge University Press. Krstić, V. (2019). Can you lie without intending to deceive? Pacifc Philosophical Quarterly, 100, 642–60. Lackey, J. (2013). Lies and deception: An unhappy divorce. Analysis, 73(2), 236–48. Ludwig, K. (1992). Impossible doings. Philosophical Studies, 65, 257–81. Mahon, J. (2011). Review of Lying and Deception by Thomas L. Carson. Notre Dame Philosophical Reviews. Maitra, I. (2018). Lying, acting, and asserting. In E. Michaelson & A. Stokke (Eds.), Lying: Language, knowledge, ethics, and politics (pp. 65–84). Oxford & New York: Oxford University Press.


Eliot Michaelson and Andreas Stokke Meibauer, J. (2011). On lying: Intentionality, implicature, and imprecision. Intercultural Pragmatics, 8, 277–92. Meibauer, J. (2014). Bald-faced lies as acts of verbal aggression. Journal of Verbal Aggression, 2(1), 127–50. Mele, A. (2010). Intention. In T. O’Connor & C. Sandis (Eds.), A companion to the philosophy of action (pp. 108–113). Oxford: Wiley Blackwell. Neale, S. (2005). Pragmatics and binding. In Z. G. Szabó (Ed.), Semantics versus pragmatics (pp. 165–285). Oxford and New York: Oxford University Press. Sorensen, R. (2007). Bald-faced lies! Lying without the intent to deceive. Pacifc Philosophical Quarterly, 88, 251–64. Stokke, A. (2010). Intention-sensitive semantics. Synthese, 175, 383–404. Stokke, A. (2013). Lying and asserting. Journal of Philosophy, CX(1), 33–60. Stokke, A. (2018). Lying and insincerity. Oxford and New York: Oxford University Press. Velleman, J. (1989). Practical refection. Princeton, NJ: Princeton University Press. Williams, B. (2002). Truth and truthfulness: An essay in genealogy. Princeton, NJ: Princeton University Press. Williamson, T. (1996). Knowing and asserting. The Philosophical Review, 105, 489–523.


8 PROPAGANDA Anne Quaranto and Jason Stanley

1 Introduction Let’s defne propaganda as an argument that bypasses reason in urging its audience towards a conclusion (the conclusion could either be a proposition, such as “Germans are dangerous”, or a command, “join the cause!”) Propaganda bypasses reason, in the sense that it appeals to various mechanisms – stereotypes, afect, and, more generally, fawed ideology, to mask from its audience reasons they ought to consider in an argument for the propagandists’ desired conclusion. Propaganda presents a problem. By assumption, propaganda bypasses reason. But how do propagandistic arguments compel? To make the matter more puzzling, propaganda often compels in the mask of reason. Consider Frederick Hofmann’s 1896 book, Race Traits of the American Negro. In it, Hofmann argues that Black people have less “vital force” than white people. This is a work of scientifc racism, a work of racial propaganda, flled with statistics and massive amounts of evidence. In one chapter, Hofmann argues that Black people have “excessive mortality.” In another, he argues that Black people have vastly greater propensity toward criminality. In each case, he claims that there is no environmental explanation—for example, he argues that “[i]n Washington, the colored race has had exceptional educational, religious, and social opportunities,” and so environment cannot explain racial diferences in arrests. In his discussion of mortality, he argues that relevant white and Black populations in his studies have the same environmental conditions. Hofmann’s book is presented as the epitome of reason. And yet it is racial propaganda. In his discussion of Hofmann’s book, the historian Khalil Muhammad (2010) provides a clue about why Hofmann’s book is propaganda, and how it uses the appearance of reason to be convincing. In his work, Hofmann repeatedly argues that white European immigrant populations in Northern cities face worse social and environmental conditions than Blacks. And Hofman argues that the solution to analogous social problems for these communities is an improved environment. In other words, Hofmann’s work presupposes a stereotype about Black people, one that imprints itself on the very evidence he gathers. This stereotype rules out environmental explanations of high Black mortality or crime rates; it rules such environmental explanations out as implausible from the beginning. But the stereotype is not present for immigrant populations. With immigrant populations, Hofmann is suddenly able to recognize environmental conditions that explain social problems. Such background stereotypes mask the reality that fies in the face of the conclusion of the argument. In The Mismeasure of Man, Stephen Jay Gould provides similar explanations for other examples of scientifc racism—for example, in discussing Samuel George Morton’s craniometry studies, he shows that Morton gathered his evidence on the assumption that Black skulls would be smaller, ruling out larger Black skulls as corruptions of the data pool (Gould, 1981, p. 65). 125

Anne Quaranto and Jason Stanley

Propaganda in fact characteristically appeals to virtuous ideals—an ideal of reason, or humanity, or freedom—in the service of a goal that is inconsistent with those ideals. Nazi political theorist Carl Schmitt writes: Humanity as such cannot wage war because it has no enemy….That wars are waged in the name of humanity is not a contradiction of this simple truth; quite the contrary, it has an especially intensive political meaning. When a state fghts its political enemy in the name of humanity, it is not a war for the sake of humanity, but a war wherein a particular state seeks to usurp a universal concept against its military opponent. At the expense of its opponent, it tries to identify itself with humanity in the same way as one can misuse peace, justice, progress, and civilization in order to claim these as one’s own and to deny the same to the enemy. (Schmitt, 1996, p. 54) Propaganda functions by exploiting stereotypes that mask reality—for example, the humanity of an enemy. In his 1852 speech, “What to the Slave is the Fourth of July?,” Frederick Douglass wonders how the founders could have venerated liberty and yet tolerated slavery. His answer is that they sufered under stereotypes that justifed racial hierarchies.1 Propaganda typically employs a logic of what Stanley (2015) calls “undermining,” where fawed stereotypes, or ideologies, allow a virtuous ideal to be employed in the service of a goal that is inconsistent with it.2 The persuasiveness of propaganda involves stereotypes, and the appeal to ideals that seem inconsistent with those stereotypes. To study propaganda is to study this process of the formation of stereotypes and their employment, often necessarily masked, in propaganda. The theory of meaning enters in at a number of junctures in the study of propaganda. First, there are linguistic techniques to mask problematic stereotypes, allowing them to enter in unchallenged. Thus, theorists of meaning who have studied propaganda have focused on linguistic ways to smuggle in content. In her 1999 paper, “Ideology and the Persuasive Use of Presupposition,” Marina Sbisà argues that presupposition accommodation is central to many cases of political propaganda. On her account, strategic use of presupposition is a characteristic way to smuggle controversial normative presuppositions into political debate. For example, consider being asked, “What is your solution to the Jewish Question?” In their classic study of pornography as propaganda, Rae Langton and Caroline West argue that propaganda is efective as a way to denigrate women by leading its consumers to accommodate sexist presuppositions (Langton & West, 1999). Masking can also occur in other ways, including “dog whistles” that allow speakers to deny that they appealed to such stereotypes, such as “inner city” used to convey a racial message (see Stanley 2015; Khoo 2017; Saul 2018); and other linguistic devices, such as what Saul (2017) calls “racial fgleaves.” The study of the linguistic means to mask stereotypes or render them difcult to challenge is a task for the theory of meaning. There are also linguistic processes at work in the formation of stereotypes, such as the formation of friend–enemy distinctions, or the justifcation of various hierarchies. The function of slurs is important here, and the question of how slurs contribute to this process is also a question for the theory of meaning. Finally, the study of propaganda challenges the theory of meaning (Beaver & Stanley, forthcoming). Standardly, the theory of meaning has focused on cases of cooperative, rational, transparent communication, between a speaker and a hearer. Propaganda, on the other hand, involves a speaker (or perhaps more accurately, an author) that’s rarely a single individual and an audience that’s typically an entire community or at least a group within that community. It’s much less clear whether the communication of propaganda still involves, for example, hearers’ identifying the speaker’s communicative intention, and the speaker’s and hearers’ mutual knowledge of such a communicative intention.3 The study of propaganda involves understanding the communicative processes in persuading people by bypassing rationality. It has led philosophers and linguists to add new elements to the theory of meaning, such as Elizabeth Camp’s notion of a perspective (Camp, 2013), introduced 126


to explain the communicative efect of slurs. And it has led theorists of meaning to alter their understanding of classic notions, such as Sbisà’s argument that propaganda forces us to recognize a normative character to the rule of accommodation (Sbisà, 1999), or the more dramatic reworking of the theory of presupposition and accommodation in Beaver and Stanley (forthcoming). In studying propaganda, theorists of meaning continue a long tradition of inquiry. We may have new terms, such as “fake news,” or study it in piecemeal, as in the case of slurs.4 But in studying these phenomena, theorists of meaning connect to a longer tradition. Our goal in this chapter is to contextualize and systematize this study in the context of its recent history.

2 Propaganda’s Epistemic or Rational Harms There is a long tradition of defning propaganda in terms of its epistemic or rational defects, and taking this characteristic to distinguish propaganda from rational persuasion or education. Indeed, in 1932 Bertrand Russell wrote that emotional propaganda is dangerous because “it tends to close the mind to argument” (p. 217), and Hitler himself noted in Mein Kampf that propaganda is not the medium of rational refection: The function of propaganda is… not to weigh and ponder the rights of diferent people, but exclusively to emphasize the one right which it has set out to argue for. Its task is not to make an objective study of the truth… its task is to serve our own right, always and unfinchingly. (1925, p. 182) Propaganda is often thought to be persuasion that merely asserts a conclusion without ofering an argument or reasons in support. But propaganda can present its audiences with arguments— arguments whose persuasive force depends not on their soundness, but on some other appeal which allows them to afect audiences’ beliefs, attitudes, sentiments, or behavior. This view of propaganda as irrational persuasion appears at least as early as 1933, when the sociologist Frederick Lumley identifed the content of propaganda as “unsupported, partially supported, or trickily supported conclusions” (pp. 148–9). Also part of this view of propaganda is the idea that propaganda “closes minds” by precluding rational engagement, or as Lumley put it, by “making further thought unnecessary” (p. 149). Not long after, the psychologist F. C. Bartlett developed this view in more detail, describing the ways in which propaganda “strives continually to paralyse critical analysis and to stimulate all tendencies to thoughtless and slavish acceptance” (1940, p. 66). This view of propaganda as bypassing rationality has remained reliably popular. Several decades after the emergence of propaganda scholarship, Jacques Ellul, one of the most infuential propaganda theorists, reiterated that propaganda “must constantly short-circuit all thought and decision” and work “at the level of the unconscious” (1965, p. 27). Ellul also identifed a key psychological efect which enables propaganda to circumvent reason—which he called “crystallization,” a process of organizing and entrenching stereotypes and other patterns of thought and action. Ellul’s concept of crystallization plays the role that ideology does in other theories of propaganda (e.g. Stanley, 2015). As a result, propaganda obviates critical refection, “hardens prevailing stereotypes, and… codifes social, political, and moral standards” (Ellul, 1965, p. 163). This crystallization of stereotypes—or ideology—makes propaganda efective. And propaganda itself further reinforces the very structures that make it efective in the frst place. Norms of rationality continue to play a central role in more recent and contemporary theories of propaganda. For example, both Ted J. Smith III and J. Michael Sproule take the lack of sufcient reason or evidence as propaganda’s central feature (Smith, 1989, p. 81; Sproule, 1994, p. 6). In the same spirit, Sheryl Tuttle Ross describes propaganda as “epistemically defective or lacking epistemic merit,” not only when it’s false, but also when there’s no rational connection between 127

Anne Quaranto and Jason Stanley

the target belief and other beliefs (2002, p. 23). In one of the most detailed such accounts, Stanley Cunningham identifes an abundance of epistemic norms that propaganda violates: it cultivates confusion; it exploits expectations; it poses as information and knowledge;… it systematically disregards superior epistemic values such as truth and truthfulness, understanding and knowledge; it discourages rationality, reasoning, and a healthy respect for rigor, evidence, and procedural safeguards; it promotes the easy acceptance of unexamined beliefs and supine ignorance. (2002, p. 176) 5 In a more recent elaboration of this view of propaganda as circumventing reason, philosopher Randal Marlin explains that propaganda can in fact bypass reason even as it appeals to reason, because the irrational efects are produced by the inclusion of “a hidden, misleading, or otherwise unexamined presupposition” which afects audiences’ reasoning in ways of which they’re not conscious (2013, p. 12). Propaganda can thus disguise itself as cooperative communication, purporting to ofer rational arguments while in reality ofering arguments that rely on various irrational techniques which we discuss throughout this chapter—such as framing, loaded language, stereotypes, symbols, innuendo, scientifc-seeming data—to sway audiences to draw certain conclusions which they wouldn’t draw on solely rational grounds. An ideology is a set of practices, including linguistic practices, which embed stereotypes and social meanings into actions (including speech acts). To summarize the above discussion— propaganda’s efectiveness relies on ideology (Stanley, 2015). It is because of ideology that propaganda can hide its unreason under the cloak of universal reason, its partiality under the mask of universality. Ideology assigns authority, characteristically illegitimate authority, to certain positions in society; for example, patriarchal ideology assigns special authority to men. The efectiveness of propaganda typically involves associating the source or content of its message with ideological sources of authority. Like the other techniques we’ve considered so far, this presents propagandistic claims as more reliable, or arguments as more reasonable, than in fact they are, thereby bypassing audiences’ rational faculties. Bartlett, for example, describes propaganda as most often taking the form of suggestion “based upon a relationship of superiors and inferiors” analogous to a doctor-patient relationship (1940, pp. 51–2). The propagandists’ having or at least arrogating some status or position above the audience lends their proposal an air of trustworthiness and expertise, or even more simply provides the role model audiences are implicitly or explicitly encouraged to follow.6 Rae Langton ascribes the efectiveness of the Nazi propaganda of Julius Streicher in part to its strategic confation of epistemic and practical authority. Its perceived epistemic authority comes from its source, the ruling political party, and that, in turn, gives it its practical authority (Langton, 2018). This use of an authority stance is perhaps most obvious in propaganda openly issued by governments or private institutions. However, even anti-establishment propaganda can use prestige to bolster its infuence. Though these conceptions of propaganda focus on propagandees’ coming to believe in certain ways, propaganda needn’t instill beliefs in order to produce epistemic or rational harms. Instead, it might pursue a more limited goal—instilling doubt. Because it doesn’t require convincing audiences of any specifc, refutable claims, this goal is more attainable now that audiences can easily access competing views and additional information online. The motto of RT, Russia’s propaganda station, is “Question More.” Stanley writes, describing its propaganda efects: RT’s strategy was not devised to produce knowledge. It was rather devised as a propaganda technique, to undermine trust in basic democratic institutions. Objective truth is 128


drowned out in the resulting cacophony of voices. The efect of RT, as well as the myriad conspiracy-theory-producing websites across the world, including in the United States, has been to destabilize the kind of shared reality that is in fact required for democratic contestation. (2018, p. 68)7 Yet another set of irrational efects of propaganda concern not the accuracy or reliability of a claim, but rather who gains from audiences’ coming to believe a certain way. That is, propaganda also bypasses or actively sabotages audiences’ rationality when it persuades them to form beliefs, attitudes, or intentions to act in ways that are counter to their own interests (though it is a thorny matter to characterize the relevant sense of “interest”). Here we see another clear contrast between propaganda and ideal, cooperative communication: whatever the audience interests might be, propaganda aims to further the propagandist’s interests only; it’s disinterested or even opposed to the interests of the audience. Cooperative communication, on the other hand, is typically assumed to serve the speaker’s and hearer’s shared interests.

2.1 Disguised and Covert Propaganda We began with the example of scientifc racism, which is an instance of propaganda presented as something else—as true, informative, as making a rational appeal or argument, or as being cooperative communication. Propaganda used in mass communication also employs this efect, as when xenophobic attacks on immigrants are presented as public health warnings about disease, for example. For J. Michael Sproule, this covertness—propaganda presented as something other than propaganda—is key evidence of the manipulative nature of propaganda, and the ability to bypass reason that this covertness enables is, in turn, precisely what distinguishes propaganda from rational persuasion: “Whereas the direct persuasion of a speech alerts our critical faculties that someone is trying to win us over, propaganda’s covertness hides the manipulative element in mass communication” (1994, p. 3). Indeed, as Randal Marlin observes, such covertness is crucial for achieving propaganda’s efects: “For propaganda to be successful, a targeted audience must not recognize what is communicated as propaganda” (2013, p. 187). There are distinct theories about the mechanisms that undergird propaganda’s covertness. Of course, we may expect multiple mechanisms to be in play. For example, even theorists focusing on other mechanisms would not deny that propaganda often relies on presupposition. One set of linguistic mechanisms are those introducing not-at-issue content: presupposition and various forms of conventional implicature, such as nonrestrictive relative clauses and appositives, expressives, epithets and honorifcs. The content contributed by these kinds of expressions includes both at-issue content, which addresses the Question Under Discussion, and not-at-issue content, which does not.8 At-issue content and not-at-issue content are governed by diferent norms; proposing atissue content invites direct challenges in a way that proposing not-at-issue content does not.9 This allows a propagandist to subtly produce certain efects on the audience, without the accountability that would typically accompany making that content at-issue. Not-at-issue content can be introduced in more or less subtle ways. At one end of the spectrum, emotions and normative beliefs or attitudes evoked by caricatures and stereotypes are typically blatant, though still distinct from their descriptive content. At the other, the normative or afective impact of symbolic and loaded language—like euphemisms, epithets, dog whistles, and other expressions with strong connotations or robust social meaning—may go unnoticed even by those who respond strongly to it. Further examples of this kind of language are Name Calling and Glittering Generalities, two of the seven classic techniques of propaganda identifed by the Institute of Propaganda Analysis in 1937. These devices function to sway audiences by associating the referent with certain ideals—whether vices, as in Name Calling with phrases like “Fascist, demagogue, 129

Anne Quaranto and Jason Stanley

dictator, Red,” or virtues, as in Glittering Generalities like “truth, freedom, honor, liberty, social justice” and “the American way.” Stanley (2015, Chapter 4) uses not-at-issue content to analyze code words and dog whistles. Other theorists identify pragmatic linguistic mechanisms as enabling propaganda’s covertness. Justin Khoo (2017) rejects the view that code words are best to be understood in terms of notat-issue content (what he calls the “multidimensionality theory of code words”). Not-at-issue content is characteristically non-cancelable. For example, it’s odd to say (1), but perfectly fne to utter (2) (ibid., p. 55): 1 2

John stopped smoking. He never smoked at all. Our welfare system provides needed services to many unfairly disadvantaged citizens. Every negative stereotype about poor Black people is false.

If “welfare” were a code word for negative stereotypes about poor Black Americans, (2) would be problematic to utter in the way (1) is. On Khoo’s view of code words, the felicity of (2) shows that the negative stereotypes that may be triggered by uttering “welfare” are in no way semantically encoded. Rather, code words invite their audiences to draw certain inferences from their pre-existing stereotypical beliefs, but these inferences remain independent of code words’ content. The mechanism Khoo identifes is not limited to code words. Framing efects and “spin” also subtly encourage the audience to draw certain inferences, without the propagandist having to outright present them for more careful consideration. Consider, for example, the diference George Lakof (2001) notes between framing the September 11th attacks as an act of war, versus as a crime: “The crime frame entails law, courts, lawyers, trials, sentencing, appeals, and so on.” The “war” frame, on the other hand, involves “‘casualties,’ ‘enemies,’ ‘military action,’ ‘war powers,’” and shifts the appropriate response from the courts to the military.10 Garth Jowett and Victoria O’Donnell argue that such framing efects can shape not only audiences’ cognitions, but also their perceptions, citing as examples “perception-shaping phrases that sanitize the reality of war” like “‘collateral damage’ standing for civilians killed or injured; ‘friendly fre’ for soldiers killed or injured by troops from their side; ‘turkey run’ for randomly killing a massive number of people; and ‘sorties’ for bombing missions,” as well as “military acronyms such as WMD and IED” (2015, p. 10). They argue that propaganda manipulates cognitions by targeting perceptions (or vice versa) because the two interact within a single complex process; cognitions (and more generally, attitudes) are based in part on perceptions, but existing beliefs or attitudes, in turn, infuence how we perceive the world (ibid., p. 12. See also Stanley, 2015, pp. 211–6). Another pragmatic account of some of the language of propaganda is ofered by Jennifer Saul (2018), who argues that dog whistling is a kind of perlocutionary speech act, one which is successfully performed only when the performance is covert, or not consciously noticed by the audience. On this account, one can explain the “cancelability” of dog whistling and code words as in (2) because the utterance in (2) is not a dog whistling speech act of this kind. According to the theory in Beaver and Stanley (forthcoming), code words and dog whistles do encode negative messages as part of their not-at-issue content. But Beaver and Stanley adjust the theory of not-at-issue content to give it the fexibility to account for what appears to be cancelability, by incorporating a speech-act theoretic account into not-at-issue content. Whether these mechanisms are semantic or pragmatic, what they ultimately have in common is that they allow the propagandist to connote certain normative claims without having to assert them, thereby hiding the manipulative aspect of propaganda (Sbisà, 1999). As F. C. Bartlett pointed out in 1940, symbols are efective propagandistic devices precisely because their ability 130


to convey more than one meaning allows the propagandist to hide the manipulative aspect of the message beneath its surface-level meaning: Indirect suggestion looks always as if it is aiming at one thing, but its real purpose is something diferent…. It is the mark of the symbol always to have at least two meanings: one of them looks to be obvious and open to everybody, but the other produces efects without ever coming out into the light. (p. 63) Formal models of discourse structure can help clarify how these strategies work. For example, we might view propagandistic spin from within Craige Roberts’ theory of discourse as structured by a series of motivating questions which the participants are interested in answering. On this view, what the propagandist does is subtly, even covertly, change the Question Under Discussion (QUD), by saying something which answers a diferent question than the one that has actually been structuring the discourse up to that point. The implicit proposal to change to the QUD is often simply accommodated by hearers, and so may easily go unnoticed. At the same time, the choice of QUD makes a diference to whether some discourse is manipulative. The speaker’s implicit QUD may automatically rule out certain answers in a way that precludes the audience from being able to fully rationally evaluate the matter, but because the change in QUD remains implicit, the claim may seem neutral. A similar mechanism—implicitly ruling out certain answers without even considering them— is at play in censorship and biased reporting. Messages produced under these conditions are propagandistic when they oversimplify reality, while at the same time being presented as rational, informative, or cooperative communication. Edward Herman and Noam Chomsky (1988) argue this form of propaganda also encompasses other, more subtle ways in which individuals and institutions with political and economic power constrain mass media to stife dissent and preclude debate, such as ownership of the media by wealthy elites and media’s reliance on advertising for income. These conditions, like outright censorship and bias, also produce messages which purport to be informative but are actually one-sided. Such propaganda is manipulative and epistemically harmful because it “irrationally closes of certain options that should be considered,” in Stanley’s words, thereby persuading audiences to accept a view on poor grounds (2015, p. 49). Lastly, propaganda’s being manipulative is not the only thing that’s often hidden. The more common form of propaganda today is covert or “black” propaganda, in which the author or source of the propaganda is hidden (as contrasted with the overt or “white” propaganda issued by self-identifed ministries of propaganda during World Wars I and II).11 Covert propaganda can take the form of whisper campaigns and rumors, or it can simply be presented as coming from a diferent source than it actually does (for example, from a panel of experts, rather than from a corporation). This form of propaganda is useful in part because the audience’s inability to identify the source makes it that much harder to check the truth or reliability of the claims made. But it’s also useful because, unlike in cooperative communication, the audience of covert propaganda cannot know the speaker’s intentions. Thus, it’s particularly difcult to rationally evaluate the reliability of such propaganda’s claims, even apart from questions of fact. The result is that when it’s efective, such covert propaganda leads the audience to think that the acquired belief or attitude is their own.

2.2 False or Misleading Propaganda Besides hiding its argumentative or manipulative nature, yet another way propaganda bypasses reason is by spreading false beliefs or encouraging irrational inferences. However, it’s a misconception that propaganda must be false (see, e.g., Ellul, 1965, pp. 52–4; Stanley, 2015, pp. 42–3). More 131

Anne Quaranto and Jason Stanley

often, propagandistic assertions are neither clearly false nor clearly true. Such “gray” propaganda or disinformation may convey a mix of truths and falsehoods, or even just carefully selected truths, with the overall efect of misleading the audience or presenting information whose reliability is uncertain. Simplifcation is one classic propagandistic technique that can mislead or manipulate audiences without asserting any falsehoods.12 Propaganda can also make use of literally true claims. In fact, truths will often make for more efective propaganda than falsehoods, as apparently Lenin and Goebbels both recognized, because they’re harder for audiences to reject or discount.13 Other propagandistic messages are simply not truth-evaluable, such as slogans in the form of imperatives—“Don’t tread on me,” “Proletariat of the world, unite!” and “Make America Great Again,”—or otherwise falling short of the subject–predicate sentence form—“liberté, égalité, fraternité” (liberty, equality, brotherhood) in the French Revolution or “ein Volk, ein Reich, ein Führer” (one people, one empire, one leader) in the Third Reich.14 Symbols more generally are particularly useful in propaganda for this very reason: an audience thinking in terms of abstract, vague symbols is easier to manipulate, because it’s harder to check whether a claim is true when it involves symbols like liberty or equality.15 Propaganda’s use of narratives, and especially revisionist histories, similarly complicates the question of its truth or falsity. Political scientists James Combs and Dan Nimmo identify the suspension of disbelief as a key technique for making propaganda appealing and efective, one for which the narrative form is particularly well-suited (1993, p. 89).16 Narrative form also provides unique opportunities for misleading and manipulating, related to the technique of framing we discussed earlier. For example, which events are included or excluded from a narrative can radically shift both the emotional impact and the moral that the audience will take away. Combs and Nimmo note that narrative form expands the range of techniques available to a propagandist, to include not only verbal content, but also gestures, which, like dog whistles and symbols, “convey meanings often quite apart from their apparent ones.” Combs and Nimmo thus show how propaganda can bypass rationality by taking the narrative form: “Persuasion is less information oriented than persona and performance oriented…. the purpose of the spectacle, or story, is to override disconfrming facts” (1993, p. 120). Propaganda thus challenges not only theories of meaning, but theories of truth as well. Propagandistic use of narratives is but one kind of case in which a simple correspondence theory of truth cannot sufce to explain the way in which propaganda is false, or at least inaccurate. As Cunningham observes, a pragmatist or coherence theory of truth could capture a kind of truthfulness that propaganda shares with civic rhetoric: “Goebbels, agitators, and political advertisers may indeed defend their messages as conducive to desires or socially desirable results… or they may champion their message as consistent with other reports and ideologies” (2002, p. 112). Moreover, the coherence in question needn’t be restricted to coherence among beliefs; good propaganda, at least, certainly exhibits an emotional coherence as well. Propaganda also dramatizes the interaction between semantics and pragmatics, and the communicative functions made possible by this interaction. Like the use of narrative form, the various techniques for adding not-at-issue content to a discourse or inviting inferences, discussed above, enable speakers or authors to produce messages that are half-true, or true but misleading. Innuendo is another way that literally true claims can function as propaganda. For example, Stanley invites us to consider “a non-Muslim politician in the United States saying, ‘There are Muslims among us’” (2015, pp. 42–3; also see Ellul, 1965, pp. 56–7). The literal truth or falsity of such an utterance doesn’t yet tell us whether or why it’s propaganda. Innuendo, among other propagandistic techniques employing truths, thus highlights the limited explanatory power of semantic notions alone. Explaining why a claim like this is propaganda will need to involve something besides the semantic content, value, or meaning of the words involved. Gricean conversational maxims might be able to do this explanatory work; perhaps an utterance of “There are Muslims among us” 132


violates the principle of relevance or quality by stating the obvious, and hearers thus interpret the utterance as implicating something more than just its literal meaning. Moreover, the pragmatic efects of some utterances derive not from their semantic content, but instead from a lack of certain content. Eric Swanson (2017) has shown that an important category of propaganda can be explained in terms of broadly Gricean notions—in terms of what he calls omissive implicature. As he explains, the Nazis often implicated permission for violent actions of their supporters by not apologizing for them: The omission of an apology when it’s manifest that one is expected conveys, through omissive implicature, that the speaker (or would-be speaker, if the relevant conversational participant is silent) does not have sufcient reason to apologize, because the speaker (or would-be speaker) is thwarting the manifest expectation of an apology. So by failing to apologize for the “excesses of the lower ranks,” the Nazis implicated that they did not have sufcient reason to so apologize. This omissive implicature in turn invites interpreters to strengthen the implicature by asking: why not? By failing to apologize, did the Nazis implicate that they did nothing wrong? (Swanson, 2017, p. 129) Swanson’s discussion reveals the point familiar from propaganda studies: what’s left unsaid can be the propagandizing (e.g. Stanley, 2015, p. 55). A further complication is that propaganda can also use truths sincerely, where the audience hears a true fact without innuendo or implicature. One form of this kind of propaganda is what Jacques Ellul calls rational propaganda, which is “based exclusively on facts, statistics, economic ideas” that—at least on their face—appeal to reason (1965, p. 84). However, since most in the audience aren’t technical experts, the rational aspect of such propaganda is spurious. What makes rational propaganda efective is not the objective, scientifc, or factual argumentation, but rather the impression of truth and reliability. What remains with the individual afected by this propaganda is a perfectly irrational picture, a purely emotional feeling, a myth. The facts, the data, the reasoning—all are forgotten, and only the impression remains…. Thus propaganda in itself becomes honest, strict, exact, but its efect remains irrational. (Ellul, 1965, p. 86)17 Another propagandistic use of truths is in bureaucratic propaganda, a notion theorized by the sociologists David Altheide and John Johnson. They defne bureaucratic propaganda as “any report produced by an organization for evaluation and other practical purposes that is targeted for individuals, committees, or publics who are unaware of its promotive character and the editing processes that shaped the report” (Altheide and Johnson, 1980, p. 5). For example, Altheide and Johnson discuss the infation of crime statistics by police departments to acquire federal grants or the defation of the same statistics to manage the reputation of a city (ibid., pp. 23–4). Echoing Ellul’s idea that propaganda takes a utilitarian attitude toward truth, they observe that “bureaucratic propaganda uses truth for organizational goals… by presenting managed and often contrived reports as though they were done ‘scientifcally’ and therefore depict ‘objective’ truth” (ibid., p.  23).18 A related kind of propaganda, the subject of Stanley (2016), is technocratic propaganda, where decisions are presented as the result of a kind of technocratic decision making whose details are not accessible to non-experts. Ultimately, these kinds of propaganda persuade not by ofering good reasons, but by embodying the authority of objective, factual data. Despite occasionally relying on true claims, these techniques are still aimed at bypassing rationality, like the other, more obviously manipulative ones. 133

Anne Quaranto and Jason Stanley

2.3 Propaganda’s Exploiting Emotion Besides manipulating audiences into believing incorrectly or reasoning irrationally, propaganda also bypasses audiences’ rational faculties by appealing to or exploiting their emotions and sentiments. Note that this set of techniques and the ones described in the preceding two sub-sections are not merely compatible, but synergistic. Indeed, writing in 1962, Ellul had already noticed that “purely impassioned and emotional propaganda is disappearing” and that “Hitler’s most infammatory speeches always contained some facts which served as base or pretext” (p. 85).19 Even rational propaganda, which, as we discuss above, presents audiences with facts and seemingly rational appeals, ultimately relies on emotional pressure to incite action (Ellul, 1965, p. 86). Propaganda often appeals to emotion precisely in order to manipulate audiences into accepting falsehoods or drawing irrational inferences. Bartlett, for example, points out that whether or not propaganda ofers an argument or merely asserts a claim, its acceptance relies on the existence of “some approved outburst of popular emotion” (1940, pp. 73–4). Appeals to emotion are also key according to Stanley, who notes that propaganda’s persuasive force lies not in its providing reasons, but rather in its “seeking to overload various afective capacities, such as nostalgia, sentiment, or fear” (2015, pp. 53). Propaganda’s reliance on emotional appeals also highlights the priority of emotional coherence and efcacy, rather than rational coherence and truth. At the same time, if exploiting emotions is propaganda’s central goal, then it becomes clear when and why truth is valuable to the propagandist: propaganda employs true claims not with an eye to the epistemic virtues of rational persuasion, but only when they’re the more efective way to achieve the propagandist’s goal. A number of theorists agree that propaganda can’t create completely new sentiments in its audience, but instead must use and reshape existing ones.20 So, propaganda can mobilize audiences’ existing emotional associations with certain people, things, words, concepts, or images, or it can displace existing emotions from one object or issue to another, or from one part to the rest.21 In particular, it often aims to transfer reasonable emotions to a new object in an unreasonable way. For example, propaganda may take anxiety about an uncertain economic future, and attach it to immigrants as the source. The emotional force attached to a propagandistic message can also mask the fact that the audience is being persuaded to believe or act in ways that serve not their interests, but those of the propagandist. One technique particularly useful to this end is appealing to in-group sentiments or stereotypes about out-groups. Such appeals can distract audiences from some internal threat by shifting their discontent onto a scapegoat. They can also encourage action which, though it’s at the out-group’s expense, is actually counter to the in-group’s interest. More generally, propaganda’s use of emotion allows it to bypass rationality because the emotional force helps mask the irrational aspects of propaganda, such as simplifcation, exaggeration, or a lack of argument or supporting reasons. As Bartlett argues, were it not for this emotional impact, propaganda’s rational or epistemic defciencies would surface and would “excite at least as much popular criticism as any other form of statement” (1940, p. 78). Emotional appeals can thus help bypass rationality not only directly, but also indirectly by bolstering another technique already discussed above—masking the persuasive and manipulative aspects of propaganda. Many of the linguistic techniques we’ve already discussed above also work to activate and exploit emotions in the service of propagandistic manipulation—framing and spin; loaded language like euphemisms, epithets, and stereotypes; symbols and metaphors; and claims to authority. Symbols are perhaps the most clearly useful for such emotional appeals, as their contribution to discourse consists not of propositional, truth-conditional content, but rather of values, norms, ideals, and more generally afective elements. We’ve already seen propaganda’s reliance on single words referring to broad, vague, and normatively laden concepts, such as “freedom” or “democracy,” as 134


well as slogans like “Don’t tread on me,” “Proletariat of the world, unite!” and “Make America Great Again.” Such loaded, symbolic language bypasses audiences’ critical faculties by encouraging them to transfer sentiments or emotions from one object or issue to another. The infuential journalist and political commentator Walter Lippmann, for example, identifes the centrality of symbols to propaganda’s ability to exploit pre-existing emotional associations, describing propagandistic technique as “the use of symbols which assemble emotions after they have been detached from their ideas” (1922/2012, p. 38). At the same time, symbols’ afective force persuades audiences to accept claims or conclusions without adequate supporting reasons. Like propagandistic emotional appeals more generally, propagandistic use of symbols serves to bypass rationality by masking propaganda’s irrational aspects.

2.4 Propaganda as Incitement to Action Some scholars have argued that propaganda primarily targets audiences’ actions and behaviors, not their beliefs or emotions. Perhaps most infuential among those advancing this view is Jacques Ellul, who argues that “new” propaganda (as contrasted with the “old” war-time propaganda) is centrally concerned with “efectiveness” (1965, p. x)22—that is, with manipulating audiences into behaving in certain ways.23 Of course, beliefs and emotions are still a secondary target, because transforming audiences’ behavior will also transform their beliefs, desires, and attitudes. Ellul ofers a particularly insightful explanation of why targeting behavior is ultimately more efective than targeting beliefs or attitudes directly: “He who acts in obedience to propaganda can never go back,” Ellul writes. He is now obliged to believe in that propaganda because of his past action. He is obliged to receive from it his justifcation and authority, without which his action will seem to him absurd or unjust, which would be intolerable…. He is what one calls committed. (Ibid., p. 29) Perhaps surprisingly, this view of propaganda as behavior-manipulation is not a competitor to, but rather another variant of, the view of propaganda as reason-bypassing. Despite the focus on action and behavior, on Ellul’s view what’s distinctive of propaganda is still the fact that, rather than persuading rationally, it manipulates, “short-circuiting the intellectual process” (ibid., p. 30). This short-circuiting occurs in a two-stage process. Before propaganda can actually incite any actions, much less produce any habitual behaviors, there must frst be a conditioning process that sets in place the attitudes and refexes which later propaganda will exploit (ibid., p. 32). This initial phase thus aims to associate words and other symbols with certain behavioral and afective responses, without yet calling for concrete action or connecting its messages to any specifc goal or issue. This process quite often relies on myths—normatively and afectively powerful images shared by a population, like myths of work, happiness, the nation, youth, or the hero (ibid., p. 40). Having been abstracted away from a concrete reality, myths come to encompass all that is of value—all that is good, just, and true (ibid., p. 31). “Strongly colored, irrational, and charged with all of man’s power to believe,” myths can thus spur to action much more efectively than rational persuasion could (ibid., p. 40). With the right myth in place, propaganda can mobilize audiences by connecting a particular situation or issue with the values encompassed by the myth. Because myths are emotionally powerful, appealing to a myth will give the audience a sense of urgency and make them feel that action is required. But the myth will also show audiences how to act and assure them that such action is appropriate and will lead to success (ibid., pp. 184, 209). 135

Anne Quaranto and Jason Stanley

3 Propaganda’s Socio-political Efects In the previous section, we considered various irrational efects that propaganda has on individual audience members. But as has been observed from the early days of propaganda analysis, propaganda doesn’t target individuals merely as individuals, but as members of certain groups, communities, or societies. (See, e.g., Bernays, 1928, pp. 44, 55; and Ellul, 1965, pp. 7–9.) Propaganda exploits the values and sentiments we hold in virtue of our social relations and group memberships, both to get our attention and to afect us. Such appeals, in turn, afect those groups, communities, and societies as a whole.

3.1 Legitimizing and Delegitimizing The techniques we’ve seen so far—whether they’re aimed at manipulating audiences’ beliefs, attitudes, or behavior—can serve to legitimize or delegitimize political or social institutions. During a revolution or by a subversive group, agitation propaganda aims to delegitimize an established social or political order. It fxates on, exaggerates, or even manufactures crises to convince audiences of the urgency of a situation, and it directs audiences’ anger and hatred into drastic, even violent action, often against some scapegoat or common enemy (Ellul, 1965, pp. 72–3). While agitation propaganda is only efective in brief bursts, integration propaganda works over long periods of time to produce conformity in audiences’ thought and behavior, with the aim of legitimizing institutions ( Jowett & O’Donnell, 2015, p. 315). Although it’s less dramatic and although examples may not come to mind as easily, integration propaganda is no less important or paradigmatic than agitation propaganda. In fact, it is the more common form, not least because its success depends on its being widespread and continuous, repeating its message in a variety of ways in order to gradually, imperceptibly establish or maintain the legitimacy of some institution, in the audience’s mind (Ellul, 1965, pp. 17–18). Integration propaganda interprets and explains events and issues to the audience, as well as rationalizing and justifying the actions of an institution. To perform these tasks, integration propaganda exploits and adapts as needed the myths, symbols, values, and ideals that its audience already holds. Unsurprisingly, integration propaganda is produced by politically, economically, or socially dominant groups, but it targets both marginalized groups and the dominant groups themselves. For example, economic elites justify maldistributions of wealth by appealing to the ideals of individualism, self-sufciency, and work ethic—legitimizing the existing economic order by attributing concentrations of wealth to merit and just desert. Such propaganda serves to pacify those without much economic power, redirecting their energies into pursuing the individualist ideals. But it also serves an important function for the economic elites; it relieves them of guilt by assuring them of the justice and merit of the existing order.24 Two key outlets for propagating and maintaining these ideals and myths are schools and mass media, both of which reach nearly everyone in a modern polity.25 Censorship, though a more drastic and hence rarer technique, also serves this legitimizing function, by precluding any views which would threaten the standard justifcation. Conformity in values and ideals, success of the proposed rationalization or justifcation, and the stability of the existing order are interdependent: the more widely accepted these myths and ideals are, the more convincing will be the justifcation, and thus the more efective the propaganda at maintaining the status quo (Ellul, 1965, p. 200). Another kind of legitimizing propaganda is produced by political bureaucracies, as theorized by David Altheide and John Johnson’s work on bureaucratic propaganda introduced above. Bureaucratic propaganda takes the form of ofcial reports by governments and other organizations, which produce such propaganda precisely to legitimize themselves to other institutions and to the public more generally (Altheide and Johnson, 1980, p. 18). In particular, what legitimates an 136


organization is the scientifc, objective appearance of its reports—and by extension, its work. By presenting themselves this way, organizations appeal to our commitments to rationality and aim to persuade us to accept their work as legitimate as science. At the same time, insofar as we already accept similar organizations as legitimate, an organization can legitimize itself by demonstrating how its reports and its work function in the same way (Altheide and Johnson, 1980, pp. 26–32). As an example, Altheide and Johnson point to the increasing bureaucratization and marketization of higher education (ibid., p. 230). Although bureaucratic reports are not produced with an express intention to mislead or manipulate, they are aimed to convince superiors and other ofcials of the legitimacy and efciency of the ofce producing them. But when the audience of these reports changes, so does their efect. Altheide and Johnson argue that when such reports are disseminated to the public via the mass media, their purpose changes…. For example, police crime statistics are no longer merely one way of organizationally accounting for the kind and amount of certain types of work done by individuals as a way of justifying expenditures and salaries; crime statistics are now regarded as “objective” indicators of the amount of crime in our society and the threat it poses to all her safety unless the tide is turned. (Ibid., p. 18) In such a change of context, bureaucratic propaganda might end up legitimizing a rather diferent institution than intended, or even—as in the case of crime statistics—reinforcing harmful stereotypes.

3.2 In- and Out-Grouping Propaganda nearly always exploits some aspect of audiences’ social identities, to appeal to the associated emotions and values and thereby make itself more efective. But propaganda can also aim primarily at producing or maintaining in- and out-group distinctions. While war-time propaganda is most clearly directed at a specifc enemy group, some theorists see it as central to all propaganda that it presents a specifc socio-political group (e.g. ethnic, national, or religious) as Other or even as enemy.26 At the same time, propaganda implicitly defnes its audience in opposition to this out-group; by identifying the Other, it also identifes Us.27 The use of stereotypes is clearly central to these goals. Part III of Walter Lippmann’s classic 1922 book Public Opinion is called “Stereotypes,” and it is generally considered the source of this term in connection with the formation of public opinion. Lippmann argues that stereotypes are inevitable, as they allow us to structure the “great blooming, buzzing confusion of the outer world” (Lippmann, 1922/2012, p. 63). As he adds, “what matters is the character of the stereotypes” (p. 70). Lippmann characterizes stereotypes as “an ordered, more or less consistent picture of the world, to which our habits, our tastes, our capacities, our comforts and our hopes have adjusted themselves” (p. 73). Lippmann argues that stereotypes form the “core of our personal tradition, the defenses of our position in society” (ibid.). “No wonder,” he adds, “that any disturbance of the stereotypes seem like an attack upon the foundations of the universe” (ibid.). Lippmann is clear about the role of stereotypes in justifying hierarchies; indeed, he argues that Aristotle’s description of the “natural slave” in his justifcation of slavery in The Politics is “the perfect stereotype.” And Lippmann’s diagnosis of the mistake in Aristotle’s argument is that it appeals to this stereotype, which obscures basic facts about the humanity of those who are enslaved (p. 75). Key to propaganda’s reason-bypassing function, stereotypes are resistant to evidence, in Lippmann’s description “[imposing] a certain character on the data of our senses before the data reaches the intelligence” (ibid.). And Lippmann is vivid about the role stereotypes play in the 137

Anne Quaranto and Jason Stanley

formation of “friend–enemy distinctions,” a particular form of in-group/out-group distinction that underlies war and other forms of mass violence. “Out of the opposition we make villains and conspiracies” (p. 101). Positive stereotypes of an in-group encourage pride in that social identity, strengthening the identifcation. In forming an “enemy” out-group, negative stereotypes unite the in-group in their fear, contempt, fear, or hatred.28 Nazi propaganda minister Joseph Goebbels emphasized the usefulness of anxiety to this technique and the importance of striking the right balance between too much and too little anxiety (Doob, 1950, pp. 438–9). Often alongside stereotypes, historical narratives—whether real or revisionist—serve to unite the target audience against some outgroup that is presented as having wronged them. Anti-Semitic Nazi propaganda often cast Jewish people not only as the enemy, but also as the scapegoat. For example, a Nazi poster from 1943 proclaims “Der ist Schuld am Kriege!” (“The war is his fault!”), directing Germans’ war-time anxieties and frustrations into hatred for Jews. More subtly, such propaganda can employ the variety of techniques for introducing not-at-issue content, which we’ve discussed earlier. Racist dog whistles like “inner city” are an especially clear example; they activate audiences’ racist prejudices, with plausible deniability for the speaker and often without hearers consciously recognizing the problematic message. More drastically, such propaganda sometimes employs derogatory language that demeans or even dehumanizes members of the out-group. Considering one such example, Lynne Tirrell argues that leading up to and during the Rwandan genocide, propaganda calling Tutsis derogatory terms like the Kinyarwanda “inyenzi” (cockroach) and “inzoka” (snake) functioned not only to encourage attacks on Tutsi communities, but also to normalize and justify such attacks. This dehumanizing language associated Tutsis with creatures that evoke disgust and deserve extermination. Echoing Ellul’s analysis, Tirrell stresses that derogatory language afects not only audiences’ beliefs or emotions about the out-group, but their behavior toward the out-group as well. “Because of the action-engendering force of derogatory terms,” she argues, “actions hitherto unthinkable (i.e. the extermination of a people) came to be regarded as socially appropriate and even required” (Tirrell, 2012, p. 176). Propagandistic appeals to in-group identity and negative attitudes about the out-group entrench existing in- and out-group divisions and associated attitudes, but this typically isn’t the main goal. Rather, the polarization, scapegoating, and pride involved serve some other practical interests of the propagandist. Goebbels, for example, noted that one of the functions of propaganda is to ease feelings of aggression by directing them into hatred for a specifed enemy (Doob, 1950, p. 440).29 Bartlett observes that, although propaganda often invokes a crisis threatening the audience’s in-group and portrays the threat as coming from outside, “the real aim is to organise the enthusiastic elements of the group in support of the current direction of internal development” (1940, p. 80). At the level of an entire community, another function of in- and out-grouping becomes apparent—the homogenization of the in-group’s thought, or what Ellul termed crystallization. As we’ve already seen, homogenization is itself a central goal of integrative propaganda, even in instances that don’t target any out-group. Stereotypes help produce this homogenization because they don’t require much efort to interpret and internalize. Via stereotypes, propaganda ofers a simple, undemanding framework through which the audience can understand the world around them. The more such stereotypical thinking spreads, the more it pushes out competing theories. In Ellul’s rather cynical conclusion, the result of such a process is that “man can no longer modify his judgments and thought patterns” (1965, p. 164). Members of the in-group thus become less and less able to take on the thought patterns of members of other groups, further deepening the divisions between them.30 Of course, the efects on the in-group are not the only ones that matter in polarizing a community; the manipulation of out-group members also contributes to the polarizing efects. 31 Indeed, 138


members of marginalized groups often internalize the negative stereotypes about them that are promulgated in the interests of the dominant group.32 The efects produced on the in-group work alongside those produced on the out-group to strengthen the divisions between them, while also often weakening the out-group’s social or political standing. That is, propaganda doesn’t simply refer to existing divisions between in- and out-groups, nor does it merely produce such divisions. Rather, it also enforces or even produces a hierarchy between groups. Indeed, the very concept of an out-group already suggests a hierarchy: to identify a group as Other isn’t to make a neutral claim about the world, but rather to present that group as lesser, as undeserving of the social status, rights, privileges, and powers that the in-group accords its own members. Especially when it’s internalized, Othering alienates those members of the out-group who are part of the same polity or society as the in-group, and seeing themselves vilifed is likely to inspire insecurity and even fear among some. When this occurs, members of the out-group begin to see themselves as Other, losing trust in their neighbors and political representatives, and losing confdence in their own social or political status. When they lose faith in their own social and political power, the out-group is liable to really lose such power, as their fear and insecurity might mean they make fewer demands, become compliant, and accept a weakened status. For example, reports of domestic abuse by non-naturalized residents in the US fell in response to increased activity by and publicity for ICE (Engelbrecht, 2018). Of course, members of the out-group might instead resist their vilifcation and alienation, reasserting their membership in the polity or society, demanding fair and equal treatment, and possibly even launching counter-propaganda. But now that they’ve been marked out for exclusion, the out-group will have to work harder to prove their membership in a shared identity. They’ll be held to higher standards for behavior that has higher stakes. Moreover, as Stanley argues, if the propaganda they’re responding to makes use of implicit stereotypes, members of the out-group will have to accommodate “however provisionally, the negative stereotype of their group, simply to enter smoothly into any conversation about their group with members of the dominating group” (2015, p. 163).

4 Normative Status Propaganda has a strong negative connotation to many, though not all, English speakers today. This is largely a post-World War II phenomenon, recognizable both in ordinary intuitions about its connotations, and scholarly writing. Cunningham, for example, maintains that propaganda is “an inherently unethical social phenomenon” and argues that “to describe propaganda without reference to its unethical complexion is to fractionalize it, and to minimize it” (2002, p. 176). Yet it is common to fnd positive post-war references to propaganda in the work of scholars and activists in the African-American tradition. In her 2006 book Bodies in Dissent, Daphne Brooks calls the performances of Henry “Box” Brown, who reconstructed his dramatic escape from slavery on stage, “part abolitionist propaganda” (p. 66). And in a 1954 sermon, “Propagandizing Christianity,” Rev. Martin Luther King wrote, “…propaganda does not have to be evil. There is a noble sense in which propaganda can be used.” Indeed, “the great debate” of the Harlem Renaissance, between W.E.B. Du Bois and Alain LeRoy Locke, was about the efcacy of using art as anti-racist propaganda; hence the title of Alain Locke’s 1928 reply to Du Bois, “Art or Propaganda?” (See Harris, 2004.) A relatively simple way to argue for the unacceptability of propaganda is to focus on its ill effects, including on the moral status of the actions or behaviors that propaganda provokes (or aims to provoke). This kind of argument is perhaps most convincing in the case of propaganda that creates or strengthens ethnic, racial, or religious in-/out-group efects, and particularly propaganda that relies on or incites feelings of hatred to do so. Anti-Semitic Nazi propaganda is a prime example here. Insofar as it dehumanized and demonized a particular group, that propaganda is 139

Anne Quaranto and Jason Stanley

reprehensible. Moreover, insofar as it encouraged citizens’ support for and complicity in the Holocaust, Nazi propaganda is reprehensible because it resulted in morally reprehensible actions on the part of those citizens. A related argument grounds propaganda’s reprehensibility in its harming or at least opposing the greater social interest. Leonard Doob considers this essential to propaganda, which he defnes as directed at “ends considered unscientifc or of doubtful value in a society at a given time” (1948, p. 240). Perhaps surprisingly, on Doob’s defnition it turns out that certain messages, which seem to share propaganda’s manipulative techniques or aims, aren’t actually propaganda. For example, certain governments require cigarette manufacturers to print warnings on each pack of cigarettes, to discourage smoking or to encourage smokers to quit. Like many of the other pieces of propaganda we’ve considered, these warnings aim to persuade audiences by targeting their emotions and producing an aversive response; they take a catastrophic tone and are often accompanied by graphic images depicting the most serious consequences of smoking (Stanley, 2015, pp. 58–9). If propaganda is counter to the interest of a polity, and encouraging smoking cessation is in the polity’s interest, then such warnings are not, according to a view like Doob’s, propaganda. Parallel arguments have been made in support of the opposite conclusion of Doob’s, that there can be morally praiseworthy uses of propaganda, ones which have benefcial outcomes. Both Stanley and Marlin develop theoretical accounts of propaganda that include politically, morally, and even (as they argue) democratically acceptable uses of propaganda. Drawing on the “civic rhetoric” tradition, Stanley argues that propaganda can be used not just to harm democracy, but to repair it, by calling attention to perspectives that ideology has rendered invisible (Stanley, 2015, pp. 111f ).33 For example, on the analysis profered by Daphne Brooks (2006), “abolitionist propaganda,” such as Henry Box Brown’s emotionally gutting performances of his mailing of himself to freedom, forced his audiences to challenge their own racist ideologies, by showing that those who are enslaved not only have agency, but also have a powerful desire to exercise it.34 Marlin argues that, although propaganda persuades audiences not through rational means, there are cases where such persuasion is not only acceptable, but desirable. This is the case when the beliefs or values that the propagandist seeks to inculcate are morally good or socially desirable ones—such as in elementary education, in particular when the aim is to inculcate a moral belief in children (Marlin, 2013, p. 175). He also argues that in some cases, propagandistic persuasion is acceptable precisely because it’s efective in a way that non-propagandistic persuasion couldn’t be, such as in “the teaching and inculcation of religious belief… where purely rational forms of persuasion seem inadequate for the purpose” (ibid.). As Rev. Martin Luther King did before him, Marlin concludes that, insofar as there is a positive value to religious faith, and propaganda is the only way to instill such faith, religious propaganda is not only acceptable, but even morally praiseworthy.35 If propaganda is a form of communication, then its being neutral is unsurprising; after all, other, non-propagandistic and even non-persuasive forms of communication also come in good or bad variants. Accordingly, this view of propaganda as neutral is often accompanied, if not justifed, by framing propaganda as merely a tool, which can be used for good or for evil.36 Whether or not a particular piece of propaganda is problematic will thus depend on who’s using the tool and to what ends. For example, Edward Bernays, one of the earliest students of propaganda and the “father” of public relations, insisted in 1928 that “whether, in any instance, propaganda is good or bad depends upon the merit of the cause urged, and the correctness of the information published.”37 Though Stanley (2015), Rogers (2012), Garsten (2006) and others argue that propaganda can be democratically acceptable, such as when a democracy is fawed, others argue that propaganda is inherently anti-democratic, as it uses anti-democratic methods to achieve results. For example, Ellul notes in his preface, “If I am in favor of democracy, I can only regret that propaganda renders the true exercise of it almost impossible” (1965, p. xvi).38 140


A frst pass at spelling out this confict between democracy and propaganda relies on classical liberal notions of legitimate power deriving from the social contract by which the true bearers of power (the people) invest governments with power. On this assumption about legitimate power, propaganda is incompatible with legitimate government more broadly, and democracy in particular, because it’s inherently coercive. If propaganda really manufactures consent or shapes public will, but democracy depends on citizens’ [tacitly] consenting to be governed, then a contradiction arises: in the liberal ideal, power fows from the people to the government, but with propaganda, the direction seems reversed. Another tension arises between the conception of a democratic citizen and the conception of a propagandee. On the one hand, democracy seems predicated on the idea that every citizen makes rational decisions based on their own self-interest. On the other hand, propaganda circumvents our rationality, exploits our irrationality, or even makes us irrational (as discussed above). So, propaganda either demonstrates the falsity of this fundamental liberal assumption, or actively makes it impossible for the assumption to be reality. However, the classical liberal views of legitimate political power, individual citizens’ psychologies and rationality, and individuals’ relation to and participation in state activities don’t seem to refect the reality of modern democracies. More realistically, even in a modern developed democracy, few if any citizens have the time or even the inclination to be informed well enough—not to mention educated adequately—to make rational decisions on every political issue. Bernays and Ellul both point to the complexity of modern societies and polities, concluding that it would be impossible for any one individual to fully understand all aspects of all issues on which they might have an opinion. A democracy, Bernays argues, requires an “invisible, intertwining structure of groupings and associations” to have “organized its group mind and simplifed its mass thinking” (1928, p. 44). This structure, which includes political parties as well as social clubs and informal associations, allows opinions, values, norms, and desires to be spread and homogenized, just as fashions are. Moreover, to the extent that we do have opinions on a number of political issues, they are constantly in fux, often irrational, and rarely accurately expressed in vote tallies (Ellul, 1965, p. 124). Propaganda ofers a way for citizens to participate (or feel as though they’re participating), without government policy having to obey such unreasonable and unreliable opinions. Ellul argues that “even in a democracy, a government that is honest, serious, benevolent, and respects the voters” will need to use propaganda (ibid., p. 126), precisely to shift public opinion (or public will) in the desired direction and “make the masses demand of the government what the government has already decided to do” (ibid., p. 132). This argument, however, takes a rather cynical view of the average citizen’s psychology. It depicts, or at least tacitly assumes, that audiences are irrational, ill-informed, and incapable of considering complex, nuanced evidence. Bernays is not alone in pointing out that propaganda must present simplifcations because the average person thinks in terms of clichés (1928, p. 74). Ellul, too, argues that, the average citizen in a democracy wants to participate but is incapable of considering the full complexities of, say, a foreign policy. Propaganda fulflls this need by providing a simple explanation of all the information available to them, a coherent framework that can explain and synthesize all the information available to such a citizen, without demanding any work on their part (Ellul, 1965, pp. 139–47). The audience are seen not just as poor reasoners, but also as ignorant of their own desires or values. Referencing his uncle Sigmund Freud, Bernays discusses at length a number of examples of how people are “rarely aware of the real reasons which motivate their actions” (1928, p. 74). They’re passive, impressionable, and in need of guidance from a superior; they have to be told what they want, or at least what they should want. These (often implicit) assumptions about the audience of propaganda bring to light another confict—one between propagandistic technique and the value of individual autonomy central to 141

Anne Quaranto and Jason Stanley

liberal ideals. Propaganda presenting false or misleading claims most clearly violates the audience’s autonomy, in much the way lying does. (See, e.g., Marlin, 2013, pp. 178–9; and Stanley, 2015, pp. 57–8.) That is, because it presents false claims as true or misrepresents true claims in misleading ways, propaganda fails to treat its audience like free, rational agents. But this argument can also be extended to show that all propaganda is wrong, for the same reasons. As Stanley describes this position, “Insofar as a form of propaganda is a kind of manipulation of rational beings toward an end without engaging their rational will, it is a kind of deception” (2015, p. 58). Thus, even if a piece of propaganda conveys something that is strictly speaking true, it may still violate hearers’ autonomy, because it manipulates them to take on beliefs, desires, values, emotions, or other mental states that they would not have acquired through non-propagandistic persuasion. However, even this broad, principled condemnation of propaganda, seems to have some exceptions. Marlin points out, for example, that in time of war, rational agents might freely choose to be propagandized to, if “the individual rationally accepts both the need for preserving the state and, therefore, the obligation to come to its defence when under attack” (2013, p. 179). In such cases, we might be willing, or even actively desire, to be presented with believable atrocity stories in order to remove qualms and stimulate more enthusiasm among the troops, thus increasing the odds of winning (Ibid., 178–9). Stanley makes an analogous argument about dire warnings on cigarette packs: “presumably the idea is that we have tacitly granted our permission to the ministry of health to take such steps.” (2015, p. 59) concluding that such propaganda is only problematic if it’s made without our giving at least tacit consent. Most importantly, however, the use of propaganda to repair fawed democracies may be democratically legitimate, if its use allows its audience to realize their agency. Flawed ideologies—stereotypes that distort the humanity of others, for example—also threaten autonomy. People in the grip of fawed ideologies are not fully free—no one thinks that the citizens of North Korea, for example, act freely, as they are under the grip of systematic lies. Similarly, a racist, ableist, or sexist ideology will lead someone to think they are acting in accord with reason, when they are in fact being duped by fawed stereotypes. Civic rhetoric, propaganda in the service of dismantling such fawed ideologies, is thus in the service of restoring democratic agency. This is a democratically acceptable goal. Whether the method to achieve it is democratically acceptable is, as we have seen, a matter of continued dispute.

Acknowledgments We are especially grateful to David Beaver for his insightful feedback and extensive input on multiple drafts throughout the project. Thank you also to Justin Khoo for editorial comments that improved the fnal version, as well as Manuel Garcia-Carpintero, Teresa Marques, and Jennifer Saul for discussion.

Notes 1 Douglass takes care to note that these stereotypes did not prevent white Americans from recognizing that Black persons had human agency – since they had laws that punished them for misbehavior. 2 See also Marques (2020) on “meaning perversions.” 3 Terence Moran (1979) theorizes propaganda as “pseudocommunication” and identifes ten distinctions between communication and pseudocommunication, including features related to the control of discourse, the stated and observed goals, the kinds of symbols used, and the kinds of justifcation ofered and reasoning encouraged. Stanley Cunningham (2002) has similarly argued that propaganda is at best



4 5 6

7 8

9 10 11 12 13 14 15 16 17

pseudocommunication. Like Moran’s, Cunningham’s argument relies on identifying communication in terms of certain norms and values which propaganda violates (pp. 176–8). For recent literature on “fake news,” see Gelfert (2018); Pepp, Michaelson, and Sterken (2019); and Rini (2017). For the argument that this literature is really just the study of propaganda, see Habgood-Coote (2019). See also Stanley (2015, pp. 197–216), explaining the epistemic harms resulting from holding fawed ideological beliefs, which undermining propaganda exploits and promulgates. There is a debate about how essential such practical authority is to the efectiveness of propaganda. Langton (1993) argues that subordinating speech, pornography in particular, requires some sort of authority or social status in order to be efcacious. Butler (1997, p. 86f.) contests the centrality of authority required by Langton. Maitra (2012) argues that speech can be subordinating even if the speaker lacks authority. See Pomerantsev (2014, 2019) for the best accessible accounts of RT. On not-at-issue content generally, see Simons, Tonhauser, Beaver, and Roberts (2010), who employ the Question Under Discussion theory of discourse structure from Roberts (1996) and argue that what’s common to the varieties of conventional implicature identifed in Potts (2005) is this pragmatic property of not-at-issueness. For a more thorough discussion of propagandistic uses of not-at-issue content, see Stanley (2015), pp. 130–69. This notion of common ground is due to Stalnaker (1978). For a broader discussion of framing and related techniques used in mass media, see Jowett & O’Donnell (2015), pp. 204–6. Indeed, one of the earliest theorists of propaganda and one of the founders of Public Relations, Ivy Lee, identifed the “failure to disclose the source of information” as “the essential evil of propaganda” (Lee 1925, p. 23). See, e.g., Doob (1950), p. 436, reporting Goebbels’ description of propaganda as “painting in black and white,” and Ellul (1965), pp. 146–7, explaining how propaganda relieves the anxiety of too much information in a complex world by ofering simple answers to difcult questions. Ellul attributes to Lenin the dictum that “in propaganda, truth pays of” (1965, p. 53), and Doob reports that Goebbels advised true claims to be used as much as possible (1950, p. 428). Slogans like these are neither true nor false not only because they do not predicate anything of a subject, but also because they seem to have a normative, rather than a descriptive, force. Paraphrasing the former as “There is liberty, equality, brotherhood” just doesn’t seem to be what the revolutionaries meant. Of course, symbols are also useful because they appeal to audiences’ emotions. We return to this point below. See also Balkin (1998, Chapter 9, “Narrative Expectation”) for a profound discussion of how narratives make messaging familiar and persuasive. See also Bartlett (1940), pp. 93–4, citing Albig (1939), p. 319, in his explanation of how statistical data helps bypass reason: Yet comparatively few people are given much genuine facility in the management of numbers, and fewer still are aforded any opportunity to understand critically the use of even the simplest kinds of statistical measurements. When a statement is “quantifed” it seems to carry, to the majority of persons, a superior certainty, and it passes without question. In his description of rational propaganda, Randal Marlin also emphasizes the discrepancy between its content seeming to appeal to rationality and its actual efects bypassing rational faculties: Rational propaganda has the appearance of genuine scientifc truth, but it is often mystifcation. Citations of facts and fgures leave the impression of great rationality, but the hearer is unable or unwilling to analyze the fgures and is persuaded by the appearance of rationality rather than by coming to grips with genuine reality. Marlin (2013), p. 29

18 See also Cunningham (2002), pp. 166–7, 176, writing, “It handles truths and information as mere instruments”; and Doob (1950), p. 428, describing propaganda’s focus on expediency, not morality, as the rationale for using truth as frequently as possible, and p. 433, recounting Goebbels’ dictum that credibility matters more than actual truth. 19 Ellul’s Propaganda was frst published in French in 1962. 20 For but a few examples, see Russell (1932), p. 211; Bartlett (1940), pp. 56–7; and Ellul (1965), pp. 33–6, 49. 21 See, for example, Russell (1932), p. 215; Bartlett (1940), pp. 60–3; and the discussion of Transfer, one of the seven propagandistic devices identifed in Institute of Propaganda Analysis, 1937, p. 6.


Anne Quaranto and Jason Stanley 22 Ellul goes on to present Nazi Minister of Propaganda Joseph Goebbels as agreeing with this claim, quoting him as saying, “We do not talk to say something, but to obtain a certain efect.” For an earlier insightful account of the techniques by which propaganda incites audiences to action, see Chapter 17 in Doob (1948). 23 Ellul (1965), p. 25: The aim of modern propaganda is no longer to modify ideas, but to provoke action. It is no longer to change adherence to a doctrine, but to make the individual cling irrationally to a process of action. It is no longer to lead to a choice, but to loosen the refexes. 24 For a discussion of some of the psychological mechanisms at work in elites’ self-justifcation, see Stanley (2015), pp. 224–31. 25 For the classic discussion of mass media’s role in producing conformity in acceptance of the legitimacy of economic and political elites, see Herman and Chomsky (1988). For classic discussions of the role of schools in propaganda, see Woodson (1933), Althusser (1971). For an in-depth discussion of the role of education, see Stanley (2015), pp. 269–91. 26 For example, Ellul maintains that “all propaganda is aimed at an enemy” (1965, p. 152). Megan Hyska (2018) also takes polarization as central to propaganda, though she considers a somewhat diferent kind of polarization than we do here. She argues that deliberative polarization, which prevents sub-groups from being able to engage in rational debate with one another, is itself a form of propaganda. 27 Lynne Tirrell (2012), p. 190, calls this the “insider/outsider function,” which she argues all deeply derogatory terms perform. 28 For a brilliant recent study of the formation and character of these types of stereotypes, see Livingstone-Smith (2012). 29 Ellul (1965), p. 152, agrees with Goebbels on this point: “man always has a certain need to hate… Propaganda ofers him an object of hatred… And the hatred it ofers him is not shameful, evil hatred that he must hide, but a legitimate hatred, which he can justly feel.” In a footnote to these words, Ellul echoes Goebbels even more explicitly: “Propaganda thus displaces and liberates feelings of aggression by ofering specifc objects of hatred to the citizen.” 30 Or, in Stanley’s words, the result erodes the in-group’s empathy and respect for other groups. Stanley (2015), p. 139. 31 The diference of the emotional efects on the in-group from those on the out-group can also be thought of in terms of propaganda’s having two distinct audiences—the group that it aims to alienate from the rest, and the group that it aims to unite against them. 32 For an explanation of how and why marginalized groups accept the fawed ideologies of dominant groups, see Stanley (2015), pp. 232–51. 33 For the civic rhetoric tradition, see e.g. Stanley (1983), Garsten (2006), Rogers (2012). 34 Though see Ellul, 1965, p. 33, arguing that “it is almost impossible to break down racial prejudice by propaganda.” 35 Within limits, of course—“notably the point where bigotry and intolerance are encouraged.” Ibid. 36 See, e.g., Bernays (1928), p. 39: “The instruments by which public opinion is organized and focused may be misused. But such organization and focusing are necessary to orderly life.” 37 Bernays (1928), p. 48. See also Jowett and O’Donnell (2015), p. 397, echoing Bernays in one of the conclusions of their rich theory of propaganda as communication: “Propaganda is not necessarily an evil thing. It can only be evaluated within its own context according to the players, the played upon, and its purpose.” 38 However, he later argues that propaganda is inevitable in post-industrial, democratic societies; see below.

References Albig, W. (1939). Public Opinion. McGraw-Hill. Altheide, D. L., & Johnson, J. M. (1980). Bureaucratic Propaganda. Allyn and Bacon. Althusser, L. (1971). Ideology and ideological state apparatuses. In Lenin and Philosophy and Other Essays (pp. 121–76). New York University Press. Balkin, J. (1998). Cultural Software: A Theory of Ideology. Yale University Press. Bartlett, F. C. (1940). Political propaganda. Cambridge University Press. Beaver, D., & Stanley, J. (2019). Toward a non-ideal philosophy of language. Graduate Faculty Philosophy Journal, 39(2), 503–47. Beaver, D., & Stanley, J. (forthcoming). The Politics of Language: An Essay in Non-Ideal Theory of Meaning. Princeton University Press.


Propaganda Bernays, E. L. (1928). Propaganda. Ig Publishing. Brooks, D. (2006). Bodies in Dissent. Duke University Press. Butler, J. (1997). Excitable Speech: A Politics of the Performative. Routledge. Camp, E. (2013). Slurring perspectives. Analytic Philosophy, 54(3), 330–49. Combs, J. E., & Nimmo, D. D. (1993). The New Propaganda: The Dictatorship of Palaver in Contemporary Politics. Longman. Cunningham, S. B. (2002). The Idea of Propaganda: A Reconstruction. Praeger. Doob, L. (1948). Public Opinion and Propaganda. Henry Holt and Co. Doob, L. (1950). Goebbels’ principles of propaganda. The Public Opinion Quarterly, 14(3), 419–42. Ellul, J. (1965). Propaganda: The Formation of Men’s attitudes. Random House. Engelbrecht, C. (2018, June 3). Fewer immigrants are reporting domestic abuse. Police blame fear of deportation. The New York Times. Garsten, B. (2006). Saving Persuasion: A Defense of Rhetoric and Judgment. Harvard University Press. Gelfert, A. (2018). Fake news: A defnition. Informal Logic, 38(1), 84–117. Gould, S. J. (1981). The Mismeasure of Man. W.W. Norton & Co. Habgood-Coote, J. (2019). Stop talking about fake news! Inquiry, 62(9), 1033–65. Harris, L. (2004). The great debate: W.E.B. Du Bois vs Alain Locke on the aesthetic. Philosophia Africana, 7(1), 15–39. Herman, E. S., & Chomsky, N. (1988). Manufacturing Consent: The Political Economy of the Mass Media. Pantheon. Hitler, A. (1925). Mein Kampf. (Manheim, R., Trans.) Houghton Mifin. Hyska, M. (2018). This Machine Kills Fascists: Detecting Propaganda with Formal Models of Mass Discourse Structure [Unpublished doctoral dissertation]. University of Texas at Austin. Institute of Propaganda Analysis. (1937). How to detect propaganda. Propaganda Analysis, 1(2), 5–6. Khoo, J. (2017). Code words in political discourse. Philosophical Topics, 45(2), 33–64. Jowett, G., & O’Donnell, V. (2015). Propaganda and Persuasion (6th ed.). SAGE Publications. Lakof, G. (2001). Metaphors of terror. The Days After. University of Chicago Press. Langton, R. (1993). Speech acts and unspeakable acts. Philosophy and Public Afairs, 22(4), 293–330. Langton, R., & West, C. (1999). Scorekeeping in a pornographic language game. Australasian Journal of Philosophy, 77(3), 303–19. Langton, R. (2018). The authority of hate speech. In J. Gardener, L. Green & B. Leiter (Eds.), Oxford Studies in Philosophy of Law (Vol. 3) (pp. 123–52). Oxford University Press. Lippmann, W. (2012). Public Opinion. Renaissance Classics (originally published 1922). Lee, I. (1925). Publicity: Some of the Things It Is and Is Not. Industries Publishing. Maitra, I. (2012). Subordinating speech. In I. Maitra & M. K. McGowan (Eds.), Speech and Harm: Controversies Over Free Speech (pp. 94–120). Oxford University Press. Marlin, R. (2013). Propaganda and the Ethics of Persuasion (2nd ed.). Broadview. Marques, T. (2020). Amelioration vs. perversion. In T. Marques & A. M. Wikforss (Eds.), Shifting Concepts: The Philosophy and Psychology of Conceptual Variability (pp. 260–84). Oxford University Press. Moran, T. P. (1979). Propaganda as pseudo communication. ETC: A Review of General Semantics, 36(2), 181–97. Muhammad, K. G. (2010). The Condemnation of Blackness: Race, Srime, and the Making of Modern Urban America. Harvard University Press. Pepp, J., Michaelson, E., & Sterken, R. (2019). What’s new about fake news? Journal of Ethics and Social Philosophy, 16(2), 67–94. Pomerantsev, Peter (2014) Nothing Is True and Everything Is Possible: The Surreal Heart of the New Russia. Public Afairs. Pomerantsev, Peter (2019) This Is Not Propaganda: Adventures in the War against Reality. Public Afairs. Potts, C. (2005). The Logic of Conventional Implicatures. Oxford University Press. Rini, R. (2017). Fake news and partisan epistemology. Kennedy Institute of Ethics Journal, 27(2) Supplement, E-43–E-64. Roberts, C. (1996). Information structure in discourse: Toward a unifed theory of formal pragmatics. The Ohio State University Working Papers in Linguistics, 49, 91–136. Rogers, M. L. (2012). The people, rhetoric, and afect: On the political force of Du Bois’s The Souls of Black Folk. American Political Science Review, 106(1), 188–203.


Anne Quaranto and Jason Stanley Ross, S. T. (2002). Understanding propaganda: The epistemic merit model and its application to art. Journal of Aesthetic Education, 36(1), 16–30. Russell, B. (1932). Education and the Modern World. Norton. Saul, J. M. (2017). Racial fgleaves, the shifting boundaries of the permissible, and the rise of Donald Trump. Philosophical Topics, 45(2), 97–116. Saul, J. (2018). Dogwhistles, political manipulation, and philosophy of language. In D. Fogal, D. W. Harris, & M. Moss (Eds.), New Work on Speech Acts (pp. 360–83). Oxford University Press. Sbisà, M. (1999). Ideology and the persuasive use of presupposition. In Language and Ideology. Selected papers from the 6th International Pragmatics Conference (Vol. 1, pp. 492–509). International Pragmatics Association Antwerp. Schmitt, C. (1996). The Concept of the Political. University of Chicago Press. Simons, M., Tonhauser, J., Beaver, D., & Roberts, C. (2010). What projects and why. Proceedings of Semantics and Linguistic Theory (SALT), 20, pp. 309–27). Smith, David Livingstone. (2012). Less than Human: Why We Demean, Enslave and Exterminate Others. St. Martin’s Press. Smith, Ted J. III. (1989). Propaganda and the techniques of deception. In Ted J. Smith III (Ed.), Propaganda: A Pluralistic Perspective (pp. 65–98). Praeger. Sproule, J. M. (1994). Channels of Propaganda. Indiana University Press. Stalnaker, R. (1978). Assertion. Syntax and Semantics, 9, 315–32. Stanley, J. (2015). How Propaganda Works. Princeton University Press. Stanley, J. (2016). The emergency manager: Strategic racism, technocracy, and the poisoning of fint’s children. The Good Society, 25(1), 1–45. Stanley, J. (2018). How Fascism Works: The Politics of Us and Them. Penguin Random House. Stanley, M. (1983). The mystery of the commons: On the indispensability of civic rhetoric. Social Research, 50(4), 851–83. Swanson, E. (2017). Omissive implicature. Philosophical Topics, 45(2), 117–38. Tirrell, L. (2012). Genocidal language games. In I. Maitra & M. K. McGowan (Eds.), Speech and Harm: Controversies Over Free Speech (pp. 174–221). Oxford University Press. Woodson, C. G. (1933). The Miseducation of the Negro. The Associated Publishers.


9 CODE WORDS Justin Khoo

1 An Example to Fix Ideas At a rally in Houston, TX in October 2018, Donald Trump said the following: America is winning again. America is respected again because we are putting America frst. We’re putting America frst. It hasn’t happened in a lot of decades. We’re putting them frst. We’re taking care of ourselves for a change folks… But radical Democrats want to turn back the clock… for the rule of corrupt, power-hungry, globalists. You know what a globalist is? You know what a globalist is? A globalist is a person that wants the globe to do well, frankly, not caring about our country so much. And you know what, we can’t have that. You know they have a word. It sort of became old-fashioned. It’s called a nationalist, and I say really, we’re not supposed to use that word. You know what I am? I’m a nationalist, okay? I’m a nationalist. Nationalist. Nothing wrong. Use that word. Use that word. What was he talking about? He seemed to be contrasting two visions of America and its relationship to the rest of the world. According to the “globalist” vision, the interests of the US are seen as continuous with those of the rest of the world, and the US should be actively engaged in foreign afairs and open to opportunities to integrate its economy into the global economy (including expanding opportunities for immigration). According to the “nationalist” vision, there are irreconcilable diferences with respect to culture, race, religion, and ideology drawn along national borders, and the US should close its borders, retreat from the global economy, and minimize involvement in foreign afairs. That’s at least something close to a dictionary defnition of “globalist” and “nationalist.” But something more is going on. Over the course of the last century, these words have taken on racial and religious connotations, with “globalist” becoming associated with Jewish liberal conspiracy theories (many involving Hungarian-American banker George Soros), and “nationalist” associated with white Christian conservatives, in particular, the ideology of white nationalism: a loose coalition of commitments encompassing belief in white racial superiority, a fear of waning cultural and demographic dominance of white people in America, and the goal of dividing nations by race and establishing a white America.1 When Trump says “I’m a nationalist,” he seems to say only that he believes the US government should be less involved in foreign afairs. But despite this, he manages to convey more:


Justin Khoo

perhaps that he believes white people are (and white culture is) superior and that he will fght to keep whites in positions of authority and power. And when Trump labels his opponents “globalists,” he says merely that they believe the US government should take an active role in foreign afairs, yet somehow conveys that they favor policies that will promote the interests of non-white, non-Christian, non-Americans at the expense of harming the interests of white Christian Americans. On the face of it, this linguistic behavior is surprising. Trump opted to convey a crucial piece of information to his audience tacitly and implicitly. But tacit and implicit information is easily missed. If Trump’s communicative aim was to convey this information as clearly as possible, he should have come out and said it explicitly. There must be some reason he did not. And Trump is not alone here. Using coded speech to convey information implicitly is a widespread practice in political speech. Here are three other prominent examples: “Arizona bears the brunt of this country’s illegal immigration problem. Its citizens themselves feel themselves under siege by large numbers of illegal immigrants who invade their property, strain their social services, and even place their lives in jeopardy.” – Justice Antonin Scalia.2 •

“Illegal immigrant” => undocumented immigrants from Latin America.

“there’s power, wonder-working power, in the goodness and idealism and faith of the America n people.” – President George W. Bush.3 •

“Wonder working power” => I am Christian.

“We have got this tailspin of culture, in our inner cities in particular, of men not working and just generations of men not even thinking about working or learning the value and the culture of work, and so there is a real culture problem here that has to be dealt with.” – Congressman Paul Ryan. ( •

“Inner cities” => Poor African American neighborhoods.

An immediate and natural question about coded speech is why it exists at all. Following Pinker et al. (2007), we might think of the situation game theoretically: insofar as communication is a rational endeavor, there must be some beneft for leaving part of your message implicit, for your audience to work out, rather than committing yourself to it explicitly. In Section 1.1, drawing on work by Tali Mendelberg, I’ll argue that coded speech arises as a way to circumvent certain social norms by conveying disavowable information. A second question concerns the mechanism by which coded speech works: in what sense are coded messages conveyed to their audience? In Section 1.2, I argue for a simple theory of coded speech, which I call the tacit inference theory. According to this theory, the implicit content of a coded utterance is inferred by the hearer from what the speaker says plus some background belief. I argue that this simple theory can, somewhat surprisingly, account for a wide range of examples of coded speech. Finally, we have the practical question of what we should do about coded speech. If our discussion is on the right track, coded speech is the result of social norms that sanction expressions of, and actions based upon, certain beliefs or attitudes. Since coded speech provides an easy way to circumvent such norms, we might wonder what value these norms have in the frst place. In Section 1.3, I discuss this issue and argue that such social norms have value independently of whether they can easily be circumvented.


Code Words

2 Plausible Deniability Focus on a simplifed version of Paul Ryan’s statement quoted above: (1) There is a culture of laziness in our inner cities. In saying this, I will suppose that Paul Ryan was trying to convey that African Americans are lazy. Call this the covert message of Ryan’s utterance. It seems correct that many individuals hearing this will take this message to be conveyed by Ryan here. That is, they will think that Ryan meant for them to believe it and believe that he believes it. However, since the message is covert, there is room for Ryan to plausibly deny commitment to it. That is, Ryan may say (2) without contradicting himself or taking anything back: (2) There is a culture of laziness in our inner cities; it has nothing to do with race, and everything to do with poverty. Notice that such denials of covert messages are coherent, sensible, and felicitous, unlike denials of entailed or presupposed content: (3) #Smith owns a red Honda, but he doesn’t own anything red. (4) #John quit smoking yesterday, and he never smoked a day in his life. This suggests that whatever way in which coded speech conveys covert messages, it is very diferent from how entailed or presupposed content of the uttered sentence is conveyed by the speaker. We’ll come back to this point in the next section. Here, I want to focus on the following question: why speak in code? Given that one central aim of communication is to be understood, it is prima facie surprising that speakers would choose to leave some aspect of their communicative aims implicit – doing so runs the risk of being misinterpreted, for instance. However, as we’ve just seen, unlike explicit claims, leaving a message implicit allows you to plausibly deny commitment to it. So, it seems, there must be some beneft of plausible deniability. Pinker et al (2007) ofer a theory on which one beneft of being able to maintain plausible deniability arises when one doesn’t know how one’s audience will take their message. They give an example of a person who has been pulled over by a police ofcer. The person desires most of all to not be incarcerated, second most to be ticket free, and third most to keep her pocket money. If the ofcer is corrupt, an explicit bribe (“Don’t write me the ticket and I’ll give you $50”) will ensure going ticket free without incarceration, at the cost of pocket money. But if the ofcer is honest, an explicit bribe will result in incarceration – the worst outcome. An implicit bribe (“Maybe we could just take care of the matter here”) allows for the possibility of going ticket free if the ofcer is corrupt without the risk of incarceration if the ofcer is honest. The reason the implicit bribe allows for this possibility is that it is sometimes not easy to tell when a statement is a bribe, and just as it would be bad to not arrest a briber, it is bad to arrest a non-briber for bribery. So, even if the implicit bribe is a bribe, and hence the person ought to be arrested for bribery, since it’s not clearly a bribe (due to his being able to plausibly deny it), the ofcer might error on the side of avoiding false positives and not arrest the briber. So much for implicit bribes, what about coded political speech? We might think the situation is analogous in the following respect: (i) there is a norm N that prohibits certain behavior that certain agents want to engage in, (ii) it is bad to sanction an agent with N who hasn’t violated


Justin Khoo

N, and (iii) coded speech allows those agents to violate N while plausibly denying doing so, thus making it unclear whether their behavior is genuinely N-violating. Given this framework, we expect to fnd social norms N that coded speech aims to subvert in exactly this way. The work of Tali Mendelberg provides a framework that accounts for implicit racial appeals such as the use of code words like inner city in exactly this way (see in particular Mendelberg 2001, 2008). Mendelberg proposes that individuals recognize the existence of social norms that prohibit explicitly racist behavior.4 This means that individuals that behave in racist ways, or express racist views, will be subject to social sanctions – shaming, potentially losing their job, and so on.5 Yet, nonetheless, some individuals may continue to want to behave in these ways – maybe by voting against policies that beneft minorities that they think are not deserving of state support, or by expressing racist ideologies – without being subject to the sanctions imposed by violating these social norms. Coded speech generates plausible deniability that makes it unclear whether the norm-violating behavior really is norm-violating. Let’s look at an example to see the theory in action. In a controlled study, Ismail White presented participants with a story in which a politician advocated for a food stamp program, varying the language used to describe the program (White 2007). The politician either explained that the food stamp program beneftted African American, inner-city, poor, or American families: “… programs like food stamps and Medicaid represent important safety nets for many African American / inner city / poor / American families, keeping them from falling deeper into poverty.” Participants were independently asked a question to measure their degree of racial resentment toward African Americans.6 White found that when the food stamp program was described as beneftting inner-city families, greater racial resentment toward African Americans was a signifcantly stronger predictor of opposition to that program. In particular, this means that the most racially resentful participants were more strongly opposed to the program when it was described as beneftting inner-city families than when it was described as beneftting African American families, or poor families. This surprising result can be explained within Mendelberg’s theory as follows. It’s plausible that the politician’s speech in the inner city condition conveyed (at least) two distinct things: i ii

Non-Racial: that food stamp programs help families living in poor, densely populated, city centers. Racial: that food stamp programs help poor African American families.

The two upshots are related by the common belief that poor urban areas are mostly populated by poor African Americans.7 By contrast, in the African American condition, plausibly, all that is conveyed is that food stamp programs help African American families. Next, we suppose that racially resentful participants are opposed to food stamp programs that help African American families because they believe such families could do just as well as white families if only they would try harder (to paraphrase one of the racial resentment prompts). However, given awareness of a norm prohibiting explicitly racist behavior, we expect such individuals to be cautious not to behave in explicitly racist ways. Given that opposing a policy describing solely as helping African American families would be explicitly racist, we expect such participants to inhibit their true views about food stamp programs in the African American condition, and thus not signal their opposition to it. But, when the program is described as helping “inner city” families, opposing the policy is ambiguated: the opposition might be racial (picking up on upshot (ii)


Code Words

above), or rather based on economic grounds (picking up on upshot (i) above). Thus, opposition in that condition allows for political cover which does not violate the norm of racial equality – since there is room for opposing the program on non-racial grounds, such opposition is not an explicit avowal of racism. Thus, describing the food stamp program using “inner city” creates space to allow racially resentful individuals to oppose the policy while at the same time plausibly deny charges of violating the norm of racial equality. In this case, the plausible deniability is generated by the speaker for her audience. However, in other cases, the plausible deniability might be generated by the speaker for herself. Suppose again that Trump calls himself a nationalist. It is possible that there are two hearers H1 and H2, for whom the upshot of Trump’s remark is very diferent. H1 comes to believe that Trump is a nationalist in the dictionary defnition sense – thus, that he will prioritize domestic issues over foreign afairs. H2 comes to believe that Trump believes in white supremacy and will fght to keep whites in positions of authority and power. Since he hasn’t explicitly committed himself to white nationalism (by calling himself a nationalist), Trump can act as if all he intended to communicate is that he is a nationalist in the dictionary defnition sense, even if he in fact intended to communicate that he was a white nationalist. If accused of admitting to being a white supremacist, Trump could easily disavow that (possible) upshot of his utterance. Similarly, H2 knows that some of Trump’s audience may be like H1. Thus, H2 can act as if all Trump’s claim conveyed to him just what it did to H1, even if this isn’t so. Thus, H2 can say that he supports Trump because he opposes globalization, rather than, because of his (Trump’s) commitment to white supremacy. As such, Trump’s claim gives H2 cover – in the sense of an alternative explanation of his behavior – to support Trump on racist grounds.8 And, as before, there are benefts to being able to communicate to H2 that you’re a white supremacist while also being able to plausibly deny doing so. By communicating this to H2, Trump may gain H2’s vote. But by being able to deny doing so, he avoids being sanctionable by social norms against racism. We’ve thus applied Mendelberg’s theory to two examples, showing how it predicts the rationale behind coded speech: the value of coded speech lies in how it can be used to create opportunities for norm-violating behavior that avoids sanction by that norm. We saw that these opportunities for norm-violating behavior might arise for the speaker using the code word, or their audience.

3 Mechanism: Te Tacit Inference Teory We saw above that part of the Mendelbergian theory of coded speech involves positing that coded speech conveys information that the speaker may plausibly deny commitment to. We saw this with “nationalist” but it also holds for the racial connotations of “inner city”: notice that a speaker may say things like,9 (5) Food stamp programs help many inner-city families, none of whom are African American. (6) We a culture of laziness in our inner cities, but it has nothing to do with race. The question before us, then, is in what sense does coded speech convey such disavowable information? We saw above that the disavowable information conveyed by a coded utterance cannot be part of the semantic content of the code word it involves: semantic content is not cancelable in this way, as evidenced by the fact that one cannot coherently deny it. (7) #Smith drives a red Honda, but he doesn’t drive a red vehicle.


Justin Khoo

We might think, then, that coded speech conveys its disavowable information via an implicature, since implicatures are cancelable. That is, although uttering: (8) Sue ate some of the cake. Implicates that Sue didn’t eat all of the cake, this implicature is cancelable – I can say without contradicting myself that: (9) Sue ate some, in fact all, of the cake. So, perhaps coded speech generates plausible deniability because the coded part is communicated via implicature? That might be so in some cases, but I’ll argue that in other cases, no implicatures are involved. Instead, I will argue for a very simple theory of coded speech. According to the tacit inference theory, code words need not mean anything special to be a source of disavowable upshots, and nor do they require pragmatic mechanisms like implicature to work. Instead, all the work generating such upshots happens on the hearer’s side, coming about by way of the hearer’s beliefs about what the code word is about or who uses it. A further beneft of such a theory is that it generalizes to account for non-linguistic sources of disavowable information as well, something semantic theories of code words struggle to do.10 Let U be some coded utterance. Distinguish the assertoric content conveyed by that utterance, call this X, from the coded information it conveys, Y, and assume that X does not entail Y. Remember, assertoric content is not disavowable, but coded information is. My proposal is that hearers get from X to Y by way of a background belief in a conditional: X -> Y. Thus, hearers will come to believe Y on the basis of the speaker’s utterance of U, even though the speaker does not commit herself to Y and thus may disavow any commitment to Y. This tacit inference theory is compatible with additional semantic and pragmatic properties of coded speech, if in fact any are needed – that is, it could be that in some cases the coded content is conveyed via implicature, and in other cases via presupposition. (However, as we’ll see, I think for at least the cases discussed in this chapter, we don’t need any additional mechanisms to account for the data.) Let’s look at how this simple theory applies to the examples above. Start with “inner city” and the data from White (2007). According to the simple theory, “inner city” does not mean anything about race. Rather, perhaps, it just means, “densely populated impoverished city center.” However, the simple theory posits that many individuals believe that inner cities are mostly populated by African Americans (and other minorities). Thus, when the politician says, “food stamp programs help many inner-city families,” he says only that food stamp programs help many families living in densely populated, impoverished, urban cores. Yet, to hearers with the relevant belief about the racial demographics of inner cities, an upshot of his utterance will be that food stamp programs help many poor African American families. As a result, these hearers may act on this upshot, thus coming to oppose the food stamp program on that basis, while gaining cover for doing so – that is, they may at the same time assert that their reasons for opposing the program are economic (perhaps they think more federal funds should address poverty in rural areas). Thus, while their opposition to food stamp programs has as a foreseeable consequence that African American families will sufer, they may coherently deny that this is the reason for their opposition. Instead, they may hold that this is a regrettable consequence of their otherwise principled opposition to such programs. By contrast, if the program is described as primarily helping African American families, then it will be harder to deny the racist upshot of their action. Notice, in particular, that the coded information need not be implicated or even communicated in any way, for this efect to come about. The hearer learns that the food stamp program will help poor African American families from the speaker’s utterance whether or not she thinks 152

Code Words

the speaker intended to communicate this information (indeed, whether or not the speaker even believes that inner cities are populated mostly by African Americans). This is an important point: the speaker may in some cases communicate the coded information, but this is not required to generate the coded efects.11 Next, turn to “nationalist.” As we saw, calling someone a “nationalist” may convey (as disavowable information) that they believe in white supremacy and will fght to keep whites in positions of authority and power. How do we account for this within the simple theory? Again, the view is that all there is to the meaning of “nationalist” is, perhaps, “person who prioritizes domestic US interests over global interests.” So, how do we explain the disavowable information it conveys? Here, I propose that “nationalist” works diferently from “inner city.” The latter generates disavowable information via a bridging inference that connects the demographic/economic meaning of “inner city” to something about race. Regarding “nationalist,” I think it’s less plausible that there is any background belief connecting nationalist beliefs with white supremacy. Instead, my hypothesis is that this case involves a meta-linguistic bridge principle. I propose that “nationalist” is a word that is believed to be used members of a certain community (white supremacists) to covertly identify each other, like a password. Its coded efect comes about by the meta-linguistic knowledge that a white supremacist would call someone a “nationalist” only if they believed that person was a white supremacist.12 This bridge is meta-linguistic because it comes from a belief about the pattern of usage of the expression “nationalist,” not from its meaning.13 When Trump says, “I’m a nationalist,” to those who recognize that this expression is used primarily by members of the white supremacist community to identify one another, his utterance will convey that he is a white supremacist. But, since the word only means “person who prioritizes domestic US interests over the interests of other countries,” Trump can disavow this upshot of his utterance.14 It’s worth keeping in mind how the explanations of the disavowable upshots of “inner city” and “nationalist” are alike and diferent. They are alike in that neither appeals to anything about the meanings of these code words, instead appealing to bridging beliefs that connect them to their disavowable upshots. However, they difer in how those bridging beliefs work. In “inner city,” the bridging belief connects the subject matter of the term (poor urban areas) to race (African Americans). In “nationalist,” the bridging belief connects uses of the expression with the speaker’s beliefs (that they are a white supremacist) to the target’s beliefs (that they are also a white supremacist). In the latter case, it is a belief about the expression itself (that it’s used by white supremacists to identify each other) that explains why it conveys that disavowable information. Notice, fnally, that in both cases, for individuals lacking the relevant bridging beliefs, neither will be conveyed by such utterances. Thus, we predict some degree of individual variation among what’s conveyed by uses of these expressions, as I think is correct. Similarly, I want to emphasize that, just as with “inner city,” the disavowable information conveyed by the coded utterance may, but need not, be implicated. According to my theory, a white supremacist may say, “I am a nationalist” to knowingly and intentionally communicate to his audience that he is a white supremacist. In such a case, this information would be communicated as a conversational implicature (most likely a Manner implicature).15 However, I want to emphasize that the relevant efect of the utterance is the same whether or not it communicates the relevant disavowable upshot. Suppose that X says, “I am a nationalist” with no intention that her audience come to believe she is a white supremacist on the basis of her utterance – perhaps she thinks they don’t believe that “nationalist” is used by white supremacists to identify each other.16 Yet, suppose that Y, X’s audience, does believe this about “nationalist” and takes X’s utterance as evidence that she is a white supremacist. In that case, this will be conveyed by X’s utterance to Y, even though it is not communicated by it. And, in this case, X’s utterance still provides cover for Y to support X on the grounds that she is a white supremacist while denying doing just that, so the phenomenon we are out to explain arises even when the relevant upshot is not communicated.17,18 153

Justin Khoo

We can contrast the tacit inference theory with a competitor theory, according to which code words like “inner city” have multiple dimensions of semantic meaning. This view is endorsed by Stanley (2015). Stanley proposes that “inner city” has two dimensions of conventional meaning: a non-racial at-issue meaning and a racial not-at-issue meaning. A plausible version of Stanley’s view is that the non-racial meaning is “poor, densely populated, city center,” and the racial meaning is “African American”: •

“Inner city”: • At-issue: poor, densely populated, city center. • Not-at-issue: African American.

Not-at-issue content is distinguished from at-issue content in a number of ways (see Beaver et al. 2009). Both kinds of contents are communicated by utterances of sentences that encode them, but they do so in diferent ways. Not-at-issue content is backgrounded. This means that a speaker (generally) expects the not-at-issue content of her utterance to already be accepted by her interlocutors (or otherwise be accepted by her interlocutors without discussion). By contrast, at-issue content is foregrounded – the speaker (generally) proposes that the at-issue content of her utterance be accepted by her interlocutors, in the sense that she puts it out there for uptake (to be accepted or rejected). This distinction manifests itself in the availability of direct denial for at-issue content, whereas not-at-issue content is generally not available for direct denial. An example of this latter contrast comes from presuppositions: (10) A: John stopped smoking last week. B: That’s false. / No, he didn’t (stop / #ever smoke). A’s sentence presupposes that John used to smoke, and thus this content of his utterance is notat-issue. By contrast, A’s sentence entails that John no longer smokes, and thus this content of his utterance is at-issue. As a result, B’s rejections target only the at-issue content of A’s utterance (that John no longer smokes). The not-at-issue content is not available for direct denial and is expected to be accepted without controversy.19 Stanley’s view is that code words communicate racial information covertly as not-at-issue content. So, when the politician above says, (11) Food stamp programs help many inner-city families. The at-issue content he communicates is that food stamp programs help many families living in poor, high density, city centers, whereas the not-at-issue content he communicates is that food stamp programs help many African American families. By analogy with (10), the not-at-issue content is not available for direct denial, a prediction which is borne out: (12) A: Food stamp programs help many inner-city families. B: That’s false. / No, they don’t (help families in poor, high density, city centers / #help African American families). Stanley’s theory ofers the following explanation of how code words like “inner city” generate cover for their audience. In this case, since two propositions are communicated, an audience member may plausibly oppose such a policy on either ground – either because the policy is helping families in poor, high density, city centers, or because it is helping African American families. Since opposing the policy on the former grounds does not violate the norm of racial equality, this 154

Code Words

provides cover for audience members to oppose it on the latter grounds without obviously violating the norm of racial equality. However, Stanley’s theory predicts that the coded information conveyed by a coded utterance should not be disavowable by the speaker (see Khoo 2017 for more discussion). Semantically encoded presuppositions and conventional implicatures cannot be canceled in the sense that someone uttering a presupposing or conventionally implicating sentence cannot felicitously negate that content: (10) #John stopped smoking, and he never smoked at all. The following is incoherent, like saying p and ~p. This suggests that conventionally encoded notat-issue content is strong in the sense that someone assertorically uttering a sentence that conventionally encodes not-at-issue content thereby commits themselves to that content. Thus, by tying the racial connotation of “inner city” to not-at-issue content, Stanley’s theory similarly predicts that it is strong in this sense. As such, Stanley’s theory predicts that it should be incoherent to say things like: (11) Food stamp programs help many white inner-city families. (12) Food stamp programs help many inner-city families, and here I don’t mean African American families. It is an empirical question whether these sentences are incoherent in the way that (10) is. To my ear, they are unexpected, but coherent. But my judgments are not the relevant ones. The judgments we should care about are those of the racist respondents in the White (2007) study who were inclined to mediate their opposition to the food stamp program in the African American condition, but not in the inner-city condition. As of the writing of this article, I am not aware of any empirical evidence of this kind. But, the predictions of the theories are clear and testable, so there is a way forward.20 Relatedly, notice that the notion of a disavowable upshot of code words is hard to square with Stanley’s theory. In particular, recall above we had: (13) I am a nationalist. But I don’t believe in white supremacy. If “nationalist” achieves its coded efects by communicating the not-at-issue content that the speaker is a white supremacist, then utterances like these should sound incoherent. Yet, to me, at least, they do not. In response to these objections, Beaver and Stanley (2020) argue that plausible deniability of “inner city” and other code words is generated slightly diferently. It’s not some particular content that is semantically encoded by the expression, but rather that uses of the expression “inner city” presupposes a speech practice of (perhaps) thinking of racial disparities in income as due to racial or cultural diferences in work ethic, rather than (say) systemic racism. Since one may engage in a speech practice unintentionally and unknowingly, this accounts for unintentional coded efects, as well as plausible deniability, since to engage in a speech practice does not require endorsing any particular content (see Beaver and Stanley 2019: 523–4). I am not totally sure I understand what a speech practice is, or what it is to presuppose or even engage in one, but insofar as doing so licenses certain inferences in one’s audience, and insofar as that is the explanation for the efects of coded speech, I think Beaver and Stanley agree with my theory. If that is right, however, the appeal to specifcally linguistic notions of presupposition at this junction (even pragmatic presupposition) still seems to me confused, since to presuppose p seems to require at least being committed to p, 155

Justin Khoo

and one need not be committed to the content conveyed by a speech practice they engage in (it’s worth pointing out that the politician in White 2007’s example is a Democrat, arguing against a reduction in funding for food stamp programs, and thus would presumably sincerely reject the racist ideology inherent in the speech practice of using “inner city”). So, ultimately, it seems there may be room for various positions in this debate to converge, and indeed, I suspect they already have.

4 Coded Speech: What to Do? We’ve seen how Mendelberg’s theory, together with the simple theory of code words, explains how coded speech conveys disavowable information, thus generate plausible deniability for speakers and hearers, thus enabling them to violate social norms while plausibly deny doing so, and thus avoid sanction by those norms. A natural thought about social norms is that their value lies in their ability to promote positive social change by incentivizing certain positive behaviors and disincentivize certain negative behaviors. From this perspective, coded speech undermines the ability to enforce violations of social norms, thus undermining their positive social efects. What, then, should we do to resist coded speech? We might call out the disavowed information conveyed by coded speech in an efort to challenge deniability. Tali Mendelberg suggests this strategy, and Congresswoman Barbara Lee opts for it in response to Paul Ryan’s “inner city” speech: Lee: “Let’s be clear; when Mr. Ryan says, ‘inner city,’ when he says, ‘culture,’ these are simply code words for what he really means: ‘black.’”21 However, Ryan’s response reveals a potential downside to this strategy. First, it is easy for the speaker to simply disavow the information being called out, thus neutralizing the response. After all, this is exactly what coded speech sets the speaker up to do. And this is exactly what Paul Ryan does in response: Ryan: “This isn’t a race-based comment; it’s a breakdown of families, it’s rural poverty in rural areas, and talking about where poverty exists – there are no jobs and we have a breakdown of the family. This has nothing to do with race.” Second, the speaker may call out the responder, accusing them of making the issue about race/gender/ etc. There are two kinds of strategies here; I’ll focus on the race case for each. One I call the “exasperated retort” in which the speaker says that not everything is about race, and admonishes the responder for (constantly) bringing race into the discussion. The other I call the “how dare you” response, in which the speaker admonishes the responder for accusing her of racism/sexism/etc. This response wields the social norm against the responder – the strategy presupposes that violations of these norms are serious, but then concludes that this is exactly why the speaker’s original claim/behavior was not serious enough to warrant sanction by the norm, and furthermore, adds the twist that it is the responder who is not taking the norm seriously by trying to use it against non-violating behavior. These available responses show the danger of calling out the implicit information conveyed by coded speech. As such, they illustrate a crucial challenge we face when it comes to enforcing social norms. Indeed, it seems that coded speech is inevitable in a society with prevailing social norms that are not universally accepted. This is cause for pessimism – if the value of social norms lies in their promoting and sustaining positive social change, but coded speech is an unavoidable strategy for undermining the enforcement of social norms, then we may worry that social norms cannot actually bring about these positive efects. 156

Code Words

Things get worse, though, for social norms. It’s plausible that a social norm may actually be a rallying point for individuals who disagree with it and resent being subject to its sanctions. Call a subgroup within a society who resent the punishments that come with violations of some norm N the N-resentful (maybe they think the norm is too broad, or too punitive). Now, suppose there are people who are powerful enough to shrug of the sanctions that come from openly violating norm N – call these people N-iconoclasts. Suppose the N-iconoclast continually and openly violates N in order to generate support among the N-resentful. As such, the N-resentful come to see the N-iconoclast as someone who gives voice to their thoughts (thoughts which they could not voice due to threat of social sanctions), has their interests in mind, and will work to erode the norm N, which they may see as hurting them. In an excellent New York Times Stone column from 2015, Jason Stanley presciently suggested that Donald Trump may be a norm-iconoclast for many white Americans frustrated by what they perceived to be social trends favoring norms that go too far in punishing behavior they think of as acceptable.22 Trump’s coarse and unfltered language throughout his campaign showed open disdain for various norms against racism and sexism (I hope I don’t need to mention the most salient examples here.) As such, it’s plausible that some of his support among white Americans derived from his norm-iconoclasm serving as a rallying point.23 At this point, we might wonder whether there is any value at all to social norms, such as Mendelberg’s norm of racial equality, that sanction otherwise legal behavior that a majority of society has agreed is morally problematic. After all, it seems such norms can be easily violated without reproach using coded speech, and that such norms may also serve as rallying points for normresentful individuals, and thus, partially responsible for strong-man leaders like Donald Trump. Nonetheless, I want to argue that social norms may in fact promote positive social change not by afecting behavior but instead by providing reassurance of dignity. My thought is that social norms, like the norm of racial equality, are tacit expressions of a society’s willingness to defend the dignity and respect of all its members. I follow Jeremy Waldron in thinking that dignity is a person’s “social standing, the fundamentals of basic reputation that entitle them to be treated as equals in the ordinary operations of society” (Waldron 2012: 5). Expressions of racist attitudes and beliefs makes public that one believes certain races are inferior, or not worthy of the same respect and dignity as others. Hence, such expressions may erode the dignity of individuals in the targeted groups in the society. The existence of a norm of racial equality that punishes such expressions provides assurance members of non-dominant groups that the society is committed to defending their dignity. Even if this norm can be subverted (via coded speech) and twisted to create solidarity among racists (by fouting it), its existence is valuable in this sense. Furthermore, insofar as the promotion of dignity is a general social good, and one which may contribute to social fourishing, we may continue to hold that social norms like these promote positive social change, even if they do not actually curb the behavior they aim to disincentivize. Notice that my idea here draws on Waldron’s work, but is actually quite diferent from Waldron’s own deployment of the notions of dignity and assurance. Waldron’s aim is to fnd grounds for legal prohibitions on hate speech. However, the case for legal prohibitions on speech is famously fraught with complications. For instance, we might worry about issues of content discrimination and the ability to challenge a legislative body’s decisions regarding speech prohibitions. However, this same concern does not arise for social prohibitions on speech, which are compatible with a maximally liberal legal free speech policy. In fact, such social norms might themselves be justifed as legally protected behavior under the very principles justifying a liberal policy toward free speech, since these norms are simply the result of the free exercise of individuals to choose who to associate with, employ, and so on. Thus, it seems plausible that social norms promote social goods by providing assurance of dignity to non-dominant groups. This assurance remains in place even if the enforceability of the norm is undermined by coded speech; this is important: it’s the dispositions underlying the social norm are what grounds the assurance, not the enforcement of the norm. 157

Justin Khoo

5 Conclusion This chapter has taken us on a tour of coded speech. We’ve asked, and attempted to answer, the following questions: (1) Why does coded speech arise? Answer: to subvert social norms by generating plausible deniability. (2) How does coded speech achieve this efect? Answer: by strategically exploiting tacit beliefs to get one’s audience to come to infer something from one’s speech without explicitly committing oneself to it. (3) What is the value of social norms that coded speech aims to subvert? Answer: the value of such norms lies in the assurance of dignity they bring about for non-dominant groups. However, we’ve explored only a few examples of coded speech. It remains to be seen whether there’s a plausible unifed theory of coded speech, and perhaps the simple theory articulated here will require supplementation to cover other cases. Further work is also needed to unpack the content of various social norms, and such work might prove helpful in fnding additional ways of responding to coded speech, preventing it from undermining these norms. Finally, we might consider whether and to what extent we may want adjust social norms in response to coded speech. If my argument in the fnal section is correct, social norms have value independently of the behavior they curb, but we might wonder whether better articulated norms may be more resistant to coded speech.

Notes 1 See: https://www. how-trump-defnes-globalist-and-nationalist/ 2 This quote comes from a Supreme Court decision: Arizona v. US, 132 S. Ct. 2492, 2522 (2012). 3 This quote was from George W. Bush’s 2003 State of the Union Address (http://www.washingtonpost. com/wp-srv/onpolitics/transcripts/bushtext_012803.html). 4 Other potential relevant social norms include norms against sexist behavior, … 5 For instance, there are numerous instances of people who exhibit racist behavior losing their jobs. One example is radiologist Kyle Thomas, who lost his job at Mississippi Baptist Medical Center after being videotaped calling a black woman the n-word ( 6 The question was: “how well the words ‘hardworking,’ ‘intelligent,’ and ‘lazy’ describe either Blacks or Whites as a group?” This is a standard measure of unobtrusively measuring racial resentment in an individual: see Kinder and Sanders 1996. 7 Note that the racial upshot of “inner city” is disavowable. A real-world example of this comes from Paul Ryan, who said: We have got this tailspin of culture, in our inner cities in particular, of men not working and just generations of men not even thinking about working or learning the value and the culture of work, and so there is a real culture problem here that has to be dealt with. When called out that his speech was racist against African Americans by Congresswoman Barbara Lee, Ryan responded, This isn’t a race based comment; it’s a breakdown of families, it’s rural poverty in rural areas, and talking about where poverty exists – there are no jobs and we have a breakdown of the family. This has nothing to do with race. 8 This is how code words have earned the moniker “dog whistles.” Richard Morin, who is credited with coining the expression, writes, “Subtle changes in question-wording sometimes produce remarkably diferent results … researchers call this the `Dog Whistle Efect’: Respondents hear something in the question that researchers do not” (Morin 1988).


Code Words 9 A real-world example of this comes from Paul Ryan, who said: We have got this tailspin of culture, in our inner cities in particular, of men not working and just generations of men not even thinking about working or learning the value and the culture of work, and so there is a real culture problem here that has to be dealt with. When called out that his speech was racist against African Americans by Congresswoman Barbara Lee, Ryan responded, This isn’t a race based comment; it’s a breakdown of families, it’s rural poverty in rural areas, and talking about where poverty exists – there are no jobs and we have a breakdown of the family. This has nothing to do with race. 10 Here’s an example. George W. Bush was a rich New Englander (born in New Haven, CT) who went to Yale and Harvard. Yet, throughout his presidency, he cultivated the image of a blue-collar man of the people by taking photographs clearing brush and speaking in a thick Texas accent. In so doing, he aimed for others to come to believe he was a blue-collar man at heart, but not by recognizing his intention, for in that case he would instead be seen as disingenuous. Thus, Bush didn’t implicate this by his actions. Instead, they were merely conveyed by them – and, indeed, Bush was seen by many Americans to be a “guy you’d want to have a beer with” (,3777471&dq=who+would+you+rather+have+a+beer+with+bush+or+kerry&hl=en). 11 Thus, the mechanism of coded speech may cross cut that of insinuation, which involves refexive communicative intentions where the communicated content may not become common ground; see Camp (2018). 12 This is similar to Bolinger (2017) and Nunberg (2018)’s views about slurs, which they claim implicate membership in a group that hates members of the slur-targeted group. 13 The explanation here is thus similar to other code words, in particular “wonder working power,” which was used by George W. Bush in his campaign in presidency to convey his Christianity to his Christian supporters while not alienating his non-Christian supporters (who simply didn’t recognize the Christian source of the expression). See Albertson (2015) for empirical evidence of these efects of “wonder working power.” 14 Also, it’s plausible that “globalist” is used by members of certain conservative groups to identify individuals who are anti-American. I’m not sure whether the best explanation of the disavowable upshots of “globalist” is the frst or second model. 15 The reasoning might go as follows: why did the speaker use this somewhat archaic and wonky term, “nationalist” rather than say that they are fghting to bring jobs back to the US? A possible reason is that they intend to identify as a white supremacist, given that white supremacists use this term to identify themselves to each other. Thus, the audience may reason to this conclusion on the basis of the utterance. 16 Of course, if X’s audience thinks that X doesn’t know that white supremacists use “nationalist” to identify each other, then they won’t infer this from the fact that X self-ascribed it. 17 Things are diferent when what’s communicated matters. Suppose Y asked X, “Is Sue seeing anyone these days?” and X said, “Sue has been spending a lot of time in New York recently.” Suppose X misheard Y’s question, and thus didn’t intend her utterance to be relevant to the question Y actually asked. In this case, nothing is implicated, even though Y will take X’s answer to implicate something relevant to her question – perhaps that Sue is seeing someone in New York. In this case, this will still be conveyed by X’s utterance for Y. Yet, if X didn’t believe this, and what matters is whether X misled Y, then whether this upshot was communicated or not will matter a lot – since X didn’t communicate this, it is false that X misled Y, even though there was miscommunication between them. 18 This point is even clearer when it comes to “inner city” which depends on existing beliefs about the audience, and not at all on what the speaker thinks or intends her audience to think. 19 This is not to say that not-at-issue content cannot be rejected, it is just harder to do so. However, Kai von Fintel (2004) notes that some kind of conversational backtracking is needed: “Hold on-John never smoked in the frst place!” or “Hey wait a minute – I didn’t know John smoked!” Also, the other main source of conventionalized not-at-issue content is conventional implicatures (see Potts 2005 for discussion). Non-restrictive relative clauses generate conventional implicatures: i

Smith, who owns Capital Bank, lives in New York.

Here, (i) conventionally implicates that Smith owns Capital Bank. Similarly with presuppositions, direct denials target at-issue content, not conventionally implicated content: ii That’s false. / No he doesn’t (live in NY / #own Capital Bank). Conventional implicatures difer from presuppositions in their projection profle: the range of operators from which embeddings of conventionally implicating / presupposition triggering expressions continue


Justin Khoo to communicate their implicatures/presuppositions. For instance, both conventional implicatures and presuppositions project out of questions: iii a Did John stop smoking? => John used to smoke. b Does Smith, who owns Capital Bank, live in New York? => Smith owns Capital Bank However, conventional implicatures, but not presuppositions, project out of belief reports: iv a Sue thinks that John stopped smoking; but he never smoked in the frst place. b #Sue thinks that Smith, who owns Capital Bank, lives in New York; but he doesn’t own Capital Bank. 20 I do want to point out that the fact that Paul Ryan was able to say what he did (see footnote 8) without sounding like he took back his original comment is further evidence against Stanley’s view. Generalizing, the fact that politicians who say “inner-city” and later deny the racial connotations of their utterance are not accused of fip-fopping is evidence against Stanley’s view. 21 12/3394871/ryan-poverty -inner-city/ 22 23 This thought goes some way to explaining why Donald Trump won majorities among rural, older, middle-class, white voters. (See

Bibliography Albertson, Bethany L. 2015. Dog-whistle Politics: Multivocal Communication and Religious Appeals. Political Behavior, 37(1), 3–26. Beaver, David, Roberts, Craige, Simons, Mandy, & Tonhauser, Judith. 2009. Presupposition, Conventional Implicature, and Beyond: A Unifed Account of Projection. In: Klinedinst, Nathan, & Rothschild, Daniel (eds), Proceedings of New Directions in the Theory of Presuppositions. Toulouse: ESSLLI, 1–15. Beaver, David & Stanley, Jason. 2019. Toward a Non-Ideal Philosophy of Language. Graduate Faculty Philosophy Journal, 39(2), 501–45. Beaver, David & Stanley, Jason. 2020. Hustle: The Politics of Language. Princeton, NJ: Princeton University Press. Bolinger, Renée Jorgensen. 2017. The Pragmatics of Slurs. Noûs, 51(3), 439–62. Camp, Elisabeth. 2018. Insinuation, Common Ground, and the Conversational Record. Pages 40–66 of: Fogal, Daniel & Harris, Daniel (eds), New Work on Speech Acts. Oxford: Oxford University Press. Grice, H. Paul. 1957. Meaning. The Philosophical Review, 377–88. Grice, Paul. 1989. Studies in the Way of Words. Cambridge, MA: Harvard University Press. Haslanger, Sally. 2007. “But Mom, Crop-tops Are Cute!” Social Knowledge, Social Structure and Ideology Critique. Philosophical Issues, 17(1), 70–91. Khoo, Justin. 2017. “Code Words in Political Discourse.” Philosophical Topics, 45(2), 33–64. Kinder, Donald R, & Sanders, Lynn M. 1996. Divided by Color: Racial Politics and Democratic Ideals. Chicago, IL: University of Chicago Press. Mendelberg, Tali. 2001. The Race Card: Campaign Strategy, Implicit Messages, and the Norm of Equality. Princeton, NJ: Princeton University Press. Mendelberg, Tali. 2008. “Racial Priming Revived.” Perspectives on Politics, 6(1), 109–123. Morin, Richard. 1988. Behind the Numbers: Confessions of a Pollster. The Wall Street Journal, October 16. Nunberg, Geofrey. 2018. The Social Life of Slurs. In: Fogal, Daniel, Harris, Daniel, & Moss, Matt (eds), New Work on Speech Acts. Oxford: Oxford University Press. Pinker, Steven, Nowak, Martin, & Lee, James. 2007. “The Logic of Indirect Speech.” PNAS, 105(3), 833–8. Potts, Christopher. 2005. The Logic of Conventional Implicatures. Vol. Oxford Studies in Theoretical Linguistics. Oxford: Oxford University Press. Sartre, Jean-Paul. 1956 (2012). Being and Nothingness. New York, NY: Open Road Media. Saul, Jennifer. 2017. Racial Figleaves, the Shifting Boundaries of the Permissible, and the Rise of Donald Trump. Philosophical Topics, 45(2), 97–116. Stanley, Jason. 2015. How Propaganda Works. Princeton, NJ: Princeton University Press. von Fintel, Kai. 2004. Would You Believe It? The King of France Is Back! (Presuppositions and Truth Value Intuitions). Pages 315–42 of: Reimer, Marga, & Bezuidenhout, Anne (eds), Descriptions and Beyond. Oxford: Oxford University Press. Waldron, Jeremy. 2012. The Harm in Hate Speech. Cambridge, MA: Harvard University Press. White, Ismail K. 2007. When Race Matters and When It Doesn’t: Racial Group Diferences in Response to Racial Cues. American Political Science Review, 101(02), 339–54.



Please note: this chapter quotes examples of racist and sexist speech. In the United States—which has been more thoroughly studied in this regard than many countries—a norm against being racist coexists with quite widespread racial resentment among White people, at varying levels of conscious awareness. It is overwhelmingly likely that the same is true elsewhere as well. As a result, tapping into the racial resentment while not too obviously violating the anti-racism norm—for example, via dogwhistles—has proved a highly efective electoral strategy. This has also been fairly widely studied.2 Another, less widely examined, technique is the fgleaf—something which just barely covers a thing that you’re not supposed to show in public. Figleaves allow one to get away with what would otherwise be seen as racist utterances, thereby shifting the norms of what counts as racism for those that accept the fgleaf. A classic but overly simple fgleaf, efective only with a limited group, would be “I’m not racist, but R,” where R is an utterance that would otherwise be seen as racist. Although readers of this chapter will most likely not fnd this fgleaf efective, it is still very widely used, and efective with some—as well will see, fgleaves do not need to be all that widely successful in order to be useful. Figleaves, then, play a very important role by enabling much more blatant expressions of racial resentment than those that can be dogwhistled. There is good reason to believe that this makes them a potent and dangerous force in shifting norms. This article explains how fgleaves work, and explores the application of the notion to other forms of prejudice, such as sexism.3 In addition, it discusses possible extensions of the concept.

1 Racial Figleaves Racial fgleaves serve to block inferences to the claim that some particular utterance R is racist, or that the person who uttered R is racist. More precisely: A racial fgleaf is an utterance which (for some portion of the audience) blocks the conclusion that (a) some other utterance, R, is racist; or (b) the person who uttered R4 is racist.5 A racial fgleaf may be synchronic, occurring at the same time as the utterance R; or it may be diachronic, occurring at a diferent time from R. In both cases, it works in the same way. The utterance R is the sort of thing that might cause an audience to think that the speaker is racist (or the sort of thing that might itself be deemed racist). A fgleaf is uttered, which allows the audience to avoid making this inference. (In the case of a diachronic fgleaf, it may even be that the audience 161

Jennifer Saul

makes the inference and then, based on the fgleaf, reconsiders it and decides that the speaker/R is not racist after all.) The need for racial fgleaves comes from the widespread (though by no means universal) acceptance of what Tali Mendelberg calls the Norm of Racial Equality (Mendelberg 2001). This is a norm which forbids overt expression of very obvious racism and is endorsed even by people who harbor quite high levels of racial resentment (explained below). Mendelberg does not ofer a precise formulation of this norm, though she notes that it forbids open expression of support for legal discrimination and segregation, and claims of biological inferiority. It also makes people uncomfortable about voting for those who hold such views. I have argued elsewhere (Saul 2017, 2019a) that it is best understood as Don’t be racist, leaving unspecifed what it takes for an utterance to count as racist. The Norm’s lack of precision means that individual speakers will form their own views on the subject, and White speakers will generally adopt a very high bar for what counts as racism. Jane Hill (2008) itemizes some of the ways that White people’s folk theories set a high bar for what counts as racism: for example, they tend to hold that racism is a matter of fully conscious beliefs and intentions, and that to be a racist one must condemn all members of a racial group, and view all members of a racial group as biologically inferior. Anything which can be seen as falling short of this can be seen as not racist. Hill also discusses further strategies used by White people to avoid acknowledging racism. One of these, for example, is to insist that racism is a matter of what is one’s “heart” rather than one’s “head” (Hill 2008: 103). The “head” may cause one to make racist-sounding utterances, but the truth about a person lies in their “heart,” which one can insist (even in the face of very substantial contrary evidence) is non-racist. We can see this sort of strategy at work in Kellyanne Conway’s comments below, about why we should accept Trump’s afterthe-fact explanation of ofensive comments regarding a disabled reporter (showing, incidentally, that the strategy is not confned to the dodging of racism allegations): Why don’t you believe him? Why isn’t it taken at face value?” Conway said in exasperation. “You can’t give him the beneft of the doubt on this and he’s telling you what was in his heart? You want to go with what’s come out of his mouth rather than what’s in his heart. (Meyer 2017) Given this tendency, and the very strong desire not to be forced to confront the racism of one’s friends or politicians one supports, it is easy to see how fgleaves might serve to block inferences to the claim that some speaker one cares about is racist. Almost no matter what they have said, it will be possible for them to follow this with an utterance that ofers a way for at least some audience members (though rarely all) to avoid being forced to accept that the speaker is racist. No matter what their racial sentiments (conscious or unconscious), most White people do not want to think of themselves as racist, do not happily self-identify as racists, and do not want to associate with those that they see as racist.6 However, they are nonetheless likely to harbor high levels of what psychologists call “racial resentment.” This is measured by agreement with items like the following: It’s really a matter of some people not trying hard enough; if blacks would only try harder they could be just as well of as whites. (Tesler and Sears 2010: 19) The fact that many racially resentful people endorse the Norm of Racial Equality means that politicians can beneft from fnding ways to tap into racial resentment without violating the


Racist and Sexist Figleaves

Norm. As noted that the start of this article, one of the most-discussed methods is that of racial dogwhistles, but a further method seems to be what I have called a “fgleaf ”. Dogwhistles require one to express racist sentiments in subtle ways. Figleaves, however, can be added on to utterances that would otherwise be seen as racist, blocking their condemnation and thereby enabling the expression of much more clearly racist sentiments and views. To see how this works, imagine that a speaker makes an otherwise obviously racist utterance R. They then (either immediately or at some later time) utter a fgleaf, F. Some of the audience for R would have inferred that the speaker is racist but were reassured by the fgleaf that she was not. The fgleaf thereby blocks this inference. A key reason that this works is that many White people, as Hill argues, are so very willing to fnd ways  to explain away what looks like racism. This makes them extremely receptive to fgleaves as a way of avoiding the conclusion that someone is racist.7 We will discuss examples of these in Part III.

2 Gender Figleaves I will spend somewhat more time on the background to gender fgleaves, because these have not been explored in print before. Gender fgleaves will block the inferences to the claim that someone has said something sexist, or that the person in question is sexist. More precisely: A gender fgleaf is an utterance which (for some portion of the audience) blocks the conclusion that (a) some other utterance, R, is sexist; or (b) the person who uttered R is sexist. The need for gender fgleaves arises from a desire to follow the norm Don’t be sexist, despite widespread sexist attitudes. Just as not many people happily self-identify as racist, so too not many people happily self-identify as sexist. They may think of themselves as traditionalist about gender, but they are likely to deny that they are sexist, and likely to want to avoid seeing themselves as sexist. Mendelberg argues that there is widespread commitment to a Norm of Gender Equality in the United States. As evidence (Mendelberg 240), she cites surveys showing that by the 1990s, 80% of Americans agreed that “women and men should have an equal role” (up from 49% in 1972). All of this is true, but we will see that (as with the Norm of Racial Equality) it does not rule out support for a lot of things which many would indeed view as sexist. Mendelberg also acknowledges that “gender equality norms coexist uneasily with lingering negative predispositions” (240). But, she says, these difer from the sorts of negative predispositions that are present in the case of race: she suggests that they involve role expectations rather than fear and resentment.8 (As an example she cites studies showing very large majorities who think that a couple’s choices about where to move should be led by the man’s and not the woman’s job opportunities.) Since the 1990s (e.g. Glick and Fiske 1996, 1997), psychologists have distinguished between two forms of sexism, Hostile and Benevolent. These will prove important to the workings of gender fgleaves. Hostile Sexism is “antipathy toward women who are viewed as usurping men’s power” (Glick and Fiske 2001: 109). It “seeks to justify male power, traditional gender roles, and men’s exploitation of women as sexual objects through derogatory characterisations of women” (Glick and Fiske 1997: 121). It’s measured through agreement with items like Many women are actually seeking special favors, such as hiring policies that favor them over men, under the guise of asking for “equality” and disagreement with items like There are actually very few women who get a kick out of teasing men by seeming sexually available and then refusing male advances (Glick and Fiske 1997: 135). Benevolent Sexism is the kinder, gentler face of sexism. It is “a subjectively favorable, chivalrous ideology that ofers protection and afection to women who embrace conventional roles” (Glick


Jennifer Saul

and Fiske 2001: 109). It is measured through agreement with items like Many women have a quality of purity that few men possess and disagreement with items like Men are complete without women (Glick and Fiske 1997: 135). As Glick and Fiske write, both Hostile and Benevolent Sexism “presume traditional gender roles and both serve to justify and maintain patriarchal social structures” (Glick and Fiske 1997: 121). Unsurprisingly, perhaps, women are much more likely to endorse Benevolent Sexism than Hostile Sexism. The efects of women’s own Benevolent Sexism are worth noting, as “benevolent paternalism may reduce women’s resistance to patriarchy” (2001: 111). Signifcantly, women are more forgiving of discriminatory acts (like “losing a promotion to a less qualifed man” (2001: 111)) if given a protective justifcation. Benevolent Sexism can provide just such a protective justifcation. Women who endorse Benevolent Sexism are also more forgiving of overt hostility by male partners. Both women and men seem to direct the diferent kinds of sexism at diferent types of women—with Hostile Sexism directed at feminists and women in traditionally male roles, and Benevolent Sexism directed at women in traditionally female roles. Interestingly, men who are highest in Hostile Sexism (as directed at feminists or “career women”) are also highest in Benevolent Sexism (as directed at stay-at-home mothers). As with the Norm of Racial Equality, Mendelberg never gives us a precise formulation of the Norm of Gender Equality. But it is clear that she takes it to include the idea that women and men should have an equal role. It undoubtedly rules out such things as not allowing women to vote or own property, and (for most jobs, but not for example the priesthood) discriminating explicitly on the basis of sex. It also undoubtedly roles out explicit endorsements of male domination of women, or of confning women to nurturing roles. I will opt to formulate it in a manner parallel to the race norm: Don’t be sexist. Again, what counts as sexist will be left to individuals’ judgments. One bit of indirect evidence for this norm is that Hostile and Benevolent Sexism are not measured by the very most explicit statements of sexist attitudes. A person may count as high in Hostile Sexism even if they don’t assent to claims like “men should dominate women,” although an ideology of male domination is at the root of Hostile Sexism. And a person may count as high in Benevolent Sexism even if they don’t assent to claims like “women need to be protected by men,” although this is the idea at the root of benevolent Sexism. Instead, these forms of sexism are measured through assent to very carefully formulated items, clearly designed to avoid running afoul of a norm that makes it—in general—socially unacceptable to endorse absolutely explicit statements of sexism. Because the Norm of Gender Equality, Don’t be sexist, is extremely vague, there will be considerable variation in how it is interpreted. This is very much like the Norm of Racial Equality, except that perhaps even more variation is possible. The Norm of Racial Equality precludes claims of biological inferiority and support for segregation, according to Mendelberg (we can take this as meaning that most people would consider such claims to be clearly racist). It’s not at all clear that the Norm of Gender Equality does likewise: claims of naturally diferent abilities and interests between men and women are very widely accepted and utterly mainstream. Gender-segregated toilets are so normal that their absence is surprising in many contexts—and doing away with them is an immensely controversial proposal. Moreover, it is remarkable how widely (though of course not universally) acceptable expressions of Benevolent Sexism are. Take, for example, Paul Ryan’s widely (and often approvingly) reported response to the Access Hollywood tape. Trump’s comments on this tape were not just clear instances of Hostile Sexism, but actual boasting about sexual assault. Two quotations: I moved on her like a bitch, but I couldn’t get there, and she was married. Then all of a sudden I see her, she’s now got the big phony tits and everything. 164

Racist and Sexist Figleaves

I’m automatically attracted to beautiful [women]—I just start kissing them. It’s like a magnet. Just kiss. I don’t even wait. And when you’re a star they let you do it. You can do anything… Grab them by the pussy. You can do anything. (Mathis-Lilley 2016) Ryan condemned the utterances, saying that “women are to be championed and revered” ( Jacobs et al. 2016). Ryan received extremely widespread praise in the media for this utterance. As many feminist critics noted, however, Ryan’s utterance was a clear expression of the sort of Benevolent Sexism that holds women back from being taken seriously as agents, and that indeed can serve to support rape culture. A sexist fgleaf is an additional utterance that is added on to an utterance S that would otherwise be seen as sexist. A synchronic fgleaf comes at the same time as S, while a diachronic fgleaf comes at a diferent time—often later, in response to criticism. The addition of the fgleaf casts doubt—for at least some of the audience—on the idea that the utterance and/or the speaker is sexist. There is a widespread (though I think false) belief that racism and sexism are solely a matter of individual speakers’ conscious beliefs and intentions.9 For those who hold this belief, all a fgleaf needs to do is to cast some doubt on whether the speaker consciously holds racist or sexist beliefs or intentions. It will generally be the case that a fgleaf will work for some audience members and not for others, at least when the audience is as large and diverse as voters that a politician is attempting to appeal to. Despite the fact that it won’t work on all voters, a fgleaf can still be highly efective in a political campaign, since these are often won or lost through the shifting of specifc, and often quite small, groups of voters.

3 Varieties of Figleaf In this section I attempt to give some idea of the range of kinds of fgleaves, and to ofer some real-life examples. It is by no means intended to be comprehensive.

3.1 Denial Figleaf The classic over-obvious racist fgleaf is “I’m not a racist but…,” followed by R (a statement that would otherwise be seen as racist).10 This is denial fgleaf, one which seeks to block an inference to the claim that the speaker is racist simply by denying the speaker’s racism. As noted earlier, this sort of fgleaf will be highly inefective for many people, but it will work for others. Its sexist counterpart is, of course, “I’m not a sexist but…” (and its relatives) followed by S, a statement that would otherwise be seen as sexist. Donald Trump provided a great example of this as a synchronic fgleaf back in 1994: I have days where, if I come home—and I don’t want to sound too much like a chauvinist,” the President said, “but when I come home and dinner’s not ready, I go through the roof. (Cohen 2017) This fgleaf is too obvious, and indeed probably makes the hearer expect something sexist to come afterward. For most of us it will not succeed. But for someone who shares the sentiment that follows the fgleaf, it may well succeed. The man who agrees that he’d really like his dinner to be ready when he gets home, and who’s been feeling resentful of it not being ready can listen to this and tell himself that it’s not sexist after all to think this. And, perhaps, he will soon begin to feel more empowered to say it and to act on it 165

Jennifer Saul

3.2 Statement of General Care for Group This is an especially popular fgleaf for sexism, and it seems a favorite of Donald Trump. In early 2017, the Washington Post (Blake 2017) collected 21 examples of it (though did not label them as “fgleaves”). Here are just a few: 1

After the Access Hollywood tape: I have great respect for women. Nobody has more respect for women than I do. I’ve said things that, frankly, you hear these things I said. And I was embarrassed by it. But I have tremendous respect for women. And women have respect for me.

2 3 4

In response to Clinton’s criticism of his behavior, during a debate: “Nobody has more respect for women than I do. Nobody.” In response to criticism of his comments regarding primary opponent Carly Fiorina: “I have tremendous respect for women, and I am going to protect women.” After criticism of his attack on Megyn Kelly: “I cherish women. I want to help women. I’m going to do things for women that no other candidate will be able to do.”

Interestingly, this sort of utterance seems to be one that Trump uses over and over as a diachronic fgleaf, producing it on a regular basis in response to criticisms of sexist utterances. Also interestingly, it sometimes includes some quite clear Benevolent Sexism, as in (3) and (4). The idea is meant to be that someone who declares his respect or care for women in general could not possibly be a sexist. While such a statement may be efective in reassuring people who are already strongly committed to supporting Trump and looking for a way to brush of accusations of sexism, it’s unlikely to be convincing to those who are generally worried that the accusations may be true.

3.3 Afection Assertion11 The most widely known afection assertion fgleaf is the classic “some of my best friends are black.” This may occur synchronically, being immediately followed by “but R” or it may be diachronic, occurring at a diferent time from R. We get a nicely complex version of this from Representative Mark Meadows’s questioning of Michael Cohen (Donald Trump’s former lawyer, who began cooperating with the special prosecutor Robert Mueller). Cohen had accused Trump of being a racist. Meadows stated that Trump could not be racist, literally pointing to a Black former Trump employee as evidence of this, stating that she would not work for a racist. (This utterance is not a fgleaf on my preferred understanding, because it is not uttered by the utterer of a potentially racist utterance R, though see the fnal section for more expansive understandings of “fgleaf ” which might render a diferent verdict.) Meadows were accused of racism for defending Trump in this way, and he responded with the family version of the friendship assertion fgleaf: My nieces and nephews are people of color. Not many people know that. (Gregorian 2019) For many White people (though not Meadows), a friendship assertion fgleaf about Black people is either unavailable or a lie—such is the extent of racial segregation in our hierarchically organized world. Gender, however, works diferently. Most men have women or girls who they genuinely love or at least care for—or who they have been loved or cared for by. Even if they do not have daughters or wives, most have mothers—or have had mothers. And most can plausibly claim at least some degree of afection for these women (or girls). This means that there will be a fgleaf 166

Racist and Sexist Figleaves

available, containing a heartfelt description of their afection for at least one woman—which can be taken as insulating them from accusations of sexism. Wives and daughters are very often mentioned when a man has been accused of misogynistic behavior. If the tactic works, the mention of wives and daughters can serve as a fgleaf, convincing the audience either that the accusation must be false or that it doesn’t show the misogynistic attitudes it’s been taken to show. Famously, Representative Ted Yoho used several abusive swear words when speaking to Representative Alexandria Ocasio-Cortez. Ocasio-Cortez gave a viral speech about the encounter. Yoho’s response gave a central role to his wife and daughters, using the fact of his relationship to them as evidence that he must have been misheard (Michallon 2020): Having been married for 45 years with two daughters, I’m very cognizant of my language. The ofensive name-calling, words attributed to me by the press were never spoken to my colleagues and if they were construed that way, I apologise for their misunderstanding. This fgleaf only works if one accepts (at least for the moment) that those with wives and daughters could not be abusive to other women. The history of the world is flled with contrary evidence.

3.4 More Broadly: Not all X It is a part of the folk theory of White racism, according to Hill (2008), that a racist will condemn all members of a racial group. It seems likely to me that the same thing holds for sexism, broadly construed. Indeed, Kate Manne (2017) discusses this with respect to misogyny. She notes that it’s widely believed that misogynists are people who hate all women, but that it’s very unlikely anyone will meet this description—since there will be some women who devote themselves to tending to men’s needs, and even the most stereotypically misogynist man will not hate these women. (He is far more likely, in fact, to adopt a benevolent sexist attitude toward these women.) This can lead to the thought that misogyny does not really exist. Both Manne and Hill (rightly, in my view) reject the view that racism, sexism, and misogyny actually require such universal condemnation. Importantly, the belief that racism and sexism involve universal condemnation provides ample opportunity for fgleaves. These can be seen as underlying the friendship assertion fgleaf. It is this belief which (for those who hold it) allows this fgleaf to block inferences to the speaker’s racism. Those of us who know that racists have Black friends will not think that having Black friends is a good reason to think that someone can’t be racist. Taking advantage of the intersectional nature of oppression, someone who only makes ofensive comments about Black women (and praises White women and Black men) might deny both racism and sexism by uttering a fgleaf which cites the White women and Black men that they have praised. More generally, many fgleaves take the form of asserting that one doesn’t hold some negative belief about all members of a group. Because of the widely held view that racists will condemn all members of the group, any utterance indicating a positive (or even neutral) attitude toward some members of the group can serve as a fgleaf to block the inference that the speaker is prejudiced against that group. This sort of fgleaf is used in Donald Trump’s candidacy announcement, in which he focused on the Mexicans who are sent rather than all Mexicans (Washington Post Staf 2015). When Mexico sends its people, they’re not sending their best. They’re not sending you. They’re not sending you. They’re sending people that have lots of problems, and they’re bringing those problems with [them]…They’re rapists. And some, I assume, are good people. I have a more extensive discussion of this example in my 2017, but here I want to note just that it’s another instance of the broad not all X category of fgleaf. And it’s clearly an efective one. As 167

Jennifer Saul

I discuss in that paper, Trump supporters reassured each other online that Trump was not racist because he did not call all Mexicans rapists. The not all X move provides a powerful general means of generating fgleaves.

3.5 Mention Figleaf/ Paralipsis In my (2017), I discuss the use of a “mention” fgleaf, in which a person states that they are not going to say something, mentioning the thing that they say they are not going to say. My example there was Glenn Beck, interviewing Muslim congressman Keith Ellison: I feel like saying is, “Sir, prove to me that you are not working with our enemies.” And I know you’re not. I’m not accusing you of being an enemy, but that’s the way I feel, and I think a lot of Americans will feel that way. The fact that Beck says he is not asking Ellison to prove he’s not working with America’s enemies serves as a fgleaf for the racism of his utterance. This sort of fgleaf is a variety of paralipsis, a well-known and much-used rhetorical technique.12 We even see it in Chaucer (Nordquist 2019). “The music, the service at the feast, The noble gifts for the great and small, The rich adornment of Theseus’s palace … All these things I do not mention now.”

We have a nice example of a sexist mention/paralipsis fgleaf in Trump’s comments on Megyn Kelly (Anonymous 2017): I refuse to call Megyn Kelly a bimbo, because that would not be politically correct. Instead I will only call her a lightweight reporter! The fact that he says he is not calling Kelly a bimbo serves as a fgleaf for the sexism of his remark, allowing supporters to say he was not sexist because he did not call her a bimbo.

3.6 Benevolent Sexism and Figleaves As we’ve noted in passing, some popular fgleaves include statements of Benevolent Sexism. Even just asserting that one has a wife or daughter will sometimes function to invoke Benevolent Sexism— by suggesting that one has a male protective attitude toward them. Because Benevolent Sexism is so much more acceptable than Hostile Sexism, it is able to act as a fgleaf for Hostile Sexism that would otherwise be too readily recognized as such. The theory of Ambivalent Sexism explains why an assertion of Benevolent Sexism can be such a potent fgleaf. It turns out that exposure to expressions of Hostile Sexism leads women in particular to be more likely to endorse Benevolent Sexism. Peter Glick, who frst formulated the notion of Ambivalent Sexism with Susan Fiske, explains that this can be a sort of psychological self-defense mechanism—since Benevolent Sexism can be seen as a form of protection (Crockett 2016). So women who hear the Benevolent Sexist fgleaf will be especially likely to fnd it appealing. Moreover, Benevolent Sexism also ofers another fall-back in case the fgleaf fails to work. While the attitudes and behaviors of the Hostile Sexist may be viewed as unacceptable by one who is a Benevolent Sexist, they are also often excused. If women need protection from men, after all, men need to be strong, assertive, and aggressive. Benevolent Sexism helps to explain the acceptance of these traits as part of a “boys will be boys” mentality (Crockett 2016). 168

Racist and Sexist Figleaves

All this may be why we often see people who make very obviously hostile sexist remarks also making statements of Benevolent Sexism. From Donald Trump, for example, we get both “You have to treat [women] like shit”; and “I cherish women. I want to help women” (Kruse 2015). Hostile and Benevolent Sexism work extremely well together, despite their surface appearance of being in tension, and they are often combined in the behavior and utterances of individuals. Because Benevolent Sexism ofers potent fgleaves, the combination is a powerful way of defusing concerns about sexism.

3.7 Condemnation of Others’ Sexism/Racism Condemning racism or sexism in someone else can serve as an excellent fgleaf for one’s own. Often, these fgleaves are diachronic, rather than occurring along with a problematic utterance Trump ofered one of these fgleaves when he criticized Antonin Scalia’s remarks on afrmative action: I thought it was very tough to the African-American community, actually… I don’t like what he said. No, I don’t like what he said. I heard him, I was like, “Let me read it again” because I actually saw it in print, and I’m going—I read a lot of stuf—and I’m going, “Whoa!”. (Scott 2015) When the Access Hollywood tape surfaced, Trump spent time condemning Bill Clinton, accusing him of sexual assault, and accusing Hillary Clinton of abusing his victims (Associated Press 2016). Trump over the weekend said “Bill Clinton has actually abused women” and Hillary Clinton “bullied, attacked, shamed and intimidated” her husband’s “victims.” This utterance served more than one purpose. It served to associate Hillary Clinton with Bill Clinton’s problematic behavior, and it accused Hillary Clinton of similar behavior. But it also— more relevantly for our purposes here—served as a fgleaf for Trump’s own behavior.

3.8 Deride Special Snowfakes, Political Correctness A currently popular technique for dodging accusations of racism, sexism, or other forms of bigotry is to position oneself as an opponent of “political correctness” or “snowfakes.” An utterance which does this can readily be understood as a fgleaf. Recall that a gender fgleaf is an utterance that blocks an inference from an utterance, S, that would otherwise be seen as sexist, to the claim that the person who uttered S is sexist. Sexist utterances are very often accompanied by what I will call a political correctness fgleaf. Here’s Trump in 2016 (Smith 2016). I’m not going to say it because I’m not allowed to say it because I want to be politically correct,” he began, launching into the naughty-schoolboy shtick he often employs. “So I refuse to say that I cannot stand her screaming into the microphone all the time.” Then he covered his ears, as if shielding them from his opponent’s harpy shrieking. “I just couldn’t stand it. I got such a headache, oh please. As with any fgleaf, this one will only work to block the inference for some audiences. Feminist audiences are likely to be unmoved by the political correctness fgleaf and they will conclude that Trump is demonstrating his sexism. Other audiences—who, in some cases, might otherwise 169

Jennifer Saul

have made the same inference—will focus on his reference to political correctness, their attention turning to this topic rather than Trump’s sexism. In part, then, this works as a distraction. But it is more than that: for someone who thinks it is bad to be “politically correct,” labeling the inference (to the claim that the speaker is racist/sexist) as one that a politically correct person would make is enough to make the inference unattractive, and to thereby keep them from making it. This is an extremely useful all-purpose fgleaf, applicable to any case that might otherwise be recognized as involving a socially unacceptable prejudice.

3.9 Force Figleaves13 So far, we have looked at fgleaves which work by providing a bit of purported evidence which can help the audience avoid having to infer that the speaker is racist. These don’t require any revisiting of what speech act R (the otherwise obviously racist utterance) was. Another sort of fgleaf works diferently, however. These fgleaves work by causing the audience to see R as a diferent sort of speech act than they otherwise would, one which (perhaps) does not serve as an indicator or racism.14 Consider the case of a performance artist. If you walk in on a piece of performance art and don’t realize what is happening, you might think that the performance artist utterances are indicators of what they believe. If, once the performance is over, they explain that you’ve just witnessed performance art, your reevaluate the utterances that you witnessed, taking them to be something other than straightforward assertions and no longer taking them as good indicators of the performance artist’s beliefs. That’s not a fgleaf, but it’s the sort of situation that can help us to see why this kind of fgleaf can be appealing. A force fgleaf is an utterance that functions to change the way that the audience understands the speech act being performed by R, in such a way as to make them not in for that R or its utterer is racist. Some examples of this follow.

3.10 Joking, Locker Room Talk A classic, and very popular, force fgleaf is to fag the context as non-serious. “I was only joking” or “I was being ironic” are extremely popular fgleaves for ofensive utterances of all sorts. These combine especially well with stereotypes of feminists as humorless to provide efective fgleaves for sexist utterances: the feminist objection to a sexist comment can be responded to with “it was just a joke” (a fgleaf, which allows the audience to avoid inferring sexism), and the feminist insistence that it wasn’t just a joke—or that jokes can still be indicators of sexism can be dismissed as a predictable result of feminists’ famed humorlessness. Closely related was the way that Donald Trump defended himself after the Access Hollywood tape surfaced. When the revelation of these tapes led to an uproar, just before the election, Trump responded by saying that the comments were merely “locker room talk” (Diaz 2016): It was locker room talk, as I told you. That was locker room talk. I’m not proud of it. I am a person who has great respect for people, for my family, for the people of this country. And certainly, I’m not proud of it. But that was something that happened, This is meant to work as a fgleaf, by revealing fgleaf to be something other than a serious assertion. And for some, this clearly worked—after all Trump was elected, and in fact a majority of White women voted for him despite these comments. But, like most fgleaves, it did not succeed with everyone. Interestingly, in this case one point of contention was the nature of locker room


Racist and Sexist Figleaves

talk. Many professional athletes came forward to dispute the idea that Trump’s utterance should be seen as “locker room talk”. One example (Gregory 2016): On one side is graphic talk,” says Brendon Ayanbadejo, who spent a decade in the NFL and won a Super Bowl with the 2012 Baltimore Ravens. “The language might even be disgusting. But everyone’s having a good time talking about it. No one is talking about assaulting anyone. That’s criminal talk. Those are completely diferent places. This response shows an efort to challenge a fgleaf by questioning its truth: if locker room talk is not as Trump describes it, then perhaps his utterance needs to be seen as sexist after all. Whether or not such challenges—or others—are efective is a matter that requires empirical study, which has not yet been done.

3.11 Mimicry John Turri (draft), in an interesting exploration of fgleaves from the left of the political spectrum, discusses the case of New York Times reporter Sarah Jeong, whose social media utterances included the following: white men are bullshit white ppl are fucking dumb the world could get by just fne with zero white ppl and the only thing stopping POC is… a disinclination toward genocide!? Jeong responded with what Turri identifes as a fgleaf, explaining that she was only mimicking the abuse she received on social media. As a woman of color on the internet, I have faced torrents of online hate […] I engaged in what I thought of at the time as counter-trolling. While it was intended as satire, I deeply regret that I mimicked the language of my harassers. Turri is right that without this comment, pointing to context, many would likely identify view Jeong’s tweet as racist. So this comment, by blocking that (for some but not all, as is often the case) serves as a fgleaf. A further complication of this case, though, is that while many people would not be inclined to view Jeong’s utterance as racism, this is for diferent reasons from those we’ve discussed so far. So far, I have mentioned the way that many White people adopt defnitions of “racism” which require fully conscious very sweeping condemnations of all members of a racial group. More relevantly for this case, many anti-racist activists adopt defnitions of “racism” that rule out the idea of racism against White people by people of color—these people will have no need of a fgleaf to see Jeong’s utterance as non-racist. As I argue more fully below, in my section on veridical fgleaves, I do not think it is necessary to take a stand on the correct defnition of racism in order to understand how racial fgleaves work. Despite my concerns about the complexities of this particular example, I agree with Turri that the mimickry defense can serve as a fgleaf.

3.12 Overall Incoherence Donald Trump makes use of a rarely used sort of diachronic fgleaf, one that serves to undermine all inferences regarding what Trump’s views actually are. This fgleaf consists of a very large and


Jennifer Saul

well-known collection of contradictory utterances.15 No person could believe all the things expressed by these utterances, and the total collection has led many to hesitate to draw any conclusions at all from what Trump says because his utterances are not indicators of his belief. Arguably, this can be understood as the claim that Trump is known to be what Frankfurt (2005) calls a bullshitter: someone who (unlike a liar) has no regard whatsoever for the truth or falsity of their utterances. Such people are rare, since it is much more common for a person to be a liar—someone who very much cares what the truth is, and cares about making sure that their audience believes something false instead. If Trump is a true bullshitter, this line of thought goes, nothing he says indicates anything about what he thinks.16

4 Efects of Figleaves 4.1 Shifing of Norms Figleaves used by politicians can have efects that go well beyond the particular election campaigns that they may be a part of. In particular, they can allow for changes in our norms and their interpretation which are difcult to achieve otherwise. Langton (2012), McGowan (2012), and others have argued powerfully for a picture on which speech acts may change conversational presuppositions and rules in signifcant, indeed dangerous ways. McGowan is especially concerned with the ways that speech acts of ordinary speakers can change the conversational score (a term borrowed from Lewis), making e.g. racist utterances acceptable where they were not before. Moreover, she suggests, the same process can make racist behavior acceptable. According to McGowan, ordinary utterances function as conversational exercitives, and bring about changes in conversational score unless someone speaks up to object to them. So, for example, an ordinary speaker’s explicitly racist utterance will make racism acceptable in a context unless someone speaks up to object to it. However, I have argued in my (2017) that this picture is not fully adequate on its own: because of the dynamics of conversation and human interaction, this sort of change in conversational score is not automatic whenever people fail to speak up. People who accept the Norm of Racial Equality will generally recognize an explicitly racist utterance as racist, and want to dissociate themselves from the utterer. They may well not speak up—out of fear, shyness, or discomfort with confrontation. But the fact that they accept the Norm means that the efortless shift McGowan predicts will often not take place. Instead, they may change the subject, end the conversation, or pretend the utterance wasn’t made. Of course, speakers who reject the Norm will not want to dissociate themselves from the utterer—but they are irrelevant to the issue of shifting norms, since they already ready reject the Norm of Racial Equality. There’s a gap, then, in McGowan’s picture of how these shifts take place.17 Figleaves can fll this gap. Suppose that a politician makes an utterance that would normally be seen as violating the Norm of Racial Equality, R. However, he accompanies it with a fgleaf, F, that leads the audience to see R as not racist, or the speaker as not racist. Now the audience will be much more comfortable than they would have been otherwise with accommodating the utterance of R, thereby shifting their understanding of what it is acceptable to say, and perhaps of what counts as racism. To make this more vivid, take the case of Donald Trump’s fgleaf-laden commentary about Mexicans (Washington Post Staf 2015): When Mexico sends its people, they’re not sending their best. They’re not sending you. They’re not sending you. They’re sending people that have lots of problems, and they’re bringing those problems with [them]…They’re rapists. And some, I assume, are good people.


Racist and Sexist Figleaves

For the audience who accepts the fgleaves, the rest of the utterance is now seen as something a non-racist might say. That is, this audience now thinks saying that Mexicans are rapists does not show one to be a racist. This is a very substantial shift, aided by fgleaves.

4.2 Enabling More Blatant Speech, Provoking More Opposition As we have noted, fgleaves allow politicians to make extreme norm-violating utterances that they could not otherwise have made. This success at making these utterances means they will resort less and less to dogwhistles, since they can get away with more blatant utterances. This habituates audiences to these sorts of utterances, making them less shocking and less likely to inspire emotional reactions. But this has certain further efects that may pull in the opposite direction, as noted below. The voters who reject the fgleaves now fnd themselves facing a huge collection of shockingly racist utterances that have been made by a particular candidate. Because of the fgleaves, these utterances were made in an explicit manner, rather than by dogwhistling. This means that the utterances are far more widely recognizable as racist than dogwhistle utterances would be, and far more shocking and horrifying. And this, I think, is part of what has galvanized the resistance in the United States. Just after Trump’s election, the resistance poured into the streets (and airports) in protest. As Brexit loomed and then-Prime Minster Theresa May continued her long career of xenophobic policy-making (Consterdine 2018), the lack of outcry and public protest over the xenophobia was notable. It was also notable, though, that May’s utterances (not her policies) were very much less blatant than Trump’s. Quite plausibly, this reduced the expression of outrage and the mobilization of opposition.

5 Open Questions The idea of a fgleaf is new and many issues remain unresolved. In this section, I sketch out four such issues.

5.1 Non-utterances as Figleaves? As defned, fgleaves must be utterances. In the original paper introducing the concept (Saul 2017), I foat the possibility of non-utterances as fgleaves. If we expand the defnition to allow this, then we can make sense of, for example, a human fgleaf—like a Black person announcing a policy that might be seen as racist. A human fgleaf is a person who makes an utterance R that would otherwise be seen as racist. The fact that R is uttered by a human fgleaf blocks at least some of the audience from inferring that R is racist. There are strengths and weaknesses to broadening the defnition of “fgleaf ” in this way. There is a structural similarity that makes this extension appealing, but at the same time there is a worry that too many things may end up counting as fgleaves once we open the door to such extensions—we can imagine circumstances in which images, clothing choices, or music serve as fgleaves. Perhaps this is an accurate refection of our situation and such an extension should be made, but it is not yet clear. A further sort of non-utterance that might be a fgleaf is a fgleaf that plays a role in thought, without being uttered (Saul 2017). A nice real-life example is the following gender fgleaf,


Jennifer Saul

described by someone in the process of recounting how he initially had trouble realizing that he had misogynist sentiments. Here’s the thing. I work with women all day, and I don’t have an issue with anyone. I could’ve told you back then if someone had said to me, “Of, you’re a misogynist. You hate women.” And I could say, “Nuh-uh, I love my mom. I love my sisters. I’ve loved my—the girlfriends that I’ve had in my life.” But you can’t claim to be OK with women and then go online and insult them—seek them out to harm them emotionally.18 According to this man’s report, he would not have recognized the sexism involved in his online abuse of women, because if it had been pointed out he would have used mental fgleaves—about his mother, sisters, co-workers, and girlfriends—to block the inference to the claim that he was misogynist. The fgleaves do not need to be uttered in order to serve the function of blocking this inference.

5.2 Figleaves for Non-utterances? A further question is whether the notion of fgleaves should be extended to actions which might otherwise be seen as racist, as in this possible formulation: A fgleaf is an utterance that blocks at least some observers from inferring that a particular action, A, or a particular person who carried out that action, P, is racist. It is highly intuitive to classify some such utterances as fgleaves. Take, for example, the case of Ralph Northam, the Democratic Governor of Virginia. When it became public that Northam had appeared either as a Klansman or in blackface (it was unclear from the photo which person Northam was), one of the ways he defended himself was to describe his friendships with Black people (Eligon 2019). I was in public school during desegregation,” he said during a news conference, at which his discomfort discussing race showed at times. “I have a lot of African-American friends that I went to school with, played ball with, and I suspect I’ve had as much exposure to people of color as anybody. If fgleaves by defnition must function to prevent particular utterances (or their utterers) being seen as racist, this is not a fgleaf: if it works, it prevents a photo or an action (dressing up for the photo) being seen as racist. Intuitively, however, it feels very much like a fgleaf. A case might, then, be made that the notion should be extended in this way.

5.3 Utterances of Diferent Speakers? As I have understood a fgleaf, it must be uttered by the same person who utters the problematic utterance R. However, some (like Turri) have suggested that we should also countenance the notion of a third-party fgleaf. This extension is certainly possible, and a case can be made for it. Indeed, the defnition of “fgleaf ” I have given is silent as to the identity of the fgleaf ’s utterer. I have always tacitly assumed that the utterer would need to be the same person who utters R, but perhaps other parties can also provide fgleaves. If so, then many additional cases become candidates for fgleaves—including anyone who asserts that someone else’s utterance was not really racist, that the person has Black friends, or that they were only joking. Once more, this may well 174

Racist and Sexist Figleaves

be an accurate refection of how things work, but it may also extend the notion beyond the bounds of usefulness. All this remains to be seen.

5.4 Viridical Figleaves? One of the trickiest issues concerns whether fgleaves must serve as cover for the actual racism/ sexism of a person or utterance. A fgleaf which does this successfully leads away from the truth by blocking inferences to the claim that the person/utterance is racist/sexist. But if the utterance or person actually is not racist or sexist, then the fgleaf, if accepted leads toward the truth. For this reason, I call such fgleaves veridical fgleaves. As I have defned “fgleaf ” above, veridical fgleaves are possible: a fgleaf has the function of blocking an inference to the claim that the speaker/utterance is racist/sexist, for some utterance that would otherwise be seen as racist or sexist.19 Quite clearly, this could happen in a case where that inference would be a mistake. One such possible case would be a person (perhaps a child or a non-native speaker) who makes a linguistic error and mistakenly utters a slur without realizing that it is a slur. If this person then said “Oh no—I didn’t realize what that word meant!,” this utterance would block the inference to the claim that she was racist. It would, then, meet the defnition of a fgleaf. This may seem a little strange—a fgleaf, one might think, should be covering for something. But of course even in its original form on a work of art it needn’t be: there needn’t actually be genitalia under the painted leaf. This consequence, it seems to me, can be reasonably accepted. Other cases, however, push more strongly in another direction. Consider the case of an anti-racist speech on race and crime. This speech contains a sentence noting the diferent incarceration rates for Black and White men in America, then goes on to explain the way that these rates result from both structural and individual racism. The sentence about incarceration rates, on its own, might well seem like a racist utterance. It is the presence of the rest of the speech that blocks the inference to the claim that the utterance, or the speaker, is racist. However, it seems very misleading to call the rest of this speech a fgleaf for one sentence that, taken out of context, would otherwise be seen as racist. One solution to this would be to adopt a defnition which rules out veridical fgleaves. This defnition could require that the utterance R or speaker S would be correctly taken to be racist, if the fgleaf was not present. Since it would be incorrect to take this particular utterance about incarceration rates in this example to be an indicator of racism, the rest of the speech will not count as a fgleaf. This alteration would mean that whether or not a particular utterance is a fgleaf will turn, in part, on whether another utterance (or speaker) is in fact racist.20 Speakers who disagree about the correct understanding of racism, then, will disagree about whether or not a particular utterance is a fgleaf. This does not mean, however, that a theory of racism must be built in to a theory of fgleaves. Whatever theory of racism one has, fgleaves will be the utterances that function in a way that blocks recognition of the racism of certain utterances (with respect to certain audiences). A problem for this defnition, however, lies in the widespread nature of racism, at least on many views. If one considers implicit racism to be racism, then it is by no means unlikely that the author of an anti-racist text criticizing the racism of the prison industrial complex might themself be racist. If this is the case, then the defnition given above will struggle to avoid the verdict that the rest of the text is a fgleaf. And yet this seems wrong. An alternative would be a defnition that requires R, in the absence of a fgleaf, to be an indicator of racism. This might work, but substantial further work would be needed to properly cash out a notion of indicator of racism, for this purpose. (Is it enough that many people who utter a particular phrase do so because of their racism? Then the sentence about incarceration rates might well be an indicator of racism without the fgleaf.) 175

Jennifer Saul

6 Conclusion The concept of a fgleaf is a new one in social and political philosophy of language. It seems to be clearly useful. Yet there are still many unanswered questions, and many ways in which the contours of the concept remain to be defned. Moreover, our political world is evolving very quickly with respect to racism, sexism, and the felt need (or lack thereof ) to cover for them. My hope is that the concept will continue to evolve in ways that help us to understand and combat the forces of bigotry currently on the rise in so many parts of the world.

Acknowledgments I’m very grateful to the extremely insightful and helpful audiences at the Cambridge Meaning Reading Group, the Southwestern Ontario Feminism and Philosophy workshop, and the German Society for Analytic Philosophy. I’m also grateful to Kelly Gaus, Justin Khoo, and Rachel Sterken for helpful comments on a previous draft, and for their patience.

Notes 1 I’m very grateful to the audience at the German Analytic Philosophy Conference in Cologne, where I presented material on sexist fgleaves, and to the Southwest Ontario Feminism and Philosophy Group as well as the Cambridge Meaning Reading Group, for work-in-progress discussions of this chapter. I’m also grateful to the editors and a referee for this book, for helpful feedback. 2 See for example Henderson and McCready (2019), Khoo (2017), Lopez (2014), Mendelberg (2001), Saul (2018, 2019a, b), Stanley (2015), Valentino et al. (2002), though not all of these use the term “dogwhistle”. 3 Figleaves are also used with other subject matters, e.g. “I care about poverty, but…”; “The needs of the disabled are important, but….” Whether or not fgleaves are needed or used will turn in large part on the prevailing norms in one’s community. This is why fundamentalist Christians opposing trans rights are far less likely to say “I support trans rights, but…” than are those who oppose trans rights but consider themselves to be feminist. In my 2019a and 2019b, I discuss the absence of such norms for many proBrexit utterances, and the way that this facilitates their acceptance. 4 I have tacitly assumed that the utterer of the fgleaf must be the same as the utterer of R, but this can be questioned. See 5.3 for a discussion of this point. 5 This is a slightly altered version of the defnitions I gave in my work (2017) and (2019a). 6 Of course, there are some who happily self-identify as racist, and indeed Steve Bannon has suggested (McAuley 2018) that people should be more comfortable with being called “racist”. For those who are not worried about being racist, there won’t be a need for fgleaves. 7 Importantly, though, the fgleaves will not work on everyone, or even on all White people. 8 She does not consider the literature on hostile and Benevolent Sexism. Hostile Sexism, as discussed below, actually seems much closer to fear and resentment, and is certainly not just about role expectations. 9 On racism, see Hill, and my 2017. On sexism, see Manne (2017). Manne draws a distinction between sexism and misogyny, using the term “sexism” slightly diferently from the way that it is used here (this use is broader). But both of these categories—as she uses them—ft well with hostile and Benevolent Sexism. She, like me, thinks that it is a mistake to understand these as solely a matter of individual speakers, but acknowledges that this understanding is widespread. 10 For a fuller discussion of this sort of fgleaf, see my work (2017). 11 I previously (in my 2017) called this “friendship assertion,” but I’ve changed the name slightly to broaden its scope. 12 I’m grateful to Paul Bloomfeld for discussion of this. 13 I previously called these “context” fgleaves, but have found that this generated confusion. I thank the Cambridge Meaning Reading group, several of whose members convinced me to focus instead on the shift of the speech act performed. The term “force fgleaf ” comes from a draft paper by John Turri. 14 This can be seen as structurally similar to the blocking Rae Langton discusses (Langton 2018) but with important diferences. First, it is done by the same person, rather than as a resistant action by someone who wants to change the speech act. And second, when done deliberately, it is generally meant to change the perception of the speech act by those who might object, while leaving in place many of its efects. It is also related, even more closely, to Laura Caponetto’s work (2020) on undoing things with words. This work focuses on frst personal eforts to change a speech act that one has already made.


Racist and Sexist Figleaves 15 On this point see also Lynch (2016). 16 I explore this more fully with Tim Kenyon in a paper currently in progress, “Bald-Faced Bullshitters: Making Sense of Trump and Johnson.” 17 And, it’s worth noting, sometimes the shift may happen efortlessly. This is more likely in cases that also involve an authoritative speaker. For more on the efects of authoritative speakers and on how this authority functions, see Langton (1993) and Maitra (2009, 2012). 18 Interview with Lindy West, on This American Life. Quoted in Manne (2017): 52. 19 Some earlier formulations of this are less neutral, seeming to require that R or the person who utters R is actually racist. 20 I’m focusing here on racism, but the same points will hold regarding sexism.

References Anonymous. 2017. “Decoding What Donald Trump Is ‘Not Saying, Just Saying’”, News.Com.Au, https:// news-story/6f1f4c30a07d80388e22a3d27773f9cf. (Consulted 11 September 2020). Associated Press. 2016. “Defant Trump Siezes on Clinton Sex Scandal in Wake of Tape Leak”, Fortune, 9 October, (Consulted 11 September 2020). Blake, A. 2017. “21 Times Donald Trump Has Assured Us He Respects Women”, Washington Post, https:// (Consulted 29 August 2020). Caponetto, Laura, 2020. “Undoing Things with Words”, Synthese 197, 2399–414. Cohen, C. 2017. “Donald Trump Sexism Tracker: Every Ofensive Comment in One Place”, The Telegraph, (Consulted 29 August 2020). Consterdine, E. 2018. “Hostile Environment: The UK Government’s Draconian Immigration Policy Explained”, The Conversation, (Consulted 11 September 2020). Crockett, E. 2016. “Why Misogyny Won”, Vox 15 November, 13571478/trump-president-sexual-assault-sexism-misogyny-won. Diaz, D. 2016. “3 Times Trump Defended His ‘Locker Room’ Talk”, CNN Politics, 10 October, https:// (Consulted 11 September 2020.) Eligon, J. 2019. “The ‘Some of My Best Friends Are Black’ Defense”, New York Times 16 February, https:// (Consulted 11 September 2020.) Frankfurt, H. 2005. On Bullshit. Princeton: Princeton University Press. Glick, P., and Fiske, S. T. (1996). “The Ambivalent Sexism Inventory: Diferentiating hostile and benevolent sexism”, Journal of Personality and Social Psychology 70:3, 491–512. Glick, P., and Fiske, S. T. (1997). “Hostile and benevolent sexism: Measuring ambivalent sexist attitudes toward women”, Psychology of Women Quarterly 21:1, 119–35. Glick, P., and Fiske, S. T. (2001). “An Ambivalent Alliance: Hostile and Benevolent Sexism as Complementary Justifcations for Gender Inequality”, American Psychologist 56:2, 109–18. Gregorian, D. 2019. “Heated Racial Dispute between Reps. Tlaib and Meadows Mars Cohen Hearing”, NBC News, (Consulted 29 August 2020). Gregory, S. 2016. “Donald Trump Dismisses His ‘Locker Room Talk’ As Normal. Athletes Say It’s Not”. Time, October 10, 2016. (Consulted 15 January 2021). Henderson, R. and McCready, E. 2019. “Dogwhistles and the At-Issue/Not-At-Issue Distinction”, in Gutzmann, D. and Turgay, K., Secondary Content, Leiden: Brill, 222–245. Hill, Jane. 2008. The Everyday Language of White Racism. Chichester: Wiley- Blackwell. Jacobs, B., Siddiqui, S. and Bixby, S. 2016. “You Can Do Anything”: Trump Brags on Tape about Using Fame to Get Women”, The Guardian, (Consulted 29 August 2020). Kenyon, T. and Saul, J. In Progress. “Bald-Faced Bullshitters: Making Sense of Trump and Johnson”, coauthored with Tim Kenyon, to be published in Horn, L., From Lying to Perjury: Linguistic and Legal Perspectives on Lies and Other Falsehoods, Berlin: De Gruyter. Khoo. 2017. “Code Words in Political Discourse”, Philosophical Topics 45:2, 33–64.


Jennifer Saul Kruse, M. 2015. “The 199 Most Donald Trump Things Donald Trump Has Ever Said”, Politico, 14 August, (Consulted 11 September 2020). Langton, R. 1993. “Speech Acts and Unspeakable Acts”, Philosophy and Public Afairs 22:4, 293–330. Langton, R. 2012. “Beyond Belief: Pragmatics in Hate Speech and Pornography”, in Maitra, I. and McGowan, M.K., Speech and Harm: Controversies Over Free Speech, Oxford: Oxford University Press, 72–93. Langton, R. 2018. “Blocking as Counter-Speech”, in Fogal, D., Harris, D., and Moss, M., New Work on Speech Acts, New York: Oxford University Press, 144–64. Lopez, I. 2014. Dog Whistle Politics: How Coded Racial Appeals Have Reinvented Racism and Wrecked the Middle Class. New York, NY: Oxford University Press. Lynch, M. 2016. “Trump, Truth, and the Power of Contradiction”, New York Times, https://www.nytimes. com/2016/05/08/opinion/sunday/trump-truth-and-the-power-of-contradiction.html. Maitra, I. 2009. “Silencing Speech”, Canadian Journal of Philosophy 39:2, 309–38. Maitra, I. 2012. “Subordinating Speech”, in Maitra, I. and McGowan, M.K., Speech and Harm: Controversies Over Free Speech, Oxford: Oxford University Press, 94–120. Manne, K. 2017. Down Girl: The Logic of Misogyny. New York: Oxford University Press. Mathis-Lilley, B. 2016. “Trump Was Recorded in 2005 Bragging about Grabbing Women “by the Pussy””, Slate, by_the_pussy.html. (Consulted on 29 August 2020). McAuley, J. 2018. “’Let Them Call You Racists’”: Steve Bannon Delivers Fighting Speech to France’s National Front”, Washington Post March 10, 2018, y-is-with-us-steve-bannon-delivers-rhetoric-f illed-speech-to-frances-nationalfront/2018/03/10/4f21e016-2480-11e8-946c-9420060cb7bd_story.html. (Consulted 8 August 2020.) McGowan, Mary Kate. 2012. “On ‘Whites Only’ Signs and Racist Hate Speech: Verbal Acts of Racial Discrimination”, in Maitra, I. and McGowan, M.K., Speech and Harm: Controversies Over Free Speech, Oxford: Oxford University Press, 121–47. Mendelberg, T. 2001. The Race Card: Campaign Strategy, Implicit Messages, and the Norm of Equality. Princeton: Princeton University Press. Meyer, K. 2017. “Conway Asks: Why Do You Believe What Trump Says ‘Rather Than What’s in His Heart’?”, Mediaite, (Consulted 1 August 2020). Michallon, C. 2020. “Ted Yoho Used His Daughters as a Shield Against AOC. How Original”, The Independent 24 July, (Consulted 11 September 2020). Nordquist, R. 2019. “Paralepis (Rhetoric)”, Glossary of Grammatical and Rhetorical Terms, ThoughtCo, (Consulted 11 September 2020). Saul, J. 2017. “Racial Figleaves, the Shifting Boundaries of the Permissible, and the Rise of Donald Trump”, Philosophical Topics 45:2, 97–116. Saul, J. 2018. “Dogwhistles, Political Manipulation, and Philosophy of Language”, in Fogal, D., Harris, D., and Moss, M., New Work on Speech Acts, New York: Oxford University Press, 360–83. Saul, J. 2019a. “What Is Happening to Our Norms against Racist Speech?”, Aristotelian Society 93:1, 1–23. Saul, J. 2019b. “Immigration in the Brexit Campaign: Protean Dogwhistles and Political Manipulation”, in Fox, C, and Saunders, J., Media Ethics, Free Speech, and the Requirements of Democracy, New York, NY: Routledge, 21–37. Scott, E. 2015. “Trump Hits Scalia Over Comments On Black Students”, CNN Politics, 13 December, (Consulted 11 September 2020). Smith, C. 2016. “Coded and Loaded: How Politicians Talk about Race and Gender without Really Talking about Race and Gender”, California Magazine, (Consulted 11 September 2020). Stanley, J. 2015. How Propaganda Works. Princeton, NJ: Princeton University Press. Tesler, M. and Sears, D. O. 2010. Obama’s Race: The 2008 Election and the Dream of a Post-Racial America. Chicago, IL: University of Chicago Press. Turri, J. Draft. “Figleaves Right and Left: A Case-study of Viewpoint Diversity Applied to the Philosophy of Language.” Valentino, N., Hutchings, V. and White, I. 2002. “Cues That Matter: How Political Ads Prime Racial Attitudes During Campaigns”. American Political Science Review 96: 1. Washington Post Staf. 2015. “Full Text: Donald Trump Announces a Presidential Bid”, Washington Post, 16 June, (Consulted 11 September 2020).


11 PROTEST AND SPEECH ACT THEORY Matthew Chrisman and Graham Hubbs

During the fnal match of the 2018 FIFA World Cup, four women in security uniforms temporarily halted play early in the second half by running across the pitch. The goal of their intervention was not immediately obvious to the average spectator. At the moment it happened, it was not unreasonable to think that their aim was the attention of infamy. One of the women, Veronika Nikulshina, managed to high-ten French star Kylian Mbappé—perhaps she wanted to convey her afection for his play. After the match, a Facebook post clarifed the point of their action: these were members of Pussy Riot, and they had disrupted the match to express dissatisfaction about and make political demands of Vladimir Putin’s regime. The pitch invasion, it turned out, was an act of political protest.1 Our interest in this chapter is to explicate the communicative structure of protest by recourse to speech act theory. Even in non-verbal protests such as the Pussy Riot example, we will argue, protests contain an implicit communicative structure that speech act theory can make explicit. Our analysis will also allow us to distinguish protest from similar sorts of complaints and demands. We proceed in three stages. First, we make some preliminary distinctions that help to identify protests as a complex, dual-aspect speech act. We next improve on this analysis by mobilizing the idea of a felicity condition of a speech act; specifcally, we will argue that the speech act of protest can be understood as having a complex felicity condition stemming from the felicity conditions of making a specifc kind of evaluative claim and a connected prescription. We then kick away the ladder of felicity conditions, further improving the analysis by arguing that the structure of the two connected communicative aspects of protest can be understood in terms of the normative statuses presumed by and afected through protest. Our account here belongs to previous work we have done on the place of protest in non-ideal political theory. As we have discussed elsewhere (Chrisman and Hubbs 2018, 168), it is imaginable that an ideal polity would have no need, and so no place for, protest. In non-ideal polities, however, institutions that allow for the expression of at least some minority and dissenting opinions are needed if the polity is to be even minimally just. Legal protection for protest is one such institution; channels for recognizing the political legitimacy of some forms of civil disobedience are, arguably, another. John Rawls, for example, famously characterized civil disobedience as “one of the stabilizing devices of a constitutional system, although by defnition an illegal one” (1999a, 336). It is clearly non-ideal to rely on illegal practices to “stabilize” a constitutional system, and it is also not ideal for legitimate opinions and demands to go unheard even if these are expressed legally. The theory we present below is meant to apply both to legal and illegal forms of the nonideal speech act of protest. 179

Matthew Chrisman and Graham Hubbs

1 Protests: Some Preliminary Distinctions Protests come in varying degrees of complexity: a political rally, for example, can involve a variety of protesters protesting a variety of things. To take advantage of the resources of speech act theory, we will start by focusing on a single agent performing a single act of protest. In the standard individual case, this minimally involves conveying opposition to something and, in response, advocating for some sort of corresponding change. John Searle acknowledges these aspects when he writes that “…a protest involves both an expression of disapproval and a petition for change” (Searle 1975, 368), though we might add that expressing disapproval is one among potentially many ways a protest can convey opposition to something, and petitioning for change is just one among potentially many ways a protest can prescribe redress. Below, as we develop successive stages of the analysis, we will seek to refne these notions in order to distinguish protest from other related speech acts. For now, what is important to notice is that these communicative aspects of a protest are distinct, not only from each other, but also from the way in which a person protests, i.e., the means of her protest. When one protests, then, there is something one is against—the object—there is something one prescribes—the redress—and there is a way of communicating these connected stances—the means. In some basic cases, the object that the protester opposes is some wrongdoing, and the redress the protestor pursues is simply an acknowledgment of and/or apology for that wrong. For example, protests of historical wrongs such as the internment of Japanese Americans during World War II have this favor. A protestor expresses disapproval of what happened to these people, and they may want merely for this wrongdoing to be more widely acknowledged. In other basic cases, the object of a protest is a prohibition, and the redress the protester pursues is to lift that prohibition. For example, the object of a sufrage protest might be, e.g., the prohibition on women voting, and the corresponding redress might be to grant women the vote. In this case, the diverse acts of protest (carrying placards, chaining oneself to parliament, smashing windows, etc.) are various ways of communicating opposition to an aspect of the status quo and prescribing that it be changed—these are the various means of protest. In other cases, the object can be the status quo more generally, and the corresponding prescription may not be specifcally defned. Tommie Shelby (2016, 223–5) and Candice Delmas (2018, 44–8) suggest that aspects of urban riots can be understood this way, as deeply frustrated reactions to gross and systemic injustice and as demanding that things, generally, change. In some cases of protest, a single object may correspond with a number of distinct prescriptions for redress. Take, for example, a protest in response to the police killing of an unarmed person of color. Such a protest might have a variety of objects, e.g. the event of the killing, or the systemic behavior of the police. For the sake of simplicity, let the object in this case be the event of the killing. Even with respect to this single object, a protester might seek any of a variety of redresses: she might want the ofcer who committed the killing to lose his badge, or for that ofcer to be put on trial, or for the police chief to step down, or for the police force to undergo an independent investigation, or all of these, or something else. A given object of protest, then, might be connected to a variety of prescribed courses of redress. Because of this, as mentioned above, an event in which several protesters participate may be united by a single object but may involve several diferent individual acts of protest, each with its own prescription for redress. Moreover, such events may involve acts that difer not just in their prescriptions for redress but also in their objects. At anti-globalization protests, for example, one may fnd objects ranging from specifc trade treaties to the activities of specifc corporations to the laws governing those corporations to the governments that produce those laws to the politicians that make up those governments. Each of these objects, in turn, might be connected to a multiplicity of corresponding demands. 180

Protest and Speech Act Theory

Moreover, a single protester with a single object and a single desired redress may employ a multiplicity of means over the course of an act of protest. Staying with the police-protest example: suppose a protester participates in a rally whose explicit demand is to have charges brought against a specifc police ofcer for a shooting. This protester may march legally in the street, or they may do so illegally, and as they march either legally or illegally, they may chant, and/or carry a sign, and/or perform a symbolic gesture (e.g. put their hands up in the air while saying “Hands up; don’t shoot”). Each of these actions involves the same object and the same prescription, but they are diferent means of drawing attention to the object and pursuing the redress. In Pussy Riot’s World Cup protest, some members invaded the pitch of the fnal match, while another made a Facebook post to clarify the point of the invasion. These are, clearly, separable actions; together, they conveyed opposition to the Putin regime and prescribed specifc corresponding remedies. For some acts of protest it is useful to distinguish between the direct communicative goal of protest and the broader goals of the act of communication. This distinction is helpful, for example, for characterizing the role of solidarity in protesting. Suppose someone who is not a Muslim protests an immigration ban on Muslims. Their communicative goal is to object to the ban and to demand that it be changed; however, their broader goal in performing that communicative act might be to show the Muslim members of their community that someone who is not directly affected by the ban stands in solidarity with them. The redress that is prescribed by an act of protest, then, need not be the only communicative goal that the protesting agent pursues by her protest. It is also useful to distinguish between the intended audience of an act of protest and the agent who is called upon to perform the protest’s redress, whom we shall call the addressee of the protest. In some cases, the protest’s addressee may not be the primary intended audience. A group of citizens in a small Idaho community may gather to protest Donald Trump’s immigration policies even if none expect that Trump himself will ever become aware of the protest. Their primary audience may be the members of their community even as they call on their addressee (Trump) to change his policies. For this act to count as a protest, however, the protestors must be committed to thinking Trump could come to know about their demands and so in that weak sense be a possible member of the audience. Moreover, for an act to count as a protest, it must be sufciently public for some general audience to be aware of the protest. Complaining to one’s co-workers about a company policy is not protest, and telling one’s boss privately and confdentially that one disagrees with the policy is not protest, but expressing this disagreement with a megaphone outside of company headquarters is. When we speak of a protester’s intended audience, we do not mean to suggest that a protester must have someone explicitly in mind when protesting. The intended audience might be implicit; the protester might only think explicitly of the protester if asked, “Who do you want to hear your protest?” Moreover, the intended audience need not be very specifc—it might be, e.g., whoever will listen or the public. On some occasions, however, the intended audience may explicitly include specifc individuals: a group of students might, e.g., protest explicitly and specifcally to a university president (who, here, would also be the addressee) to demand that a building’s name be changed. As we understand the practice, protests necessarily involve diferent relations of power, or authority, or something similar, between the protester and the agent from whom redress is desired. We take no stand here on what hierarchical notion—power, or authority, or something else—best articulates this diference. The examples above demonstrate what we have in mind: consider the relation between the Russian government and Russian citizens, or the relation between a government and those it governs who lack the right to vote, or the relation between a police force and the people over whom it enforces the law. In each of these, the addressee has authority or power that the protesters lack, and the protesters are demanding that that authority or power be exercised in a way it has not been and/or is not being exercised. To put the point starkly: Pussy Riot can 181

Matthew Chrisman and Graham Hubbs

protest Putin’s politics, but it is impossible for Putin to protest Pussy Riot’s politics. If he objects to something they do, he does not protest that action; if he wants redress, he can efect it directly through, e.g., imprisonment. All of the examples of protest we have considered so far are explicitly political. However, if there are non-political hierarchies, then non-political protest is possible as well. A recreational youth football team might not be a political entity; if it is not, its players can non-politically protest by, e.g., sitting down when the coach instructs them to run laps around the pitch. This act of sitting could be a mere display of laziness or exhaustion, but it could also have the communicative elements of a protest, when, for example, it expresses opposition to the coach’s command and prescribes that the coach be less demanding with his lap-running requirements, either just in this specifc instance or more generally. Our focus for this chapter will remain fxed on explicitly political protest, but we intend what we say to be sufciently general to allow for non-political protest as well.

2 Felicity Conditions, Uptake, and the Speech Act of Protest So far, the frst stage of our analysis has been to trace out some initial implications of the idea that a protest can be viewed as a complex speech act, where a person takes some communicative means in an efort to (i) convey opposition to some object and (ii) prescribe redress related to that opposition.2 This analysis, however, is not sufciently precise, for some communicative actions that aim at both (i) and (ii) do not seem to be protests. Imagine a traveler who has been bumped of of an overcrowded fight in accordance with the terms and conditions of his ticket. Imagine this traveler yelling at an airline attendant, saying, “This is crap! I have to make it to my destination tonight. You have to put me on this fight!” This conveys opposition to the travel situation and prescribes a course of redress, but it is not obviously a protest; indeed, we think there is good reason not to group this case with the examples of protest from the previous section. In this section, we will argue for this claim by pursuing a second stage of the analysis, in which we attempt to identify some felicity conditions distinguishing protest from other similar communicative actions. 3 J. L. Austin (1975, 14–24) argued infuentially that we can identify felicity conditions for different kinds of speech acts. Felicity conditions themselves can be divided into distinct classes. One familiar class comprises those conditions that, when violated, undermine one’s counting as performing the act one aims to perform in speaking. For example, imagine a priest in Belize who attempts to wed two people of the same sex. The priest does not have the legal authority to wed these people, for homosexual marriage is illegal in Belize. So, when he says of this couple, “I now pronounce you married,” he has not wed them, for the felicity conditions for wedding a couple have been violated. To count as wedding two people in Belize, one must have a certain normative status (i.e. one must be legally authorized to do so), and one must follow the rules in performing the wedding ceremony. Failure to have the needed status or to follow the ceremonial rules results, not in a bad marriage, but in no marriage at all. Austin (1975, 16–18) calls these sorts of “unhappy” speech acts misfres; because these speech acts fail to satisfy norms that constitute the intended action type, the acts do not count as instances of the relevant type. A diferent kind of felicity condition derives from norms whose applicability is partially constitutive of an action counting as a specifc type of speech act. One can fail to meet this sort of norm while still counting as performing the relevant speech act; in such a case, one just counts as performing the act poorly or incorrectly relative to the norm.4 To see what we have in mind, consider the view that truth is a constitutive aim of assertion.5 Someone who holds this view does not claim that, in order to count as an assertion, a speech act must be true; rather, the idea is that the fact that someone’s saying something is correct/incorrect in virtue of its relation to truth is (part  of ) what makes that speech act an assertion. On this view, it is the applicability of, not conformity 182

Protest and Speech Act Theory

to, the norm “it ought to be true” that is partially determinative of something’s counting as an assertion. Speakers need not themselves endeavor to speak truly for their speech to count as assertion, as evidenced by lying.6 What is important, rather, is the type of normative evaluation that is appropriate for the speech act. Taking this position allows one to distinguish assertion from other uses of declarative sentences, such as hypothetical supposition and overt fction. This is why, when Doyle says, “Holmes lives on Baker Street,” it is not appropriate to wonder whether what he said is true, for Doyle is not asserting that Holmes lives on Baker Street but rather is telling a fctional tale. Another idea from speech act theory is relevant here: for some speech acts, only minimal uptake on the part of the audience is required for them to count as instances of their kind. For example, for a speech act to count as a warning, it is enough for the audience to understand that the speaker, in saying what she is saying, is trying to alert the audience of some perceived danger. When Cassandra admonishes the Trojans not to take the giant horse left behind by the Greeks, she succeeds in warning the Trojans even though they ignore her advice. By contrast, if we describe the same act as persuading the audience to guard against the danger, then more audience uptake is required for the speech act to satisfy this description; the audience must become convinced to react in accordance with the warning. Although Cassandra manages to warn the Trojans, she does not persuade them to leave the horse outside the city walls, for they do not heed her warning.7 We can mark this diference in uptake between warning and persuading by comparing the intelligibility of the following sentences. “Cassandra warned the Trojans, but they didn’t listen to her” is intelligible—it indicates that the Trojans understood Cassandra’s warning but did not heed it. By contrast, “Cassandra persuaded the Trojans, but they did not listen to her” is much more difcult to make sense of, as the latter clause appears to preclude the former.8 We will return to this distinction at the end of this section. With these points in mind, let us return to our disgruntled traveler. The object of his anger is the weather delay. He expresses disapproval of his travel situation and demands redress. Why not count this, then, as a protest, grouping it together with the examples from the previous section? In discussing civil disobedience, which we view as a species of protest, Rawls provides a clue. He writes: “…it is important that the action be properly designed to make an efective appeal to the wider community” (1999a, 330). Rawls is imagining protest actions as making appeals to one’s constitutional community, wherein there exists some overlapping conception of the basic principles of just interaction. For this reason, famously, in A Theory of Justice he claims that civil disobedience should address “the sense of justice in the community, an invocation of the recognized principles of cooperation amongst equals” (1999a, 337). In order to cover other sorts of community, and other sorts of protest, one might propose something more general than what Rawls says here.9 That is, one way to explain why the traveler counts as merely complaining rather than protesting is to argue that protest involves conveying opposition to something, not merely by expressing disapproval (however justifed), but by doing so in a way that amounts to publicly making an evaluative claim, one which appeals to a wider audience, and in efect says that the object of protest is in some way unfair or unjust.10 To be more precise, the present suggestion is that, unlike a mere complainer, a protestor presumes there is some overlapping conception of what is unfair or unjust between them and their audience, and they are publicly characterizing (at least implicitly) the object of protest as falling under this conception. The traveler’s travel situation may be highly annoying to him, but it is implausible to view his expression of disapproval as designed to make an efective appeal to some overlapping conception of what is unfair or unjust of the wider community. We are not claiming that the relevant conception of what is unfair or unjust needs to be very precise or explicit in the minds of protestors. Instead, we are suggesting that this kind of grounding is part of the normative structure of the social practice, rather than a conscious feature of the intentions of those engaged in the practice. By comparison, those who engage in assertion might not have a very precise understanding of truth or have some explicitly self-conscious intention to 183

Matthew Chrisman and Graham Hubbs

say what is true. Mirroring the metaphor of assertions constitutively aiming at truth, we can then generate a more precise analysis as follows: protests constitutively aim at truth about what is unfair or unjust relative to some conception of what is fair or just that is presumed to overlap with the protest’s audience. This can be only half the story, however, for along with this evaluative component, and as we have been emphasizing throughout, protests also have a prescriptive component. Continuing with a generalization of the Rawlsian idea, we claim that protests prescribe redress in a specifc way, namely, as a way to address the unfairness or injustice claimed. In the language of constitutive aims, the present suggestion is that, in addition to constitutively aiming at truth about what is unfair or unjust, protests also aim at the imposition of an obligation for their addressee to do something about the object of protest. Reformulated as a pair of “ought” claims, the present suggestion is that the negative evaluation that the object is unfair or unjust implicit in a protest ought to be true, and the prescribed redress in a protest ought to be such so as to impose an obligation on the protest’s addressee. Putting together the two halves of the story, we get the view that the applicability of this pair of norms is part of what makes a speech act a protest. No one, the disgruntled traveler included, presumes that bad weather is unfair or unjust or that his demands impose any justice- or fairness-based obligation on the airline employees. His complaints and demands about his travel situation are thus not even subject to the speech act norms constitutively applying to protest, and this provides grounds for not classifying his action as a protest. As is the case with assertion, a particular protestor might not themself explicitly aim at conforming to the constitutive norms, but the present suggestion is that the fact that their speech act is understood by participants in the social practice to be evaluable with respect to these norms is part of what constitutes it as a protest. In speaking only of a presumed overlapping conception of what is unfair or unjust, someone pursuing this broadly Austinian analysis of protest would have to acknowledge that the intended audience and/or addressee of a protest can disagree with the protestor’s view that the object of protest is unfair or unjust. If I protest an extant law that you think is perfectly good but I think is unjust, you are likely to be unpersuaded by my protest, because we do not agree about the justice of the law. For my speech act to count as a protest, it must be clear that I hold the law in question to be unjust and think that you should agree. If you and I do not already have a common view about the injustice of a situation—in this case, whether the relevant law is or isn’t just—then I need you to come to share my view if I want my protest to succeed. There are at least two ways I can attempt this. I can seek a broader conception of justice we already share and try to persuade you that, according to it, the law I am protesting is unjust by the lights of this conception and the redress prescribed is a reasonable way to respond. Alternatively, I can attempt to bring you over to my conception of what is unjust, or at least the relevant part of this conception for evaluating the object of protest and the redress pursued. My presumption, then, is either that our conceptions of justice already overlap in some relevant respect or that you should come to share at least this element of my conception of what is unjust. Even if, on some occasion, this presumption fails to hold—i.e. even if the audience never comes to share an overlapping conception of what is unfair or unjust—the protester may still succeed in performing an act of protest. Indeed, even if the protestor knows the presumption of an overlapping conception of what is unjust will remain false with respect to a particular audience, the protestor may still succeed in performing an act of protest.11 Like warnings, protests are speech acts whose execution requires minimal audience uptake. Returning to the distinction illustrated above by the Cassandra examples: “The crowd protested Bush’s proposal to invade Iraq, but the White House didn’t listen” is intelligible. For a speaker here (e.g., some member of the crowd) to count as protesting, her audience (e.g., Bush and/or members of his administration) must understand why she is saying what she is saying but not necessarily change their minds as a result of this understanding. More specifcally, the audience must understand what the speaker is saying as the expression of 184

Protest and Speech Act Theory

negative evaluative claim about something being unfair or unjust and as prescribing some redress to this presumed unfairness or injustice. It is not, by contrast, necessary that the audience be persuaded of the unfairness or injustice of the object or the reasonableness of the prescribed redress as a way to address this purported unfairness or injustice. One counts as performing a speech act of protesting even if one fails to convince the audience of either of these points.

3 Protest and the Pragmatic Topography of Speech Acts We have just examined the possibility of improving the speech act analysis of protest by drawing on the Austinian notion of felicity conditions. We appealed to an analogy with assertions, which have commonly been thought to be partly constituted by the applicability of a norm of truth, to suggest that protests have a similar felicity condition: they ought to convey true claims about something’s being unfair or unjust relative to a presumed overlapping conception of fairness or justice. On refection, however, this doesn’t seem quite right. For one thing, not all protests are obviously about unfairness or injustice: some have suggested to us that one protest the color of the national team’s football jersey or the closing of a widely beloved bar in one’s neighborhood without obviously making a claim about what is unfair or unjust relative to the presumed overlapping conception of fairness or justice.12 For another thing, it is not clear, at all, that protests typically make claims (even just implicitly) about what the presumed shared conception of fairness or justice would rule out. Instead, what seems crucial to a speech act’s being a protest is that protestors make a negative evaluation that purports to be grounded in the sorts of considerations that could convince one’s audience to agree with the evaluation. Perhaps such negative evaluations always come from some particular normative perspective, and perhaps one’s audience will come to share the negative evaluation only if they consider the issue from a relevantly overlapping normative perspective, but it does not seem that protests typically aim at truth about normative perspectives. Another worry about the previous section’s analysis concerns the idea that protests are infelicitous if they do not (at least implicitly) impose obligations of redress on their addressee. On refection, this too can’t be quite right. Normally we think the speech act of commanding is successful only if it imposes obligations, but in order for this to happen, the speaker must have some kind of authority over the addressee. When it comes to protests, however, the power relation is clearly diferent. Putin has a kind of authority over Pussy Riot, and because of this, as noted at the end of Section 1.1, he cannot protest their actions; it is only those who lack control or authority over the addressee who are in a position to protest. But if they lack control or authority, how could they impose obligations on the addressee of their protest? And how could doing so be a constitutive aim of the speech act of protest? To address these concerns, we have to replace the idea of felicity conditions with a more precise tool from speech act theory that builds on the idea that assertion can be helpfully characterized in normative scorekeeping terms.13 The basic idea is that certain conditions must obtain in order for a speaker to be entitled to assert something, and when a speaker is presumed to assert something with entitlement, then other normative statuses change in rule-governed ways.14 Quill Kukla (writing as Rebecca Kukla) and Mark Lance (2009) have suggested a generalization of this idea, whereby every speech act is characterized in normative scorekeeping terms. Our competence with the linguistic practices in which we take part can then be viewed as a matter of implicit scorekeeping, and particular speech acts can be abstractly characterized not in terms of the conditions under which they are felicitous but in terms of normative functions from input conditions for a speaker’s entitlement to perform the act in question to output statuses afected by a presumed entitled performance of that act. This is what we might term a normative functionalist approach to understanding the nature of speech acts, because it replaces felicity conditions with normative functions in order to individuate speech acts. 185

Matthew Chrisman and Graham Hubbs

More precisely, for any particular speech act, we can attempt to characterize the facts, considerations, or reasons that entitle one to perform a speech act. These facts, considerations, and reasons are going to be diverse and often particular to specifc speech situations, but we might attempt to articulate partially constitutive entitlement conditions for various speech act kinds. However, this is only half of the normative functionalist story. The other half is the outputs—that is, the entitlements, obligations, and reasons to do things, including but not limited to performing other speech acts—that are generated by an entitled performance of a speech act. Again, these are going to be diverse and often particular to specifc speech situations, but we might attempt to articulate the general kinds of entitlements, obligations, and reasons the generation of which is partially constitutive of various speech act kinds. By pairing these up with constitutive entitlement conditions and comparing and contrasting speech act kinds in terms of their input-output functions between constitutive normative statuses, we can generate what Kukla and Lance call a “topography” of the normative terrain of linguistic practices. For example, rather than say that an assertion is felicitous only if true, we can instead say that (input:) an assertion is what it is because its being true is (a central part of ) what one presumes when one takes its performance to be entitled. Arguably, at least in core cases of assertion, the speaker incurs an obligation or at least a reason to defend their assertion or retract it when presented with reasonable challenges. Moreover, (output:) accepting an assertion as entitled is accepting its content as true and holding that others are also entitled to accept its content as true. This, in turn, supplies a reason for anyone to think it true. Similarly, rather than say that a command is felicitous only if made with authority to impose obligations, we now say that (input:) a command is what it is because the speaker’s having authority to impose obligations on the addressee is (a central part of ) what one presumes when one takes its performance to be entitled. Moreover, (output:) accepting a command as entitled is accepting that the addressee of the command has acquired an obligation to do what they have been commanded to do. Those examples are overly simplifed and might need “ceteris paribus” or “pro tanto” clauses to specify the normative functions constituting assertion and command more precisely, but the general normative functionalist strategy is to individuate diferent speech acts in the linguistic practice of which they are a part by identifying what it takes to be entitled to perform the speech act and what such a performance does to the downstream normative statuses of speaker and audience. On this normative functionalist conception of linguistic practice, speakers and hearers “keep score” on ongoing conversations by tracking what speech acts are being performed, which they do by understanding (at least implicitly) the associated function characterizing what the normative score has to be for the speaker to be entitled to perform the act in question and what efect such performance has on the normative score. Diferent speech acts are individuated by diferent normative functions. In addition, at a very abstract level, the input and output normative statuses that individuate speech acts can be usefully classifed as particular or general depending on whether having them depends on something particular to the person who has them above and beyond their generic status of being the speaker of a language. For example, if we say that an assertion presumes entitlement based in the truth of what is said, and we think that what is true for one speaker is true for any other speaker (i.e., that the relevant truth isn’t speaker-relative), then assertions presume a general entitlement in the sense that this entitlement is available to anyone if it is available to the speaker. Because of this, in asserting p, a speaker is in efect accepting that the normative score is such that any other speaker would also be entitled to assert p. Moreover, if we say that, when one successfully asserts something, others become entitled to believe what was asserted, then this output normative status is also general in the sense that it doesn’t depend on who it is that’s acquiring the entitlement. If one person gets this new normative status by virtue of hearing and understanding the assertion as an assertion, then so does anyone else who hears and understands the assertion as 186

Protest and Speech Act Theory

an assertion. In successfully asserting, a speaker gives anyone who recognizes the assertion a reason to believe what is said, not just particular people, such as friends or subordinates or the Queen. By contrast, as already observed, the speech act of commanding presumes entitlement based in the speaker’s authority over a particular audience member, viz. the addressee. So entitlement to command someone to do something is particular in the sense that it depends on something particular to the speaker, namely, their relationship to the addressee, which the speaker need not presume others to have. Moreover, when one successfully commands someone to do something, one changes the obligations of the addressee but no one else. So, this output normative status is also particular in the sense that the particular person addressed by the command, and no one else, acquires the obligation. In successfully commanding someone to do something, a speaker gives an obligation to a particular person (such as a subordinate). The normative function representing asserting is thus from general input status to general output status, whereas the normative function representing commanding is from particular input status to particular output status. Many speech acts, however, are more complex in their normative input/output statuses than these simple examples suggest. Indeed, a full articulation even of the practice of asserting and commanding may reveal other normative statuses that are partially constitutive of these speech acts within a linguistic practice that have contrasting particular or general aspects. In any case, at the most basic level, it can be useful to classify normative input conditions (what entitles a speech act) and normative output statuses (what entitlements, obligations, and reasons change as a result of an entitled performance of the speech act) as particular or general. For one thing, this helps to see the possibility of speech acts with a general input status like assertion, but particular output status like commands, and vice versa. For another, this helps to map the way complex speech acts can involve both general and particular elements in both the input and output statuses of the normative functions constituting them as the speech acts that they are.15 Examples of these mixed-element and complex speech act kinds are easy to come by. First, consider moral, prudential, or legal claims about what some particular person should do. Insofar as these have truth about what is morally, prudentially or legally prescribed as an entitlement condition, they might be viewed as general in their normative input condition, yet, insofar as they give obligations to some specifc person, they might be viewed as particular in (at least this part of ) their normative output.16 In contrast, consider the speech act of baptizing. What must be true in order for someone to be entitled to baptize something depends on the speaker’s particular relation to the thing to be baptized (roughly, they must be recognized by others as having the right to name the thing in question). In baptizing something, however, the speaker creates for everyone a general obligation to call the baptized thing by its new name.17 These are examples of mixed-element speech acts; for an example of a complex speech act, consider the case of someone marrying two people. As noted above, being entitled to perform a traditional marriage requires a socially recognized status (e.g. historically in Christianity, that of being a priest), so one main consideration that entitles one to perform the speech act is particular. The output-status, in contrast, is a mix of general and particular obligations. The particular obligations include the specifc obligations each member of the married couple acquires vis-à-vis the other, including (traditionally) commitments of monogamy. But, marrying people generates a general obligation for others to treat the married couple as married (for various social and legal reasons, e.g. tax status, deathbed privileges).18 A complex speech act type thus can have diverse types of input and output normative statuses. The precise details of these examples don’t matter here. What’s important is to recognize the strategy of representing speech act kinds with normative functions, whose inputs and outputs can be either general or particular. We now turn back to the case of protests with this more precise normative functionalist conception of speech acts in hand. We think protests can be characterized as complex speech acts with a diversity of input and output-status types, and making these 187

Matthew Chrisman and Graham Hubbs

explicit can help to make the communicative structure of protest clearer than our previous appeal to felicity conditions did. As we noted at the beginning of Section 1, an act of protest involves both i ii

an expression of opposition, and a connected prescription for redress.

Previously, we appealed to the machinery of felicity conditions to explain these two elements. In what follows we want to refne this analysis by using the machinery of normative functionalism. Regarding (i), previously we suggested that a protest might be viewed as claiming, either explicitly or implicitly, that its object is unfair or unjust relative to a presumed overlapping conception of what is unjust or unfair. This is why we suggested that a protest is infelicitous if the object is not the sort of thing that can be unfair or unjust. The normative functionalist approach just sketched provides a way to improve on this idea by arguing that protestors are not (mainly) making claims about what is ruled out by some conception of fairness or justice but rather negatively evaluating the object of protest in a general way. More precisely, our suggestion is that the way protests negatively evaluate the object presumes a kind of entitlement that is general rather than particular. If the object of protest is in fact bad or wrong, then it is bad or wrong in a way that would entitle anyone to express opposition, not just people with some special normative status.19 This is why it is not only the oppressed who can legitimately protest oppression; if oppression entitles one person to express opposition, then it entitles anyone to express opposition. Moreover, the core normative output status of element (i) of a protest is also general. A protest seeks to entitle others to also negatively evaluate the object of protest, and this is an entitlement that does not depend on who in particular those others are. Of course, a protestor may specifcally want their friends and family to agree that the object of protest is wrong, and those who are directly and negatively afected are more likely to appreciate one’s reasons for protesting something, but protests aim for wider persuasion based in general reasons. So, in sum, our idea with respect to (i) is thus to draw on the normative functionalist conception of speech acts above, in order to characterize protest as having a negative evaluative element that is general in its core constitutive input and output normative statuses. It is like assertion in this regard, but we are not claiming that this element is essentially an assertion that something is unfair or unjust; rather, it is a negative evaluation of the object which is supposed to enjoy entitlement based in general reasons for opposing the object of protest and to change general entitlements for other people to similarly evaluate the object of protest without regard to who in particular those other people are. One can protest, then, without making a claim about what’s true relative to a conception of fairness or justice. This makes sense of the possibility that one can protest things one thinks are wrong without even attempting to rely on a presumed overlapping conception of fairness or justice—no such considerations are relevant if one can protests the color of the national team’s jersey or the closing of one’s favorite bar. What is key is that in protesting one presumes a general entitlement to do so and, when successful, one changes general entitlements for others with respect to their evaluations of the object of protest. Let’s turn now to the second element of a protest. Regarding (ii), previously we foated the idea that protests prescribe redress, seeking to impose an obligation on their addressee. This suggests that protests are felicitous only if the protestor has authority to impose obligations, but then we raised the worry that it is puzzling how such authority could be enjoyed by protestors who, in order to protest at all, must lack authority over the people from whom they desire redress. The normative functionalist approach to individuating speech acts also provides a way to improve the analysis of this element of protest. Indeed, we now want to suggest that, in order to have the status to protest, the protester must not have authority, power, etc., over the addressee of the protest (at 188

Protest and Speech Act Theory

least with respect of the topic of the protest). It must be rather the addressee’s prerogative, and not that of the protester, to do what is being prescribed. It is not Pussy Riot’s prerogative to grant Russian citizens greater political freedom or to order Putin to do so. Were it their prerogative, they would not pursue their goals by protesting Putin, but they would instead enact their goals (perhaps by speech act types, such as decrees). Putin, by contrast, cannot protest Pussy Riot’s activities, for he has the authority to respond to any injustice or unfairness of their behavior, not by seeking redress, but by, e.g., directly punishing them. So the input status for the prescriptive element is not general. Although it is not general, it does not require that the protester be subject to the addressee’s power or authority—Australians can march on the Russian embassy in Grifth to protest Putin’s treatment of Pussy Riot. On the output side, the normative statuses which are generated by the prescriptive element of a successful protest are particular, in the sense that they are obligations on particular people. However, the particularities of the people on whom the obligation is imposed are clearly diferent than standard examples of commands. As just noted, the addressee of a protest ought to be someone with authority or control over the prescribed redress to the protest, and a protest assumes they have general rather than particular reasons to give the prescribed redress. This helps us see more clearly why the disgruntled traveler from the previous section is better seen as complaining rather than protesting his unfortunate situation. If the airline employee to whom he is complaining were to put him on the fight, they would have to bump someone else, which the traveler knows. This suggests that whatever reasons there are to give redress in this case are particular to the employee’s relationship with the traveler rather than something general, such as what is based in what’s fair or for the common good. Moreover, if the traveler is simply ignoring the meteorological facts of the situation—imagine him saying, “I don’t care that you can’t take of in this weather—I have to get home tonight, and you have to get me there!”—then the case is one of complaint, and not protest, because there cannot be an appropriate addressee for the protest. The proper addressee of a protest is someone who has some sort of status that allows them to react to and/or afect the object of protest and who is not using or has not used this power in the way that the protester thinks they should. In this last case, the traveler cannot direct his addressee to use their normative powers in a certain way, because the relevant normative powers simply do not exist—no one controls or is responsible for the weather. In this case, then, there is no proper addressee for the speech act, and so there can be no protest.

5 Conclusion Understanding protest is a topic for non-ideal political philosophy, but protests are clearly communicative acts, which makes this also a topic for applied philosophy of language. In this chapter, we have sought to use the resources of speech act theory to better understand the communicative aspects of protest because we think these will be useful for evaluating the justifability of various protests in various non-ideal political circumstances. We pursued our analysis in three stages. First, we distinguished the object, redress, and means of a protest. This provided a way to think of atomic acts of protest as having dual communicative aspects, viz. a negative evaluation of the object and a connected prescription of redress. Second, we used Austin’s notion of a felicity condition to further characterize the dual communicative aspects of protest. This allowed us to distinguish protest from some other speech acts which also involve a negative evaluation of some object and a connected prescription of redress. However, several questions remained. We thus turned to Kukla and Lance’s idea of a normative functionalist analysis of speech acts to advance the view that protests are a complex speech act constituted by dual input normative statuses and dual output normative statuses, with some general elements and some particular elements. On the input side, we have claimed that a protest presumes entitlement to negatively evaluate the 189

Matthew Chrisman and Graham Hubbs

object of protest (where we think this has something to do with general grounds that pertain to everyone and not just the protestor), and a protest presumes entitlement to prescribe redress from the addressee (where we think that has something to do with grounds for imposing the obligation that derives from the wrong claimed in the object of protest rather than from one’s status as having authority over the addressee). On the output side, we have claimed a successful protest generates entitlements and reasons for others to negatively evaluate the object of protest (where we think this has something to do with general grounds that go for everyone or everyone in the audience), and a successful protest generates obligations for the addressee to perform redress (where we think this has something to do with particular grounds that apply only to the addressee of the protest and no one else because of their authority or control over the prescribed redress).

Notes 1 2 In the background of these two elements of the speech act of protest is a third that we will mention here and then suppress for the remainder of this chapter. All protests also have a vocative element, for all protests exhort the attention of some audience to whom the relevant expression of opposition and corresponding prescription of redress will be articulated. Vocatives have been largely ignored in the study of speech acts, but this is starting to change. For a recent philosophical discussion of some of the ethical dimensions of vocative speech acts, see Sanford Goldberg (2020, ch. 2). For recent work on vocatives in linguistics, see Barbara Sonnenhauser and Patrizia Noel Aziz Hanna (2013). In Matthew Chrisman and Graham Hubbs (ms) we discuss the vocative element of protest in more detail, especially as it connects to political riots. 3 See also Mary Lynne Gasaway Hill (2018, ch. 2), who pursues a similar strategy to us of studying protests with the resources of speech act theory and especially by appeal to felicity conditions. 4 Failure to meet these norms is related to the other species of “unhappy” speech acts Austin describes as abuses. For example, a promise made with no intention to do what one promises still counts as a promise, but the speaker abuses the norm of promise making that says one ought to promise to do things only if one intends to do them. We remain unsure how exactly to think about Austin’s category of abuses. There are violations norms whose applicability is partially constitutive of a speech act’s being a particular type of speech act that are nonetheless difcult to conceptualize as abuses. 5 See Michael Rescorla (2009) for discussion of various proponents of this idea. For a more recent discussion of the topic, see Sanford Goldberg (2015). 6 For more on this, see Eliot Michaelson and Andreas Stokke’s contribution to this volume. 7 Following Austin, this diference between warning and persuading is sometimes understood as the difference between illocutionary and perlocutionary acts. In our experience, the terminology is misleading, in that it suggests a binary distinction when it seems more plausible that there is a spectrum between speech acts that require minimal audience uptake and those that require signifcant audience uptake. 8 This form of analysis is inspired by Moore’s Paradox. G. E. Moore argues that there is something absurd in saying “I went to the pictures last Tuesday, but I don’t believe that I did” (Moore 1942, 543). This absurdity, however, does not appear to be that of a straightforward contradiction, so the philosophical task is to locate the source of the absurdity. The structural parallels between Moore’s sentence and the sentences above should be clear. For recent discussions of Moore’s Paradox itself, see Mitchell Green and John Williams (2007) and Jack Woods (2014). For another example of extending the analytic strategy of Moore’s Paradox, see Graham Hubbs (2019). 9 Rawls has a few slightly more general things to say in The Law of Peoples. For example, when discussing the structure of “decent consultation hierarchy,” he says that judges must be willing to listen to and to respond to dissent, and if dissenters do not agree with the judges’ decisions, the dissenters may “renew their protest,” which “is permissible provided it stays within the basic framework of the common good idea of justice” (1999b, 72). These remarks go undeveloped, however; here, we attempt to work out some of the details of a more general account. 10 In an earlier discussion, we say that the object must explicitly concern justice, and we defend this claim by explaining how broadly we understand the concept of justice (Chrisman and Hubbs 2018, 176). This risks inviting confusion. We hope to avoid any such confusion here by spelling out the relevant idea in terms of agent-neutral normative statuses, as we do in the next section.


Protest and Speech Act Theory 11 Indeed, we suspect this is part of the story of why infuential protests have the persuasive power that they do. By presuming an overlapping conception of what is unfair or unjust, a protestor treats that conception as not at issue in making their claims and prescriptions, which is a common rhetorical technique for motivating an audience through nonrational means to agree with something by taking it for granted. See Rae Langton and Caroline West (1999, 309–11) discussing pornography, Stephen Finlay (2014, 184–8) discussing moral ought-claims, and Jason Stanley (2015, ch. 4) discussing propaganda for similar ideas about the persuasive power of various forms of pragmatic presupposition. 12 We wish to remain neutral about the correctness of applying the English word “protest” to these cases, and we don’t see our methodology as one of refning necessary and sufcient conditions for the correct application of this word in light of competent speaker intuitions. Our hope is to identify a recognizable normative phenomenon of which political protests of the sort we mentioned at the beginning of this chapter are paradigms but which might admit of other non-political examples. 13 See especially David Lewis (1979), Robert Brandom (1983, 1994), and John MacFarlane (2011). 14 The notion of entitlement here is practice-internal. There may be reasons of morality, prudence, law, or etiquette that someone shouldn’t assert something that they are entitled, as far as the practice of assertion goes, to assert. However, that doesn’t mean that entitlement conditions have to do only with what people have and will say. For example, one might suggest that the practice of assertion is such that one is entitled to assert only that for which one justifedly believes or knows to be true. 15 The distinction we make here between general and particular statuses is similar to Kukla and Lance’s (2009) distinction between agent-neutral and agent-relative normative statuses. Since the terms “agent-neutral” and “agent-relative” are most commonly applied to reasons, and Kukla and Lance don’t explain how that distinction maps onto normative statuses such as entitlements and obligations, we have adopted an alternative terminology that we hope is less apt to confusion. 16 See Kukla and Lance (2009, 100–102). 17 See Kukla and Lance (2009, 39, 43). 18 See Kukla and Lance (2009, 20-1). 19 There may be some protests, such as Rawlsian civil disobedience, that presume something like membership in a particular constitutional community as a limitation on the entitlement to negatively evaluate the object of protest. This would be a way that the normative statuses changed by the protest would be less than fully general, as they depend on a particular relationship between the speaker and those acquiring new reasons as a result of the protest. Here we are focused on protest writ large, but we suspect a more fne-grained treatment of particularity and generality might provide theoretical resources for distinguishing diferent kinds of protest.

Works Cited Austin, J. L. (1975). How to do things with words. Cambridge, MA: Harvard University Press. Brandom, R. (1983). Asserting. Noûs 17: 637–50. Brandom, R. (1994). Making it explicit. Cambridge, MA: Harvard University Press. Chrisman, M. and Hubbs, G. (2018). Speaking and listening to acts of political dissent. In C. R. Johnson (ed.), Voicing dissent: The ethics and epistemology of making disagreement public. New York: Routledge, 164–81. Chrisman, M. and Hubbs, G. (unpublished). “The language of the unheard”: Rioting as a speech act. Delmas, C. (2018). A duty to resist: When disobedience should be uncivil. New York: Oxford University Press. Finlay, S. (2014). Confusion of tongues: A theory of normative language. New York: Oxford University Press. Gasaway Hill, M. L. (2018). The language of protest: Acts of performance, identity, and legitimacy. Cham, Switzerland: Palgrave-Macmillan. Goldberg, S. (2015). Assertion: On the philosophical importance of assertoric speech. New York: Oxford University Press. Goldberg, S. (2020). Conversational pressure: Normativity in speech exchanges. Cambridge, MA: Harvard University Press. Green, M. and Williams, J. (Eds.). (2007). Moore’s paradox: New essays on belief, rationality, and the frst person. New York: Oxford University Press. Hubbs, G. (2019). Anscombe on how St. Peter intentionally did what he intended not to do. American Catholic Philosophical Quarterly 93: 129–45. Kukla, R. and Lance, M. (2009). Yo! and Lo! The pragmatic topography of the space of reasons. Cambridge, MA: Harvard University Press. Langton, R. and West, C. (1999). Scorekeeping in a pornographic language game. Australasian Journal of Philosophy 77(3): 303–19.


Matthew Chrisman and Graham Hubbs Lewis, D. (1979). Scorekeeping in a language game. Journal of Philosophical Logic 8: 339–59. MacFarlane, J. (2011). What is assertion? In J. Brown and H. Cappelen (eds.), Assertion. Oxford: Oxford University Press, 79–96. Moore, G. E. (1942). A reply to my critics. In P. A. Schlipp (ed.), The philosophy of G. E. Moore Evanston, IL: Northwestern University Press. Rawls, J. (1999a). A theory of justice. Cambridge, MA: Harvard University Press. Rawls, J. (1999b). The law of peoples. Cambridge, MA: Harvard University Press. Rescorla, M. (2009). Assertion and constitutive norms. Philosophy and Phenomenological Research 79: 98–130. Searle, J. (1975). A taxonomy of illocutionary acts. In K. Gunderson (ed.), Language, mind, and knowledge: Minnesota studies in philosophy of science, vol. 15, Minneapolis: University of Minnesota Press, 344–69. Shelby, T. (2016). Dark ghettos: Injustice, dissent, and reform. Cambridge, MA: Harvard University Press. Sonnenhauser, B. and Hanna, P. N. A. (2013). Vocative! Addressing between system and performance. Boston: De Gruyter Mouton. Stanley, J. (2018). How propaganda works. Princeton, NJ: Princeton University Press. Woods, J. (2014). Expressivism and Moore’s paradox. Philosopher’s Imprint 14(5): 1–12.



1 Introduction Successful communication depends not only on our knowledge of language, but also on our knowledge of context. If a speaker utters the sentence “he is going to get burnt,” we will have to rely on our knowledge of the context in order to grasp what proposition they are trying to express. If there is a mutually salient individual in front of us, whose trousers have just caught on fre, then we will know that this salient individual is the intended referent. If we were talking about our mutual friend Frank, and somebody has just asked how Frank’s latest business deal is going, it will be clear that Frank is the intended referent. Two completely diferent propositions are expressed in these situations, and without contextual knowledge, we would not be able to tell which proposition was expressed. Contexts can be fruitfully modeled as bodies of shared information presupposed by the conversational participants (Stalnaker 1973, 1978, 2002). A speaker presupposes a proposition if they take it for granted for the sake of the conversation. We can defne defective and non-defective contexts in these terms (cf. Stalnaker 1978): NON DEFECTIVE CONTEXT: One in which all the presuppositions of the participants are the same. DEFECTIVE CONTEXT: One in which the presuppositions of the participants difer. It is commonly assumed that most communicative interactions take place within non-defective contexts (or contexts which are close enough to being non-defective for it not to matter). Stalnaker (1978) ofers the following line of reasoning in support of this assumption: A defective context will exhibit a kind of instability, and will tend to adjust to the equilibrium position of a non-defective context. Because hearers will interpret the purposes and content of what is said in terms of their own presuppositions, any unnoticed discrepancies between the presuppositions of speakers and addressees is likely to lead to a failure of communication. Since communication is the point of the enterprise, everyone will have a motive to try and keep the presuppositions the same. And because, in the course of a conversation, many clues are dropped about what is presupposed, participants will usually be able to tell that divergences exist if they do. So, it is not unreasonable, I think, to assume that in the normal case contexts are non-defective, or at least close enough to being non-defective. 193

Andrew Peet

A context is CLOSE ENOUGH to being non-defective if the divergences do not afect the issues that actually arise in the conversation. Suppose, for example, that you know that Jones has won the election, believe mistakenly that I know it as well, and are prepared to take the truth of this proposition for granted if the occasion should arise, say by using it as a suppressed premise in an argument, or by using the description the man who won the election to refer to Jones. On my dispositional account of speaker presupposition, if you are prepared to use the proposition in this way, then you DO presuppose that Jones won the election, even if you never have a reason to display this disposition because the subject never comes up. Since I do not know that Jones won the election I do NOT presuppose it, and so the context is defective. But the defect may be harmless. It will not necessarily be harmless: If the news is of sufciently urgent interest, your failure to raise the subject may count as a display of your disposition to take it for granted. There will not exactly be a failure of communication, but there will be a misperception of the situation if I infer from the fact that you do not tell me who won that you do not know either. (Stalnaker 1978, pp. 85–6) If this line of reasoning is sound then defective contexts (or, at least, contexts which are defective in some consequential manner) will constitute deviations from normality. They will be the sort of marginal cases we can harmlessly ignore in much (though not necessarily all) of our linguistic theorizing. After all, communication will typically occur in non-defective contexts. So, if we want to understand how communication functions then these are the contexts on which we should focus. I believe that defective contexts should occupy a more central place in our linguistic theorizing. This is because I reject the claim that defective contexts constitute deviations from normality. I believe that it is fairly standard for our communicative interactions to take place within defective contexts. Thus, to better understand how linguistic communication functions we must understand how it works in defective as well as non-defective contexts. In this chapter I hope to persuade you that defective contexts are more ubiquitous than we typically assume. In doing so, I will draw attention to a number of pressing social and theoretical issues which arise once we start to consider defective contexts. I will proceed by pointing to a number of ways in which defective contexts can emerge without self-correcting in the manner envisioned by Stalnaker. First I will consider situations in which some, but not all interlocutors recognize that the context is defective. I will then consider cases of opaquely defective contexts, and the question of what it is for a context to be “close enough” to being non-defective. I will close by suggesting some further avenues for exploration.

2 Transparently Defective Contexts Stalnaker writes that when one party recognizes a failure of alignment in presuppositions, their presuppositions will be adjusted so as to render the context non-defective. This can happen two ways. First, an audience might make explicit the fact that they do not share the speaker’s presuppositions. The speaker will, if cooperative, respond by removing the ofending item from their presuppositions. Alternatively, the audience may adjust their own set of presuppositions to accommodate the speaker’s presupposition. We will consider each approach, and see that they can (and in some important cases will) fail to render the context non-defective. In the former case, an interlocutor’s attempt to make the defect explicit might fail to be recognized or understood, in which case the defect will remain. In the latter case, we will see that accommodation is not an all or nothing afair. An audience might accommodate p for certain purposes, but not for all the purposes the speaker has in mind. This will 194

Defective Contexts

not only fail to render the context non-defective, but it may also damage the speaker’s epistemic position with respect to the ofending presupposition.

2.1 Hermeneutical Impasses Our ability to fx a defective context by making a defect explicit depends on our ability to make mutually apparent the existence and nature of the defect. That is, if we recognize and reject an agent’s presupposition of p, we must be able to make them recognize that they are presupposing p, that we don’t presuppose p, and that we refuse to accommodate their presupposition. Furthermore, to render the context non-defective, we must be able to do this without introducing further defects. That is, it is no good removing an agent’s presupposition of p if this simply leads them to adopt an alternative but equally problematic presupposition. It is not always easy to make problematic presuppositions mutually apparent in this way. First, not all presuppositions are easily articulated, meaning that attempts to make them explicit can lead to further confusion. Second, certain presuppositions are more central to the presupposer’s understanding of the world than others. These central presuppositions are, in a sense, part of the bedrock against which the agents understand and interpret the rest of the world. When an interlocutor attempts to reject a presupposition their act must be interpreted by their audience. Yet, if the presupposition is central to the audience’s understanding of the world then that presupposition will likely form part of the basis for their attempt to interpret the speaker’s act. The speaker’s attempt to reject the presupposition and fx the context will, thus, be unsuccessful. As an illustrative but unrealistic example, we might consider how difcult it would be for a solipsist to make transparent their rejection of an ordinary (non-philosophical) interlocutor’s presupposition that there is an external world occupied by many minds. This presupposition is so central to the way most of us understand and interpret the world that we cannot be easily shifted from it. A basic presupposition of the way we understand communicative interactions is that they are interactions between minded individuals in a public space. Thus, the solipsist’s attempts to make explicit and remove this presupposition will usually result in failure. Call cases like this “Hermeneutical Impasses” (I take this terminology from Anderson [2017]). The case of the solipsist is hypothetical. It illustrates the possibility and structure of hermeneutical impasses, but it is the sort of case which can be harmlessly ignored in serious theorizing about language. Yet hermeneutical impasses are not uncommon. All they require is for a presupposition (or set of presuppositions) to be central to one interlocutor’s understanding of the world, and yet utterly objectionable from the point of view of another. This kind of situation arises with a worrying frequency. Salient examples arise with respect to the difering assumptions about race which have shaped dialogue between African Americans and white Americans. It is worth considering some such examples. I will start with a much discussed case; the trial of Tom Robinson in Harper Lee’s (1960) To Kill a Mocking Bird. Robinson is a Black man falsely accused of rape in depression-era Alabama. He is in an impossible position. He is facing an all-white jury which harbors the common racial beliefs of the time: that Black people are violent, untrustworthy, and by their very nature inferior to whites. These beliefs are not feeting or easy to shift. They are central to the way in which the jury members understand the social world they inhabit. Without these beliefs they would not be able to live the lives they live; they would not be able to sustain their position of dominance. Their failure to fully grasp the humanity of Black people forms a sort of self-serving ignorance which is essential for the maintenance of the social order from which they beneft (Mills 2007). These beliefs also form the background presuppositions against which they interpret Robinson’s utterances, and assess his credibility as an informant (see Fricker [2007] on this latter point). Pohlhaus Jr. (2012) calls our attention to a particular instance in the text where these presuppositions manifest themselves and bring about misunderstanding: during his cross examination, Robinson is asked why he was 195

Andrew Peet

at his accuser’s home. He tells the jury that he felt sorry for her. Having done so, he immediately realizes his mistake, and shifts uncomfortably in his seat. The notion that a Black man could feel empathy for a white woman runs contrary to the racist background assumptions which form the basis for the jury’s interpretation of his speech. As a result, his utterance is interpreted as suggesting that he felt superior to his white accuser (see Peet [2017] for an account of how such a misinterpretation can arise). Following his utterance, it now becomes a presupposition of the jury, something they take to be common ground, that Robinson felt superior to his accuser. It is hard to see how Robinson could correct this misinterpretation. Even if he somehow managed to make it clear that he didn’t feel this way, the jury still needs some way to interpret his utterance. And the correct interpretation, that he felt empathy for her and understood her situation, is one they are prevented from reaching by the racist presuppositions that are so central to their understanding of the social world.1 In order for the context to be rendered non-defective the jury must be parted from their racist beliefs (an impossible task in this setting). Anderson (2017) draws our attention to a similar situation which has arisen with respect to the Black Lives Matter movement: to be properly understood, the slogan “Black Lives Matter” must be understood against the background presupposition that, at present, Black people in America, unlike white people, are treated as if their lives don’t matter (especially by the police). It has to be understood that white people are in a privileged position and that, at present, Black people often (and rationally) fear for their lives at the hands of the police. This is hard for many white Americans to accept. Even for those who recognize that racial inequality still exists in the US, a full recognition of the extent of such inequality can be hard to grasp and come to terms with. This ignorance is, much like the southern racists’ ignorance, self-serving. It is central to the ability of many white Americans to get on with their lives, and beneft from the social order as it stands. This ignorance leads to misinterpretation. A common response to “Black Lives Matter” has been that “all lives matter,” as if this second claim was inconsistent with that of the Black Lives Matter activists. The suggestion is that Black Lives Matter activists are claiming that only Black Lives Matter, or that Black Lives Matter in some special sense. Such an interpretation only makes sense against the assumption that African Americans are not treated any diferently from white Americans (at least with respect to the value of their lives). Fixing this context by making the defect salient would require forcing those who reply that “all lives matter” to recognize their privileged position relative to Black people, and the realities of the situation of Black people in the United States. Given the role that such ignorance plays, it will not be straight forward to remedy. Hermeneutical impasses don’t arise only when a presupposition is central to an interlocutor’s understanding of the world. Sometimes the concepts required to render explicit the rejection of a presupposition are not available as part of the common conceptual repertoire of the interlocutors. In such cases conceptual engineering may be required. The clearest examples involve hermeneutical injustice: the phenomenon whereby a defciency in our shared conceptual repertoire renders members of marginalized social groups unable to make sense of or articulate to others important social experiences Fricker (2007) presents the example of a woman trying to make sense of and articulate her experience of what we would now call “sexual harassment” before this concept was frst coined. The perpetrator of the harassment would likely have thought of their behavior as firting (or some other socially acceptable form of behavior). The victim rejects this characterization of the interaction. But it is hard to do so convincingly without an alternative concept to fll the gap. Hermeneutical impasses illustrate one clear and important way in which defective contexts can fail to be self-correcting. In these cases, the interlocutors will, at least sometimes, share a mutual interest in communicating. It will be in their interests to render the context non-defective by making the defective presupposition mutually apparent, and removing it from everyone’s representation of the common ground. Yet doing so will be close to impossible. The only other alternative is for the problematic presupposition to be accommodated. But this would, in cases like those 196

Defective Contexts

discussed above, involve giving up too much. Moreover, it is not clear that accommodation will, in practice, always render a context non-defective. This is the theme of the next section.

2.2 Accommodation and Omissive Implicature When interlocutors recognize that their presuppositions do not align they have a number of options. The frst is to object to the presupposition. We have just seen that this won’t always fx the context. Moreover, this is an extreme measure. It will often feel confrontational, and can involve a certain degree of risk. Another alternative is to accommodate the presupposition. That is, when one interlocutor recognizes that another interlocutor is presupposing something they were not previously presupposing, they can render the context non-defective by quietly adding the relevant proposition to their set of presuppositions. This is a common feature of ordinary conversations. If you didn’t know that I had a child, and I stated “I am sleep deprived because my son wakes up and cries every hour or so,” then you will likely add to your set of presuppositions the proposition that I have a son. You will recognize that I was presupposing this, and in order to render the context non-defective you will add this proposition to your set of presuppositions. However, accommodation is not an all or nothing afair. It can be partial or feeting, and when this is the case it can fail to render the context non-defective. Imagine that you have a friend who is a known bull-shitter. They will often make up stories or exaggerate their exploits in order to make themselves appear more interesting than they actually are. You fnd this annoying, but you are willing to let it slide because they have other redeeming virtues. Suppose they state, “when I climbed El Capitan I used a kilo of chalk.” You do not believe that they ever climbed El Capitan, however you do not want to call attention to or explicitly reject this presupposition. Doing so would risk generating open confict, and you do not wish to jeopardize your friendship. You could just accommodate the presupposition, and act, for the rest of the conversation, as if your friend did climb El Capitan. However, there is also a range of intermediate options. First, of course, one could adopt the passive aggressive strategy of making clear one’s rejection of the presupposition by making statements which presuppose or imply (but don’t explicitly state) that it is false. You might, for example, ask about your friend’s progress on a beginner level climb you know they have been struggling with. This serves to make it clear that you reject their presupposition, but it does so without generating open confict. More generally, one can make statements which insinuate the falsity of the presupposition in such a way that one retains plausible deniability about the insinuation. This way, if explicitly challenged, one can deny that one rejects speaker’s presupposition (see Camp [2018] for discussion of insinuation and plausible deniability). This strategy does not involve accommodating the presupposition in any substantive sense, but it simply allows one to avoid explicit rejection. Whether or not this strategy produces a non-defective context will depend on whether one’s rejection of the presupposition is recognized and accommodated by one’s interlocutor. However, the passive aggressive strategy still carries risks. There are better options if one wishes to avoid confict. For example, one might simply change the subject. Doing so would not render the context non-defective, for your friend would continue to presuppose something you do not. They will continue to presuppose that they have climbed El Capitan, whilst you may, unbeknownst to them, be disposed to object to the presupposition were it raised again. However, changing the subject might render the context “close enough” to being non-defective, for it will remove the conversational relevance of the problematic proposition. Similarly, one might accommodate the presupposition to a certain extent, but place a limit on how accommodating one is willing to be. For example, you may adopt a disposition to not object, or make one’s rejection of the presupposition known unless put on the spot, and forced to make a statement which carries with it the relevant presupposition. For example, you may let your friend continue with their tall tales until they ask what you were doing whilst they were on El Capitan, 197

Andrew Peet

at which point you might make your rejection of their presupposition apparent. This would be to act, in some respects, as if the presupposition is true, but it would not amount to full accommodation. One is not disposed to act as if the proposition is true for the sake of conversation. Such partial accommodation will only partially mend the context. But it will be enough to allow for smooth communication, and it will be enough to avoid confict. The above examples are mundane and relatively inconsequential. However, the strategy of partial accommodation, or of attempting to avoid the subject, can be problematic. As Stalnaker hints at the end of the quoted passage, we can communicate things with our silence. It is unlikely that we will communicate much when we fail to call out our lying friend (other than, perhaps, that we don’t disbelieve them). However, sometimes there will be an expectation that, if p were false, we would object to the presupposition or statement of p. This will particularly be the case when somebody presupposes a proposition they know to be controversial or infammatory. If we fail to object, then this can communicate that we share the speaker’s belief in relevant presupposition (loosely following Swanson [2017] I’ll call this “omissive implicature”). Consider the following case from Maitra (2012): An Arab woman is on a subway car crowded with people. An older white man walks up to her, and says “Fucking terrorist, go home. We don’t need your kind here.” He continues speaking in this manner to the woman, who doesn’t respond. He speaks loudly enough that everyone else in the subway car hears his words clearly. All other conversations cease. Many of the passengers turn to look at the speaker, but no one interferes. (Maitra 2012, p. 101) The victim’s silence is unlikely to communicate much. Nobody would expect her to object, as it is likely that she will feel highly intimidated. The silence of the other passengers will communicate somewhat more. It is true that, as Swanson (2017) notes, there will usually be a number of possible explanations for the passengers’ silence. One explanation might be that they share the speaker’s beliefs. But an alternative explanation may be that they too feel threatened into silence, or that they consider the speaker so far below them that they are not worth challenging. Thus, to anybody for whom these possible explanations are salient, the passengers’ silence will not communicate much. However, not all of these explanations will be salient to the speaker. The speaker is unlikely to see themselves as being threatening (at least, not toward the non-Arab passengers), and they are unlikely to see themselves as being unworthy of the other passengers’ engagement. Among the most salient hypothesis for the speaker will be that at least a reasonable number of the other passengers share their beliefs, or at least respect their right to state their hateful views. This will have a number of detrimental efects. First, it will render the context defective. It may be that, if put on the spot, any of the passengers would explicitly reject the speaker’s racist beliefs. In this sense, they are not presupposing that the speaker’s beliefs are acceptable, nor are they presupposing the authority of the speaker to make such statements (contra Langton [2018]). After all, they are not disposed to act as if these statements are either true or acceptable, nor that the speaker has the authority to make such statements. Thus, there will be a failure of alignment in presuppositions between the speaker and audience. Second, and more interestingly, the audience’s omissive implicature may strengthen the speaker’s belief that their racist views are acceptable or correct, and that they have the authority to publicly state them. That we can, in general, rationally strengthen our belief in p by asserting that p is illustrated by the following example from Goldberg (2016): SCIENCE TEAM: Research Team X specializes in a certain subfeld of physical chemistry. The members of X have assembled excellent evidence for their hypothesis H, where (they recognize that) this evidence would be sufcient to convince even the most demanding 198

Defective Contexts

physical chemist of the acceptability of H. Consequently, they go to a prestigious national physical chemistry conference and announce their fndings, concluding that H is acceptable on these grounds. Now, the members of X acknowledge that in such a large audience there will be other people who have diferent background knowledge, diferent evidence, and different expertise. Given (their knowledge of ) the prestige of the conference and the excellence of the audience, as well as the norms of conference behaviour, they reasonably think to themselves that if there had been a faw in their experimentation or an alternative explanation for their data someone in the audience would likely have known of this and would have spoken. So, seeing no one raise any such objection, and observing the audience’s favourable reaction to their announced fndings, they increase their confdence in the thought that there are no relevant pieces of counter-evidence, alternative explanations for their data, or faws in their experimental set-up. (Goldberg 2016, pp. 3–4) Like the scientists, the angry racist is performing their act in front of a large and diverse body of peers. If they were wrong, they may reason, somebody would have objected. In this way the audience’s silence is damaging to the speaker. It provides them with misleading evidence for the acceptability of their views. Perhaps this is not such a plausible hypothesis when applied to the angry racist on the train. Perhaps, in reality, any such individual would have sufcient self-awareness to realize that the other passengers are simply intimidated by them, or do not want to escalate the situation. But more mundane versions of this type of scenario are worryingly common. I imagine many readers will have had the experience of being forced to listen to a taxi driver outline their unhinged views on immigration, or of trying to politely escape a political discussion with an intoxicated bigot at a bar. We may on occasion object in such situations. But many of us will not. Rather, we will try to change the subject, or politely remain silent unless explicitly challenged. In doing so we risk communicating our agreement, and we harm the speaker. The more the speaker’s statements go unchallenged, the stronger the misleading evidence that their statements are correct. A one-of incident may not have a noticeable efect. However, if a speaker fnds themselves in such situations frequently, and they are regularly met with a non-confrontational silence, this could cumulatively lead to signifcant epistemic damage. This section has ended on something of a digression. We have moved from consideration of whether or not defective contexts are typically self-correcting, to a discussion of the harms which can be brought about by failure to fx the context (even when the context is close enough to being non-defective for communication to proceed smoothly). However, the examples are nonetheless illuminating. They illustrate the manner in which Stalnaker’s argument rests on a problematic simplifcation about conversation: the claim that, in conversation, communication is the aim of the game. Clearly communication will typically be one of our primary aims in conversation. However, it is one amongst many. Conversation also serves as a means of bonding, of generating familiarity, as forming a basis for cooperation, or as an opportunity for self-expression. Our desire to, for example, avoid confict, or allow the speaker to feel that we are on good terms, may override our interest in communicating smoothly. The result will often be defective contexts which fail to self-correct. It may be objected that Stalnaker’s argument, charitably understood, is more limited in scope. That is, perhaps we should understand Stalnaker to simply be claiming that in cooperative contexts where communication is the aim of the game, defective contexts will typically be selfcorrecting (at least, where this is possible). This is a far weaker and less interesting claim. After all, very few (if any) of our conversational exchanges will be purely communicative in aim. So there will be little motivation for the claim that we can ignore defective contexts in our linguistic 199

Andrew Peet

theorizing.2 Furthermore, it might be objected that the situations discussed so far are somehow atypical, and that Stalnaker’s claim should be taken only to pertain to typical, or normal contexts. In the following section I will argue that even typical contexts, where communication is one of the primary aims, can be problematically defective without self-correcting.

3 Opaquely Defective Contexts In the previous section we considered what might be called transparently defective contexts: contexts in which at least one of the conversational participants realizes that the context is defective, but is either unable or unwilling to correct it. However, contexts can also be opaquely defective. That is, defects can arise without anybody realizing it. Here I will focus on one way in which opaquely defective contexts might arise: through the potentially approximate and inexact nature of typical linguistic communication. It seems quite plausible that communication is, fairly often at least, approximate and imprecise. Minor and seemingly inconsequential misunderstandings are not out of the ordinary. Much of our speech is loose, and many of us regularly communicate with concepts we only partially understand. As Yalcin (2014) puts a related point, it may be that, typically, “coordination on items of content is a highly approximate, more-or-less afair, with perfect coordination on content not being especially important, and rarely or never happening” (Yalcin 2014, 24). If this is right, then defective contexts will abound: in typical communicative exchanges, one interlocutor’s representation of the common ground will merely approximate the other interlocutors’ representations of the common ground. After all, the proposition the audience adds to their representation of the common ground (i.e., the propositions they immediately entertain and automatically attribute to the speaker through their sub-personal systems of comprehension) will merely approximate the proposition intended by the speaker (i.e., the proposition the speaker will think of themselves as asserting and which they will, as a result, add to their representation of the common ground). However, if the interlocutor’s representations of the common ground approximate each other closely enough, these diferences will usually go unnoticed. There are three questions which are natural to ask when presented with this argument: 1 2


Do we have good reason to believe that communication is typically imprecise in these ways? If communication is typically imprecise in this way, wouldn’t interlocutors take this imprecision into account whilst updating their representations of the common ground, thus rendering the resulting contexts non-defective? Surely, if defective contexts commonly arise from the imprecision typical of linguistic communication, the resultant contexts will still be “close enough” to being non-defective? This seems especially obvious if, as you say, the interlocutors’ representations of the common ground will typically approximate one another closely enough for any diferences to go unnoticed.

In the following subsections I will address each of these questions, in turn.

3.1 Te Imprecision of Communication I won’t attempt to establish beyond doubt that linguistic communication is typically imprecise and approximate in the manner discussed above. To do so would take us beyond the scope of this chapter. Instead, I will merely aim to illustrate the plausibility of the hypothesis through a discussion of the prevalence of conceptual variation: the idea that diferent interlocutors will typically assign slightly diferent concepts to the same words. There are other potential sources 200

Defective Contexts

of approximation and imprecision in communication, such as vagueness, loose talk, and context sensitivity. However, I have discussed many of these issues at length elsewhere, so I will refrain from doing so here.3 I take it to be uncontroversial that diferent interlocutors will often understand the same word in slightly diferent ways. They may have slightly diferent patterns of association, their conceptions may center around slightly diferent prototypes, they may be disposed to make slightly diferent inferences with the same concept, etc. For example, my prototype for “cat” is likely modeled on my own cat, a relatively small, furry animal, fairly passive in behavior (except when provoked by mischievous hands), and very afectionate. These will be among the features which make up my “cat” prototype. Likewise, my familiarity with this cat afects the associations and inferences I will most immediately and automatically make when somebody uses the word “cat.” Your prototype for “cat” will likely be similar. That is, it will list many of the same features. However, for some readers at least, there is likely to be a small degree of divergence. At the very least, these features may be weighted diferently. The same will likely be true of the typical associations and inferences automatically drawn when interpreting utterances using the word “cat.” If factors such as those discussed above determine what a word means in a speaker’s idiolect, then, often, the same word will mean diferent things to diferent interlocutors. However, this latter claim is controversial. Externalists, for example, reject the claim that meaning is determined (solely) by such factors, instead holding that the meaning a word has for a speaker will be determined in large part by their environment. That is, the meaning “cat” has when I use it (in thought or talk) does not just depend on the various associations and inferences which afect my use of the term. For externalists the meaning of “cat” depends on the use of the term “cat” in the wider linguistic community I implicitly defer to when using the term. And these patterns of use will, in turn, be determined by the various patterns of inference and association which underlie and guide the use of “cat” by members of this community. As a result, as long as all the interlocutors are implicitly deferring to the same linguistic community, the meaning of “cat” will be the same for each interlocutor. Thus, when a speaker says “my cat is sleeping” the audience, by implicitly deferring to the same linguistic community as the speaker, will be able to update their representation of the common ground with precisely the same proposition as that intended by the speaker. I am doubtful that this will solve the problem. There are two reasons for this. First, the externalist response depends on the assumption that interlocutors will typically defer to the same linguistic communities. But it is not clear that this assumption is correct. An individual act of deference will always be directed at some linguistic community. There will be some linguistic community it picks out. There must be some story to tell about which linguistic community is picked out by a given act of deference. One natural story to tell here is that the relevant linguistic community will be determined by the speaker’s conception of the community most relevant to their present use of the relevant term. But this leaves space for small idiosyncrasies to emerge in diferent interlocutor’s conceptions of the communities to which they are deferring. Such idiosyncrasies seem inevitable, much as they do with our prototypical conceptions of, say, cats. The result is that interlocutors will typically defer to slightly diferent (although largely overlapping) linguistic communities in typical situations. Tiny variations in the overall patterns of use in each community will determine tiny diferences in meaning assigned by each interlocutor. So, when one speaker says “my cat is sleeping” the proposition each interlocutor updates their representation of the common ground with will be slightly diferent (but extremely similar). Of course, it might be objected that the community deferred to is not determined by the audience’s conception of the relevant community. But then we’re owed a story about what does determine the relevant community. It cannot be deference again, for then we would be led to a regress. Alternatively, it could be that some groups are reference magnets for states of implicit deference. It is not clear to me how such reference magnetism would work. However, it may be possible to tell a story here. 201

Andrew Peet

So I do not wish to put too much weight on this consideration. My point is simply that it is an open (and pressing) question for externalists exactly what determines which precise community is deferred to in a given instance of thought or talk. And whether or not the externalist is able to hold that interlocutors typically coordinate precisely in communication will depend on how they respond to this challenge.4 My second worry regarding the externalist response is that even if externalists are able to guarantee that there is some proposition (the proposition determined by use of the relevant sentence in the relevant community), which is always added to each interlocutor’s representation of the common ground, there is nothing to stop additional propositions determined by each interlocutor’s idiosyncratic uses being added to their individual representations of the common ground, or shaping their understanding of the common ground in problematic ways. To see this, consider Tyler Burge’s classic example of the patient who thinks she has arthritis: Betsy believes she has arthritis in her thigh. She goes to the doctor, and tells her doctor as much, only to be told that arthritis is an infammation of the joints, meaning that one cannot get arthritis in one’s thigh. Betsy takes this on-board, and ceases to believe she has arthritis in her thigh. Prior to her visit to the doctor Betsy had an idiosyncratic understanding of the concept “arthritis.” Even though she deferred in her use of the term in everyday thought and talk, the idiosyncrasies of her prior understanding would have afected her representation of the common ground when she communicated with the term. For instance, if somebody told her, “I’m pain-free other than the arthritis in my lower extremities,” Betsy would, through her implicit deference to her linguistic community, update her representation of the common ground with the proposition that the speaker is only experiencing arthritis based pain in their lower extremities (cf. Burge 1979). However, whilst the speaker would take themselves to be eliminating from the common ground worlds in which they have pain in their thigh, for Betsy these situations would still seem to be live possibilities. In a sense, Betsy’s representation of the common ground will be inconsistent, although not in a way that is transparent to her.5 The arthritis case involves a fairly major idiosyncrasy in understanding. However, smaller idiosyncrasies in understanding (such as those arising from slightly diferent associations and inferences underlying use of the term “cat”) are common. And I see no reason to doubt that they would infuence our representations of the common ground in a similar way (but to a much smaller degree). Indeed, it appears that such diferences in understanding are often manipulated by strategic speakers. If a term is understood by one segment of the population as carrying a particular association, or as giving rise to particular inferences, then a speaker can use this term in order to communicate something to that segment of the population that may be missed by others. This allows the speaker to maintain plausible deniability regarding the hidden message. Such terms are referred to as “dogwhistles” or “code words” (see Mendelberg 2001; White 2007; Stanley 2015; Khoo 2017; Saul 2018). For example, suppose a politician wishes to target voters with underlying racial prejudices, but does not want to be seen as overtly racist. They might highlight the fact that their opponent’s proposed welfare policies disproportionately target “inner city families,” hoping that their intended audience (racially resentful white voters) will understand them as communicating that Black families will beneft disproportionately from their opponent’s policies. There is much to be said about code words, but I will not discuss them further here since they have a chapter of their own in this volume. The important point for our purposes is that it is unclear how they could have their powerful efects if everybody understood them in the same way. That is, if everybody understood code words in the same way it is not clear that speakers could use them as a means to maintain plausible deniability. Thus, I conclude that the thesis of conceptual variation is plausible, even on externalist views of meaning. Furthermore, it seems clear that if the thesis of conceptual variation is correct, linguistic communication will typically be approximate (to varying degrees).


Defective Contexts

3.2 Epistemic Responsibility When Communicating Approximately At this point it is natural to object that if communication is typically approximate in the way I have suggested, then surely interlocutors will take this into account when updating their representations of the common ground. If I know that your use of the word “cat” is likely to difer slightly from my own, but I don’t know how, then it would be epistemically irresponsible of me to update my representation of the common ground in accordance with my own specifc use of the word “cat.” Rather, I should update the context in a more coarse-grained way. I should eliminate fewer possibilities from my conception of the common ground than I would otherwise. In doing so I am able to prevent the context from becoming defective. There are a number of problems with this objection: 1



The above objection assumes that language users are standardly aware of the fact that we typically use the same words in ever so slightly idiosyncratic ways. However, it is far from clear that this is the case. Ultimately this is an empirical question. But I am highly doubtful that most unrefective language users ever think about this, and even more doubtful that it efects their linguistic practice. Although forming a more coarse-grained belief may reduce the risk of forming a false testimonial belief, it is unlikely to render one’s context non-defective.6 In order for the context to be rendered non-defective the speaker and hearer must each adjust their representations of the common ground in precisely the same way. Suppose we have two interlocutors, Ashima and Kai. Ashima asserts the sentence S. For Ashima, S expresses the proposition p1. For Kai it expresses the proposition p2 . But they are each aware of the possibility of divergence in meaning. So neither of them updates their representation of the common ground with the precise proposition they associate with S. Rather, they each adjust with a more coarse-grained proposition (i.e., a proposition which eliminates fewer worlds from the set of open possibilities). In order for this to work, they must update the context with exactly the same proposition, which means they must each make slightly diferent deviations from their standard use of the term. It is hard to see how, without knowing the precise idiosyncrasies of one-another’s use of S, they could reliably converge on the same proposition with which to update their representations of the common ground.7 Even if it was possible for interlocutors to adjust their updates to the common ground in such a way that they converge on precisely the same meanings, this would be highly impractical. Communication moves quickly. Interpretation is largely unrefective. If communication required that we refect precisely on the ways in which our uses of various bits of language may difer, and adjust our representations of the common ground accordingly, then communication would be massively cognitively demanding. This would be highly impractical. As long as the interlocutors coordinate approximately, it will not be worth the extra expenditure of cognitive resources to secure precise coordination. Thus, even if such adjustments were possible in principle, it is unlikely that they typically occur in practice.

For these reasons, I fnd the epistemic responsibility concern unpersuasive. However, my third response to this objection lends further force to the fnal potential worry I highlighted in the opening of §3. Surely, to the extent that communication is typically approximate, the contexts which result are close enough to being non-defective for it not to matter. After all, if they were not close enough to being non-defective it would be worthwhile to expend the additional cognitive resources necessary to render them close enough to being non-defective. I will consider this objection in the fnal section.


Andrew Peet

3.3 “Close Enough” Communication In the previous sections I have suggested that typical communicative contexts will be defective; diferent interlocutors will typically have slightly diferent representations of the common ground. However, Stalnaker never claimed that typical communicative contexts would be non-defective. Rather, he claimed that they would typically either be non-defective, or close enough to being non-defective for it not to matter. It is natural to feel that the considerations I have adduced at best suggest that contexts will rarely be completely non-defective, but will nonetheless be close enough to being non-defective for it not to matter. This line of thought is further buttressed by two considerations I appealed to in the previous sections. First, to the extent that typical interlocutors’ representations of the common ground differ, they will usually approximate one another well enough for any diferences to go unnoticed. Secondly, the cognitive capacities which underlie our ability to communicate, and the social and epistemic norms that govern and guide our communicative practices, are adaptive. They have developed to allow us to achieve our communicative ends. If typical communicative contexts were not close enough to being non-defective for it not to matter, then our norms and capacities would surely adapt in such a way as to ensure that we do typically coordinate well enough for any divergences in meaning to be inconsequential (call this “the adaptiveness argument”). There is a great deal to be said for this line of reasoning. However, it forces us to ask what it really amounts to for a context to be close enough to being non-defective for it not to matter. This question is not easily answered. One might of course claim that a context is close enough to being non-defective if the interlocutors’ representations of the common ground are close enough to allow for successful communication. This was hinted at in §2.2 when I noted that Stalnaker’s remarks on defective contexts may best be taken as concerning contexts where communication is the name of the game. But this just pushes the problem back a step: what is successful communication? If successful communication requires precise coordination on meaning or content, then contexts typically will not be sufciently non-defective for it not to matter. We can make progress by considering the core idea behind the adaptiveness argument: linguistic communication has important functions. There are many important things it allows us to do. The various mechanisms and norms which underlie our communicative practices have evolved and adapted to allow us to do these things smoothly and successfully. That is, they have evolved to allow us to successfully achieve our communicative ends. So, communicative contexts will typically be close enough to being non-defective for us to be able to successfully achieve our communicative ends. The problem we now run into is that linguistic communication has many diferent functions. And these apparent functions will typically require diferent levels (and potentially diferent forms) of coordination on meaning and content.8 For example, in communicating with others we attempt to coordinate our actions and plans so that we can work together to achieve ends of mutual interest, we acquire true beliefs (and even knowledge) from others, we express our feelings and emotions to others, we establish power relations and shape others’ cognitive worlds in ways which give us an advantage over them, we agree and disagree, we consent, and we issue commands. Some of these functions will no doubt be more central to the way our norms and linguistic abilities have developed than others. With respect to those functions which are more peripheral to the pressures and circumstances which gave rise to our linguistic norms, it is an open question whether typical communicative contexts really are sufciently close to being non-defective for these functions to be reliably achieved. For many of these functions to be reliably achieved it may not even be required that audiences typically form true beliefs via communication. As long as the beliefs they form are close enough to the truth to have the same behavioral consequences in a wide enough range of circumstances it is not clear that there would be sufcient pressure to force further adaptation. Furthermore, even if 204

Defective Contexts

the acquisition of true beliefs is a central function of linguistic communication (as I am inclined to think it is), it is not clear that our communicative practices would be sufciently attuned to reliably support the acquisition of knowledge. Truth and knowledge have the same behavioral consequences. Yet it might be thought that the reliable acquisition of knowledge would place considerably higher demands on the systems of norms and cognitive capacities which underlie our communicative practices than the acquisition of true belief. For example, knowledge has a modal component: it requires that we get it right not only in the actual world, but also all nearby worlds. If our communicative practices developed in such a way that for instance, we acquired true beliefs 90% of the time, approximately true beliefs 9% of the time, and dangerously false beliefs 1% of the time, they would have high statistical reliability. However, this would not necessarily secure modal reliability. It would be consistent with such a situation that when we form true beliefs there will usually be a small set of nearby worlds in which we acquire only approximately true beliefs. If this was the case, then typical communicative contexts would be sufciently non-defective to support the reliable spread of true belief, but not knowledge.9 Of course, this is purely speculative. But it serves to illustrate that just because communicative contexts are typically close enough to being non-defective to support the central functions of linguistic communication, we should not conclude that we can safely ignore defective contexts in our philosophical and linguistic theorizing. Indeed, I hope that throughout this chapter it has become clearer to the reader just how important it is to be cognizant in our theorizing of the various different ways in which communicative contexts can be rendered defective, and the important impact such defects can have. Defective contexts are ubiquitous. They are not typically self-correcting. And they have signifcant potential to undermine our communicative aims in various ways.

4 Conclusion This chapter has merely scratched the surface, and I have raised more questions than I have answered. I hope that future research on defective contexts will answer some of the questions I have raised. However, the scope of this chapter has been very narrow. There are many issues for future work on defective contexts to contend with which have not been broached here. I would suggest the following as promising additional avenues for future research in this area: 1



In this chapter I have focused on defective contexts defned as those in which the presuppositions of the interlocutors do not coincide. But there are sophisticated alternatives to presuppositional approaches to context. For example, linguists such as Cragie Roberts (2004, 2012), and philosophers such as Schoubye and Stokke (2016) see contexts as being structured by the questions under discussion. Such approaches to context have potential to shed light on contextual defciencies missed by focusing only on contexts as sets of presuppositions. I have primarily focused on defects which negatively afect our ability to spread information. But, as mentioned at various points above, conversation has many diferent aims. It is conceivable that contexts in which information transfer occurs smoothly may nonetheless be defective along other dimensions. For example, a discourse could be framed in terms of concepts, which are ofensive or otherwise harmful to some of the interlocutors. We need to have a better grasp of the various aims of communication in order to fully understand the various ways in which contexts, and communication more generally, can be defective. I have primarily focused on face to face communication between individuals. However, our means of communication are rapidly expanding. And with new forms of communication come new ways for communication to go wrong. We must be cognizant of the ways in which developing communication technologies (such as new forms of social media) constitute defective platforms for reliable (or otherwise unproblematic) communication. 205

Andrew Peet

Notes 1 Imagine, for example, that he simply states “I don’t mean that I felt superior to her.” This would force the audience to identify a satisfactory re-interpretation of his original utterance which doesn’t carry the implication of felt superiority. That is, it would require that they conceive of a Black person as being able to feel sorry for a white woman without felt superiority. But if this possibility is in tension with their central presuppositions then no such interpretation will be forthcoming, and his attempted clarifcation will come across as disingenuous. 2 To be clear, Stalnaker himself never claims that we can largely ignore defective contexts in our theorizing. This is simply something that follows fairly naturally from the stronger reading of his argument. So this more charitable construal of his argument may well be more accurate. This is not of great consequence here. I take the stronger version of the argument to be prima facie compelling regardless of whether or not it is endorsed by Stalnaker. And it is the stronger version of the argument that might tempt us to sideline defective contexts in our theorizing. So it is the stronger version of the argument that is my focus. 3 See Peet (2015, 2016, 2018, and 2019). See Davies (2019a, 2019b) for responses to some of these considerations. 4 This challenge becomes especially pressing when we consider items of language other than, say, natural kind terms. For natural kind terms the relevant community may be experts regarding the subject matter at hand. Such communities may have epistemic properties which make them reference magnets for states of deference. It is less clear what community would be reference magnetic for uses of words like “bottle” or “chair,” and how such reference magnetism would actually work. Thanks to Sandy Goldberg and Joey Pollock for discussion here. 5 It might be held that the context in this situation is non-defective, since each interlocutor’s representation of the common ground is identical, but their conceptions of (or understandings of ) the common ground difer. Strictly speaking this might be correct. However, there is a clear sense in which such contexts are still defective. Coordination on content is not the be-all and end-all of communication. Mutual understanding matters as well (Pollock 2015). 6 For discussion of the “coarse-grained proposition” approach to testimonial belief in cases of inexact communication arising from context sensitivity, see Peet (2016) and Davies (2019b). 7 It might be thought that vagueness can salvage the situation. That is, Kai and Ashima each update their representations of the common ground in a vague way, and thus avoid updating their respective representations of the common ground with diferent propositions. This does not solve the problem, for we are still left with the question of how they are to update their representations of the common ground in the same vague way. Indeed, on some views of vagueness (such as many valued approaches) introducing vagueness may make coordination even more miraculous (MacFarlane 2016). 8 For two recent forms of pluralism about communicative success, see Abreu Zavaleta (2018), and Pollock (Forthcoming). 9 How pressing this worry is will depend on how demanding one takes the conditions on knowledge-yielding communication to be. In Peet (2019) I argue that the conditions for knowledge-yielding communication are actually fairly undemanding. However, many previous authors have taken the conditions on knowledge-yielding communication to be signifcantly more demanding (for example, Heck 1995; Goldberg 2007). If they are correct, then this worry may be pressing.

Papers Cited Abreu Zavaleta, M. (2018). Semantic Variance. PhD Thesis. New York University. Anderson, L. (2017). Hermeneutical Impasses. Philosophical Topics, 45 (2), 1–19. Burge, T. (1979). Individualism and the Mental. Midwest Studies in Philosophy, 4, 73–121. Camp, E. (2018). Insinuation, Common Ground, and Conversational Record. In D. Fogal, D. Harris, and M. Moss (eds.) New Work in Speech Acts (pp. 40–67). Oxford: Oxford University Press. Davies, A. (2019a). Testimony, Recovery, and Plausible Deniability: A Response to Peet. Episteme, 16 (1), 18–38. Davies, A. (2019b). Testimonial Knowledge and Context Sensitivity: A New Diagnosis of the Threat. Acta Analytica, 34 (1), 53–69. Fricker, M. (2007). Epistemic Injustice: Power and the Ethics of Knowing. Oxford: Oxford University Press. Goldberg, S. (2007). Anti-Individualism: Mind and Language, Knowledge and Justifcation. New York: Cambridge University Press.


Defective Contexts Goldberg, S. (2016). Can Asserting that p Improve the Speaker’s Epistemic Position (And Is That a Good Thing). Australasian Journal of Philosophy, 95 (1), 157–70. Heck, R. (1995). The Sense of Communication. Mind, 104 (413), 79–106. Khoo, J. 2017. Code Words in Political Discourse. Philosophical Topics, 45 (2), 33–64. Langton, R. (2018). The Authority of Hate Speech. In J. Gardner, L. Green, and B. Leiter (eds.) Oxford Studies in Philosophy of Law, Volume 3 (pp. 123–152). Oxford: Oxford University Press. Lee, H. (1960). To Kill a MockingBird. New York: J.B. Lippincott & Co. MacFarlane, J. (2016). Vagueness as Indecision. Proceedings of the Aristotelian Society, 90, 255–83. Maitra, I. (2012). Subordinating Speech. In I. Maitra and M.K. McGowan (eds.) Speech and Harm: Controversies over Free Speech (pp. 94–121). Oxford: Oxford University Press. Mendelberg, T. (2001). The Race Card: Campaign Strategy, Implicit Messages, and the Norm of Equality. Princeton, NJ: Princeton University Press. Mills, C. (2007). White Ignorance. In S. Sullivan and N. Tuana (eds.) Race and Epistemologies of Ignorance (pp. 11–39). New York: SUNY Press. Peet, A. (2015). Testimony, Pragmatics, and Plausible Deniability. Episteme, 12 (1), 29–51. Peet, A. (2016). Testimony and the Epistemic Uncertainty of Interpretation. Philosophical Studies, 173 (2), 395–416. Peet, A. (2017). Epistemic Injustice in Utterance Interpretation. Synthese, 194 (9), 3421–43. Peet, A. (2018). Testimonial Knowledge without Knowledge of What Is Said. Pacifc Philosophical Quarterly, 99 (1), 65–81. Peet, A. (2019). Knowledge-Yielding Communication. Philosophical Studies, 176 (12), 3303–27. Pohlhaus Jr, G. (2012). Relational Knowing and Epistemic Injustice: Toward a Theory of Wilful Hermeneutical Ignorance. Hypatita, 27 (4), 715–35. Pollock, J. (2015). Social Externalism and the Problem of Communication. Philosophical Studies, 172 (12), 3229–51. Pollock, J. (2020). Context and Communicative Success. In T. Ciecierski and P. Grabarczyk (eds.) The Architecture of Context and Context-Sensitivity (pp. 245–263). Switzerland: Studies in Linguistics and Philosophy, Springer. Roberts, C. (2004). Context in Dynamic Interpretation. In L.R. Horn, and G. Ward (eds.) The Handbook of Pragmatics (pp. 197–220). Oxford: Blackwell Publishing. Roberts, C. (2012). Information Structure in Discourse: Toward an Integrated Formal Theory of Pragmatics. Semantics and Pragmatics, 49 (5), 1–69. Saul, J. (2018). Dogwhistles, Political Manipulation, and Philosophy of Language. In D. Fogal, D. Harris, and M. Moss (eds.) New Work on Speech Acts (pp. 360–83). New York, NY: Oxford University Press. Schoubye, A. and Stokke, A. (2016). What Is Said? Noûs, 50 (4), 759–93. Stalnaker, R. (1973). Presuppositions. Journal of Philosophical Logic, 4, 447–57. Stalnaker, R. (1978). Assertion. Syntax and Semantics, 9, 315–32. Stalnaker, R. (2002). Common Ground. Linguistics and Philosophy, 25 (5–6), 701–21. Stanley, J. (2015). How Propaganda Works. Princeton, NJ: Princeton University Press. Swanson, E. (2017). Omissive Implicature. Philosophical Topics, 45 (2), 117–37. White, I. (2007). When Race Matters and When It Doesn’t: Racial Group Diferences in Response to Racial Cues. American Political Science Review, 101(2), 339–54. Yalcin, S. (2014). Semantics and Metasemantics in the Context of Generative Grammar. In A. Burgess and B. Sherman (eds.) Metasemantics: New Essays on the Foundations of Meaning (pp. 17–55). Oxford: Oxford University Press.



Linguistic Harms


This chapter aims to clarify diferent notions of pejoratives and pejoration and to taxonomize certain classes of pejorative expressions, with special focus on pejorative nouns for persons and groups of persons, especially slurs. It aims to highlight the rich variety of phenomena and to encourage deeper investigations into the relationships between varieties of pejoratives. Trigger warning: In what follows, I mention numerous deeply derogatory expressions, including slurs. Understandably, many may wish to stop here. Nevertheless, while various projects on deeply derogatory expressions can be executed without such mentions, a comparative analysis of pejoratives like the kind I undertake here cannot. And there is also value in doing so: seeing and hearing these words encourages our viscerally feeling how loathsome they are. Understanding these expressions – their linguistic properties, individual connections to cognition, social structures, ideologies, and oppression – requires talking about them, just as understanding rape, slavery, and genocide demands opening our eyes to horrors it is natural to wish to run from.

1 Pejorative Lexical Items, Pejorative Uses, Pejorative Speech Acts In speaking about language and communication, ‘pejorative’ is used in a number of ways. While I will not cover all of them,1 I distinguish and focus on three notions: (i) Pejorative lexical items, (ii) Pejorative uses of an expression, and (iii) Pejorative speech acts. Pejorative lexical items include words and sufxes that have, or may be added to words to create, negatively valenced meaning or connotations. The range of pejorative lexical items is vast and diverse and complicated. Among the most prominent are pejorative nouns used to designate people and groups of people. We have: pejorative nicknames (Shifty for Congressman Adam Schif; Drumf for Donald Trump), all-purpose pejoratives for individuals (asshole, jerk), and pejoratives for individuals that specify a basis for the disparagement, including their behavior (brat); character (blowhard, sellout); physique (bean pole), occupation (bean-counter, pig); and persons’ situations, class, gender, roles, and struggles (hobo, lowlife, breeder, hag, crackhead, spinster). Some pejoratives are or are derived from so-called thick terms, expressions that mix categorization with an evaluation (coward, brute), while others receive their pejorative character from societal disdain for what’s described (liar, rapist, murderer, racist). We even have pejorative words to put down the good, like the old-fashioned goody two-shoes, for those (deemed to be) excessively striving toward conventional virtue and our contemporary do-gooder and justice warrior, both political pejoratives, used by the right for those (deemed) overly moralizing lefties obsessively promoting policies of racial justice.2 211

Robin Jeshion

An important class of pejorative nouns for groups of people include a wide range of deeply derogatory terms. Even these expressions do not naturally form a single univocal class. They include a diverse range, including canonical slurs (Kike, Spic, faggot), slurs that invoke descriptors (curry-muncher for South Asians, Christ Killer for Jews), gendered slurs (slut, wuss), and explicitly stereotyping expressions (Uncle Tom, Jewish American Princess, Karen).3 In addition to individual words, many languages contain sufxes that combine with bases to generate pejorative meanings. For instance, the English sufx -aster, as in poetaster and philosophaster, indicates inferior, would-be poets and philosophers, respectively. Via increasing prevalence of portmanteau, the sufx -tard has recently become a productive derogatory sufx – libtard, republitard, Trumptard, fucktard.4 Pejorative lexical items can also be formed from combining individual words. Nouns combined with certain pejorative adjectives and expressives generate complex expressions like dirty Jew, stinking Chinese, goddamn liberal that derogate in many of the ways that slurs do.5 There are other systematic linguistic devices by which speakers belittle, mock, derogate, and so forth. Our capacities to communicate specifc emotions via language, both intentionally and sub-personally, have been richly researched within psychology for over a century (Scherer, 2003). One central fnding is that various prosodic efects like tone of voice, pitch, and sound-duration convey a plethora of distinct attitudes, including disgust and contempt that are discernable by interlocutors. Speaking with minimal changes in pitch or with an exaggerated lengthening in sound-duration, for instance, has been shown to indicate speaker-disgust (Scherer, 2013; Sendlmeier, Stefen, & Bartels, 2016). Caustic sing-song-voice is a conventional form of mockery.6 Contemptuous intonation placed on specifc words – for instance, Chinese, Jew, liberal – forms another conventional way to derogate.7 Other pejorative lexical items do not designate persons. Like those that do, they take numerous forms, including proper names, like La La Land for Los Angeles, and general terms, such as fyover states, for the interior continental US states (deemed) too boring to visit, and death tax for the estate tax, both politicized (Nunberg, 2018). For many, the pejoration is productive and exploits mechanisms piggybacking on pejoratives for people or groups: chick-fick for movies (supposed to) appeal to women, and evaluated as fuf; Yuppie fu used to designate chronic fatigue syndrome while conveying suspicion of its reality and that those claiming its afiction are hyper-sensitive lightweights needing to man-up; guido haircut and dyke jeans; and Hymietown for New York City, all constructed from slurs. Some pejorative adjectives and verbs are formed similarly – bitchy, jew-down, gyp me8 – while others, like uppity, when used by whites to describe Blacks who ‘don’t know their place’, are drenched in racism without incorporating slurs at all. The reverse occurs too. Ok, Boomer, a retort used to dismiss older folks as helplessly out of touch gave birth to the ageist pejorative Boomer, from which in turn we now have the morbid slang term Boomer Remover for the coronavirus.9 All of these pejorative lexical items possess typical uses that convey something negative: a negative attitude, evaluation, or stance. Nevertheless, as we’ll see, they may defy a common linguistic analysis of how their negative polarity is encoded or conveyed, and on what exactly individuals or groups the pejoration attacks. What does unite them all, however, is that they are the tools – for some of them, the weapons – of pejorative utterances insofar as they are commonly used to degrade, diminish, insult, mock, and so forth, often in highly specifc ways. Of course, people put others down without using any pejorative expressions at all, and people use pejorative lexical items without any intentional or unintentional derogating, or without causing any ofense. Pejorative lexical items may be used to praise or convey solidarity. We need, therefore, to distinguish two further notions: a pejorative use of an expression and a pejorative speech act. I will say that within an utterance, certain words are used pejoratively within that context, regardless of whether any of the words themselves are pejorative expressions. Sincerely calling someone libtard involves a pejorative word used pejoratively. Yet boy and girl, which are not themselves 212

Varieties of Pejoratives

pejorative lexical items, are systematically used as pejoratives, as in calling a Black man boy as a means of derogating and exerting racial social dominance, and calling a schoolboy girl to belittle, feminize, or otherwise mark him as weak. Similarly, metaphoric uses of expressions like weasel (for a sneak), rat (squealer), granola (hippie), and clown (fool) are notoriously efective devices of insult (Camp, 2017). And Trump’s sarcastic tweeted reference to Obama as our great African-American President is a pejorative use of a laudative. Quite generally, since sarcasm is often a polarity reversal switch – think of sarcastic Genius! – almost any non-pejorative lexical item can have pejorative uses. Additionally, when canonical pejorative lexical items like slurs are used by their targets, the terms themselves are typically not being used as pejoratives (though they may secondarily carry an ironic echo of their standard derogation). The same would apply to one-of or localized uses of pejorative nicknames. A friend and political ally of Congressman Adam Schif could jokingly call him Shifty and, as in the slur case, would be using that pejorative lexical item non-pejoratively, as a way to convey solidarity. Finally, we need to distinguish pejorative speech acts, speech acts that ridicule, mock, harass, humiliate, belittle, subordinate, stigmatize, marginalize, degrade, dismiss, insult, derogate, or dehumanize. In the political realm, while frequently performed by the dominant to oppress, pejorative speech acts may also be used by the subordinated to ‘punch up’ or back, or more generally, as a form of resistance. Pejorative speech acts need not involve the utterance of any pejorative lexical items or any words used pejoratively at all. When addressed to an intended target, Oscar Wilde’s quip “Some cause happiness wherever they go; others, whenever they go” insults, but not in virtue of any systematic pejorative content or associations in any of its constituent words. Trump’s comment that the journalist Megyn Kelly had ‘blood coming out of her wherever’ aims to induce disgust, and thus humiliate and dismiss Kelly. While its efect trades on and communicates the same misogyny as that of many gendered slurs, it does so without invoking any pejorative words or uses of words. And of course, pejorative speech acts may be wordless: intentional expressive facial gestures of contempt, coded symbolic hand gestures like a twisting of your forefnger against your temple to indicate someone is ‘loco’, and a refusal to shake someone’s hand may all ridicule, insult, or subordinate, doing so with the same specifcity of derogatives. As Desmond Morris (1978, 186) noted long ago, ‘almost any action can operate as an Insult Signal’, so long as it is performed ‘in the wrong time or in the wrong place’. There are a wide variety of pejorative speech acts that merit special analysis and that are increasingly receiving philosophical attention. Amongst these are the intentional political uses of code words that may be innocent-sounding yet impart and evoke racial bias, like inner city and Trump’s calling the coronavirus the Kung Flu (Mendelberg, 2001; Stanley, 2015; Khoo, 2017; Swanson, 2020); intentional dogwhistles that subordinate groups by covertly infaming bias and bigotry (Saul, 2018)10; speech acts that subordinate and silence, like Senator Mitch McConnell’s leading the Senate in silencing Elizabeth Warren from reading a Coretta Scott King letter during the 2017 Jef Sessions confrmation hearing (Langton, 1993, 2012; Langton & West, 1999; Maitra, 2012; McGowan, 2012, 2019; Tirrell, 2018); speech acts that are dismissive, like ending a conversation with a ‘say it to the hand’, and acts that use purported ignorance to undermine and discredit speakers (Cull, 2019); misgendering, the intentional calling or referencing of someone with a pronoun or classifer (man, woman) with which they don’t identify or have disavowed; and deadnaming, the intentional calling or referencing of someone by their previous name, as in calling a transwoman with their male birth-name (Conrod, 2017; Davis & McCready, 2020). My focus in this article is on pejorative speech acts performed with pejorative lexical items used pejoratively. Many such pejorative speech acts are exercises of power, privilege, and authority. A full analysis of them must explicate how they reinforce power structures and oppress, and how some uses fght back, and help fatten hierarchies. Similarly, part of the increasing interest in pejoratives and pejorative speech acts concerns the relationship of some of them to ideologies. While jerk is ideology-neutral, spinster is decidedly not; while Wilde’s remark insults sans ideology, Trump’s sneer 213

Robin Jeshion

to Kelly completely rests on systemic ideologies pathologizing women’s bodies for its misogynistic force. But I leave exploration of these vital matters to the side – not because they are not absolutely central, for they are, but because here my aim is to lay a taxonomic linguistic foundation. Before starting of, a brief note on polysemy and loose talk. Many words are polysemous: they have multiple conventional meanings and associations which are related to one another. For instance, lamb is used to refer to both the animal and its meat. Consequently, a word’s meaning within an utterance is highly sensitive to context. Most pejoratives are highly polysemous. Faggot has a conventional pejorative meaning on which it is a tool for dehumanizing gay men, has another on which it is an all-purpose put down, much like loser, and is also used positively as a playful term of endearment amongst gay men. The same goes for loose talk, whereby we speak fguratively, or broaden or narrow the ranges of application of a term, sometimes on the fy. My cat Willow plays fetch. To highlight her unusual play, I speak fguratively, calling her a dog. Likewise, I speak loosely when I describe my friend’s long-standing partner as his husband, even though they have never ofcially tied the knot. While a full theory needs to account for fgurative and loose pejorative talk, as well as pejorative polysemy, I focus here primarily on pejorative expressions with their (possibly numerous) pejorative meanings.

2 Pejorative Nicknames Pejorative nicknames are enduring tools with which rivals taunt each other, schoolyard bullies mock the scrawny, the grubby, and the weak, and online trolls smear and ridicule their nemeses. Especially in very recent years, they loom large in the polarized American political landscape. The right ridiculed Barack Obama with Obozo, Oblowme, and Barack Hussain Obama, while, for Donald Trump, the left dishes up Drumf, tRump, and Trumplethinskin. For any competitor or threat, Trump himself famously brandishes nicknames including Crooked Hilary for Hilary Clinton, Low IQ Maxine for Congresswoman Maxine Waters, Shifty for Congressman Adam Schif, and Pocahontas for Senator Elizabeth Warren. And the class of pejorative nicknames extends beyond names for persons, including terms for places, like Bezerkely for Berkeley and political policies, Obamacare, which, though later reclaimed, started its life as a pejorative nickname. Surprisingly, pejorative nicknames are almost entirely neglected from the burgeoning philosophical literature on social-political pejoratives.11 This is a rich area calling out for development. Here I only sketch enough of their properties to illuminate their usefulness in understanding slurs. Linguistically, pejorative nicknames should be classifed alongside other nicknames. Kennedy (2015) ofers a fruitful defnition on which nicknames are proper nouns that difer in structure from the ‘formal names associated with an individual’, yet share the same reference as the formal name. Diferences in structure from the formal name may be minimal, governed by rules like the shortening of Shaq for Shaquille O’Neal, or the use of initials like RBG for Ruth Bader Ginsberg. Or they may be more extensive, like the laudatives Magic for Earvin Johnson and Notorious RBG for Bader Ginsberg. In each, the referent of the nickname is the same as that of the formal name. Nicknames have an extra-semantic component that difers from that of the formal name. It may be descriptive, adding supplementary content that colors perception of the referent, or it may be sociolinguistic, revealing features of the relationship between the coiner and users of the nickname and the nickname’s referent. Often it is both. For example, although it contains no descriptive content, Shaq indicates informality, and thus conveys familiarity, or desired closeness, between users and referent. It also follows the common pattern of shortening (Wierzbicka, 1992). By contrast, Magic supplies descriptive content associated with Johnson’s stellar basketball skills. While nicknames with descriptive contents include all kinds of content, it is often rooted in ‘metonymic associations with the referent, alluding to their physical or behavioral traits or to their characteristic activities’ (Kennedy, 2015, 38.6). Whatever the content, it is always extra-referential 214

Varieties of Pejoratives

and supplementary: the nickname’s referent is not determined satisfactionally, by whosoever ‘fts’ the description. Further, failing to satisfy a description in no way impinges on who or what the nickname refers to. Like honorifcs, terms of endearment, and other social deictics like the French tu/vous, nicknames frequently signify social relationships, especially familiarity and hierarchy, and thus refect social norms and power relationships and diferentials. They often reveal speaker’s evaluative and afective stances toward the nicknames’ referents. Similarly, speakers use nicknames to plaster social identities onto their referents, and with such use, speakers often also highlight aspects of their own social identity. For instance, the coining of Notorious RBG and its proliferation contributed to fashioning Bader Ginsberg’s social identity as a powerful badass defender of justice and, eventually, as feminist icon. Users of the laudative in turn revealed their respect and admiration for her, and also often their position within social-political space as one who stands with her in the fght for gender equality. Now, pejorative nicknames are just those nicknames that communicate something negative about their referent, whereas laudative nicknames do the opposite. There are myriad modes by which they do so. Name corruption, as in Drumf, ridicules via a non-standard derivation of the nickname from the formal name. Others fuse name corruption with supplemental pejorative content: Obozo, tRump. Here, the scornful descriptive content (bozo, rump) is precisely what corrupts the name. With Crooked Hilary, the ridicule is fat-footedly descriptive adjectival supplementation, yet with crooked, Trump and followers fomented perception of Clinton as beholden to Wall Street. With Low IQ Maxine, the descriptive content similarly does double duty: it plainly aims to diminish Waters’ intelligence, yet by exploiting racist tropes and stereotypes of Black persons, it is also a coded ‘race-card’, functioning to rouse and heighten an audience’s bigotry.

3 Tick Terms and Pejorative Nouns for Persons Thick terms are expressions that yoke together a descriptive dimension and an evaluative dimension. Our terms for specifc actions, things, and persons often incorporate not only a description of them but also a particular tenor or mode of evaluation of them. We assess actions as courageous; persons as loyal, or pious, or chaste; a city as vibrant; and a sculpture as poignant. The thick terms courageous, loyal, pious, chaste, and vibrant all describe a quality together with an evaluation  – for these, normally as positive evaluations. Negatively valenced thick terms like dogmatic, brutal, disloyal, cowardly, lewd, and blasphemous function similarly. All thick terms contrast with thin terms like good, bad, right, wrong, praiseworthy or problematic, which are merely evaluative, too lean and general to indicate the dimension along which something is assessed as positive or negative. Philosophical theorizing about thick terms tackles numerous questions: how do they combine evaluation and description? Do they ever express truths, and if so, does that mean there are evaluative facts? How are thick terms and concepts related to thin terms and concepts, and, in particular, do they inherit their evaluative component from them? See Eklund (2013), Zangwill (1995), Väyrynen (2013, 2016) for a small sampling of the abundant and mature philosophical literature on these questions. What is less developed and of special interest here is how negatively valenced thick terms, functioning as adjectives, nearly always also occur in some form as nouns that are frequently used reductively, to characterize what sort of person someone is – and as such are closely linked with other deeply pejorative expressions, including slurs. Those who act brutally are called brutes; the dogmatic are diminished as dogmatists; the disloyal as fnks, squeelers, and snitches. The cowardly are not just deemed cowards, but are also mocked as chickens and wimps. Furthermore, some thick terms’ meanings are ideologically loaded. Blasphemous is inextricably tied to religious doctrine about the sacredness of deities, and lewd and chaste to religious ideologies on sexuality. While the blasphemous are judged sinners, the lewd are often reduced to sluts and whores, lechers and rakes. Since many 215

Robin Jeshion

of these qualify as deeply derogatory terms, there is reason to think that the linguistic analyses of at least some thick terms possess parallels with many pejorative nouns, including some slurs (Hay, 2013; Cepollaro & Stojanovic, 2016; Cepollaro, 2020)

4 Varieties of Slurs Slurs have special standing amongst many other pejoratives. To call a word a slur, as opposed to just a pejorative, is to locate it within cultural and political space. Unlike pejoratives more generally, slurs typically disparage persons on the basis of central socially important categories including race, ethnicity, religious afliation, nationality, sexual orientation, political stance, ability, and weight. As such, we regard their weapon-uses as socially transgressive. As Nunberg (2018, 239) noted: To describe a word as a slur isn’t just to say that it’s ofensive, but to assign a particular moral or political tenor both to the ofense it gives and the ofense one commits in uttering it. Using a slur isn’t simply a break of personal manners or a sign of coarseness….For us, a slur is a kind of verbalized thought-crime: it perpetuates social inequalities, infects even innocent minds, and undermines the conduct of public discourse. And as such, slurring….becomes a speech act in which institutions and the law may take an ofcial interest. In short, derogating speech acts with slurs are not just ofensive; they frequently cause harm (Delgado and Stefancic, 2004; Waldron, 2012). Should all expressions deemed slurs receive the same linguistic analysis? Here, I distinguish several classes of them, what I call canonical slurs, descriptive slurs, gendered slurs, and stereotyping expressions/slurs.12 Although I cannot advance a full argument, the discussion here illustrates that there is reason to treat slurs as forming diferent classes with distinct linguistic properties.

4.1 Canonical Slurs Canonical slurs are expressions that derogate their targets on the basis of socially signifcant categories including (but not limited to) ethnicity, race, religion, sexual orientation, gender identity, political afliation, national status, class, and ability. Chink is a slur for Chinese persons, and the n-word for persons of African descent, kike and yid for Jewish persons, faggot for gay persons, and so on. Others include dago, dyke, gook, goy, honky, tranny, cripple, and illegal. Often, the targets of slurs value and embrace the very social identities they are derogated for as part of their own autonomously determined self-identity. Many people who are, say, gay or Jewish self-identify on the basis of their sexual orientation or religion (or ethnicity) as a member of the group. Canonical slurs are typically taken to have neutral counterparts, a non-pejorative correlative expression. The neutral counterpart of wop is Italian, of tranny is transgender, and so on. Nevertheless, not all slurs possess a syntactically simply neutral counterpart. Take gook, a term originating within the US military used indiscriminately for black- and brown-skinned natives of countries they occupied or were at war with, including Filipinos, Haitians, Costa Ricans, and Nicaraguans, and later for Filipinos, Koreans, and Vietnamese.13 Other slurs that lack a syntactically simple neutral counterpart include the racist dago (for Spanish, Portuguese, Italians) and the classist redneck. On most theories of slurs, neutral counterparts occupy a critical role in explaining slurs’ semantics. The reason for this is simple: particular slurring terms are systematically related to particular groups as devices for picking out, and derogating, those groups – the very groups that their neutral counterparts designate. Someone who, while aiming to speak literally, regularly applies kike to refer to Black persons is linguistically incompetent with the term on account of failing to know the group to which the term is applied. A similar point can be framed in terms of linguistic 216

Varieties of Pejoratives

understanding. Someone who understands utterances containing the n-word as being about Jews misunderstands that word. To account for such linguistic incompetence and misunderstandings, semantic theories need some specifcation of a slur’s target. Yet this condition is thin. While theories on which slurs and neutral counterparts are co-extensive easily satisfy it, the condition does not demand co-extension. Analyses on which slurs have empty extensions – for example, on which kike means ‘ought to be the target of contempt on account of being Jewish’ – satisfy it (Hom & May, 2013; Bach, 2018), as do those that give slurs’ meaning in terms of stereotypes of the group referenced by their neutral counterparts (Hom, 2008; Croom, 2013; Neufeld, 2019). A theory of slurring terms also minimally needs to explain how slurs difer from their neutral counterparts. Intuitively, slurs, but not their neutral counterparts, are standard tools of derogation.14 Utterances of them have a distinctive toxicity, with the power to infict deep harm on targets, ofend even when merely mentioned, and poison the communicative landscape, even amongst non-targets. There are three main avenues for accounting for slur’s toxicity: semantically, pragmatically, or socio- and psycho-linguistically. Semantic analyses include those containing explicitly derogatory content, like that encoded in ‘contemptable because Jewish’ (Hom & May, 2013, 2015; Bach, 2018); those containing dominantly negative group-stereotypes associated with the group, like being cheap, loud, and neurotic (Hom, 2008; Camp, 2013; Croom, 2013; Neufeld, 2019; Burnett, 2020); those signaling the speaker’s alignment with and endorsement of a negative socio-political perspective on the group (Camp, 2013, 2018; Rappaport, 2019); those encoding directives to interlocutors to adopt a derogatory perspective toward members of the group (Kirk-Giannini, 2019); those encoding speaker’s negative afective stances toward the group, typically the expression of contempt (Kaplan, 2005; Potts, 2005, 2007; Saka, 2007; Schlenker, 2007; Richard, 2008; Jeshion, 2013, 2016, 2018; Davis & McCready, 2020; Marques & Garcia-Carpintero, 2020). The derogatory element may be encoded as semantic descriptive content in the slurs’ meaning analyses, as presuppositions, as rules governing use, as a conventional implicature, or as what follows inferentially from an inferential role semantics. Pragmatic analyses construe speakers as fouting conversational maxims, typically manner maxims, to afliate with and signal their allegiance to bigoted attitudes about the group, or alternately, to cue and align themselves with bigoted ideologies (Bolinger, 2017; Nunberg, 2018; Pullum, 2018; Swanson, 2020).15 Socio-linguistic analyses account for slurs’ derogation by appealing to social properties of slurring words, including most prominently social prohibitions on their use, where the derogation derives from infractions of such prohibitions (Anderson & Lepore, 2013). Finally, to explain slurs’ toxicity, some suggest they have neuro-linguistic features that in hearing them, we undergo distinct neurological processing akin to that of hearing curse words, which accounts for their eliciting deep emotional responses (Rappaport, 2019). Theorists may ofer up a purist analysis, whereby a single semantic, pragmatic, or socio- and neuro-linguistic mechanism explains how uses of slurs derogate. Alternatively, they may advocate a theory drawing on a combination of semantic, pragmatic, and socio- and neuro-linguistic (and possibly other) mechanisms.16 Slurs occur not only in simple atomic predications, but in all varieties of complex constructions, including those involving negation, disjunction, conditionals, and modals, as well as imperatives, interrogatives, and vocatives. (1) Jake is a Kike. (2) Jake is not a Kike – he only looks like one. (3) If there are faggots in the bar, my father won’t go. In a construction like (2), assumed to be discourse-initial, the negation operator acts on the group that the neutral counterpart designates. The anti-semite that utters (2) expresses that Jake is not Jewish. In (3), the homophobe says what will happen if there are gay persons at the bar. In both, 217

Robin Jeshion

the operator does not interact with – does not remove – the derogating element in the slurs. The toxicity in utterance (1) is also present in (2). This reveals that the derogation projects across negations, modals, conditionals. The same phenomenon obtains with slurs occurring in nondeclaratives like interrogatives, imperatives, and vocatives. (4) Did you see all those Spics waiting outside the Home Depot? (5) Get lost, Kike! The utterances of slurs in these constructions are all derogating, and appear to be so for the same reason. So it appears, prima facie, that the derogatory element of slurs cannot be explained just by appealing to a descriptivist semantics encoding negative content, because while simple atomic predications like (1) are derogatory (‘Jake is contemptable because Jewish’), others like (3) are not (‘If there are people contemptible on account of being gay outside the bar,…’). Proponents of such descriptive analyses suggest that a constant derogatory element obtains across all these constructions because in each environment, in just using the slur, speakers make existential presuppositions  – that there are persons in the extension of the slur (Hom, 2008; Neufeld, 2019). With uses of slurs, speakers map social identities onto targets in a way that reductively deindividualizes, bypassing individuals’ unique qualities and histories, while reducing their social identity to their group category. This can be an additional mode of disrespecting individuals as individuals, but should not be wholly assimilated to the derogation of targets that speakers enact with slurs. Pervasive speaking and thinking of individuals as gay or Black or Jewish can also reductively de-individualize, yet isn’t exactly the same phenomenon as the distinctive scornful denigration inherent to slurs that manifest that the target is lesser. Tokenism and even politically valuable eforts at workforce diversifcation framed with neutral counterparts can encourage targets to see themselves as reductively de-individualized, regarded only, or primarily valuable as an instance of a kind. While morally egregious and a form of disrespect, the wrong of such de-individualization isn’t one and the same as that performed and enacted with slurs in their weapon-uses (Tirrell, 1999; Jeshion, 2013, 2018; Kukla, 2018). Slurs are also said to essentialize their targets to communicate that their targets have traits, dominantly negative, that are due to and explained by their ‘underlying nature’. Kike, say, suggests that Jews have an essence, one that explains their being or being disposed to be cheap, incessant talkers who dominate world banking. Some theorists have explained this phenomenon by building it directly into the semantics, as falling out of licensed inferences of uses of the slur, or treating slurs as failed natural kind terms like water, elm, and jade that incorporate (supposed) metaphysical biologically determined essences (Tirrell, 1999; Camp, 2013; Neufeld, 2019; Burnett, 2020).17 Another option is to construe how slurs communicate what kind of person one is as falling out of the structure of contemptuous regard itself, something that is alive not only in slurs but in many linguistic ‘degradation ceremonies’. On such an analysis, linguistic acts manifesting contempt for a particular attribute or failing map a character-defning identity onto the target, saying what a person is’ (Jeshion, 2018).18 Finally, most theories of canonical slurs also aim to explain the conditions under which utterances with slurs are true. In the interests of space, on this matter I refer the reader to alternative sources.19

4.2 Descriptive Slurs Another large and pervasive class of slurs has received little attention. These are expressions like curry-muncher, raghead, Christ killer, guido, and porch monkey that, like canonical slurs, are systematically applied to socially defned groups rooted on ethnicity, race, religion, sexual orientation, etc., by non-target members, and for the same purpose – to derogate targets qua group member. Unlike canonical slurs, however, the expressions themselves contain descriptions that purport to 218

Varieties of Pejoratives

somehow characterize the intended targets. Christ killer, for instance, is a slur for Jews that carries the description of them as (descendants of ) killers of Christ. Kike, by contrast, is an expression that ofers no characterization of Jews. Let’s call slurs like these descriptive slurs. Should the analysis of descriptive slurs’ truth conditions and mode or modes of derogation be accounted for in roughly the same fashion as canonical slurs? What linguistic role does the additional descriptive material play in the expression? Do descriptive slurs exhibit the same properties as canonical slurs? Some argue that descriptive slurs possess distinctive semantic properties. A recent argument by DiFranco (2015) aims to establish the diference thus: because curry-muncher is a slur for South Asians, for theories that ofer a common semantic analysis of canonical and descriptive slurs and maintain that canonical slurs are truth-conditionally equivalent with their neutral counterparts (assuming they have one), (6) and (7) will be truth-conditionally equivalent: (6) Rohit is a curry-muncher. (7) Rohit is a South Asian. That is, whatever contribution the descriptive slur makes to the truth conditions of the sentence containing it is the same as the contribution the neutral counterpart makes to the  truth conditions of a corresponding sentences that substitutes the neutral counterpart for the slur. But now consider, (8) Rohit used to be a curry-muncher, but hasn’t been one since he changed his diet. Yet if (6) is truth-conditionally equivalent to (7), (8) must be false since a change in diet cannot change one’s ethnicity. But, the argument goes, (8) can be true, so curry-muncher in (6) cannot make the same contribution to the truth conditions as South Asian does in (7). There is reason to doubt the argument’s cogency. For one, it implicitly supposes that curry-muncher must have a single meaning and make the same contribution to truth conditions in both (6) and (8). But it is possible that on the intended reading on which (8) is true, the slur has a diferent meaning than it does in (6). On such a reading of (8), the expression must at least pick out those who eat curry, while in (6) it need not: it may simply be co-extensive with South Asian (Caso & Lo Guercio, 2016). Though fallacious due to possible equivocation, the argument helps shed light on descriptive slurs. I’ll suggest that we treat them at the truth-conditional level as akin to canonical slurs, for doing so unifes and sheds light on a wide range of descriptive slurs. Bigots who use curry-muncher as in (6) aim to pick out just the class of South Asians, not the subclass of curry-eating South Asians. The point generalizes for the class of descriptive slurs and receives support from a wealth of alternative examples. If a bigot describing the scene in his high school cafeteria says, (9) The curry-munchers sit far away from the noodle-niggers. he aims to speak about South Asians as a class, and East Asians as a class, not subclasses of these groups captured by treating the descriptive content of the slurs as contributing another property computed compositionally. The South Asians who despise curry and East Asians who dislike noodles correctly understand that they too are being derogated. Similarly, the bigot who utters (10) does not aim to pick out the group of those named Guido or even the set of Italians who are named Guido, (10) There are more Guidos in Brooklyn than in all of California. 219

Robin Jeshion

but simply the set of Italians (or male Italians).20 That is, the descriptive material in the descriptive slur does not factor into determining the slur’s extension or into the truth conditions of the assertion containing it. Further examples bolster the point. As a slur, camel jockey refers to those of Arabic heritage, not its intersective class with those who actually ride camels; beaner refers to those of Hispanic heritage, not its bean-eating subclass. In these examples, and those that follow, the descriptive material is, and is intended to be, referentially and truth-conditionally inert. In this regard, descriptive slurs are analogous to pejorative nicknames. With Shifty, Trump refers to Adam Schif, notwithstanding Schif’s lack of shiftiness. Because the descriptive material plays no role in reference-determination, theorists can, in general, ofer a common account of descriptive slurs’ capacity to derogate that they do for canonical slurs (Caso & Lo Guercio, 2016). For instance, for a pejorative use of curry-muncher, an expressivist and a pragmatic afliationist can both maintain that the bigot references South Asians, yet the former theorist holds he expresses contempt for South Asians as South Asians while the latter afliates with those social groups that hold South Asians in disregard. If this route is adopted, it still needs to be supplemented with an analysis of what and how the descriptive material contributes to these slurs’ derogation. After all, there’s nothing itself diminishing about being one who eats curry or is named Guido. I would suggest an alternative analysis. While the descriptive content is not extension-determining, it systematically functions as a tool to additionally mock, ridicule, derogate, or dehumanize the targeted group. Here again, pejorative nicknames ofer a model: while ‘shifty’ plays no role in Shifty Schif ’s reference-determination, it is used as a tool to mock Schif himself, and create a social identity of him as one who’s not to be trusted. The way these slurs do so will be a function of a range of factors, most prominently what the descriptive content is and how it interacts with ideologies about the group. Here again, it mirrors the structure of pejorative nicknames. While descriptive slurs can select content about virtually any features of or associated with the group, salient patterns emerge, and codifying them illuminates the nature of the mockery, ridicule, derogation, and dehumanization. First, some descriptive slurs explicitly contain stereotyping content: Stereotyping: penny pincher, Christ Killer ( Jewish), pansy, fairy (gay men), orange-picker (Hispanic), camel jockey (Arabic) Uses of these descriptive slurs disseminate, perpetuate, and reinforce the group-stereotypes encoded in their descriptive contents. Yet the descriptive material is not exclusively a way of branding its targets with a stereotype. It dominantly functions to mock them. Second, descriptive contents of these slurs are often used to mock customs and cultural habits of the group. Many are embraced by group members as communal identity-markers, and some constitute or represent afrmations of their group’s values and commitments. A few may even signify their group’s unity. Consider that slurs with descriptive content are systematically associated with: Names common to the group: Guido (Italian), Hymie, Ikey ( Jewish), Mick (Irish)21 Foods favored by the group: beaner (Hispanic), curry-muncher (South-Asian), frog (French), kraut (German), noodle nigger (Chinese) Symbolic dress or customs of the group: beanie ( Jewish), half-dick ( Jewish men), raghead (Muslims, Hindus, more broadly to Pakistani, and Mid-Easterners) Others attack the language of the group and the name for the group. Language of the group: ching-chong (Chinese)22 Name for the group: Injun (Native-American Indian), Jap ( Japanese), Hebe (Hebrew), Lesbo (lesbian), Spic (Hispanic) 220

Varieties of Pejoratives

Of course, ching-chong itself lacks descriptive content – the words don’t have any independent meaning at all, and Paki, Jap, Hebe, Lesbo, and Spic involve name-clipping, and Injun name-corruption, and thus similarly introduce no additional descriptive content beyond the group name. They merit inclusion here as descriptive slurs because, again on analogy with pejorative nicknames, the slurring words themselves contribute an additional source of derogation – by mocking the target group’s language and name. Yet other descriptive slurs mock aspects of the body or reduce the person or group to their body parts. Body: eight ball, nappy head (African or African descent), slanty, slope (Asian), redskin (Native American) Especially for certain gendered slurs for women, as applied to women, the descriptive content metonymically reduces the person to a sex object, while simultaneously saturating that reduction with hatred:23 Sexualized body: cunt, slit Almost all of the foregoing descriptions derogate in a particularly invidious way, by mocking and attacking what might best be called group intimacies: our bodies, the foods we consume, the language we speak, what we name our children, what we and society name ourselves qua group, how we cloth ourselves and express religious devotion.24 As tools of mockery, pejorative speech acts with descriptive slurs function in many ways like non-linguistic mockery. Think of Trump’s 2015 mocking the New York Times journalist Serge Kovaleski for his arthrogryposis, to the glee of his rally followers.25 Think of the Nazi in the movie Son of Saul, humiliating Saul with a mocking Jewish shtetl dance, eliciting laughter from his fellow Auschwitz SS commandos.26 Mockery hooks the emotions, attacking what is intimate, and thus vulnerable, and makes the audience laugh, and partake in the scorn. Other descriptive contents are stronger still. Some dehumanize with lacerating metonymic or synecdochic reduction of persons to animals, especially with the simianization of Blacks. Others dehumanize with tormenting references to dehumanizing treatment of the group in a way that mocks or that suggests the treatment as ftting. Animal: inyenzi (meaning ‘cockroach’, Tutsi by the Hutus), judenschwein ( Jew-pig), ape, porch monkey (Black) Treatment: cotton picker, jungle bunny, Jim Crow (African-American), lamp shade, oven, oven-dodger ( Jews, reference to Holocaust) The upshot is that descriptive slurs27 are in two ways akin to pejorative nicknames: while the descriptive material is referentially and truth-conditionally inert, it adds extra-referential supplementary content which colors perception of the referent and is used as a stinging tool to ridicule, mock, derogate, or dehumanize the targeted group.

4.3 Gendered Slurs I distinguish gendered slurs from canonical and descriptive slurs. Whereas the latter are dominantly used as weapons to reference and derogate a whole group G qua G, and in particular for identities that target members themselves often embrace, slurs of gender are not deployed to reference a whole gender group. Bitch, slut, cunt, and twat have a range of basic uses on which they do 221

Robin Jeshion

not function to reference girls and women as a class; pussy, wuss, and the old-fashioned sissy do not reference boys and men as a class, as (11), (12) exemplify. (11) Cover up, you look like a slut! (12) If you back out now, you’re even more of pussy than you already are. Gendered slurs are rather expressions dominantly used to police behaviors and attitudes that defy certain misogynistic and patriarchal norms on gender (Freeman, 1968; Ashwell, 2016).28 Bitch has been canonically applied to women and girls whose actions, attitudes, or demeanors defy misogynistic norms governing assertiveness. As second-wave feminist Jo Freeman noted, the word “serves the social function of isolating and discrediting a class of people who do not conform to the socially accepted patterns of behavior”.29 Women and girls are dubbed bitches when they are deemed domineering, stuck-up, overly aggressive, independent, strong-minded, or outspoken. Slut is applied to those who break or are perceived to be or feared to be breaking a host of religious and patriarchal social norms regarding their sexuality, specifcally concerning the amount of sex, the number and variety of sex partners, when she frst starts having sex, as well as the autonomy and assertiveness she exhibits about wanting sex or wanting certain kinds of sex. Girls and women are branded sluts for wearing leggings (too tight) or a drapey shirt (too loose) or simply for puberty hitting them earlier than their peers (too much boob for ffth grade). Perversely but all too commonly, women are dubbed sluts after being raped. In short, slut is systematically applied to those deemed or feared to look or be too sexually promiscuous by the standards of patriarchy in place in a particular context. Similarly, pussy and wuss are applied to boys who break patriarchal social norms on masculinity for being insufciently daring, brave, sporty, adventurous, strong-willed, tough, or overly emotional, sensitive, attached. While the threshold for counting as breaking the norms has evolved over time, the patriarchal norms themselves and the slurs that echo them have tenaciously endured.30 A striking piece of evidence that gendered slurs do not reference the whole gendered group is that group members regularly wield such slurs. This is atypical for canonical and descriptive slurs in their basic uses. Jews do not commonly derogate other Jews with kike; Hispanics don’t typically insult their fellows with Spic.31 Yet women slut-shame and smear other women by branding them sluts and bitches as much as men; men commonly abuse and stigmatize other men by brandishing pussy and wuss to police male behaviors away from the feminine and non-hetero danger zones. There is another important diference between gendered slurs and canonical and descriptive slurs. Slurs of gender often possess single or a class of corresponding opposite laudative expressions that characterize properties that are valued and rewarded by the standards of patriarchy. Pure, innocent, virginal, and chaste, when applied to girls and women,32 glorify along the same spectrum and according to the same patriarchal norms on women’s sexuality by which slut or whore shame. While these terms may, to some ears, sound anachronistic, the glorifcation of purity and virginity is hardly a thing of the past. It is alive and well, gaining steam from conservative culture urging girls to wear chastity rings and attend purity balls that culminate in girls’ pledges to their fathers to stay sexually abstinent until marriage, and their fathers’ reciprocally pledging to protect their virginity. Opposed to sissy, wuss, and pussy are the alpha male and the stud, the old-fashioned macho, and the new-fashioned Manosphere’s Chad and slayer, all terms that celebrate (along various diferent dimensions) uber-masculine traits, and are used to rank men as dominant, especially sexually dominant. While basic uses of canonical slurs derogate targets as lesser in virtue of their group membership, they aren’t typically contrasted with a correlative opposite or superlative. (Aryan marks an exception, indicating being both Germanic and superior.) Let’s say that pairs like sissy and stud are opposite polarity same-spectrum evaluatives.


Varieties of Pejoratives

Together these points underscore that, unlike most canonical slurs, gendered slurs lack neutral counterparts. The point is not just that gendered slurs do not reference whole genders as a class, and so slut, for example, does not have woman as a neutral counterpart. It is also not just that gendered slurs lack syntactically simple neutral counterparts or that there is no non-evaluative counterpart to these terms. The point is stronger: even the most promising candidate syntactically complex counterpart expressions do not capture the meaning of their corresponding gendered slurs. I’ll argue this point by illustrating with slut. The most promising neutralized meaning analysis is promiscuous woman. The common dictionary defnitions construe promiscuity when applied to persons quite neutrally to mean ‘having or having a propensity toward frequent casual sex with many partners’. Some specify a negative polarity like the OED’s ‘indiscriminate in sexual relations’, which has the favor of a thick term, combining description and evaluation. The point I will argue is that on either defnition, slut is not synonymous with promiscuous woman. The problem is that the meaning-analysis of promiscuous woman is insufcient for capturing the ideologically rooted meaning of slut which is fundamental to the concept. By contrast, promiscuous woman need not encode that the standards for being deemed promiscuous derive from misogynistic and patriarchal sexual norms, whereas, I contend, the meaninganalysis for slut must. To see this, consider a twin earth thought experiment. Imagine society S* that is, in many ways, like our society S. S* is however (radically) unlike ours insofar as it is egalitarian along gender lines. S* has no patriarchal ideology I, had in many cultures on Earth. Up through time t, S* has no social norms inhibiting women’s sexuality and constraining their sexual freedoms, norms that partly characterize patriarchal ideologies I and systems of women’s oppression, and the women in S* fully exercise their sexual freedoms – at least in the same manner and to the same degree as men in S*. Yet, at a certain point in time t, women in S* are strongly discouraged from having a lot of sex with casual partners. Why the change? Some years before t, a novel virus spreads through S’s population. This femivirus, as they call it, is transmitted through sexual contact and can infect everyone; however, it is just women, not men, who develop severe, sometimes life-threatening symptoms from it. As yet, no vaccines or protective measures save for knowing your partner is femivirus-free will protect the women in S*. In time, a purely medical ideology I* arises according to which women’s own health will be compromised by their having frequent sex with numerous casual sex partners. In time, a word-form slut* is used in S* to police behaviors away from what is believed to, and does, compromise women’s health. Imagine further that slut* eventually becomes a term of derision, used to derogate and shame those who break the medically grounded norms on sexual health and even those whose appearance make others think or fear that they would break those norms. In fact, in S*, slut* is applied to police exactly the same behaviors as slut is used to police – the only diference is the underlying ideology. My intuition is that even if slut and slut* are used in the same contexts to police the same kinds of sexual behaviors, the two expressions do not mean the same thing and the folks of S* lack beliefs and other attitudes having the concept slut as part of their mental content. The beliefs they express with slut* cannot be the same as those expressed in our community with slut. However, on our understanding of the meaning of promiscuity, those women that breach the medical norms in S* are appropriately called promiscuous women. Thus, while slut* and slut both apply to sexually promiscuous women, they are not identical in meaning.33 If this is right, there needs to be some way to work the patriarchal ideology into the meaning of gendered slurs. One option would be to posit an extremely rich descriptive content that encodes foundational ideas of the ideology.34 This, however, will implausibly force too much content into the notion. A more promising option would be to treat these slurs’ meanings as indexically tied to their ideology-infused environment.


Robin Jeshion

Finally, although slurs of gender lack neutral counterparts, they share the ofense projection profle of canonical slurs and descriptive slurs. Each of (13)–(16) involves a weapon-use of the gendered slur, (13) Jane’s not a slut. She’s bona fde marriageable material. (14) Jane puts out, but she’s not a slut. (15) Why is this party full of so many bitches? (16) Do you think Edgar’s a wuss? I’ve never seen him taking control of that nagging wife of his. and what is derogating in the utterance is the same as what is derogating in (11) and (12), where the speaker uses the slur to attribute a property. While both (13) and (14) involve a denial that Jane is a slut, the speaker nevertheless leaves intact commitment to this misogynistic way of classifying women. Although the speaker in (16) only inquires whether Edgar is a wuss, they register adherence to a patriarchal ideology that ridicules men who don’t dominate women. In fact, the gendered slurs’ projection profle is also manifest in their opposite polarity same-spectrum evaluatives. Responses to (16) like (16′) or (16″) (16′) Hell no, I’m pretty sure that Edgar’s a total alpha. He was just being polite with her parents around. (16″) Yeah, but it sure would be better if he were an alpha and put her in her place. reveal the same commitment to the questionable patriarchal norms even though the laudative alpha is contained in epistemic and modal constructions. In this way, gendered slurs possess important similarities to thick terms like lewd and chaste rooted on patriarchal ideologies. I close this section be noting that what I am here classifying as slurs of gender are not uniquely associated with gender. Other pejoratives having nothing to do with gender have common structure. Consider uppity. This deeply racist term does not refer to Blacks as a group, but instead functions to police behaviors by putting down just those Black people who refused to kowtow to whites. Any terms that function to put down and police people for stepping out of their bigoted-ideologically defned social roles will count. Slurs of gender, as I have been calling them, are simply the most prevalent instances.

4.4 Stereotyping Expressions Our fnal group of pejorative expression explicitly evokes a rich stereotype of a subset of a social group or intersectional group. They gain their pejorative favor in part by their relationship to a whole web of social constructions and ideologies on one or the other category, or their intersections. Because the stereotypes are so fne-grained and understanding them requires extensive social knowledge, ordinary dictionaries either fail to list them or provide unhelpfully thin defnitions. I have therefore relied heavily here on the Urban Dictionary, which ofers surprisingly exacting, often brilliantly wicked defnitions of terms that ofer a lens into the social structures and social invective surrounding them. Consider Welfare Queen, the Reagan-era label for women who abuse the welfare system by fraudulently collecting excess payment or staying on welfare indefnitely, yet was, from its inception, highly politicized. To justify welfare reform, Reagan deployed it to create the stereotype of a lazy young Black woman, donning furs and driving a Cadillac, who ‘plays the system by remaining unwed and having as much promiscuous sex as possible to pump out as many children as possible’, to milk white-tax-payers for everything she can get [Urban Dictionary]. Since being rightly castigated as deeply racist and misogynist, the use of the term has diminished, but the stereotype 224

Varieties of Pejoratives

is frequently parroted more covertly if less explicitly on the political stage with fear-mongering mentions of welfare and other political messaging like Speaker of the House Paul Ryan’s 2019 complaint that ‘inner city’ residents are ‘not even thinking about working or learning the value of the culture of work’ (Covert, 2019). The expression and its stereotype have been used within the Black community as well. In his 1996 sketch Black People verses Niggaz, the Black comedian Chris Rock famously tongue-lashed, loudly and explicitly, a subgroup that includes exactly the stereotype that Welfare Queen encompasses. The comment section to the Rock YouTube video reveals a wealth of pro-remarks, with many commentators underscoring their Black identity. In part, because Rock is the speaker, the diatribe is plausibly interpretable as exclusively classist, not racist. Now take Jewish American Princess, the term encoding a stereotype that emerged in the mid20th century in part by Jewish-American writers like Herman Wouk, Philip Roth, Saul Bellow, and Norman Mailer, who each crafted central young female characters who are conventional, materialistic, and indulged. The term itself was popularized in the 1970s, typically by Jewish writers and journalists, and came to represent an archetype of Jewish women from privileged backgrounds, usually suburban, ‘who have been spoiled by material wealth and overbearing parents to the point that they are self-absorbed, high-maintenance, materialistic, and snobby’ [Urban Dictionary].35 While some regard it as an ethnoreligious sexist slur, especially because it echoes centuries old tropes about Jews’ obsessions with money and, of late, dominating world-banking, others do not. Strikingly, it is rarely used outside the Jewish world (Keiles, 2018).36 When it is, however, it is frequently regarded as anti-semitic. Another stereotyping expression in this genre is Tiger Mom, introduced by Amy Chua in 2011 for a mother, typically East or South Asian, who relentlessly and ruthlessly manages her children’s studies with authoritarian parenting methods, demanding high academic achievement and excellence in high-status extra-curriculars like classical music performance, while sacrifcing their social lives. Chua originally crafted her notion within her half-autobiographic, half how-to Battle Hymn of the Tiger Mother to show why Chinese parenting methods are superior to lax, ‘good job!’ western methods, in efect, aiming to make Tiger Mom a badge of honor. The book ended up being a partial lament, with Chua adopting a more nuanced, mixed perspective, and the expression itself is now highly variable in its valence, depending upon the context. Like Jewish American Princess, its origins are in-group and it amplifes tropes of the Asian math-wiz, and ‘a grade of A- being an ‘Asian-F’’. Yet the expression difers insofar as it is in wide circulation, usually used without taboo. Others in this linguistic genre include Uncle Tom, described in the Urban Dictionary as a derogatory term for a Black man ‘who is a sell-out to his own race…who will turn his back on other Black people to keep the approval of whites’; we have the similarly valenced whitewashed, for those Black people who have mannerisms and behaviors of white people, roughly synonymous with Oreo, Black on the outside, white on the inside. Amongst the bloated assortment of stereotyping terms for girls and women, we have Valley Girl, a white teenage girl from Southern California who “spends 99% of her waking hours at the mall and says ‘totally’, ‘omg’ and ‘whatever’ waaay too much”; and the UK Essex Girl, ‘a dumb, gum-chewing sexually promiscuous girl’ with ‘tacky jewelry, garish make-up…stiletto heels, frequently inebriated’. While almost always regarded as pejorative, many see these as innocent, maybe sometimes apt, while others, like Bate (2016), regard them as ‘not really about geography or a fake tan’ but rather ‘about class and gender…where misogyny meets snobbery’, a cross between the gendered slur slut and the classist slurs hillbilly and trailer trash. Stereotyping words that are often regarded as non-pejorative are the large swath for mothers, usually white mothers – soccer mom, hockey mom, helicopter mom, stage mom, and MILF. All but the last are batted around casually, employed in the respectable, woke press like The Atlantic and the NYT, often with blindness to how demeaning the stereotypes are and to the near absence of corresponding typologizing of fathers.37 Indeed, sometimes the terms, especially soccer mom, are 225

Robin Jeshion

regarded as just marking a demographic category. Yet a quick look at the telling Urban Dictionary shows otherwise. Soccer Mom receives six defnitions, each dripping in condescension. Consider this small sampling from the entries: ‘the only people who have no real purpose in life other than to pick there(sic) children from school, take them to an afterschool program….’; ‘the most despicable species of women known to humankind’; ‘white, middle-classed woman’; ‘usually blond’, living in the suburbs, who ‘drives a minivan or SUV,….usually Christian, blames the internet and video games for all the problems in the world’ [Urban Dictionary]. We now have the 2020 meme-driven Karen, for entitled, demanding middle-aged white women who use their privilege to get what they want, often at others’ expense; Becky, for conventional white women who are decidedly unwoke, unaware of their privilege; and Basic Bitch, for those women, lacking in imagination, who always veer toward mainstream ideas, clothes, music. Though the compendium for boys and men is smaller, there’s the 1980s and 1990s frat boy, the updated Bro and Chad, for young white men who hang around in packs; overuse “‘sick’, ‘dope’, ‘epiccc’”; and are blind to their privilege and misogyny [Urban Dictionary]. And then there’s the unambiguously classist Joe-six-pack, for the ‘average’ – to twit, low – intelligence American blue-collar worker who drinks a lot of beer and lives in a fyover state. I present this robust catalog of expressions to highlight their diversity and the challenges that arise for theorizing about them. To my knowledge, there have been no studies within linguistics or analytic philosophy that systematically analyze how they encode or convey their rich meanings; how they are related to theories of canonical, descriptive, or gendered slurs; and how they are or become pejorative in specifc speech acts. I wholeheartedly encourage development of this fertile area of research. I close here by noting six common features of these stereotyping expressions: frst, and most obviously, these terms are not derogatives for the whole social identity groups that canonical and descriptive slurs reference. Second, while most are dominantly pejorative, they difer from canonical, descriptive, and many gendered slurs insofar as most do not carry or communicate deep derogation – though of course some can be used in a context to do so. In general, to call or refer to someone as a Tiger Mom, a Becky, or a Jewish American Princess is not to express, therewith, that they are lesser persons. Third, members of the broader class of which the subgroup or stereotype is a part frequently use these expressions as pejoratives – unlike with canonical and descriptive slurs, but very much like gendered slurs, often to police behaviors, and sometimes they have more authority and entitlement to do so.38 Fourth, many stereotyping pejoratives cross stereotypes of, say, race, ethnicity, or class with gender, and a fuller analysis of them would have to incorporate important theorizing about intersectionality.39 Fifth, the extent to which these terms diminish and ridicule their intended referents difers radically depending on the stereotype itself and how they are related to wider stereotypes of the supergroups from which they stereotype: being an authoritarian, unyielding parent obsessed with your children’s education is generally construed less negatively than being a welfare fraud. Lastly, like many stereotypes, most fexibly widen to include those even beyond their primary group: as Chua noted, Tiger Mom can refer to parents of all stripes, women, men, and the non-binary, and all ethnicities.

5 Conclusion I have articulated three key distinctions about pejoratives, and have detailed a large class of pejoratives, with emphasis on pejorative nicknames, thick terms incorporating evaluation and classifcation, and four varieties of slurs. My main aim has been to taxonomize, with the hopes that it educates the uninitiated and inspires more wide-ranging research by the experts. I hope also to have illustrated that pejorative nicknames and thick terms may ofer rich sources for further understanding of slurs, and, moreover, that each class of slurs possesses distinctive linguistic and structural features that may present challenges to views that aim for a uniform analysis. 226

Varieties of Pejoratives

6 Acknowledgments I presented material from §4.2 at Rutgers University, Institut Jean Nicod, University of Chicago, University of Barcelona, Stockholm University, and from §4.3 at the 2019 Lisbon Pragmasophia Conference. I am grateful to the audiences for their valuable feedback, especially Liz Camp, Jef King, Manolo Garcia-Carpintero, Katrin Gluer-Pagin, Peter Pagin, Francois Recanati, and Ken Taylor. Special thanks to Justin Khoo and Rachel Sterkin for their astute editing and immense patience.

Notes 1 I set aside the diachronic phenomenon of pejoration, a specifc kind of linguistic change whereby a previously neutral or positive polarity word comes to acquire a new pejorative lexical meaning, connotation, or association. Idiot, now decidedly pejorative, was previously a polite formal way of speaking about severe intellectual disability within legal and medical contexts. Pejoration contrasts with amelioration, the process of linguistic change whereby a negatively valenced expression comes to be neutralized or have a positive meaning, tone, or linguistic association. While bad has a pejorative meaning, it has positive slang uses. ‘Whoa, your dress is bad!’ can be a fabulous compliment. While both processes are related to our discussion, especially to pejoratives’ polysemy and context-sensitivity, I refrain from addressing them here. 2 A comprehensive analysis of pejorative lexical items will ideally explain their relationship to laudatives (angel, goddess, whiz, hottie) and honorifcs (Ms., Your Honor), as in Potts and Kawahara (2004). 3 See Allen (1983) for a robust survey of ethnic derogatory terms. 4 The sufx -ard is itself pejorative, and etymologically gave rise to bastard and retard, yet portmanteau of of retard itself is what is now generating the new terms. 5 See Potts (2005, 2007) for seminal accounts of expressives and intensifers, Jeshion (2013) for comparison of such complex expressions to slurs. 6 The prosody of ‘sing-song’ conventionally mocks and is discernable without reliance on semantic descriptive content. 7 Alternatively, contemptuous regard may be conveyed in facial expressions (Ekman & Friesan, 1986; Ekman & Heider, 1988; Ekman, 1994). Miller (1997) ofers a fascinating analysis of the psychology and social functions of disgust. 8 Sennet and Copp (2019) discuss slurs as verbs and adjectives. 9 Many pejorative expressions are difcult to classify because they branch multiple categories. For instance, fag hag, which originated as gay slang, is a pejorative for women whose prime social relationships are with gay or bisexual men. 10 Communicating bias, bigotry, or hierarchy is of course not inherent to code words or dogwhistles. Medical slang, for instance, includes code words that shield patients from fear (BONITA, big ole’ needle in the ass) and alarm (code pink, missing baby). 11 A notable exception is Nunberg (2018), who aims for a uniform treatment of slurs and political pejorative nicknames. I believe the analysis falls short in part because nickname pejoration is itself multifarious, but development of this point must be left for another time. 12 These are not hard and fast categories that neatly classify words; numerous terms belong to multiple categories. 13 Gook is sometimes said to be 20th-century US ‘soldier’s slang’ for anyone who is not American, though this misses the racist element in the term. 14 Of course, slurs can be used without derogating targets, and neutral counterparts can be used to derogate, as we saw above. The point is just that slurs have a specifc pejorative function that neutral counterparts lack. 15 Kukla (2018) seems to have a similar view. 16 Hom, Hom and May, Camp, Tirrell, Neufeld aim to explain slurs’ properties by appeal to their semantic analyses alone, Nunberg, Bolinger, and Swanson by appeal to pragmatics alone, Anderson and Lepore by appeal to socio-linguistics alone. Jeshion, Garcia-Carpintero and Marques, and Rappaport draw on a combination of semantics, pragmatics, socio-linguistics, neuro-linguistics. 17 Neufeld draws on empirical research on psychological essentialism by Gelman (2003) and others to frame and support her analysis of slurs as encoding essences. 18 The idea of mapping essentializing identities on people with ‘degradation ceremonies’ goes back to classic work in sociology by Harold Garfnkel (1956) and, more extensively, by Erving Gofman in his


Robin Jeshion

19 20 21



24 25 26 27 28 29 30

31 32 33 34 35 36 37

38 39

discussions of stigma (1963) and interaction ritual (1967). Contemporary analysis of the structure of contempt within psychology and moral psychology emphasizes how contempt is ‘whole personal’, has a basis which is perceived as character defning, and contributes to hierarchy formation (Mason, 2003, 2016; Fischer & Roseman, 2007; Bell, 2013; Darwall, 2018; Roseman, 2018). See most of the pieces cited in this section, especially Richard (2008) and Hom and May (2018) for interesting heterodox views. Sentence (10) can have other meanings as well. It can be used narrowly to refer only to those Italians who satisfy a certain stereotype of Italian males, or more widely to refer to anyone who satisfes that stereotype, regardless of whether they are Italian. Simply deploying a name common to the group is not sufcient for counting as a descriptive slur or derogative. The term must be wielded like a slur. The grocery store Trader Joe’s has for years used labels like Trader Giotto’s for Italian- and Trader Jose for Mexican-food products. Like other companies, they have been charged with micro-racial bias for exoticizing other cultures, but the names were not being used as slurs. Ching-chong is typically used as a conventionalized one-of word to mock, and thus difers from most canonical and descriptive slurs insofar as it usually is not used to reference a group and does not combine with other words syntactically. Still, it is now also commonly coerced to act as a noun (“You’re going to date that ching-chong?”). What of prick, dick, and schmuck? As used in American-English, these expressions do not metonymically reduce men to body parts. And, while clearly gendered, they disparage for certain behaviors, and are used to police behaviors, yet not in any way governed by social norms oppressing men, unlike bitch (Horn, 2018). Group intimacies contribute to group identity construction. These are cousins to socio-linguists’ notion of “acts of identity” (Le Page & Tabouret-Keller, 1985; Eckert, 2000, 2008; Bucholtz & Hall, 2005). The moment is documented here: The scene is depicted here: Similarly, complex expressions like dirty Jew possess important linguistical parallels to pejorative nicknames and descriptive slurs. Manne (2018) ofers a compelling analysis of misogyny as the ‘policing’ or ‘law enforcement’ branch of patriarchy. As she notes in her unpublished BITCH Manifesto, 1968, the ‘most prominent characteristic of all Bitches’ is that they ‘violate conceptions of proper sex role behavior’. Ashwell insightfully argues against regarding gendered slurs as having woman and man as their neutral counterparts, and also emphasizes the normative favor inherent to gendered slurs. While I share elements of Ashwell’s understanding of gendered slurs, I deny that just because a class of expressions like slurs ‘are united in lacking warrant for their application’ they ‘ought to be analyzed similarly’. Indeed, part of my goal here is to show that slurs come in various varieties and might, indeed, likely do, resist a comprehensive single linguistic analysis. There are exceptions. Kennedy (2003) discusses weapon-uses of the n-word within the Black community. Although these expressions are adjectives, when they implicitly or explicitly modify girl or woman, they describe a type, and thus are akin to the old-fashioned good girl, making them opposites of slut and whore. In new research, I ofer more detailed development of this argument. Note that there may be a corresponding twin-earth argument for canonical slurs, but I cannot delve into its intricacies here. The idea is to provide a souped-up version of Hom’s early Combinatorial Externalism analysis of racial slurs, which richly incorporates content about underlying racist ideologies. Urban Dictionary. The term itself was also perpetuated by Jewish journalists, as in Baumgold (1971) and brought together in Lukatsky and Toback (1982). Keiles (2018) ofers a more recent and thoughtful article on the stereotype. Non-Jews living in Jewish areas sometimes use it, typically playfully, and sometimes in widened ways, to include non-Jewish people who, say, dress like the archetype. Deadbeat dad is one for those who absolve parenting responsibilities, and usually not used in the respectable press unless in scare quotes. DILF is new on the scene. Some of these expressions have been equalized along gender lines – helicopter parenting is commonplace – but there should be no doubt about the gender-disparities in these terms’ frequency. The relevant contrast is with weapon-uses of canonical and descriptive slurs, not, say, uses of the terms with their reclaimed meanings. See especially the pioneering work by Crenshaw (2017).


Varieties of Pejoratives

References Allen, I. (1983) The Language of Ethnic Confict: Social Organization and Lexical Culture. Columbia University Press. Anderson, L. & Lepore, E. (2013) Slurring Words. Nous, 47, 25–48. Ashwell, L. (2016) Gendered Slurs. Social Theory and Practice, 42(2), 228–39. Bach, K. (2018) Loaded Words: On the Semantics and Pragmatics of Slurs. In Sosa, D. (Ed.), Bad Words, Oxford University Press, 60–76. Bate, M. (2016) The Problem with the Phrase ‘EssexGirl’. Baumgold, J. (1971) The Persistence of the Jewish Princess. New York Magazine 4(12), 25–31. Bell, M. (2013) Hard Feelings: The Moral Psychology of Contempt. Oxford University Press. Bolinger, R. (2017) The Pragmatics of Slurs. Noûs, 51(3), 439–62. Bucholtz, M. & Hall, K. (2005) Identity and Interaction: A Sociolinguistic Approach. Discourse Studies, 7, 585–614. Burnett, H. (2020) A Persona-Based Semantics for Slurs. Grazer Philosophische Studien, 97(1), 31–62. Camp, E. (2013) Slurring Perspectives. Analytic Philosophy, 54(3), 330–49. Camp, E. (2017) Why Metaphors Make Good Insults Perspectives, Presupposition, and Pragmatics. Philosophical Studies, 174, 47–64. Camp, E. (2018) A Dual Act Analysis of Slurs. In Sosa, D. (Ed.), Bad Words, Oxford University Press, 29–59. Caso, R. & Lo Guercio, N. (2016) What Bigots Do Say: A Reply to DiFranco. Thought, 5, 265–74. Cepollaro, B. (2020) Slurs and Thick Terms: When Language Encodes Values. Lexington Books. Cepollaro, B. & Stojanovic, I. (2016) Hybrid Evaluatives: In Defense of a Presuppositional Account. Grazer Philosophische Studien, 93(3), 458–88. Conrod, K. (2017) Pronouns and Misgendering. Master’s thesis, University of Washington. Covert, B. (2019) The Myth of the Welfare Queen. New Republic. myth-welfare-queen Crenshaw, K. (2017) On Intersectionality: Essential Writings. The New Press. Croom, A. (2013) How To Do Things With Slurs: Studies in the Ways of Derogatory Words. Language and Communication, 33, 177–204. Cull, M. (2019) Dismissive Incomprehension: A Use of Purported Ignorance to Undermine Others. Social Epistemology, 33(3), 262–71. Darwall, S. (2018) Contempt as an Other-Characterizing, ‘Hierarchizing’ Attitude. In Mason, M. (Ed.), The Moral Psychology of Contempt, Rowman and Littlefeld Publishers, 193–215. Davis, C., & McCready, E. (2020) The Instability of Slurs. Grazer Philosophische Studien, 97, 63–85. Delgado, R. & Stefancic, J. (2004) Understanding Words That Wound. Westview Press. DiFranco, R. (2015) Do Racists Speak Truly? On the Truth-Conditional Content of Slurs. Thought, 4, 28–37. Eckert, P. (2000) Linguistic Variation as Social Practice. Blackwell. Eckert, P. (2008) Variation and the Indexical Field. Journal of Sociolinguistics, 12/4, 453–76. Eklund, M. (2013) Evaluative Language and Evaluative Reality. In Kirchin, S. (Ed.), Thick Concepts, Oxford University Press, 161–81. Ekman, P. & Friesan, W. (1986) A New Pan-Cultural Facial Expression of Emotion. Motivation and Emotion, 10(2), 159–68. Ekman, P. & Heider, K. (1988) The Universality of Contempt Expression: A Replication. Motivation and Emotion, 12, 303–8. Ekman, P. (1994) Strong Evidence for Universals in Facial Expression: A Reply to Russell’s Mistaken Critique. Psychological Bulletin, 115, 268–87. Fischer, A. & Roseman, I. (2007) Beat Them or Ban Them: The Characteristics and Social Functions of Anger and Contempt. Journal of Personality and Social Psychology, 93, 103–15. Freeman, J. (1968) BITCH Manifesto. In Firestone, S. and Koedt, A. (Eds.), Notes from the Second Year: Women’s Liberation, Major Writing of the Radical Feminists. Garfnkel, H. (1956) Conditions of Successful Degradation Ceremonies. American Journal of Sociology, 61, 420–24. Gelman, S. (2003) The Essential Child: Origins of Essentialism in Everyday Thought. Oxford University Press. Hay, R. (2013) Hybrid Expressivism and the Analogy between Pejoratives and Moral Language. European Journal of Philosophy, 21(3), 450–74. Hom, C & May R. (2013) Moral and Semantic Innocence. Analytic Philosophy 54, 293–313.


Robin Jeshion Hom, C. (2008). The Semantics of Racial Epithets. The Journal of Philosophy, 105, 416–40. Hom, C. & May R. (2018) Pejoratives as Fiction. In Sosa, D. (Ed.), Bad Words, Oxford University Press, 108–31. Horn, L. (2018) Nice Words for Nasty Things. In Sosa, D. (Ed.), Bad Words, Oxford University Press, 193–229. Jeshion, R. (2013) Expressivism and the Ofensiveness of Slurs. Philosophical Perspectives, 27, 231–59. Jeshion, R. (2016) Slur Creation, Bigotry Formation: The Power of Expressivism. Phenomenology and Mind, 11, 130–9. Jeshion, R. (2018) Slurs, Dehumanization, and the Expression of Contempt. In Sosa, D. (Ed.), Bad Words, Oxford University Press, 77–107. Kaplan, D. (2005) The Meaning of ‘Ouch’ and ‘Oops’. Unpublished manuscript. Keiles, J. (2018) Reconsidering the Jewish American Princess. Vox. the-goods/2018/12/5/18119890/jewish-american-princess-jap-stereotype. Kennedy, R. (2003) Nigger: The Strange Career of a Troublesome Word. Vintage. Kennedy, R. (2015) Nicknames. In Taylor, J. (Ed.), The Oxford Handbook of the Word, Oxford University Press, 650–668. Khoo, J. (2017) Code Words in Political Discourse. Philosophical Topics, 45, 33–64. Kirk-Giannini, C. (2019) Slurs Are Directives. Philosophers’ Imprint, 19(48), 1–28. Kukla, R. (2018) Slurs, Interpellation, and Ideology. The Southern Journal of Philosophy, 56, 7–32. Langton, R. (1993) Speech Acts and Unspeakable Acts. Philosophy and Public Afairs, 22, 293–330. Langton, R. (2009) Sexual Solipsism: Philosophical Essays on Pornography and Objectifcation. Oxford University Press. Langton, R. (2012) Beyond Belief: Pragmatics in Hate Speech and Pornography. In Maitra, I. and McGowan, M. (Eds.), Speech and Harm: Controversies Over Free Speech, Oxford University Press, 72–93. Langton, R. & West, C. (1999) Scorekeeping in a Pornographic Language Game. Australasian Journal of Philosophy, 77, 303–19. Le Page, R. & Tabouret-Keller, A. (1985) Acts of Identity: Creole-Based Approaches to Language and Ethnicity. Cambridge University Press. Lukatsky, D. & Toback, S. (1982) The Jewish American Princess Handbook. Printed Matter Pub, Co. Maitra, I. (2012) Subordinating Speech. In Maitra, I. and McGowan, M. (Eds.), Speech and Harm: Controversies Over Free Speech, Oxford University Press, 94–120. Manne, K. (2018) Down Girl: The Logic of Misogyny. Oxford University Press. Marques, T. & Garcia-Carpintero, M. (2020) Really Expressive Presuppositions and How to Block Them. Grazer Philosophische Studien, 97(1), 138–58. Mason, M. (2003) Contempt as a Moral Attitude. Ethics 113, 234–272. Mason, M. (2016) Contempt: At the Limits of Reactivity. In Mason, M. (Ed.), The Moral Psychology of Contempt, Rowman & Littlefeld Publishers, 173–192. McGowan, M. (2012) On ‘Whites Only’ Signs and Racist Hate Speech: Verbal Acts of Racial Discrimination. In Maitra, I. & McGowan, M. (Eds.), Speech and Harm: Controversies Over Free Speech, Oxford University Press, 121–47. McGowan, M. (2019) Just Words: On Speech and Hidden Harm. Oxford University Press. Mendelberg, T. (2001) The Race Card: Campaign Strategy, Implicit Messages, and the Norm of Equality. Princeton University Press. Miller, W. (1997) The Anatomy of Disgust. Harvard University Press. Morris, D. (1978) Manwatching. Panther Books. Neufeld, E. (2019) An Essentialist Theory of the Meaning of Slurs. Philosophers’ Imprint, 19(35), 1–29. Nunberg, G. (2018) The Social Life of Slurs. In Fogel, D., Harris, D., & Moss, M. (Eds.) New Work on Speech Acts. Oxford University Press, 237–95. Potts, C. (2005) The Logic of Conventional Implicatures. Oxford University Press. Potts, C. (2007) The Expressive Dimension. Theoretical Linguistics, 33, 165–98. Potts, C. & Kawahara, S. (2004) Japanese Honorifcs as Emotive Defnite Descriptions. In Proceedings of SALT XIV, 253–270. Pullum, G. (2018) Slurs and Obscenities. In Sosa, D. (Ed.), Bad Words, Oxford University Press, 168–92. Rappaport, J. (2019) Communicating with Slurs. The Philosophical Quarterly, 69, 795–819. Richard, M. (2008) When Truth Gives Out. Oxford University Press. Roseman, I. (2018) Rejecting the Unworthy: The Causes, Components, and Consequences of Contempt. In Mason, M. (Ed.), The Moral Psychology of Contempt, Rowman & Littlefeld Publishers, 107–30.


Varieties of Pejoratives Saka, P. (2007) How to Think About Meaning. Springer. Saul, J. (2018) Dogwhistles, Political Manipulation, Philosophy of Language. In Fogel, D., Harris, D., & Moss, M. (Eds.), New Work on Speech Acts, Oxford University Press, 360–83. Scherer, K. (2003) Vocal Communication of Emotion: A Review of Research Paradigms. Speech Communication, 40, 227–56. Scherer, K. (2013) Emotion in Action, Interaction, Music, and Speech. In Arbib, M. (Ed.), Language, Music, and the Brain: A Mysterious Relationship, MIT Press, 107–39. Schlenker, P. (2007) Expressive Presuppositions. Theoretical Linguistics, 33(2), 237–45. Sendenlmeier, W., Stefen, I., & Bartels, A. (2016) Pejorative Prosody. In Finkbeiner, R., Meibauer, J., & Wiese H. (Eds.), Pejoration, John Benjamins, 1–18. Sennet, A. & Copp, D. (2019) Pejorative Verbs and the Prospects for a Unifed Theory of Slurs. Analytic Philosophy, 61(2), 130–51. Stanley, J. (2015) How Propaganda Works. Princeton University Press. Swanson, E. (2020) Slurs and Ideologies. Forthcoming in Celikates, R., Haslanger, S., & Stanley, J. (Eds.), Analyzing Ideology: New Essays, Oxford University Press. Tirrell, L. (1999) Derogatory Terms: Racism, Sexism, and the Inferential Role Theory of Meaning. In Hendricks, C. & Oliver, K. (Eds.), Language and Liberation: Feminism, Philosophy, and Language, SUNY Press, 41–79. Tirrell, L. (2018) Authority and Gender: Flipping the F-Switch. Feminist Philosophy Quarterly, 4(3), 1–32. Väyrynen, P. (2013) The Lewd, the Rude, and the Nasty. Oxford University Press. Väyrynen, P. (2016) Thick Ethical Concepts. In Zalta, E. (Ed.), The Stanford Encyclopedia of Philosophy. Waldron, J. (2012) The Harm in Hate Speech. Harvard University Press. Wierzbicka, A. (1992) Semantics, Culture, and Cognition: Universal Human Concepts in Culture-Specifc Confgurations. Oxford University Press. Zangwill, N. (1995) The Beautiful, the Dainty, and the Dumpy. British Journal of Aesthetics, 35(4), 317–29.



1 Introduction For such a seemingly pervasive phenomenon, the concept of a microaggression has only come into relief within the last 50 years. The goal of this chapter is to explain why microaggressions have suddenly become a more remarkable feature of our social and political environment. Psychologist Chester Pierce coined the term in 1970, focusing specifcally on anti-Black microaggressions. Pierce’s analysis of the phenomenon shows how microaggressions became a more pressing moral problem during a time in U.S. history in which (1) overt forms of racial aggression were increasingly outlawed and/or deemed less socially acceptable, and (2) there was new emphasis on increasing racial diversity within predominately white institutions and professions. The concept of microaggressions appears at a moment of transformation in the mechanisms of structural oppression where overt behaviors start to become less legally and socially acceptable, and this occurs alongside a new emphasis on creating diverse professional and educational environments. In tracing the history of this concept, I argue for an adaptive theory of microaggressions. Following Michelle Alexander’s adaptive theory of oppression (2010), microaggressions, as a mechanism of structural oppression, represent an adaptation from overt to covert actions, and, as a form of social practice, can be adaptive in their own right. While microaggressions are not a new social phenomenon, it is unsurprising that the concept arises alongside a broader shift within the United States toward more covert forms of oppressive social practices, mirroring a similar shift, for example, from slurs to code words. In this way, microaggressions are more likely to fourish in environments when overt tactics of oppression are too socially, politically, or legally costly to ofenders. The layout of my chapter is as follows. Section 2 provides an overview and analysis of Chester Pierce’s original theory of anti-Black microaggressions, connecting his account to Michelle Alexander’s adaptive theory of oppression in the case of American white supremacy. Section 3 ofers a brief survey of the current literature on microaggressions, focusing on a fundamental shift between Pierce’s original theory and current analyses of microaggressions. Pierce argues that once targets become aware of microaggressions as an ofensive mechanism, they can more easily identify particular instances and are thus able to better defend themselves against their harms. However, contemporary research frames targets’ perception in terms of attributional ambiguity. Attributional ambiguity occurs when it is unclear to targets, ofenders, and bystanders whether a microaggression has occurred. Targets experience this attributional ambiguity as a unique form of epistemic uncertainty. That is, unlike Pierce’s framework for perception, general awareness of microaggressions more frequently results in epistemic uncertainty than easy identifcation. Finally, 232

Problem of Attributional Ambiguity

Section 4 argues that this shift towards increasingly attributionally ambiguous microaggressions ought to be understood as an insidious adaptation in the social phenomenon. Microaggressions already mark a larger shift toward more covert tactics. In order for microaggressions to remain an efective tacit, they must also become increasingly covert, making it harder for targets to perceive particular instances. If one goal of research on microaggressions is to better identify types of microaggressions and prevent and/or mitigate their harms, then the rise in attributionally ambiguous microaggressions ought to be of concern to theorists because of the ways this adaption undermines this goal.

2 Microaggressions as an Ofensive Mechanism The term ‘microaggression’ was frst coined by Chester Pierce in the 1970s to describe a particular class of anti-Black, racist behaviors on the part of white Americans (1970). Drawing on his own experiences as an assistant football coach at Harvard, Pierce likens race relations in America to a football game in which the ofensive mechanisms of white people force Black people into a constant state of defense (1970, 1989). For Pierce, ofensive mechanisms are the myriad ways white people (individually and collectively) maintain a racial hierarchy between Black and white Americans psychologically, physiologically, and materially.1 As a form of ofensive mechanism, microaggressions are “subtle, minor, stunning, automatic assaults… by which Whites stress Blacks unremittingly and keep them on the defensive, as well as in a psychologically reduced condition” (1989; 308). Microaggressions are stress-inducing interruptions into people’s daily lives, draining their energy through repeated occurrence and putting targets in a constant state of deference or hypervigilance2: “by monopolizing our perception and action through regularly irregular disruptions, [microaggressions] contributed to relative paralysis of action, planning, and self-esteem” (1989; 308). Microaggressions function as a covert way to maintain racial hierarchy by repeated, invasive but ‘micro’ interruptions. Broadly, ofensive mechanisms are the processes by which a structural form of oppression is maintained. Microaggressions, as such a mechanism, constitute a set of micro social practices that reinforce or contribute to one or more forms of oppression, taking the form of violent and aggressive learned behavior on the part of the dominate social group. This learned behavior can be conscious, unconscious, or pre-conscious. However, it is often expressed automatically because of repetitive training and correction (Pierce 1970; 271–2). That is, football players practice their ofensive maneuvers, correcting their behavior in accordance with their coach’s feedback, so enacting these maneuvers during a game comes easily and automatically. Ofensive and defensive maneuvers work similarly in the case of oppressor-oppressed dynamics between social groups. The aim of ofensive mechanisms is to maintain control of the game in order to win or, in this case, dominate.3 Thus, microaggressions work because they force the oppressed to play defense, failing to see the large-scale outcome of these practices. In this case, the outcome is the forms of cumulative harm that afect both the fourishing of the oppressed group as a whole and individual members of that group. If Pierce is correct in his account of ofensive mechanisms, then it follows that these mechanisms are subject to change based on what the most efective ofensive plays for maintaining control of the game are at any particular time. In this way, ofensive mechanisms are subject to change in light of larger social and political shifts. Given that the goal of these ofensive mechanisms is about preserving a pre-existing social hierarchy and they are subject to change in order to do so most efectively within a particular context, Michelle Alexander’s adaptive theory of oppression can provide a valuable way of understanding how such shifts work. In The New Jim Crow (2010), Alexander argues that white supremacy, as a form of structural oppression, is adaptive: “Since the nation’s founding, African Americans repeatedly have been controlled through institutions such as 233

Christina Friedlaender

slavery and Jim Crow, which appear to die, but then are reborn in new form, tailored to the needs and constraints of the time” (21). With the apparent death of one institution, there follows a period of heightened backlash through which new institutions for social control emerge. Alexander notes that their emergence might appear sudden, but their birth is set in motion during the power struggles of maintaining the previous form of control.4 What is uniquely valuable about Alexander’s account is she argues for an adaptive theory of structural oppression in which such adaptations can be increasingly insidious: “as systems of control have evolved, they have become perfected, arguably more resilient to challenge, and thus capable of enduring for generations to come” (22). These adaptations are also insidious because those who live under new, adaptive mechanisms have a harder time perceiving them,5 which gives the appearance of moral and social progress. In the case of white supremacy, those who remain largely unharmed by and/or beneft from racial oppression often fail to see these adaptations, even as they are the ones enacting them. This, as Pierce notes, leads white people in particular to misperceive the extent to which moral and social progress had occurred: In almost any black-white negotiation each participant views things diferently, depending on whether he is white (the ofender) or black (the ofended). For instance, the psychological hallmark of racism is the altogether too well-known tendency for whites to congratulate themselves, before a black [person], concerning what marvelous “progress” is being made. To the perception of the white ofender, this is true and reasonable. From the vantage point of the black ofended, however, this is both untrue and unreasonable. (1970; 267) Alexander argues that claims to moral and social progress must be framed in light of these adaptions, which function by way of “preservation through transformation” (21).6 This is neither to claim that all forms of social control are essentially the same nor that no moral or social progress has occurred. Rather, this is a call for us to examine the extent of their diferences without exaggeration, noting the appearance of diference in some cases is more superfcial than not (200–4). Put in Pierce’s terms, there is a transition in the ofensive mechanisms in order to maintain control of the game. The mechanisms of oppression adapt in response to attempts to implement a social, moral, and political change, but it does not follow that these new mechanisms are necessarily less harmful than the mechanisms of the former social order. It is not surprising that the concept of a microaggression comes into relief in the 1970s, during a larger period of social and political upheaval in which the legal status and social acceptability of overt forms of oppressive violence and aggression started to shift. Maintaining pre-existing forms of social control thus required an increase in covert ofensive mechanisms, ones that provided the ofender with the cover of plausible deniability. In the case of racism, microaggressions start to emerge as a major ofensive mechanism alongside both the end of legal segregation and an increase in calls for racially diverse institutions. That is, microaggressions are a transformation of previous modes of maintaining a racial hierarchy that preserve its goal of white supremacy.7 Drawing on his own experiences as a Black psychiatrist within a predominately white profession (1989), Pierce argues that, in practice, integrated environments often place an additional burden on Black people because they are efectively tasked with maintaining “unity in diversity.”8 Here, ‘diversity’ is defned as non-white and white people actively seek to ‘increase diversity’ in predominately white spaces (often for their own beneft) but place no demand on themselves or their institutions to change.9 Maintaining this pseudo-unity tasks Black people with managing white people’s responses to their presence within historically white spaces and with extra labor related to further increasing this so-called ‘diversity.’ In this way, microaggressions work as an ofensive mechanism through constant distraction, by demanding the 234

Problem of Attributional Ambiguity

attention, time, and energy of the target, arising frequently within the demand for unity (1989). One might liken Pierce’s analysis of microaggressions here to Toni Morrison’s 1975 lecture in which she argues that racism itself functions as a form of distraction by masking what it really is: a power struggle (Morrison et al.). And this distraction, as Morrison explains: “It keeps you from doing your work. It keeps you explaining over and over again, your reason for being” (Ibid). Both Pierce and Morrison characterize this unending distraction as an ofensive, warlike tactic, serving to grind people down by constantly interrupting their lives, demanding a response, and exploiting their labor by placing the burden of maintaining “unity in diversity” on them.10 Thus, once targets are made aware of this ofensive mechanism, they are in a better position to recognize and, ultimately, refuse these distractions. This refusal, especially at a collective level, is essential for breaking up the power dynamic in play (Pierce, 1970; 279–82). Pierce’s decision to develop a psychological theory of ofensive mechanisms rests on his concern that simply focusing on how individuals must act defensively fails to adequately capture the source of their reaction, the ofensive players, and, most importantly, the ofensive-defensive dynamic itself (1970). Understanding this dynamic is essential for addressing the outcome of ofensive mechanisms, in part by providing tools for how to play better defense when necessary or how to refuse to participate in the game itself. On the one hand, Pierce argues that microaggressions are an efective ofensive mechanism because the attempt to manipulate the behavior of targets is only a successful tactic if targets generally comply, which they are more likely to do when they fail to see they are being forced into a defensive position in the frst place. Since microaggressions produce sustained, elevated stress levels, targets often develop adaptive techniques, which can “dilute, postpone, or defect stress” (1995; 277). Yet, such techniques merely “allow victims to coexist with their oppressors, even at the cost of participating in and permitting much of their own catastrophe” (Ibid.). On the other hand, Pierce argues that an increase in both the individual and collective awareness of this dynamic makes it much easier for targets to identify particular instances, in part because they become largely obvious to the targets. Perpetrators, however, will have a signifcantly harder time seeing this general dynamic and its particular manifestations. Pierce argues that white people, for instance, have a “perceptual illness” (1970; 267), which functions as a form of horrifc ignorance. He raises this point frst when noting white people’s misperception about racial progress, but more broadly puts into sharp contrast white people’s obliviousness of their behavior and its meaning and the obviousness of this behavior to Black people, including the automaticity with which white people often commit microaggressions. This contrast appears again when Pierce notes the shocking gap between the harm caused by microaggressions and the obliviousness with which white people commit them: “It is difcult, if not impossible, for a black [person] to understand how a white, particularly a privileged white, can exhibit ofensive micro-aggressions without considering himself a murderer” (268). In later work, Pierce likens both racist and sexist microaggressions to a form of torture and terrorism. Being in a ‘diverse’ setting that exists within a larger structure of oppression is efectively like learning to how to live alongside one’s torturers, but, in this case, the tortures are largely oblivious to the fact that they are torturers, seeing themselves as innocent: …most Whites and males are largely oblivious to the stress they create merely by their potential for exerting power. Almost all oppressors are unmindful of the withering efects of cumulative, individual, and collective microaggressions toward victims, even honestly believing that they themselves have inficted no damage upon socially devalued persons. (1995; 277) It is worth noting that Pierce’s contrast between obviousness and obliviousness mirrors Baldwin’s (1963) discussion of crime and innocence. The obliviousness of white perpetrators is a horrifc 235

Christina Friedlaender

kind of innocence, as Baldwin writes: “it is the innocence [of whites] which constitutes the crime… [White people] are, in efect, still trapped in a history which they do not understand; and until they understand it, they cannot be released from it” (20, 22). In this way, microaggressions are a transformation of more overt ofensive mechanisms but ofenders fail to see both the meaning of their own behavior and the history that has shaped it. Thus, in looking back at overt ofensive mechanisms, ofenders are more likely to see themselves as more socially and politically progressive in comparison, failing to see how the transformation from overt to covert practices can insidiously preserve the function of overt mechanisms as well as how their own behavior betrays their stated or conscious commitment to progressive values.

3 Te Current Landscape Since Pierce’s initial use in the case of anti-Black racism, the concept of microaggressions has been extended to include (but is not limited to): other forms of racism, sexism, heterosexism, cissexism, classism, ageism, ableism, fatphobia, colorism, settler-colonialism (or anti-indigeneity), Islamophobia, xenophobia, and anti-Semitism. The term ‘microaggressions’ entered more popular discourse in 2007, thanks to psychologist Derald Wing Sue. Sue’s analysis and taxonomy of microaggressions has profoundly shaped the current landscape of research on microaggressions, specifcally due to the widespread adoption of his taxonomy. True to Pierce’s initial usage, targets of microaggressions are still largely taken to be members of groups who are subject to structural oppression or identity-based marginalization.11 This is also true of Pierce’s focus on the role of automaticity, as much of the contemporary literature frames microaggressions as largely unintentional or unconscious and as the result of implicit bias. The goal of this section is to provide a brief overview of the contemporary literature, focusing on a fundamental shift between Pierce’s original theory and the current landscape: the ease with which targets perceive microaggressions. I frst ofer a brief overview of the ways microaggressions harm oppressed groups and their individual members, followed by a discussion of the social ontology of microaggressions, which concerns the defnition of microaggressions, diferences between micro- and macroaggressions, and the criteria for determining whether a particular act or statement was, in fact, a microaggression.

3.1 Harms Microaggressions can cause a number of diferent kinds of harm as well as compound pre-existing structural inequalities. Such harms can be categorized broadly as either individual or structural. Harms to the individual include both the short-term and long-term efects of microaggressions. These include harms to one’s psychological or physical health,12 one’s ability to function and fourish within an economic system, including one’s ability to exercise choice over the kind of work one does and the workplace itself,13 and harms to one’s educational development and intellectual growth.14 When a microaggression occurs, a target might experience immediate psychological and epistemic harms. Such harms largely impact one’s relation to self and one’s relation to context in which the microaggression occurred. Psychological harms negatively impact one’s sense of self-worth, increase feelings of self-doubt (that can be akin to imposter syndrome), causing the target to feel contextually excluded or unwelcome. For example, if a male philosophy professor commits a sexist microaggression during class, it might impact women’s participation or confdence within the classroom as well as increase feelings of exclusion for an already underrepresented group within philosophy. Once a microaggression occurs, targets are further burdened with deciding whether to point it out, reporting that they often suppress their reaction, which comes with its own set of cognitive and emotional costs (Wang et al., 2011). This is because, frst, when a target calls out a microaggression, they are sometimes told they are over-reacting, 236

Problem of Attributional Ambiguity

exaggerating, or falsely perceiving oppression where there is none (Sue, 2010a). If a microaggression has occurred, such reactions constitute epistemic microaggressions in their own right, i.e., microaggressions that constitute a form of testimonial injustice (Freeman and Stewart, 2019) and open the door to further epistemic gaslighting (Friedlaender and Ivy, 2020). Second, having to frequently second-guess whether another’s behavior was simply rude or directed at them in virtue of their membership in an oppressed social group can put one in a state of hypervigilance. Because microaggressions do not occur in isolation, the cumulative harm from microaggressions is often the most concerning for individuals. Moreover, continuous experiences of microaggressions within a particular context might cause individuals to withdraw or leave entirely. For example, the experience of microaggressions in clinical medicine can undermine patient trust in physicians, especially when epistemic microaggressions result in misdiagnosis or downplaying pain and other symptoms. It is not surprising that, as a result, some patients stop going to the doctor entirely (Freeman and Stewart, 2018). In the context of work, microaggressions can cause targets to withdraw from certain activities that are benefcial career-wise (e.g., informal networking over drinks after a department talk) or simply choose a diferent career path (e.g., leaving philosophy for a more welcoming career path). Structural harms constitute the larger-scale outcomes created by microaggressions, which generally compound pre-existing forms of inequality (Brennan, 2016). Microaggressions in work and education can contribute to institutional demographic disparities and thus further compound economic inequality. For example, the intersection of sexist and racist microaggressions within philosophy might contribute to fewer women of color in the discipline overall due to a hostile professional climate.15 The cumulative harm experienced by individuals as members of social groups can also contribute to a disparity in health outcomes, including, for example, the disparity in mortality rates between Black and white Americans (Pierce, 1970).

3.2 Social Ontology Derald Wing Sue’s defnition and taxonomy of microaggressions has become standard in the literature.16 On Sue’s model, microaggressions are “the brief and commonplace daily verbal, behavioral, and environmental indignities, whether intentional or unintentional, that communicate hostile, derogatory, or negative… slights and insults to the target person or group” (Sue et al., 2007; 273). Sue’s (2010a) taxonomy ofers three subsets of microaggressions: microassaults, microinsults, and microinvalidations. Microassaults are conscious and deliberate. They are characterized by their overt or explicit nature, such as the use of a slur, display of a hate symbol, or intentionally ignoring members of marginalized groups within a shared space, such as a classroom. By contrast, microinsults and microinvalidations are characterized as largely unintentional and unconscious. Microinsults are subtle forms of insulting or derogatory communication towards the targeted group, such as expressing surprise that a woman is skilled in mathematics. Similarly, microinvalidations subtly ignore, downplay, or express unwarranted skepticism toward the experiences, thoughts, and feelings of members of a targeted group. Sue also argues that microaggressions can be verbal, nonverbal, or environmental (Ibid.). Environmental microaggressions do not usually have a single or clear perpetrator. For example, if a philosophy department’s decorations are mostly pictures of white people and men and these decorations have accumulated over the course of many years, there is no single perpetrator for what is a collectively harmful outcome. While Sue’s defnition and taxonomy of microaggressions has been widely adopted – both in academic literature and popular culture – it is also increasingly the target of critique (see, e.g., Minikel-Lacocque, 2013; Wong et al., 2013; McTernan, 2017; Freeman and Stewart, 2018; Friedlaender and Ivy, 2020). Much of the controversy focuses on Sue’s classifcation and inclusion of microassaults, e.g., a man’s intentional use of a sexist slur or a neo-Nazi’s intentional display of a 237

Christina Friedlaender

swastika. Since microassaults are overt and intentional, Sue’s taxonomy treats them as a deviation from the standard forms of microaggression (i.e., microinsults and microinvalidations, which are largely covert and unintentional), arguing that microassaults are similar to more overt forms of oppressive social practices. The predominant critique of microassaults is that they are not appropriately micro and should be recategorized as macroaggressions (see, e.g., Wong et al., 2013). However, simply rejecting microassaults from the taxonomy is not sufcient because there is a two-fold problem with their inclusion. First, as Rachel McKinnon and I argue, because Sue’s widely adopted taxonomy treats them as a deviation, much of the ensuing research on microaggressions has focused on the standard forms, which are regarded to be largely unintentional ( forthcoming). Moreover, given that the predominant moral concern about microaggressions is (as the story goes) that they are largely perpetrated by those who are consciously committed to more progressive or egalitarian social and political values (Sue, 2010a), the hyper-focus on unintentional microaggressions is unsurprising, in part because it seems that our moral analysis of and practical solutions for unintentional kinds might difer from intentional kinds. Because of this focus on unintentional types, the causal story for microaggressions has centered on the role of implicit attitudes and biases (see, e.g., Sue, 2003, 2010a). However, this hyper-focus on unintentional microaggressions means that intentional microaggressions are often included only in a passing reference to Sue’s taxonomy and are rarely the subject of investigation. The ensuing assumption that microaggressions are a largely unintentional phenomenon is particularly concerning because there is no clear evidence to suggest this is accurate.17 The second problem is that even if there is good reason to exclude microassaults from the taxonomy, it does not follow that an updated taxonomy ought to retain the assumption that microinsults and microinvalidations are largely unintentional. Therefore, merely rejecting microassaults is not sufcient. Theorists of microaggressions will also need to revise the role of intentionality.18 Both Pierce and Sue frame microaggressions as largely unintentional and unconscious. Sue takes up Pierce’s claim that microaggressions seem automatic by linking microaggressions with implicit bias. However, this assumption seems largely based on Pierce’s initial claim that, from the target’s perspective, perpetrators engage in automatically expressed microaggressions and are largely oblivious to having done so. While Sue takes up this assumption in connecting microaggressions with implicit bias, he does not retain Pierce’s perceptual framework in which microaggressions are obvious to targets but perpetrators are oblivious of their behavior. Sue instead ofers a diferent perceptual framework for microaggressions: attributional ambiguity. Rather than seeing microaggressions as obvious, targets experience attributional ambiguity when they are uncertain of whether a microaggression has just occurred (e.g., “Did that just happen? Was that what I think it was?”). There are several layers to attributional ambiguity. First, since microaggressions are subtle and can occur quickly in interpersonal interactions, it can be hard to fully perceive what has happened, leaving the target second-guessing whether they fundamentally misperceived the behavior (e.g., did they simply mishear what the person said?). Even if the target did not misperceive the behavior at this fundamental level, there is a second level of ambiguity in distinguishing microaggressions from other sorts of covertly rude or disrespectful behavior. In such cases, there are plausible alternative explanations for the behavior and no clear evidence to suggest which might be correct. For example, if a male supervisor I just met ignores my comments in a meeting, I might not have sufcient evidence to determine whether he treats most people that way regardless of gender or if his ignoring me is due to my gender.19 Third, in other cases, the problem is not distinguishing microaggressions from other forms of disrespectful behavior, but rather the behavior has two plausible explanations: either it is genuinely innocuous or it is a microaggression (and is therefore also disrespectful). For example, certain forms of forgetfulness might be innocuous (e.g., forgetting or misremembering someone’s name) but other forms might constitute microaggressions (e.g.,


Problem of Attributional Ambiguity

the tendency to forget names associated with certain cultural or ethnic groups, misremembering names in a stereotyped way, or confusing the names of members of the same group). Sue argues that attributional ambiguity makes microaggressions potentially more stressful than macroaggressions because the latter and its causes are generally “clear and obvious” to the targets; whereas microaggressions are perceptually “nebulous,” leaving the target to second-guess what has happened and how to respond to an event for which others are inclined to accept an alternative, innocuous explanation or have failed to perceive the behavior altogether (Sue, 2010a; 106). Following Sue, much of the current research maintains this focus on the feature of attributional ambiguity. For instance, Saba Fatima argues that attributional ambiguity creates a unique form of epistemic uncertainty in targets (2017). This uncertainty is unique to targets because they often must seek out ‘sanity checks,’ i.e., asking another to confrm whether one’s perception is correct (usually the ‘sanity check’ is sought from other members of the target’s group) (Sue, 2010a; 74–5; Nadal, 2011; 332). This shift concerning perception connects to the question of how one determines whether a microaggression has, in fact, occurred. For Pierce, identifcation of individual microaggression was not a signifcant problem. Once a person is aware of the general ofensive mechanism, they are in a better position to identify particular cases. For Sue, on the other hand, general awareness of microaggressions does not necessarily increase one’s ability to identify particular instances, but it does seem to increase the target’s epistemic uncertainty. That is, if microaggressions are harder to detect, awareness of the general phenomenon increases targets’ experiences of epistemic uncertainty. Unsurprisingly, much of the contemporary analysis of microaggressions thus far has focused on how to determine whether a microaggression has occurred. For psychologists, answering this question is to the beneft of targets, specifcally in learning how to best respond to their occurrence and minimize their harm. For philosophers, answering this question is about moral analysis. That is, if microaggressions are a moral wrong, then identifcation seems crucial for determining what our moral response ought to be. There are two primary approaches to the question of determination: (1) structuralist, mindindependent accounts, and (2) non-structuralist, mind-dependent accounts.20 On a structuralist account, microaggressions are the outcome of certain causal processes, but determining whether the causal structure in question has occurred requires evidence that is largely inaccessible. One response has been to develop a structuralist account of microaggressions that still classifes them as mind-independent (see, e.g., Huber and Solorzano, 2015; McTernan, 2017; Friedlaender, 2018; Tremain, 2019). For example, if microaggressions occur as the micro-expressions of a broader form of structural oppression and that form of oppression exists independently of any specifc oppressed person’s perception of its existence (e.g., women are objectively oppressed even if individual women deny they experience oppression qua gender), then microaggressions exist independently from any particular individual’s perception in a similar way (Friedlaender, 2018). On a structuralist model, microaggressions are a set of social practices that contribute to and reinforce the unjustifed hierarchies and inequalities constituted by structural oppression. A structuralist account aims to identify possible microaggressions by noting that individual microaggressions are often individual instances of larger patterns of verbal and nonverbal behavior. As Pierce (1970) notes, targets frequently experience similar kinds of microaggressions. When a particular instance is similar to a pattern of behaviors that we already identify as microaggressions, we have good reason to identify it as a microaggression as well, and can do so without having to determine either what went on in the mind of the perpetrator or exactly how a particular environmental microaggression arose. On the other hand, if microaggressions are constituted by the subjective experience of the target, then we neither run into the same problem concerning evidence for cognitive mechanisms


Christina Friedlaender

nor are we forced to provide an equally complex causal story about the mechanisms of structural oppression. Sue argues that microaggressions are subjective insofar as a target determines whether a microaggression has occurred (Sue et al. 2007; Sue, 2017). Taking up a non-structuralist approach, Rini (manuscript) reinterprets Sue’s claim as one in accordance with standpoint theory’s analysis of group-level knowledge creation, arguing that the experience itself can serve as the basis for a form of knowledge or evidence.21 Because members of oppressed groups have continuous experiences of oppression, they are more likely to be able to notice forms of discrimination related to their oppression and are more likely to be accurate in attributing the event or behavior as being causally related to mechanisms of oppression. That is, members of oppressed groups are more likely to accurately identify when a microaggression has occurred. In the fnal section, I retain my previous account of microaggressions as a set of social practices that occur in accordance with broader mechanisms of oppression (2018), building in Alexander’s adaptivity component. Standpoint theory ofers a way to produce knowledge claims about these patterns of micro-behaviors, aiding in reliable identifcation for particular instances. However, my worry is that the increasingly common experience of attributional ambiguity makes this knowledge production harder in the frst place.

4 Attributional Ambiguity as an Adaptation This shift toward increasingly attributionally ambiguous microaggressions ought to be understood as an insidious adaptation in the social phenomenon.22 Microaggressions already mark a larger shift in ofensive mechanisms moving toward more covert tactics. In order for microaggressions to continue working well as an ofensive mechanism, they must also become increasingly covert, making it harder for the targets to perceive particular instances. This is concerning for two reasons. First, Pierce claims targets must be able to identify particular instances in order to better defend themselves against the harms of microaggressions. If attributional ambiguity makes identifcation harder, then this is not only a problem for individual targets, but it also prevents the process of group-level knowledge production on which standpoint theory relies. This makes it harder to notice patterns of similar behavior as well as identify specifc instances of that pattern. Second, attributional ambiguity makes it increasingly easier for people to intentionally commit microaggressions, which calls into question the standard narrative about who commits microaggressions and why. Following the frst claim, if microaggressions are the micro-social practices that track border forms of structural oppression, then standpoint theory provides a way to produce knowledge claims about (1) patterns of similar micro-behavior that qualify as microaggressions, and (2) the ability to rely on those patterns for purposes of particular identifcation. However, attributional ambiguity makes it signifcantly harder to achieve a group knowledge claim in this case because such claims require participating individuals to be able to notice their occurrence in the frst place. The problem of attributional ambiguity is that it can distort the target’s perception, potentially leaving them with the belief that a microaggression might not have occurred, and this information does not get included in the production of knowledge in this case. Furthermore, ofenders can respond to a target after an occurrence in ways that subject the target to a form of epistemic gaslighting, which can alter the target’s experience of uncertainty altogether, leaving the target to believe their initial experience was simply wrong. Thus, attributional ambiguity disrupts the capacity for members of oppressed groups to engage in group-level knowledge production, interfering with the type or amount of knowledge about microaggressions and the capacity to identify particular instances. This insidious adaption retains the harmful nature of microaggressions while leaving targets increasingly unable to adequately defend themselves from these harms. 240

Problem of Attributional Ambiguity

Following Morrison’s and Pierce’s analysis of distraction, this is further concerning because when microaggressions constantly interrupt targets’ lives, attributional ambiguity makes it harder for targets to refuse these distractions. In the case of intentional microaggressions, this shift makes them more insidious. The harder it is to determine whether one has occurred, the more cognitive energy is drained from the target. That is, the potentiality of a microaggression becomes a source of distraction in itself and the fact that there is often not an open discussion of what occurred not only leaves the target to second-guess what happened without necessarily any resolution but opens the door to epistemic gaslighting. That is, intentional microaggressions are a distraction based on the target’s epistemic uncertainty. Epistemic gaslighting hijacks this uncertainty. Gaslighting causes targets to second-guess their perception of reality, making it harder for targets to identify these distractions because gaslighting can reframe them as non-distractions. This makes it increasingly harder overall for targets to see these moments of distraction and makes it harder to refuse them. That is, it is hard to refuse a distraction if one is unclear whether it should be classifed as a distraction in the frst place. In this way, the potential for microaggressions causes a second-order distraction (i.e., second-guessing) that is independent of microaggressions as frst-order distraction (i.e., the microaggression itself ). One can begin to see why this is cause for concern. To return to Morrison’s lecture: You don’t waste your energy fghting the fever; you must only fght the disease. And the disease is not racism. It is greed and the struggle for power. And I urge you to be careful. For there is a deadly prison: the prison that is erected when one spends one’s life fghting phantoms, concentrating on myths, and explaining over and over to the conqueror your language, your lifestyle, your history, your habits. (1975) By reading Morrison and Pierce together, one can see that ofensive mechanisms are ultimately about a struggle for power, about maintaining hierarchy and inequality. Microaggressions are a tacit of distraction but one that also serves to reinforce the prevailing hierarchy within the habits of both perpetrators and targets. Unintentional perpetrators remain trapped in their obliviousness, enacting, as Baldwin argued, a history they fail to understand but serve as a fundamental source of self hood and identity. Intentional perpetrators, who see the game for what it is, can further exploit this ofensive mechanism to their own end, draining their targets through constant distraction and contributing to ongoing structural inequalities between oppressors and the oppressed. When targets are able to easily identify microaggressions, refusing them as a source of distraction becomes easier, even if the perpetrators remain oblivious to the refusal. As microaggressions become increasingly attributionally ambiguous, it becomes harder for the targets to, as Morrison writes, distinguish the fever from the disease. In this way, attributional ambiguity causes a second-order distraction (i.e., second-guessing) that is independent of microaggressions as frst-order distraction, i.e., the microaggression itself, and this second-order distraction is harder to see as a distraction in its own right. Second, attributional ambiguity opens the door to the feature of plausible deniability. If called out, ofenders can exploit the terms of ambiguity to argue for a plausible alternative explanation for their behavior, which allows them to potentially avoid moral reproach or responsibility for their actions. Philosophical literature on slurs and propaganda helps to show why this is the case. The insidiousness of propaganda is a function of the process by which problematic social meanings are smuggled into the not-at-issue content of seemingly neutral or innocuous words and phrases, making it difcult to call attention to the problematic content and masking the linguistic harm taking place (Stanley, 2015). In this way, the attributional ambiguity of microaggressions is akin to propaganda or dog whistles. As Alexander (2012) argues, overtly neutral language can be covertly 241

Christina Friedlaender

racist or sexist, and the guise of neutrality provides the speaker with plausible deniability. For instance, in response to Ronald Reagan’s call for states’ rights: [Reagan’s] critics promptly alleged that he was signaling a racial message to his audience, suggesting allegiance with those who resisted desegregation, but Reagan frmly denied it, forcing liberals into a position that would soon become familiar—arguing that something is racist but fnding it impossible to prove in the absence of explicitly racial language. (ibid., 48) Slurs lack this guise of neutrality, even if the user tries to claim they did not know the word was considered a slur, or if the user is not necessarily culpable for the use, e.g., as in the case of appropriation of a slur by the targeted group (see, e.g., Anderson and Lepore, 2013a, 2013b). Microaggressions are, in one sense, like Reagan’s call for states’ rights. The ofender knowingly trades on the ambiguity in order to avoid reproach by ensuring that the target must second-guess the meaning of the action, and the target will likely be met with scrutiny if they attempt to call attention to it. After all, bystanders are also likely to experience this ambiguity insofar as they might believe the perpetrator’s alternative explanation for the behavior and deny the reality of the target’s experience (Friedlaender and Ivy, 2020). However, because the feature of attributional ambiguity tracks a target’s experience of epistemic uncertainty, microaggressions are concerning in a diferent way. Dog whistles, code words, and propaganda are often not as experientially ambiguous as contemporary microaggressions. When an article in Breitbart uses the term ‘globalists,’ for instance, it is likely obvious to many Jewish Americans that this is a coded way to refer to conspiracy theories about Jewish power and is likewise obvious for the intended hearer who picks up on and agrees with the anti-Semitic content. The more attributionally ambiguous microaggressions become, the more the targets themselves are left second-guessing the reality of what occurred. The problem is no longer explaining how one knows harmful content has been covertly communicated, but determining whether there was any harmful content at all. This creates further conditions for harm because plausible deniability allows perpetrators to intentionally gaslight targets about what has occurred, which targets are likely to internalize (Friedlaender and Ivy, 2020). The shift from obviousness to ambiguity shows an insidious adaption in microaggressions. Targets are left second-guessing whether their perception about mechanisms of oppressions is correct, which makes it increasingly harder for targets to recognize and address those mechanisms altogether. Since Pierce’s initial contrast between obliviousness and obviousness has been replaced with attributional ambiguity, microaggression theorists ought to seriously consider to what extent this obliviousness still pervades the actions of perpetrators. My goal here is not to suggest that all who commit microaggressions are doing so intentionally. Rather, it is to suggest that the hyper-focus on unintentional microaggressions has produced a narrative about who the ofenders of microaggressions are and why they ofend. This narrative broadly claims that ofenders are often wellintentioned individuals who consciously hold progressive and egalitarian social and political values and their implicit attitudes or other unconscious associations are the cause of these ofenses. However, this narrative allows intentional perpetrators to either (1) pretend their behavior was unintentional and potentially change the moral response they receive, or (2) deny they have caused any ofense at all, exploiting the ambiguity to further gaslight the target. Even more concerning, the ‘well-intentioned, progressive-value-holding ofender’ narrative frames microaggressions as a sign of social and political progress rather than another transformative preservation in the mechanisms of oppression. This is because these covert or microaggressive behaviors are used as a sign that people have shifted their social and political attitudes toward oppressed groups, that such people would never commit macroaggressions and would stop committing microaggressions if it were 242

Problem of Attributional Ambiguity

within their conscious control. Recognizing that intentional microaggressions are also common disrupts this narrative, pointing to an insidious adaptation instead of large-scale progress. Sufcient attention to intentional bad actors ensures that microaggressions are seen for what they are: an adaptive mechanism of oppression that retains the goal of its overt predecessors, whose own changes make them increasingly hard to perceive for those harmed by it. Microaggressions are not a sign that overt mechanisms of oppression are less acceptable. There is no clear evidence to suggest those who commit microaggressions are less likely to commit macroaggressions. The general narrative that microaggressions are largely committed by well-intentioned people with consciously progressive values is not only wrong, but it represents microaggressions as a sign of moral and social progress over the intentional bad actors of the past. If Michelle Alexander is right, we ought to be wary of progressive narratives because we risk thinking the harms of today are less than those of the past and mechanisms of oppression have lessened in their intensity. Contemporary microaggressions are thus uniquely concerning because attributional ambiguity ultimately leaves those in the best epistemic position to perceive the mechanisms of oppression – the oppressed – second-guessing the truth of their perception.

5 Acknowledgments Thank you to Steph Butera, Morgan Elbot, Joshua Felix, Rose Lenehan, Jessica Pedrosa, Jasper St. Bernard, members of The University of Memphis’ Minorities and Philosophy chapter and Graduate Brown Bag Series for their feedback on earlier drafts, and to the editors, Justin Khoo and Rachel Sterken, for the invitation to contribute.

Notes 1 The term ‘ofensive mechanism’ frst appears alongside the term ‘microaggression’ (Pierce, 1970). 2 Pierce argues that Black Americans can develop post-traumatic stress disorder (PTSD) as a result of these subtle but regular forms of aggression (1995). Hypervigilance is one symptom of PTSD, wherein the person is forced into a constant state of defense. 3 As Pierce writes, To do this requires deception while carrying out a pattern of plays with the deliberate intention of destroying and demolishing any opponent who chances to present a frustration. Often this guile and deception is most successful when the defense is obliged to accustom itself to your play patterns, since you, the ofence, not them the defense, know what your fnal plan with be. (270) 4 Alexander does argue that these adaptions are not inevitable but stopping this cycle would likely require mass social and political resistance and upheaval. 5 Alexander seems to suggest that all those living under the new form of control will have a harder time seeing the shift in terms of adaptation rather than progress. However, one might argue, following standpoint theory, that those who do not beneft from the racial hierarchy might be in a better position to achieve knowledge of these adaptions. For further reading on race, ignorance, and standpoint theory, see Sullivan and Tuana (2007). 6 Alexander takes this phrase from: Reva Siegal, “Why Equal Protection No Longer Protects: The Evolving Forms of Status-Enforcing Action.” Stanford Law Review 49 (1997): 1111. 7 I am not suggesting that microaggressions appear for the frst time at this moment in history, but that they start to take hold more insidiously and pervasively, marking, in line with Alexander’s analysis, a transformation from overt to covert practices that preserves the goal of the overt practices, e.g., the maintenance of white supremacy. 8 As Pierce writes, I was obligated to handle the extra stress that came from needing to reckon racial concerns of my own and of many others of all races, about what I decided, how I did it, why I did it. No White in a similar position would have had these concerns in the same qualitative and quantitative way that


Christina Friedlaender any Black [person] would have. Nor would the Whites’ decisions be as likely to be scrutinized and sharply criticized – or reversed – because of racial aspects in the decision. All minority people will concur that the stress in maintaining unity in diversity is because in the U.S.A. skin color compounds and complicates every facet of behavior. (1989; 305–6) 9 James Baldwin similarly notes that the relationship between acceptance and integration can be wrongfully contorted, remaking ‘integration’ as ‘assimilation to whiteness’ and ‘acceptance’ as ‘acceptable to white people’ (1963; 22). 10 For a related discussion of epistemic exploitation see Berenstain (2016) and Lorde (2007; 110–4). 11 For example, Derald W. Sue and Christina M. Capodilupo (2008) argue that potential targets of microaggressions are members of marginalized groups (see also: Sue, 2010b). Michael Dover (2016), Christina Friedlaender (2018), and Emily McTernan (2017) argue that targets are members of groups who are subject to structural oppression, dehumanization, or exploitation. 12 See, e.g., Brondolo et al. (2003), Sue (2010a), and Nadal (2018). 13 See, e.g.: Pierce (1988), Sue et al. (2009), Decuir-Gunby and Gunby (2016). 14 See, e.g.: Solórzano et al. (2000), Guzman et al. (2010), Kim and Kim (2010). 15 For further discussion of microaggressions targeting women of color in particular, see, e.g.: Willis (2015), Holling (2019), Nuru and Arendt (2019), Yep (2019). 16 There have been alternative taxonomies proposed by psychologists for detailing the kinds of microaggressions that occur for a particular oppressed group, as Kevin Nadal et al. (2012) do in the case of cis-sexist, or transphobic microaggressions; however, Sue’s taxonomy is the only general one, intended to apply across all forms of oppression, and has been used widely throughout the literature on microaggressions as such. However, Sarah Amira de la Garza ofers a radically diferent approach to microaggressions drawing on the work of Gloria Anzaldúa (2019). 17 This is not to discount the experience of automaticity as a source of information about the nature of microaggressions. Rather, it is to point out that this experience in and of itself does not provide sufcient evidence that microaggressions are, in fact, largely unconscious or unintentional. 18 Rethinking the role of intentionality in microaggressions connects to the question of what causes microaggressions. Because of the assumption that microaggressions are largely unconscious or unintentional, theorists tend to focus on implicit bias as the main precursor to microaggressions. However, microaggressions can be the result of other cognitive errors (e.g., the fundamental attribution error), explicit biases and consciously endorsed stereotypes, other forms of ignorance, be they individual or structurally produced (McTernan, 2017; Friedlaender, 2018; Rini, 2018; Friedlaender and Ivy, 2020; Rini, manuscript), or result from certain kinds of hermeneutical impasses between speakers (Anderson, 2017). Of course, attempting to defnitively discern the cause of any particular microaggression requires us to have impossible-to-obtain epistemic insight into the cognitive processes of perpetrators. This is why others argue that we ought to abandon defning microaggressions in terms of perpetrator motivations and instead focus on the experiences of targets in developing a defnition and taxonomy (Freeman and Stewart, 2018; Rini, manuscript); or, alternatively, that we ought to defne microaggressions as an intentional social practice, even if particular perpetrators express them unconsciously (Tremain, 2019). 19 Of course, one might argue this is a false binary because the function of male entitlement might be such that this person’s general sense of entitlement could be due to their gender privilege (i.e., they are generally disrespectful) and they are engaging in a sexist microaggression in this particular case. 20 The distinction in this case is whether a microaggression is a mind-independent social fact or a social fact determined by the particular experience of a target. 21 For additional discussion of standpoint theory and microaggressions, see Freeman and Stewart (2020). 22 If the shift is in the phenomenon (rather a conceptual shift), this ofers better explanatory power for why older examples of microaggressions do not merely seem outdated but would be macroaggressions in today’s context. For example, many of Pierce’s early examples of verbal microaggressions do not seem covert today. If microaggressions are a form of social practice that maintains an oppressor-oppressed dynamic in accordance with broader mechanisms of structural oppression, then it is unsurprising to see shifts in those practices over time. Even Sue’s widely cited early examples seem less covert today, which is likely due to the fact that increased awareness of a particular social practice as a microaggression eventually makes it overt. This means that forms of microaggression that were largely obvious to targets potentially turn into low-level macroaggressions (or perhaps they need a diferent name), paving the way for more insidious practices to take hold, which work better, in part because they tend to be more attributionally ambiguous.


Problem of Attributional Ambiguity

Work Cited Alexander, Michelle. (2012). The new Jim crow: Mass incarceration in the age of colorblindness, revised Ed. New York: The New Press. Anderson, Luvell. (2017). Hermeneutical impasses. Philosophical Topics, 45(2), 1–19. Anderson, Luvell, and Ernie Lepore. (2013a). Slurring words. Noûs, 47(1), 25–48. Anderson, Luvell, and Ernie Lepore. (2013b). What did you call me? Slurs as prohibited words. Analytic Philosophy, 54(3), 350–63. Baldwin, James. (1963). The fre next time. New York: The Dial Press. Berenstain, Nora. (2016). Epistemic exploitation. Ergo, 3, 22, 569–90. Brennan, Samantha. (2016). The moral status of micro-inequities: In favor of institutional solutions. In Michael Brownstein and Jennifer Saul (Eds.), Implicit bias and philosophy, vol.2: Moral responsibility, structural injustice, and ethics (pp. 235–53). Oxford: Oxford University Press. Brondolo, Elizabeth, Ricardo Rieppi, Kim P. Kelly, and William Gerin. (2003). Perceived racism and blood pressure: A review of the literature and conceptual and methodological critique. Annuals of Behavioral Medicine, 25, 55–65. Decuir-Gunby, Jessica T., and Norris W. Gunby. (2016). Racial microaggressions in the workplace: A critical race analysis of the experiences of African American educators. Urban Education, 51(4), 390–414. de la Garza, Sarah Amira. (2019). Facing tlahtlacolli (microaggressions) with nepantla and conocimiento: A xicana epistemological approach. In Leandra Hinojosa and Robert Gutierrez-Perez (Eds.), This bridge we call communication: Anzaldúan approaches to theory, method, and praxis (pp. 57–76). Lanham, MD: Lexington Books. Dover, Michael. (2016). The moment of microaggression: The experience of act of oppression, dehumanization, and exploitation. Journal of Human Behavior in the Social Environment, 26 (7–8), 575–86. Fatima, Saba. (2017). On the edge of knowing: Microaggression and epistemic uncertainty as a woman of color.” In Kirsti Cole and Holly Hassel (Eds.), Surviving sexism in academia: Feminist strategies for leadership (pp. 147–57). New York: Routledge. Freeman, Lauren, and Heather Stewart (2020). “Sticks and stones can break your bones and words can really hurt you: A standpoint epistemological reply to critics of the microaggression research program.” In Lauren Freeman and Jeanine Weekes Schroer (Eds.), Microaggressions and philosophy (pp. 36–66). New York: Routledge. Freeman, Lauren, and Heather Stewart (2018). Microaggressions in clinical medicine. The Kennedy Institute of Ethics Journal, 28(4), 411–49. Friedlaender, Christina. (2018). On microaggressions: Cumulative harm and individual responsibility. Hypatia, 33(1), 5–21. Friedlaender, Christina, and Veronica Ivy. (2020). A defense of intentional microaggressions and microaggressive harassment: The fundamental attribution error, harassment, and gaslighting of transgender athletes. In Lauren Freeman and Jeanine Weekes Schroer (Eds.), Microaggressions and philosophy (pp. 184–204). New York: Routledge. Guzman, Fernando, Jesus Trevino, Fernand Lubuguin, and Bushra Aryan. (2010). Microaggressions and the pipeline for scholars of color. In Derald W. Sue (Ed.), Microaggressions and marginality: Manifestation, dynamics, and impact (pp.145–67). Hoboken, NJ: Wiley. Holling, Michelle A. (2019). “You intimidate me” as a microaggressive image to discipline womyn of color faculty. Southern Journal of Communication, 84(2), 99–112. Huber, Lindsay Pérez, and Daniel G. Solorzano. (2015). Racial microaggressions as a tool for critical race research. Race, Ethnicity, Education, 3, 297–320. Kim, Suah, and Rachel H. Kim. (2010.) Microaggressions experienced by international students attending U.S. institutions of higher education. In Derald W. Sue (Ed.), Microaggressions and marginality: Manifestation, dynamics, and impact (pp. 171–91). Hoboken, NJ: Wiley. Lorde, Audre. (1984/2007). Sister outsider. Berkeley, CA: Crossing Press. McTernan, Emily. (2017). Microaggressions, equality, and social practices. Journal of Political Philosophy, 26(3), 261–81. Minikel-Lacocque, Julie. (2013). Racism, college, and the power of words: Racial microaggressions reconsidered. American Educational Research Journal, 50(3), 432–65. Morrison, Toni, Primus St. John, John Callahan, Judy Callahan, and Lloyd Baker. (1975). Black studies center public dialogue: Part 2. Portland State University. Special Collections: Oregon Public Speakers. 90. Retrieved on 18 May 2019: Nadal, Kevin. (2011). The racial and ethnic microaggressions scale (REMS): Construction, reliability, and validity. Journal of Counseling Psychology, 58(4), 470–80.


Christina Friedlaender Nadal, Kevin, Avy Skolnik, and Yinglee Wong. (2012). Interpersonal and systemic microaggressions towards transgender people: Implications for counseling. Journal of LGBT Issues in Counseling, 6(1), 55–82. Nadal, Kevin. (2018). Microaggressions and traumatic stress: Theory, research, and clinical treatment. Washington, D.C.: American Psychological Association. Nuru, Andra K., and Colleen E. Arendt. (2019). Not a safe space: Women activists of color’s response to racial microaggressions by white feminist allies. Southern Journal of Communication, 84(2), 85–98. Pierce, Chester. (1970). Ofensive mechanisms. In Floyd B. Barbour (Ed.), The black seventies (pp. 265–82). Boston: Porter Sargent. Pierce, Chester. (1988). Stress in the workplace. In Alice F. Coner-Edwards and Jeanne Spurlock (Eds.), Black families in crisis: The middle class (pp. 22–34). New York: Brunner/Mazel. Pierce, Chester. (1989). Unity in diversity: Thirty-three years of stress. In Gordon LaVern Berry and Joy Keiko Asamen (Eds.), Black students: Psychosocial issues and academic achievement (pp. 296–312). New York: Sage. Pierce, Chester. (1995). Stress analogs of racism and sexism: Terrorism, terror, and disaster. In Charles V. Wille (Ed.), Mental health, racism, and sexism (pp. 277–96). New York: Routledge. Rini, Regina. (2018). How to take ofense: Responding to microaggression. Journal of the American Philosophical Association, 4(3), 332–351. Rini, Regina. (Manuscript). The ethics of microaggressions. New York: Routledge. Siegal, Reva. (1997). Why equal protection no longer protects: The revolving forms of status-enforcing action. Stanford Law Review, 49, 1111–48. Solórzano, Daniel, Miguel Ceja, and Tara Yosso. (2000). Critical race theory, racial microaggressions, and campus racial climate: The experiences of African American college students. The Journal of Negro Education, 69(1), 60–73. Stanley, Jason. (2015). How propaganda works. Princeton, NJ: Princeton University Press. Sue, Derald Wing. (2003). Overcoming our racism: The journey to liberation. San Francisco: Jossey-Bass. Sue, Derald Wing. (2010a). Microaggressions in everyday life: Race, gender, and sexual orientation. Hoboken, NJ: Wiley and Sons. Sue, Derald Wing. (2010b). Microaggressions, marginality, and oppression: An introduction. In Derald Wing Sue (Ed.), Microaggressions and marginality: Manifestation, dynamics, and impact, 1st ed. (pp. 3–25). Hoboken, NJ: John Wiley & Sons. Sue, Derald Wing. (2017). Microaggressions and “evidence”: Empirical or experiential reality? Perspective on Psychological Science, 12(1), 170–2. Sue, Derald W., Annie I. Lin, and David P. Rivera. (2009). Racial microaggressions in the workplace: Manifestations and impact. In Jean L. Chin (Ed.), Diversity matters: Education and employment (pp. 157–72). Westport, CT: Praeger. Sue, Derald Wing, and Christina M. Capodilupo. (2008). Racial, gender, and sexual orientation microaggressions: Implications for counseling and psychotherapy. In Derald Wing Sue and David Sue (Eds.), Counseling the culturally diverse: Theory and practice (pp. 190–131). Hoboken, NJ: John Wiley & Sons. Sue, Derald Wing, Christina M. Capodilupo, Gina C. Torino, Jennifer M. Bucceri, Aisha M. B. Holder, Kevin L. Nadal, and Marta Esquilin. (2007). Racial microaggressions in everyday life: Implications for clinical practice. American Psychologist, 62(4), 271–86. Sullivan, Shannon and Nancy Tuana, eds. (2007). Race and epistemologies of ignorance. Albany, NY: SUNY Press. Tremain, Shelly. ( January 13, 2019). ‘Microaggressions’: Just another word for practices. Biopolitical Philosophy. Retrieved on: 19 May 2019: microaggressions-just-another-word-for-practices/. Wang, Jennifer, Janxin Leu, and Yuichi Shoda. (2011). When the seemingly innocuous “stings”: Racial microaggressions and their emotional consequences. Personality and Social Psychology Bulletin, 37(12), 1666–78. Willis, Tasha Y. (2015). “And we still rise…”: Microaggressions and intersectionality in the study aboard experiences of black women. Frontiers: The Interdisciplinary Journal of Study Aboard, 26 (Fall), 209–30. Wong, Gloria, Annie O. Derthick, E. J. R. David, Anne Saw, and Sumie Okazaki. (2013). The what, the why, and the wow: A review of racial microaggressions research in psychology. Race and Social Problems, 6(2), 181–200. Yep, Gust A. (2019). A thick intersectional approach to microaggressions. Southern Communication Journal, 84(2), 113–26.



1 Introduction This chapter defends a novel analysis of a phenomenon that Miranda Fricker calls hermeneutical injustice. In Sections 2 and 3, I identify two distinct defnitions of hermeneutical injustice supplied by Fricker in her groundbreaking book Epistemic Injustice: Power and the Ethics of Knowing (2007). I argue that it is not obvious how these defnitions are related, or how to resolve the discrepancy between them. I defend an alternative, recursive defnition of hermeneutical injustice and argue that it is more ideologically parsimonious and has greater explanatory power than either of Fricker’s defnitions. I conclude by distinguishing hermeneutical injustice from closely related phenomena such as white ignorance (Mills 1997, 2007, 2017), willful hermeneutical ignorance (Pohlhaus 2012), and contributory injustice (Dotson 2012).

2 First Defnition of Hermeneutical Injustice In Epistemic Injustice: Power and the Ethics of Knowing, Miranda Fricker ofers two distinct defnitions of hermeneutical injustice. She introduces the frst defnition of hermeneutical injustice in conjunction with her discussion of Carmita Wood, who, according to Fricker, is unable to properly understand or describe her experience of workplace mistreatment because she lacks the concept sexual harassment. In her discussion of this case, Fricker defnes hermeneutical injustice as the injustice of having some signifcant area of one’s social experience obscured from collective understanding owing to a structural identity prejudice in the collective hermeneutical resource. (2007, 155) However, Fricker ofers very few clarifying remarks about the above defnition. First, she does not explain what it means to say that someone’s social experience is “obscured from collective understanding.” Second, she does not explain what hermeneutical resources are, or what makes them collective. Third, Fricker does not give an account of the nature of “structural identity prejudice.” I will address each of these elements in turn. Let us say that someone’s social experience is “obscured from collective understanding” when most people (which may include the person who has the experience in question) do not understand the nature or normative signifcance of that experience. For example, in the 1970s, most 247

Rebecca Mason

people did not understand the nature or normative signifcance of subjecting women to unwanted sexual advances at work. In particular, most people did not understand that subjecting women to unwanted sexual advances at work constitutes a form of wrongful and harmful harassment (not firting). Thus, women’s experiences of being sexually harassed were obscured from collective understanding. Similarly, until very recently, most people did not understand the nature or normative signifcance of women’s experiences of extreme sadness, mood swings, intense irritability, hopelessness, anxiety, and low energy following childbirth. In particular, most people did not understand that those experiences constitute a physiological condition and are not something for which women are morally blameworthy. Thus, women’s experiences of postpartum depression were obscured from collective understanding. Next, we need to clarify the nature of the collective hermeneutical resource. Let’s say that hermeneutical resources are the cognitive and linguistic tools (i.e. concepts and words) that we use to understand the world and to communicate with one another about it. For example, the concept sexual harassment and the corresponding term “sexual harassment” are hermeneutical resources. They enable us to think and talk about sexual harassment. According to Fricker, prior to the 1970s, the concept sexual harassment and the corresponding term “sexual harassment” did not belong to the collective hermeneutical resource. On her view, this is why women’s experiences of being sexually harassed were obscured from collective understanding. In virtue of what do some concepts and words belong (or fail to belong) to the collective hermeneutical resource? In later work, Fricker says that the collective hermeneutical resource includes “meanings that just about anyone can draw upon and expect those meanings to be understood across social space by just about anyone else. The collective hermeneutical resource contains those concepts and conceptualizations that are held in common” (Fricker 2016, 163 emphasis in original). In other words, some concepts and terms belong to the collective hermeneutical resource if and only if the majority of people possess those concepts and are competent with the terms that are used to express them.1 Lastly, we need to clarify the nature of structural identity prejudice, and how the collective hermeneutical resource instantiates it. Identity prejudice, according to Fricker, is a property of individuals. In particular, identity prejudice characterizes someone’s beliefs or judgments (2007, 35). Fricker argues that “prejudices are judgments, which may have a positive or a negative valence, and which display some (typically, epistemically culpable) resistance to counter-evidence owing to some afective investment on the part of the subject” (35). On her account, negative identity prejudice produces testimonial injustice. A negative identity prejudice is a prejudice “with a negative valence held against people qua social type” (35). For example, someone’s belief that women are not reliable testifers exhibits negative identity prejudice. Given this characterization of identity prejudice, however, it is not at all clear what it means to say that the collective hermeneutical resource instantiates structural identity prejudice, except to suggest, rather obliquely, that the structural variety of identity prejudice is non-agential. Fricker says that the collective hermeneutical resource is structurally prejudiced when it supplies interpretations of a group’s social experiences that are biased (2007, 155) and that these biased interpretations are caused by “persistent and wide-ranging hermeneutical marginalization” (155). However, besides giving a few sketchy examples, she does not develop an account of what it is for the collective hermeneutical resource to be biased due to hermeneutical marginalization. Thus, the nature of structural identity prejudice remains opaque on Fricker’s account. However, Fricker’s second defnition of hermeneutical injustice makes no mention of structural identity prejudice in the collective hermeneutical resource, and it is that defnition to which I turn in the next section. 248

Hermeneutical Injustice

3 Second Defnition of Hermeneutical Injustice So much for Fricker’s frst defnition of hermeneutical injustice. Fricker provides a second defnition of hermeneutical injustice only a few pages after the frst, this time in the context of her discussion of a case of hermeneutical injustice involving a character named Joe from Ian McEwan’s novel Enduring Love. Fricker’s second defnition of hermeneutical injustice, unlike the frst, does not make reference to the collective hermeneutical resource or to structural identity prejudice. Instead, it invokes the phenomenon of hermeneutical marginalization, which is not included in Fricker’s frst defnition of hermeneutical injustice. Fricker states that her second defnition of hermeneutical injustice is a “generic” one that defnes hermeneutical injustice “per se” (2007, 158). According to the generic defnition, hermeneutical injustice is The injustice of having some signifcant area of one’s social experience obscured from collective understanding owing to hermeneutical marginalization. (2007, 158) In other words, a subject, S, sufers hermeneutical injustice when most people fail to understand the nature S’s social experiences because S belongs to a social group, G, that is hermeneutically marginalized. According to Fricker, social group is hermeneutically marginalized when members of that group, in virtue of their membership in that group, are unable to participate equally in those social institutions and practices through which hermeneutical resources are generated and publicized, e.g. government, law, journalism, and academia. As a result, members of these groups are unable to participate equally in the creation and dissemination of hermeneutical resources (i.e. by constructing legislation and social policy, creating media content, producing scholarly research, etc.). Though Fricker’s frst and second defnitions of hermeneutical injustice are related, they are nonetheless distinct. According to the frst defnition, but not the second, hermeneutical injustice essentially involves structural identity prejudice in the collective hermeneutical resource. Let’s call Fricker’s frst defnition of hermeneutical injustice “HI-1.” According to the second defnition, but not the frst, hermeneutical injustice is essentially caused by hermeneutical marginalization. Let’s call Fricker’s second defnition of hermeneutical injustice “HI-2.” This is not to say that there is no connection between HI-1 and HI-2. To the contrary, hermeneutical marginalization is supposed to cause structural identity prejudice in the collective hermeneutical resource. However, this does not resolve this discrepancy between the frst and second defnition of hermeneutical injustice. It remains the case that there is more than one defnition of hermeneutical injustice on ofer, and that Fricker has not provided a clear explanation of how we should understand the relationship between them. However, given that Fricker describes the second defnition of hermeneutical injustice as generic, one possibility is that HI-1 is a species of HI-2. The genus-species relation is traditionally understood in conjunctive terms; that is, in terms of a conjunction of a generic property and one or more diferentia, i.e. properties that diferentiate the species. In other words, a species, s, is defned by a generic property, F-ness, and those properties, G1…Gn (the diferentia) that distinguish s from other species of F. For example, the species triangle belongs to the genus polygon. The generic property that diferentiates triangles from other species of polygons (e.g. pentagon, hexagon, chiliagon) is the property of being three-sided. Thus, a triangle is a three-sided polygon; a pentagon is a fvesided polygon; a hexagon is a six-sided polygon, etc. What each species of the genus polygon has in common is the property that defnes their genus, viz. the property of being a plane fgure that is bounded by a fnite, closed chain of straight-line segments; what diferentiates the various species of polygons from one another is the number of straight-line segments by which they are bounded. 249

Rebecca Mason

HI-1 and HI-2 have a property in common which could serve as the generic property, viz., having some signifcant area of one’s social experience obscured from collective understanding. Moreover, HI-1 has a property that diferentiates it from HI-2; in particular, it has the property of being produced by structural identity prejudice in the collective hermeneutical resource. However, HI-1 is not a species of HI-2 because HI-2 also has a property that diferentiates it from HI-1. In particular, it has the property of being produced by hermeneutical marginalization. Therefore, HI-1 and HI-2 do not stand in the genus-species relation. There is more than one way to proceed at this point. First, it is possible that these defnitions are related in some other way that I have not considered. Second, the discrepancy between them could be avoided if we were to reject one defnition in favor of the other. However, in Sections 4 and 5, I pursue a third strategy. Specifcally, I develop an alternative defnition of hermeneutical injustice. I argue my defnition is more ideologically parsimonious than either of Fricker’s because it does not rely on the notions of structural identity prejudice or hermeneutical marginalization.2 Moreover, it has greater explanatory power. This gives us a reason to prefer my defnition of hermeneutical injustice over either of Fricker’s defnitions.

4 Hermeneutical Injustice Redefned In this section, I defend an alternative defnition of hermeneutical injustice and argue that is preferable to Fricker’s because it is more ideologically parsimonious. In particular, it does not require an account of either structural identity prejudice or hermeneutical marginalization. I begin by stating two necessary conditions which aim to capture the salient features of Fricker’s central case of hermeneutical injustice, i.e. Carmita Wood. A subject, S, sufers hermeneutical injustice only if i ii

S is unable to understand the nature or normative signifcance of their social experience, e, or S is unable to describe the nature or normative signifcance of e in a way that most people can understand.3

Wood was unable to understand or to describe the nature and normative signifcance of her experience of being sexually harassed at work in a way that most other people could understand. When Wood tried to explain why she had quit her job, she “was at a loss to describe the hateful episodes” (Brownmiller 1999, 280–1). Thus, Carmita Wood’s case satisfes (ii). Though (i) and (ii) are necessary for hermeneutical injustice, they are not sufcient. This is because it is possible for (i) or (ii) to be satisfed in a way that is not unjust. Consider the following case: Jef is the wealthy founder, CEO and President of a multi-national corporation. Jef regularly enjoys dining at Michelin Star restaurants where he drinks expensive wines with each course of his meal. After a particularly enjoyable dinner one night, however, he fnds himself unable to describe, in a way that most other people can understand, “that je ne sais quoi of drinking a particular expensive wine at a lavish dinner” (Prescott-Couch 5). Thus, Jef satisfes (ii). That is, Jef is unable to describe his social experience of drinking a particular wine at a lavish dinner in such a way that most other people can understand. However, his inability to do so is not unjust.4 Therefore, the defnition of hermeneutical injustice needs to include a condition that specifes that the satisfaction of conditions (i) or (ii) constitutes hermeneutical injustice only if they are satisfed due to injustice.5 In other words, the defnition of hermeneutical injustice needs to include a recursive condition specifying that someone is subject to hermeneutical injustice only if the frst or second condition is satisfed in a way that constitutes injustice: 250

Hermeneutical Injustice

i ii iii

S is unable to understand the nature or normative signifcance of their social experience, e, or S is unable to describe the nature or normative signifcance of e in a way that most people can understand, and The satisfaction of conditions (i) or (ii) is unjust.

It is beyond the scope of this chapter to give a comprehensive account of when the satisfaction of (i) or (ii) constitutes injustice. I am inclined to think that there is more than one way of doing so, and that philosophers with diferent conceptions of justice may disagree about what makes the satisfaction of either of the frst two conditions unjust. Perhaps (iii) is satisfed when someone is unfairly prevented from understanding or articulating the nature or normative signifcance of their inchoate social experiences.6 However, unfairness need not be the only reason why the satisfaction of (i) or (ii) constitutes injustice. To the contrary, it is possible that diferent cases in which (i) or (ii) is satisfed may be unjust for diferent reasons. The addition of condition (iii) in the above defnition yields the desired result that Jef, the wealthy CEO, does not sufer hermeneutical injustice when he fnds that he cannot describe the “je ne sais quoi” of drinking a particular expensive wine, for the inefability of Jef’s experience is not plausibly unjust. However, (i)–(iii) still do not state necessary and sufcient conditions for hermeneutical injustice. To see why, consider a case in which (ii) is satisfed because S is physically gagged. In that case, S is unable to describe the nature or normative signifcance of her social experiences in a way that anyone can understand, for the gag makes her speech incomprehensible. Moreover, suppose that it is unjust that S is physically gagged. In that case, (iii) is satisfed as well. However, intuitively, this is not an instance of hermeneutical injustice. To avoid this result, I argue that a fourth condition should be added to the defnition of hermeneutical injustice; in particular, I argue that someone sufers hermeneutical injustice only if (i) or (ii) is satisfed because the collective hermeneutical resource is defcient or distorted.

5 Defcits and Distortions in the Collective Hermeneutical Resource According to Fricker, instances of hermeneutical injustice have a common etiology. In particular, she argues hermeneutical injustice is always caused by “a paucity of shared concepts” (2016, 170). For example, according to Fricker, Carmita Wood was subject to hermeneutical injustice because the collective hermeneutical resource did not contain the concept of sexual harassment, or the corresponding term. Similarly, she argues that Wendy Sanford was unable to understand the nature and normative signifcance of her experience because the concept of postpartum depression was not widely shared. Thus, according to Fricker, hermeneutical injustice occurs only if the collective hermeneutical resource is defcient. In light of this, one might argue that the defnition of hermeneutical injustice should be amended so that a subject, S, sufers hermeneutical injustice if and only if i ii iii iv

S is unable to understand the nature or normative signifcance of their social experience, e, or S is unable to describe the nature or normative signifcance of e in a way that most people can understand, and Conditions (i) or (ii) are satisfed because the collective hermeneutical resource is defcient, and The satisfaction of conditions (i) or (ii) and (iii) is unjust.

This defnition yields the desired result that individuals who are unjustly gagged or otherwise prevented from speaking do not sufer hermeneutical injustice, for, although they are unable to describe the nature or normative signifcance of their social experiences, they are not unable to do so because the collective hermeneutical resource is defcient. 251

Rebecca Mason

However, although the cases of hermeneutical injustice involving Carmita Wood and Wendy Sanford arise from gaps in the collective hermeneutical resource, Fricker also considers a case of hermeneutical injustice that is not produced by conceptual impoverishment—that of the anonymous protagonist in Edmund White’s autobiographical novel, A Boy’s Own Story. In White’s novel, the unnamed protagonist struggles to make sense of his nascent homosexual desires. The boy describes his bewilderment as follows: I never doubted that homosexuality was a sickness; in fact, I took it as a measure of how unsparingly objective I was that I could contemplate this very sickness. But in some other part of my mind I couldn’t believe that the Lysol smell must bathe me, too, that its smell of stale coal fumes must penetrate my love for Tom. Perhaps I became so vague, so exhilarated with vagueness, precisely in order to forestall a recognition of the final term of the syllogism that begins: If one man loves another, he is a homosexual, I love a man… (White 1982, 117–8) Here, the confusion the protagonist describes does not stem from the fact that he lacks the concept of homosexuality. He not only possesses the concept of homosexuality, that concept belongs to the collective hermeneutical resource. Thus, insofar as the protagonist in White’s novel suffers hermeneutical injustice, it is not caused by “a paucity of shared concepts,” as Fricker maintains. Rather, what is going on in this case is that the young protagonist is reluctant to apply the shared concept of homosexuality to himself. Moreover, he is reluctant to do so because of the way in which homosexuality was understood at the time.7 In other words, the protagonist fails to understand his love for Tom as homosexual because he has internalized the descriptions of homosexuality as a “sickness” or a “stage,” and, as a result, is loath to apply the concept of homosexuality to his feelings for Tom. If hermeneutical injustice is always caused by conceptual deficiency, then this is not a case of hermeneutical injustice. However, intuitively, this is a case of hermeneutical injustice. Like Carmita Wood and Wendy Sanford, White’s protagonist is not able to understand the nature and normative significance of a particularly important social experience of his. Therefore, it is not the case that hermeneutical injustice is always caused by conceptual deficiency. My view is that although some instances of hermeneutical injustice involve failures of concept possession (Wood and Sanford lacked the concepts of sexual harassment and postpartum depression, respectively), others involve failures of concept application.8 For example, the boy in White’s novel possesses the concept he needs to understand his love for Tom, i.e. homosexuality. However, he finds himself unable to make sense of that experience because he is unwilling to apply that concept to himself given its pathological and heteronormative associations. Cases such as these proliferate. In her book, I Never Called it Rape, Robin Warsaw (2019) recounts the experience of Lori, a college student who was raped by an acquaintance while on a date. Warsaw describes how Lori struggled to understand the nature and normative significance of this experience (i.e. that it was sexual assault and that it was not her fault but his). Indeed, she takes the name of her book from that experience: Lori never called it “rape” (at least not initially). As Katharine Jenkins argues, women like Lori are often unable to understand the nature or normative significance of their experience of being raped by people they know (family, friends, spouses) due to prevalent myths about rape and domestic violence. Jenkins points out that “such myths function to obscure understandings of these phenomena including victims’ understanding of their own experiences” (2017, 192).9 In particular, rape myths perpetuate the idea that rape is perpetrated by strangers (often racialized strangers) rather than friends, family members, or intimate partners. Women like Lori possess the concept of rape. Indeed, that concept is widely shared; therefore, it belongs to the collective hermeneutical resource. Nonetheless, they fail to apply it to their 252

Hermeneutical Injustice

experiences of sexual assault due to, e.g., the myth that rape is committed by strangers. Thus, women like Lori present another case of hermeneutical injustice involving a failure of concept application rather than a failure of concept possession.10 A theory of hermeneutical injustice should capture cases such as these. To meet this desideratum, my view is that hermeneutical injustice can occur either because the collective hermeneutical resource is deficient or because it is distorted. For example, both Wood and Sanford failed to understand the nature and normative significance of their experiences because the collective hermeneutical resource did not include the concepts of sexual harassment and postpartum depression. These are examples of hermeneutical injustice that arise from deficits in the collective hermeneutical resource. By contrast, in cases of hermeneutical injustice that arise from failures of concept application, subjects possess the requisite concepts but fail to apply them in ways that illuminate the nature and normative significance of their social experiences.11 They fail to do so, I argue, because the collective hermeneutical resource is distorted. The protagonist in White’s novel and women like Lori provide examples of hermeneutical injustice caused by distortions in the collective hermeneutical resource.12 To understand how the collective hermeneutical resource can be distorted, we need a more refined understanding of it. Recall that hermeneutical resources are the cognitive and linguistic tools (i.e. concepts and words) that we use to understand the world and to communicate with one another about it, and that the collective hermeneutical resource is comprised of those concepts and terms that are shared by everyone. In what follows, I argue that the collective hermeneutical resource is not merely a jumble of words and concepts but also instantiates a structure. This idea can be usefully clarified by adapting Katherine Ritchie’s view of organized social groups (Ritchie 2013, 2015, 2020). According to Ritchie, organized social groups instantiate structures. On her view, structures are networks of nodes and edges: “Nodes represent positions or places that can be occupied by objects. Edges represent relations that hold between nodes (or node-occupiers)” (forthcoming, 4). She argues that nodes are occupied by people and edges represent relations between people, or between the nodes that they occupy. Consider an organized social group like a baseball team.13 On Ritchie’s view, individual baseball teams are comprised of nodes, occupied by players, and relations between the nodes the players occupy. More specifically, baseball teams instantiate a structure which has nodes for pitcher and catcher. On the minor league baseball team the Albuquerque Isotopes, for example, the pitcher node is occupied by Logan Cozart (among others) and the catcher node is occupied by Chris Rabago. As on any baseball team, those who occupy the catcher node are functionally related to those who occupy the pitcher by the return ball to relation. My suggestion is that the collective hermeneutical resource should be understood in an analogous way. In particular, the collective hermeneutical resource instantiates a structure comprised of nodes and edges. On Ritchie’s view of organized social groups, the nodes are occupied by people. By contrast, on my view of the collective hermeneutical resource, the nodes are occupied by concepts and the terms we use to express them. The edges represent inferential relations among the concepts and terms which occupy the nodes. For example, the concept koala (and the corresponding term “koala”) belongs to the collective hermeneutical resource and is inferentially related to other concepts such as animal and marsupial (along with the corresponding terms), which also belong to the collective hermeneutical resource. Which relations the collective hermeneutical resource instantiates are determined by which concepts and terms belong to the collective hermeneutical resource, and the inferential dispositions of those individuals who rely on those concepts and terms to understand the world and communicate about it with one another. Let’s say that the inferential relations instantiated by the collective hermeneutical resource are thick when very many people are disposed to make the 253

Rebecca Mason

relevant inferences. By contrast, the inferential relations instantiated by the collective hermeneutical resource are thin when few people are disposed to make the relevant inferences. Conceiving of the collective hermeneutical resource in this way enables me to explain how the collective hermeneutical resource can be distorted. On my view, the collective hermeneutical resource is distorted when the terms and concepts that comprise it are inferentially related in ways that are invalid or inductively weak. For example, in the 1950s, most people were disposed to infer illness, unnatural, and deviant from the concept of homosexuality. Therefore, at that time, the collective hermeneutical resource instantiated thick inferential relations between concepts such as homosexuality, illness, unnatural, and deviant. The collective hermeneutical resource was distorted because inferential relations between these concepts and terms are invalid (NB: someone’s being a bachelor entails that they are unmarried, but someone’s being homosexual does not imply that they have an illness). Similarly, very many people are disposed to infer stranger (rather than boyfriend or husband) and resistance (rather than submission) from the concept of rape. Therefore, the collective hermeneutical resource instantiates thick inferential relations between rape, stranger, and resistance and only thin inferential relations between rape and boyfriend or husband. The thick inferential relations between rape, stranger, and resistance distort the collective hermeneutical resource because they are inductively weak: rape victims are much less likely to have been assaulted by a stranger than by a friend, family member, or intimate partner. When the collective hermeneutical resource is distorted in these ways, hermeneutical injustice arises not from a failure of concept possession but from a failure of concept application. Thus, though White’s protagonist has the concept of homosexuality he does not apply it to his experience of loving Tom. As a result, he struggles to understand the nature and normative signifcance of his feelings. Similarly, though Lori has the concept of rape, she does not apply it to her experience of being sexually assaulted while on a date with an acquaintance. Thus, she struggles to understand the nature and normative signifcance of what happened to her. Putting this all together, we can say that a subject, S, sufers hermeneutical injustice if and only if S is unable to understand the nature or normative signifcance of their social experience, e, or S is unable to describe the nature or normative signifcance of e in a way that most people can understand, and iii Conditions (i) or (ii) are satisfed because the collective hermeneutical resource is defcient or distorted, and iv The satisfaction of conditions (i) or (ii), and (iii) is unjust. i ii

This defnition yields the desired result that the anonymous protagonist from White’s novel and women like Lori sufer hermeneutical injustice even though these cases do not involve conceptual defciency. Both Lori and White’s protagonists are unable understand the nature or normative signifcance of their social experiences not because the collective hermeneutical resource is defcient, but because it is distorted. Finally, my defnition of hermeneutical injustice is preferable to either of Fricker’s defnitions because it does not require an account of structural identity prejudice or hermeneutical marginalization. That is, my defnition of hermeneutical injustice is more ideologically parsimonious than either of Fricker’s defnitions of that phenomenon. Moreover, it has more explanatory power insofar as it can account for a greater range of cases of hermeneutical injustice; in particular, it can account for cases such as those involving the protagonist from A Boy’s Own Story and women like Lori. This gives us a prima facie reason to prefer my defnition to either of Fricker’s. 254

Hermeneutical Injustice

6  Hermeneutical Injustice and Related Phenomena Before concluding, I would like to consider the question of how to distinguish hermeneutical injustice from related phenomena such as white ignorance (Mills 1997, 2007, 2017), willful hermeneutical ignorance (Pohlhaus 2012), and contributory injustice (Dotson 2012). Fricker argues that hermeneutical injustice and white ignorance are distinct at least partly because the former but not the latter always involves conceptual deficiency.14 On her view, white ignorance “does not generally involve any paucity of concepts on anyone’s part,” whereas in instances of hermeneutical injustice “there is always, definitively, a paucity of shared concepts” (Fricker 2016, 170). However, if hermeneutical injustice does not always arise from gaps in the collective hermeneutical resource, as I have argued, how should we distinguish white ignorance from hermeneutical injustice? According to Charles Mills (1997, 2007, 2017), white ignorance is a form of non-accidental, racially motivated irrationality. Consider, for example, the persistent belief, despite considerable evidence to the contrary, that racial inequality in the United States is caused by a “culture of poverty” among Black Americans rather than by historical injustice (e.g. slavery, Jim Crow laws) and anti-Black racism. This, according to Mills, is an instance of white ignorance. That is, a cognitive dysfunction that occurs because individuals are unwilling to understand the ways in which the world and their place in it is structured by race and racial oppression. On my view, white ignorance and hermeneutical injustice are distinct because white ignorance causes deficits and distortions in the collective hermeneutical resource, whereas hermeneutical injustice is caused by deficits and distortions in the collective hermeneutical resource. For example, one reason why the collective hermeneutical resource is deficient and distorted is that white people are unwilling to understand the ways in which the world and their place in it manifests racism and racial oppression. Thus, the collective hermeneutical resource instantiates thick inferential relations between the concepts of racial inequality and black culture and comparatively thin inferential relations between racial inequality, slavery, and racism. For the same reason, hermeneutical injustice is distinct from willful hermeneutical ignorance (Pohlhaus 2012) and contributory injustice (Dotson 2012). Willful hermeneutical ignorance occurs when members of dominant social groups dismiss or ignore the interpretive resources created by members of marginalized groups to conceptualize and describe their experiences (Pohlhaus 2012). For example, consider someone who dismisses or ignores concepts like gender identity or pansexual as used by members of the LGBTQIA+ community. Contributory injustice occurs when someone willfully refuses to use or even recognize alternative hermeneutical resources, and uses dominant, often ill-­ fitting ones instead (32). For example, consider someone who insists that racial disparities in college graduation rates are “meritocratic” and refuses to use the concept of structural racism to understand why Black and Hispanic college students are less likely to graduate than their White peers. Like white ignorance, willful hermeneutical ignorance and contributory injustice produce and maintain deficits and distortions in the collective hermeneutical resource, whereas hermeneutical injustice is produced by those deficits and distortions. For example, the collective hermeneutical resource will be deficient—and therefore liable to give rise to hermeneutical injustice—insofar as most people refuse to employ or recognize concepts and terminology created by members of the LGBTQIA+ community to understand and describe their experiences. To put the point another way, white ignorance, willful hermeneutical ignorance, and contributory injustice are done by members of dominant groups. When members of dominant social groups perpetrate white ignorance, willful hermeneutical ignorance, or contributory injustice, they produce and maintain deficits and distortions in the collective hermeneutical resource that give rise to instances of hermeneutical injustice. By contrast, hermeneutical injustice is something that happens to members of oppressed and marginalized social groups because the collective hermeneutical resource is deficient or distorted. 255

Rebecca Mason

Other phe